article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
the cold dark matter ( cdm ) paradigm successfully describes many aspects of the formation of large - scale structure in the universe ( lahav _ et al . _ 2001
; peacock _ et al . _ 2001
; navarro , frenk & white 1997 ; moore _ et al . _
galaxy - scale dark matter halos , however , exhibit several apparent inconsistencies with cdm , for example : the number of milky way satellites appears to be at least an order of magnitude lower than cdm predictions ( kauffman , white , & guideroni 1993 ; moore _ et al . _ 1999a
; klypin _ et al . _ 1999 ) , and dark matter halos in dwarf and low surface brightness galaxies are much less cuspy than in cdm simulations ( burkert 1995 ; mcgaugh & de blok 1998 ; moore _ et al . _
some reports ( tyson , kochanski & dellantonio 1998 ; smail _ et al . _ 2000 ) even suggest that cdm fails on galaxy cluster scales for some clusters , but the latter are controversial ( broadhurst _ et al . _ 2000
; shapiro & iliev 2000 )
. proposed modifications of cdm include self - interacting dark matter ( spergel & steinhardt 2000 ; firmani _ et al . _ 2000
) , warm dark matter ( hogan & dalcanton 2000 ) , annihilating dark matter ( kaplinghat , knox & turner 2000 ) , scalar field dark matter ( hu & peebles 2000 ; goodman 2000 ) , and mirror matter ( mohapatra , nussinov & teplitz 2002 ) , each of which is invoked to soften the core density profile .
many of these modifications will soften the core profile of galaxy clusters as well , although other astrophysical processes such as the adiabatic contraction of core baryons ( hennawi & ostriker 2002 ) may ameliorate this effect .
baryons , however , introduce a host of complications to cdm simulations ( frenk 2002 ) . in order better to discriminate among cdm , its modifications , and other astrophysical influences , we are mapping the dark matter profiles of a sample of galaxy cluster cores .
we use imaging spectroscopy from the _ chandra x - ray observatory _ ( our own observations
arabadjis , bautz & garmire 2002 and those in the _ chandra _ archive ) to deproject the radial profile of the intracluster medium ( icm ) density and temperature for each cluster . in order to extract a dark matter profile from spatially resolved
x - ray spectroscopy one usually assumes that the galaxy cluster is spherically symmetric and in hydrostatic equilibrium , and so for the most part we have restricted our sample to clusters for which , to judge from their _ chandra _ x - ray images , these assumptions appear to be valid .
as noted by allen ( 1998 ) , clusters in hydrostatic equilibrium often contain `` cooling flows , '' or cool components in their cores , and these components must be properly modelled if reliable inferences about cluster mass are to be drawn .
if the model of the icm contains only a single emitting component ( at temperature @xmath4 and density @xmath5 ) at each radius , the inferred temperature profile will tend to dip significantly toward the center of a cluster with a cool component .
if , however , gas in the the core contains a second ( cooler ) component which is cospatial and isobaric with the first , then a hot icm component coexists with the cool component in the core .
the latter case tends to produce a larger central mass ( see figures [ fig1 ] ) and steeper density profile than the former .
our problem is to choose between a simple model m@xmath6 , representing a single - com- ponent core icm , and a complex model m@xmath7 , representing two - phase gas . in order to test for the presence of a second emission component we adopt a simplified geometry containing only two spherical shells ( inner = 1 , outer = 2 ) . in both models ( m@xmath6 and m@xmath7 ) , shell 2 contains a ( hot ) thermal plasma at temperature @xmath8 and density @xmath9 .
model m@xmath6 contains only one emission component in shell 1 , characterized by a temperature @xmath10 and a density @xmath11 , whereas model m@xmath7 contains a hot and a cool emission component in shell 1 , described by @xmath10 , @xmath11 , @xmath12 , and @xmath13 .
the x - ray emission from each component is modelled spectroscopically using the mekal model ( see , e.g. , liehdal , osterheld & goldstein 1995 ) as implemented in the xspec software package ( arnaud 1996 ) .
the best - fit parameter values of each model are calculated using a @xmath14 minimization routine .
hereafter we will refer to the simple and complex model parameters using vectors @xmath15 and @xmath16 , respectively ; i.e. , @xmath17 and @xmath18 .
it might seem straightforward to apply a conventional test such as the likelihood ratio test or the @xmath19-test ( bevington 1969 ; cash 1979 ) to choose between the two models .
however , since @xmath15 lies on a boundary of @xmath16 ( with @xmath20 ) , these tests can not be employed ( protassov _ et al . _
instead , we construct an _ empirical _ @xmath19-distribution using markov chain monte carlo ( mcmc ) sampling , and gauge the significance of the complex model from the location of the @xmath19 value of the data within that distribution ( protassov _ et al . _
2002 ) .
we briefly summarize our computation of an empirical `` @xmath19-distribution '' here ; details will be presented elsewhere ( arabadjis & bautz , in preparation ) .
starting with the best - fit parameters @xmath21 of model m@xmath6 , we sample the 4d parameter space in its vicinity using mcmc sampling .
this is done by running a tcl script within xspec which calculates the probability distribution function @xmath22 of a trial perturbation @xmath23 about @xmath21 given the observed data .
the trial point is chosen using the trial distribution function @xmath24 .
the choice of @xmath25 is arbitrary ; we restrict ourselves to functions which are symmetric in parameter space transitions , i.e. @xmath26 .
this new parameter set is accepted if @xmath27 exceeds a random number on [ 0,1 ] . if not , the trial point is rejected and new one is selected .
the sequence of accepted @xmath28 is a markov chain whose stationary distribution follows @xmath29 ( lewis & bridle 2002 ) .
we repeat this procedure until we have 100 values of @xmath30 for model m@xmath6 . for each of the parameter sets in the sample we simulate an x - ray spectrum , taking proper account of the instrument response and photon statistics .
we then model each of the simulated data sets using both m@xmath6 and m@xmath7 , and tabulate the @xmath19-value of each data set : where @xmath32 is the number of degrees of freedom of the simple model .
( in practice , the normalization can be ignored . ) the @xmath19-distribution for the cooling flow cluster emss 1358 + 6245 is shown in panel b of figure [ fig1 ] .
the @xmath19-value of the original _ chandra _ data set is indicated with a dashed line .
one- and two - temperature mass profiles ( dotted / triangles and solid / circles , respectively ) of emss 1358 + 6245 ; [ b ] empirical @xmath19-distribution for models @xmath33 and @xmath34 of emss 1358 + 6245 .
[ fig1 ] ] of the 100 mcmc simulations that were run , only two resulted in an @xmath19-value which exceeded that of the data that is , if @xmath33 were the correct description , an @xmath19-value as large as that observed would occur with a probability of only 2% meaning that the model with a separate , co - spatial cool component is preferred .
the result is that a model with a steeper density profile and a larger central mass is favored . if this trend obtains for most cooling flow clusters , it may rule out several of the cdm modifications .
we have yet to apply the foregoing analysis to discriminate between 1- and 2-temperature models to all clusters in our sample . instead ,
here we merely illustrate , in figure [ fig2 ] , the sensitivity of the derived mass profile and inferred density slope in the cluster core to the assumed state of the intracluster medium for five reasonably relaxed clusters .
these results have been obtained using the deprojection methods described in arabadjis , bautz & garmire ( 2002 ) .
the left panel of figure [ fig2 ] shows that , generally , the inferred mass is larger , the encircled mass profile marginally flatter , and , in consequence , the density somewhat steeper in the 2-temperature model . the right panel of figure [ fig2 ] shows the logarithmic density slope parameter @xmath35 ( here defined so that the central density @xmath2 ) inferred from the mass profile for each cluster .
( we plot @xmath35 against redshift merely as a convenient means of display . )
the indices shown were obtained by fitting the mass profile within @xmath36 h@xmath1 kpc , which is typically about one - tenth the virial radius inferred from the best - fit nfw profiles for these objects .
we note that the core slopes are generally consistent with , or slight steeper than , the standard cdm results ( @xmath37 ; navarro , frenk & white 1997 ; moore _ et al . _
1999b ) .
we have illustrated a rigorous and quantitative procedure for distinguishing between emission models of cool components in clusters .
we have presented mass profiles derived from _
chandra _ x - ray images for five clusters , using both 1- and 2-temperature models for the central x - ray emission . in almost all cases
we find that the slope of the total central mass density is consistent with , or slightly steeper than , standard cdm predictions . | galaxy cluster mass distributions offer an important test of the cold dark matter picture of structure formation , and may even contain clues about the nature of dark matter .
x - ray imaging spectroscopy of relaxed systems can map cluster dark matter distributions , but are usually complicated by the presence of central cool components in the intracluster medium . here
we describe a statistically correct approach to distinguishing amongst simple alternative models of the cool component , and apply it to one cluster .
we also present mass profiles and central density slopes for five clusters derived from _
chandra _ data , and illustrate how assumptions about the cool component affect the resulting mass profiles . for four of these objects , we find that the central density profile ( @xmath0h@xmath1 kpc ) @xmath2 with @xmath3 , for either of two models of the central cool component
. these results are consistent with standard cdm predictions .
[ 1996/06/01 ] a&a |
we thank d. stauffer for a careful reading of this manuscript .
part of this work was realized during the `` complexity and chaos '' workshop at isi , under grant ofes 95.0046 . | a simple cellular automata model for a two - group war over the same `` territory '' is presented .
it is shown that a qualitative advantage is not enough for a minority to win . a spatial organization as well a definite degree of aggressiveness are instrumental to overcome a less fitted majority .
the model applies to a large spectrum of competing groups : smoker - non smoker war , epidemic spreading , opinion formation , competition for industrial standards and species evolution . in the last case
, it provides a new explanation for punctuated equilibria .
.5cm*pacs : * 01.75+m , 05.50+q , 89.90+n .5 cm physics has dealt with quite a success in describing and understanding collective behavior in matter . very recently
many physicists have used basic concepts and techniques from the physics of collective disorder to study a large spectrum of problems outside the usual field of physics such as social behavior @xcite , group decision making @xcite , financial systems @xcite and multinational organizations @xcite . see
@xcite for a review of these applications .
a few years ago , galam has developed a hierarchical voting model based on the democratic use of majority rule @xcite . in the simplest case of two competing parties @xmath0 and @xmath1 with respective support of @xmath2 and @xmath3 ,
it was shown that , for the @xmath1 , winning the elections at the top of the hierarchy ( i.e. after several tournaments ) does not depend only on @xmath4 but also on the existence of some local biases .
in particular , in the case of voting cells of four persons , a bias is introduced ( usually in favor of the leading party , e.g. @xmath1 ) to solve the @xmath5-@xmath6 situations . then , the critical threshold of support for the ruling party to win can be as low as @xmath7 .
the model showed how a majority up to @xmath8 can self - eliminate while climbing up the hierarchy , using locally the democratic majority voting rule .
this self - elimination occurs within only few hierarchical levels .
following this previous study , we address here the universal and generic problem of the competing fight between two different groups over a fixed area .
we present a `` voter model '' which describes the dynamical behavior of a population with bimodal conflicting interests and study the conditions of extinction of one of the initial groups .
this model can be thought of as describing the smoker - non smoker fight : in a small group of persons , a majority of smokers will usually convince the few others to smoke and vice versa .
the point is really when an equal number of smokers and non - smokers meet . in that case
, it may be assumed that a social trend will decide between the two attitudes . in the us , smoking is viewed as a disadvantage whereas , in france , it is rather well accepted . in other words , there is a bias that will select the winner party in an even situation . in our example , whether one studies the french or us case , the bias will be in favor of the smokers or the non - smokers , respectively .
the same mechanism can be associated with the problem of competing standards ( for instance pc versus macintosh for computer systems or vhs versus beta mag for video systems ) . the choice of one or the other standard is often driven by the opinion of the majority of people one meets .
but , when the two competing systems are equally represented , the intrinsic quality of the product will be decisive .
price and technological advance then play the role of a bias .
here we consider the case of four - person confrontations in a spatially extended system in which the actors ( species @xmath0 or @xmath1 ) move randomly .
the process of spatial contamination of opinion plays a crucial role in this dynamics . in the original galam model @xcite ,
the density threshold for an invading emergence of @xmath1 is @xmath7 if the @xmath1 group has a qualitative bias over @xmath0 . with a spatial distribution of the species , even if @xmath9 , @xmath1 can still win over @xmath0 provided that it strives for confrontation .
therefore a qualitative advantage is found not to be enough to win . a geographic as well a definite degree of aggressiveness are instrumental to overcome the less fitted majority .
the model we use to describe the two populations @xmath0 and @xmath1 influencing each other or competing for some unique resources , is based on the reaction - diffusion automata proposed by chopard and droz @xcite .
however , here , we consider only one type of particle with two possible internal states ( @xmath10 ) , coding for the @xmath0 or @xmath1 species , respectively .
the individuals move on a two - dimensional square lattice . at each site , there are always four individuals ( any combination of @xmath0 s and @xmath1 s is possible ) .
these four individuals all travel in a different lattice direction ( north , east , south and west ) .
the interaction takes place in the form of `` fights '' between the four individuals meeting on the same site . at each fight ,
the group nature ( @xmath0 or @xmath1 ) is updated according to the majority rule , when possible , otherwise with a bias in favor of the best fitted group : * the local majority species ( if any ) wins : @xmath11 where @xmath12 . * when there is an equal number of @xmath0 and @xmath1 on a site , @xmath1 wins the confrontation with probability @xmath13 .
the quantity @xmath14 $ ] is the bias accounting for some advantage ( or extra fitness ) of species @xmath1 .
the above rule is applied with probability @xmath15 .
thus , with probability @xmath16 the group composition does not change because no fight occurs . between fights
both population agents perform a random walk on the lattice .
this is achieved by shuffling randomly the directions of motion of the fours individuals present at each site and letting them move to the corresponding neighboring sites @xcite .
initially , populations @xmath0 and @xmath1 are randomly distributed over the lattice , with respective concentrations @xmath2 and @xmath3 .
it is clear that the model richness comes from the even confrontations .
if only odd fights would happen , the initial majority population would always win after some short time .
the key parameters of this model are ( i ) @xmath15 , the aggressiveness ( probability of confrontation ) , ( ii ) @xmath17 , the @xmath1 s bias of winning a tie and ( iii ) @xmath4 , the initial density of @xmath1 .
the strategy according to which a minority of @xmath1 s ( with yet a technical , genetic , persuasive advantage ) can win against a large population of @xmath0 s is not obvious .
should they fight very often , try to spread or accept a peace agreement ?
we study the parameter space by running cellular automata implementing the above system . in the limit of low aggressiveness ( @xmath18 ) ,
the particles move a long time before fighting . due to the diffusive motion ,
correlations between successive fights are destroyed and @xmath1 wins provided that @xmath19 and @xmath20 this is the mean - field level of our dynamical model which corresponds to the theoretical calculations made by galam in his election model @xcite . more generally , and for @xmath21 , we observe that @xmath1 can win even when @xmath22 , provided it acts aggressively , i.e. by having a large enough @xmath15 .
thus , there is a critical density @xmath23 such that , when @xmath24 , all @xmath0 are eliminated in the final outcome . below @xmath25 , @xmath1 looses unless some specific spatial configurations of @xmath1 s are present .
this is a general and important feature of our model : the growth of species @xmath1 at the expense of @xmath0 is obtained by a spatial organization .
small clusters that may accidentally form act as nucleus from which the @xmath1 s can develop . in other words , above the mean - field threshold @xmath7 there is no need to organize in order to win but , below this value only condensed regions will be able to grow .
when @xmath15 is too small , such an organization is not possible ( it is destroyed by diffusion ) and the strength advantage of @xmath1 does not lead to success . figure [ pbvsk0 ] summarizes , as a function of @xmath4 and @xmath15 , the regions where either @xmath0 or @xmath1 succeeds .
it turns out that the separation curve satisfies the equation @xmath26 .
it is also interesting to study the time needed to annihilate completely the looser . here
, time is measured as the number of fights per site ( i.e. @xmath27 where @xmath28 is the iteration time of the automaton ) .
we observed that , in this case , the dynamics is quite fast and a few units of time are sufficient to yield a collective change of opinion .
the previous results assume a contant bias .
however , with the assumption that an individual surrounded by several of its congeners becomes more confident and thus less efficient in its fight , one may vary the bias @xmath17 as a function of the local density of @xmath1 .
for example , within a neighborhood of size @xmath29 , the bias can decrease from 1 to 0 as follows : @xmath30 if @xmath31 ( local minority of @xmath1 s ) and @xmath32 if @xmath33 ( local majority of @xmath1 s ) , where @xmath34 designates the number of @xmath1 s in the neighborhood .
this rule produces an interesting and non - intuitive new behavior . depending on the value of @xmath35 ,
there is a region near @xmath36 such that the @xmath0 species can win by preventing the @xmath1 s from spreading in the environment .
this is achieved by a very aggressive attitude of the @xmath0 s .
note that this effect is already present in the previous case ( @xmath37 and @xmath38 ) , but only on the line @xmath36 and for @xmath39 .
figure [ pbvsk3 ] summarizes the regions where either @xmath0 or @xmath1 succeeds when @xmath40 .
in addition to the separation line shown in light gray , the time needed to decimate the other opinion is indicated by the gray levels .
we observe that this time may become large in the vicinity of the critical line .
depending on the time scale associated with the process , such a slow evolution may be interpreted as a coexistence of the two species ( if a campaign lasts only a few days or a few weeks , the conflict will not be resolved within this period of time ) .
we have shown that the correlations that may exist between successive fights may strongly affect the global behavior of the system and that an organization is the key feature to obtain a definite advantage over the other population .
this observation is important .
for instance , during a campaign against smoking or an attempt to impose a new system , it is much more efficient ( and cheaper ) to target the effort on small nuclei of persons rather than sending the information in an uncorrelated manner .
also , according to figure [ pbvsk3 ] , an hypothetical minority of smokers in france must harass non - smokers during social meetings ( coffee break , lunch , ... ) rather often but not systematically , in order to reinforce their position . on the contrary , for an hypothetical majority of smokers in the us , either a smooth or a stiff harassment against the non - smokers
is required to survive .
aggressiveness is the key to preserve the spatial organization .
refusing a fight is an effective way for the @xmath0 species to use its numerical superiority by allowing the @xmath1 individuals to spread . with this respect
, a minority should not accept a peace agreement ( which would results in a lower @xmath15 ) with the leading majority unless the strength equilibrium is modified ( i.e. @xmath1 is better represented ) .
motion is also a crucial ingredient in the spreading process .
there is a subtle tradeoff between moving and fighting .
when little motion is allowed between fights ( @xmath41 ) , the advantage is in favor of @xmath0 again . in an epidemic system , our model shows that two solutions are possible to avoid infestation : either one let the virus die of isolation ( dilute state due to a small @xmath15 ) or one decimates it before it spreads ( large @xmath15 ) .
finally a simple variant of the above model provides a possible scenario to explain punctuated equilibria @xcite in the evolution of living organisms .
it is well known that the transition between two forms of life may be quite abrupt .
there is no trace of the intermediate evolutionary steps . to give some insights into this problem we modify our voter model by including a creation rate for the @xmath1 individuals ( @xmath42 , with probability @xmath43 ) . in this context , the @xmath1 species is fitter than the @xmath0 species ( the bias @xmath44 ) but the numerical advantage of @xmath0 is too strong for @xmath1 to survive .
however , if the simulation is run for a long enough time , nucleation in this metastable state will happen , which will produce locally a very favorable spatial arrangement of @xmath1 s .
these @xmath1 s will then develop and , very rapidly , eliminate all @xmath0 s .
in other words , a very numerous species may live for a considerable amount of time without endangering competitors and suddenly , be decimated by a latent , fitter species .
this scenario needs a strong statistical fluctuation but no additional external , global event . in conclusion ,
although the model we propose is very simple , it abstracts the complicated behavior of real life agents by capturing some essential ingredients . for this reason
, the results we have presented may shed light on the generic mechanisms observed in a social system of opinion making . |
ultracold atomic quantum gases with spin degrees of freedom provide exceptionally clean and idealized testing beds for studying quantum magnetism @xcite .
optical trapping from ac stark shifts of off - resonant laser fields are capable of equal confinement for all atomic pseudo - spin components which facilitates research into exciting spinor physics with atomic quantum gases . as is often employed in studying a trapped bose gas , when treating the condensate , we first take a mean - field ( mf ) approximation assuming only one eigenvalue of the single - particle density matrix is macroscopic , being of order @xmath0 .
second quantization is then limited to the condensate mode .
such a simple scenario already allows for many interesting quantum many body phenomena@xcite .
two popular atomic species often employed in experimental research on spinor bose - einstein condensates ( becs ) @xcite , are @xmath1rb and @xmath2na atoms . within each species ,
their interactions are dominated by the density - dependent interaction in comparison to the much weaker spin - dependent interactions . as a result , single - spatial mode approximation ( sma ) , whereby the spatial dependence of the condensate wave function is determined independent of the spin degrees of freedom , was introduced @xcite and remains reasonable as long as the number of atoms is not too large @xcite . within the mf approximation
, the ground state of a spinor bose - einstein condensate ( bec ) is found to be ferromagnetic , polar , or cyclic phases , _ etc _ , determined by the spin - dependent interactions and the total ( hyperfine-)spin @xmath3 of the atom .
further theoretical work armed with full quantum calculations reveal interesting many - body states @xcite , beyond the scope of those from mf approximations . for spin-1 condensates , the exact eigenstates will contain paired spin singlets @xcite , which become more complex for higher spin condensates .
for example , the spin-2 case involves spin singlets which can be formed by either two or three atoms @xcite .
a general procedure exists for more detailed information of the building blocks of eigenstates determined by their associated generating functions @xcite .
more generally , we can always resort to the means of numerics to diagonalize the ground - state single - particle density matrix , which then reveals fragmented ground states if more than one eigenvalues are being of order @xmath0@xcite .
several groups have recently studied spinor condensate mixtures @xcite , which also display non - mf features , such as anomalous quantum fluctuations for each spin components and quantum entangled ground states .
this is first discovered in spinor condensates with more than one orbitals , for instance , the case of pseudo spin-1/2 condensates @xcite for which kuklov and svistunov
@xcite predicted that in the ground states all atoms will have to condense into two orthogonal spatial orbitals or more due to the conservation of the total spin .
this could result in a condensate ground state being a maximally entangled many - body state .
_ replaced the two orbitals with two different atomic species , a ground state with entangled order parameter follows @xcite . under mf approximation ,
we have previously elaborated the ground - state phase diagram for a condensate mixture of two spin-1 becs @xcite .
the interesting phases are named appropriately as ff , aa , pp , cc , and mm phases , distinguishing different structures and interaction parameter spaces .
furthermore , we provide many beyond mf results based on a full quantum spin - dependent hamiltonian @xcite , which contains non - commuting terms , forbidding a simple derivation of the exact eigenstates .
for two special cases , commutations are restored among the generally non - commuting terms .
first , when the interspecies singlet pairing interaction is ignored ( @xmath4 ) , the hamiltonian is then simply composed of three operators which obey the angular momentum algebra . making use of the eigenstates of single spin-1 condensates
, we directly construct the eigenstates of a binary spin-1 mixture using the angular momentum coupling representation .
second , when the interspecies anti - ferromagnetic spin - exchange interaction is strong enough , the ground state will be forced to develop entanglement between the two species @xcite , a result consistent with what is discovered in a spin-1 condensate placed inside a double well @xcite .
other interesting features are discussed for @xmath4 revealing fragmentation and quantum entanglement @xcite .
in this revisit we hope to understand quantum entanglement between two spin-1 condensates when both interspecies spin - exchange and singlet - pairing interactions are present .
our study is based on the same model system of a binary mixture of spin-1 condensates confined in optical traps .
the corresponding field operators that annihilates a boson of species 1 and species 2 at position @xmath5 are described respectively by @xmath6 and @xmath7 , where @xmath8 denoting the three zeeman hyperfine states .
the sma is adopted for each of the two species , employing two spatial mode functions @xmath9 and @xmath10 respectively , and the field operators are expanded as @xmath11 and @xmath12 , with @xmath13 and @xmath14 respectively the annihilation operators for an atom in the spin component @xmath15 . in the absence of external magnetic field , the spin - dependent hamiltonian for a binary mixture of spin-1 condensates then becomes the following @xmath16 under the sma @xcite .
the interaction coefficients are @xmath17 , @xmath18 , and @xmath19 .
@xmath20 ( @xmath21 ) is intra - species spin - exchange interaction parameter of species 1 ( 2 ) . @xmath22 and @xmath23 denote inter - species spin - exchange and singlet - pairing interaction parameters , respectively .
the singlet pairing operator becomes @xmath24 , and two angular momentum like operators @xmath25 and @xmath26 obey the usual angular momentum algebra @xcite .
they commute with atom number operators @xmath27 and @xmath28 . in the above
, @xmath29 denotes the @xmath30 component of the spin-1 matrix @xmath31 .
as presented in the earlier study of @xcite , for the spin - dependent hamiltonian of eq .
( [ hamiltonian ] ) , the first three terms commute with each other , but they do not commute with the fourth term .
this shows the ground state determined will depend on the interaction parameters . as a result
, we resorted to the special cases of no interspecies singlet - pairing interaction ( @xmath4 ) and @xmath32 @xcite . for the first case of @xmath4 ,
it has already attracted much attention due to the appearance of fragmented ground states and the associated entanglement between two species and exotic atomic number fluctuations @xcite . in this study
, we will discuss the general case of the full spin - dependent hamiltonian of eq .
( [ hamiltonian ] ) .
whenever the ground state depends on the interaction parameters , we have to perform a full quantum calculation numerically , usually this amounts to a full exact numerical diagonalization for each atom numbers @xmath33 and @xmath34 . before discussing the numerical results , we want to point out that there still exist two conserved quantities : the total spin angular momentum and its @xmath35-component , as @xmath36 commutes with the spin - dependent hamiltonian . as a result
, we can elaborate spin structures from building blocks derived by generating function of the maximum spin states @xmath37 , where we have used quantum numbers @xmath38 and @xmath39 to denote the common eigenstates of the angular momentum operators @xmath40 and @xmath41 .
we recall the suitable generating function @xmath42 for a binary mixture of two spin-1 condensates derived earlier in ref . @xcite . from this generating function ,
we have figured out all six building blocks for constructing the eigenstate @xmath43 , which is given by @xmath44 where the six building blocks are @xmath45 and @xmath46 , @xmath47 , and @xmath48 satisfy the constrains @xmath49 and additionally @xmath50 .
spin states @xmath51 of other magnetization can be constructed by simply applying @xmath52 on the state @xmath43 as @xmath53 ( un - normalized ) . in most numerical studies , we assume each species contain 100 atoms @xmath54 . due to the so(3 ) symmetry of our model , we restrict the hilbert space into the subspace of zero magnetization with @xmath55 @xcite . in fig .
[ fig1 ] , we summarize the extensive numerical results . since the total spin angular momentum is conserved , we can use the eigenvalue of the operator @xmath40 to distinguish different phase , which is then accompanied by the information of the building blocks .
a total of three constrains exist for the allowed values of @xmath46 , @xmath47 , and @xmath48 @xmath56 .
only three are needed for a solution , we choose the three as @xmath57 , @xmath58 and @xmath59 , which are determined numerically from evaluating the associated expectation values of @xmath60 , @xmath61 , and @xmath62 , respectively . from extensive numerical results we construct the ground - state phase diagram as shown in fig .
[ fig1 ] . perhaps not surprisingly , it is almost the same as the mf approximation studied in ref .
each phase is then labeled the same as before @xcite , albeit that the meanings can be different due to the non - commuting operators in the spin - dependent hamiltonian eq .
( [ hamiltonian ] ) . in tab .
[ tabl ] , we summarize the properties for the four special phases : ff , aa , pp , and cc . the remaining mm phase still denotes the phase whose parameters evolve continuously across the phase boundaries .
.the expectation values for the special operators in the ground state within different phases . [ cols="^,^,^,^,^",options="header " , ]
and @xmath63 . blue solid lines denote continuous phase transition boundaries .
red solid lines denote discontinuous phase transition boundaries between two phases with fully determined total spin angular momentum @xmath38 .
the green solid line denotes the discontinuous phase transition boundary between the phase cc3 and the mm phase .
the black dash - dotted lines correspond to @xmath64 , which serve as guides for the eye .
the von neumann entropy of the ground states are painted by gray scale density plots , where black ( white ) color refer to low ( high ) entanglement respectively .
the three subplots denote fixed intra - specie spin exchange interaction parameters of @xmath65 @xmath66 : ( a ) ( @xmath67,@xmath68 ) ; ( b ) ( @xmath69 ) ; and ( c ) ( @xmath70 ) .
the red dashed lines are four lines connecting the points @xmath71 , @xmath72 , @xmath73 and @xmath74 in the parameter space of @xmath75.,width=307 ] to quantify entanglement between the two species , we numerically computed the von neumann entropy @xmath76 , where @xmath77 is the reduced density matrix resulting from partial tracing of the ground - state density matrix @xmath78 over the basis of species 2 . the amount of entanglement is then shown as density plots over the phase diagram of fig .
[ fig1 ] , with a legend shown on the right , the black ( white ) color refer to low ( high ) entanglement . in the absence of interspecies singlet - pairing interaction @xmath79 , the spin - dependent hamiltonian of eq .
( [ hamiltonian ] ) contains only three operators commuting with each other . as a result we can use quantum numbers @xmath80 , @xmath81 , @xmath38 , and @xmath39 to quantify its eigenstate @xmath82 , with @xmath83 , @xmath84 , @xmath85 and @xmath86 . in the ground states
, we have @xmath87 for ferromagnetic interspecies spin - exchange interaction @xmath88 ; and @xmath89 for anti - ferromagnetic interspecies spin - exchange interaction @xmath90 , while the value of @xmath80 and @xmath81 are determined by the three interaction parameters . in fig .
[ fig2 ] , we display the expectation values of the intra- and interspecies single - pairing number operators in the ground state , divided by their corresponding maximum values shown in tab .
in addition , we show the total spin angular momentum and the von neumann entropy .
first , when @xmath91 , irrespective of the interspecies spin - exchange interaction , atoms in the same species will not pair into a singlet , but atoms in different species will pair into singlets with @xmath92 close to reaching its maximum values @xmath93 when the interspecies spin - exchange interaction is anti - ferromagnetic .
the corresponding total spin angular momentum @xmath40 is equal to its maximum value @xmath94 with a relatively low von neumann entropy when the interspecies spin - exchange interaction is ferromagnetic . the total spin angular momentum @xmath40 is equal to @xmath95 with the von neumann entropy @xmath96 for anti - ferromagnetic interspecies spin - exchange interaction .
second , when @xmath97 , in the two limits of large ferromagnetic or anti - ferromagnetic interspecies spin - exchange interaction , the ground state is the same as above in the previous case . in the other limit with low interspecies spin - exchange interaction , atoms in the same species tend to pair into singlets giving rise to no entanglement between the two species .
this implies the ground state can be written as a product state : @xmath98 . for the remaining phase , we call it the mm phase , which can show higher ( lower ) entanglement compared to the ff phase between the two species when @xmath99 ( @xmath100 ) . for the final case
when @xmath101 , the ground state show similar properties as that in the second case
. , @xmath102 , @xmath103 , denoted by blue solid , red dash - dot , and black dash - dot - dot lines , respectively .
right : the normalized total spin angular momentum @xmath104 and von neumann entropy of the ground state , denoted by blue solid and red dash - dot lines , respectively .
the three subplots denote zero interspecies singlet - pairing interaction @xmath79 and fixed intra - species spin - exchange interaction parameters @xmath105 @xmath66 : ( a ) ( @xmath67,@xmath68 ) ; ( b ) ( @xmath69 ) ; and ( c ) ( @xmath70 ) .
, width=326 ] the most attractive phase when @xmath4 is the entangled ground state denoted by @xmath106 @xcite , @xmath107 which show high entanglement with @xmath96 between the two species . as demonstrated in fig .
[ fig1 ] by numerical calculations , however , there remain other phases which show larger entanglement between the two species .
this shows that the state @xmath106 is not a maximal entangled state , in contrast to previously studied case of two pseudo spin-1/2 condensates @xcite .
this is not a surprise @xcite . due to the redundant degrees of freedom in the spin-1 case ,
the total spin angular momentum of each species can take other values besides the largest value of @xmath0 . to demonstrate the entanglement between the two species , we show their corresponding expectation values and von neumann entropy along four lines connecting the points @xmath108 , @xmath72 , @xmath109 and @xmath110 in the parameter space of @xmath75 .
the four lines are marked as red dashed lines in the fig .
[ fig1](a ) . in fig .
[ fig3 ] , we illustrate the ground - state properties of two ferromagnetic condensates with intraspecies spin - exchange interactions at @xmath111 . first of all , we consider the aa phase .
when @xmath4 , the spin - dependent hamiltonian contains three operators commuting with each other , and the ground state can be expressed as @xmath106 for large enough anti - ferromagnetic interspecies spin - exchange interaction , and show high entanglement between the two species . when @xmath112 , although the fourth term of the spin - dependent hamiltonian does not commute with the other three , we find that the ground state not only show similar expectation values of the operators , it also contain similar entanglement between the two species , over a large area in the phase diagram demonstrated in fig . [ fig1](a ) .
for @xmath113 , irrespective of its value , the ground state falls into the aa phase . for @xmath114 , the ground state is still classified as the aa phase , as long as @xmath115 does not exceed a critical value , which increases in proportion to the interspecies spin - exchange interaction parameter @xmath116 . in the first column of fig .
[ fig3 ] , we evaluate the properties of the aa phase , where the ground - state expectation value of intra- and interspecies singlet - pairing number operators are close to @xmath95 , @xmath95 , and @xmath93 , respectively .
the total spin angular momentum is exactly equal to 0 , and the von neumann entropy is close to 1 . when interspecies singlet - pairing interaction exceeds a critical value , the ground state changes into the mm phase .
as long as @xmath117 , it tries to decrease the interspecies singlet - pairing interaction . in the first column of fig .
[ fig3 ] , we follow the line of @xmath118 and illustrate the phase transition from the aa phase into the mm phase . in the mm phase , as long as we increase @xmath115 accordingly , atoms in different species will continuously avoid to pair into singlets , while atoms in species 1 will try to pair into singlets .
atoms in species 2 first try to pair into singlets and then avoid to pair . meanwhile , the total spin angular momentum first increases and then decreases . for a relatively large area of the mm phase , we find the two species show high entanglement compared to the aa phase . as interspecies singlet - pairing interaction
is increased , the ground state will fall into the cc1 phase , where its total spin angular momentum @xmath40 will be equal to @xmath119 , and atoms in the species 1 ( 2 ) will pair ( not pair ) into singlets . at the same time the expectation value of @xmath120 will be near to its maximum @xmath93 .
the ground - state von neumann entropy in the cc1 phase remains at a low value .
going along the line @xmath121 , with decreasing interspecies spin - exchange interaction , the ground state changes from the cc1 phase , to the mm phase , and finally to the ff phase .
atoms of species 1 will become unpaired continuously , while the total spin angular momentum increases from @xmath119 in the cc1 phase to its maximum @xmath94 in the ff phase .
the ground - state entanglement between the two species in the cc1 and the ff phase are almost at the same level . while in the mm phase
, the entropy first increases to near 1 and then decreases .
we then follow the line @xmath122 . with decreasing interspecies singlet - pairing interaction ,
the ground state covers the ff , mm , and pp phases successively . in the pp phase
, atoms in the same / different species will all try to pair into singlets .
the ground - state expectations values for intra- and interspecies singlet - pairing number operators will reach close to their corresponding maximum , meanwhile the two species show higher entanglement . in the mm phase ,
the expectations values for operators or the von neumann entropy change continuously to connect the ff phase and the pp phase .
lastly , we consider the line @xmath123 . as long as the interspecies spin - exchange interaction increases ,
the ground state will be changed continuously from the pp to the mm , and to the aa phase . from the fourth column of fig .
[ fig3 ] , we find that in the whole line of @xmath123 , the two species show a relatively high entanglement with the von neumann entropy @xmath124 remaining higher than 1 .
especially so in the mm phase , where the highest entropy reaches 1.4568 , which is close to the maximum entropy of @xmath125 . in figs .
[ fig4 ] and [ fig5 ] , we illustrate the two other cases with intraspecies spin - exchange interactions @xmath126 fixed respectively at @xmath127 and @xmath128 .
we find that the ground - state show similar properties to that in the two ferromagnetic condensates shown in fig .
the only difference is for the cc2 or cc3 phase . in the cc2 phase , the ground state expectation values for both intra- and interspecies singlet - pairing number operators are close to their corresponding maximum @xmath119 , @xmath119 , and @xmath93 . while in the cc3 phase
, they are close to @xmath95 , @xmath119 , and @xmath93 , respectively .
the total spin angular momentum of the ground state is equal to @xmath95 in the cc2 phase , and @xmath119 in the cc3 phase .
meanwhile , in the cc2 phase the von neumann entropy is less than 1 but larger than that in the ff phase , while in the cc3 phase , it is close to the value in the ff phase . before conclusion , we hope to stress that the maximal entangled state in this system is given by @xmath129 it is the eigenstate or the ground state ( if @xmath113 ) of the @xmath23-term in the hamiltonian of the eq .
( [ hamiltonian ] ) , which means the maximal entangled state @xmath130 is the eigenstate of two spin-1 condensates with only interspecies spin - singlet pairing interaction ( @xmath131 and @xmath112 ) , with the corresponding eigenvalue @xmath132 @xcite .
in conclusion , we have studied the ground - state phase diagram for a binary mixture of two spin-1 condensates more carefully , going beyond the mf approximation .
when there exists no interspecies singlet - pairing interaction , the spin - dependent hamiltonian contains three operators commuting with each other . in this special case ,
the most interesting phase is the aa phase , where two species show high entanglement .
when interspecies singlet - pairing interaction is turned on , the added operators do not commute with the previous three ones , which forbids us from obtaining exact eigenstates analytically for the model spin system . in this study
, we perform full quantum diagonalization to find the ground states numerically . to quantify the ground states , we work out the building blocks to construct the maximum spin states , which can be used rightfully to discuss entanglement scales between the two species .
we have evaluated the associated ground - state von neumann entropy .
after detail calculations , we find that the aa phase can persist for large areas of the parameter space for interspecies spin - exchange and singlet - pairing interactions .
in addition , there is another interesting phase : the pp phase , which show similar level of entanglement between the two species .
what s more , we find that aa phase is not the maximum entangled state .
the ground state with highest entanglement we found lies in the mm phase .
this work is supported by nsf of china under grants no . 11004116 , no .
10974112 and no . 91121005 , nkbrsf of china , and the research program 2010thzo of tsinghua university .
s. trotzky , p. cheinet , s. flling , m. feld , u. schnorrberger , a. m. rey , a. polkovnikov , e. a. demler , m. d. lukin and i. bloch , science * 319 * , 295 ( 2008 ) ; maciej lewenstein and anna sanpera , science * 319 * , 292 ( 2008 ) ; we have previously miscalculated the operator @xmath134 , which is equal to @xmath135 not @xmath136 in the ref .
@xcite . as a result , @xmath137 @xmath138 , @xmath139 and @xmath140 .
also the eq .
( 13 ) in the ref .
@xcite need to be changed to @xmath141 . | we revisit in detail the non - mean - field ground - state phase diagram for a binary mixture of spin-1 bose - einstein condensates including quantum fluctuations .
the non - commuting terms in the spin - dependent hamiltonian under single spatial mode approximation make it difficult to obtain exact eigenstates . utilizing the spin z - component conservation and the total spin angular momentum conservation , we numerically derive the information of the building blocks and evaluate von neumann entropy to quantify the ground states .
the mean - field phase boundaries are found to remain largely intact , yet the ground states show fragmented and entangled behaviors within large parameter spaces of interspecies spin - exchange and singlet - pairing interactions . |
in the past few years , cyclic spin exchanges in quantum antiferromagnetic magnets have attracted considerable attention . from various experiments
, it has been realized that four - spin cyclic exchanges play an important role in understanding the spin dynamics of insulating cuprate materials , such as the two - leg spin ladders cu@xmath0o@xmath1 in the compounds srcu@xmath2o@xmath3 , @xcite ( sr , ca , la)@xmath4cu@xmath5o@xmath6 , @xcite and the cuo@xmath2 plane in la@xmath2cuo@xmath7 .
@xcite similar mechanism of multiple spin exchanges in @xmath8he ( ref . ) and wigner crystal , @xcite which was found to be large , has also been proposed .
spin ladders can be viewed as intermediates between one - dimensional spin chains and two - dimensional spin lattices .
they are particular of interest in the understanding of the high - temperature superconductivity in two dimensions and the haldane s conjecture @xcite in spin chains .
the two - leg spin ladder is the simplest system on which cyclic exchanges can be materialized . in the absence of cyclic exchanges , with nonzero interchain couplings ,
the @xmath9 spin ladder lies in a haldane phase , which has a short - range resonating - valence - bond ( rvb ) ground state with a spin gap , and superconductivity emerges upon doping.@xcite contrary to the case in a @xmath9 chain , the spinons in the ladder are confined back to the well - defined quasiparticles ( massive magnons ) in the neighborhood of @xmath10 , where a spectral gap exists.@xcite it has been shown that four - spin cyclic exchanges can make the spin gap decreases rapidly.@xcite furthermore , tuning of the four - spin interactions can drive the system into different phases .
@xcite at the nearest critical point , the system becomes gapless to cross from the rvb rung - singlet phase to a staggered dimer phase . by a field theory using the majorana fermion representation , nersesyan and tsvelik referred to the critical phenomenon as the level @xmath11 su(2 ) wess - zumino - novikov - witten model or su(2 ) @xmath12 conformal field theory .
in other words , the phase transition is of the second - order type and belongs to the universality class of the takhtajan - babujian critical point.@xcite this statement has been confirmed by numerical simulations .
@xcite the hamiltonian describing an isotropic @xmath9 spin ladder with additional cyclic ( four - spin ) interaction is represented by @xmath13\right.\nonumber \\ & + 2j_{\text{cyc}}\left[\left(\mathbf{s}_{1,i}\cdot\mathbf{s}_{1,i+1}\right)\left(\mathbf{s}_{2,i}\cdot\mathbf{s}_{2,i+1}\right)\right.\nonumber \\ & + \left(\mathbf{s}_{1,i}\cdot\mathbf{s}_{2,i}\right)\left(\mathbf{s}_{1,i+1}\cdot\mathbf{s}_{2,i+1}\right)\nonumber \\ & \left.\left.-\left(\mathbf{s}_{1,i}\cdot\mathbf{s}_{2,i+1}\right)\left(\mathbf{s}_{2,i}\cdot\mathbf{s}_{1,i+1}\right)\right]\right\ } , \label{eq:1}\end{aligned}\ ] ] where the indices 1 and 2 distinguish upper and lower legs , and @xmath14 labels rungs ; @xmath15 and @xmath16 are the interchain and intrachain couplings , respectively ; @xmath17 denotes the strength of the cyclic ( four - spin ) exchange . the model is schematically shown in fig .
[ ladder ] . in the rest of the paper , for convenience , we call eq .
( [ eq:1 ] ) as type - i hamiltonian .
schematic structure of a two - leg ladder with cyclic ( four - spin ) interaction .
the round arrows stand for the action of the four - spin terms as given in eq .
( [ eq:1 ] ) , which is not completely identical to the usual cyclic permutations.,width=283 ] the above hamiltonian distinguishes from another one where the ring exchange term is directly represented by a cyclic permutation operator @xmath18 in the plaquette : @xmath19 we call this form as type - ii hamiltonian .
the cyclic permutation operator can be expressed by spin-@xmath20 operators including bilinear frustrating and biquadratic terms .
@xcite so type - ii hamiltonian can be re - written as @xmath21\right\ } \nonumber \\ % & + k\sum_{\langle ijkl\rangle}\left(p_{ijkl}+p_{ijkl}^{-1}-\frac{1}{4}\right)\nonumber\\ h & = \sum_{i}\left\ { \left(j_{\text{rung}}+2k\right)\mathbf{s}_{1,i}\cdot\mathbf{s}_{2,i}\right.\nonumber \\ & + \left(j_{\text{leg}}+k\right)\left(\mathbf{s}_{1,i}\cdot\mathbf{s}_{1,i+1}+\mathbf{s}_{2,i}\cdot\mathbf{s}_{2,i+1}\right)\nonumber \\ & + k\left(\mathbf{s}_{1,i}\cdot\mathbf{s}_{2,i+1}+\mathbf{s}_{2,i}\cdot\mathbf{s}_{1,i+1}\right)\nonumber \\ & + 4k\left[\left(\mathbf{s}_{1,i}\cdot\mathbf{s}_{1,i+1}\right)\left(\mathbf{s}_{2,i}\cdot\mathbf{s}_{2,i+1}\right)\right.\nonumber \\ & + \left(\mathbf{s}_{1,i}\cdot\mathbf{s}_{2,i}\right)\left(\mathbf{s}_{1,i+1}\cdot\mathbf{s}_{2,i+1}\right)\nonumber \\ & \left.\left.-\left(\mathbf{s}_{1,i}\cdot\mathbf{s}_{2,i+1}\right)\left(\mathbf{s}_{2,i}\cdot\mathbf{s}_{1,i+1}\right)\right]\right\}. \label{eq:2}\end{aligned}\ ] ] there have been extensive theoretical studies on the type - ii hamiltonian . by the density matrix renormalization group ( dmrg ) @xcite and exact diagonalization method ,
the energy spectra about one - magnon branch and two - magnon continuum in the rung - singlet phase have been obtained .
@xcite the phase diagram for @xmath22 was studied in detailed @xcite and a global picture for general cases has been proposed .
@xcite the phase transitions , mainly concerning the case from the rung - singlet to the dimerized phase , together with the critical exponent were studied by field theory , perturbative and series expansion method .
@xcite on the other hand , the type - i hamiltonian catches less attention .
and it is widely believed that the physics should be the same for the type - i and type - ii hamiltonians , although there is a lack of detailed comparisions .
theoretically , the ring exchange interaction naturally arises as an effective low - energy model for the multiband hubbard model near the insulating filling and in the strong coupling limit .
it has been shown that the four - spin terms are most significant ones.@xcite in order to highlight the effect of this four - spin term , in this paper , we employ the type - i hamiltonian for our numerical simulations .
as mentioned above , the zero - temperature behaviors of the spin ladders with cyclic exchange interactions have been analyzed in detail , especially for the type - ii hamiltonian .
a study on the thermodynamic properties of the systems has not been carried out extensively , except for the study by a high - temperature series expansion method .
@xcite these quantities can be directly compared with the experimental observations and therefore are of particular use .
detailed and reliable numerical results on these quantities may offer an expedient way to determine the model parameters for the compounds we are interested in . in this paper , we show that by using the transfer - matrix renormalization group ( tmrg ) method , we can not only obtain the temperature dependence of the specific heat and susceptibility of the spin ladders down to the low - temperature regime and with high accuracy , but also extract extra information from the data for the magnitude of the gap and the one - triplet magnon excitations , which can be compared with the experiments directly .
as proposed by experiments , for the insulator cuprate ladder materials , @xcite both @xmath15 and @xmath16 are positive and close in magnitude .
the values of @xmath23 range in a region @xmath24 , which roughly corresponds to the parameter region @xmath25 for the type - i hamiltonian and the system is still in the rung - singlet phase . for this reason , in our numerical simulations
, we confine the parameter @xmath26 $ ] , and @xmath27 is taken three different values : 0.5 , 1.0 and 1.2 , where @xmath28
we take @xmath29 in order to make a comparison with the typical rung - singlet phase .
the numerical algorithm we employ to study the thermodynamical properties of the present model is the tmrg method , @xcite which is a powerful numerical tool for studying the thermodynamic properties of one - dimensional quantum systems .
one is able to evaluate nearly all thermodynamic quantities by the maximum eigenvalue and the corresponding left and right eigenvectors of the transfer matrix .
we will show later that the magnitude of the gap and the low - lying excitations can also be extracted from the obtained data . in our numerical calculations , @xmath30 ,
the error caused by the trotter - suzuki decomposition is less than @xmath31 . during the tmrg iterations ,
@xmath32 states are retained and the truncation error is less than @xmath33 down to @xmath34 .
( color online ) the specific heat coefficient @xmath35 for type - i hamiltonian for three values of @xmath27 with @xmath36 0 , 0.02 , 0.04 , 0.06 , 0.08 , and 0.1 , in ascending order along the direction of arrows.,width=283 ] figure [ c / t ] shows the tmrg results on the temperature dependence of the specific heat coefficient @xmath35 for @xmath37 0.5 , 1.0 , 1.2 with @xmath36 0 , 0.02 , 0.04 , 0.06 , 0.08 and 0.1 . besides the exponential decrease as @xmath38 , which indicates the existence of a gap , the general behavior of the curves for a given @xmath27 is as follows : as @xmath39 increased , the slope of @xmath40 shifts to lower temperatures
the reason is obvious : that with the increase of the four - spin cyclic exchange , the spin gap and the whole dispersion decrease .
@xcite another interesting feature is that at @xmath37 1.0 and 1.2 , a shoulder gets more and more distinct at low temperatures with the increase of @xmath39 .
we attribute this shoulder to the tendency of deconfinement of bound spinons from the magnons , as the crossover occurs from the strong coupling regime @xmath41 to the weak coupling one @xmath42 with the extreme limiting case of a finite @xmath35 value at @xmath43 .
temperature dependence of the susceptibility @xmath44 for type - i hamiltonian for three values of @xmath27 with @xmath36 0 , 0.02 , 0.04 , 0.06 , 0.08 , and 0.1 in ascending order along the direction of arrows.,width=283 ] the corresponding results for spin susceptibility @xmath45 are presented in fig .
[ chi_1 ] .
the exponential decrease of @xmath45 as @xmath46 , and the shift of the peak to lower temperatures with the increase of @xmath39 are consistent with the above behaviors of @xmath35 .
our numerical result on the susceptibility well agrees with that of ref .
obtained by high - temperature series expansion and exact diagonalization .
@xmath47 dependence of @xmath48 and @xmath49 ( in the unite of @xmath15 ) for @xmath29 with @xmath36 0 , 0.02 , 0.04 , 0.06 , 0.08 , and 0.1 in ascending order along the direction of arrows.,width=283 ] the @xmath39 dependence of the gap is a central quantity , since basically it determines the relevancy of the cyclic exchange term . in the vicinity of the critical point @xmath50 , generally , @xmath51 for the type - ii hamiltonian in both strong and weak coupling limits , by the perturbative cluster expansion and pad approximation in the strong coupling regime , together with the bosonization technique in the weak coupling limit , mller _ _ et al.__@xcite showed that the critical exponent @xmath52 for the gap is approximately equal to 1 ; i.e. , the gap is linearly dependent on @xmath39 near the critical point . for the @xmath53 case , this result
was also obtained by hijii _ _ et al.__@xcite here , we will see that for type - i hamiltonian , in the region considerably far away from the critical point , @xmath54 , the linear behavior of the gap vs. @xmath39 still holds in almost all the parameter range . for the spin-@xmath20 ladder in the rung - singlet phase , similar to the case in a @xmath55 chain
, we can assume that the low - lying excitation spectrum @xmath56 for @xmath57 near @xmath58 is approximately given by @xmath59 where @xmath60 is the spin wave velocity and @xmath61 is the spectrum gap .
the low temperature behaviors of the specific heat and susceptibility per spin can be expressed , respectively , as @xcite @xmath62 for @xmath63 . in this situation
, @xmath48 and @xmath49 should be linear to @xmath47 with a slope @xmath64 .
figure [ 0.5 g ] shows the @xmath47 dependence of @xmath48 and @xmath49 ( in units of @xmath15 ) for @xmath29 with @xmath36 0 , 0.02 , 0.04 , 0.06 , 0.08 , and 0.1 .
we observe a well - behaved linearity at @xmath65 . for @xmath37 1.0 and 1.2
, we have the similar results .
the values of the gap at various @xmath27 and @xmath39 can thus be extracted by fitting the data in the interval @xmath66 .
the results for @xmath29 and @xmath67 are presented in fig .
the gaps obtained from the specific heat and susceptibility are well coincident with each other . to both verify the low - temperature scaling behavior of the system and check the validity of eq .
( [ eq:5 ] ) , in fig .
[ chi_2 ] , the rescaled susceptibility @xmath68 versus @xmath69 for @xmath29 , @xmath70 at various @xmath39 is presented . for a fixed value of @xmath27
, the curves converge to a single one at low temperatures , demonstrating the dominance of @xmath61 in determining the susceptibility at low temperatures .
furthermore , it also suggests that the values of the gap we obtained by this approach are quantitatively reliable .
this situation is quite similar to the case of the @xmath55 antiferromagnetic heisenberg chain with single - ion anisotropy .
@xcite here @xmath71 serves as a common factor which incorporates the unknown effect on the prefactor in eq .
( [ eq:5 ] ) . from fig .
[ gap ] , we observe that the gap is approximately linearly dependent on @xmath39 for fixed @xmath27 in the parameter space we studied .
we also notice the different descending rates of @xmath61 with @xmath39 for various @xmath27 . at @xmath72 , @xmath73 with the difference @xmath74 .
this indicates that the gap decreases as the intrachain coupling @xmath16 increases , which is consistent with the perturbative picture from the dimer limit of uncoupled rungs ( strong coupling regime ) .
on the other hand , at @xmath75 , the difference of the gaps becomes very small .
we expect for higher value of @xmath39 a reverse of the gaps with @xmath76 should occur . for @xmath67 and @xmath77 ( not shown in fig . [ gap ]
) , the values of the gap obtained by the present method well coincide with the numerical results from the density - matrix renormalization group ( dmrg ) study for the type - ii hamiltonian for the corresponding values of the cyclic exchange coupling.@xcite for instance , when @xmath67 and @xmath72 , the gap obtained here by tmrg is @xmath78 , as compared with the dmrg result of @xmath79.@xcite the @xmath39 dependence of the spin gap for @xmath37 0.5 and 1.0 obtained by fitting the eqs .
( [ eq:4 ] ) ( signed as @xmath80 ) and ( [ eq:5 ] ) ( given as @xmath44 ) , respectively.,width=283 ] @xmath68 versus @xmath69 for @xmath29 , @xmath70 with @xmath36 0 , 0.02 , 0.04 , 0.06 , 0.08 , and 0.1 in ascending order along the direction of arrows.,width=283 ] to come to a close , we estimates the errors in the process of gap fitting . on the one hand , as mentioned in the previous section , combining the errors from the trotter - suzuki decomposition with the tmrg truncation , we estimate that the error caused by the tmrg algorithm in the fitting interval ( in our work @xmath81 ) is less than @xmath82 at most . on the other hand , in obtaining the fitting formulas ( [ eq:4 ] ) and ( [ eq:5 ] ) , the low - temperature limit and quadratic approximation ( see eq .
( [ eq : quadraticdispersion ] ) ) have been adopted .
for the specific heat , the expression including the higher order contributions of @xmath69 is given explicitly as @xcite @xmath83 e^{-\delta / t}.\end{aligned}\ ] ] the uncertainty in eq .
( [ eq:4 ] ) is roughly estimated as @xmath84 , \label{eq : delta}\ ] ] where @xmath85 represents some value of temperature @xmath86 lying in the fitting interval . from this , in our case the uncertainty of gap due to the fitting formulas can be estimated approximately as @xmath87 . based on these considerations ,
we conclude that the error in the gap fitting is about @xmath82 , which is not exceeding the symbol size in fig .
in tmrg calculations , the partition function @xmath88 is the original and the most accurate numerical quantity . in previous section
we obtained the specific heat and spin susceptibility , which can also be got from the derivatives of the partition function directly . in the following
we will discuss the single - particle state in this system and show that its dispersion , as well as the ground - state energy and the magnitude of the gap , can be obtained from the analyses of the partition function . in the rung singlet phase
, the elementary excitations are gapped .
the energy gap and dispersion for the single particle state can be analyzed perturbatively .
@xcite when @xmath89 , the system is consisted of an array of rungs with a vacuum state being the product of spin singlet state on each rung . above this vacuum state , a hard core boson of spin 1 is the only possible excited state on each rung .
the system has a flatband dispersion @xmath90 for single particle excitations . due to the hard core scattering between the bosons , the statistics of the system
is essentially fermi - dirac type , @xcite which allows us to write down the partition function for the system : @xmath91 where @xmath92 is the ground - state energy per rung in the thermodynamic limit and the prefactor @xmath93 before @xmath94 comes from the three internal states of the single particle excitations . in the rung singlet phase ,
switching upon the couplings @xmath16 and @xmath17 will induce a dispersion @xcite for the single particle state , and the interaction between the particles
. classified by the internal symmetries , the two - particle interaction is either spin - channel independent or spin - channel dependent .
a bound state may also appear in certain spin - channels .
this has been observed in a previous study , @xcite in which it is shown that the bound state is also momentum dependent .
therefore , the interaction is very complicated and depends on the total momentum of the scattering particles .
since we are dealing with a system with massive particles at zero chemical potential , the density of the particles would be dilute at low temperatures .
taking into account of the form of the effective partition function for single particle dispersion @xmath95 , we expect it would match the accurate tmrg results at both high and low temperatures by including higher order harmonics in the dispersion : @xcite @xmath96 the energy gap is given by @xmath97 at @xmath98 . starting with a set of trial values for @xmath99 s given by perturbative calculation , @xcite we
minimize the squared 2-norm of the residual in free energy calculated from the trial dispersion : @xmath100 ^ 2 , % & \sqrt{\frac{1}{n}\sum_i % \left [ { \frac{\ln z_{\text{eff}}(\beta_i ) - \ln z_{\text{tmrg}}(\beta_i ) } % { \ln z_{\text{tmrg}}(\beta_i ) } } \right]^2 } , \label{eq : error}\end{aligned}\ ] ] where @xmath101 are smoothly distributed in the interval @xmath102 with a spacing @xmath103 .
typically , we show the fitting results for the @xmath67 case in table [ tab_1.0 ] .
the corresponding dispersions are plotted in fig .
[ disper ] . in our final numerical results ,
the values of @xmath104 in eq .
( [ eq : error ] ) are of order @xmath105 , and the mean deviation at each data point is less than @xmath33 .
0.22 cm .[tab_1.0]the partition function fitting results by eq .
( [ eq : zeff ] ) for the @xmath67 case . @xmath106 in the last column is the energy gap obtained by eq .
( [ eq:4 ] ) , as plotted in fig .
[ gap ] . [ cols="^,^,^,^,^,^,^,^",options="header " , ] ( color online ) dispersion ( in units of @xmath15 ) for @xmath67 with @xmath36 0 , 0.02 , 0.04 , 0.06 , 0.08 , and 0.1 in descending order along the direction of arrows.,width=283 ] from table [ tab_1.0 ] , we find that @xmath107 and @xmath108 monotonically decreases and increases with @xmath39 , respectively . this is consistent with the perturbative results .
@xcite on the other hand , the third order coefficient @xmath109 increases with @xmath39 . at @xmath75
, it reaches 0.15 , about one - third of @xmath110 .
this indicates that the approximation up to the third order in the dispersion is necessary for @xmath67 case . on the contrary , for @xmath29 , @xmath111 , @xmath112 , the absolute value of both @xmath110 and @xmath109
are less than @xmath113 , and the dispersion can be satisfactorily approximated by single - component harmonics with only @xmath108 .
this fact reflects that with the increase of @xmath16 and @xmath17 , higher orders in the harmonic expansion are needed to describe the one - particle excitation in the spectrum .
the dispersion relations shown in fig . [ disper ] should correspond to the one - triplet magnon branch in the spectrum of the ladder .
we compare the results with the previous studies by the dmrg .
@xcite we find they have quite similar shapes and are quantitatively agreement in the vicinity of @xmath98 .
the deviation when approaching @xmath114 can be attributed partially to the two - triplet bound state arising from the magnon - magnon interaction that we have ignored .
@xcite the energy gaps obtained in this way are presented in table [ tab_1.0 ] , and are compared with their counterparts from the specific heat data .
we see that they are rather coincident .
in addition , we have obtained the ground state energy ( @xmath92 in eq .
( [ eq : zeff ] ) ) , which is compatible with the results from the zero - temperature dmrg scheme .
in this paper , we explored the thermodynamic properties of the spin ladders with cyclic four - spin interactions by the tmrg method .
the temperature dependences of the specific heat and the susceptibility of the ladder in the rung - singlet phase are analyzed numerically .
based on the above results , we extracted the values of the spin gap and found that it manifests approximately linear behavior with the cyclic interaction in the parameter space we considered .
we also showed that the dispersion for low - lying excitations can be obtained from the partition function very effectively by combining with the mean - field or perturbative studies . comparing with the high temperature series expansion and the exact diagonalization approaches ,
@xcite we believe that the results presented in this study give more direct and accurate physical quantities in the thermodynamic limit .
further experimental results on specific heat and susceptibility may allow one to determine the various parameters for the two - leg spin ladder materials via this approach .
in addition , the dispersion relations extracted from the free energy may be used to be compared with the experimental data of the inelastic neutron scattering spectra .
10 a. gling , u. kuhlmann , c. thomsen , a. lffert , c. gross , nd w. assmus , phys .
b * 67 * , 052403 ( 2003 ) .
s. brehmer , h .- j .
mikeska , m. mller , n. nagaosa , and s. uchida , phys .
b * 60 * , 329 ( 1999 ) .
m. matsuda , k. katsumata , r. s. eccleston , s. brehmer , and h - j .
mikeska , phys . rev .
b * 62 * , 8903 ( 2000 ) ; m. matsuda , k. katsumata , r. s. eccleston , s. brehmer , and h - j .
mikeska , j. appl . phys .
* 87 * , 6271 ( 2000 ) .
t. s. nunner , p. brune , t. kopp , m. windt , and m. grninger , phys . rev .
b * 66 * , r180404 ( 2002 ) . y. honda , y. kuramoto , and t. watanabe , phys . rev .
b * 47 * , 11329 ( 1993 ) .
r. coldea , s. m. hayden , g. aeppli , t. g. perring , c. d. frost , t. e. mason , s .- w .
cheong , and z. fisk , phys .
lett . * 86 * , 5377 ( 2001 ) .
g. misguich , b. bernu , c. lhuillier , and c. waldtmann , phys .
lett . * 81 * , 1098 ( 1998 ) .
s. chakravarty _
philos . mag .
b * 79 * , 859 ( 1999 ) ; m. katano and d. s. hirashima , phys .
b * 62 * , 2573 ( 2000 ) ; b. bernu , ladir .
c@xmath115ndido , and d. m. ceperley , phys .
lett . * 86 * , 870 ( 2001 ) .
f. d. m. haldane , phys . lett . *
93a * , 464 ( 1983 ) ; phys .
lett . * 50 * , 1153 ( 1983 ) ; j. appl .
* 57 * , 3359 ( 1985 ) .
g. roux , s. r. white , s. capponi , a. luchli , and d. poilblanc , phys .
b * 72 * , 014523 ( 2005 ) .
s. r. white , r. m. noack , and d. j. scalapino , phys
lett . * 73 * , 886 ( 1994 ) .
e. dagotto and t. m. rice , science * 271*,618 ( 1996 ) , and references therein .
d. g. shelton , a. a. nersesyan , and a. m. tsvelik , phys .
b * 53 * , 8521 ( 1996 ) .
a. luchli , g. schmid , and m. troyer , phys .
b * 67 * , r100409 ( 2003 ) .
v. gritsev , b. normand , and d. baeriswyl , phys .
b * 69 * , 094431 ( 2004 ) .
a. a. nersesyan and a. m. tsvelik , phys .
78 * , 3939 ( 1997 ) .
p. p. kulish , n. y. reshetikhin , and e. k. sklyanin , lett . math
. phys . * 5 * , 393 ( 1981 ) ; l. a. takhtajan , phys . lett . * 87a * , 479 ( 1982 ) ; h. m. babujian , _ ibid .
* 90a * , 479 ( 1982 ) . k. hijii and k. nomura , phys
b * 65 * , 104413 ( 2002 ) .
k. hijii , shaojin qin , and k. nomura , phys .
b * 68 * , 134403 ( 2003 ) .
m. mller , t. vekua , and h .- j .
mikeska , phys .
b * 66 * , 134423 ( 2002 ) .
s. r. white , phys .
* 69 * , 2863 ( 1992 ) ; phys .
b * 48 * , 10345 ( 1993 )
. k. r. schmidt , h. monien , and g. s. uhrig , phys .
b * 67 * , 184413 ( 2003 ) .
e. mller - hartmann and a. a. reischl , eur .
j. b * 28 * , 173 ( 2002 ) .
a. bhler , u. lw , k. p. schmidt , and g. s. uhrig , phys .
b * 67 * , 134428 ( 2003 ) .
m. windt , m. grninger , t. nunner , c. knetter , k. p. schmidt , g. s. uhrig , t. kopp , a. freimuth , u. ammerahl , b. bchne , and a. revcolevschi , phys .
lett . * 87 * , 127002 ( 2001 ) .
r. j. bursill , t. xiang , and g. a. gehring , j. phys . :
matter * 8 * , l583 ( 1996 ) .
x. wang and t. xiang , phys .
b * 56 * , 5061 ( 1997 ) .
t. xiang and x. wang , in _ density - matrix renormalization : a new numerical method in physics _ , edited by i. peschel , x. wang , m. kaulke , and k. hallberg ( springer , new york , 1999 ) , pp .
149 - 172 .
t. xiang , phys .
b * 58 * , 9142 ( 1998 ) .
d. coombes , t. xiang , and g. a. gehring , j. phys . : condense .
matter * 10 * , l159 ( 1998 ) .
m. troyer , h. tsunetsugu , and d. wrtz , phys . rev .
b * 50 * , 13515 ( 1994 ) .
jizhong lou , shaojin qin , tai - kai ng , zhaobin su , and ian affleck , phys .
b * 62 * , 3786 ( 2000 ) ; k. okunishi , _ ibid . _ * 60 * , 4043 ( 1999 ) . | the temperature dependence of the specific heat and spin susceptibility of the spin ladders with cyclic four - spin interactions in the rung - singlet phase is explored by making use of the transfer - matrix renormalization group method .
the values of spin gap are extracted from the specific heat and susceptibility , respectively .
it is found that for different relative strength between interchain and intrachain interactions , the spin gap is approximately linear with the cyclic four - spin interaction in the region far away from the critical point .
furthermore , we show that the dispersion for the one - triplet magnon branch can be obtained by numerically fitting on the partition function . |
consider the the beta distribution @xmath4 , with the density function , @xmath5 the mean of @xmath0 is readily obtained by the formula @xmath6 , but there is no general closed formula for the median .
the median function , here denoted by @xmath7 , is the function that satisfies , @xmath8 the relationship @xmath9 holds . only for the special cases @xmath10 or @xmath11 we may obtain an exact formula : @xmath12 and @xmath13 .
moreover , when @xmath14 , the median is exactly @xmath15 .
there has been much literature about the incomplete beta function and its inverse ( see e.g. @xcite for a review ) .
the focus in literature has been on finding accurate numerical results , but a simple and practical approximation that is easy to compute has not been found .
trivial bounds for the median can be derived @xcite , which are a consequence of the more general mode - median - mean inequality @xcite . in the case of the beta distribution with @xmath16 , the median is bounded by the mode @xmath17 and the mean @xmath6 : @xmath18 for @xmath19 the formula for the mode does not hold as there is no mode . if @xmath20 , the order of the inequality is reversed .
equality holds if and only if @xmath14 ; in this case the mean , median , and mode are all equal to @xmath15 .
this inequality shows that if the mean is kept fixed at some @xmath21 , and one of the shape parameters is increased , say @xmath22 , then the median is sandwiched between @xmath23 and @xmath21 , hence the median tends to @xmath21 . from the formulas for
the mode and mean , it can be conjectured that the median @xmath7 could be approximated by @xmath24 for some @xmath25 , as this form would satisfy the above inequality while agreeing with the symmetry requirement , that is , @xmath26 .
since a @xmath4 variate can be expressed as the ratio @xmath27 where @xmath28 and @xmath29 ( both with unit scale ) , it is useful to have a look at the median of the gamma distribution .
@xcite studied the median function of the unit - scale gamma distribution median function , denoted here by @xmath30 , for any shape parameter @xmath31 , and obtained @xmath32 , rapidly approaching @xmath33 as @xmath22 increases .
it can therefore be conjectured that the distribution median may be approximated by , @xmath34 figure ( [ fig - betaerrors ] ) shows that this approximation indeed appears to approach the numerically computed median asymptotically for all distribution means @xmath35 as the ( smaller ) shape parameter @xmath36 . for @xmath37 ,
the relative error is less than 4% , and for @xmath38 this is already less than 1% .
figure ( [ fig - betaerrp ] ) shows the relative error over all possible distribution means @xmath35 , as the smallest of the two shape parameters varies from @xmath3 to @xmath39 .
this illustrates how the relative error tends uniformly to zero over all @xmath21 as the shape parameters increase .
the figure also shows that the formula consistently either underestimates or overestimates the median depending on whether @xmath40 or @xmath41 .
however , the function @xmath42 approximates the median fairly accurately if some other @xmath43 close to @xmath44 ( say @xmath45 ) is chosen .
figure ( [ fig - betadisterr ] ) displays curves of the logarithm of the absolute difference from the numerically computed median for a fixed @xmath46 , as the shape parameter @xmath22 increases .
the absolute difference has been scaled by @xmath21 before taking the logarithm : due to this scaling , the error stays approximately constant as @xmath21 decreases so the picture and its scale will not essentially change even if the error is computed for other values of @xmath40 .
the figure shows that although some approximations such as @xmath45 has a lower absolute error for some @xmath22 , the error of @xmath47 tends to be lower in the long run , and moreover performs more consistently by decreasing at the same rate on the logarithmic scale . in practical applications
, @xmath48 should be a sufficiently good approximation of @xmath49 .
another measure of the accuracy is the tail probability @xmath50 of a @xmath0 variate @xmath51 : good approximators of the median should yield probabilities close to @xmath15 .
figure ( [ fig - betatail ] ) shows that as long as the smallest of the shape parameters is at least 1 , the tail probability is bound between @xmath52 and @xmath53 .
as the shape parameters increase , the probability tends rapidly and uniformly to @xmath54 .
finally , let us have a look at a well - known paper that provides further support for the uniqueness of @xmath55 .
@xcite and @xcite provide approximations for the probability function @xmath56 of a @xmath4 variate @xmath51 . although they do not provide a formula for the inverse , it is the probability function at the approximate median . according to @xcite , @xmath56
is well approximated by @xmath57 where @xmath58 is the standard normal probability function , and @xmath59 is a function of the shape parameters and the quantile @xmath60 .
consider @xmath61 : @xmath62 should be close to zero and at least tend to zero fast as @xmath22 and @xmath63 increase .
now assume that @xmath21 is fixed , @xmath22 varies and @xmath64 .
the function @xmath65 equals , rewritten with the notation in this paper , @xmath66 \right ) \left(\frac{1+f(a , p;d)}{m(1-m)}\right)^{1/2},\ ] ] where the function @xmath67 tends to zero as @xmath22 increases , being exactly zero only when @xmath68 or @xmath69 .
it is evident that for the fastest convergence rate to zero , one should choose @xmath49 .
this is of the order @xmath70 ; if @xmath71 , for example if we choose the mean @xmath21 as the approximation of the median ( @xmath72 ) , the rate is at most @xmath73 . | a simple closed - form approximation for the median of the beta distribution @xmath0 is introduced : @xmath1 for @xmath2 both larger than @xmath3 has a relative error of less than 4% , rapidly decreasing to zero as both shape parameters increase .
keywords : beta distribution , distribution median |
due to the complication of a non - abelian @xmath2 gauge theory , quantum chromodynamics ( qcd ) , which describes the strong interaction , has many non - perturbative features such as the dynamical breaking of chiral symmetry and quark confinement . in the study of hadron structure at low energy scale ,
two kinds of models of qcd are often used , each incorporating some important qcd features .
the constituent quark potential model which incorporates the qcd quark confinement has been impressively successful in hadron spectroscopy and decays @xcite , except for the pseudo - scalar @xmath0 meson which is a goldstone boson and has very low mass .
it was shown recently that the rpa @xmath3-component must be considered in the treatment of pions @xcite .
on the other hand , the nambu - jona - lasinio ( njl ) model @xcite and several of its extensions @xcite which incorporate the chiral symmetry can describe the @xmath0 meson very well as a massless goldstone boson of dynamical breaking of chiral symmetry . however , the njl model lacks the important qcd property of quark confinement . in ref .
@xcite , we proposed an extension to the constituent quark model to comprise the rpa @xmath3-component quark excitations .
as in njl model , we start from an effective quark hamiltonian . for meson structure
, this leads to a hamiltonian with two channels the ordinary valence quark @xmath4-channel and a new @xmath3-channel , which is an extension of the quark potential model .
this extended quark potential model has a new coupling potential connecting the ordinary valence quark @xmath4-channel and the small component @xmath3-channel .
the coupling potential can be related to the ordinary valence quark potential in the model . in this paper
, we implement the chiral symmetry in the extended constituent quark potential model .
first we introduce the extended quark potential model . in sec . 3
, we analyze @xmath0 and @xmath1 mesons using this framework . the restriction from chiral symmetry
is considered . with the meson wave functions containing @xmath3-channel quark excitations , ksrf relation @xcite about @xmath0 and @xmath1 meson decay constants
is derived .
finally , we summarize our discussion .
qcd perturbation theory is very successful when applied in the high energy processes . in the low energy regime
, hamiltonian approach can provide an alternative method to understand the non - perturbative features of qcd , such as the quark confinement .
many efforts were made to obtain an effective hamiltonian from the exact qcd lagrangian @xcite .
here we start from the effective quark hamiltonian , assuming the quark fields have been separated approximately from the gluon fields @xmath5 where @xmath6 is the free current quark hamiltonian and @xmath7 is the current quark mass .
@xmath8 includes all effective quark interactions .
@xmath9 is the kernel function for coupling vertex @xmath10 .
apart from a constant , the hamiltonian ( [ h ] ) can be written in normal order as @xmath11 where @xmath12 represents the free particle energy of constituent quark and its general form is @xmath13 \psi(\bm{x}),\ ] ] where the dependence on the three - vector momentum arises from the instantaneous approximation of quark interactions .
the two functions @xmath14 and @xmath15 should be obtained from a self - consistent mean field calculation .
however in a constituent quark potential model , the quark interactions contain the linear quark confinement which has severe infrared divergence in momentum space . on the other hand
, the constituent potential quark models always assume that the constituent quarks have fixed masses . as an extension of the constituent potential quark model , here
we assume that the @xmath16 and @xmath17 functions can be approximated as constants .
so we treat the constituent quark mass as a parameter , @xmath18 ( with @xmath19 ) .
the free particle hamiltonian becomes @xmath20 \psi(\bm{x}).\ ] ] in this paper , we will only consider meson structures . in nave quark model , meson fields are approximately expressed in terms of local quark fields as @xmath21 where @xmath22 for iso - vector pseudo - scalar @xmath0 mesons or @xmath23 for iso - vector vector @xmath1 mesons . according to @xmath24 expansion , the above meson structures are exact in the limit @xmath25 @xcite . in the non - relativistic limit , as quark momentum @xmath26 , one has@xcite @xmath27,\end{aligned}\ ] ] for @xmath0 and @xmath1 mesons . on the other hand , the field operators of @xmath0 and @xmath1
can also be expressed in the @xmath28 and @xmath29 of meson creation and annihilation operators @xmath30.\ ] ] if the mesons are composed of creation of valence quark pairs as in the potential models , one deduces from the equivalence of eqs .
( [ meson - f1 ] ) and ( [ meson - f2 ] ) @xmath31 however , generally one can only obtain @xmath32.\end{aligned}\ ] ] this is just the rpa excitation operator the meson state is a superposition of creation and annihilation of @xmath33 pairs on the vacuum .
the second part , i.e. the @xmath3-component is simply discarded in the ordinary quark potential model as a small component .
thus , as an extension to the potential model , we take the mesons as excitation modes of the vacuum of the rpa type : @xmath34 for a meson in the rest frame , after we take account of the quark interactions , @xmath29 excitation operators are @xmath35,\end{aligned}\ ] ] where the time reversals are defined as @xmath36 and @xmath4 and @xmath3 are the rpa amplitudes determined from the well - known rpa equation of motion@xcite @xmath37 \right ] \left| 0 \right\rangle = e \left\langle 0 \right| [ \delta q , q^\dagger ] \left| 0 \right\rangle.\ ] ] we obtain the equations for rpa amplitudes @xmath38 , \\
\label{yrpa - m } & & \left(e + 2 k^0 \right ) y(\bm{k},s_1,s_2 ) \nonumber \\ & = & -\sum_{s_3s_4 } \int \frac{d^3k^{\prime}}{(2\pi)^3 } \left [ \langle \bm{k},s_1,s_2 \mid v \mid \bm{k}',s_3,s_4 \rangle x(\bm{k}',s_3,s_4 ) \right .
\nonumber \\ & & \left .
+ \langle \bm{k},s_1,s_2 \mid u \mid \bm{k}',s_3,s_4 \rangle
y(\bm{k}',s_3,s_4 ) \right].\end{aligned}\ ] ] here we have introduced two effective potentials @xmath39 and @xmath40 .
their matrix elements are @xmath41 thus , with the rpa approximation , the quark potential model is extended to a coupling system with two channels .
the first part , i.e. x - amplitude @xmath42 which we will call the x - channel wave function , is the large component which is the sole valence quark contribution in the ordinary quark potential model .
the second part , i.e. the y - amplitude @xmath43 which we will call the y - channel wave function , is the small component which is discarded in the ordinary quark potential model .
the potential @xmath39 interacts only within each individual channel while the new coupling potential @xmath40 couples the two channels together .
if the coupling between the two channels @xmath40 is small , the system decouples .
the extended potential model reduces to the ordinary potential model and the potential @xmath39 is just the ordinary quark potential . here
we will show that the coupling potential @xmath40 is important due to chiral symmetry .
one can use the dirac kets and bras to simplify notations . @xmath42 and @xmath43 are represented by two kets @xmath44 and @xmath45 respectively , @xmath46 eqs .
( [ xrpa - m ] ) and ( [ yrpa - m ] ) can be written concisely as @xmath47 or the familiar matrix form @xmath48 \left[\begin{array}{c } |x\rangle \\
|y\rangle \end{array}\right ] = e \left[\begin{array}{c } |x\rangle \\ |y\rangle \end{array}\right],\ ] ] where @xmath49 . by definitions ( [ u ] ) and ( [ v ] ) ,
the potentials @xmath39 and @xmath40 will be evaluated from the same kernels in the quark interaction @xmath8 , thus they can be related to each other . in the non - relativistic limit @xmath50 , we have @xmath51 where on the right hand side of equations , the kernels are divided into five types as in ref .
@xcite , i.e. , scalar ( @xmath52 ) , pseudo - scalar ( @xmath53 ) , vector ( @xmath40 ) , axial - vector ( @xmath16 ) , and tensor ( @xmath54 ) ; @xmath55 s are the pauli spin matrices . to normalize the model wave functions ,
we apply the standard boson commutation relation @xmath56 = ( 2\pi)^3 2p^0 \delta^3(\bm{p}-\bm{p}')\ ] ] to rest mesons .
we obtain @xmath57 then we find the normalization relation of rpa type @xmath58 with the normalized meson wave functions , one can calculate the static properties of mesons .
the @xmath0 weak decay constant is defined as @xmath59 for pions at rest , the above equation becomes @xmath60 in the non - relativistic limit , we obtain @xmath61 where @xmath62 and @xmath63 are the coordinate wave functions in x- and y- channel of the @xmath0 meson with its total spin @xmath64 . the @xmath1 electro - magnetic decay
constant is defined as @xmath65 where @xmath66 are the isospin indices , and @xmath67 is the @xmath1 meson polarization index . for static @xmath1 mesons @xmath68 in the non - relativistic limit ,
one obtains @xmath69
following the quark potential model @xcite , we choose the effective quark interaction @xmath8 as a vector interaction plus a scalar confinement interaction @xmath70 according to lattice calculation , the scalar confinement kernel is linear in @xmath71 @xmath72 the vector kernel is taken from the one - gluon exchange plus a constant @xmath73 where @xmath74 is the qcd running coupling constant . in our analysis
, we use the simple harmonic oscillator wave functions for pions and rho mesons as a first approximation . in this way
the underlying physics is emphasized .
@xmath75 where @xmath52 , @xmath76 are the spin quantum numbers of the meson , and @xmath77 is the ground state wave function of harmonic oscillator in momentum space @xmath78 in the non - relativistic limit , the two potentials @xmath39 and @xmath40 are @xmath79 where @xmath80 here , one must be careful with the one - gluon exchange interaction @xmath81 .
it arises from the covariant form @xmath82 in momentum space , where @xmath83 is the 4-momentum of the exchanged gluon . for potential
@xmath39 , @xmath84 and @xmath85 are the 4-momenta of incoming and outgoing quarks respectively .
thus in non - relativistic limit , @xmath86 which is a coulomb potential in coordinate space .
however for coupling potential @xmath40 , @xmath87 and @xmath88 are the 4-momenta of @xmath33 quark pair which were created or annihilated on the vacuum .
so in non - relativistic limit , @xmath89 . the term @xmath81 is suppressed in @xmath40 .
it is well known in njl model , the chiral symmetry can be broken dynamically with a nonzero vacuum condensate coming from quark interaction .
the current quarks gain dynamical masses and become constituent quarks which are the valence quarks in potential model .
the goldstone boson pions are still massless . to account for this dynamical chiral symmetry breaking in our model
, the coupling potential @xmath40 must be large .
this is why we introduce a constant term into the vector kernel . in this way
, our model differs from the ordinary potential model ( see ref .
@xcite ) which had the constant term in the scalar interaction . in an ordinary potential model
, this change will cause no difference ( the potential @xmath39 is not changed ) .
but in this extended potential model , the coupling potential @xmath40 can be made strong enough to give @xmath0 mesons very low masses . inserting the simple model wave functions ( [ test - x - wvf ] ) and ( [ test - y - wvf ] ) into eq .
( [ eqm - rpa ] ) with the potentials given by ( [ u - pot - simp ] ) and ( [ v - pot - simp ] ) , we obtain an eigen equation of a @xmath90 matrix @xmath91 \left[\begin{array}{c } x \\ y \end{array}\right ] = e \left[\begin{array}{c } x \\ y \end{array}\right],\ ] ] where @xmath92 the energy @xmath93 can be easily obtained @xmath94 the wave function parameter @xmath95 is determined by the minimum of the energy @xmath96 since @xmath17 is a constant , we get @xmath97 we approximate the average quark kinetic energy by the following expression . @xmath98
numerically the deviation is very small and less than 5% ( see fig . [ eps-1 ] ) .
comparison of the integral @xmath99 ( solid line ) in the average quark kinetic energy with @xmath100 ( dashed line ) used in eq .
( [ kk ] ) . ] the current quark mass @xmath7 can be obtained from the partial conservation of axial - vector current ( pcac ) .
consider the axial - vector current @xmath101 according to pcac , its divergence is @xmath102 where @xmath103 and @xmath104 are current masses of up and down quarks .
let us calculate the matrix element @xmath105 for @xmath106 meson with momentum @xmath107 .
one obtains @xmath108 > from right hand side of pcac relation ( [ pcac ] ) , one also has @xmath109 both of the matrix elements can be easily calculated using the model wave functions for a static @xmath106 meson .
we obtain @xmath110 before the numerical calculation , one can make some qualitative analysis of the properties of @xmath0 and @xmath1 mesons in the model .
let @xmath111 the matrix equation ( [ eqm - rpa - simp ] ) can be written for @xmath112 , @xmath113 @xmath114 also the normalization relation ( [ normalization ] ) becomes @xmath115 first , let us consider the @xmath0 meson which is a goldstone boson .
we have @xmath116 where @xmath117 if the current quark mass @xmath118 , chiral symmetry is a strict qcd symmetry . from the goldstone theorem
, @xmath0 meson should be massless .
as an effective theory of qcd , the @xmath0 meson mass in our model should also be zero .
assuming @xmath119 is the constituent quark mass in the limit of @xmath118 , we have @xmath120 in the real world , current quarks have small nonzero masses .
so does the @xmath0 meson .
as the average mass of current quarks ( u and d ) slightly changes to @xmath121 , the constituent quark mass will also change a little bit to @xmath122 with @xmath123 .
so we have @xmath124 and @xmath125 eqs .
( [ eqn - pi-1 ] ) and ( [ eqn - pi-2 ] ) become @xmath126 we obtain @xmath127 with the normalization @xmath128 , we obtain @xmath129 > from pcac ( eq .
( [ pcac - m0 ] ) ) , the mass of current quark is related to @xmath130 as @xmath131 next we can calculate the weak decay constant of @xmath0 meson to obtain @xmath132 eq .
( [ uv_pi ] ) means @xmath133 and @xmath134 , which shows the importance of the contribution to the @xmath0 meson from the small component of y - channel . for the @xmath1 meson , with the help of eq .
( [ eqn - k ] ) , the equations are @xmath135 one simply obtains @xmath136 with the normalization @xmath137 , @xmath138 next the electromagnetic decay constant of @xmath1 meson can be calculated as @xmath139 a relation among masses of @xmath0 , @xmath1 mesons and current quark mass @xmath7 can be easily obtained @xmath140 > from the experimental @xmath1 meson mass @xmath141 mev , one can estimate that @xmath142 mev .
then we can estimate the current quark mass to be @xmath143 mev from the physical pion mass @xmath144 mev .
another relation about @xmath0 and @xmath1 decay constants can also be easily obtained from eqs .
( [ f_pi_cal ] ) and ( [ f_rho_cal ] ) , @xmath145 the ksrf relation @xcite gives @xmath146 while the experiment value is @xmath147 . in our approach @xmath148 from eq .
( [ uv_rho ] ) .
this means the contribution to the rho meson from y - channel is rather small .
if the small component is completely neglected , i.e. , @xmath149 , we may obtain @xmath150 and @xmath151 .
then @xmath152 thus @xmath146 which is exactly the ksrf relation .
the experimental value is between our result and ksrf relation . in numerical calculation
we parametrize the running coupling constant as @xmath153 as in ref .
@xcite ( with @xmath28 in gev ) . also the constituent quark mass @xmath119
is fixed to be @xmath154 mev .
the results are listed in table [ table-1 ] .
first ( set 1 ) , we choose the confinement parameter @xmath155gev@xmath156 according to lattice calculation .
the @xmath1 meson mass is somehow larger .
next ( set 2 ) , we adjust the confinement parameter slightly to @xmath157gev@xmath156 .
.[table-1]numerical results for @xmath0 and @xmath1 mesons .
experimetal data are taken from ref .
@xcite . [ cols="<,^,^,^,^,^,^ " , ] both of pion and rho meson masses agree with the experimental values well .
the current and constituent quark masses are also reasonable in the ranges of theoretical and experimental estimations .
the decay constants @xmath158 and @xmath159 deviate from the experimental data by a factor 2 . this is not very surprising . in the derivation of the effective hamiltonian ( which we skip in this work )
, the current quark field @xmath160 is renormalized into constituent quark field .
this means that the constituent quark may include the contributions from configurations other than the single quark , like @xmath161 , @xmath162 etc as indicated through the deep inelastic scattering experiments .
for example , we may assume the normalized constituent quark field @xmath163 @xmath164 here we have introduced the factor @xmath165 .
roughly speaking , @xmath166 is the probability of finding the `` pure '' current quark field in a complicated constituent quark .
however , from the electro - weak lagrangian we know that what really participates in the electro - weak interactions are the current quarks
. in other words , only the `` pure '' current quark will contribute to the pion weak decay constant and rho meson electromagnetic decay constant . in our calculation
we need two real current quarks to annihilate .
so there will be a difference of factor @xmath167 .
decay constants @xmath158 and @xmath159 can then be fit to the experimental values if we set the additional normalization factor @xmath168 ( in table [ table-1 ] the results are listed in parentheses ) .
the chiral symmetry is studied in an extended constituent quark potential model which includes the small component effect with rpa approximation . with some modification of quark interaction in potential model ,
@xmath0 meson is still a goldstone boson .
the small mass of @xmath0 meson can be connected to nonzero current quark mass from pcac .
all other mesons such as the vector @xmath1 meson can also be studied in this model like in the ordinary constituent quark models . with the small component effect of y - channel contribution , the well - known ksrf relation of the @xmath0 and @xmath1 meson decay constants
is reestablished in the model .
this extension to constituent quark potential model aims at improving the constituent quark model to comprise the small component , i.e. , the @xmath169 channel quark excitations which are essential to @xmath0 ( and k ) mesons . to thoroughly investigate the effect of this extension , we need perform a full calculation of meson spectroscopy .
this work is supported by national natural science foundation of china , ministry of education of china , fanedd and srf for rocs , sem . | the chiral symmetry is applied to an extended constituent quark potential model . with random phase approximation ( rpa )
, a small component effect is added to the constituent quark model . to obtain the pseudoscalar @xmath0 meson as a goldstone boson ,
the quark effective potentials are modified in the model to account for the dynamical breaking of chiral symmetry .
also the vector @xmath1 meson is calculated and the ksrf relation about @xmath0 and @xmath1 meson decay constants is derived in the model . , , , , chiral symmetry , quark potential model , random phase approximation 12.39.pn , 14.40.aq |
achieving high directivity with compact radiators has been a major concern of the antenna community since its early days @xcite .
still today , many modern applications , such as automotive radars , satellite communication , millimetre - wave point - to - point communication , and microwave imaging , strive for simple and efficient low - profile antennas producing the narrowest possible beams @xcite . extending the size of the radiating aperture leads to an enhanced directivity , but only if the aperture is efficiently excited .
to date , uniform illumination of large apertures is achievable with reflectors and lenses ; although these can be made compact using concurrent metamaterial concepts , they still require substantial separation between the source and the aperture , resulting in a large overall antenna size @xcite .
in addition , feed blockage and spillover effects must also be considered , usually complicating the design and reducing the device efficiency .
high aperture efficiencies can also be achieved using antenna arrays @xcite ; nevertheless , the requirement for elaborated feed network significantly increases the complexity of this solution and limits its compactness @xcite , and may also introduce considerable feed - network losses .
leaky - wave antennas ( lwas ) , on the other hand , can produce directive beams using a low - profile structure fed by a simple single source @xcite .
their typical configuration consists of a guiding structure with a small perturbation , facilitating coupling of guided modes to free - space radiation . in the much - discussed fabry - prot ( fp )
lwas , a localized source is sandwiched between a perfect electric conductor ( pec ) and a partially - reflecting surface ( prs ) , forming a longitudinal fp cavity @xcite . by tuning the cavity height at the design frequency @xmath0 , favourable coupling of the source to a single parallel - plate waveguide mode
is achieved , forming a dominant leaky wave emanating from the source ; the typical device thickness lies around half of a wavelength .
the leaky mode is characterized by a transverse wavenumber whose real part @xmath1 corresponds to the waveguide dispersion , and is accompanied by a small imaginary part @xmath2 determined by the prs .
this leads to an azimuthally - symmetric directive radiation through the prs towards the direction defined by @xmath3 , where @xmath4 is the free - space wavenumber , with a beamwidth proportional to @xmath2 .
broadside radiation is achieved when @xmath5 is small enough such that the splitting condition @xmath6 is satisfied , and the peaks of the conical beam merge @xcite .
another class of lwas which has received significant attention lately is based on modulated impedance metasurfaces ( mometas ) @xcite .
these so - called holographic antennas use a point source to excite surface waves on a thin dielectric sheet covered with metallic patches and backed by a pec ground plane , establishing effective surface impedance boundary conditions @xcite ; guiding surface waves , these structures can be very thin , below fifth of a wavelength .
similar to fp - lwas , the guiding structure is designed such that only a single surface mode @xmath7 is allowed to propagate ; small modulation of the surface impedance , implemented by variation of patch sizes or dielectric thickness , couples the bound modes to radiative modes . to facilitate such coupling in the case of surface waves ,
whose transverse momentum is greater than that of free - space , the impedance modulation should have a periodicity @xmath8 comparable to the wavelength . the interaction between the surface wave and the perturbation results in an infinite number of floquet - bloch ( fb ) harmonics @xmath9 ; the periodicity should be designed such that one of them radiates to the desirable direction , while the others become evanescent , ensuring good directivity .
the leakage rate @xmath2 , and correspondingly the beamwidth , are determined by the depth of the modulation @xcite .
both fp - lwas and mometas have an appealing compact configuration and their radiation characteristics can be rather simply controlled by tuning the properties of the guiding structure and the perturbation . nonetheless , due to their leaky - wave nature , they suffer from a fundamental efficiency limitation when considering practical finite apertures : designing a moderate leakage rate @xmath2 with respect to the aperture length @xmath10 yields uniform illumination of the aperture ( high aperture efficiency ) but results in considerable losses from the edges ( low radiation efficiency ) ; on the other hand , large values of @xmath2 lead to high radiation efficiencies but in this case only a portion of the aperture is used for radiation , leading to a wider beam @xcite . to mitigate edge - taper losses , shielded fp - lwa structures have been recently proposed , using pec side walls which form a lateral cavity @xcite .
nevertheless , the tight coupling between the propagation of the leaky mode inside the fp cavity and the angular distribution of the radiated power manifested by @xmath3 poses serious limitations on the achievable aperture efficiency .
this is most prominent for antennas radiating at broadside , in which only low - order lateral modes , carrying transverse wavenumbers which are small enough to satisfy the splitting condition , can be used .
consequently , such antennas are designed to excite exclusively the @xmath11 lateral mode , which inherently limits the aperture efficiency , defined as the relative directivity with respect to the case of uniform illumination , to about @xmath12 @xcite .
although this problem can be partially solved by the use of artificial magnetic conductor ( amc ) side walls instead of pecs @xcite , one of the most limiting constraints on the antenna design which still remains is the requirement for mode purity . as the dominant spectral components of the field in the cavity directly translate to prominent lobes in the radiation pattern , only a single mode should be allowed to propagate in the cavity to guarantee high directivity . however , as demonstrated by @xcite , suppression of parasitic modes in a cavity is usually a very difficult problem .
in particular , single mode excitation becomes increasingly challenging as the desirable aperture size increases , due to the small differences between the wavenumbers associated with low - order modes ; thus , in practice , this solution can not be used for arbitrarily - large apertures . from the discussion
so far it follows that it would be very beneficial if we could optimize separately the fields inside the cavity and the fields formed on the aperture .
this would allow us to achieve good illumination of the aperture without the necessity to meet restricting conditions ( e.g. , the splitting condition , or single mode excitation ) , stemming from the coupling between excitation and radiation fields . but how to achieve such a separation ?
the equivalence principle suggests that for a given field exciting a surface , desirable ( arbitrary ) aperture fields can be formed by inducing suitable electric and magnetic surface currents , supporting the required field discontinuities @xcite .
based on this idea , the concept of huygens metasurfaces ( hmss ) has been recently proposed as a means for versatile wavefront manipulation @xcite .
huygens metasurfaces are planar structures composed of subwavelength elements ( meta - atoms ) , engineered to generate the surface currents required by the equivalence principle to achieve a prescribed functionality . in general , for a given incident field and desirable transmitted field , elements exhibiting effective loss and gain are required for the implementation @xcite .
however , for certain applications , the fields can be judiciously stipulated such that the metasurface can be constructed from passive and lossless elements , i.e. electric and magnetic polarizable particles @xcite . in fact , we have recently shown that if the reflected and transmitted fields are set such that the wave impedance and the real power are continuous across the two facets of the metasurface , the aperture phase can be tailored by a passive and lossless hms to produce directive radiation towards a prescribed angle @xmath5 , for any given excitation source ; the design procedure is straightforward once the source plane - wave spectrum is assessed @xcite .
indeed , in this paper we propose to harness the equivalence principle to efficiently convert fields excited in a cavity by a localized source to highly - directive radiation using a huygens metasurface : cavity - excited hms antenna .
the device structure resembles a typical shielded fp - lwa configuration , with an electric line source surrounded by three pec walls and a huygens metasurface replacing the standard prs ( fig .
[ fig : physical_configuration ] ) . for a given aperture length @xmath10 and a desirable transmission angle @xmath5 ,
we optimize the fp cavity thickness and source position to predominantly excite the highest - order mode of the lateral cavity , with the hms reflection coefficient ensuring the wave impedance is equalized along the metasurface ; this guarantees the aperture is well illuminated .
once the source configuration is established , we stipulate the aperture fields to follow the power profile of the cavity mode , ensuring the real power is conserved at each point , and impose the suitable linear phase to promote radiation towards @xmath5 . with the cavity fields and aperture fields in hand , we invoke the equivalence principle and evaluate the electric surface impedance and magnetic surface admittance required to support the resultant field discontinuity @xcite .
our previous work @xcite guarantees that these would be purely reactive , hence could be implemented using passive and lossless meta - atoms . utilizing the equivalence principle as described results in formation of aperture fields , the magnitude of which follows the power distribution inside the cavity , whereas their phase is independently determined to vary in a plane - wave - like fashion .
this has two important implications .
first , as the power profile of the highest - order lateral mode creates hot spots of radiating surface currents approximately half a wavelength apart , a uniform virtual phased - array is formed on the hms aperture ; based on array theory @xcite , such excitation profile is expected to yield very high directivity with no grating lobes regardless of the scan angle @xmath5 .
second , in contrast to lwas of any type , the antenna directivity does not deteriorate significantly even if other modes are partially excited , as these would merely vary the amplitude of the virtual array elements , without affecting the phase purity .
this semianalytical design procedure can be applied to arbitrarily - large apertures , yielding near - unity aperture efficiencies , in agreement with full - wave simulations ; due to the pec side walls , no power is lost via the edges .
this offers an effective way to overcome the efficiency tradeoff inherent to fp - lwas and mometas , while preserving the advantages of a single - feed low - profile antenna .
to design the hms - based antenna , we simply apply the general methodology developed in @xcite to the source configuration of fig . [
fig : physical_configuration ] ; for completeness , we recall briefly its main steps .
we consider a 2d scenario ( @xmath13 ) with the hms at @xmath14 and a given excitation geometry at @xmath15 embedded in a homogeneous medium ( @xmath16 , @xmath17 ) . under these circumstances ,
the incident , reflected and transmitted fields in the vicinity of the hms can be expressed via their plane - wave spectrum @xcite @xmath18 where @xmath19 is the inverse spatial fourier transform of @xmath20 , @xmath21 is the source spectrum , @xmath22 is the hms reflection coefficient , and @xmath23 $ ] is the transmission spectrum . as before , @xmath1 denotes the transverse wavenumber , and the longitudinal wavenumber is @xmath24 . for simplicity
, we only consider here transverse electric ( te ) fields ( @xmath25 ) ; the nonvanishing magnetic field components @xmath26 can be calculated from @xmath27 via maxwell s equations . for a given source spectrum
, it is required to determine the reflected and transmitted fields , through the respective degrees of freedom @xmath22 and @xmath28 , that would implement the desirable functionality .
once the tangential fields on the two facets of the hms are set , the equivalence principle is invoked to evaluate the required electric and magnetic surface currents to induce them @xcite .
the polarizable particles comprising the hms are then designed such that the average fields acting on them induce these surface currents @xcite .
analogously , the hms can be characterized by its electric surface impedance @xmath29 and magnetic surface admittance @xmath30 , relating the field discontinuity and the average excitation via the generalized sheet transition conditions ( gstcs ) @xcite , to promote directive radiation towards @xmath5 we require that the aperture ( transmitted ) fields approximately follow the suitable plane - wave - like relation@xmath31 where @xmath32 is the aperture window ( envelope ) function ( yet to be determined ) and @xmath33 is the te wave impedance of a plane - wave directed towards @xmath5 . in previous work , we have shown that if the wave impedance and the real power are continuous across the metasurface , these aperture fields can be supported by a passive lossless hms ( purely reactive @xmath34 and @xmath35 ) @xcite .
the first condition , local impedance equalization , means that the total ( incident+reflected ) fields on the bottom facet of the metasurface should exhibit the same wave impedance , i.e. @xmath36 ; this is achieved by setting the reflection coefficient to a fresnel - like form @xmath37 determining the reflected fields everywhere , fixing our first degree of freedom . to satisfy the second condition , local power conservation , we require that the aperture window function follows the _ magnitude _ of the total ( incident+reflected ) fields at @xmath38 , namely , @xmath39f\left(k_t\right)\right\rbrace\right| \nonumber \\ & = \left(\mathcal{f}^{-1}\left\lbrace \textstyle \left[\frac{1}{2\beta}\left(1-\gamma\right)f\right]\star\left[\frac{1}{2\beta}\left(1-\gamma\right)f\right]\right\rbrace\right)^{1/2 } , % \frac{1}{2\pi}\left|\int_{-\infty}^{\infty}\frac{dk_t}{2\beta } % \left[1-\gamma\left(k_t\right)\right]f\left(k_t\right)e^{jk_ty}\right|e^{-j\xi_\mathrm{out } } , \nonumber \label{equ : aperture_window_function}\end{aligned}\ ] ] where @xmath40 is the autocorrelation of the spectral - domain function @xmath41 @xcite ; this determines the transmitted fields everywhere , fixing our second degree of freedom .
the absolute value operator in the last equality is of utmost significance : it indicates that the transmission spectrum of the aperture fields follows , up to a square root , the _ power spectral density _
( psd ) of @xmath42 , and _ not _ the spectral content of the incident and reflected _
is directly related to the balanced ( plane - wave - like ) contribution of the electric and magnetic fields to the power flow that we stipulated in eq . , and results in a significantly favourable plane - wave spectrum , as will be discussed in detail in the next subsection .
finally , we use these semianalytically predicted fields and the equivalence principle , manifested by the gstcs , to calculate the required hms surface impedance , yielding the desirable purely reactive modulation given by , @xmath43\right\rbrace \frac{z_{se}\left(y\right)}{z_\mathrm{out}}=\frac{y_{sm}\left(y\right)}{y_\mathrm{out}}=-\frac{j}{2}\cot\left[\frac{\phi_{-}\left(y\right)-\phi_{+}\left(y\right)}{2}\right ] % \nonumber \label{equ : hms_design}\ ] ] where @xmath44 are the phases of the stipulated fields just above and below the metasurface @xcite . once the general design procedure is established , applying it to the configuration of fig .
[ fig : physical_configuration ] , which includes an electric line source at @xmath45 surrounded by pec walls at @xmath46 , @xmath47 , is straightforward : it reduces to finding the corresponding source spectrum .
the latter is quantized due to the lateral cavity , and includes multiple reflections between the hms at @xmath14 and the pec at @xmath46 @xcite ; explicitly , @xmath48 \delta\left(k_t-\frac{n\pi}{l}\right ) \!\!\!\ !
\end{array } \right\rbrace .
\label{equ : cavity_source_spectrum}\ ] ] we refer to the sum of the fields corresponding to the @xmath49 terms in the summation as the field of the @xmath50th mode of the lateral cavity , where @xmath51 .
although this procedure is applicable for any transmission angle , we restrict ourselves from now on to the case of broadside radiation @xmath52 , where the performance of shielded and unshielded fp - lwas is the most problematic due to the splitting condition @xcite ( design of oblique - angle radiators is addressed in appendix [ sec : oblique_angle ] ) . in this case , the desirable radiation pattern is symmetric with respect to the @xmath53 plane , thus it is only natural to use a laterally - symmetric excitation ; subsequently , we set the lateral position of the source to be @xmath54 throughout the rest of the paper . with this selection , the even modes vanish , and the odd modes follow a cosine profile in the lateral dimension .
one of the key differences between the cavity - excited hms antenna and fp - lwas is that by harnessing the equivalence principle we control the _ individual _ contributions of the electric and magnetic fields to the flow of power , expressed by the lateral distribution of the @xmath55-component of the poynting vector on the aperture .
more specifically , as mentioned in the previous subsection , the resultant ( transmitted ) aperture fields corresponding to eq . actually follow the square root of the _ power _ profile dictated by the cavity mode , and _ not _ the profile of the cavity _ fields_.
this distinction is very important , as the _ power _ profile of a standing wave is always positive , whereas the _ field _ profile changes signs along the lateral dimensions .
hence , the spectral content of the aperture fields , which determines the far - field radiation pattern , is fundamentally different . to illustrate this point , and to elucidate the guidelines for optimizing the cavity excitation
, we compare the fields formed on the device aperture for a shielded fp - lwa , where a standard prs is used , and for a cavity - excited hms antenna with the same excitation .
figure [ fig : mode_analysis_prs_hms ] presents the spatial profile of the tangential electric field , its spatial fourier transform , and corresponding radiation patterns ( calculated following @xcite ) , for single mode excitation of the @xmath56 ( solid blue lines ) , @xmath57 ( dashed red lines ) , and @xmath58 ( green solid lines ) modes , for an aperture length of @xmath59 .
all plots are normalized to their maximum , as the radiation pattern is sensitive to the variation of the fields , and not to their magnitude . as follows from eq . , the spatial profile of the @xmath50th - mode aperture field is proportional to @xmath60 for a standard prs , but for an hms designed according to eq .
, it is proportional to @xmath61 [ fig .
[ fig : mode_analysis_prs_hms](a),(d ) ] . except for the lowest - order mode @xmath56 , for which the two functions coincide
, the difference in the spatial profile translates into distinctively different features in the spectral content [ fig .
[ fig : mode_analysis_prs_hms](b),(e ) ] . for the @xmath50th mode
, the transmission spectrum of the hms aperture corresponds to the autocorrelation of the prs aperture spectrum , leading to formation of peaks centered around the second harmonics ( @xmath62 ) and dc ( @xmath63 ) .
as both the right - propagating and left - propagating components of the standing wave coherently contribute to the dc peak , the latter dominates the transmission spectrum , and the radiation patterns corresponding to the hms aperture exhibit highly - directive radiation towards broadside [ fig .
[ fig : mode_analysis_prs_hms](f ) ] .
in contrast , the prs aperture devices exhibit symmetric ( conical ) radiation to angles determined by the dominant spectral components of the aperture fields @xcite , i.e. towards @xmath64 $ ] [ fig .
[ fig : mode_analysis_prs_hms](c ) ] .
the transverse wavenumber @xmath65 corresponding to the lowest - order mode @xmath56 is small enough such that the two symmetric beams merge @xcite , which enables the prs aperture to radiate a single beam at broadside .
indeed , small - aperture shielded fp - lwas utilize this @xmath66 mode to generate broadside radiation .
however , as demonstrated by @xcite , the aperture efficiency of this mode is inherently limited to about @xmath12 , due to the non - optimal cosine - shaped aperture illumination of the lowest - order mode @xcite , leading to broadening of the main beam [ inset of fig .
[ fig : mode_analysis_prs_hms](f ) ] .
this highlights a key benefit of using an hms - based antenna , as it is clear from fig .
[ fig : mode_analysis_prs_hms](f ) that we can use high - order mode excitations , which provide a more uniform illumination of the aperture , for generating narrow broadside beams with enhanced directivities . in fact , as the index @xmath50 of the mode exciting the hms increases , the autocorrelation of eq . drives the second harmonic peaks outside the visible region of the spectrum [ shaded region in fig .
[ fig : mode_analysis_prs_hms](b),(e ) ] , funnelling all the radiated power to the broadside beam , subsequently increasing the overall directivity .
this improvement in radiation properties can be explained using ordinary array theory . as seen from fig .
[ fig : mode_analysis_prs_hms](d ) , the peaks of the field profile generated by the @xmath50th mode on the hms aperture form hot spots of radiating currents separated by a distance of @xmath67 .
the radiation from such an aperture profile would resemble the one of a uniform array with the same element separation .
as known from established array theory , to avoid grating lobes the element separation should be smaller than a wavelength @xcite . for an aperture length of @xmath68 , where @xmath69 is an integer , the hot spot separation satisfies this condition for mode indices @xmath70 ; specifically , for @xmath71 ( fig . [ fig : mode_analysis_prs_hms ] ) , grating lobes would not be present in the radiation pattern for mode indices @xmath72 . in agreement with this argument , fig .
[ fig : mode_analysis_prs_hms](f ) shows that for @xmath57 grating lobes still exist , while for the highest - order mode @xmath58 they indeed vanish .
excited by a _
single _ mode as a function of the mode index @xmath50 .
solid lines denote the respective radiation characteristics of a uniformly excited aperture @xcite and dash - dotted lines mark the hpbw ( blue ) and directivity ( red ) of _ multimode _ excitation corresponding to the hms antenna of fig .
[ fig : physical_configuration ] with @xmath59 , @xmath73 , and @xmath74.,width=264 ] these observations are summarized in fig .
[ fig : mode_performance ] , where the radiation characteristics of an hms aperture of @xmath59 excited by a single mode are plotted as a function of the mode index @xmath50 ( only fast modes @xmath75 are considered ) . for comparison ,
solid lines denoting the half - power beamwidth ( hpbw ) and directivity values achieved for a uniformly - illuminated aperture of the same size are presented as well .
indeed , it can be seen that the lowest - order lateral mode exhibits the worst performance by far , and the performance improves as the mode index increases .
while the half - power beamwidth saturates quickly , the directivity values continue to increase with @xmath50 until the point in which grating lobes disappear @xmath76 is crossed ; for mode indices @xmath72 the radiation characteristics of the hms aperture are comparable with those of the optimal uniformly - excited aperture . from an array theory point of view ,
excitation of the highest - order mode is preferable , as the corresponding equivalent element separation approaches @xmath77 , implying that such aperture profile would be suitable for directing the radiation to large oblique angles @xmath78 without generating grating lobes @xcite .
another reason to prefer excitation of the highest - order mode in the case of cavity - excited hms antennas is that the hms reflection coefficient @xmath22 grows larger with @xmath79 ; therefore , the power carried by the highest - order mode @xmath80 is best - trapped in the cavity , guaranteeing uniform illumination even in the case of very large apertures .
nevertheless , generating a single - mode excitation of a cavity via a localized source can be very problematic @xcite .
fortunately , the cavity - excited hms antenna can function very well also with multimode excitation , as long as high - order modes dominate the transmission spectrum .
this is demonstrated by the dot - dashed lines in figs .
[ fig : mode_analysis_prs_hms ] and [ fig : mode_performance ] , corresponding to a multimode excitation generated by the configuration depicted in fig .
[ fig : physical_configuration ] with @xmath59 , @xmath73 , and @xmath74 .
as expected from the expression for the source spectrum [ eq . ] , for a given aperture length @xmath10 , the field just below the aperture due to a line source would be a superposition of lateral modes , the weights of which are determined by the particular source configuration , namely the cavity thickness @xmath81 and source position @xmath82 . the multimode transmission spectrum in fig .
[ fig : mode_analysis_prs_hms](b ) indicates that for the chosen parameter values , high - order modes ( @xmath83 ) predominantly populate the aperture spectrum , however low - order modes ( @xmath84 ) are present as well , to a non - negligible extent ( this takes into account the fact that the transmission coefficient @xmath85 is higher for lower - order modes ) .
considering that the far - field angular power distribution @xmath86 is proportional to @xmath87 , the multimode excitation of the prs aperture results in a radiation pattern resembling the one corresponding to single mode excitation of the highest - order mode ( @xmath58 ) but with significant lobes around broadside [ fig .
[ fig : mode_analysis_prs_hms](c ) ] ; consequently , even if a conical beam is desirable , multimode excitation would result in significant deterioration of the directivity .
on the other hand , the same multimode excitation does not degrade substantially the performance of the hms antenna . the autocorrelated spectrum , relevant to the field induced on the hms aperture , results in merging of all spectral components into a sharp dc peak , with the high - order grating lobes pushed to the evanescent region of the spectrum [ fig .
[ fig : mode_analysis_prs_hms](e ) ] .
this retains a beamwidth comparable with that resulting from a single - mode excitation of the highest - order mode , with only slight deterioration of the directivity due to increased side - lobe level [ fig .
[ fig : mode_analysis_prs_hms](f ) and inset ] . continuing the analogy to array theory ,
such multimode excitation introduces slight variations to the magnitude of the array elements , forming an equivalent non - uniform array @xcite . the corresponding multimode hpbw and directivity values
are marked , respectively , by blue and red dash - dotted lines in fig .
[ fig : mode_performance ] , verifying that indeed , cavity - excited hms antennas achieve near - unity aperture efficiencies with a practical multimode excitation ; this points out another key advantage of the cavity - excited hms antenna with respect to shielded fp - lwas . with these observations in hand ,
we are finally ready to formulate guidelines for optimizing the cavity excitation for maximal directivity . for a given aperture length @xmath68 , with respect to eq . , we maximize the coupling to the @xmath80 mode ( which exhibits the best directivity ) by tuning the cavity thickness @xmath81 as to minimize the denominator of the corresponding coupling coefficient ; equally important , we minimize the coupling to the @xmath56 mode ( which exhibits the worst directivity ) by tuning the source position @xmath82 as to minimize the numerator of the corresponding coupling coefficient . to achieve these with minimal device thickness we derive the following design rules @xmath88
although this is somewhat analogous to the typical design rules for ( unshielded ) fp - lwas @xcite , the key difference is that for hms - based antennas we optimize the source configuration _
regardless _ of the desirable transmission angle @xmath5 .
this difference is directly related to the utilization of the equivalence principle for the design of the proposed device , which provides certain decoupling between its excitation and radiation spectrum [ _ cf . _
[ fig : mode_analysis_prs_hms](b),(e ) ] .
this decoupling becomes very apparent when the hms antenna is designed to radiate towards oblique angles @xmath89 , in which case the same cavity excitation yields optimal directivity as well ( see appendix [ sec : oblique_angle ] ) .
two important comments are relevant when considering these design rules .
first , even though following eq
. maximizes the coupling coefficient of the highest - order mode and minimizes the coupling coefficient of the lowest - order mode , it does not prohibit coupling to other modes .
the particular superposition of lateral modes exhibits a tradeoff between bemawidth and side - lobe level ( as for non - uniform arrays @xcite ) .
thus , final optimization of the cavity illumination profile is achieved by fine - tuning the source position @xmath82 for the cavity thickness @xmath81 derived in eq . , with the aid of the efficient semianalytical formulas . in fact , the source position @xmath82 is another degree of freedom that can be used to optimize the radiation pattern for other desirable performance features , such as minimal side - lobe level ; this feature is further discussed and demonstrated in appendix [ sec : reduced_sll ] .
second , although the optimal device thickness increases with increasing aperture length , the increase is sublinear , following an asymptotic square - root proportion factor .
therefore , applying the proposed concept to very large apertures would still result in a relatively compact device , while efficiently utilizing the aperture for producing highly - directive pencil beams .
we follow the design procedure and the considerations discussed in section [ sec : theory ] to design cavity - excited hms antennas for broadside radiation with different aperture lengths : @xmath59 , @xmath90 , and @xmath91 .
the cavity thickness was determined via eq . to be @xmath74 , @xmath92 , and @xmath93 , respectively ; the source position was set to @xmath94 , @xmath95 , and @xmath96 , respectively , exhibiting maximal directivity .
( @xmath97 ) .
the electric response is controlled by the capacitor width @xmath98 of the electric dipole , while the magnetic response is determined by the magnetic dipole arm length @xmath99 ( see appendix [ sec : spider_cells_modelling ] for detailed description).,width=283 ] the required electric surface impedance and magnetic surface admittance modulations are implemented using the `` spider '' unit cells depicted in fig .
[ fig : unit_cell ] . at the design frequency @xmath100 ( @xmath97 ) , the unit cell transverse dimensions are @xmath101 and the longitudinal thickness is @xmath102 .
each unit cell consists of 3 layers of metal traces defined on two bonded laminates of high - dielectric - constant substrate ( see appendix [ sec : spider_cells_modelling ] ) .
the two ( identical ) external layers provide the magnetic response of the unit cell , corresponding to the magnetic surface susceptance @xmath103 , which is tuned by modifying the arm length @xmath99 ( affects magnetic currents induced by tangential magnetic fields @xmath104 ) .
analogously , the middle layer is responsible to the electric response of the meta - atom , corresponding to the electric surface reactance @xmath105 , which is tuned by modifying the capacitor width @xmath98 ( affects electric currents induced by tangential electric fields @xmath27 ) . by controlling the lengths of @xmath99 and @xmath98
, the spider unit cells can be designed to exhibit huygens source behaviour , with balanced electric and magnetic responses ranging from @xmath106 to @xmath107 ( see appendix [ sec : spider_cells_modelling ] ) .
figure [ fig : antenna_results ] presents the design specifications , field distributions , and radiation patterns for the three cavity - excited hms antennas ; table [ tab : antenna_performance ] summarizes the antenna performance parameters ( for reference , parameters for uniformly - excited apertures @xcite are also included ) .
the semianalytical predictions are compared to full - wave simulations conducted with commercially - available finite - element solver ( ansys hfss ) , where the hms was implemented using the aforementioned spider cells ( see appendix [ sec : antenna_simulations ] ) . as demonstrated by fig .
[ fig : antenna_results](a)-(c ) , the realized unit cells are capable of reproducing the required surface impedance modulation , except maybe around large values of @xmath108 ; however , it has been shown that such discrepancies usually have little effect on the performance of huygens metasurfaces @xcite .
l|c c c|c c c|c c c & & & + & full - wave & semianlytical & uniform & full - wave & semianlytical & uniform & full - wave & semianlytical & uniform + + [ cols= " < " , ] & @xmath109 & @xmath110 & @xmath111 + ) . ( a ) hms design specifications @xmath112 ( black solid line ) derived from eq . , and the realized electric surface reactance ( blue circles ) and magnetic surface susceptance ( red circles ) using the spider unit cells .
( b ) radiation patterns produced by semianalytical formalism ( blue dashed line ) and full - wave simulations ( red solid line ) . ( c )
field distribution @xmath113 produced by full - wave simulations .
( d ) semianalytical prediction of @xmath113 @xcite .
, width=151 ]
the spider unit cells depicted in fig .
[ fig : unit_cell ] were defined in ansys electromagnetic suite 15.0 ( hfss 2014 ) with two 25mil - thick ( @xmath114 ) rogers rt / duroid 6010lm laminates ( green boxes in fig .
[ fig : unit_cell ] ) bonded by 2mil - thick ( @xmath115 ) rogers 2929 bondply ( white box in fig .
[ fig : unit_cell ] ) .
the electromagnetic properties of these products at @xmath116 , e.g. permittivity tensor and dielectric loss tangent , as were provided to us by rogers corporation , have been inserted to the model .
specifically , a uniaxial permittivity tensor with @xmath117 and loss tangent of @xmath118 were considered for rogers rt / duroid 6010lm laminates , while an isotropic permittivity of @xmath119 and loss tangent @xmath120 were considered for rogers 2929 bondply .
the copper traces corresponded to @xmath121 oz .
cladding , featuring a thickness of @xmath122 ; the standard value of @xmath123 bulk conductivity was used in the model . to comply with standard pcb manufacturing processes ,
all copper traces were 3mil ( @xmath124 ) wide , and a minimal distance of 3mil was kept between adjacent traces ( within the cell or between adjacent cells ) .
this implies that the fixed gaps between the capacitor traces ( along the @xmath125 axis ) of the electric dipole in the middle layer , as well as between the two arms ( along the @xmath126 axis ) of the magnetic dipole in the top and bottom layer ( fig .
[ fig : unit_cell ] ) , were fixed to a value of @xmath127 ( @xmath124 ) ; the distance from the arm edge to the edge of the unit cell was fixed to @xmath128 ( @xmath129 ) .
unit cells with different values of magnetic dipole arm length @xmath99 and electric dipole capacitor width @xmath98 were simulated using periodic boundary conditions ; hfss floquet ports were placed at @xmath130 and used to characterize the scattering of a normally - incident plane wave off the periodic structure ( the interface between the bondply and the bottom laminate was defined as the @xmath14 plane ) . for each combination of @xmath99 and @xmath98 ,
the corresponding magnetic surface susceptance @xmath131 and electric surface reactance @xmath132 were extracted from the simulated impedance matrix of this two - port configuration , following the derivation in @xcite .
the magnetic response @xmath131 was found to be proportional to the magnetic dipole arm length @xmath99 , with almost no dependency in @xmath98 @xcite . thus , to create an adequate lookup table for implementing , we varied @xmath99 by constant increments , and for a given @xmath99 , plotted as a function of @xmath98 .
the value of @xmath98 for which the two curves intersected corresponded to a balanced - impedance point ( @xmath133 ) , where the unit cell acts as a huygens source , and thus suitable for implementing our metasurface . a lookup table composed of @xmath134 pairs and the corresponding unit cell geometries @xmath135 was constructed , and refined through interpolation .
the interpolated unit cell geometries were eventually simulated again , to verify the interpolation accuracy and finalize the lookup table entries , as presented in fig .
[ fig : spider_cell_lookup_tables ] .
finally , for a given hms with prescribed surface impedance modulation @xmath136 , a corresponding structure could be defined in hfss using the unit cells @xmath137 found via the lookup table in terms of least - squares - error . and @xmath138 .
capacitor width values @xmath139 required for achieving balanced electric and magnetic responses @xmath140 are presented as a function of the magnetic dipole arm length @xmath141 for @xmath142 ( blue open squares ) and @xmath138 ( red open triangles ) radiators , as obtained by finite - element simulations .
the corresponding @xmath140 values are denoted using a blue dashed line for @xmath142 ( @xmath143 ) and using a red solid line for @xmath138 ( @xmath144 ) .
, width=302 ]
to verify our semianalytical design via full - wave simulations , each of the cavity - excited hms antennas designed in this paper was defined in hfss using a single strip of unit cells implementing the metasurface , occupying the region @xmath145 , @xmath146 ( @xmath10 being the aperture length of the antenna ) , and @xmath147 ( in correspondence to the laminate and bondply thicknesses ) .
the simulation domain included @xmath145 , @xmath148 , and @xmath149 ( @xmath81 being the cavity thickness ) , where pec boundary conditions were applied to the @xmath150 planes to form the equivalence of a 2d scenario .
pec boundary conditions were also applied to the @xmath46 plane , and to two @xmath151-thick side - walls at @xmath47 , forming the cavity .
the line - source excitation was modelled by a @xmath152-wide @xmath153 current sheet at @xmath154 , with the current aligned with the @xmath125 axis .
radiation boundary conditions were applied to the rest of the simulation space boundaries , namely @xmath155 , and @xmath156 , allowing proper numerical evaluation of the fields surrounding the antenna . to reduce the computational effort required to solve this configuration
, we utilized the symmetries of our te scenario .
specifically , we placed a perfect - magnetic - conductor ( pmc ) symmetry boundary conditions at the @xmath53 plane , and a pec symmetry boundary conditions at the @xmath157 plane ( the pmc symmetry boundary conditions is only applicable for broadside radiators , _ cf . _ appendix [ sec : oblique_angle ] ) .
we also noticed that adding a thin layer ( @xmath158 ) of copper between the electric dipole edges and the pec parallel - plates at @xmath150 enhanced the convergence of the simulation results . with that minor modification , all of the simulated antennas converged within less than 40 iterations ( maximum refinement of @xmath159 per pass ) , where the stop conditions was 3 consecutive iterations in which @xmath160 .
several assumptions made during the derivation of the hms design formulas contribute to discrepancies between predicted and actual performance of the presented antennas .
first , the predicted fields are derived assuming the hms is capable of implementing continuous surface impedance boundary conditions , with unbound surface impedance values ; nevertheless , the physical implementation requires discretization of the continuous modulation into unit - cells , and the range of achievable surface impedance values is limited ( see fig .
[ fig : spider_cell_lookup_tables ] ) .
second , the hms is assumed to be passive and lossless , however realistic conductors and dielectrics , used for the implementation of the devices in ansys hfss , include unavoidable losses .
third , to facilitate the plane - wave - like relation between the transmitted fields on the aperture [ eq . ] , while still guaranteeing they obey maxwell s equations , we have used the approximation @xmath161 \right\ } \right| \label{equ : sve_1}\ ] ] which is satisfied when @xmath162 which is a refinement of the slowly - varying envelope ( sve ) approximation utilized in @xcite .
this approximation is self - consistent with our design scheme , as when the transmitted fields are directive towards @xmath163 as desirable , the dominant components of the transmission spectrum @xmath164 are in the vicinity of @xmath165 , where the reflection coefficient @xmath166 completely vanish ( the numerator of eq . vanishes ) .
interestingly , the impacts of these three assumptions can be assessed by reviewing the predicted and simulated power flow across the metasurface .
the @xmath55-directed power profiles below ( blue ) and above ( red ) the hms as predicted by the semaianalytical formalism ( open circles and squares , respectively ) and as extracted from full - wave simulations ( dashed and solid lines , respectively ) are presented in fig .
[ fig : power_profiles ] , for the three antennas reported in section [ sec : results ] .
the fact that the general trend and quantitative data of the semianalytical and simulated results compare well ( note that the profiles are plotted using a common @xmath167 unit scale ) , indicates that the first assumption is valid ( this is also supported by @xcite ) .
the semianalytical predictions made based on a homogenized continuous surface impedance boundary conditions mostly agree with the simulation data recorded below and above the metasurface , where effective medium theory predicts discretization effects to be negligible @xcite .
violations of the second assumption , regarding the lossless nature of the hms , would manifest themselves as differences between the simulated power profile below and above the metasurface , which must originate in dissipation in the unit - cell conductors and dielectrics . on the other hand , violations of the third assumption , related to the sve approximation , would manifest themselves as differences in the semianalytically predicted power profile below and above the metasurface , as they correspond to violations of local power conservation @xcite .
local deviations from these two assumptions are found to be rather small , they contribute to a non - negligible reduction of the total power flow across the metasurface ( integrated over the aperture length ) .
the values denoted in the legends of fig .
[ fig : power_profiles ] indicate that according to full - wave simulations about @xmath168 of the power below the hms is dissipated in the lossy conductors and dielectrics , while the semianalytically predictions reveal about @xmath159 discrepancy between the power below and above the metasurface .
while these relative deviations can be considered small albeit non - negligible , it seems that they actually balance each other .
the theoretical derivation assumes and prescribes a lossless hms , but the minor violations of the sve approximation contribute to predicted ( maxwellian ) fields which must be supported by small losses . on the other hand ,
the implemented hms does include realistically unavoidable losses , which turn out to dissipate a comparable amount of power .
we hypothesize that this balance allows overcoming the minor deviations from the theoretical assumptions , facilitating the very good agreement between predicted and simulated results reported herein .
61ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1109/jrproc.1940.228959 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1109/tap.2009.2029277 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) `` , '' ( , , ) chap .
link:\doibase 10.1109/tap.2013.2287296 [ * * , ( ) ] `` , '' ( , , ) chap .
link:\doibase 10.1109/map.2015.2397154 [ * * , ( ) ] `` , '' ( , , ) chap
. link:\doibase 10.1109/jproc.2010.2103530 [ * * , ( ) ] link:\doibase 10.1109/tap.2006.874350 [ * * , ( ) ] link:\doibase 10.1109/tap.2010.2055812 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1109/tap.2014.2377718 [ * * , ( ) ] link:\doibase 10.1109/22.798001 [ * * , ( ) ] @noop * * , ( ) in link:\doibase 10.1109/aps.2010.5560977 [ _ _ ] ( , ) pp . in link:\doibase 10.1109/eucap.2012.6206359 [ _ _ ] ( , ) pp . in link:\doibase 10.1109/eucap.2006.4584495 [ _
( , ) pp . @noop * * ( ) link:\doibase 10.1109/tap.2011.2173133 [ * * , ( ) ] in @noop _ _
( , ) pp .
link:\doibase 10.1109/tap.2011.2173108 [ * * , ( ) ] in link:\doibase 10.1109/aps.2013.6711032 [ _ _ ] ( , ) pp .
link:\doibase 10.1155/2014/479189 [ ( ) ] `` , '' ( , , ) chap . in @noop _ _ ( , ) pp . @noop * * , ( ) in @noop _ _ ( , ) pp .
link:\doibase 10.1103/physrevlett.110.197401 [ * * , ( ) ] link:\doibase 10.1364/oe.21.014409 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevb.89.075109 [ * * , ( ) ] link:\doibase 10.1038/srep04971 [ * * , ( ) ] @noop ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1002/adom.201400584 [ ( ) , 10.1002/adom.201400584 ] link:\doibase 10.1103/physrevx.3.041011 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1109/tap.2014.2354419 [ * * , ( ) ] link:\doibase 10.1109/tap.2003.817560 [ * * , ( ) ] @noop _ _ ( , ) @noop _ _ , ed .
( , , ) `` , '' ( , , ) chap . , ed .
link:\doibase 10.1109/tap.2006.869925 [ * * , ( ) ] @noop `` , '' ( ) , @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1364/ol.39.002719 [ * * , ( ) ] link:\doibase 10.1038/nmat3839 [ * * , ( ) ] link:\doibase 10.1126/science.1253213 [ * * , ( ) ] link:\doibase 10.1364/oe.23.002293 [ * * , ( ) ] link:\doibase 10.1126/science.aaa2494 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.87.347 [ * * , ( ) ] link:\doibase 10.1103/physrevb.84.115421 [ * * , ( ) ] link:\doibase 10.1109/lawp.2013.2293631 [ * * , ( ) ] | one of the long - standing problems in antenna engineering is the realization of highly - directive beams using low - profile devices . in this paper
we provide a solution to this problem by means of huygens metasurfaces ( hmss ) , based on the equivalence principle .
this principle states that a given excitation can be transformed to a desirable aperture field by inducing suitable electric and magnetic surface currents . building on this concept ,
we propose and demonstrate cavity - excited hms antennas , where the is designed to optimize aperture illumination , while the hms facilitates the current distribution that ensures phase purity of aperture fields .
the hms breaks the coupling between the excitation and radiation spectrum typical to standard partially - reflecting surfaces , allowing tailoring of the aperture properties to produce a desirable radiation pattern . |
the production of jets with large transverse momenta @xmath1 in @xmath2 reactions with both photons being quasi - real has been studied at tristan @xcite for quite some time and more recently at lep @xcite .
the presence of a large scale in the process allows perturbative qcd calculations .
these calculations are available at next - to - leading order ( nlo ) in qcd @xcite and the agreement of the predictions with experimental data is good ( see e.g. @xcite ) .
( 161,62 ) ( 0,-8 ) ( 70,-4 ) ( 0,10 ) in order to reliably understand the hadronic structure of the photon also the process should be studied where one of the incoming photons has a large virtuality @xcite .
the scattering of virtual on real photons , which is the analog to deep inelastic electron - proton ( @xmath3 ) scattering , can be achieved in @xmath0-colliders by a single - tag experiment , i.e. , one of the leptons is detected under a certain angle and the other escapes unobserved in the beampipe .
single - tag experiments have recently been performed at the lep storage rings and the results of these measurements are expected in the near future @xcite .
some studies of deep inelastic electron photon jet cross sections in leading - order ( lo ) , i.e. @xmath4 , have been done in @xcite .
a nlo qcd calculation of jet cross sections in @xmath5 collisions for relatively small virtualities up to @xmath6 gev@xmath7 has been performed in @xcite , but without considering the longitudinal photon polarization .
the simplest diagram contributing to the deep inelastic electron photon jet cross section in lo is the quark parton model ( qpm ) diagram , see figure 1 .
the real photon and the virtual photon couple directly to the charge of the bare quark , producing two final state jets .
this so - called direct process can be distinguished from resolved processes , where the real photon serves as a source of partons ( quarks and gluons ) that interact with the virtual photon .
in addition to the high @xmath1 jets from the hard scattering , the resolved contribution is accompanied by a `` spectator jet '' ( or photon remnant jet ) with small @xmath1 .
it has been discussed in @xcite that both contributions are of the same order in the strong coupling constant .
note , that the contribution involving the resolved real photon is traditionally called _ single - resolved _ in @xmath2 scattering . we will call this contribution simply _ resolved _ for ease of writing .
similar to @xmath3-scattering , one motivation for studying deep - inelastic @xmath8-scattering is to obtain information on the partonic , especially gluonic , structure of the real photon from the resolved contribution of the cross section .
furthermore , the interplay between the hadronic and the point - like part of the real photon structure function can be studied .
as is well known , only qualitative conclusions can be drawn from lo calculations , since the results strongly depend on the factorization scale , especially in the resolved contribution .
furthermore , no dependence on jet definitions , present in the experimental data , can be seen in the lo results .
only the nlo cross sections , which contain an additional parton in the final state , depend on the jet definition .
the nlo qcd corrections to the qpm diagram , i.e. , the @xmath9 contributions , consist of the radiation of a gluon in the final state and of the virtual corrections .
the quark - antiquark pair emitted from the real photon can produce a collinear singularity in the real corrections , which has to be factorized and absorbed into the real photon structure function . in this way , the direct and resolved contributions achieve a dependence on the factorization scale @xmath10 .
thus the direct and resolved cross sections are not independent from each other in nlo qcd and the clear distinction mentioned above for the lo case does no longer hold .
the nlo corrections to the subprocesses of the resolved contribution are equal to those obtained in deep inelastic @xmath3-scattering , where the photon interacts directly with the partons from the proton .
the singularities in the resolved contribution that do not cancel after adding the real and the virtual corrections are absorbed into the photon structure function , as for the direct case . in this paper
we will concentrate on the calculation of the @xmath9 terms in the direct part of the cross section for both , the transverse and the longitudinal polarization of the virtual photon . for separating the singular phase - space regions we apply the phase - space slicing method @xcite .
the resolved contributions are taken from the @xmath3-scattering case , which has been calculated using the phase - space slicing method in @xcite .
the paper is set out as follows . in section 2
we discuss cross sections in deep inelastic @xmath8 scattering in general and review the lo cross sections .
we proceed in section 3 by performing the nlo calculation of the direct contribution for both polarizations of the virtual photon .
the techniques for doing the calculations follow essentially those layed out in @xcite .
some results for inclusive single- and dijet cross sections are presented in section 4 for kinematic conditions and energies encountered at lep1 and lep2 .
section 5 contains the summary and an outlook .
the results of the analytic calculations are presented in the appendix .
to fix the notation we start by writing down the process of jet production in @xmath0 scattering : @xmath11 we are interested in the case where one lepton radiates a virtual and the other a real photon .
of course , it does not matter which of the leptons radiates the virtual photon , but for definiteness we suppose this to be the positron .
thus , the subprocess we have to consider is @xmath12 , with @xmath13 , @xmath14 and the virtualities @xmath15 and @xmath16 . the electron - positron center - of - mass ( c.m . )
energy is @xmath17 .
the hadronic , i.e. , @xmath18 , c.m .
energy is @xmath19 .
furthermore we define the variables @xmath20 where @xmath21 and @xmath22 are the energies of the incoming and outgoing electron in the @xmath0 center - of - mass system ( c.m.s . ) , respectively .
the variable @xmath23 gives the momentum fraction of the real photon in the electron .
the cross section @xmath24 for the process described above is given by the convolution @xmath25 here , @xmath26 describes the spectrum of the real photons emitted from the electron according to the weizscker - williams approximation @xcite , which in its simplest form reads @xmath27 the electron mass is @xmath28 and @xmath29 is the maximum scattering angle of the electron .
the function @xmath30 is the parton distribution function ( pdf ) of the real photon which describes the probability to find a parton with momentum @xmath31 inside the real photon , where @xmath32 $ ] .
the direct process is included in formula ( [ xsec ] ) by viewing the photon as a parton with momentum fraction @xmath33 .
it is obtained by summing not only over the quark flavors and the gluon , but also over the photon with @xmath34 .
last , @xmath35 gives the positron - parton cross section , which is given by @xmath36 where @xmath37 is the lepton tensor of the positron . in the resolved case , @xmath38 is the hadron tensor familiar from deep - inelastic @xmath3-scattering . in the direct case ,
@xmath38 stands for the appropriate tensor describing the photon - photon scattering amplitudes ( for a definition see e.g. @xcite ) .
note , that the lepton tensor of the electron radiating the real photon integrated over the azimuthal and polar angle of the outgoing electron gives the weizscker - williams formula ( [ ww ] ) .
the @xmath39-particle phase space of the final state particles of the subprocess is @xmath40ps@xmath41 and the positron phase space is given by @xmath42 the azimuthal angle of the outgoing positron , @xmath43 , can be integrated out in the hadronic c.m.s . using the definitions of the trace part of the hadron tensor @xmath44 ( denoted by _
g - part _ in the following ) , the longitudinal part @xmath45 and the unpolarized part @xmath46 , we find @xmath47 the expression @xmath48 is used here to rewrite the result in a form familiar from dis in @xmath3-collisions .
it should be noted that the real photon is unpolarized .
the longitudinal cross section of the real photon is proportional to @xmath49 and vanishes since @xmath50 .
the lo cross sections are ( per definition ) finite .
however , we will use dimensional regularization for the nlo corrections and thus we state also the lo results in @xmath51 space - time dimensions .
we will neglect all quark masses in the following .
the born approximation for the production of two final state quarks in the direct process is given by the process @xmath52 .
the resolved gluon and quark initiated subprocesses @xmath53 and @xmath54 can be found e.g. in @xcite .
note that the photon induced process @xmath52 is very similar to the one where the real photon is replaced by a gluon .
the photon induced cross section can be obtained from the gluon induced one by keeping only the abelian terms and setting @xmath55 ( hence @xmath56 ) for the virtual photon vertex . in this way also
the nlo corrections can be deduced .
using the usual definitions of the mandelstam variables @xmath57 , @xmath58 and @xmath59 , where @xmath60 and @xmath61 are the momenta of the final state particles , the two - body phase space reads @xmath62 the matrix elements for the qpm process @xmath52 read @xmath63 for the gluon induced resolved matrix elements one finds @xmath64 whereas the quark initiated resolved matrix elements are given by @xmath65 the definitions of @xmath66 and @xmath67 can be found in tab .
[ tab1 ] .
.[tab1]definition of the lo matrix elements @xmath68 and @xmath67 in terms of @xmath69 and @xmath70 , defined in the appendix . [ cols="<,^,^,^",options="header " , ]
the nlo corrections are calculated with the help of dimensional regularization .
the ultraviolet singularities in the one - loop contributions are regularized and subtracted in the modified minimal subtraction ( @xmath71 ) scheme .
the infrared and collinear singularities from the real corrections are likewise calculated in dimensional regularization .
the @xmath72 perturbative qcd corrections to the direct born process @xmath52 are given by the gluon bremsstrahlung process @xmath73 ( see fig . 2 , upper row ) and by the one - loop contributions to this process ( see fig .
2 , lower two rows ) .
( 161,115 ) ( 0,55 ) ( 35,55 ) ( 70,55 ) ( 0,25 ) ( 35,25 ) ( 70,25 ) ( 0,-5 ) ( 35,-5 ) ( 70,-5 ) ( 0,10 ) the nlo corrections to the resolved processes constitute the one - loop contributions to the born terms of section 2.2 and the four three - body contributions @xmath74 the nlo resolved parton cross sections are of order @xmath75 , but , after they have been folded with the pdf s of the real photon , of the same order as the nlo direct cross sections @xcite . in the following subsections
we will consider only the direct part in some detail .
the nlo resolved matrix elements are taken from graudenz @xcite .
we deduce the virtual correction to @xmath52 as mentioned above by keeping the abelian term of the corresponding gluon induced process in order @xmath75 and adjusting the color factors and the coupling .
the one - loop process @xmath54 can be calculated by crossing from the well - known one - loop corrections in @xmath76 @xcite , which has been done in @xcite .
following the above procedure we obtain the results for the g - part and the longitudinal part : @xmath77 the expressions @xmath78 and @xmath79 are stated in the appendix . the two - body phase - space is again given by eqn ( [ 2bpp ] ) .
the real corrections from the gluon bremsstrahlung process contain singular regions of phase - space that lead to two classes of singularities after integration .
the final state gluon can become soft or collinear to one of the other final state particles , which will produce a final state singularity .
due to the on - shellness of the real initial state photon , the splitting term @xmath80 will produce an initial state singularity .
the method used here for handling these regions is the phase - space - slicing method @xcite , where a cut - off @xmath81 is introduced in the phase - space integration to separate finite and singular regions .
the finite integrals are handled numerically which opens the opportunity to introduce experimental cuts and jet definitions in a flexible way .
the singular integrals are done analytically in @xmath51 dimensions .
both , the analytical and the numerical contributions will depend on the slicing parameter .
since @xmath81 is a non - physical parameter the dependence of these parts on @xmath81 should cancel in the sum .
we will show this explicitly for our results later on . after adding the @xmath9 terms only single poles proportional to the photon splitting function @xmath82 will remain .
these are absorbed into the photon structure function .
we now handle the final and initial state singularities separately .
we consider the gluon bremsstrahlung process @xmath83 and concentrate on the case , where e.g. the invariant @xmath84 vanishes and produces a singularity .
the case @xmath85 can be obtained by simply substituting @xmath86 . the analytical integration of the three body contributions @xmath87 and @xmath88 for the process ( [ 3b ] ) over @xmath40ps@xmath89 in the singular phase - space region is done by separating the three - particle phase - space into the two - particle phase - space @xmath40ps@xmath90 and a singular phase - space @xmath40ps@xmath91 . in the c.m.s . of the final state particles
@xmath61 and @xmath92 we define the variable @xmath93 , where @xmath94 is the angle between @xmath95 and @xmath92 , and @xmath96 .
furthermore , @xmath43 defines the angle between the two planes defined by the particles @xmath97 and @xmath98 .
we define the variables @xmath99 from the three - body momenta . for @xmath100 the separation of the phase - space
is given by @xmath101 with @xcite @xmath102^{-{\epsilon}}\ ] ] with the function @xmath103^{-{\epsilon } } = 1 + { \cal o}(r)\ ] ] and the normalization constants @xmath104^{-{\epsilon } } = \frac{{\gamma}^2(1-{\epsilon})}{{\gamma}(2 - 2{\epsilon } ) } , \qquad n_\phi = \int\limits_0^\pi d\phi \sin^{-2{\epsilon}}\phi = \frac{4^{\epsilon}\pi{\gamma}(1 - 2{\epsilon})}{{\gamma}^2(1-{\epsilon } ) } \quad .\ ] ] the two - body phase - space @xmath105 is once again given by eqn ( [ 2bpp ] ) .
defining the variable @xmath106 $ ] , the limits of integration in @xmath107 are given by @xmath108 $ ] , @xmath109 $ ] and @xmath110 $ ] . expressing the @xmath111 matrix elements @xmath87 and @xmath88 by the variables @xmath112 and @xmath43 in the limit @xmath100
, we obtain the approximated matrix elements @xmath113 and @xmath114 .
these are then integrated over the singular region of phase - space : @xmath115 the final results for @xmath116 and @xmath117 are listed in the appendix .
they contain the ir collinear and soft singularities that cancel against those of the virtual corrections .
it is essential that the singular terms are proportional to the lo matrix elements and that @xmath118 and @xmath119 reduce to the usual two - body invariants @xmath120 and @xmath121 for the above limit @xmath100 . in the direct case
the real photon can split into a @xmath122 pair that gives rise to a collinear singularity if the partons are emitted parallel .
the singularity appears when the invariant @xmath123 vanishes .
we define the new variable @xmath124 \quad , \ ] ] that gives the fraction of the momentum @xmath125 that participates in the subprocess after a particle has been radiated in the initial state .
the variable @xmath126 $ ] is connected to @xmath127 through @xmath128 .
furthermore , we define the invariants @xmath129 from the three - body momenta .
all other definitions concerning the phase - space slicing remain as in the previous subsection . in the limit
@xmath130 the variable @xmath131 reduces to @xmath132 . in the same limit
the three - body phase - space separates according to @xmath133 , where @xcite @xmath134 with @xmath135^{-{\epsilon } } = 1 + { \cal o}(z_1 ) \quad .\ ] ] the two - body phase - space is given by equation ( [ 2bpp ] ) .
the integration over @xmath107 with @xmath136 $ ] , @xmath137 $ ] and @xmath110 $ ] is restricted to the singular region of @xmath138 in the range @xmath139 $ ] with @xmath140 $ ] . expressing the matrix elements @xmath87 and @xmath88 with the variables @xmath141 and @xmath43 and taking the limit @xmath130 , one obtains the approximated matrix elements @xmath142 and @xmath143 .
these are integrated according to @xmath144 where @xmath145 and @xmath146 are stated in the appendix .
the full nlo cross section @xmath147 can be written as a sum of the @xmath81-dependent two- and three - body cross sections @xmath148 and @xmath149 : @xmath150 the @xmath81-dependence in @xmath147 cancels up to terms of order @xmath151 which have been neglected in the analytical calculations .
thus , @xmath81 has to be chosen sufficiently small for the approximations to be valid .
the two - body cross section consists of the final state , virtual , initial state and born contributions @xmath152 , @xmath153 , @xmath154 and @xmath155 : @xmath156 when the virtual and final state cross sections are added , the pole terms cancel , as can be seen by directly comparing formul ( [ vg ] ) and ( [ vl ] ) with ( [ final ] ) from the appendix .
the simple pole appearing in the initial state singularities @xmath145 and @xmath146 is proportional to the splitting function @xmath157 .
this function appears in the evolution equation of the pdf of the real photon as an inhomogeneous ( point - like ) term .
therefore , the photon initial state singularities can be absorbed into the real photon pdf , according to the procedure given in @xcite .
we define the renormalized pdf @xmath158 of a quark @xmath159 in the electron as @xmath160 where @xmath161 is the lo pdf before the absorption of the collinear singularity .
the nlo transition function is given by @xmath162 with @xmath163 in the @xmath71 scheme .
in the discussed order , @xmath164 .
the initial state contribution @xmath165 for the photon initial state singularity is calculated from the unrenormalized cross section @xmath166 by @xmath167 the cross section @xmath155 contains the lo photon - parton scattering matrix elements given in section 2 , .
the factor @xmath168 in eqn ( [ nlo - trans ] ) is combined with the factor @xmath169 in eqn ( [ int22 ] ) and leads to a factorization scale dependent term of the form @xmath170 = - p_{q\leftarrow { \gamma}}(z_b ) \ln \left ( \frac{m_{\gamma}^2}{q^2}\right ) + { \cal o}({\epsilon } ) \quad .\ ] ] in this way , the subtracted partonic cross section will depend on the scale @xmath171 , as does the pdf of the resolved photon .
after this subtraction procedure , all terms in the two - body cross section @xmath148 are finite and the limit @xmath172 can be performed .
eqn ( [ xsec2 ] ) then yields a physically meaningful cross section .
we now come to a discussion of numerical results for single- and dijet inclusive jet cross sections . before analyzing these physical cross sections , we discuss some checks of the analytical results , one being the dependence of the cross sections on the phase - space slicing parameter @xmath81 . furthermore we show the dependence of the cross sections on the factorization scale .
the multi - dimensional integrals are performed using the vegas @xcite monte carlo integration routine . at lep
, electrons are scattered on positrons with equal energy .
events are selected in a certain angular range of the tag and rejected if there is also an antitag above a certain energy and scattering angle .
this corresponds to the situation , where one photon is quasi - real and the other has a finite , non - vanishing virtuality . in this paper
we will give predictions at lep1 and lep2 energies for certain characteristic values of the photon virtuality @xmath173 , namely @xmath174 and @xmath175 gev@xmath7 .
at lep1 , the electron and the positron both have the energy @xmath176 gev , corresponding to a c.m.s .
energy of @xmath177 gev .
the energies of the two incoming leptons at lep2 are @xmath178 gev , giving a c.m.s .
energy of @xmath179 gev .
the antitag conditions are taken into account by using the weizscker - williams formula ( [ ww ] ) with @xmath180 milliradians in the case of lep1 and with @xmath181 milliradians for lep2 .
we have not imposed cuts on @xmath182 and @xmath23 , i.e. , @xmath183 $ ] .
we will calculate the cross sections in the hadronic ( @xmath18 ) c.m.s . for the theoretical studies here
we will refrain from taking into account the exact tagging conditions from the lep experiments @xcite and will concentrate on the above stated energies , although also higher energies will become available at lep2 .
the jets are clustered at lep using the snowmass cone algorithm @xcite , where hadrons @xmath184 are combined into a single jet , if their distance from the jet center in azimuth - rapidity space , given by @xmath185 is smaller than a cone radius @xmath186 , i.e. , for @xmath187 . here ,
@xmath188 and @xmath189 are the rapidity and the azimuthal angle of the combined jet respectively , defined as @xmath190 the radius @xmath186 is chosen as in the experimental analysis , namely @xmath191 . for the parton densities of the photon we use the nlo parametrization of glck , reya and vogt @xcite , transformed from the dis@xmath192 to the @xmath71 scheme .
the corresponding @xmath193 value is @xmath194 mev .
since the energies are quite large , we will choose @xmath195 flavors for our computations .
the renormalization and factorization scales are set equal to @xmath196 .
in particular for the larger values of @xmath173 this is a slightly better choice than @xmath197 , since the minimum transverse energy of @xmath198 gev is rather small .
we could have chosen @xmath197 as a scale , especially for the @xmath197-spectra , but we prefered to take one scale consistently throughout the work .
the choice of scale only has a marginal effect on the results presented in the next sections . from section 3.3
it is already clear that the poles encountered in the real corrections cancel against those from the virtual corrections , which is a test of the singular terms in the nlo corrections .
the remaining photon initial state singularities have been absorbed into the structure function of the real photon . as a further test
we have checked that the analytic formul given in the appendix reduce to the expressions encountered in @xmath2 scattering ( photoproduction limit , i.e. , @xmath199 ) stated in @xcite .
a test of the @xmath81 dependent terms is the calculation of the two - body and three - body cross sections as a function of @xmath81 .
the dependence should cancel in the sum of the two contributions . in the analytical calculation , terms of order @xmath200
have been omitted and thus the slicing parameter , which is normalized with the hadronic c.m.s .
energy , has to be chosen much smaller than one . in fig . 3 a and b we show the single inclusive direct jet cross section integrated over @xmath201 gev as a function of @xmath81 , integrated over the two different @xmath173-regions @xmath202 $ ] gev@xmath7 and @xmath203 $ ] gev@xmath7 for the above stated lep1 conditions .
the analytic two - body contributions ( dashed ) are negative and the absolute value is quite large for the smaller values of @xmath81 .
the three - body contributions ( dotted ) are large and positive .
the sum of these ( full ) is independent of the slicing parameter in the shown region of @xmath81 , as expected .
although it is theoretically safer to choose smaller values of @xmath81 , because of the omission of the @xmath204-terms , it is from the practical ( numerical ) point of view convenient to have larger values of @xmath81 , since the compensation between the two - body and the three - body parts are then smaller and the statistical errors are likewise smaller
. we will choose @xmath205 in the further calculations . in fig .
3 we also compare the nlo cross section with the lo result ( the dash - dotted line ) , which does not depend on @xmath81 . for the lo cross sections we have chosen the same two - loop @xmath206 formula and the same pdf as in the nlo cross sections , so differences between the lo and nlo results are due to differences in the lo and nlo matrix elements .
the nlo curves are smaller than the lo order ones and the corrections are around 10% for the smaller @xmath173 region . for the larger @xmath173 region the nlo corrections reduce the born result by approximately 30% . the resolved contributions are also independent of @xmath81 , which we explicitly checked but do not show here .
similar tests of the resolved contribution have already been performed for the case where the real photon is substituted by a proton in @xcite .
having checked the pole terms , the limiting behavior and the @xmath81-independence , which shows that the program for the numerical calculations is reliable , we proceed by discussing the dependences of the cross sections on the renormalization- and factorization - scales in lo and nlo .
this has been done before in the limiting case of @xmath199 , i.e. , for @xmath2-scattering , in @xcite . in fig .
4 a and b we show the variation of the nlo direct and lo resolved cross sections with the factorization scale @xmath10 ( normalized with @xmath196 ) for the same kinematic conditions and regions of virtuality as in fig . 3 .
in both regions , the lo resolved cross section ( dotted line ) grows by about 50% , going from @xmath207 to @xmath208 .
this dependence is due to the factorization scale dependence of the photon structure function alone , since the lo partonic cross sections are finite and no initial state singularities have to be treated .
the dependence of the lo resolved contribution is completely canceled by the nlo direct contribution , shown as the dashed line .
the sum ( full line ) shows no dependence on the factorization scale in the plotted region .
the dependence of the nlo direct contribution on the factorization scales comes in through the absorption of the singular terms in the initial state into the photon structure function , as explained in section 3.3 .
the resolved nlo corrections to the partonic cross section do depend on the factorization scale , in contrast to the resolved lo partonic cross sections .
one sees however only a small overall dependence on the factorization scale of under 4% by going from @xmath207 to @xmath208 , when the nlo direct and resolved contributions are added to obtain the full nlo cross section ( we do not show these plots here ) .
the renormalization scale dependence of the resolved cross section is reduced by going from lo to nlo , as has been shown in @xcite . for the direct cross section ,
the lo result is independent of the renormalization scale , because the strong coupling constant @xmath206 does not appear in the qpm diagram .
thus , the nlo cross section is actually lo in the strong coupling .
as we will see later on , the nlo correction to the lo direct process is small and thus the renormalization scale dependence , which is only seen in the nlo corrections , will likewise be small .
in addition , the running of the strong coupling is compensated by including the virtual graphs in the nlo direct process .
we found only a small variation of the nlo direct cross section of about 10% between the smallest and largest scale considered here .
thus , summing the direct and resolved contributions we find a reduced renormalization scale dependence due to the reduced dependence of the resolved components .
we start this section by discussing the single - jet inclusive @xmath209-distributions @xmath210 in the region @xmath211 , integrated over @xmath201 gev for lep1 conditions .
we show the direct and resolved cross sections and their sum in fig . 5 and look at the longitudinal and transverse contributions to the individual direct and resolved components in fig .
6 and fig . 7 .
5 a , b , c and d show the direct ( dashed ) and resolved ( dotted ) contributions as well as their sum ( full line ) for the four @xmath173 values defined above , @xmath174 and @xmath175 gev@xmath7 .
the direct contributions are peaked at @xmath212 and are quite symmetric , which is to be expected , since we are in the @xmath18 c.m.s .
a small asymmetry arises from the mass of the virtual photon .
the resolved cross section is peaked at @xmath213 and shows a stronger asymmetry .
this is due to the additional integration over the momentum fraction of the parton in the resolved photon .
the peak is in the positive @xmath209-region , since the virtual photon is chosen to travel along the positive @xmath214-axis .
four all @xmath173 values , the direct contribution is the dominant one , although the resolved becomes more important for larger @xmath173 values and in general for positive @xmath209 s .
this shows that the direct contribution falls of slightly stronger with rising @xmath173 compared to the resolved contribution .
the reason for this is the relatively low cut on @xmath197 , where the resolved component is expected to be more important . in fig .
6 a , b , c and d we show the direct cross section , split up into the transverse and longitudinal parts , compared to the full lo cross section .
the transverse part ( dashed line ) is denoted by @xmath215 and is obtained from the full cross section by setting @xmath216 , so only the g - part is in @xmath215 .
likewise , the longitudinal part @xmath217 ( dotted line ) is obtained by setting @xmath218 .
the sum of both gives the full direct contribution , denoted here as @xmath219 ( full line ) .
for the plots a and b , the transverse part is dominant , although the longitudinal part is important , especially in the central region .
one observes a plateau around @xmath220 for the transverse part , which becomes a dip for the two larger @xmath173-values in fig .
6 c and d. this local minimum is compensated in the full direct cross section @xmath219 through the longitudinal component which has a maximum around @xmath212 . from this one
sees that the longitudinal polarization of the virtual photon gives an important contribution to the direct cross section .
the nlo direct cross section @xmath219 is compared to the lo cross section ( dash - dotted line ) , containing both polarizations of the virtual photon . in fig . 6
a and b , the lo and nlo cross sections can not be distinguished in the logarithmic plot , but the nlo curves lie around 5 - 10% lower than the lo ones , which is consistent with the results shown fig . 3 .
for the two larger @xmath173-values , the nlo curve lies approximately 25% below the lo cross section in the central @xmath209-region .
furthermore , the nlo cross section has a slightly stronger fall - off towards the edges of the @xmath209 phase space than the lo cross section .
the dip observed in the nlo transverse cross section is already present in lo , which we do not show here .
it is not present in photoproduction and for the smaller @xmath173 values . in
lo the reason for the local minimum at larger @xmath173 is twofold .
first , the interference term in the born matrix elements @xmath221 is negative and sharpley centered around @xmath220 . for small @xmath173
this contribution is negligable , but going to larger virtualities the subtraction of the interference term in the central @xmath209-region leads to the observed local minimum .
second , the individual terms @xmath222 and @xmath223 in the born matrix elements are peaked on opposite sides , slightly away from @xmath220 . for @xmath224
their sum still gives a curve with an absolute maximum at @xmath220 . at larger
@xmath173 values though the maxima of the terms @xmath222 and @xmath223 move to larger ( smaller ) @xmath209-values , so that their sum leads to a slight local minimum near @xmath220 .
these effects , observed in lo , have their correspondences in nlo .
next , we consider the resolved components , again separated into transverse and longitudinal parts , in fig .
7 a , b , c and d. the curves are labeled as in fig .
the transverse part of the cross section is dominant for all @xmath173 values and the longitudinal part even becomes less important for the larger virtualities and is one order of magnitude smaller compared to the transverse part at @xmath225 gev@xmath7 .
the longitudinal part is peaked more in the positive @xmath209 region for the smaller @xmath173 values than the transverse part and it is shifted to the smaller @xmath209 s for the larger @xmath173 values . here , also the lo curves and the nlo curves are not distinguishable in the logarithmic plots of fig . 7 a and b. for the resolved cross section ,
the nlo cross sections are around 10% larger than the lo cross sections , in contrast to the direct case , where the nlo cross sections lie lower than the lo cross sections .
it should be mentioned that the transverse part of the gluon induced resolved cross section also shows a dip near @xmath220 , for the same reasons as in the case of the transverse direct cross section .
however , the gluon induced process gives only a rather small contribution to the resolved cross section and the transverse part of the quark induced cross section does not have a local minimum in the @xmath209-distribution , since here the interference term is positive . in connection with the @xmath197-distributions it will turn out to be important that the direct cross sections have a larger plateau in the central @xmath209-region than the resolved components , which are peaked more sharply around @xmath213 and fall off stronger towards the edges of the @xmath209 phase space .
the effect is stronger for the larger @xmath173 values .
this is due to the special form of the transverse direct cross section , which is relatively broad already for the smaller @xmath173 values and has two maxima at the edges of the @xmath209 spectrum for the larger @xmath173 values .
we further mention that the @xmath209 distributions become narrower when integrated over @xmath226 for larger values of @xmath227 due to the stronger kinematical restrictions in the region of larger @xmath197 .
this holds for both , the direct and the resolved components .
we now come to the @xmath197 distributions of the single - jet inclusive cross section @xmath228 for the same @xmath173-values as in fig .
the rapidity is integrated over the central region @xmath229 in the hadronic c.m.s .
we show the nlo distributions of the direct ( dashed ) and the resolved ( dotted ) component of the cross section and the sum ( full line ) of the two in the @xmath197-range @xmath230 $ ] gev in fig . 8
a , b , c and d. as can be seen , the direct component is the dominant one in the whole @xmath197-range for the two smaller @xmath173-values .
in addition the resolved contribution falls off stronger with rising @xmath197 than the direct component .
the stronger fall - off of the resolved component holds for the two larger @xmath173-values as well , but here the resolved component becomes comparable to the direct contribution for the smallest @xmath197 values .
the flattening of the @xmath197 distribution of the direct component for the small @xmath197 s , already visible at @xmath231 gev@xmath7 , even leads to a crossing of the direct and resolved cross sections at @xmath232 gev for the largest @xmath173-value , fig . 4d , so that the resolved is larger than the direct .
this behavior is a purely kinematical effect and stems from the fact that the @xmath209-integration is restricted to @xmath229 .
we have mentioned above that the direct cross sections have a broader @xmath209-spectrum than the resolved ones .
cutting on @xmath209 thus leads to a stronger cut on the direct than on the resolved components .
the effect becomes more important for the larger @xmath173 values because the direct contributions have an even broader @xmath209-spectrum for @xmath231 and @xmath175 gev@xmath7 than for the smaller @xmath173 values , as already discussed above .
since the @xmath209-spectra become narrower for the larger @xmath197 s , the cut on @xmath209 does no longer have such a large effect on the direct cross section and the direct component again dominates over the resolved component for the larger @xmath197 s . as a main result we find that the resolved contribution is small compared to the direct contribution and therefore there is little hope to learn about the parton distributions , especially the gluon distributions , in the photon from deep - inelastic @xmath8-scattering .
furthermore , the resolved component is largely suppressed for larger transverse energies @xmath197 , which agrees with the expectation that the point - like part of the real photon is dominant at large scales .
we now come to the presentation of inclusive dijet cross sections for lep1 and lep2 energies . in fig .
9 a , b , c and d we show the dijet cross section @xmath233 integrated over @xmath234 $ ] as a function of the transverse momentum of the trigger jet @xmath235 for the same @xmath173-values as in fig . 5 ( a)(d ) for lep1 conditions .
we see much the same behavior as in the case of the single - jet cross sections , namely that the direct component is dominant and the resolved one falls off stronger with rising @xmath197 . as in the single - jet case
, the direct cross section has a flattening towards the smaller @xmath197 s for the largest @xmath173 values , which seems to be even stronger in the two - jet case .
the reasons for the flattening are the same as discussed above in the single - jet case , only here we have a cut on the @xmath209 s of both jets , which increases the effect of the @xmath209-cuts and thus the flattening of the @xmath197 distributions . in fig .
10 a , b , c and d the dijet cross section integrated over @xmath234 $ ] is shown in the range @xmath236 $ ] gev for lep2 conditions , i.e. , for slightly different antitagging conditions than for lep1 and for larger energies .
therefore , the cross sections are larger than for the dijet cross sections for lep1 .
all other conclusions remain unchanged .
the phenomenon of the flattening turns out to be quite obvious here . as can be seen in fig .
10 d , the direct contribution even has a slight maximum in the @xmath197 spectrum for @xmath237 gev and by going to smaller @xmath197 values the direct cross section becomes smaller .
this is compensated by the rise in the resolved component , so that the sum of the direct and the resolved component rises by going to smaller @xmath197 s .
we have presented the calculation of the direct component of jet production in deep inelastic electron - photon scattering in nlo qcd .
transverse and longitudinal polarizations of the virtual photon have been taken into account .
the singular regions of phase space have been extracted with the help of the phase - space slicing method .
the initial state singularities of the real photon have been absorbed into the pdf of the real photon .
cross sections have been obtained by adding the direct and resolved real photon parts .
the spectrum of the real photon has been approximated by the weiscker - williams formula .
we have shown that the scale dependence of the nlo cross sections are reduced .
we are thus in the position to compare nlo calculations to experimental data for deep inelastic @xmath8 scattering , taken e.g. at lep1 and lep2 .
we have presented @xmath209- and @xmath197-distributions of inclusive one- and two - jet cross sections for lep1 and lep2 energies for photon virtualities between @xmath6 and @xmath175 gev@xmath7 .
the direct component gives the dominant contribution to the cross section .
the nlo direct cross sections are about @xmath238% smaller than the lo cross sections for @xmath239 gev@xmath7 and about the same size for @xmath173 around @xmath238 gev@xmath7 .
the longitudinal part of the cross section plays an important role and can contribute up to 50% in the full cross section .
it is of course possible to study the influence of different parametrizations of the real photon pdf on the resolved cross section .
we have refrained from showing such comparisons here , because the resolved plays only a minor role in the cross sections .
furthermore it is possible to study the effects of cuts on other observables , such as @xmath240 or the invariant jet mass , in nlo , as has been done for @xmath2-scattering in @xcite .
however , data on these kind of variables are not expected in the near future .
it has been discussed recently in @xcite in connection with @xmath3-scattering at hera that contributions from the resolved virtual photon might contribute to jet cross sections up to virtualities of @xmath241 gev@xmath7 .
the resolved virtual photon contributions have to be taken into account when terms of the type @xmath242 become large and have to be absorbed into a virtual photon structure function , as is the case for @xmath243 .
the subtraction procedure has been worked out in @xcite for the case of @xmath3-collisions and is very similar for @xmath8-reactions @xcite . in this paper
we have not taken into account the effects of a resolved virtual photon .
however , the case of @xmath3-scattering corresponds to the small resolved contribution in the @xmath18 case .
the resolved virtual photon effects in the direct cross section will presumably be small for larger @xmath173 values and it is not clear how the double - resolved contribution , i.e. , the contribution where the real and the virtual photons are both resolved , behaves .
we will extend our studies presented here in the future to include the effects of resolved virtual photons .
i am grateful to g. kramer for discussions and comments on the manuscript .
i thank e. mckigney , d.j . lauber and d. miller for explaining to me some details of the lep single - tag experiment .
the mandelstam variables @xmath120 and @xmath121 , as well as the invariants @xmath244 and @xmath182 are defined as in section 2 in the main part . the lo matrix elements in @xmath51
space - time dimensions are given by @xmath245 the one - loop contributions to the process @xmath52 depend on the two - body variables @xmath120 and @xmath121 .
the function @xmath246 appearing in the virtual corrections is defined as @xcite @xmath247 + \frac{\pi^2}{6 } \quad , \end{aligned}\ ] ] where @xmath248 is the dilogarithm function . for the g - part of the virtual corrections we find @xmath249 \nonumber \\ & + & 4\ln \frac{s}{q^2 } \left ( \frac{2s}{u+t } + \frac{s^2}{(u+t)^2 } \right ) \nonumber \\ & + & \ln \frac{-u}{q^2 } \left ( \frac{4s+2u}{s+t } - \frac{ut}{(s+t)^2 } \right ) + \ln \frac{-t}{q^2 } \left ( \frac{4s+2t}{s+u } - \frac{ut}{(s+u)^2 } \right ) \nonumber \\ & - & 2l(-s ,- u ) \frac{s^2+(s+t)^2}{ut } - 2l(-s ,- t)\frac{s^2+(s+u)^2}{ut } \nonumber \\ & + & \left ( \frac{4s}{u+t } + \frac{s}{u+s } + \frac{s}{s+t } \right ) - \left ( \frac{s}{u}+\frac{s}{t}+\frac{u}{t}+\frac{t}{u } \right ) \quad . \label{vg}\end{aligned}\ ] ] for the longitudinal part we find @xmath250 \quad . \label{vl}\end{aligned}\ ] ] the final state corrections to the process @xmath73 depend on the invariant mass cut - off @xmath251 and on @xmath120 and @xmath121 . neglecting terms of order @xmath252
, we find @xmath253 the longitudinal correction in the final state , @xmath117 , is simply obtained by replacing the born term @xmath254 by @xmath70 .
the photon initial state singularity depends on the cut - off parameter @xmath255 , the additional variable of integration @xmath127 and on @xmath120 and @xmath121 .
again , terms of order @xmath252 have been neglected : @xmath256 the longitudinal term , @xmath146 , can be obtained by replacing the @xmath254 by @xmath70 , as for the final state corrections , keeping the invariants @xmath257 and @xmath132 exchanged .
the altarelli - parisi splitting function in ( [ a1 ] ) is given by @xmath258 | a complete next - to - leading order qcd calculation of deep inelastic electron - photon scattering including direct and resolved real photon components is presented .
soft and collinear singularities are extracted using the phase space slicing method .
application of the results for the prediction of single- and dijet cross sections at the lep @xmath0 collider are presented , using the snowmass jet definition .
the dependence of the cross sections on the transverse momentum and on the rapidities of the jets are discussed . |
wx cen was initially identified by @xcite as a possible optical counterpart of the hard x - ray transient source cen xr-2 , although this identification was latter discarded .
because of its spectral characteristics , the object was then classified as a wolf - rayet star of type wn 7 @xcite , while @xcite classified it as a nova - like variable .
ds95 showed that it is a binary system with an orbital period of 10.0 hr .
the photometric orbital variation , determined by these authors , based on their spectroscopic observations , is approximately sinusoidal with an amplitude of @xmath16 mag . in spite of the period being relatively long ( in the context of cataclysmic variables ) it was not possible to identify any spectral signature of the secondary star .
a distance of 1400 pc was determined from the na i d line equivalent width , and an @xmath17 was suggested from the average of three distinct determinations ( 4430 dib and na i d equivalent widths and he ii 4686 , 10124 lines ratio ) .
doppler tomography of wx cen produced by ds95 suggests that the gas of the secondary has normal chemical abundance , while the wind of the primary star is probably over - abundant in helium ( he@xmath18/h = 0.56 ) .
ds95 also showed that , if the primary component of wx cen is a white dwarf , then the secondary should have a mass bellow @xmath19 and , therefore , be evolved . if , however , one assumes the hypothesis that the secondary is a main sequence star , it would have @xmath20 , and the primary , a mass of @xmath21 .
@xcite included wx cen in a group of 4 galactic binaries defined as the v sge stars .
the other three objects are v sge @xcite , v617 sgr @xcite and di cru @xcite .
they are characterized by the presence of strong emission lines of o vi and n v. besides , he ii is at least two times more intense than h@xmath22 .
the v sge stars are similar to the compact binary supersoft sources ( cbss ) , seen in the magellanic clouds , but not that frequent in the galaxy .
cbsss are interpreted as suffering hydrostatic hydrogen nuclear burning on the surface of a white dwarf .
this burning is due to the high mass - transfer rate , a consequence of the fact that the systems have inverted mass ratios ( see @xcite for a revision and references ) . @xcite also considered the hypothesis of galactic cbss nature for the stars v sge , t pyx and wx cen , established on their extremely blue colors , high luminosities , orbital light curves and highly excited emission line spectra . in an observational program to search for galactic wr stars
, @xcite rediscovered wx cen as a new wn3 , when this star received the wr 48c designation . in our search for v sge type stars
, we selected wr 48c as a candidate , on the basis of spectroscopic criteria , for a detailed observational study .
we , then , realized that wr 48c was the already known star wx cen .
, width=317 ] , width=317 ]
we observed wx cen photometrically for 9 nights on 2000 april and 2003 may and july with the boller & chivens 60 cm and zeiss 60 cm telescopes at the pico dos dias observatory mct / lna located in brazpolis , southeast brazil ( see table [ jophoto ] ) .
a thin back - illuminated eev ccd 002 - 06 chip and a wright instruments
thermo - electrically cooled camera were employed to obtain the images .
the timing was provided by a global positioning system ( gps ) receiver .
the frames were obtained through the johnson @xmath23 filter , with exposure times of 40 , 50 and 80 seconds .
we also observed this star in white light with 2 seconds exposure time , for which we used the same ccd , operated in the frame transfer mode , resulting in dead time between integrations of the order of milliseconds .
bias and flatfield exposures were obtained for correction of the stellar images .
the images , except the ones obtained in frame transfer mode , present an overscan region used for the ccd s zero point correction .
the data reduction was performed with the standard procedures , using iraf routines .
the star is located in a quite rich field ; we , therefore , carried out differential psf ( point spread function ) photometry of wx cen and four comparison stars of the field using the lcurve package , written and kindly provided by m.p .
it makes use of daophot routines to treat automatically long time series data .
@lccccl@ name & _ v _ & _ b - v _ & _ e(b - v ) _ & d ( kpc ) & sp + + + stock 16 & ... & ... & 0.49 & 1.64 & ... + ruprecht 107 & ... & ... & 0.46 & 1.44 & ... + collinder 271 & ... & ... & 0.29 & 1.17 & ... + basel 18 & ... & ... & 0.51 & 2.23 & ... + ngc 4815 & ... & ... & 0.81 & 3.08 & ... + trumpler 21 & ... & ... & 0.20 & 1.26 & ... + bh 144 & ... & ... & 0.60 & 9.35 & ... + hogg 16 & ... & ... & 0.41 & 1.58 & ... + collinder 272 & ... & ... & 0.47 & 2.04 & ... + pismis 18 & ... & ... & 0.50 & 2.24 & ... + + + 114800 & 8.00 & 0.12 & 0.35 & 1.17 & b2 iv ne + 114737 & 8.01 & 0.17 & 0.46 & 2.00 & o9 iv + 114886 & 6.87 & 0.12 & 0.40 & 1.23 & o9 iv + 114478 & 8.68 & 0.49 & 0.65 & 3.21 & b1 ib + 115034 & 8.80 & 0.09 & 0.35 & 1.90 & b0.5 vn + 114122 & 8.57 & 0.59 & 0.73 & 3.10 & b1 iab + 115363 & 7.79 & 0.61 & 0.74 & 3.20 & b1 ia + 113754 & 9.50 & 0.63 & 0.90 & 4.43 & o9.5 ia + cp-623111 & 10.20 & 0.19 & 0.44 & 2.69 & b1 v + cp-613512 & 10.03 & 0.35 & 0.58 & 3.34 & b1 iii + 115071 & 7.96 & 0.24 & 0.50 & 1.03 & b0.5 vn + 113432 & 8.86 & 0.80 & 0.91 & 2.35 & b1 1b + cp-632544 & 9.97 & 0.49 & 0.67 & 5.64 & b1 ib + cp-613598 & 10.3 & 0.84 & 0.96 & 6.40 & b5 iap + 113411 & 9.05 & 0.46 & 0.71 & 2.72 & o9.5 iv + wx cen & 13.6/14.2 & 0.4 & 0.63 & @xmath24 & wn3h pec + | we confirm the orbital period of wx cen @xmath0 wr 48c determined by @xcite
ds95 and refined its value to @xmath1 d. the light curve of this object has a peak to peak variation of approximately 0.32 magnitudes .
it is non - sinusoidal in the sense that it has a v - shaped narrow minimum , similar to the ones seen in v sge , v617 sgr and in compact binary supersoft sources cbss .
most of the emission lines in the optical spectrum are due to balmer , he ii , c iv , n v , o v and o vi .
an analysis of the he ii pickering series decrement shows that the system has significant amount of hydrogen .
the emission lines of he ii 4686 became weaker between the 1991 and 2000/2002 observations , indicating distinct levels of activity .
the spectra of wx cen show variable absorption features in the balmer lines with @xmath2 km s@xmath3 and in emission with @xmath4 km s@xmath3 .
these highly variable events remind the satellites in emission of cbss .
we estimate the color excess as @xmath5 on the basis of the observed diffuse interstellar band at 5780 .
given the distance - color excess relation in the direction of wx cen , this implies a distance of @xmath6 kpc .
interstellar absorption of the na i d lines show components at @xmath7 km s@xmath3 , which corresponds to the velocity of the coalsack , and three other components a @xmath8 , @xmath9 and @xmath10 km s@xmath3 .
these components are also seen with similar strengths in field stars that have distances between 1.8 and 2.7 kpc .
the intrinsic color of wx cen is @xmath11 and the absolute magnitude , @xmath12 .
extended red wings in the strong emission lines are seen .
a possible explanation is that the system has a spill - over stream similar to what is seen in v617 sgr .
we predict that when observed in opposite phase , blue wings would be observed .
a puzzling feature that remains to be explained is the highly variable red wing ( @xmath13 km s@xmath3 ) of the o vi emission lines as well as of the red wings of the h and he lines .
the velocity of the satellite - like feature is consistent with the idea that the central star is a white dwarf with a mass of @xmath14 . with the high accretion rate under consideration ,
the star may become a sn ia in a time - scale of @xmath15 years .
[ firstpage ] stars : individual : wx cen stars : binaries : close stars : emission - line . |
there is convincing evidence from recent radio and optical observations that the diffuse interstellar medium ( ism ) exhibits significant subparsec - scale variations down to limits of a few au ( frail _ et al_. 1994 ; davis _ et al_. 1996 ; meyer & blades 1996 ; watson & meyer 1996 ) . on somewhat larger scales , strong variations in interstellar profiles
are observed towards multiple late - type stars in globular clusters ( langer , prosser , & sneden 1990 ; bates et al .
1995 ) . in some cases
, the inferred densities of these structures far exceed the nominal diffuse cloud values and approach those of molecular cloud cores .
for example , frail _ et al_. ( 1994 ) found pervasive variations in the opacity on scales of 5 - 100 au utilizing multiepoch observations of 21-cm absorption toward high velocity pulsars , which implied densities of @xmath1@xmath2@xmath3 .
it is not clear how such small , dense structures can arise or be maintained in the low pressure environments of diffuse clouds .
one difficulty with using the d lines as a tracer of small scale structure is the increasing saturation of these lines as one goes to larger column densities .
thus the bulk of the gas in heavily reddened sightlines can not be sampled using the d lines , and one is generally limited to identifying variations in the weaker wings of the line .
a potential probe of these sightlines is the @xmath47698 interstellar line .
the lower cosmic abundance of k coupled with differences in the photoionization cross section leads to columns that are typically a factor of @xmath550 below those of in diffuse interstellar clouds ( hobbs 1974 ) .
thus one can use to see if the small scale structure phenomenon extends from the very diffuse components that have been well sampled using the d lines ( watson & meyer 1996 ; meyer & blades 1996 ) to highly reddened sightlines such as hd 206267 ( this paper , @xmath6 @xmath7 0.50 ) . in this _
letter _ , we present kitt peak national observatory ( kpno ) cude feed echelle observations of the interstellar absorption towards members of 5 multiple star systems which have saturated interstellar profiles .
these observations allow us to trace the small scale variations previously observed using to much more heavily reddened lines of sight .
in addition , we present a simple method for estimating the possible differences in density in the clouds using the observed column variations .
the observations were obtained in june 1998 with the 0.9 m cude feed telescope and spectrograph at kitt peak national observatory using camera 6 in echelle mode and a ford 3000 x 1000 pixel ccd chip .
the resolution of the data was measured using the thar lamp emission lines and is @xmath8 @xmath9 at the location of the @xmath10 line .
a total of 5 systems were observed : 4 binaries or common proper motion doubles and one triple system ( see table 1 ) .
observations of the stars @xmath11 aql and @xmath11 leo were obtained as a template for dividing out telluric absorption in the vicinity of the 7698 line . for the majority of the stars ,
exposures were taken on different nights at different grating tilts to reduce the effect of any flaws in the ccd chip . however , due to poor weather conditions observations of @xmath12 oph were limited to a single night and only a single exposure was obtained for hd 206267 d. reduction of the data was done with the noao iraf echelle data reduction package .
the individual frames were first bias - subtracted and flat - fielded , then the scattered light was removed .
for all stars except @xmath12 oph the individual orders were then extracted and cleaned of cosmic ray hits . since both members of the binary @xmath12 oph were in the slit , only the portions of the combined stellar profile that were dominated by either one of the two stars
were extracted .
all of the resulting one - dimensional spectra were then wavelength calibrated , shifted to heliocentric coordinates , summed , and finally continuum fitted using low order polynomials
. figures 1 and 2 show the final line profiles .
the resulting s / n - ratios for these observations varies widely , from as low as @xmath020 for the faint star hd 206267 d to as high as @xmath13 for @xmath14 sco .
column densities , b - values , and relative velocities of the various components were derived by profile fitting with the programs xvoigt ( mar & bailey 1995 ) and fits6p ( welty , hobbs , & york 1991 ) . the wavelength and oscillator strength of the 7698 line were taken from morton ( 1991 ) ; the inclusion of hyperfine splitting ( welty , hobbs , & kulkarni 1994 ) in the fits does not significantly alter the derived columns or b - values .
the column density , b - value , and velocity of the various components were free parameters in the initial fits , while in the final fits the velocities of the components were fixed since the velocity differences between members of a system were smaller than the uncertainties .
the resulting column densities are listed in table 2 ; the listed uncertainties include contributions due to signal fluctuations , continuum placement errors , and the errors induced by uncertainty in the b - values .
it must be noted that if there is unresolved component structure that the column density may be higher than that listed . in particular , higher resolution observations of @xmath12 oph a suggest that there are at least three distinct components making up what is fit here as a single main component along this line of sight ( welty 1998 ) .
as seen in figures 1 and 2 , significant differences in the absorption profile were seen between the interstellar lines towards the stars in 3 of the 5 systems .
the detection of significant variation in the profiles of the within the saturated cores of the ( previously observed ) profiles shows that the small scale structures occur even in regions of high column density .
it should be noted that the detection of widespread variation in and in multiple components make a circumstellar origin for the observed small scale structures significantly less tenable .
for example , towards @xmath15 sco variations in are observed at @xmath16 @xmath9 and in in the main component at @xmath17 @xmath9 , with the enhancements in and columns occurring towards different members of the binary .
one also sees variations in multiple components towards the hd 206267 system , but one must note that the hd 206267 system bears a striking resemblance to the trapezium ( abt 1986 ) , and like the trapezium is in a region of recent , active star formation .
thus much of the variation towards this system may be in gas which is located relatively near the stars ( although not circumstellar ) , and may be the result of interactions between the surrounding medium and stellar winds and/or associated with the region ic 1396 centered on hd 206267a ( odell _ et al_. 1993 ) .
while large differences were not detected towards @xmath12 oph ( hd 147933/4 ) or hd 161270/89 it is possible that significant variation could be hidden in the strong ( possibly saturated ) cores of these lines . in any case , the observed variations suggest that the small scale structures previously identified using are truly ubiquitous , and occur even in highly reddened sightlines .
we can use the observed variations in the to infer the corresponding variations for comparison with previous optical studies of small scale structure .
hobbs ( 1974 ) identified a relationship between the total line of sight column densities of and : where n ( ) and n ( ) are in units of @xmath20@xmath21 . assuming the above `` typical '' relationship between n ( ) and n ( ) , we see that the observed variations correspond to variations as large as @xmath22@xmath23 .
thus small scale variations occur over 3 orders of magnitude in column density ( meyer & blades 1996 ; watson & meyer 1996 ; this paper ) , and therefore represent more than a population of small , low column clouds . furthermore , if we assume the `` typical '' relationship between n ( ) and n ( ) from hobbs ( 1974 ) , we can estimate the density @xmath24 in the structures responsible for the observed profile variations using the binary separation by assuming that variations in trace variations in .
then the observed n ( ) differences imply @xmath25@xmath3 , similar to the densities inferred for a number of components in previous studies ( meyer & blades 1996 ; watson & meyer 1996 ) .
such densities are also only slightly less than the densities ( @xmath26@xmath2@xmath3 ) inferred by frail _ et al_. ( 1994 ) in their study of 21-cm absorption towards high velocity pulsars .
the lack of observable changes in the column density in components where the column varies towards the binary @xmath27 cru suggest that at least some fraction of the variation detected in the studies is due to density , temperature , or ionization fraction fluctuations , and thus not indicative of variations in the hydrogen column density between the lines of sight ( lauroesch _ et al_. 1998 ) .
if we assume that the observed variations are due to similar fluctuations , we can use the measured column density differences to estimate ( albeit somewhat crudely ) the difference in density and temperature in these structures .
the recombination rate for potassium is proportional to @xmath28 ( pquignot & aldrovandi 1986 ) ; if we assume a neutral ideal gas in pressure equilibrium then the recombination rate will go roughly as @xmath29 .
typically it is assumed that the @xmath30 ratio is roughly constant in neutral interstellar clouds , with the dominant source of electrons being carbon atoms . under this assumption
we can then estimate required density contrast from the observed columns since n()@xmath31 .
we must note that as the density increases and/or the temperature decreases in these clouds there will be a corresponding increase in the n()/n ( ) ratio , and thus a corresponding decrease in the @xmath30 ratio .
however , if n ( ) @xmath32 n ( ) in both clouds then the @xmath30 ratio will be roughly constant . in any case , we will initially assume @xmath30 is constant , and then identify ( if necessary ) any cases in which a correction is required .
table 2 lists the column densities from our profile fitting analysis of these lines of sight .
based on the measured differences in the column densities for the various components we have estimated the density contrast @xmath33 assuming an ideal gas with a constant @xmath30 , then : where we have defined the strong and weak components such that n()@xmath38n()@xmath39 .
thus all values of @xmath33 will be @xmath40 .
looking at table 2 , we note that the estimated density differences are generally quite small , generally less than a factor of 2 .
this shows how relatively small density and temperature fluctuations can give rise to large differences in the column densities of neutral species .
there have been a number of simplifying assumptions made in the preceding argument which may not be correct .
first , we have assumed that the observed column density differences reflect density and temperature fluctuations and not column variations . despite the lack of column density variations along the lightly reddened @xmath27 crucis line of sight ( lauroesch _ et al_. 1998 ) , the radio observations of pervasive variations in the 21 cm opacity towards pulsars and extra
galactic radio sources imply that some ( or even all ) of these structures may be associated with column density variations .
in addition , these structures may not be in pressure equilibrium with the surrounding medium since we have no information about the lifetimes of these structures we can not say whether they are stable structures or not . we have also assumed that the thermal gas pressure dominates in these structures , implicitly ignoring any magnetic field effects .
thus we may have underestimated the magnitude of the density fluctuations in the sightlines . on the other hand
, we have assumed that the @xmath30 ratio is constant , another possibility is that there are pervasive fluctuations in the electron density on small scales or that the ratio decreases in the denser clouds due to the recombination of carbon .
only by directly measuring the densities ( @xmath41 and @xmath42 ) in individual components towards multiple systems through the use of diagnostics such as the and fine structure lines can the nature of these structures be fully explored .
such ultraviolet observations will also enable the identification of any sightlines with column density enhancements similar to those inferred from radio studies .
however , it must be noted that optical studies remain an efficient probe of small scale structures , and can be used to trace such structures on a variety of scales .
it is a pleasure to acknowledge the support of the staff of kpno , especially daryl willmarth .
we would also like to thank ed jenkins , eric sandquist , dan welty , don york , and the referee for their valuable comments and suggestions .
lcccccc system & alternate & v & e(b - v ) & separation & separation + & i d & ( mag ) & ( mag ) & ( ) & ( au ) + hd 144217/8 & @xmath15 sco & 2.6/4.9 & 0.21 & 13.6 & 2720 + hd 145502/1 & @xmath43 sco & 4.0/6.3 & 0.25 & 41.1 & 7200 + hd 147933/4 & @xmath12 oph & 5.0/5.9 & 0.48 & 3.2 & 740 + hd 161270/89 & & 6.2/6.6 & 0.07 & 20.6 & 2880 + hd 206267 a / c / d & & 5.6/8.4/8.0 & 0.52 & & + lcccc system & velocity & n ( ) & b - value & @xmath44 + & ( @xmath9 ) & ( @xmath45 ) & ( @xmath9 ) & ( @xmath3 ) + hd 144217/8 & -9.9 & 12.3@xmath46/16.6@xmath47 & [email protected] & 1.2 + & -11.7 & 3.3@xmath49/2.3@xmath50 & [email protected] & 1.2 + hd 145502/1 & -9.3 & [email protected]/13.1@xmath51 & [email protected] & 1.1 + & -11.1 & 7.1@xmath46/2.6@xmath52 & [email protected] & 1.8 + & -13.3 & [email protected]/[email protected] & ( 0.2 ) & 1.2 + hd 147933/4 & -7.8 & 117@xmath4818/118@xmath4822 & [email protected] & @xmath531.3 + & -12.5 & [email protected]/@xmath531.3 & [email protected] & @xmath541.0 + hd 161270/89 & -16.5 & [email protected]/[email protected] & [email protected] & @xmath531.2 + hd 206267 a / c / d & -18.9 & [email protected]/[email protected]/[email protected] & [email protected] & 1.3 + & -17.2 & [email protected]/@xmath533.6/[email protected] & [email protected] & @xmath544.2 + & -16.0 & [email protected]/[email protected]/@xmath538.2 & ( 0.3 ) & @xmath531.6 + & -14.5 & [email protected]/[email protected]/[email protected] & [email protected] & 1.2 + & -13.0 & [email protected]/[email protected]/[email protected] & [email protected] & 1.2 + & -10.8 & [email protected]/[email protected]/[email protected] & [email protected] & 1.3 + & -8.9 & [email protected]/[email protected]/[email protected] & ( 0.2 ) & @xmath532.7 + & -4.8 & @xmath531.3/[email protected]/@xmath532.0 & [email protected] & @xmath541.7 + | we present high resolution ( r@xmath0170,000 ) kitt peak national observatory cude feed telescope observations of the interstellar 7698 line towards 5 multiple star systems with saturated components .
we compare the absorption line profiles in each of the two ( or three ) lines of sight in these systems , and find significant differences between the sight - lines in 3 out of the 5 cases .
we infer that the small scale structure traced by previous observations is also present in at least some of the components with saturated absorption lines , and thus the small scale structures traced by the neutral species are occurring at some level in clouds of all column densities .
we discuss the implications of that conclusion and a potential explanation by density inhomogeneities . |
fractional superstrings @xcite are string theories whose physical state conditions are generated by fractional - spin currents on the world - sheet .
the resulting critical central charges of these strings are found to be less than that of the superstring , suggesting the possibility of string theories with critical space - time dimensions less than 10 .
this paper is the second of two papers examining the properties of the simplest such fractional superstring , namely the one with spin-4/3 currents on the world - sheet . the first paper in this series @xcite described tree scattering amplitudes of this string for states in the untwisted sectors of the spin-4/3 fractional superconformal algebra , and showed that they obey duality and spurious state decoupling . also
, a specific @xmath0 ( non - critical ) model of this string with a three - dimensional space - time interpretation was constructed , where it was shown that the untwisted - sector physical states correspond to space - time bosons including the graviton and yang - mills bosons . in ref .
@xcite a no - ghost theorem for these @xmath0 untwisted - sector states was proven .
this paper examines the twisted sectors of the spin-4/3 fractional superstring . in section [ sii ]
we define the twisted sectors in terms of the monodromies of the fractional currents with states in those sectors .
we derive general properties that models having twisted sectors must obey using the methods of refs .
it is found that there are two types of twisted sectors , which we denote by @xmath1 and @xmath2 , whose occurence depends on how a @xmath3 automorphism of the spin-4/3 fractional superconformal algebra is realized in specific conformal field theory ( cft ) models . in section [ siii ]
we show how scattering amplitudes containing one channel of either @xmath1 or @xmath2 twisted - sector states and satisfying spurious state decoupling can be constructed .
various choices of physical state conditions in the @xmath2 sector are consistent with spurious state decoupling in the prescription for scattering amplitudes that we develop .
presumably , demanding duality of four - point amplitudes will further specify the set of @xmath2 physical state conditions .
the restriction in our scattering prescription to only two external twisted - sector states is due to the absence of an appropriate dimension - one vertex describing the emission of a single twisted - sector state .
this is closely analogous to the situation in the `` old covariant '' formalism of superstring amplitudes @xcite in which , due to the absence of the faddeev - popov superghost fields , an appropriate dimension - one vertex for the emission of a ramond - sector state is also lacking .
this analogy is made even closer upon consideration of the example of the @xmath0 model of the spin-4/3 string where the @xmath2 sector is found to describe space - time fermions . in this model ,
discussed in section [ siv ] , the twisted - sector states are realized by a @xmath3 orbifold twisted sector of the @xmath0 cft . in particular
, the @xmath0 cft is a tensor product of three free coordinate boson fields @xmath4 on the world - sheet which do not participate in the orbifolding , and an `` internal '' @xmath5 wess - zumino - witten model described by two bosons compactified on a triangular lattice .
the relevant twisted sector arises upon twisting by a reflection through the origin of that lattice .
some details of this construction are relegated to an appendix . in section [ siv ]
we also calculate some of the low - lying physical states in the @xmath2 sector and compute a three - point coupling between two fermions and the massless vector boson of the untwisted sector .
these computations place some restrictions on the choice of physical state conditions ; however , the precise set of @xmath2-sector physical state conditions consistent with duality of tree amplitudes remains an open question in this model .
there are many other ways of twisting the triangular @xmath5 lattice , but lorentz invariance is invariably broken in the resulting twisted sectors .
this follows from the fact that under the action of all twists other than the @xmath3 twist used to define the @xmath2 sector , at least one of the @xmath5 currents is not invariant .
currents not invariant under the twist can not have zero modes in the twisted sector , implying that lorentz invariance is broken in these sectors . whether or not the states in various twisted sectors should be included in the string spectrum is determined by one loop modular invariance ; however , since the @xmath0 model described above is not a critical ( @xmath6 ) representation of the spin-4/3 fractional superstring , it is not expected to be modular invariant in any case . in section [ sv ]
we briefly describe a @xmath7 model of the spin-4/3 fractional superstring in which the @xmath1 ( as opposed to @xmath2 ) twisted sector is realized . unlike the @xmath2 sector
, the @xmath1-sector physical state conditions are determined by the scattering amplitude prescription developed in section [ siii ] . because the @xmath7 model has only one flat dimension the resulting space - time physics is trivial .
it remains an open question whether either the @xmath1 or @xmath2 sector is actually realized in a critical ( @xmath6 ) model of the spin-4/3 fractional superstring .
we begin with a brief description of the spin-4/3 fractional superconformal ( fsc ) algebra .
a more detailed discussion can be found in ref .
the fractional currents , @xmath8 , and the energy - momentum tensor , @xmath9 , together generate the fsc chiral algebra , encoded in the singular terms of the operator product expansions ( ope ) @xcite : @xmath10 the first ope implies that @xmath9 obeys the conformal algebra with central charge @xmath11 , while the second implies that @xmath8 are dimension-4/3 virasoro primary fields .
the constants @xmath12 in the @xmath13 opes are fixed by associativity to be @xmath14 this algebra generates the physical state conditions for the spin-4/3 fractional superstring . since there is only a single cut on the right - hand side of each ope , the currents @xmath15 are abelianly braided ( or parafermionic ) . under interchange of @xmath16 and @xmath17 ( along a prescribed path ,
say a counterclockwise switch ) the only consistent phase that @xmath18 or @xmath19 can pick up with itself is @xmath20 .
the phase that develops upon interchange of @xmath18 with @xmath19 can be taken to be @xmath21 .
the group of automorphisms of the fsc algebra organizes the representation theory of its highest - weight modules .
the order - six automorphism group @xmath22 of the fsc algebra is generated by the transformations [ autom ] @xmath23 where @xmath24 , and @xmath25 . since the fsc algebra is supposed to be an organizing symmetry of the states of the spin-4/3 string , it is natural to assume that its automorphisms extend to automorphisms of the cft representation of the fsc algebra .
all states can then be assigned definite @xmath26 quantum numbers under the action of the @xmath27 group of automorphisms generated by the transformation ( [ autom]a ) .
the untwisted sectors of the fsc algebra consist of the set of states which obey the bypass relations @xmath28 where @xmath29 is a state with @xmath26-charge @xmath30 .
the bypass relation @xmath31 denotes the analytic continuation of @xmath16 along a closed path looping once around @xmath17 in a counterclockwise sense as shown in fig .
clearly the currents themselves are untwisted - sector fields with @xmath26 charges @xmath32 for @xmath15 and charge 0 for @xmath33 .
we define the twisted sectors of fsc algebra representations in terms of the bypass relations the twist fields obey with the fsc currents .
the basic property of the twisted sectors is that the twist fields are double - valued with respect to the fractional currents @xmath8 .
more precisely , with any twist field @xmath34 , the split algebra currents satisfy the bypass relations @xmath35 the path defining the double - bypass relation @xmath36 is shown in fig .
the phase in the double - bypass relation is chosen so that the @xmath15 currents have zero modes in the twisted sectors .
indeed , the double - bypass relation implies that @xmath37 is a double - valued analytic function on the @xmath16-plane with branch point at @xmath38 and thus defines the mode expansion @xmath39 to further specify the twisted sectors we must also know the single bypass relations of the fractional currents @xmath15 with the twist fields . in general under a single bypass
the @xmath15 currents go to some other pair of ( dimension-4/3 ) currents in the cft under consideration . denoting these currents by @xmath40 ,
we have @xmath41 we have included the factor of @xmath42 for later convenience .
for these bypass relations to be consistent with associativity of the cft operator product algebra , @xmath43 must be images of the @xmath15 currents under the action of some @xmath3 automorphism of the cft .
this implies , in particular , that @xmath40 satisfy the same double - bypass relation ( [ dblbypass ] ) as @xmath15 do , implying that @xmath40 have half - integral mode expansions as in ( [ twistedom ] ) .
the single bypass relations ( [ twcurbp ] ) then imply the identification of modes @xmath44 there are some special choices for the @xmath40 currents ( or , equivalently , for the @xmath3 automorphism ) which lead to particularly simple twisted - sector properties .
the first is the choice @xmath45 , where the @xmath3 automorphism group is trivially realized by the identity transformation . in this case
we simply recover the untwisted sectors .
a second choice is @xmath46 , where @xmath47 .
this choice gives rise to twisted sectors we refer to as the @xmath1 sectors of the fsc algebra , since they are equivalent to the @xmath1 sectors introduced in ref .
the general case where @xmath48 gives rise to what we call @xmath2 twisted sectors .
the remainder of this section explores the representation theory of the @xmath1 and @xmath2 sectors .
we begin with the @xmath1 sector since it is simpler , and derive the generalized commutation relations satisfied by the modes of @xmath15 when acting on an arbitrary twisted - sector state .
we then turn to the @xmath2 sector which is complicated by the need to specify the operator product algebra of the @xmath15 currents with their images @xmath40 under the @xmath3 automorphism .
this depends , in general , on specific properties of the cft representaton of the fsc algebra in question .
the @xmath1 sectors arise from choosing the @xmath3 automorphism which relates @xmath15 and @xmath40 to be an automorphism of the fsc algebra itself .
there are three @xmath3 subgroups of the @xmath22 automorphism group generated by the transformations ( [ autom ] ) of the spin-4/3 fsc algebra .
they give rise to three twisted sectors with bypass relations @xmath49 where @xmath50 .
( the @xmath1-sector `` @xmath51-disorder '' fields @xmath52 introduced in ref .
@xcite actually satisfy the bypass relations @xmath53 . with the change of basis @xmath54 these bypass relations become the bypass relations written above . ) the single bypass relations ( [ rbypass ] ) imply the mode identifications @xmath55 when acting on @xmath56 .
thus in the @xmath1 sector there is really only one independent fractional current , which we can take to be either @xmath18 or @xmath19 .
the three @xmath1 sectors labelled by @xmath57 are related by the @xmath26 automorphism ( [ autom]a ) of the fsc algebra , and thus are isomorphic in models in which that automorphism extends to the whole cft operator product algebra .
since we assume that this is always the case , we henceforth restrict the discussion to the @xmath58 @xmath1 sectors , the other two sectors being identical . in analogy to the superconformal gauge of the ordinary superstring ,
the physical states of the spin-4/3 fractional superstring are certain highest - weight states of the fsc algebra that is , they are annihilated by all the positive modes of @xmath33 and @xmath15 . in order to show spurious state decoupling in scattering amplitudes
, we will need to know the algebra satisfied by these modes .
this algebra takes the form of generalized commutation relations ( gcrs ) for the modes of @xmath15 due to cuts in the fsc operator product algebra @xcite . to derive the @xmath1-sector gcrs ,
consider the integral @xmath59 where the contours both wind twice around the origin with the @xmath60 contour inside the @xmath61 contour .
the factors in the integrand have been chosen to make these contours closed : the whole integrand is a double - valued analytic function on both the @xmath16 and @xmath17 planes with branch points at @xmath38 and @xmath62 and possible poles at @xmath63 or @xmath64 ( _
i.e. _ , at the point @xmath63 on both sheets ) . evaluating this integral by shrinking the @xmath60 contour close to the origin and using the mode definition
( [ twistedom ] ) , which can be inverted as @xmath65 gives @xmath66 where the @xmath67 are binomial coefficients defined by the expansion @xmath68 the integral @xmath69 can also be evaluated in another way by deforming the @xmath61 contour to lie inside the @xmath60 contour .
the result of this deformation is three contributions , as shown in fig .
one contribution is just the same integral with @xmath61 and @xmath60 interchanged .
the other two contributions pick up residues associated with the @xmath70 ope singularities at @xmath63 and @xmath64 .
the former singularity can simply be read off from the @xmath71 ope ; to evaluate the latter , we must continue @xmath16 once counter - clockwise around the origin to @xmath72 before letting it approach @xmath17 ( on the second sheet ) . by the single - bypass relation ( [ rbypass ] ) , this analytic continuation is @xmath73 performing the same continuation on the other factors in the integrand , picking up the residues of poles from the fsc algebra , and combining all the contributions as shown in fig .
[ fig2 ] , gives @xmath74 = 2^{-5/3}\lambda^+ ( -1)^{n+m}g^+_{n+m\over2 } \nonumber\\ & & % \qquad\qquad\qquad % % add line for galley \qquad\qquad\mbox { } + 2^{-4/3 } \left\{(-1)^n+(-1)^m\right\ } \left[l_{n+m\over2 } + { c\over128 } \left(6n^2 - 5\right)\delta_{n+m}\right ] .
\end{aligned}\ ] ] where we have used ( [ zfmodiden ] ) to write the gcrs in terms of @xmath18 modes alone .
the commutation relations of the @xmath15 modes with the @xmath75 modes of the stress - energy tensor follow in the standard way : @xmath76 = \left({n\over 3}-{m\over 2}\right ) g^\pm_{n+{m\over2}}.\ ] ] acting on highest - weight states ( states annihilated by all the positive modes of @xmath15 and @xmath33 ) , one can show from ( [ rsectgcr ] ) that the zero modes satisfy the relation @xmath77 .\ ] ] we refer to a twisted sector obeying the bypass relations ( [ twcurbp ] ) with @xmath78 as an @xmath2 sector .
the @xmath2 sector of the fsc algebra is characterized by a @xmath3 automorphism of the particular cft model in question which interchanges @xmath79 .
this symmetry between @xmath15 and @xmath80 implies that the @xmath81 opes form a second spin-4/3 fsc algebra by themselves ( at the same central charge and with structure constants @xmath82 ) , and that @xmath40 have untwisted - sector @xmath26 charges @xmath32 , and so obey all the concomitant bypass relations with @xmath15 .
the @xmath83 opes , on the other hand , remain undetermined by this symmetry , and depend on the properties of the specific cft representation of the fsc algebra under consideration .
we will assume that the @xmath84 opes are of the form @xmath85 + \ldots\biggr\}. \end{aligned}\ ] ] we will see that the @xmath2 sector of the @xmath0 model to be considered in section [ siv ] has currents obeying opes of this form .
the form of the @xmath83 opes ( [ ggtope ] ) is actually more general , and follows for representations satisfying a few physical properties .
assume that these representations have a global @xmath86-dimensional poincar symmetry realized by the tensor product of @xmath86 coordinate bosons @xmath4 with an `` internal '' cft which has a positive definite spectrum of highest weights .
@xmath15 and @xmath87 then only involve derivatives of the @xmath4 fields with no vertex contributions of the form @xmath88 , and are @xmath89 singlets .
this requirement is natural in representations which have a flat @xmath86-dimensional space - time interpretation in string theory .
in addition , if no dimension-1/3 and no dimension - one @xmath89 scalar fields exist , then ( [ ggtope ] ) is the most general form the @xmath83 opes can take .
note that these last two conditions are not as strong as they might appear .
for example , a pair of @xmath26-charge @xmath32 dimension-1/3 scalar fields would obey braiding properties and operator product selection rules identical to those of @xmath26 parafermion currents ; but associativity of the @xmath26 parafermion current algebra fixes the central charge to be @xmath90 @xcite .
a similar argument applies to potential dimension - one currents for example , a simple associativity argument shows that adding a single dimension - one current to the spin-4/3 fsc algebra fixes the central charge to be @xmath91 .
these arguments are not proofs , though , since associativity constraints can be evaded by increasing the number of independent dimension-1/3 or dimension - one currents . in any case , however the reader judges the reasonableness of the above assumptions , there are at least two cases in which they are satisfied namely the @xmath0 model to be discussed in detail in section [ siv ] , and a @xmath92 model described in appendix c of ref . @xcite .
it follows from ( [ ggtope ] ) that the combinations of fields @xmath93 are dimension-4/3 , @xmath26-charge @xmath32 untwisted - sector highest - weight operators with respect to the fsc algebra generated by @xmath15 .
the properties of the untwisted - sector highest - weight modules derived in ref .
@xcite , together with the assumed symmetry under interchanging @xmath15 and @xmath94 , implies that @xmath95 , \nonumber\\ { \cal a}^\pm & = & { \mu \lambda^\mp \over \mu + 1 } \left(g^\pm + \delta \widetilde g^\pm \right ) .
\end{aligned}\ ] ] the dimension - two descendents of the @xmath96 fields , @xmath97 ope . from the bypass relations ( [ twcurbp ] ) and the opes ( [ ggtope ] ) , and using the mode identifications ( [ ggtequiv ] ) , one deduces the gcrs ( just as was done in the last subsection ) @xmath98
\nonumber\\ & & \qquad\qquad % % remove line for galley = { 2^{-5/3}\mu \over \mu+1 } \left[(-1)^n \lambda^- g^+_{n+m\over2}+ ( -1)^m \lambda^+ g^-_{n+m\over2}\right]\nonumber\\ & & % \qquad\qquad % % add line for galley \qquad\qquad \
\mbox{}+2^{-4/3}\left\{1+(-1)^{n+m}\right\ } \left [ l_{n+m\over2 } + { c\over128}\left(6n^2 - 5\right ) \delta_{n+m}\right ] , \end{aligned}\ ] ] and @xmath99 = 2^{-5/3}\lambda^\pm(n - m ) g^\mp_{n+m\over2 } \nonumber\\ & & \qquad\qquad \mbox { } % % remove line for galley + 2^{-4/3 } \left\{(-1)^n+(-1)^m\right\}{3\mu c\over32}n\delta_{n+m}. \end{aligned}\ ] ] the commutation relations ( [ glcom ] ) with the @xmath75 modes follow in the standard way .
note that , when acting on a highest - weight state , the second gcr gives no relation for the @xmath100 zero mode product .
one can obtain a relation for these zero modes by changing the power of the @xmath101 factor in the integrand of ( [ twistint ] ) .
one finds in this way @xmath102 = 2^{-5/3}\lambda^\pm g^\mp_{n+m\over2 } \nonumber\\ & & % \qquad\qquad\qquad % % add line for galley \qquad\ \mbox { } + 2^{-4/3 } \left\{(-1)^n+(-1)^m\right\ } \left[\mu l_{n+m\over2 } + { \cal b}^\mp_{n+m\over2 } + { \mu c\over128}\left(6n^2 - 5\right)\delta_{n+m}\right ] .
\end{aligned}\ ] ] note , however , the appearance of the modes of the representation - dependent dimension - two operators @xmath103 in ( [ twistgcr2 ] ) .
we now extend the prescription developed in ref .
@xcite for dual n - point tree amplitudes of untwisted - sector states satisfying spurious state decoupling to include two twisted - sector states . in
that prescription amplitudes could be written in either of the equivalent `` pictures '' @xmath104 where @xmath105 are untwisted - sector physical states of @xmath26 charge @xmath32 and dimension 1/3 , and @xmath106 is a certain dimension - one descendent of these fields . by analogy to the ramond sector of the ordinary superstring in the `` old covariant '' formalism @xcite , we take the physical states @xmath107 of the twisted sectors to be annihilated by the positive modes of the currents @xmath15 and @xmath33 , and to be eigenstates of their zero modes with `` intercepts '' @xmath108 and @xmath109 : @xmath110 the @xmath111 condition gives rise to the mass - shell condition for the physical states a klein - gordon equation for each independent component of the physical state .
the @xmath112 conditions , on the other hand , are linear in space - time derivatives ( because the fractional currents have the form @xmath113 for models with a flat space - time interpretation ) , and so should give rise to a dirac equation for the components of the physical states .
this will indeed turn out to be the case in the models discussed in sections [ siv ] and [ sv ] .
of course , the dirac equations must be consistent with the klein - gordon equations .
this is automatically ensured in the @xmath1-sector by the relation ( [ rsectzmalg ] ) between @xmath112 and @xmath111 ; however , it is not automatically satisfied in the @xmath2-sector , and must be imposed as a separate requirement .
we expect the form of tree amplitudes with one twisted - sector channel to consist of `` in '' and `` out '' twisted - sector physical states @xmath107 sandwiching the vertex operators of physical untwisted - sector states @xmath114 which are strung together with the dirac propagator : @xmath115 where @xmath116 the choice of @xmath26 charges of the vertices and propagators ( _ i.e. _ , their plus or minus superscripts ) will be shown to be immaterial in the @xmath1 sector , and will have to be further specified for the @xmath2 sector . to show spurious state decoupling for these amplitudes , we must be able to transform them to another `` picture '' , similar to those in ( [ asum ] ) , in which the dimension - one @xmath106-vertices appear instead of the dimension-1/3 @xmath105 vertices .
we first describe this picture - changing and the resulting spurious - state decoupling theorem in the @xmath1-sector case and then move onto the more complicated @xmath2-sector case .
we start with the _ anzatz _ for the @xmath117-point amplitude with one @xmath1-sector channel @xmath118 this form for the amplitude can be rewritten in another `` picture '' using the following commutator @xmath119 & = & \left(l_0+{n\over2}-h_t\right ) w^\pm(1 ) % \nonumber\\ & & \qquad\mbox { } % % add line for galley -w^\pm(1)\left(l_0-h_t\right ) , \end{aligned}\ ] ] for any integer @xmath120 , which follows from the representation theory of the untwisted - sector module that @xmath106 belongs to .
( see , for example , the derivation of eq .
( 3.10 ) of ref .
@xcite . ) in particular , rewrite a special case of ( [ twiscom ] ) as @xmath121 insert a factor of @xmath122 before any @xmath123 propagator in ( [ apictran ] ) and commute the @xmath124 factor in the numerator to the left using ( [ spcase ] ) .
the first two terms on the right - hand side give vanishing contributions since the @xmath124 factor can be continually commuted to the left using ( [ spcase ] ) until it annihilates the @xmath125 physical state , while the second term vanishes by the `` cancelled propagator '' argument .
tree amplitudes with cancelled propagators are holomorphic in the mandelstam invariant of the cancelled propagator channel , and thus vanish by analyticity if the amplitudes have sufficiently soft high - energy behavior , as string amplitudes do .
the factor of @xmath126 in the third term of ( [ spcase ] ) cancels the @xmath123 propagator , leaving behind a klein - gordon propagator @xmath127 . applying this argument
repeatedly gives @xmath128 the fact that one of the space - time boson vertices is still a highest - weight state @xmath129 , and not one of the @xmath26-charge @xmath130 descendents @xmath106 , is familiar from tree amplitudes with one fermion line for the ordinary superstring in the old covariant formalism .
the position of the lone @xmath129-vertex is arbitrary , as can be shown by manipulations similar to those used to derive ( [ apict2r ] ) .
furthermore , the above manipulations can be reversed using eq .
( [ spcase ] ) with @xmath131 , @xmath132 , and @xmath133 to show that any choice of @xmath134 superscripts in ( [ apictr ] ) is equivalent to ( [ apictran ] ) .
it is important to note in this connection that due to the equivalence of @xmath18 and @xmath19 modes ( [ zfmodiden ] ) in the @xmath1-sector , that we must have @xmath135 .
the following trick allows us to re - express the amplitude ( [ apict2r ] ) completely in terms of @xmath106 vertices .
insert @xmath122 between @xmath136 and @xmath137 and use ( [ spcase ] ) once more to finally obtain the amplitude in the form @xmath138 this form for the amplitude is slightly delicate because both the numerator and denominator in the factor before @xmath137 annihilate it .
however , this potential ambiguity can be resolved by noting that the @xmath139 operator , when acting on the spinor wavefunction @xmath140 associated with @xmath141 , is the klein - gordon operator @xmath142 , while @xmath143 is the associated dirac operator @xmath144 . for massive states the simple zeroes of the numerator and denominator cancel , and for massless states
it can be defined as the limit as @xmath145 goes on mass - shell .
if we start with physical states in the amplitude ( [ apict3r ] ) , will they scatter only to other physical states
? we can reformulate this question in terms of spurious state decoupling .
a state @xmath146 obeying the zero - mode conditions in eq .
( [ pscr ] ) is called a spurious state if it is orthogonal to all physical states .
since the physical state conditions are the only restriction on a generic physical state , it follows that @xmath146 can be written as @xmath147 in terms of some other states @xmath148 and @xmath149 .
all states not satisfying the physical state conditions must have a spurious component .
a physical state can itself be spurious , in which case it is a null state ( since it is orthogonal to itself ) , and should also decouple from all scattering amplitudes .
thus , the decoupling of all spurious states from scattering amplitudes of physical states is a prerequisite for a sensible interpretation of those amplitudes . for this decoupling to be true
, no spurious states should contribute to residues of poles in amplitudes when an internal propagator goes on - shell .
suppose we fix the external momenta @xmath150 of the vertices @xmath151 such that some state @xmath152 in the string fock space at momentum @xmath153 is on - shell : @xmath154 of momentum @xmath155 at the propagator between @xmath156 and @xmath157 , @xmath158 then the @xmath159 term in the sum contributes a pole in momentum space .
the requirement of spurious state decoupling is that if @xmath152 is spurious , its contribution to the residue of the pole should vanish : @xmath160 to prove this , consider one term , say @xmath161 with @xmath162 , in the presentation of @xmath146 as a sum of descendent states , eq .
( [ spur ] ) , where @xmath163 must satisfy @xmath164 .
( the @xmath165 descendent pieces can be shown to decouple by the same argument using the @xmath1-sector equivalence of @xmath18 and @xmath19 ( [ zfmodiden ] ) .
the @xmath75 pieces decouple by a simpler argument . )
the @xmath166 mode can be commuted to the right in eq .
( [ spstde ] ) using eq .
( [ twiscom ] ) and the identity @xmath167 which follows from ( [ glcom ] ) .
the insertions coming from the right - hand side of eq .
( [ twiscom ] ) again vanish by a cancelled propagator argument .
finally , the @xmath166 mode can be seen to pass through the @xmath143 factor and annihilate @xmath137 by the physical state conditions using the @xmath1-sector gcr ( [ rsectgcr ] ) .
this proves spurious state decoupling .
since there is no appropriate dimension - one commuting vertex in the @xmath1 sector , we can not extend our scattering amplitude prescription ( [ apictran ] ) or ( [ apict2r ] ) to include more than two @xmath1-sector vertices . by the same token
, we can not prove cyclic symmetry ( duality ) of these amplitudes in the present formalism .
this situation is closely analogous to what happens in the old covariant formalism in the ordinary superstring .
there , dual amplitudes with spurious state decoupling can be formulated for scattering of neveu - schwarz sector states , and can only be extended to include two ramond - sector states as the `` in '' and `` out '' states in the correlator , thus losing manifest cyclic symmetry .
so presumably , just as in the ramond sector of the superstring , our inability to incorporate more than two @xmath1-module physical states in our scattering prescription means that there is a nontrivial contribution to @xmath1-module scattering amplitudes coming from the `` fractional superghost '' fields on the world - sheet .
the main difference between the @xmath1 and @xmath2 sector scattering amplitudes is that in the @xmath2 sector there is no identification of @xmath18 with @xmath19 modes analogous to ( [ zfmodiden ] ) .
this means that the zero - mode conditions ( [ pscr ] ) generate two _ a priori _ independent dirac equations for the physical state wave functions .
this will lead to overly restrictive physical state conditions unless the @xmath15 intercepts are appropriately tuned , or only a subset of all the @xmath168 modes with @xmath162 are used as @xmath2-sector physical state conditions .
since , as was emphasized at the end of section [ sii ] , the mode algebra obeyed by the fractional currents in the @xmath2 sector depends on the details of the particular cft model under consideration , we can only hope to determine the appropriate @xmath2-sector physical state conditions in the context of a specific model .
however , we can still demonstrate spurious state decoupling in amplitudes with one @xmath2-sector channel . indeed ,
all the manipulations of the last subsection for the @xmath1 sector scattering go through unchanged in the @xmath2 sector as long as all the untwisted vertices @xmath105 and dirac propagators @xmath169 in ( [ apictr ] ) have the same @xmath26 charge ( _ i.e. _ , all their superscripts are the same ) .
the point is simply that since the @xmath170 and @xmath132 modes are not related in any general way in the @xmath2 sector , the associated dirac propagators are not equivalent in particular @xmath171 is not proportional to the identity .
in fact , one can show that the _ anzatz _ for the one channel @xmath2-sector scattering amplitude @xmath172 obeys spurious state decoupling , where @xmath173 is any fixed complex number and we define @xmath174 this follows from the fact that the analog of the commutation relation ( [ twiscom ] ) is satisfied for the above combinations : @xmath175 & = & \left(l_0+{n\over2}-h_t\right ) w^\alpha(1 ) % \nonumber\\ & & \qquad\mbox { } % % add line for galley -w^\alpha(1)\left(l_0-h_t\right ) , \end{aligned}\ ] ] thus , we have a one - parameter family of satisfactory scattering amplitudes in the @xmath2 sector .
tree - level duality may restrict the value of @xmath173 and the correct set of physical state conditions in the @xmath2 sector .
in particular , by duality , if one factorizes the amplitude ( [ apictrprime ] ) in a channel other than the @xmath2-sector channel displayed , one should find an infinite tower of poles corresponding to untwisted - sector physical states .
presumably this occurs only if the @xmath2-sector states @xmath176 obey the correct set of physical state conditions .
we will discuss the choice of physical state conditions in the context of a concrete example in the next section , though to the level we compute we will not be able to put many restrictions on the possible choices . as an example of the kind of choices that could make sense ,
note that associated with each value of @xmath173 is a natural choice of physical state conditions for @xmath2-sector states : @xmath177 for @xmath162 .
note also that by the commutation relation ( [ glcom ] ) all the possible physical state conditions are generated by @xmath111 , @xmath178 , @xmath112 , and @xmath179 .
in particular , the commutator of @xmath178 with @xmath112 generates all the integer - moded @xmath180 conditions .
thus , it may also be consistent to restrict the physical state conditions to only the integer - moded @xmath33 and @xmath181 annihilation operators .
these , however , are only the simplest guesses . the fact that @xmath2-sector states are @xmath3-twist fields obeying non - trivial monodromies with each other
may indicate that the physical state conditions and scattering amplitudes for @xmath2-sector `` matter '' fields ( _ i.e. _ , after integrating out the fractional superghost pieces of the full physical vertices ) may be more complicated than what we have presented here .
strings propagating in @xmath86 flat space - time dimensions are described by a world - sheet cft which includes @xmath86 massless scalar fields @xmath182 .
the spin-4/3 fractional superstring cft also includes a set of fields @xmath183 of conformal dimension 1/3 , transforming as vectors under space - time lorentz transformations .
the simplest non - trivial such cft which is also a representation of the spin-4/3 fsc algebra can be constructed from five free massless scalar fields on the world - sheet , and hence has central charge @xmath0 .
( a list of known representations of the spin-4/3 fsc algebra is given in appendix c of ref .
three of the scalars are just coordinate boson fields @xmath182 , @xmath184 , with the standard operator products @xmath185 , where @xmath186 is the three - dimensional minkowski metric with signature @xmath187 .
the remaining two fields , @xmath188 , @xmath189 , are compactified on a triangular lattice : @xmath190 , with @xmath191 , where @xmath192 these two bosons form a representation of the @xmath5 wess - zumino - witten model .
thus the @xmath0 model has a global three - dimensional poincar symmetry .
vertex operators in the @xmath5 cft , @xmath193 for integer @xmath194 , describe virasoro - primary operators transforming as integer - spin fields under the @xmath195 symmetry . for example , some of the simplest @xmath5 virasoro - primary fields are the dimension-1/3 @xmath195-vector fields @xmath196 , the dimension-4/3 @xmath195-scalars @xmath197 , and the dimension - one @xmath198 s , given by @xmath199 the properties of these and other fields in the @xmath0 model are discussed in more detail in ref .
@xcite .
the fractional supercurrents @xmath15 satisfying the spin-4/3 fsc algebra ( [ fsca ] ) are given in terms of these fields by @xmath200 there are actually six solutions for the supercurrents , which can be obtained from the above solution by making the transformations @xmath201 or @xmath202 , where @xmath203 and @xmath40 are given by @xmath204 which differs from ( [ grepd ] ) by a sign change in the @xmath205 terms .
the existence of these six solutions is a consequence of the @xmath206 automorphism group of the @xmath0 cft generated by @xmath207 , @xmath208 , and @xmath209 , which leaves the @xmath195 generators invariant . in what follows ,
we construct the @xmath2 twisted sector corresponding to the @xmath3 automorphism @xmath209 which is generated by the transformation @xmath210 .
recalling the discussion of section [ sii ] , the fields of this twisted sector are characterized by the property that under single - bypass around a twisted - sector field , untwisted - sector fields transform according to this automorphism .
thus the defining property of the @xmath2 twisted sector is that with any twist field , @xmath34 , the free boson fields @xmath211 satisfy the basic bypass relation @xmath212 this is also the definition of the twisted sector of a @xmath3 orbifold @xcite of the @xmath5 cft under the action of the symmetry which reflects the triangular @xmath213-boson lattice through the origin .
note that this @xmath3 transformation maps the @xmath15 currents into the @xmath40 currents .
it is straight forward to show that these currents obey opes of the form ( [ ggtope ] ) with @xmath214 where @xmath215 and @xmath216 are the stress - energy tensors for the @xmath4 and @xmath5 cfts , respectively .
thus we are indeed constructing precisely the @xmath2 sector discussed in section [ sii ] .
there is a physical reason for expecting this sector to appear in the spin-4/3 fractional superstring : it includes the space - time fermionic states of the @xmath0 model .
the untwisted - sector states described so far are all space - time bosonic states , corresponding to integer - spin representations of @xmath195 , in the @xmath0 fsc algebra representation . in general
, the @xmath217 wess - zumino - witten model can be realized as the @xmath3 orbifold of the @xmath218 model , with the spinor representations of @xmath219 appearing as the @xmath3-twisted fields @xcite .
thus we expect to find the space - time fermionic physical states of the @xmath0 model of the spin-4/3 fractional superstring in the @xmath3-twisted sector of @xmath5 . in the next subsection
we construct this twisted sector in some detail , and show along the way that the @xmath2-sector bypass relations ( [ twcurbp ] ) are indeed satisfied . in the subsequent subsections we compute some low - lying @xmath2-sector physical states and discuss their scattering amplitudes using the results of section [ siii ] . from the bypass relation ( [ btwistby ] ) , it follows that the @xmath211 bosons have the mode expansion @xmath220 with modes satisfying the commutation relations @xmath221=r g^{ij}\delta_{r+s } .\ ] ] it is easy to show that @xmath222 = -g^{ij}{\rm ln}\left({\sqrt z-\sqrt w\over\sqrt z+\sqrt w}\right ) , \ ] ] implying the basic ope @xmath223 .
the quantization of the zero mode @xmath224 is crucial for correctly reproducing the operator products of untwisted sector operators when acting on states in the twisted sector . as shown in ref .
@xcite , the correct quantization of the @xmath224 zero modes of @xmath211 is @xmath225=i\pi\varepsilon^{ij},\ ] ] where @xmath226 is the antisymmetric two - index tensor normalized by @xmath227 .
these commutation relations result from the proper quantization of free boson zero modes in the presence of the constraints arising from identifications under lattice translations and the inclusion of a constant antisymmetric background field @xcite . in the twisted sector ( _ i.e. _ , with a @xmath3 twist field inserted at the origin ) the basic vertex operators @xmath228 have the normal ordered expression @xmath229 the first factor provides the signs needed to `` wick rotate '' from @xmath230 to @xmath195 .
no further cocycles are needed , since the @xmath230 signs are automatically taken care of by the commutation relations of the @xmath224 zero - modes .
the normal ordering factors @xmath231 are required to insure the factorizability of amplitudes ( associativity of the operator product ) . using this explicit form ( [ twvtx ] ) for the vertex and the mode commutation relations ( [ betacomm ] ) and ( [ narain ] ) , it is easy to check that the vertex operators indeed obey the correct untwisted - sector operator product expansion @xmath232 as they should since the operator products encode only local information and do not depend on whether any twist fields are located elsewhere on the world - sheet .
the representations of the zero mode algebra ( [ narain ] ) determine the properties of the twisted - sector ground state .
it is shown in the appendix that there are only three representations of the zero mode algebra which preserve the global @xmath195 symmetry of the @xmath5 cft .
this implies that there are three twisted sectors @xmath233 , labeled by @xmath50 , and that if @xmath56 is an arbitrary twist field in @xmath233 and @xmath234 is an untwisted - sector vertex operator , then @xmath235 .
thus the three twisted sectors are disjoint ; in fact , they are just copies of one another under the action of a @xmath26 symmetry of the @xmath5 cft . indeed , from the expression ( [ twvtx ] ) for the @xmath234 vertex acting on an arbitrary twisted - sector state @xmath236 one obtains the bypass relation @xmath237 where in the second line we have used a property of the zero - mode representations derived in the appendix .
this bypass relation implies the double - bypass relation @xmath238 , which in turn implies the existence of the mode expansion @xmath239 it is clear from ( [ sbpr ] ) that the @xmath233 twisted sectors can be `` rotated '' into each other by the @xmath26 automorphism of the @xmath5 cft : @xmath240 .
thus the three twisted sectors are equivalent and we henceforth restrict ourselves ( without loss of generality ) to the @xmath58 sector . in this sector the single - bypass relation ( [ sbpr ] ) for the basic fields are @xmath241 the last of which is precisely the defining bypass relation ( [ twcurbp ] ) of the @xmath2 sector .
these bypass relations along with the mode expansion ( [ twvtxmodexp ] ) imply that @xmath3-even fields have only integer moding , while the @xmath3-odd fields have only half - odd integer modings .
for example , @xmath242 we now discuss the properties of the twist fields . in order to characterize the twisted - sector states , we must take into account the action of the zero modes @xmath224 on the twist states .
in particular , the twisted - sector ground state forms a representation of the nontrivial zero - mode algebra ( [ narain ] ) . as a result , it is shown in the appendix , the ground states in the twisted sectors are doubly degenerate .
we label the corresponding twist fields by @xmath243 and the ground states by @xmath244 , where @xmath245 or @xmath246 .
the action of the zero mode of an arbitrary untwisted - sector operator on the twisted - sector ground states can be worked out using the methods of the appendix . in particular
, one finds that the zero modes of the @xmath195 scalar and vector fields act on the ground state as @xmath247 where @xmath248 is a dirac gamma matrix obeying @xmath249 where @xmath250 is the antisymmetric tensor in three dimensions normalized by @xmath251 .
these zero - mode actions imply that the twisted sector ground states transform as spinors under the global @xmath195 symmetry .
thus twisted sector physical states describe space - time fermionic excitations of the spin-4/3 fractional string .
since the ground states satisfy @xmath252 for @xmath253 , a simple computation reveals @xmath254 implying that the conformal dimension of the twisted - sector ground state is @xmath255 .
@xmath256 is the spinor metric which can be taken to be @xmath257 .
all other twist fields are created from @xmath258 by the repeated action of the @xmath259 creation modes @xmath260 with @xmath261 .
thus the twist states have the spectrum of conformal dimensions @xmath262 for @xmath162 an integer .
acting on a twisted - sector state , the fsc currents @xmath15 have the mode expansion @xmath263 following from ( [ twvtxmodexp ] ) .
since the space - time coordinate boson fields @xmath4 of the @xmath0 representation of the fsc algebra are unaffected by the orbifolding procedure , their mode expansion is the usual one , @xmath264 satisfying the standard commutation relations @xmath265 = ig^{\mu\nu}$ ] and @xmath266 = m\delta_{m+n } g^{\mu\nu}$ ] . combining this with the mode expansion of the @xmath5 fields acting on twisted - sector states , and recalling the form of the currents ( [ grepd ] ) , we obtain @xmath267 though we could , in principle , use the mode expansion ( [ gtwmodexp ] ) to calculate the action of the @xmath15 modes on arbitrary twisted - sector states using the expression ( [ twvtx ] ) for the vertex operators in terms of the modes of @xmath213 , this is in practice a complicated way of computing . a more efficient way which preserves lorentz invariance in intermediate steps
is to express all the twisted - sector states in terms of products of modes of @xmath268 acting on the twisted - sector ground - state spinor .
the set of all products of @xmath268 modes is not a linearly independent basis , however , so we need to compute the algebra satisfied by the @xmath268 modes .
also , in order to compute the action of the @xmath15 and @xmath33 modes on twisted - sector states , we have to express them in terms of the @xmath268 modes .
the algebra of the @xmath268 modes can be written as a set of generalized commutation relations ( gcrs ) derived from the @xmath268 opes in the same way that the @xmath15 gcrs were derived in section [ sii ] .
noting from ( [ so21modiden ] ) that the @xmath269 and @xmath270 modes are related , we define @xmath271 using the opes @xmath272 and @xmath273 , the mode algebra for the @xmath274 s is found to be @xmath275 where the coefficients @xmath67 are given in eq .
( [ ddefn ] ) .
any @xmath5 twisted - sector state can be written as a polynomial in the @xmath274 creation modes acting on the twisted - sector ground state .
the gcr ( [ epgcr ] ) plus the identity @xmath276 which is easily verified from the expression for vertex operators in terms of the @xmath213 modes ( [ twvtx ] ) , are sufficient to reduce any set of such states to a linearly independent basis .
the current modes can be expressed in terms of @xmath274 modes as follows .
since the @xmath277 and @xmath278 modes are related by the identifications ( [ so21modiden ] ) , we can define @xmath279 and from the ope @xmath280 one then derives @xmath281 similarly , since @xmath282 , @xmath283 where @xmath284 are the modes of @xmath216 .
the mode expansion of the full stress - energy tensor is then @xmath285 using ( [ stoep ] ) and ( [ ttoep ] ) in ( [ ltwmodexp ] ) and ( [ gtwmodexp ] ) gives the current modes solely in terms of @xmath286 and @xmath173 modes . as discussed in section [ siii]b , the correct set of physical state conditions in the @xmath2 sector
is not known .
so , for the moment , we assume the maximal set : @xmath287 for @xmath162 an integer .
the @xmath108 intercept determines the conformal dimension of physical state vertex operators in the twisted sector . in terms of the polarization spinors of the twisted sector states
, the @xmath111 condition gives the usual klein - gordon equation fixing the mass of the state .
the @xmath112 conditions , likewise , give dirac equations for the spinor wavefunctions , which also fix the mass of the state .
the consistency of these three zero - mode conditions determines the @xmath112 intercepts @xmath109 in terms of the @xmath111 intercept @xmath108 .
we now solve these physical state conditions for the lowest two levels of twisted - sector states .
the lowest twisted - sector state is @xmath288 where @xmath289 and @xmath290 is a spinor wavefunction .
the only non - trivial physical state conditions come from the @xmath111 and @xmath112 modes : [ tauzero ] @xmath291 consistency of ( [ tauzero]a ) with ( [ tauzero]b ) implies @xmath109 and @xmath108 are related by @xmath292 now we examine the first excited state in the twisted sector .
these are the four dimension-5/8 states @xmath293 . written in this manner
their @xmath195 properties are not apparent .
these can be made more manifest by introducing the combinations @xmath294 from the identity ( [ gtraceiden ] ) it follows that @xmath295 thus , @xmath296 describes a spin-3/2 @xmath195 representation .
if desired , they could be written in terms of @xmath213 modes as @xmath297 , @xmath298 , and @xmath299 .
the general first excited twisted - sector state is then @xmath300 where @xmath301 .
the constraint ( [ constraint ] ) implies that we are free to redefine the spin-3/2 polarization by @xmath302 , where @xmath303 is an arbitrary spinor .
we fix this freedom by taking @xmath304 the non - trivial physical state conditions come from the @xmath111 , @xmath112 and @xmath179 modes : [ tauone ] @xmath305 together with ( [ polconstr ] ) these are the correct equations of motion for a spin-3/2 particle
. consistency of the dirac equations ( [ tauone]b ) with the @xmath111 condition ( [ tauone]a ) implies the intercepts are related by @xmath306 note that the @xmath179 condition ( [ tauone]c ) is redundant , since it follows from the dirac equations ( [ tauone]b ) along with the constraint ( [ polconstr ] ) . from these two lowest @xmath2-sector levels
, it already follows that the maximal set of physical state conditions ( [ psct ] ) must be modified .
in particular , the relations ( [ interel ] ) and ( [ interel2 ] ) between the @xmath109 and @xmath108 intercepts are different and have no common solution .
if , however , we restrict ourselves to the set of physical state conditions ( [ pscrprime ] ) discussed in section [ siii]b and parametrized by @xmath173 , we find that there is then a common solution for the intercepts only for @xmath307 , @xmath308 and @xmath309 .
in other words , a consistent set of @xmath2-sector physical state conditions may be to impose only the vanishing of the @xmath18 annihilation operators and the @xmath310 condition : @xmath311 for @xmath162 an integer .
the action of the @xmath19 operators , and in particular the value of the @xmath132 intercept , would then be free to vary from state to state as determined by the gcrs for the @xmath0 model . with this choice of physical state conditons ,
the lowest - level @xmath2-sector state @xmath312 is a massless majorana spinor , and the next level describes a massive spin-3/2 particle .
note also that since the @xmath313 condition ( [ tauone]c ) is redundant , it might also be consistent to discard the half - odd - integral modings of @xmath18 as physical state conditions . a potentially useful exercise would be to compute @xmath2-sector physical states at higher levels to check these conjectures . the simplest non - trivial scattering amplitude we can write using the prescription ( [ apict2r ] ) developed in section [ siii ] is a three - point coupling for two @xmath2-sector ground states @xmath312 and the massless vector state from the untwisted sector .
this latter state was worked out for the @xmath0 model in ref .
@xcite , and is described ( in the @xmath26-charge @xmath314 sector ) by the vertex @xmath315 where the momentum and polarization satisfy @xmath316 .
the three - point coupling is then easily worked out : @xmath317 the first term is just the expected minimal coupling of the fermions to the gauge field .
the second term represents a derivative coupling which is higher - order in the string tension , and therefore is suppressed at energies below the planck scale .
this string correction to minimal coupling does not occur in the corresponding ordinary superstring amplitude , although string correction terms do appear in higher - point functions .
higher - point tree amplitudes can be calculated similarly using @xmath129 vertices and @xmath318 propagators , in accordance with the prescription developed in section [ siii]b .
as mentioned in that section , even without an understanding of the world - sheet `` fractional superghost '' system , the correctness of our twisted - sector scattering prescription can still be tested at tree level by computing four- or higher - point scattering amplitudes and checking whether duality is satisfied .
in particular , it would be interesting to work out some four - point amplitudes with two twisted - sector states and two untwisted - sector states .
one could then check for duality by looking at the spectrum of poles in the @xmath319-channel to see if it matched the spectrum of the untwisted sector for some values of the twisted - sector intercepts , while factorizing in the @xmath320-channel will give information on the physical twisted - sector spectrum and may help clarify the correct @xmath2-sector physical state conditions .
the issue of the correct set of physical state conditions in the @xmath1 sector is simpler than in the @xmath2 sector because of the mode identifications ( [ zfmodiden ] ) which imply that there is really only one independent fractional supercurrent in the @xmath1 sector . in particular , since @xmath321 , the @xmath112 intercepts must be related to each other by @xmath322 .
( recall that @xmath323 . ) also , by ( [ rsectzmalg ] ) , the @xmath170 intercept is related in a model - independent way to the @xmath111 intercept : @xmath324 this implies that @xmath325 thus ensuring that the dirac equation resulting from the @xmath326 physical state condition is automatically consistent with the mass - shell constraint coming from the @xmath327 physical state condition .
it is natural to ask whether the @xmath1-sector can be realized in the @xmath0 model described in the last section . from the discussion of section [ sii ] ,
the @xmath1 sector is characterized by the automorphism of the fsc algebra which interchanges @xmath328 .
this automorphism is extended to the whole @xmath0 cft by the _
transformations @xmath207 and @xmath210 .
thus the @xmath1 sector can be realized in the @xmath0 model , but as the twisted - sector of a @xmath3 orbifold of all five boson fields on the world - sheet .
this has unfortunate consequences for the physical interpretation of states in this sector since orbifolding the coordinate boson fields @xmath4 does not leave the generators of space - time translations @xmath329 invariant .
thus there are no translation - invariant states in the @xmath1 sector of the @xmath0 model .
however , one should not conclude from this that the @xmath1 sector is in general badly behaved from a space - time point of view
rather , this behavior is only a property of specific cft models of the spin-4/3 fsc algebra . as an example in support of this statement
, we briefly describe in this section a @xmath7 model in which the @xmath1 sector appears without orbifolding the coordinate boson fields .
unfortunately , this model has only one space - time dimension , so the resulting space - time physics is trivial ; however , we can still demonstrate the existence of @xmath1-sector physical states which are translationally - invariant in the one space - time dimension .
the @xmath7 model is written in terms of two free bosons , @xmath330 and @xmath331 , satisfying @xmath332 and @xmath333 , with @xmath331 compactified on the unit circle @xmath334 @xcite .
the vertex operators in the @xmath331-cft , @xmath335 for @xmath336 , have conformal dimensions @xmath337 , and carry the ( untwisted - sector ) @xmath26 charge @xmath338 mod 3 . denoting the dimension-1/3 and 4/3 operators by @xmath339 it is easy to check that the spin-4/3 fsc algebra currents are given by @xmath340 comparing to the expression ( [ grepd ] ) for the currents in the @xmath0 model , the important difference for our purposes is the absence of @xmath134 signs in front of the @xmath341 term in ( [ c2grep ] ) .
this means that the automorphism interchanging @xmath342 is realized in the @xmath7 cft by the transformation @xmath343 _ without _ any accompanying reflection of the @xmath330 coordinate boson .
thus , the @xmath1 sector states of the @xmath7 model are realized as states in the twisted sector of the @xmath3 orbifold of the single @xmath331-boson .
acting on this sector , @xmath331 has its mode expansion shifted in the standard way , @xmath344 with modes satisfying the commutation relations @xmath345 = ( 2r/3 ) \delta_{r+s}$ ] .
the @xmath346 zero mode commutes with everything , and so can be taken to be a constant , which we set to zero .
the basic vertex operators @xmath347 have the normal ordered expansion @xmath348 the normal ordering factors @xmath349 are required to insure the factorizability of amplitudes ( associativity of the operator product )
. from this expression acting on an arbitrary twisted - sector state @xmath176 one obtains the bypass relation @xmath350 , where @xmath24 .
in particular , the single - bypass relation for the basic fields are @xmath351 the last of which is precisely the defining bypass relation ( [ rbypass ] ) of the @xmath1 sector .
( more precisely , this is the bypass relation of the @xmath58 @xmath1 sector ; the @xmath352 @xmath1 sectors can be realized by letting the @xmath331 zero mode take the values @xmath353 and @xmath354 . ) in general , these bypass relations imply that @xmath3-even fields have only integer moding , while the @xmath3-odd fields have only half - odd integer modings . the twisted - sector ground state is non - degenerate .
we denote the corresponding twist field by @xmath355 and the twist ground state by @xmath356 .
the action of the zero mode of an arbitrary untwisted - sector operator on the twisted - sector ground state is simply @xmath357 .
since the ground states satisfy @xmath358 for @xmath253 , a simple computation analogous to ( [ dimcomp ] ) reveals that the conformal dimension of the twisted - sector ground state is @xmath359 .
all other twist fields are created from @xmath360 by the repeated action of the @xmath331 creation modes @xmath361 with @xmath362 .
thus the twist states have the spectrum of conformal dimensions @xmath363 for @xmath162 an integer .
acting on a twisted - sector state , the fsc currents @xmath15 have the mode expansion @xmath364 since the coordinate boson field @xmath330 is unaffected by the orbifolding procedure , its mode expansion is the usual one , as in eq .
( [ xbosexp ] ) . recalling the form of the currents ( [ c2grep ] ) , we obtain @xmath365 the lowest - level state in the @xmath1 sector is simply @xmath366 .
the non - trivial physical state conditions are [ rtauzero ] @xmath367 consistency of ( [ rtauzero]a ) with ( [ rtauzero]b ) implies @xmath368 and @xmath108 are related by @xmath369 which is equivalent to the relation ( [ rintrelgcr ] ) derived from the @xmath1-sector mode algebra .
the first excited state in the twisted sector is @xmath370 .
the non - trivial physical state conditions are : [ rtauone ] @xmath371 consistency of ( [ rtauone]b ) with ( [ rtauone]a ) implies the intercepts are related by @xmath372 although this is a different functional relation between @xmath368 and @xmath108 than appears in ( [ rintrelgcr ] ) , they have the common solution @xmath373 and @xmath374 , which is precisely realized when the condition ( [ rtauone]c ) is satisfied . at higher levels in the @xmath1 sector similar physical states will be found , all with specific values of the intercepts satisfying the relation ( [ rintrelgcr ] ) . _
a priori _ , there is no reason to expect all these states , or even an infinite subset of them , to have the same value of the intercept .
of course , in one space - time dimension duality does not require the existence of infinite towers of states .
we have thus shown that , to the same level of consistency , both the @xmath1 and @xmath2 sectors can be realized in models of the spin-4/3 fractional superstring .
it remains an open question whether either of these sectors is actually realized in a critical ( @xmath6 ) model of the fractional superstring .
it is a pleasure to thank k. dienes , j. grochocinski , z. kakushadze , a. leclair , and e. witten for useful discussions and comments .
would also like to thank the center for theoretical physics at m.i.t . , and b. zwiebach in particular , for their hospitality .
the work of p.c.a . was supported by nsf grant phy92 - 45317 and by the ambrose monell foundation , and the work of s .- h.h.t .
was supported in part by the national science foundation .
we take into account the action of the @xmath259 zero modes @xmath224 on the twist states . in particular ,
the twist ground state forms a representation of the nontrivial zero - mode algebra @xmath376=i\pi$ ] , and so is degenerate .
we denote this degeneracy by an index @xmath377 on the twist field @xmath243 and write for the twist ground state simply @xmath378 we choose these states to be eigenstates of @xmath379 ( the second component of @xmath224 ) : @xmath380 where , for the moment , @xmath377 can be any real number .
the zero - mode algebra @xmath376 = i\pi$ ] implies that @xmath381 and @xmath379 are conjugate variables , and that @xmath382 recall that the classical boson fields @xmath213 take values on the torus defined by the lattice identifications @xmath383 .
this implies , first of all , that only exponentials @xmath384 with @xmath385 should be considered ( since they are single - valued on the torus ) , and secondly , by ( [ bdegen ] ) that we should identify @xmath386 where @xmath387 will be determined momentarily . note that since @xmath388 = 0 $ ] for integer @xmath389
, @xmath390 is a constant .
the fact that the @xmath389 are constrained to be integers means that for each @xmath391 and every choice of @xmath387 there is a separate , inequivalent two - dimensional representation of the zero - mode algebra , consisting of the states @xmath392 and @xmath393 .
note that in terms of their @xmath379 eigenvalues , these two states differ by a half - lattice translation , as do the fixed points of the @xmath3 orbifold . in this way we match up with the familiar result that the number of ( ground state ) twist fields in an asymmetric ( chiral ) orbifold is the square - root of the number of fixed points of the orbifold action @xcite .
note also that the existence of these infinite number of representations of the zero - mode algebra is a reflection of the symmetry of the vertex operator algebra which takes @xmath394 , where @xmath390 and @xmath395 .
we now show that only three of these infinite number of inequivalent representations of the zero - mode algebra are consistent with the @xmath195 symmetry of the cft . in particular , the zero modes @xmath396 of the generators of the @xmath5 current algebra symmetry must obey the @xmath195 algebra @xmath397 = \varepsilon_{\mu\nu\rho}(u^\rho)_0 .\ ] ] using the definitions of the @xmath198 fields in terms of vertices ( [ covflds ] ) , and the identification of the zero mode of a vertex acting on the ground state as @xmath398 from ( [ twvtx ] ) where @xmath399 one can show , using the hausdorff formula @xmath400}$ ] , that the @xmath195 algebra ( [ bsoalg ] ) is satisfied only if @xmath401 acting on the ground states @xmath392 , using ( [ bzmreps ] ) and the hausdorff formula one can show that these equations reduce to @xmath402 where we have defined @xmath403 .
the only solutions to these equations are @xmath404 where @xmath50 and @xmath405 .
these three twisted - sector representations give rise to three inequivalent highest - weight modules of the fsc algebra . introducing the new notation @xmath406 for the twisted sector ground states to remove the fractional part of the index of the @xmath392 states : @xmath407 where @xmath245 or @xmath246 , we can write @xmath408 using the explicit form of the zero - mode representations given in ( [ zmreps ] ) and the expression ( [ btwzmod ] ) for the zero mode of a vertex acting on the twisted - sector ground state , it is easy deduce the zero mode actions ( [ gammazero ] ) and ( [ gammalg ] ) . electronic mail address : [email protected] . p.c .
argyres and s .- h.h .
tye , _ tree scattering amplitudes of the spin-4/3 fractional superstring i : the untwisted sectors _ , iassns - hep-93/57 , clns 92/1176 , hep - th/9310131 .
argyres , a. leclair and s .- h.h .
tye , _ phys .
lett . _ * 253b * ( 1991 ) 306 .
z. kakushadze and s .- h.h .
tye , _ ka and new determinants of fractional superconformal algebras , _ clns 93/1243 , hep - th/9310160 .
a.b . zamolodchikov and v.a .
fateev , _ sov .
j.e.t.p . _
* 62 * ( 1985 ) 215 ; _ sov .
j.e.t.p . _ * 63 * ( 1986 ) 913 .
zamolodchikov and v.a .
fateev , _ theor .
phys . _ * 71 * ( 1987 ) 451 .
a. neveu and j.h .
schwarz , _ phys .
* d4 * ( 1971 ) 1109 ; for a detailed explanation of this formalism , see , _ e.g. _ , chapter 7 of m.b .
green , j.h .
schwarz and e. witten , _ superstring theory , _ cambridge university press ( 1987 ) .
l. dixon , j.a .
harvey , c.vafa and e. witten , _ nucl .
phys . _ * b261 * ( 1985 ) 678 ; _ nucl . phys . _ * b274 * ( 1986 ) 285 ; s. hamidi and c. vafa , _ nucl .
phys . _ * b279 * ( 1987 ) 465 ; l. dixon , d. friedan , e. martinec and s. shenker , _ nucl . phys . _ * b282 * ( 1987 ) 13 .
k. itoh , m. kato , h. kunimoto and m. sakamoto , _ nucl .
_ * b306 * ( 1988 ) 362 ; m. sakamoto , _ phys .
lett . _ * b231 * ( 1989 ) 258 .
argyres , e. lyman and s .- h.h .
tye , _ phys .
_ * d46 * ( 1992 ) 4533 , hep - th/9205113 .
narain , m.h .
sarmadi and c. vafa , _ nucl .
phys . _ * b288 * ( 1987 ) 551 . | the spin-4/3 fractional superstring is characterized by a world - sheet chiral algebra involving spin-4/3 currents .
the discussion of the tree - level scattering amplitudes of this theory presented in ref .
@xcite is expanded to include amplitudes containing two twisted - sector states .
these amplitudes are shown to satisfy spurious state decoupling .
the restriction to only two external twisted - sector states is due to the absence of an appropriate dimension - one vertex describing the emission of a single twisted - sector state .
this is analogous to the `` old covariant '' formalism of ordinary superstring amplitudes in which an appropriate dimension - one vertex for the emission of a ramond - sector state is lacking .
examples of tree scattering amplitudes are calculated in a @xmath0 model of the spin-4/3 chiral algebra realized in terms of free bosons on the string world - sheet .
the target space of this model is three - dimensional flat minkowski space - time and the twisted - sector physical states are fermions in space - time .
since the critical central charge of the spin-4/3 fractional superstring theory is 10 , this @xmath0 model is not consistent at the string loop level .
epsf = 10000 |
evidence is accumulating that star - formation in galaxies and the occurrence of accretion - driven activity in their nuclei are astrophysically connected processes .
i ll give three prime lines of reasoning .
the masses of the spheroidal components in galaxies grow together with their massive black holes ( mbhs ) this is the well - known bulge black hole correlation .
nuclear accretion is often preceded by a phase of enhanced star - formation .
we suspected this from age dating of the host galaxies of nearby seyfert galaxies and qso s ( e.g. , gonzlez - delgado et al .
2001 , canalizo & stockton 2001 ) , but a wonderful proof of the symbiosis has recently been obtained by kauffmann et al .
( 2003 ) , using sdss data .
a remarkable plot from this work is reproduced in fig . 1 : this emission line ratio diagram shows the normal star - forming galaxy population , from low mass ( left ) to high mass ( bottom ) , with the `` abnormal '' populations of liner and seyfert galaxies branching off towards the top right .
the seyfert galaxies , having high [ oiii ] luminosities , were often found to display young stellar populations , i.e. , they are often post - starburst systems .
_ this is the plot all extragalactic astronomers the privileged herschel users in particular should be thinking about , concerning its origin and its cosmic evolution ! _ \3 .
the resemblance of the cosmic star - formation history plot and the qso space density plot , in particular the @xmath0 peak , may not be a coincidence .
also the nature of the distant submm galaxies , most likely containing a substantial agn - fraction ( e.g. , frayer et al .
2004 ) , supports an astrophysical link .
the following issues are also very relevant in this respect . \1 .
given the fact that @xmath1 qsos are being found , and that they possess very massive black holes , the bulge
black hole correlation just mentioned implies that massive bulges _ must _ be in place , within a gigayear after the big bang . \2 .
the emission line spectra of these @xmath1 qsos differ little or nothing from the spectra of more recent qsos : the abundances in the broad line gas are solar ( or even super - solar ) .
this is indicating that _ processed _ material is being accreted , and not pristine ism gas .
as reported in several recent studies , molecular gas is sometimes found in distant agn , both qsos and radio galaxies , in staggering quantities : there is plenty of fuel for star - formation . in summary
, it was long known that the two phenomena can co - exist ( e.g. , ngc1068 ! ) , but the study of this important star - formation agn connection evolving throughout cosmic time will be a major challenge for extragalactic astronomers in the coming decade .
as will be illustrated below , herschel ( and alma ) can and will play an important role in these investigations .
obviously this review can not be complete but it will hopefully raise interest from outside the already impressively large herschel communities .
as amply documented ( e.g. , soifer et al . 1987 ) , the all - sky iras mission opened the dusty universe , detecting dust radiation from many classes of extragalactic objects both expected and unexpected and classifying these dust properties .
ultraluminous infrared galaxies , ulirgs , emerged as an important class of obscured , gas - rich galaxies having their seds dominated by unusually strong far - infrared emission .
following in its footsteps , the observatory mission iso subsequently carried out more detailed investigations .
iso revealed the presence of cold dust ( @xmath215k ) in galaxies , and established the unexpected fast decay of starbursts .
mid- and far - infrared luminosities , colors , and temperatures were obtained for many objects in the local and nearby ( @xmath3 ) universe . in nearby galaxies ,
iso provided more insight in the physics of the star - forming regions , in particular the [ cii ] and [ oi ] cooling .
iso imaging of the warm dust in nearby galaxies proved to be a fine tool to address localized radiation properties .
iso revealed fascinating cases of obscured star - formation such as in the antennae ngc4038/4039 .
moreover , the spectroscopic capabilities of iso permitted great advance in our understanding of the gas properties of nearby dusty and gas - rich objects .
the mid - infrared aromatic features were discovered , but also the absorption characteristics , as well as the diagnostic emission lines and the cooling lines in the mid- and far - ir .
an excellent account of the iso legacy can be found in genzel & cesarsky ( 2000 ) .
nasa s spitzer mission is currently continuing this work , at much improved sensitivity ; impressive first results have appeared recently ( apjs vol .
154 no.1 issue ) , and some highlights were reported at this meeting .
spitzer will surely bring great advancement in our understanding of the spectral energy distributions ( seds ) of nearby and distant galaxies , including the magnificent 2-d information resulting from its infrared camera .
equally interesting results are to be awaited from astro - f , the japanese mission to be launched hopefully in 2005 .
astro - f will operate a sensitive large field - of - view near- and mid - ir camera , and will in addition carry out an improved all - sky far - infrared survey , fis , with a predicted yield of over @xmath4 extragalactic far - infrared sources a significant improvement with respect to iras .
the astro - f fis data will undoubtedly be important for herschel .
broadly speaking , herschel will continue where its predecessors stopped .
however , in addition to the fascinating achievements which one can foresee , herschel will obviously make completely unexpected discoveries : the telescope will point at known ( active ) objects but in addition of course follow - up on its own discoveries .
herschel will extend the iras / iso / spitzer / astro - f wavelength range into the submm , at much greater sensitivity , at considerably higher spatial resolution and at much higher spectral resolution .
these qualifications will permit studying the mid- and far - infrared out to high redshifts .
given that at these wavelengths star - formation and nuclear activity be it indirectly are optimally revealed , herschel will truly live up to its cornerstone expectations .
the paragraphs below summarize the key themes in this respect : dust and star - formation . _ in short , besides being the optimal far - ir spectrograph , herschel will become the perfect sed - machine . _ the optical infrared seds for early and late type galaxies differ most significantly in the mid- and far - ir . whereas the former are roughly flat ( in the f@xmath5 representation ) , or show a modest peak around 2@xmath6 m from stellar photospheric emission , the latter show a peak around 100@xmath6 m , indicative of cool dust .
starburst galaxies such as m82 show that peak more pronounced and at somewhat shorter wavelength ( @xmath2 60 80@xmath6 m ) , while the seds of ulirgs the champion fir - emitters are completely dominated by such a warm far - infrared peak , the latter contributing around 90% of the bolometric emission . the strength of this warm dust component is correlated with the optically thin synchrotron radio emission
this is the well known radio - fir correlation ( e.g. , condon 1992 ) .
far - infrared dust emission is also measured in active galaxies and agn . the excess thermal emission in seyfert-2 galaxy
ngc1068 is known since the 1970 s , but the systematic properties of this emission in seyfert galaxies had to await the iras survey . using iras data , de grijp et al .
( 1985 ) were able to show that the infrared color @xmath7 provides a good agn selector , that is to say : besides star - formation related warm dust , active galaxies are characterized with additional hot dust , which together yield relatively flatter infrared spectral indices . iso data permitted more elaborate multi - component dust models , both for star - forming objects ( e.g. , klaas et al .
1997 ) and active galaxies ( e.g. , prez garca et al .
( 1998 ) , spinoglio et al . 2002 ) .
moreover , the iso spectrographs allowed us to examine the richness of the mid- and far - infrared spectra .
these studies demonstrated the importance of the mid- and far - infrared wavelength range , assessing the relative and absolute strength of star - formation and nuclear activity , in a crude way , for nearby objects .
similar sed analysis tools were applied in case of more distant , more powerful agn , such as qsos , radio galaxies , and radio - loud quasars .
these studies were only moderately successful due to sensitivity constraints .
nevertheless , intriguing trends were reported such as the dusty nature of qsos ( e.g. , haas et al .
2003 ) and the star - formation connection ( van bemmel & barthel 2001 ) .
iso photometry was also applied tot test agn unification schemes ( van bemmel et al .
2000 , haas et al .
figure 2 illustrates the technique of multi - component dust model fitting , in the ( high s / n ) case of the nucleus of the cygnusa radio galaxy . recalling the substantial diffuse cold dust in galaxies
, at least three components must be present in active galaxies .
the hot circumnuclear torus dust however will display complicated aspect dependent optical thickness effects .
the latest torus models ( van bemmel & dullemond 2003 , nenkova et al .
2002 ) can not yet be properly tested with current sed data : high s / n seds for many classes of objects are eagerly awaited .
as mentioned already , the warm dust in galaxies draws from star - formation ( and also from torus emission ) .
a strong warm dust component can be inferred from infrared color - color plots and implies relatively strong star - formation activity .
this is obviously true for ulirgs , prime cases of starburst activity , but also qsos ( with an additional dust heating source , namely the agn ) can be put on such plots .
figure 3 indicates that indeed such star - forming agn can be identified : age - dating using optical spectra of the circumnuclear populations in these qsos ( canalizo & stockton 2001 ) supports their post - starburst nature , cf . expectation . during
the last few years evidence is growing that qso host galaxies may be richer in star - formation fuel than initially thought ( e.g. , evans et al . 2001 , scoville et al .
2003 ) .
herschel will address the symbiosis of mbh accretion activity and star - formation out to high @xmath8 , using the pacs and spire instruments .
this symbiosis is manifested in the global fir - submm spectral energy distributions as well as in the detailed fir - submm spectral properties .
given its ( one to two orders of magnitude ) improved sensitivity w.r.t .
previous missions , herschel will be able to measure the seds and the spectroscopic properties out to extreme redshifts , for many classes of objects .
the dust content and composition of many types of distant galaxies with and without agn come within reach .
analysis of the seds must obviously take the latest generation multi - component dust models into account .
key issues include : dust in early type galaxies , intra- and intercluster dust , dust in qso hosts ( are they really mature quiescent ellipticals ? ) , dust in radio galaxies , the cosmologically evolving dust properties and dust - to - gas ratio in galaxies , agn unification aspects ( including the search for obscured type-2 agn ) , the nature of sub - mm galaxies ( see below ) , etcetera
. the radio - fir correlation will be an important tool in these investigations , as it can help to isolate the star - formation related dust component .
deep radio observations at long wavelengths with sufficient angular resolution must provide the essential complimentary data .
the spectroscopic capabilities of herschel will be used to full advantage for distant sources to obtain the near- and mid - ir diagnostic line ratios which proved crucial to address the nature of the energy source in nearby starburst and active galaxies ( cf .
genzel et al .
the luminosities of critical low- and high ionization lines but also pah luminosities at intermediate and high redshift can be obtained from @xmath9 spectroscopy in about ten hours of exposure time .
submm galaxies may well be the distant ulirgs and their contribution to the infrared - submm background is significant ( e.g. , blain et al .
2002 ) . as such , examination of their seds and assessment of their energetics is a prime task , which spitzer has begun , but which herschel will do better , given its larger wavelength coverage .
herschel photometry in the far - infrared can be combined with alma mm - photometry to permit dust photo-@xmath8 assessment ( yun & carilli 2002 ) .
herschel may well turn out to be the only instrument capable of determining the precise redshift of obscured submm galaxies , from their molecular or ionic cooling lines .... this will require tens of hours of integration but is not unfeasible .
iso spectra have brought great advancement in our understanding of the star - formation process , both locally in the milky way and globally in nearby and low redshift galaxies . spectra taken with the spitzer irs are currently refining this knowledge and extending the distance coverage .
the improved angular resolution of herschel nevertheless is eagerly awaited for investigations of the details of the star - formation process : pdr cooling , the mass function of giant molecular clouds , the fragmentation , properties of super star clusters , etc .
herschel s hifi instrument will explore the chemistry of starburst evolution and differentiation , in the galaxy and in nearby galaxies .
template cooling line spectra containing the many co , h@xmath10o , and related lines will be obtained , and their application for distant star - forming objects will be investigated . the circumnuclear x - ray dominated regions ( xdrs ) in nearby active galaxies
will be investigated , again yielding template spectra .
pacs and spire will make imaging studies of the molecular cloud complexes , both in continuum and line ( including pah ) radiation .
pacs will obtain imaging spectroscopy of star - forming regions , focussing on the detailed physics of galactic pdrs and the star - formation
agn interplay in nearby ( active ) galaxies . concerning cooling , a lot of work must be done .
cooling through [ cii]@xmath11 and [ oi]@xmath12 is not well understood , and the [ cii ] deficit in strongly star - forming galaxies and ulirgs ( e.g. , malhotra et al .
1997 ) has now been measured at extreme redshift ( bolatto et al . 2004 ) .
taking full advantage of the herschel sensitivity , star - formation in early type galaxies will be investigated .
also the puzzling class of blue compact dwarfs will be studied : what is the origin of the low metallicity of these objects ?
herschel spectroscopic imaging observations may shed light on the driving force of nuclear activity : can we finally solve the bar issue , is there a direct physical link with circumnuclear star - formation ? also close to home , the details of the radio - fir correlation , including the small - scale break - down can be studied using herschel imaging .
such studies will refine the already impressive spitzer studies ( e.g. , gordon et al .
2004 ) . concerning distant and powerful starburst galaxies such as hylirgs , submm galaxies , lyman - break galaxies , and high @xmath8 ellipticals ,
herschel will undoubtedly quantify the level of star - formation and settle the question as to the presence of additional energy sources .
with help from the negative k - correction , galaxies having star - formation rate @xmath13 m@xmath14/yr can be observed out to @xmath15 , whereas sfrs of order @xmath16 m@xmath14/yr can be traced out to much higher redshifts .
large herschel key programs will be devoted to a census of the evolving starburst galaxy population , on the basis of data from the the large - area herschel photometric surveys . indeed
, these surveys will address the question `` how do galaxies form ? '' together with alma and other ground - based mm telescopes herschel will determine the cosmologically evolving star - formation efficiency sfe = l@xmath17/l@xmath18 .
star - formation in very high redshift agn , such as radio galaxies ( e.g. , greve et al . 2004 ) and qsos ( e.g , walter et al .
2003 ) will undoubtedly receive much herschel attention .
star - formation and agn activity were in the past at least one order of magnitude more prominent than they are now .
understanding these evolving phenomena and their interplay is a key theme of today s astrophysics . to obtain
that understanding is the raison dtre of the herschel space observarory .
herschel will deal with star and galaxy formation , throughout cosmic history , globally and in detail : mw , pdrs , gmcs , sscs , bcds , es , lbgs , eros , drgs , smgs , xdrs , qsos , prgs , lirgs , ulirgs and hylirgs , .... you name it !
these are exciting times and herschel is going to make it even more exciting .
those who wish to join in the planning of herschel studies are referred to the herschel website http://www.rssd.esa.int / herschel/.
the author acknowledges input from the herschel instrument science teams .
thanks also to guinevere kauffmann and ilse van bemmel , for permission to reproduce figs . 1 and 2 , and to marco spaans for comments on the manuscript .
blain a.w . , smail i. , ivison r.j . , et al . , 2002 ,
phys . rep . | herschel will represent a breakthrough in the study of nearby gas - rich and gas - poor galaxies , as it will for the first time permit imaging photometric and spectroscopic observations of their ism in the fir - submm wavelength range . the unprecedented sensitivity and angular resolution of herschel
will furthermore yield a breakthrough in our understanding of distant galaxies and agn , as their gas and dust both the ism- and the agn - related
will for the first time come within reach .
herschel will undoubtedly yield major discoveries concerning the cosmologically evolving gas and dust properties in galaxies , back to very early epochs .
[ 2001/04/25 1.1 ( pwd ) ] |
recent advances in nanofabrication technology increasingly enable the construction of devices operating in the quantum regime .
however , to utilize coherence effects for practical applications such as quantum information processing and communication tasks requires the ability to engineer their dynamics with high precision .
considerable progress in the area of laser technology and optimal control has shown that precise coherent manipulation of the dynamics is not infeasible for a variety of quantum systems , but such control requires accurate knowledge of the system s dynamical behavior and response to external fields , which can be used to construct accurate models from which effective control designs can be engineered .
the problem is particularly acute for manufactured systems , due to inevitable variations in the manufacturing processes , which ensure that the exact behavior of each device is unique and must be individually measured and characterized . for the manufacture of large - scale practical devices ,
the design and operation of each device should be as simple as possible meaning that the physical resources available to initialize and measure the state of a system are usually restricted to a single basis set defined by static electrode geometry . in normal operation , any state can be produced from an initial fiducial one by applying a suitable unitary rotation .
this also enables us to effectively perform measurements in an arbitrary basis , and given both these abilities , one can perform quantum process tomography @xcite .
however , the problem of characterizing a device is not trivial since initially , if one does not yet know the control response of the system , one can not generate the unitary rotations required in the first place , leading to a catch-22 situation .
what is required is a method of bootstrapping the control and characterization process so that the system dynamics and response can be incrementally assessed until full control and process tomography is possible , and only using the _ in situ _ resources .
hence , we have developed techniques based upon the analysis of generalized coherent oscillation data from rabi or ramsey - type experiments .
there are several approaches to the analysis of such experimental data including frequency - domain and time - domain analysis . in the regime of a single system transition ,
fourier analysis is effective but in the presence of multiple signals , it ceases to an optimal estimator . in previous work ,
we have shown how bayesian signal analysis can be effective in determining accurate model parameters in generic two - qubit hamiltonian systems where multiple frequencies are present . in this work ,
we extend the technique to systems with dephasing and use bayesian signal analysis to reconstruct the underlying dynamics , which are now non - unitary .
we apply this technique to three - level ( qutrit ) systems and analyze its performance for a range of dephasing rates and find that as long as coherent dynamics dominate , which would be the case for quantum information purposes , signal parameters can generally be reliably extracted and the system effectively reconstructed .
the evolution of a closed quantum system is governed by a time - dependent unitary operator @xmath0 obeying the schrodinger equation .
the evolution of an open quantum system can be highly complicated but under certain conditions it can be described by a master equation @xmath1+l_d\rho(t),\ ] ] where @xmath2=ab - ba$ ] is the usual matrix commutator , @xmath3 is a super - operator describing the interaction with the environment , and @xmath4 is a unit - trace positive operator @xmath4 on @xmath5 representing the state of the system .
this form of master equation is generally applicable to systems interacting with a memory - less ( markovian ) reservoir such as an effectively infinite bath , where it can be shown that the superoperator @xmath6 takes the form @xmath7\rho$ ] , where @xmath8 are operators on @xmath5 and the superoperators @xmath9 $ ] are defined by @xmath10\rho = v_k \rho v_k^\dag -\frac{1}{2}(v_k^\dag v_k\rho + \rho v_k^\dag v_k).\ ] ] under certain conditions we can make further simplifying assumptions . for example , dissipative effects in open systems weakly coupled to an environment are often dominated by a certain types of decoherence such as pure phase relaxation or population relaxation processes such as the spontaneous emission of photons or phonons .
these types of processes can be described by relatively simple master equations . in the case of pure dephasing the dissipation superoperator
is often determined by a single hermitian operator @xmath11 . in this case , it is easy to show that the master equation simplifies @xmath12 - \frac{1}{2 } [ v,[v,\rho(t)]].\ ] ] even with these simplifying assumptions on the open system dynamics we see that full system identification now requires the identification two generally independent hermitian operators @xmath5 and @xmath11 , which in general means the identification of @xmath13 real parameters .
fortunately , dephasing often acts in the eigenbasis of the hamiltonian , in which case @xmath5 and @xmath11 commute and are simultaneously diagonalizable , i.e. , there exists a basis @xmath14 such that @xmath15 where @xmath16 and @xmath17 are real , and in this case the identification problem reduces to finding a joint eigenbasis @xmath14 and the corresponding eigenvalues @xmath16 and @xmath17 of @xmath5 and @xmath11 , respectively .
this simplifies the problem .
if @xmath18 , @xmath19 and @xmath20 is the representation of the state in a joint eigenbasis of @xmath5 and @xmath11 then it is easy to see that the master equation ( [ eq : lme ] ) gives @xmath21 where @xmath22 and @xmath23 , i.e. , we have @xmath24 and if @xmath25 is the unitary basis transformation that maps the measurement basis to the joint eigenbasis of @xmath5 and @xmath11 , then the evolution of the density operator @xmath4 with respect to the measurement basis is given by @xmath26 . thus the evolution is determined by the transition frequencies @xmath27 , dephasing rates @xmath28 and the relation between the system and measurement basis @xmath25 , which are to be determined .
as in previous work @xcite we assume that we can prepare and measure the system in a fixed set of ( orthonormal ) computational basis states @xmath29 , where @xmath30 is the hilbert space dimension . no other measurements or resources such as non - basis states are assumed to be available initially .
the basic protocol is to prepare the system in a computational state , let it evolve for a period of time , then measure the probabilities that the system ends up in one of the computational basis states , repeating it for different times and all computational basis states .
the experimental data thus consists of @xmath31 time traces , @xmath32 , with @xmath33 , which represents the probability that the system , initially in state @xmath34 is measured in state @xmath35 after evolving under the system hamiltonian for time @xmath36 . if it was original initialized in @xmath37 . at long times , noise dominates the signal which leads to an optimal total sampling time.,scaledwidth=50.0% ] when we include dephasing in the hamiltonian eigenbasis , it can be shown that the observable probabilities are @xmath38 where the coefficients are [ eq : coeff ] @xmath39 here @xmath40 and @xmath41 are the amplitude and phase of the complex number @xmath42 and @xmath43 is the phase difference . if the hamiltonian is known to be real - symmetric in the computational basis , which is the case for many systems including atomic and molecular systems , where the off - diagonal elements of the hamiltonian are usually real transition strengths or dipole moments , and spin systems , the problem can be simplified .
the eigenvectors of a real - symmetric matrix are real , thus the phases @xmath41 must be multiples of @xmath44 so that @xmath45 , and since the sine of a multiple of @xmath44 vanishes , we have @xmath46 . in many cases the signs of the off - diagonal matrix elements are also known , e.g. for a spin chain in an anti - ferromagnetic material , the off - diagonal elements are positive , as the case for many atomic or molecular systems .
we then have @xmath47 with @xmath48 and @xmath49 , which further simplifies the reconstruction .
this shows that the identification problem for dephasing that acts in the system s natural basis is similar to the hamiltonian identification problem except that we also have to determine the dephasing rates @xmath28 . from the measurement results obtained from time
traces like in fig .
[ fig : datasignals ] , we must extract signal frequencies @xmath27 and damping rates @xmath28 as well as the amplitudes @xmath50 , @xmath51 and @xmath52 in order to be able to perform reconstruction of the system dynamics . we can do this again by bayesian estimation , maximizing the likelihood that a particular process generated the observed signal @xcite . for convenience
we label the transition frequencies of the system @xmath53 , assuming @xmath54 , and the corresponding dephasing rates @xmath55 , and define the vectors @xmath56 , @xmath57 , @xmath58 and @xmath59 where @xmath60 range from @xmath61 to @xmath30 and @xmath62 from @xmath61 to the number of transition frequencies @xmath63 . according to eq .
[ eq : probs ] , the traces should be linear combinations @xmath64 of the @xmath65 basis functions [ eq : case1 ] @xmath66 or in the case where @xmath5 is real - symmetric , the @xmath67 basis functions [ eq : case2 ] @xmath68 and our objective is to find parameters @xmath55 , @xmath53 , @xmath69 , @xmath70 and @xmath50 that maximize the likelihood of the measured data @xmath71.\ ] ] we can eliminate the explicit dependence on the linear coefficients @xmath72 , @xmath73 , @xmath50 and the noise variances @xmath74 by integration over suitable priors to obtain an explicit expression for the probability of a particular model given the observed data @xmath75 that depends only on the @xmath63 transition frequencies @xmath53 and corresponding dephasing rates @xmath55 . following standard bayesian analysis @xcite
we obtain @xmath76^{(m_b - n)/2},\ ] ] where the averages are defined by @xmath77 the components @xmath78 are essentially the orthogonal projections of the data onto a set of orthonormal basis vectors @xmath79 @xmath80 derived from the ( non - orthogonal ) basis functions @xmath81 defined above , evaluated at the respective sample times @xmath82 , via @xmath83 where @xmath84 is a @xmath85 matrix whose columns @xmath86 are the normalized eigenvectors @xmath87 of the @xmath88 matrix @xmath89 with @xmath90 thus , the parameter estimation problem for a system with decoherence acting in the hamiltonian basis is similar to that for a hamiltonian system , except that the sine and cosine basis functions for the bayesian analysis must be modified to damped sinusoids with unknown damping rates . the objective is to find the frequencies @xmath91 and damping rates @xmath92 that maximize @xmath93 , or equivalently , the log - likelihood function @xmath94.\ ] ] given a solution @xmath91 and @xmath92 that maximizes this log - likelihood , it can be shown that the corresponding optimal coefficients in the general case ( [ eq : case1 ] ) are @xmath95 where @xmath96 is shorthand notation for the expectation values @xmath97 of the linear coefficients of the basis functions , given the optimal frequencies @xmath91 and damping rates @xmath98 and the data @xmath75 . similarly in the special case ( [ eq : case2 ] ) @xmath99 since the log - likelihood function is sharply peaked with generally many local extrema , finding the global optimum using gradient - type optimization algorithms starting with a completely random guess for @xmath91 and @xmath92 is inefficient .
a global optimization such as pattern search or evolutionary algorithms might circumvent this problem , but neither proved either very effective in our case , especially for higher - dimensional search spaces .
alternatively , starting with a somewhat reasonable initial guess , especially for the frequencies , a standard quasi - newton optimization method with cubic line search @xcite proved generally very effective in finding the global maximum . to obtain an initial estimate for the frequencies we used the sum of the power spectra of the signals . although the peaks in the power spectrum are not optimal frequency estimators when there are multiple frequencies and the exact peak locations can be difficult to ascertain even for systems with only three frequencies , as fig .
[ fig : spectrum ] shows , rough estimates of the peak locations usually seem to provide a reasonable initial guess for the gradient - based likelihood optimization routine . in principle
the damping rates could be estimated from the peaks widths as well but these estimates can be tricky , especially for overlapping and minor peaks , hence we chose multiple runs with random initial guesses for the damping rates @xmath92 and selected the run with the highest final likelihood ( `` global '' maximum ) . given the extracted signal parameters we have to solve two further inverse problems : ( i ) reconstructing the level structure from the frequencies and ( ii ) constructing the matrix @xmath25 that relates the hamiltonian basis to the computational basis .
the former usually involves analyzing the relationships between the frequencies as illustrated in @xcite . in general
this is be tricky but for a qutrit system , is analysis is essentially trivial .
the basis reconstruction requires solving further optimization problems to find the coefficients @xmath100 such that eqs ( [ eq : coeff ] ) are satisfied given the estimates for the parameters @xmath101 , @xmath102 , @xmath50 and @xmath103 derived in the previous step .
due to finite sampling and noise , the inversion may not be exact , hence we recast it as a constrained optimization problem and solve it as described in @xcite .
our previous analysis @xcite also shows that we can only identify a single generic hamiltonian up to equivalence @xmath104 where @xmath105 is a diagonal unitary matrix , in the basis of the measurement . however , if the off - diagonal elements in the hamiltonian are known to be real and positive , for instance , then the hamiltonian will be uniquely determined up to a global energy level shift @xmath106 , at least in the generic case , since we have @xmath107 for @xmath108 . for a quantum control situation ,
the system dynamics can be controlled and hence different hamiltonians can be applied , and in the case subsequent hamiltonians can be fully determined up to the gauge fixed by the initial hamiltonian . by varying control parameters and tracking the change in the system dynamics , a dynamical control model can be built of the system .
we randomly generated 100 real - symmetric qutrit hamiltonians and dephasing operators with different spectral properties and the geometric average of the system @xmath109-factors ranging from 12 to 72 .
from these we generated various data traces corresponding to the stroboscopic sampling described in section [ sect : exp ] .
we considered three cases , the zero noise case ( @xmath110infinite samples per point ) , fixed finite sampling with @xmath111 experimental repetitions per time point , and an adaptive sampling strategy @xmath112 which varies the number of samples per point to reach an estimated target signal to noise ratio of @xmath113 for all @xmath60 and @xmath36 with an upper limit of @xmath114 for each data point .
we then applied our parameter estimation and reconstruction algorithms to the resulting data traces .
a range of dephasing rates was studied to see the effect on the reconstruction of the hamiltonian part of the dynamics . for the purposes of control ,
accurate determination of the hamiltonian is much more important than a precise determination of the dephasing rate , usually it suffices to know that they are below certain limits .
.median likelihoods ( @xmath115 ) and error rates ( @xmath116 ) for qutrit systems . for the 100 qutrit systems we compared the case with and without dephasing ( superscript h ) for different samples ( @xmath117 ) per data point . with no sampling noise @xmath118
, there was a small change in the median errors . for the @xmath119 case ,
the median errors increase due to the sampling noise , the addition of dephasing increases the final error by an order of magnitude to the @xmath120 region .
a simple adaptive scheme @xmath112 does similarly .
the hamiltonian is reconstructed using several runs of the optimization routine , and the solution with the minimum basis error is chosen .
[ cols="<,^,^,^,^,^,>",options="header " , ] table [ tab : median ] shows the median errors for various cases . comparing the dephasing / no dephasing cases , the errors are similar in the absence of projection noise @xmath121 ) .
the frequency @xmath122 estimation is slightly more accurate but estimation of the signal amplitudes @xmath51 is slightly less accurate since the basis functions depend on @xmath123 , hence errors in both @xmath122 and @xmath123 contribute to errors in the coefficients @xmath51 . for reduced signal to noise
, dephasing decreases the maximum likelihood and increases frequency , basis and reconstructed hamiltonian errors with a marked increase in median of the amplitude errors .
adaptive sampling overall increases the accuracy of the parameter estimation step and the reconstructed hamiltonian for both hamiltonian and dephasing systems but the improvement is more pronounced for dephasing systems
. this may be due to adaptive sampling being more beneficial for small signal amplitudes i.e. , decaying signals .
this suggests the use of adaptive sampling to increase the signal to noise ratio for samples at increasing times .
alternatively , the sample data can be weighted to give precedence to earlier samples .
further exploration of these methods will be the subject of future study .
dephasing leads to a reduction in signal at long times which can lead us to fitting noise . for strong dephasing ,
this leads to reduced accuracy in the estimation of the frequencies , and hence increased errors in the other parameters .
the spread in the fourier peaks can also lead to problems for closely spaced frequencies .
this in itself is not a problem _ per se _ for the bayesian parameter estimation step @xcite , except that it can lead to inaccurate initial search parameters coming from peak detection in the power spectrum .
this can be obviated somewhat by trying different initial parameters assuming that either of the two remaining peaks were doublets and using the most likely result . for systems of interest for quantum information processing
, the dephasing rates should be sufficiently low so that the damping of the rabi - type oscillations do not impact the scheme greatly .
for very small dephasing rates , however , it can be a problem if the algorithm overestimates the dephasing rates which means that the basis functions used are not suitable , and this is reflected in errors of the estimated amplitudes . for such systems , it is a simple enough matter to test models which are purely hamiltonian to see which gives the larger likelihood
. for various samplings shows a strong correlation and suggests that the total error in the hamiltonian is dominated by errors in the basis reconstruction step that comes about from the separate optimization of each basis function from the amplitude estimation , which may not lead to orthogonal data vectors.,scaledwidth=50.0% ] one factor which limits the reconstruction is that we may obtain a set of @xmath124 matrices @xmath125 , which ideally should be projectors onto orthogonal eigenspaces , but may not always form an orthogonal set of projectors .
we can quantify this basis error by @xmath126 fig .
[ fig : serror ] shows that there is a strong correlation between @xmath127 and the ( relative ) error in the final reconstructed hamiltonian .
thus , we can use @xmath127 to choose the best reconstructed hamiltonian from multiple optimization runs and as a rough indication of the likely accuracy of the reconstructed hamiltonian .
the data also suggests that there is little direct correlation between the likelihood and errors in the parameter estimation step and the final hamiltonian error , suggesting that the final error in the hamiltonian is dominated by errors in the basis reconstruction step .
the reconstruction step obviously depends on the parameter estimates obtained in the first step , and poor estimates for the parameters will generally result in large hamiltonian errors , but in some cases the basis reconstruction produces poor results even when the individual errors in the estimated parameters are small .
it should be possible to improve the reconstruction step by solving the @xmath31 optimization problems for the @xmath100 simultaneously rather than independently and enforcing orthonormality constraints for the basis vectors , but doing so would require solving a rather more complicated optimization problem with several nontrivial constraints .
other researchers have also begun to address the problem of system characterization with limited resources .
for example , leghtas et al .
@xcite also consider estimating parameters of three - level quantum systems using weak continuous population measurements .
however , in their case it is assumed that most of the system is already known including the transition frequencies and the precise structure of the hamiltonian , and there is no intrinsic decoherence .
they consider extracting only two real parameters of the system , the dipole transition strengths between levels 1 - 2 and 2 - 3 , which simplifies the problem enormously .
burgarth et al .
@xcite also consider hamiltonian characterization with restricted resources for heisenberg spin chains where only a small subset of spins are individually addressable .
the form and structure of the hamiltonian is known _ a priori _ to be of a particular class , and only the coupling strengths and anisotropy of the system hamiltonian are to be determined . the sign of the couplings is also known beforehand .
characterization is achieved in this case by preparing different initial states of the first spin , letting the system evolve and then performing quantum state tomography on the accessible spins .
if we consider a system of three spins , the first excitation subspace acts as a qutrit .
our protocol could be applied to this problem with some modifications .
our scheme does not require state tomography , only the determination of position of the up - spin , and there is no requirement to know the network topology .
it would be interesting to explore bayesian analysis of the response of such systems for hamiltonian characterization , and especially the role of topology in identifiability , and whether it is possible to relax the requirement for addressability of all spins . in summary
, we have shown that our current two - step procedure of bayesian parameter estimation followed by a reconstruction via optimization works in the presence of dephasing on three - level systems . however , we find that the reconstruction step is a weak point of our current implementation
. it may be possible to eliminate the parameter estimation step and directly apply bayesian maximum likelihood estimation upon the dynamical system parameters .
this would have the advantage of always giving admissible solutions at all steps .
another direction which should be explored is adaptive sampling , not only varying experimental repetitions per data point , but also using non - uniform time - domain sampling for better frequency discrimination . | we consider how to characterize the dynamics of a quantum system from a restricted set of initial states and measurements using bayesian analysis .
previous work has shown that hamiltonian systems can be well estimated from analysis of noisy data . here
we show how to generalize this approach to systems with moderate dephasing in the eigenbasis of the hamiltonian .
we illustrate the process for a range of three - level quantum systems .
the results suggest that the bayesian estimation of the frequencies and dephasing rates is generally highly accurate and the main source of errors are errors in the reconstructed hamiltonian basis . |
a knot is an embedding of a circle @xmath11 in a 3-manifold @xmath6 , usually taken to be @xmath12 or @xmath13 . in the 1920
s alexander and briggs @xcite and reidemeister @xcite observed that questions about ambient isotopy of polygonal knots in @xmath12 can be reduced to combinatorial questions about knot diagrams .
these are labeled planar graphs with overcrossings and undercrossings marked , representing a projection of the knot onto a plane .
they showed that any ambient isotopy of a polygonal knot can be achieved by a finite sequence of piecewise linear moves which slide the knot across a single triangle , which are called _ elementary moves _ ( or @xmath14-moves ) .
they also showed that two knots were ambient isotopic if and only if their knot diagrams were equivalent under a finite sequence of local combinatorial changes , now called _ reidemeister moves _ , see 7 .
a triangle in @xmath6 defines a _ trivial knot _ , and a loop in the plane with no crossings
is said to be a _ trivial knot diagram_. a knot diagram @xmath1 is _ unknotted _ if it is equivalent to a trivial knot diagram under reidemeister moves .
we measure the complexity of a knot diagram @xmath1 by using its _ crossing number _ , the number of vertices in the planar graph @xmath1 , see 7 .
a problem of long standing is to determine an upper bound for the number of reidemeister moves needed to transform an unknotted knot diagram @xmath1 to the trivial knot diagram , as an explicit function of the crossing number @xmath3 , see welsh @xcite .
this paper obtains such a bound .
[ thm101 ] there is a positive constant @xmath0 , such that for each @xmath15 , any unknotted knot diagram @xmath1 with @xmath3 crossings can be transformed to the trivial knot diagram using at most @xmath2 reidemeister moves .
we obtain the explicit value of @xmath16 for @xmath0 ; this value can clearly be improved . in this paper
we have not attempted to get an optimal value for @xmath0 , and several of our estimates are accordingly rather rough . in 1934
goeritz @xcite showed that there exist diagrams @xmath1 of the unknot such that any sequence of reidemeister moves converting @xmath1 to the trivial knot must pass through some intermediate knot diagram @xmath17 that has more crossings than @xmath1 .
it is not known how much larger the number of crossings needs to be for a general @xmath3-crossing knot , but theorem [ thm101 ] implies the following upper bound .
[ crossings.bound ] any unknotted knot diagram @xmath1 with @xmath3 crossings can be transformed by reidemeister moves to a trivial knot diagram through a sequence of knot diagrams each of which has at most @xmath2 crossings .
this holds because each reidemeister move creates or destroys at most two crossings , and all crossings must be eliminated by the final reidemeister move .
it seems possible that for each knot diagram representing the unknot there exists some sequence of reidemeister moves taking it to the trivial knot diagram which keeps the crossing number of all intermediate knot diagrams bounded by a polynomial in @xmath3 . in a general triangulated 3-manifold ,
a trivial knot is a triangle lying inside one tetrahedron and an elementary move takes place entirely within a single tetrahedron of the triangulation .
the proof of theorem [ thm101 ] is based on a general result that bounds the number of elementary moves needed to transform a closed polygonal curve @xmath4 embedded in the 1-skeleton of a triangulated @xmath5 3-manifold @xmath6 to a trivial knot .
[ thm102 ] there is a constant @xmath7 such that for each @xmath8 , any compact , orientable , @xmath5 3-manifold with boundary ( @xmath18 ) triangulated by @xmath9 tetrahedra has the following property : if @xmath4 is an unknot embedded in the 1-skeleton of interior ( @xmath19 ) , then @xmath4 can be isotoped to the trivial knot using at most @xmath10 elementary moves .
we obtain the explicit value @xmath20 . the hypothesis that @xmath4 lies in the interior of @xmath6
can be weakened to requiring only that @xmath4 lies in @xmath6 , as we indicate in 6 .
the methods we use are based on the normal surface theory of kneser @xcite and haken @xcite .
haken applied this theory to obtain an algorithm to decide if a knot is trivial .
see @xcite for a survey of algorithms to recognize unknotting , and @xcite and @xcite for possible alternate approaches based on braids .
the paper is organized as follows . in
2 we define elementary moves and outline the proof of theorem [ thm102 ] . in 3 , 4 and 5 we bound the number of elementary moves needed to isotop a curve in three distinct settings .
first we consider an isotopy of a curve to a point across a compressing disk in a 3-manifold , secondly we consider an isotopy through a solid torus from a core to a longitude , and finally we consider an isotopy across a surface between two isotopic curves . in 6 we tie these results together to complete the proof of theorem [ thm102 ] .
we then consider the more special case of knots in @xmath12 . in 7 we outline the proof of theorem [ thm101 ] and begin it by relating elementary moves to reidemeister moves . in 8
we start with a knot diagram and show how to construct a triangulation of a convex submanifold of @xmath12 containing the corresponding knot on its 1-skeleton .
this allows us to apply theorem [ thm102 ] . in 9
we complete the proof of theorem [ thm101 ] .
* note : * after completing this paper we learned that s. galatolo has obtained results about the case of reidemeister moves in @xmath12 similar to those stated in theorem [ thm101 ] and corollary [ crossings.bound ] , and at about the same time .
galatolo s constructions are also based on an analysis of normal surfaces .
an announcement of his results will appear in @xcite .
in this section we define elementary moves for knots and links .
elementary moves are well - defined in arbitrary triangulated @xmath5 3-manifolds . in section 7
we will discuss reidemeister moves , which are associated to knots and links in @xmath12 , and relate these two concepts for manifolds @xmath5-embedded in @xmath12 .
a ( piecewise linear , unoriented ) _ knot _ @xmath4 in a triangulated @xmath5 manifold @xmath6 is a closed embedded polygonal curve . an ( unoriented ) _ link _ @xmath21 is a finite union of nonintersecting knots in @xmath6 .
let @xmath22 denote the number of vertices of @xmath21 .
an _ elementary move _ ( or @xmath14-_move _ ) on a knot or link consists of one of the following operations , see burde and zieschang @xcite , murasugi @xcite .
each of these operations is required to take place in a single tetrahedron of @xmath6 .
\(1 ) split an edge into two edges by adding a new vertex to its interior .
( @xmath23 ) [ reverse of ( 1 ) ] combine two edges whose union is a line segment , erasing their common vertex .
\(2 ) given a line segment @xmath24 of @xmath21 that contains a single interior vertex @xmath25 and a point @xmath26 not in @xmath21 such that the triangle @xmath27 intersects @xmath21 only in edge @xmath24 , erase edge @xmath24 and add two new edges @xmath28 and @xmath29 and new vertex @xmath26 .
( @xmath30 ) [ reverse of ( 2 ) ] if for two adjacent edges @xmath28 and @xmath29 of @xmath4 the triangle @xmath27 intersects @xmath21 only in @xmath28 and @xmath29 then erase @xmath28 , @xmath29 and vertex @xmath26 and add the edge @xmath24 , with a new interior vertex @xmath25
. two types of elementary moves . ] corresponding to elementary moves of type ( 2 ) , ( @xmath30 ) is the piecewise - linear operation of linearly pulling the edges of a link across a triangle @xmath27 . to define this operation , an extra vertex @xmath25 on the segment @xmath24 must be specified ; this is the role of elementary moves ( 1 ) and @xmath31 .
two knots @xmath32 and @xmath33 are _ equivalent _ if one can be reached from the other by a finite sequence of elementary moves .
this gives the same equivalence classes as requiring @xmath32 and @xmath33 to be _ ambient isotopic _ @xcite in @xmath6 .
the _ unknot _ is the class of knots equivalent to a triangle contained in a tetrahedron , and a knot @xmath4 is _ unknotted _ if it is equivalent to the unknot .
equivalently , an unknot is the boundary of a @xmath5 embedded disk in @xmath6 @xcite .
we will work with knots embedded in the interior of @xmath6 .
this is not a significant restriction , since it is possible to perturb any knot which meets @xmath34 into the interior of @xmath6 after some simplicial subdivision .
we conclude this section by outlining the proof of theorem [ thm102 ] .
we first construct a compact triangulated @xmath5 submanifold @xmath35 by removing a solid torus neighborhood of @xmath4 . by removing the interior of a solid torus neighborhood @xmath36 of @xmath4
we construct a new triangulated manifold @xmath37 .
the boundary of @xmath35 consists of the original boundary @xmath34 plus a 2-torus @xmath38 , called the _ peripheral torus_. if @xmath4 is unknotted then @xmath35 contains an _ essential normal disk @xmath39_. essential means that @xmath40 is not the boundary of any disk in the torus @xmath38 . drawing on haken s normal surface theory , a result of hass , lagarias and pippenger @xcite implies that @xmath39 can be chosen to contain at most @xmath41 triangles .
we present this argument for completeness in 2 .
its boundary @xmath42 is a curve @xmath33 .
@xmath33 is isotopic to @xmath4 within @xmath43 .
there is a sequence of at most @xmath44 elementary moves which deform @xmath33 to a triangle across triangles in @xmath39 .
this is carried out in 3 .
it remains to relate @xmath33 to @xmath4 by elementary moves .
in 3 we construct a curve @xmath32 isotopic to @xmath4 in @xmath45 which is a _ longitude _ , a curve on the peripheral torus , @xmath38 , which bounds an embedded disk in @xmath35 but not inside the peripheral torus .
such curves exist if @xmath4 is unknotted in @xmath6 . in
4 we construct a longitude @xmath33 that lies in the 1-skeleton of the peripheral 2-torus @xmath38 , consists of at most @xmath46 line segments and can be isotoped to @xmath4 using at most @xmath47 elementary moves . in
5 we prove results which are used to show that there is a sequence of at most @xmath48 elementary moves which deform @xmath32 to @xmath33 while remaining entirely in the peripheral 2-torus @xmath38 . in 6 we combine these results to complete the proof .
all the constants and bounds above are explicitly computed .
the proof of theorem [ thm102 ] in effect deforms the original knot @xmath4 to a trivial knot with elementary moves made inside three surfaces : an annulus is used to go from @xmath4 to the longitude @xmath32 , a 2-torus is used to go from @xmath32 to the boundary of a normal disk @xmath33 , and a disk is used to go from @xmath33 to the trivial knot .
in this section we assume given a compact @xmath5 3-manifold with ( possibly empty ) boundary ( @xmath49 ) , which is triangulated using @xmath9 tetrahedra . we will work with a knot @xmath4 embedded in the 1-skeleton of interior@xmath50 . by barycentrically subdividing @xmath6 twice ,
we obtain a manifold @xmath51 in which @xmath4 has a regular neighborhood @xmath52 .
the regular neighborhood , obtained by taking the closed star of @xmath4 in @xmath53 , is a solid torus with @xmath38 disjoint from @xmath34 .
we let @xmath54 and note that its boundary @xmath55 .
we call the manifold @xmath56 with @xmath38 marked a _ truncated knot complement _ for @xmath4 . a _ meridian _ on the peripheral torus @xmath38 is an oriented simple closed curve @xmath57 whose homology class @xmath58 $ ] in @xmath59 generates the kernel of the map @xmath60 where @xmath61 is the inclusion map .
equivalently , it is a simple closed curve which bounds a disk in @xmath52 but not in @xmath38 .
a _ longitude _ on the peripheral torus @xmath38 is an oriented simple closed curve @xmath62 which intersects a meridian transversely at a single point and whose homology class @xmath63 $ ] in @xmath59 generates the kernel of the map @xmath64 where @xmath65 is the inclusion map , provided that this kernel is nontrivial .
note that the kernel is always infinite cyclic or trivial @xcite .
if the core @xmath4 of the solid torus @xmath52 is unknotted in @xmath6 , then a longitude exists , and it bounds a disk in @xmath35 . for general @xmath4 ,
a longitude does not exist when @xmath4 represents an element of infinite order in @xmath66 .
if @xmath67 , or more generally a homology sphere , then a longitude exists for any @xmath4 .
this curve is sometimes called a _ preferred longitude_. when a longitude @xmath62 of @xmath38 exists , its intersection number on @xmath38 with a meridian @xmath57 is @xmath68 , and the homology classes @xmath58 $ ] and @xmath63 $ ] generate @xmath69 .
a _ properly embedded disk _ in a 3-manifold with boundary @xmath70 is a disk @xmath39 pl - embedded in @xmath6 which satisfies @xmath71 .
essential disk _ in @xmath6 is a properly embedded disk such that @xmath42 does not bound a disk in @xmath34 , i.e. the homotopy class of @xmath42 in @xmath72 is non - trivial .
[ knot ] let @xmath35 be a triangulated @xmath5 3-manifold which is a truncated knot complement for @xmath4
. then @xmath4 is unknotted if and only if @xmath35 contains an essential disk @xmath39 with @xmath42 a longitude in @xmath38 .
[ [ proof . ] ] * proof . * + + + + + + + + + if @xmath4 is unknotted then it bounds a disk which is embedded in @xmath6 .
this disk can be intersected with @xmath35 to give an essential @xmath5 disk @xmath39 in @xmath35 .
conversely if @xmath35 contains an essential @xmath5 disk @xmath39 with @xmath73 , then the boundary @xmath42 of @xmath39 represents a longitude on @xmath38 , which is isotopic to @xmath4 in @xmath52 ( * ? ? ?
* , thorem 3.1 ) , and therefore is equivalent to @xmath4 .
this disk can be used to contract @xmath42 until it becomes a single triangle in @xmath35 .
the first homology group of a torus , @xmath74 , is canonically isomorphic to its fundamental group , and the statement about the homology class of @xmath75 $ ] is a consequence.@xmath76 we will use normal surface theory in the form described in jaco and rubinstein ( * ? ? ?
* section 1 ) . a _ normal surface _
@xmath39 in a triangulated compact 3-manifold @xmath6 is a @xmath5-surface whose intersection with each tetrahedron in @xmath6 consists of a disjoint set of _ elementary disks_. these are either triangles and quadrilaterals .
a quadrilateral consists of two triangles glued together along an edge lying interior to the tetrahedron .
all other edges of each triangle or quadrilateral are contained in the 2-skeleton of the tetrahedron , and do not intersect any vertices .
( see figure [ normal ] ) .
elementary disks in a normal surface . ] within each tetrahedron of @xmath6 there are only 7 kinds of possible triangle or @xmath5 quadrilateral , up to @xmath5-isotopies that map the tetrahedron to itself . these are labeled by the 7 possible ways of partitioning the vertices of the tetrahedron into the two nonempty sets determined by cutting the tetrahedron along the triangle or quadrilateral .
w. haken observed that the isotopy - type of a normal surface is completely determined by the number of pieces of each of the 7 kinds that occur in each tetrahedron .
this can be described by a nonnegative integer vector @xmath77 , which gives the _ normal coordinates _ of @xmath39 .
he also showed that the set of allowable values @xmath78 for a normal surface @xmath39 lie in a certain homogeneous rational cone @xmath79 in @xmath80 which we call the _
haken normal cone_. if @xmath81 , then haken s normal cone is specified by a set of linear equations and inequalities of the form @xmath82
@xmath83 the second set of equations expresses _ matching conditions _ which say that the number of edges on a common triangular face of two adjacent tetrahedra coming from regions in each of the tetrahedra must match .
for each triangular face there are three types of edges ( specified by a pair of edges on the triangle ) , which yields 3 matching conditions per face .
triangular faces in the boundary @xmath34 give no matching equations .
the cone @xmath79 is _ rational _ because all equations [ eq401 ] , [ eq402 ] have integer coefficients .
we let @xmath84 which is the set of integral vectors in the haken normal cone @xmath79 .
haken defined a _ fundamental surface _ to be a normal surface @xmath39 such that @xmath85 such a vector @xmath78 is called an element of the _ minimal hilbert basis _
@xmath86 of @xmath79 , using the terminology of integer programming , see schrijver ( * ? ? ?
* theorem 16.4 ) . [ haken ] ( haken )
if a triangulated , irreducible , orientable , compact 3-manifold with boundary ( @xmath49 ) contains an essential disk whose boundary is in a component @xmath87 of @xmath34 , then it contains such an essential disk @xmath39 with boundary in @xmath87 which is a fundamental normal surface .
[ [ proof.-1 ] ] * proof . *
+ + + + + + + + + see jaco - rubinstein ( * ? ? ?
* theorem 2.3 ) for a treatment of the existence of an essential normal disk in @xmath6 with boundary on @xmath87 . for a fundamental disk
, we refer to jaco and tollefson ( * ? ? ? * corollary 6.4 ) , which establishes the stronger result that there exists such a disk that is a fundamental surface and is also a _
vertex _ surface , which means that @xmath78 lies on an extreme ray of haken s normal cone @xmath79 .
@xmath76 [ [ remark . ] ] remark .
+ + + + + + + this result applies in arbitrary 3-manifolds .
the proof in @xcite assumes compactness and irreducibility of @xmath6 , but these hypotheses can be removed .
a classical theorem of alexander implies that knot complements in the 3-sphere are always irreducible .
hass , lagarias and pippenger @xcite give a simple bound for the complexity of any fundamental surface , whose proof we include for completeness .
[ tetra ] let @xmath6 be a triangulated compact 3-manifold , possibly with boundary , that contains @xmath9 tetrahedra
. * any vertex minimal solution @xmath88 of the haken normal cone @xmath89 in @xmath80 has @xmath90 * any minimal hilbert basis element @xmath88 of the haken fundamental cone @xmath89 has @xmath91 [ [ proof.-2 ] ] proof .
+ + + + + + ( 1 ) choose a maximal linearly independent subset of matching condition equations [ eq402 ] .
there will be @xmath92 of them , where @xmath93 .
any vertex ray is determined by adjoining to these equations @xmath94 other binding inequality constraints @xmath95 with the proviso that the resulting system have rank @xmath96 .
these conditions yield a @xmath97 integer matrix @xmath98 of rank @xmath96 , and the vertex ray elements @xmath99 satisfy @xmath100 in order to get a feasible vertex ray in @xmath89 all nonzero coordinates @xmath101 must have the same sign .
at least one of the unit coordinate vectors @xmath102 , @xmath103 must be linearly independent of the row space of @xmath98 . adjoin it as a first row to @xmath98 and we obtain a full rank @xmath104 integer matrix @xmath105 , \quad \det ( \tilde{{\mbox { m } } } ) \neq 0 ~.\ ] ] consider the adjoint matrix @xmath106 , which has integer entries @xmath107 ~ , \label{eqn39}\ ] ] in which @xmath108 $ ] is the minor obtained by crossing out the @xmath109 row and @xmath110 column of @xmath111 .
let @xmath112^t\ ] ] be the first column of @xmath113 . since @xmath114 this yields @xmath115 and @xmath116 because @xmath117 .
we bound the entries of @xmath118 using _
hadamard s inequality _ , which states for an @xmath119 real matrix @xmath120 that @xmath121 in which @xmath122 is the euclidean length of the @xmath110 row @xmath123 of @xmath120 .
we apply this to equation [ eqn39 ] , and observe that each row of the @xmath124 matrix @xmath125 $ ] has squared euclidean length at most 4 , because this is true for all row vectors in the system of equations [ eq402 ] and for @xmath126 . applied to equation [ eqn39 ] this gives @xmath127 however equation [ eqn39 ] shows that @xmath128 , and a vertex minimal solution @xmath129 in the extreme ray is obtained by dividing @xmath118 by the greatest common divisor of its elements , hence @xmath130 ( 2 ) a _ simplicial cone _
@xmath131 in @xmath80 is a @xmath132-dimensional pointed cone which has exactly @xmath132 extreme rays .
let @xmath133 be the vertex minimal solutions for the extreme rays .
each point in @xmath131 can be expressed as a nonnegative linear combination of the @xmath134 , as @xmath135 if @xmath129 is in the minimal hilbert basis @xmath136 of the cone @xmath131 , then @xmath137 , for otherwise one has @xmath138 and both @xmath139 and @xmath134 are nonzero integer vectors in @xmath131 , which is a contradiction .
thus , any minimal hilbert basis element @xmath140 of a simplicial cone satisfies @xmath141 the cone @xmath89 may not be simplicial , but we can partition it into a set of simplicial cones @xmath142 each of whose extreme rays are extreme rays of @xmath89 itself .
we have @xmath143 thus all hilbert basis elements of @xmath144 satisfy the bound [ eqn312 ] for hilbert basis elements of @xmath145 . using equation [ eqn35 ] to bound @xmath146
we obtain @xmath147 as claimed .
@xmath148 a @xmath5-embedded normal disk @xmath39 can be used as a template to transform its boundary @xmath42 to a single triangle by a sequence of elementary moves , in such a way that all intermediate curves lie on the surface @xmath39 .
[ piece ] let @xmath6 be a piecewise - linear triangulated compact 3-manifold with boundary , and let @xmath39 be a normal disk in @xmath6 that consists of @xmath149 triangles .
then @xmath42 can be isotoped to a triangle by a sequence of at most @xmath150 elementary moves on @xmath19 , each of which takes place in a triangle or edge contained within @xmath39 .
[ [ proof.-3 ] ] * proof .
* + + + + + + + + + we will construct a sequence of @xmath149 simple closed curves @xmath151 where @xmath152 is a triangle in @xmath39 , and such that for @xmath153 : 1 . @xmath154 is obtained from @xmath155 by at most two elementary moves ; either a move of type @xmath30 across a single triangle in @xmath39 followed by a move of type @xmath23 which removes the extra vertex created by the first move , or a move of type 1 followed by a move of type 2 across a single triangle in @xmath39 .
2 . there is a triangulated disk @xmath156 contained in @xmath39 which consists of @xmath157 triangles , such that @xmath158 .
the construction proceeds by eliminating triangles in @xmath39 one at a time by a sequence of steps , each consisting of two elementary moves , in such a way as to preserve the property that each @xmath159 is a topological disk .
we proceed by induction on @xmath160 , with the base case @xmath161 holding by hypothesis . for the induction step ,
we note that the only way that removing from @xmath159 a triangle that contains at least one edge of @xmath162 can yield a surface @xmath163 that is not a topological disk is by producing a vertex @xmath129 that is a cut - point , i.e. @xmath129 is visited twice in the curve @xmath164 . in that case @xmath129
is a vertex of @xmath162 , which implies that the triangle that is removed must have exactly one edge on @xmath162 , and the pair of elementary moves pulls this edge across to the two edges of this triangle that are adjacent to @xmath129 .
we therefore proceed by first checking if @xmath159 has any triangle having two edges on @xmath162 .
if so , we remove this triangle to obtain @xmath163 .
if this can not be done , then @xmath159 must contain an interior vertex , because a triangulated polygon that has no interior vertices and contains more than one triangle contains at least 2 triangles having two edges on the boundary . in this case
there exists a triangle having one edge on the boundary which also has an interior vertex , and we can pull across this triangle to obtain @xmath163 .
this completes the induction step .
@xmath76 putting all these results together yields : [ disk.count ] let @xmath35 be a triangulated @xmath5 3-manifold consisting of @xmath9 tetrahedra which is a truncated knot complement for @xmath4 . if @xmath4 is unknotted , then there is an essential normal disk @xmath39 in @xmath35 that contains at most @xmath41 triangles , and @xmath39 has a polygonal boundary @xmath165 which lies on the peripheral 2-torus @xmath38 and consists of at most @xmath166 line segments .
the boundary @xmath32 of @xmath39 can be transformed to a single triangle by at most @xmath44 elementary moves , with all intermediate curves being the boundary of some triangulated subdisk of @xmath39 .
[ [ proof.-4 ] ] * proof . * + + + + + + + + + by lemma [ knot ] and theorem [ haken ] there exists an essential normal disk which is a fundamental surface for @xmath35 .
the bound of lemma [ tetra ] implies that this surface contains at most @xmath167 triangles and quadrilaterals .
since each quadrilateral consists of two triangles in the triangulation , we obtain the upper bound @xmath168 for the total number of triangles .
the boundary @xmath42 contains at most two edges of each triangle , for a total of at most @xmath169 edges .
lemma [ piece ] gives the stated bound on elementary moves .
@xmath76 [ [ remark.-1 ] ] remark .
+ + + + + + + we could improve the above estimate by using a vertex minimal solution which is a disk .
in this section we bound the number of elementary moves needed to isotop a knot @xmath4 in the 1-skeleton of a triangulated compact 3-manifold @xmath70 across an annulus to a curve @xmath33 which is a longitude on the peripheral torus @xmath38 of @xmath4 .
the main result of this section is the following : [ longitude.count ] let @xmath70 be a triangulated compact @xmath5 3-manifold with @xmath9 tetrahedra , let @xmath4 be an unknotted curve embedded in interior@xmath50 , and let @xmath38 be the peripheral torus of @xmath4 in the second barycentric subdivision of @xmath6 .
then there exists a @xmath5 closed curve @xmath33 in @xmath38 which is a longitude for @xmath38 in @xmath35 such that : ( i ) .
there is an embedded @xmath5 annulus @xmath39 in the solid torus @xmath52 whose two boundary components are @xmath33 and @xmath4 , and which consists of at most @xmath170 triangles .
there is an isotopy from @xmath33 to @xmath4 across @xmath39 that consists of at most @xmath171 elementary moves . [
[ remarks . ] ] remarks .
+ + + + + + + + the closed curve @xmath33 is embedded in @xmath38 but generally does not lie in its 1-skeleton . in the special case
that @xmath70 is @xmath5-homeomorphic to a triangulated convex polyhedron in @xmath12 , as occurs with a standard knot , the exponential bounds in ( i ) , ( ii ) can be improved to bounds polynomial in @xmath9 , as we indicate after the proof .
the proof of theorem [ longitude.count ] proceeds in three steps .
we first construct a closed curve @xmath172 embedded in the 1-skeleton of @xmath52 which is _ parallel _ to the core @xmath4 , i.e. its homology class @xmath173 $ ] in @xmath174 has intersection number @xmath68 with a meridian @xmath175 $ ] .
we next observe that the homology class in @xmath174 of a longitude @xmath33 is necessarily @xmath176 = \pm ( [ \alpha ] + k [ \mu ] ) \label{eq401a}\ ] ] for some integer @xmath177 .
the second step is to bound @xmath178 by an exponential function of @xmath9 .
the third step is to obtain @xmath33 from @xmath172 by adding @xmath177 twists to @xmath172 near a fixed meridian @xmath57 .
given @xmath33 , we construct a triangulated annular surface @xmath39 embedded in @xmath52 which lies in the 2-skeleton of @xmath52 except in a `` collar '' of one meridian disk , into which the @xmath177 twists are inserted .
all vertices of the triangulated annulus @xmath39 lie on its boundary @xmath179 .
finally we obtain an isotopy giving the elementary move bound by pulling @xmath4 across the triangles in @xmath39 to @xmath33 .
we begin with some preliminary lemmas involving @xmath5 topology , which pin down some of the combinatorial structure of the regular neighborhood of @xmath4 .
background for @xmath5 topology can be found in @xcite or .
let @xmath180 be the vertices of the knot @xmath4 in @xmath6 .
since @xmath52 lies in the second barycentric subdivision of @xmath6 , each edge @xmath181 $ ] of @xmath4 is subdivided into four edges by the addition of three new vertices .
we will continue to denote the subdivided curve by @xmath4 .
denote the original vertices @xmath182 by @xmath183 and let @xmath184 , @xmath185 denote the new vertices added in the two barycentric subdivisions .
we also use the convention that @xmath186 , where @xmath187 is the total number of vertices , so that the vertices @xmath188 of @xmath4 are cyclically ordered .
we will break up the solid torus @xmath52 into a union of what we call _ wheels _ and _
wedges_. for each edge @xmath189 $ ]
let the _ wheel _ around that edge
be the closed star @xmath190 , r_k ) $ ] of the edge @xmath189 $ ] in the solid torus @xmath52 .
this is the union of all the tetrahedra in @xmath52 having @xmath189 $ ] as an edge , and is homeomorphic to a closed ball . for each vertex @xmath188 on @xmath4
wedge around @xmath188 _ be the union of all the tetrahedra in @xmath52 meeting @xmath4 only at the point @xmath188 .
let @xmath191 , r_k ) ~ , \quad 1 \leq k \leq s~,\ ] ] be the link of the edge @xmath189 $ ] in the solid torus @xmath52 . .
] [ lem.meridian ] ( 1 ) each @xmath192 lies in @xmath38 and is a meridian .
+ ( 2 ) @xmath192 and @xmath193 coincide if @xmath194 is odd and @xmath195 or if @xmath194 is even and @xmath196 .
otherwise @xmath192 and @xmath197 are disjoint .
+ ( 3 ) an edge on @xmath38 either lies on some @xmath192 or else lies on a unique triangle in @xmath52 whose opposite vertex @xmath149 lies on @xmath4 .
[ [ proof.-5 ] ] * proof .
* + + + + + + + + + \(1 ) the closed star @xmath190 , r_k ) $ ] is supported on the set of tetrahedra in @xmath52 having @xmath189 $ ] as an edge .
each such tetrahedron has one edge which is disjoint from @xmath189 $ ] , and the link @xmath192 is the union of these edges .
the surface @xmath198 which is obtained by coning @xmath192 to the vertex @xmath188 forms a topological disk in @xmath52 whose boundary is @xmath192 , and which intersect @xmath4 transversely in a single point .
thus @xmath192 is a meridian ; see figure [ meridian ] .
\(2 ) this follows since @xmath52 is a barycentric subdivision .
the vertices @xmath188 with @xmath194 odd were added during the second barycentric subdivision . prior to the second subdivision ,
these vertices lay along interior points of edges of @xmath4 .
the two meridians associated to the edges @xmath199 $ ] and @xmath200 $ ] coincide ; see figure [ section ] for a cross sectional sketch of @xmath52 . for @xmath194 even , the distance between @xmath201 and @xmath202 in @xmath203 is at least four , since the link has been barycentrically subdivided . if @xmath204 then this distance equals two .
\(3 ) each edge in @xmath38 is an edge of a triangle with a vertex @xmath188 on @xmath4 , since @xmath45 is obtained by taking the closed star of @xmath4 in @xmath53 .
if the edge is in a meridian @xmath192 , then it lies in triangles with vertices at both @xmath201 and @xmath188 or @xmath188 and @xmath202 .
if it does nt lie on a meridian then it is contained in the wedge around @xmath188 .
any triangle containing the edge and a point @xmath205 on @xmath4 can not cross the two wheels meeting @xmath188 , and the point @xmath205 must coincide with @xmath149 .
@xmath76 adjacent meridians .
the index @xmath177 is odd . ]
[ parallel curve ] there exists an embedded closed curve @xmath172 in the 1-skeleton of @xmath38 which is parallel to @xmath4 in @xmath52 .
there is an annulus @xmath206 embedded in the 2-skeleton of the solid torus @xmath52 whose boundary components are @xmath172 and @xmath4 , and all vertices of @xmath206 lie in its boundary @xmath207 . [ [ proof.-6 ] ] * proof . * + + + + + + + + + we construct @xmath172 as the union of a sequence of arcs . for each odd @xmath208 , we pick a path @xmath209 in the 1-skeleton of @xmath38 that connects a vertex @xmath210 of @xmath211 to a vertex @xmath212 on @xmath213 , does not cross @xmath214 , and is shortest among all such paths .
( we do nt allow @xmath209 to cross @xmath214 to prevent it from running between the curves in the wrong direction . )
we set @xmath215 as before .
note that @xmath209 is embedded and that it meets the collection of meridians only at its endpoints , since otherwise it can be shortened .
we then add for each odd @xmath194 a connecting path @xmath216 on @xmath192 from the endpoint @xmath217 of @xmath218 to the initial point @xmath210 of @xmath209 , and we set @xmath219 the path @xmath172 is a closed curve , and is embedded in @xmath38 .
we construct the surface @xmath206 as follows .
each path @xmath216 , @xmath220 , @xmath194 odd , lies on a meridian @xmath211 which is the boundary of a meridian disk @xmath221 centered at the vertex @xmath202 on @xmath4 , and we cone @xmath216 to @xmath202 along @xmath221 , obtaining a subdisk of @xmath221 . for each edge
@xmath209 that lies on the boundary of the wedge around the vertex @xmath202 , there is a unique triangle containing that edge and @xmath202 . for fixed @xmath194 , the union of these triangles
forms a disk in @xmath52 meeting @xmath4 at the point @xmath202 , and @xmath38 along @xmath222 , and whose boundary consists of @xmath222 together with two interior edges of @xmath52 , one running from the initial point @xmath217 of @xmath216 to @xmath202 and one running from the endpoint @xmath212 of @xmath223 to @xmath202 . finally for each meridian @xmath192
with @xmath194 odd , there is a pair of triangles @xmath224\ ] ] and @xmath225,\ ] ] whose union is a disk contained in the two wheels around @xmath226 $ ] and @xmath227 $ ] , and connecting the free edge running from the endpoint @xmath217 of @xmath218 to @xmath201 to the free edge running from the initial point ( also @xmath217 ) of @xmath216 to @xmath202 .
the union of all these disks forms an annulus which is embedded in the 2-skeleton of @xmath52 , whose boundary components are @xmath172 and @xmath4 , and all vertices of which lie in @xmath207
. see figure [ annulus ] .
@xmath76 + the annulus @xmath206 . ]
* proof of theorem [ longitude.count ] . *
we wish to convert the closed curve @xmath172 given in lemma [ parallel curve ] to a longitude @xmath33 by adding `` twists '' .
the homology class @xmath228 $ ] of a longitude @xmath33 satisfies @xmath229 = \pm ( [ \alpha ] + k [ \mu ] ) \ ] ] where @xmath57 is a fixed meridian . according to lemma [ lem.meridian ]
we may choose @xmath57 to be @xmath230 taken with a fixed orientation .
the number @xmath177 represents the number of `` twists '' that we must insert in @xmath172 to convert it to the knot @xmath33 .
we proceed to bound @xmath177 in terms of the number of tetrahedra @xmath9 in the original manifold @xmath6 .
we may assume @xmath231 for otherwise we are done .
the value of @xmath177 is uniquely determined by the requirement that the homology class @xmath173+k [ \mu ] $ ] vanish in @xmath232 .
( here we use the fact that @xmath58 $ ] is nonvanishing in @xmath232 . )
this is equivalent to the curve @xmath233 being a boundary in the cycle group .
the manifold @xmath35 contains at most @xmath234 tetrahedra , since it is contained in the second barycentric subdivision , hence the number @xmath235 of edges in @xmath35 is at most @xmath236 .
the group of boundaries in the @xmath5 chain complex of @xmath35 is generated by the paths around all 2-simplices in @xmath35 , and the number @xmath237 of these is at most @xmath238 .
we choose fixed orientations for the edges and faces , and we can then describe a basis of generators for the group of boundaries using an @xmath239 matrix @xmath240 $ ] with entries @xmath241 in which each column of @xmath98 has exactly three nonzero entries . let @xmath242 be the @xmath243-vector describing @xmath172 as an oriented cycle in @xmath35 , and @xmath244 be the @xmath243 vector describing @xmath57 as an oriented cycle in @xmath35 .
all entries in @xmath242 and @xmath244 are 0,1 or @xmath245 .
the condition for @xmath246 to be a boundary is that the linear system @xmath247 have a solution in integer unknowns @xmath248 .
the hypotheses imply that this linear system is solvable over @xmath249 for a unique value of @xmath177 . by taking a generating set of boundary faces
, we have that @xmath98 is of full column rank .
the meridian class @xmath244 is not in the column space of @xmath98 since otherwise @xmath57 would bound a surface in @xmath35 and the unknotted curve @xmath4 would represent an element of infinite order in @xmath66 .
it follows that there is at most one _ real _ value of @xmath177 for which the system ( [ eq407a ] ) is solvable , for if there were two then @xmath244 would be in the column space of @xmath98 .
thus @xmath244 is linearly independent of the column space of @xmath98 and the linear system @xmath250 \left [ { \displaystyle\frac}{{{\bf x}}}{-k } \right ] = { { \bf c}}_\alpha ~,\ ] ] has full column rank .
choosing a suitable @xmath251 rows of @xmath252 $ ] yields an invertible matrix @xmath111 and @xmath253 = \tilde{{\mbox { m}}}^{-1 } \tilde{{{\bf c}}}_\alpha ~,\ ] ] in which @xmath254 is the subset of @xmath242 determined by the same @xmath251 rows . by cramer
s rule we have @xmath255 ( \tilde{{{\bf c}}}_\alpha ) _
i ~,\ ] ] where @xmath256 $ ] denotes a minor of @xmath111 on the last column .
since @xmath257 we have @xmath258 ) .
each column of @xmath111 has at most three nonzero entries , each equal to @xmath68 , except the last column , which does not appear in the @xmath259 minor @xmath260 $ ] .
this yields the bound @xmath261 |^2 \leq 3^f ~ , ~~1 \le i \le f+1 ~,\ ] ] which with ( [ eq410a ] ) yields @xmath262 we now describe how the `` twists '' are added to @xmath172 and the surface @xmath206 of lemma [ parallel curve ] . to yield an annulus @xmath39 with boundary a longitude .
the disk @xmath263 cuts the core @xmath4 at the point @xmath264 and stretches to the boundary .
we add the twists to @xmath206 by modifying it inside the wedge around @xmath264 .
let @xmath265 denote the surface consisting of the set of closed triangles in the wedge around @xmath264 which intersect the meridian @xmath266 and which lie in @xmath38 .
then @xmath265 is an annulus , and we suppose it contains @xmath267 triangles .
the first edge @xmath235 of @xmath268 has one endpoint on @xmath269 and the other endpoint disjoint from @xmath266 and on the other boundary component of @xmath265 .
we replace the edge @xmath270 $ ] of @xmath172 with a path @xmath271 of @xmath272 edges which form a spiral going @xmath177 times around @xmath265 , starting at @xmath273 and ending at @xmath274 ; see figure [ annulus2 ] . to convert the surface @xmath206 of lemma [ parallel curve ] to the desired surface @xmath39
, we first subdivide the interval @xmath275 $ ] into @xmath272 segments by adding vertices @xmath276 .
we remove the triangle @xmath277 $ ] from @xmath206 and add @xmath272 new triangles , obtained by coning the edges of @xmath271 to @xmath264 .
this construction produces the surface @xmath39 , which is also an annulus and which also has its vertices either on @xmath4 or @xmath38 .
@xmath206 . ] adding twists replaces @xmath206 by @xmath39 .
] it is now straightforward to complete the proof . to prove ( i )
we observe that the number of triangles in @xmath39 is at most the number of triangles in @xmath52 , which is at most @xmath238 , plus the number @xmath278 of extra triangles added in the `` twisting '' step in constructing @xmath39 from @xmath206 .
the trivial bound for @xmath267 is the number of triangles in @xmath53 which is at most @xmath238 . combining this with ( [ eq412a ] ) yields at most @xmath279 triangles in @xmath39 . to prove ( ii ) we pull the curve @xmath4 across @xmath39 to @xmath33 one triangle at a time . using the fact that all vertices of @xmath39 are in the boundary @xmath42 , a suitable legal order of triangles to pull across
is easily found .
counting two elementary moves per triangle , this yields at most @xmath280 elementary moves in total .
@xmath76 + [ [ remark.-2 ] ] remark .
+ + + + + + + in the special case that the manifold @xmath6 is embedded in @xmath12 , as when we start with a standard knot complement , we can improve the bounds ( i ) , ( ii ) of theorem [ longitude.count ] to be polynomial in @xmath3 .
we sketch the idea .
the knot @xmath4 has at most @xmath238 edges in @xmath6 . using the embedding of the manifold @xmath6 in @xmath12 we can find a regular projection of the knot to a knot diagram @xmath1 , and this has @xmath281-crossings , as we show in lemma [ regular.proj ] of 7 .
we can specify a curve @xmath172 in the 1-skeleton of @xmath38 which is parallel to the core @xmath4 whose projection runs just to the right of @xmath4 in the projection plane .
choose an orientation of @xmath4 .
a longitude of a knot @xmath4 in @xmath12 is characterized by having a linking number with @xmath4 equal to zero .
the linking number is easily calculated from the projection . to each crossing of @xmath4
is associated a sign , and the sum of these signs , called the _ writhe _ of the knot , gives the linking number of a curve with @xmath4 . to ensure that the linking number is zero
, we alter the construction of @xmath172 . near every overcrossing of @xmath4 , we add a full twist of @xmath172 around @xmath4 .
this twist is positive if the sign of the crossing is negative , and vice versa .
the result is a knot @xmath33 in @xmath12 that lies on the peripheral torus whose linking number with @xmath4 is zero .
we can construct the associated annulus @xmath39 in @xmath52 with boundary components @xmath4 and @xmath33 similar to the proof given for theorem [ longitude.count ] , and it has @xmath281 triangles . in this argument
the key step is that the writhe is bounded in terms of the number of crossings of @xmath1 , which is @xmath282 , and this gives a bound @xmath283 in place of ( [ eq412a ] ) .
in this section we examine isotopies between two @xmath5 curves @xmath172 and @xmath284 on an oriented triangulated surface @xmath285 , and bound the number of elementary moves needed to move @xmath172 to @xmath284 . in the application , the surface will be the peripheral torus @xmath38 of a knot @xmath4 , @xmath172 will be @xmath32 , the boundary of a normal disk , isotopic on @xmath38 to @xmath284 , and @xmath284 will be a fixed longitudinal curve @xmath33 , parallel to @xmath4 in its regular neighborhood @xmath52 .
we derive the results of this section for a general surface , since no extra work is needed .
our main result is the following .
[ surface.bound.thm ] let @xmath285 be a triangulated orientable surface with @xmath286 triangles , and let @xmath172 and @xmath284 be embedded isotopic @xmath5 curves on @xmath285 , each having at most @xmath287 line segments .
then there is an isotopy carrying @xmath284 to @xmath172 via no more than @xmath288 elementary moves .
[ surface.bound.cor ] let @xmath285 be a triangulated orientable surface with at most @xmath289 triangles , and let @xmath172 and @xmath284 be embedded isotopic @xmath5 curves on @xmath285 , each having at most @xmath290 line segments .
then there is an isotopy carrying @xmath284 to @xmath172 via no more than @xmath291 elementary moves .
we will work with specially positioned curves and isotopies , called _
basic _ , which we now define .
let @xmath292 be a unit edge length equilateral triangle .
let @xmath293 be an equilateral triangle of half the size ; see figure [ basic.arc ] .
a _ basic arc _ in @xmath292 is one of two types : ( a ) _ type 1 . _ an arc running between the interiors of two distinct edges of @xmath292 .
this type of arc contains three straight line segments .
the first segment runs from the initial point towards the center of @xmath292 until it meets @xmath293 .
the third segment runs from the final point of the arc towards the center of @xmath292 until it meets @xmath293 .
the middle line segment connects the two in @xmath293 .
+ ( b ) _ type 2 .
_ an arc with both of its endpoints on the interior of a single edge of @xmath292 .
this type of arc contains two straight line segments , each of which makes an angle of @xmath294 with the edge .
+ arcs of type 2 are disjoint from any arcs of type 1 with non - linking endpoints , since an arc of type one makes an angle larger than @xmath295 with an edge of @xmath292 , and arcs of type 2 are disjoint from the smaller equilateral triangle .
now let @xmath293 be any triangle and fix a linear map from @xmath292 to @xmath293 .
a _ basic arc in @xmath293 _ is defined to be the image of a basic arc in @xmath292 .
type 1 and type 2 basic arcs . ]
a _ basic curve _ in a triangulated surface is a curve which meets each triangle in a union of basic arcs .
[ basic.arcs ] any two basic arcs in a triangle with distinct endpoints which lie on the interiors of the edges of @xmath292 intersect in either zero or one point .
any collection of disjoint embedded curves in a triangle can be isotoped , without moving their endpoints , to a collection of embedded basic arcs .
[ [ proof.-7 ] ] * proof . * + + + + + + + + + two basic arcs with distinct endpoints intersect if and only if their endpoints link on the boundary of the triangle .
any family of disjoint embedded arcs in a triangle is isotopic , rel boundary , to any other such family . @xmath76
any set of arcs can be isotoped to a set of basic arcs . ]
a _ basic move _ of a basic curve in a triangulated surface is defined to be one of the following curve isotopies , each of which produces a new basic curve : [ basic.moves ] ( a ) _ type 1 . _
an isotopy that slides an arc of the curve that crosses an edge along that edge .
( b ) _ type 2 . _ an isotopy that isotops an arc of the curve that crosses an edge of the triangulation across an adjacent vertex , and then isotops each of the arcs of intersection with a triangle created or altered in this process to basic arcs .
( c ) _ type 3 .
_ an isotopy that slides an arc of the curve with both endpoints on one edge of a triangle @xmath292 across the 2-gon it forms with that edge .
this results in the joining of two arcs in an adjacent triangle @xmath293 .
if the two arcs give a closed curve in @xmath293 , the closed curve is removed .
otherwise , this new arc is isotoped in @xmath293 to a basic arc .
we count the number of elementary moves required by each basic move .
we first look in a single triangle .
[ polygonal.delta ] let @xmath296 and @xmath297 each be a union of @xmath3 connected polygonal arcs in a triangle with the endpoints of @xmath209 and @xmath216 equal , and having a total of @xmath298 and @xmath299 edges respectively .
then @xmath172 can be isotoped to @xmath284 in the triangle using at most @xmath300 type 2 elementary moves , and @xmath301 elementary moves in total .
[ [ proof.-8 ] ] * proof . *
+ + + + + + + + + we will count the number of required elementary moves in two cases . in each case
we will isotop both @xmath172 and @xmath284 to an intermediate curve @xmath302 , which stays close to the boundary of the triangle . to construct @xmath302 ,
we first pick @xmath303 to be a positive angle which is small enough so that for any vertex @xmath274 in @xmath304 which lies on an edge of the triangle and any vertex @xmath149 in @xmath305 which does not lie on that edge , the angle between that edge and the line segment @xmath306 is greater than @xmath303 . given an arc @xmath209 of @xmath172 , we construct an arc @xmath307 with the same endpoints , but with interior disjoint from @xmath172 and containing just two segments .
if @xmath209 runs between distinct edges of the triangle then it cuts off a subdisk containing a single vertex of the triangle .
construct @xmath307 to consist of two segments , each of which makes an angle of @xmath303 with the edges going towards that vertex . by definition of @xmath303
, @xmath307 will be disjoint from the interiors of @xmath209 and @xmath216 .
if @xmath209 has both endpoints on a single edge of the triangle , then it cuts off a subdisk containing no vertices and a segment of one edge .
construct @xmath307 to consist of two segments contained in this disk , each of which makes an angle of @xmath303 with its edge .
again , @xmath307 will be disjoint from the interiors of @xmath209 and @xmath216 . 1 .
@xmath172 and @xmath284 are connected .
+ if @xmath172 does not meet both sides of @xmath284 , then the region between them can be triangulated with @xmath308 triangles , which can be used to isotop @xmath172 to @xmath284 using @xmath308 type 2 elementary moves . if they intersect , then applying the previous argument shows that each can be isotoped to the disjoint curve @xmath302 by using @xmath298 and @xmath299 type 2 elementary moves respectively .
it follows that we can always isotop @xmath172 to @xmath284 using at most @xmath300 type 2 elementary moves , and @xmath301 elementary moves of all types .
2 . @xmath172 and @xmath284 are not connected .
+ we will construct an isotopy from @xmath172 to @xmath284 which keeps the curves embedded at each stage .
let @xmath309 be the number of segments of @xmath209 , so that @xmath310 .
each arc @xmath209 of @xmath172 cuts off a disk meeting exactly one vertex of the triangle in which it lies , if it runs between distinct edges of the triangle , or no vertices if it runs from an edge back to that edge .
an arc of @xmath172 is said to be _ outermost _ if it cuts off a disk which contains no other arcs of @xmath172 .
let @xmath209 be outermost .
then @xmath307 is disjoint from all the other arcs in @xmath305 and meets @xmath311 only in their endpoints .
the number of segments of @xmath307 is two , and @xmath312 is embedded .
we can isotop the outermost arc @xmath209 of @xmath172 to the arc @xmath307 with the same endpoints without ever introducing any new intersections in @xmath172 , since the other arcs in @xmath172 lie on the other side of @xmath209 from @xmath307 .
this isotopy requires at most @xmath309 type 2 elementary moves , as in the previous case .
we then repeat for an outermost arc @xmath313 among the collection of arcs of @xmath314
. this isotopy takes place between @xmath313 and @xmath315 , and in particular never meets @xmath307 .
repeating for all arcs in @xmath172 , we can isotope the entire collection @xmath172 to @xmath302 using at most @xmath310 type 2 elementary moves . similarly , we can isotop @xmath284 to @xmath302 using at most @xmath316 type 2 elementary moves .
it follows that we can isotop @xmath172 to @xmath284 using at most @xmath317 type 2 elementary moves .
the total number of elementary moves required in these isotopies is at most twice the number of type 2 elementary moves .
@xmath76 define the _ length _ of a @xmath5 curve transverse to the 1-skeleton of a triangulated surface to be the number of edges of the 1-skeleton that it crosses , counted with multiplicity .
[ basic.delta ] a basic move in a surface with a triangulation of valence at most @xmath318 can be realized by at most 28 elementary moves for a type 1 basic move , @xmath319 elementary moves for a type 2 basic move , and 22 elementary moves for a type 3 basic move . in all cases @xmath320 elementary moves
suffice .
[ [ proof.-9 ] ] * proof . * + + + + + + + + + we count the number of elementary moves required for each type of basic move .
lemma [ polygonal.delta ] implies that a type 1 basic move requires at most 14 elementary moves in each of the two triangles meeting the edge along which the curve is isotoped .
the type 2 basic move can be done in the following sequence of steps .
two type 2 elementary moves slide the arc in a neighborhood of an edge to pass through the adjacent vertex .
two type 2 elementary moves slide the curve to run along edges in a neighborhood of the adjacent vertex .
@xmath321 type 2 elementary moves slide the arc over a neighborhood of the vertex in each adjacent triangle .
4 . at most 7 type 2 elementary moves
slide the arc to basic curves in each of the original two triangles , by lemma [ polygonal.delta ] .
5 . two type 2 elementary moves slide the arc to basic curves in each of the @xmath321 triangles where new intersections were introduced .
summing , this basic move requires less than @xmath322 type 2 elementary moves , and @xmath323 in total .
the type three basic move requires 1 .
one type 2 elementary move to slide the arc to coincide with a subarc of its adjacent edge .
2 . at most 10 type 2 elementary moves
slide the resulting arc to a basic curve with the same endpoints in the adjacent triangle , by an application of lemma [ polygonal.delta ] .
this sums to at most 11 elementary moves of type 2 and 22 elementary moves of all types . since @xmath324 , in all cases @xmath320
elementary moves suffice .
if two basic curves are isotopic using only basic moves of type 1 , we say that they are _ parallel _ basic curves .
[ parallel ] let @xmath172 and @xmath284 be two parallel basic curves with total length @xmath267 .
then @xmath172 can be isotoped to coincide with @xmath284 by @xmath325 basic moves of type 1 .
similarly , @xmath172 can be isotoped across @xmath284 by @xmath325 basic moves of type 1 .
[ [ proof.-10 ] ] * proof . *
+ + + + + + + + + each of @xmath172 and @xmath284 has length @xmath325 and they intersect each triangle of the surface in pairs of parallel basic arcs . using @xmath325 basic moves of type 1 , we can either move the intersections of @xmath172 along each edge to coincide with the intersection of @xmath284 with that edge , or move the intersections past to the other side of @xmath284 .
@xmath76 let @xmath172 and @xmath284 be intersecting simple closed curves in general position on a surface @xmath285 .
2-gon _ is a subdisk of @xmath285 whose boundary consists of one subarc of each of @xmath172 and @xmath284 .
a 2-gon is _ innermost _ if no arc of @xmath172 or @xmath284 meets its interior . in some cases
, it can be shown that curves with excess intersection create innermost 2-gons .
this is analyzed in some generality in @xcite .
the following is lemma 3.1 of @xcite .
[ innermost ] let @xmath285 be a compact orientable surface and let @xmath172 and @xmath284 be isotopic simple closed curves on @xmath285 , intersecting transversely . then either @xmath172 and @xmath284 are disjoint or there is an innermost 2-gon on @xmath285 bounded by one arc from each of @xmath172 and @xmath284 . [ [ proof.-11 ] ] * proof .
+ + + + + + + + the lift of @xmath172 to the universal cover of @xmath285 is a family of disjoint lines @xmath326 , and the lift of @xmath284 is a similar family @xmath327 .
assume @xmath172 and @xmath284 intersect .
then a lift @xmath268 of @xmath172 and a lift @xmath328 of @xmath284 also intersect .
the curves @xmath268 and @xmath328 are isotopic to disjoint lines by a lift of an isotopy making @xmath172 and @xmath284 disjoint .
the fundamental group of @xmath285 acts on the universal cover , and @xmath268 and @xmath328 have a common cyclic stabilizer .
it follows that they intersect infinitely often .
an arc of each between adjacent crossing points cuts off a 2-gon in the universal cover .
a finite number of arcs of @xmath329 meet this 2-gon .
each such arc is disjoint from either @xmath326 or @xmath327 , so that it cuts off a smaller 2-gon inside the previous one .
we can pass to an innermost 2-gon whose interior meets no arc of @xmath330 .
this 2-gon is disjoint from any of its translates , and must project 1 - 1 down to @xmath285 .
@xmath76 [ isotopy.length ] let @xmath285 be an orientable triangulated surface with @xmath286 triangles , and let @xmath172 and @xmath284 be isotopic simple basic curves on @xmath285 intersecting transversely .
let @xmath331 be the length of @xmath305 , and let @xmath318 be the maximal valence of a vertex in @xmath285 .
there is an isotopy carrying @xmath284 to @xmath172 via no more than @xmath332 basic moves .
[ [ proof.-12 ] ] * proof . * + + + + + + + + + we treat cases . note that @xmath333 and @xmath334 . _ case 1 : @xmath172 and @xmath284 are disjoint . _
then @xmath172 and @xmath284 bound a subsurface of @xmath285 homeomorphic to an annulus .
we will perform a series of basic moves which isotop the curves to coincide .
the edges of @xmath285 induces a cellulation of the annulus .
the cells are subdisks of triangles , bounded by arcs in @xmath305 .
if there is a vertex in this cellulation , we can slide @xmath172 or @xmath284 over a vertex by a basic move of type 2 .
repeating at most @xmath286 times , we eliminate all vertices in the annulus .
the length of @xmath305 is increased by at most @xmath321 each time we do a basic move , so the final curves have total length at most @xmath335 .
the annulus between the curves is cut by the edges of @xmath285 into a collection of cells .
since @xmath172 and @xmath284 are disjoint , and each cell boundary contains an edge lying on the 1-skeleton between each edge lying on @xmath305 , all cells are even sided .
an euler characteristic calculation shows that either all cells are 4-gons , or there is a 2-gon cell in the annulus , bounded by a basic arc of type 2 and an arc on an edge of the triangulation .
if there is a cell which is a 2-gon , we can perform a basic move of type 3 which slides an arc of @xmath172 or @xmath284 across that 2-gon .
this decreases the length of @xmath305 by two . after repeating at most @xmath336 times ,
we eliminate all 2-gon cells from the annulus , and arrive at a cellulation in which all cells are 4-gons .
when all cells are 4-gons , the two boundary components of the annulus are parallel basic curves , each of length at most @xmath337 . using at most @xmath337 basic moves of type 1 , we can make the two curves coincide .
the total number of basic moves required in the three steps above is bounded by @xmath338 _ case 2 : @xmath172 and @xmath284 intersect .
_ two basic arcs in a triangle with distinct endpoints intersect at most once , so the total number of intersections of @xmath172 and @xmath284 is at most @xmath339 .
lemma [ innermost ] gives the existence of an innermost disk @xmath340 on the surface , bounded by subarcs @xmath268 of @xmath172 and @xmath328 of @xmath284 .
the 1-skeleton of @xmath285 induces a cellulation of @xmath340 .
the cells are bounded by arcs in @xmath341 and arcs in the 1-skeleton of @xmath285 . if @xmath340 contains a vertex of @xmath285 , we can isotop @xmath268 or @xmath328 over a vertex ( possibly a different one ) by a basic move of type 2 , keeping @xmath305 embedded .
the number of vertices is bounded above by @xmath286 , so repeating at most @xmath286 times , we eliminate all vertices in @xmath340 , using at most @xmath286 basic moves of type 2 . the length of @xmath341 is increased by at most @xmath321 by each basic move , so @xmath172 and @xmath342 are isotoped to curves with total length less than @xmath343 .
since @xmath268 and @xmath328 meet at exactly two points in the boundary of @xmath340 , there are at most two cells in @xmath340 with an odd number of sides .
an euler characteristic calculation shows that one of three cases occurs : _ case 1 .
_ all of @xmath340 is contained in a single triangle , or _ case 2 .
_ there is a 2-gon cell in @xmath340 , or _ case 3 . _
all cells of @xmath340 are 4-gons , other then two 3-gons containing points of @xmath344 .
the first case is impossible because basic arcs do nt intersect in more than one point in a single triangle .
a 2-gon cell in @xmath340 gives a basic arc of type 2 .
we can use this to perform a basic move of type 3 , which slides an arc of @xmath268 or @xmath328 across a 2-gon .
this decreases the length of @xmath341 by two , and thus can be repeated at most @xmath345 times .
this leaves us in case 3 .
we can then isotop @xmath328 across @xmath268 , using at most @xmath337 basic moves of type 1 .
we have then eliminated the innermost 2-gon @xmath340 , reducing the number of intersection points of @xmath172 and @xmath284 by two , while using at most @xmath346 basic moves .
the length of @xmath305 is now bounded by the larger value @xmath347 .
if @xmath172 and @xmath284 still intersect , we repeat with another innermost disk .
eliminating this disk requires at most @xmath348 basic moves .
repeating at most @xmath339 times we eliminate all intersections and arrive at disjoint curves . summing the resulting series , which has at most @xmath339 terms ,
gives a bound of @xmath349 basic moves . with the curves now disjoint , and having length less than @xmath350 , case 1 isotops them to coincide using at most @xmath351 basic moves .
the total number of basic moves used is less than @xmath352 since @xmath353 and @xmath334 , @xmath354 so the sum is less than @xmath332 as claimed .
@xmath76 [ surface.bound ]
let @xmath285 be an orientable triangulated surface with @xmath286 triangles , and let @xmath172 and @xmath284 be isotopic basic curves on @xmath285 in general position .
let @xmath331 be the total length of @xmath305 .
then there is an isotopy carrying @xmath284 to @xmath172 via no more than @xmath355 elementary moves .
[ [ proof.-13 ] ] * proof . *
+ + + + + + + + + the required number of basic moves is at most @xmath332 , and each basic move requires at most @xmath320 elementary moves , so the total number of elementary moves required is less than @xmath356 .
we know @xmath357 so that @xmath358 gives a bound on the number of elementary moves .
we now give the proof of the main result of this section .
[ [ proof-of-theorem-surface.bound.thm ] ] proof of theorem [ surface.bound.thm ] : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we begin by isotoping @xmath172 to a basic curve .
first we perturb @xmath172 to miss the zero - skeleton of the triangulation .
there are at most @xmath286 vertices , and each has at most @xmath318 edges coming into it , where @xmath318 is the maximal valence of the triangulation of the surface .
we can perturb @xmath172 near any vertex which it passes through so that it crosses at most @xmath359 edges near that vertex .
this requires a total of at most @xmath360 type 2 elementary moves , and @xmath361 elementary moves of both types .
the number of segments on the curve has now been increased to at most @xmath362 .
next we isotop the intersection of @xmath172 with each triangle to a union of basic arcs .
this can increase the number of segments by at most a factor of three , to @xmath363 and requires less than @xmath364 elementary moves by lemma [ polygonal.delta ] .
we can do similar moves on @xmath284 , resulting in new curves @xmath172 and @xmath284 which are transversely intersecting basic curves with a total of at most @xmath365 segments .
note that the length of a basic curve is between one third and one half of the number of segments it contains , therefore @xmath305 has length at most @xmath363 .
we now apply lemma [ surface.bound ] , isotoping the curves together using at most @xmath366 ^ 4 u^3 $ ] elementary moves .
we can bound @xmath318 by @xmath286 , obtaining a bound of at most @xmath367 ^ 4 u^3 < 2^{11 } ( s + u^2)^4 u^3\ ] ] elementary moves .
we combine the results of 25 to prove theorem [ thm102 ] . [ [ proof - of - theoremthm102 - 1 ] ] proof of theorem [ thm102 ] : + + + + + + + + + + + + + + + + + + + + + + + + + + to prove theorem [ thm102 ] , we first construct a compact triangulated @xmath5 submanifold @xmath35 of @xmath6 obtained by removing a solid torus neighborhood of @xmath4 .
we barycentrically subdivide @xmath6 twice and construct a new triangulated manifold @xmath369 by removing the interior of a regular neighborhood @xmath36 of @xmath4 .
the boundary of @xmath35 consists of a subdivision of the original boundary @xmath34 plus a 2-torus @xmath38 , the peripheral torus .
the manifolds @xmath35 and @xmath52 contain between them @xmath234 tetrahedra .
as outlined in 3 , we deform the original knot @xmath4 across an annulus by elementary moves to a longitude @xmath32 on the peripheral 2-torus @xmath38 .
then we deform @xmath32 on the peripheral torus by elementary moves to the boundary @xmath33 of the embedded normal disk given by haken s theorem 3.1 .
finally we deform the curve @xmath33 across this disk using elementary moves to a triangle .
we count elementary moves . by lemma [ disk.count ] the normal disk with @xmath33
as boundary contains at most @xmath370 triangles , and the number of edges @xmath371 of @xmath33 is at most @xmath372 also by lemma [ disk.count ] it takes at most @xmath373 elementary moves to transform @xmath33 to a triangle . by theorem [ longitude.count ]
the longitude @xmath32 consists of at most @xmath374 edges , and the isotopy from @xmath4 to @xmath32 requires at most the same number of elementary moves .
it remains to bound the number of elementary moves needed to carry @xmath32 to @xmath33 , using theorem [ surface.bound.thm ] .
we note that @xmath32 is isotopic to @xmath33 if properly oriented .
theorem [ surface.bound.thm ] applies with @xmath375 to yield a bound of at most @xmath288 elementary moves taking @xmath32 to @xmath33 , where @xmath286 is the number of triangles on the peripheral torus . since @xmath376 , corollary [ surface.bound.cor ] yields a bound of @xmath377 for the number of elementary moves taking @xmath32 to @xmath33 .
for the three steps together we obtain a bound of @xmath378 .
@xmath148 [ [ remark.-3 ] ] remark .
+ + + + + + + the assumption @xmath379 was used to allow the peripheral torus @xmath52 to be constructible by two barycentric subdivisions and to have @xmath38 disjoint from @xmath34 . for the general case that @xmath380
, we could proceed by barycentrically subdividing @xmath6 twice , then isotoping @xmath4 off the boundary @xmath34 to a `` parallel '' knot @xmath381 contained strictly in the interior of @xmath53 , using @xmath382 elementary moves .
now theorem 1.2 applies to @xmath381 in @xmath383 .
the remainder of the paper is concerned with 3-manifolds embedded in @xmath12 , and is devoted to proving the reidemeister move bound of theorem [ thm101 ] .
we outline the proof of theorem [ thm101 ] .
we have given a knot diagram @xmath1 with @xmath3 vertices .
we use it to construct a triangulated @xmath5 manifold @xmath6 embedded in @xmath12 which contains a knot @xmath4 in its 1-skeleton whose projection on the @xmath384-plane in @xmath12 is the knot diagram @xmath1 . in section 8
we show that such a manifold @xmath6 can be constructed using @xmath385 tetrahedra .
now theorem [ thm102 ] shows that we can isotop @xmath4 to a trivial knot using @xmath386 elementary moves .
we project these elementary moves down onto the @xmath384-plane and obtain a sequence of reidemeister moves . the proof is completed by bounding the number of reidemeister moves in terms of elementary moves , which we do below .
in the rest of this section we describe reidemeister moves and relate them to elementary moves .
we recall basic facts about knot and link diagrams .
a projection of a link in @xmath12 into a plane is _ regular _ if all crossings are in general position , and at most 2 edges intersect the pre - image of any point .
corresponding to a regular projection of a link is a _
link diagram_. this is a labeled planar graph having the following properties : 1 . any connected component of @xmath1 with no vertices is an ( isolated ) loop .
each vertex has exactly four adjacent edges , with two edges labeled `` undercrossing '' and two labeled `` overcrossing '' at the vertex .
the labeling prescribes a cyclic ordering of edges at each vertex , and `` overcrossing edges '' are adjacent to `` undercrossing edges '' in this ordering .
conversely , for any labeled planar graph @xmath1 satisfying ( i)(iii ) there exists an unoriented link @xmath21 having projection @xmath1 .
we note that the cyclic ordering of vertices in ( iii ) actually specifies a planar embedding of the graph @xmath1 , since it determines the faces of the planar embedding .
a _ link component _ in a link diagram @xmath1 is obtained by tracing a closed path of edges such that each pair of consecutive edges match at their common vertex , either both overcrossing or both undercrossing .
the _ order _ of @xmath1 is the number of components it contains .
a _ link diagram component _ of @xmath1 is a connected component of the underlying graph ; it can be the union of several link components .
the _ crossing measure _
@xmath387 of a link diagram is @xmath388 a _ knot diagram _ is a link diagram that consists of one link component .
a _ trivial knot diagram _ is a knot diagram that has no vertices .
if @xmath1 is a knot diagram , then @xmath387 is the _ crossing number _ of the knot projection .
[ regular.proj ] if a polygonal link @xmath21 has regular projection @xmath1 , then @xmath389 * proof . * + + + + + + + + + the number of vertices @xmath22 is at least as large as the number of distinct line segments it contains .
since line segments project to line segments , each pair of line segments produces at most one crossing .
there are less than @xmath390 pairs .
furthermore line segments that project into different connected components of @xmath391 can not produce any crossings .
thus the number of nonintersecting pairs of line segments is at least @xmath392 , so [ eq202 ] follows .
@xmath76 the bound [ eq202 ] is the correct order of magnitude , because one can easily construct a sequence of knots @xmath4 with @xmath393 , which have the property that @xmath394 .
there is no upper bound for @xmath22 in terms of @xmath391 , but there does exist a link @xmath21 with a projection having diagram @xmath395 , with @xmath396 , see ( * ? ? ?
* lemma 5.1 ) and 3 .
reidemeister moves are local moves on link diagrams that are analogous to elementary moves .
they convert a link diagram @xmath395 to another link diagram @xmath397 by a local change in its structure . for unoriented links
there are four kinds of reidemeister moves , examples of which are pictured in figure [ reid.moves ] , see one of @xcite , @xcite,@xcite,*???*,@xcite .
( for oriented links there are more reidemeister moves , see wu @xcite . )
we call two knot diagrams @xmath398 and @xmath399 _ equivalent _ if and only if one can be converted into the other by a finite sequence of reidemeister moves
. this relation corresponds to knot equivalence : knots @xmath32 and @xmath33 have knot diagrams @xmath398 and @xmath399 that are equivalent if and only if @xmath398 and @xmath399 are equivalent .
reidemeister moves on a link diagram @xmath395 are related to elementary moves on a link @xmath21 whose projection is @xmath395 , but the correspondence of moves is not one - to - one .
let @xmath21 be a polygonal link and suppose that it has diagram @xmath395 as regular projection on a plane .
let @xmath400 be a link derived from l by an elementary move of type ( 2 ) or @xmath401 , and suppose that it has regular projection @xmath397 on the same plane .
then there is a sequence of at most @xmath402 reidemeister moves that converts @xmath395 to @xmath397 .
[ [ proof.-15 ] ] * proof . * + + + + + + + + + as a line segment is moved affinely in @xmath12 with one end fixed , its projection intersects the projection of any vertex of the polygonal link @xmath21 in at most one point
. therefore it can generate at most one type i or type ii reidemeister move involving the projection of that vertex .
the number of vertices is bounded by @xmath403 .
similarly , each pair of crossing segments in @xmath395 can produce at most one type iii reidemeister move . since two line segments are moved in a type ( 2 ) or @xmath401 elementary move , this yields at most @xmath404 reidemeister moves .
@xmath76 the bound of lemma [ reid.2 ] is nearly best possible .
we use these lemmas to prove a bound for reidemeister moves in terms of elementary moves .
[ elementary.to.reidemeister ] let @xmath21 and @xmath400 be polygonal links each consisting of at most @xmath3 line segments and suppose that @xmath395 and @xmath397 are link diagrams obtained by regular projection of @xmath21 and @xmath400 on a fixed plane . if @xmath21 can be transformed to @xmath400 using @xmath177 elementary moves , then @xmath395 can be transformed to @xmath397 using at most @xmath405 reidemeister moves . *
* let @xmath406 be the set of intermediate links occurring in the sequence of elementary moves , and let @xmath407 denote the projection of @xmath408 on the fixed plane .
we may assume that all projections are regular , by applying a a transversality argument and infinitesimally moving the plane of projection ( * ? ? ?
since an elementary move changes the number of vertices by at most one , and @xmath409 , we have , for @xmath410 , that @xmath411 lemma [ regular.proj ] gives @xmath412 thus we obtain @xmath413 and applying lemma [ reid.2 ] to each @xmath408 separately yields the upper bound @xmath414 . @xmath76
in this section we construct a triangulated 3-manifold @xmath35 which is a knot complement in a ball embedded in @xmath12 .
that is , @xmath415 , in which @xmath52 is an open regular neighborhood of the knot @xmath4 , @xmath416 is an open ball containing the point at infinity in its interior , and @xmath417 .
the manifold @xmath35 has boundary @xmath418 which consists of two connected components , which are topologically a 2-sphere and a torus . as in 2 , we call such a manifold @xmath35 a _ truncated knot complement_. we carry out this construction more generally for link diagrams , as it requires no extra work . given a link diagram @xmath419 with crossing measure @xmath3
, we construct a triangulated manifold @xmath6 in @xmath12 which contains @xmath385 tetrahedra and has a link @xmath21 in its 1-skeleton which has a regular projection that gives the link diagram @xmath419 .
[ triangulate ] given a link diagram @xmath419 of crossing measure @xmath3 , one can construct a triangulated convex polyhedron @xmath420 in @xmath12 such that : 1 .
the triangulation contains at most @xmath421 tetrahedra .
every vertex in the triangulation is a lattice point @xmath422 , with @xmath423 , @xmath424 and @xmath425 .
3 . there is a link @xmath419 embedded in the 1-skeleton of the triangulation which lies entirely in the interior of @xmath420 , and whose orthogonal projection on the plane @xmath426 is regular and is a link diagram isomorphic to @xmath419 . [
[ proof.-16 ] ] * proof . *
+ + + + + + + + + the link diagram @xmath419 comes with a topological planar embedding which determines the faces .
we begin by adding extra vertices to @xmath419 . to each edge or non - isolated loop
we add two new vertices of degree 2 , which splits it into three edges . to each isolated loop
we add three new vertices of degree 2 , making it an isolated triangle .
the resulting labeled graph @xmath427 is still planar , has no loops or multiple edges , and has at most @xmath428 vertices .
( the worst case consists of several disjoint single crossing projections . )
let @xmath429 denote the number of vertices of @xmath427 , and call the @xmath430 vertices added _
special vertices_. note that each vertex of the original graph has associated to it four distinct special vertices . using the topological planar embedding of @xmath427
, we next add extra edges to triangulate each face of @xmath427 , obtaining a topologically embedded triangulated planar graph @xmath431 .
the graph @xmath431 has @xmath429 vertices and @xmath432 bounded triangular faces , and the unbounded face is also a triangle .
we next encase the graph @xmath431 in a triangulated planar graph @xmath433 with @xmath434 vertices and @xmath435 bounded triangular faces , by framing the outside triangular face with a larger triangle and then adding 6 edges to subdivide the resulting concentric region into 6 triangles , obtaining @xmath436 .
see figure [ triangulation ] .
our next goal is to replace the topological embedding of @xmath436 in the plane with a straight - line embedding . in 1952
fary showed that straight - line embeddings always exist , and a recent result of de frajsseix , pach and pollack @xcite states that the vertices can be chosen to lie in a small rectangular grid of integer lattice points .
their results imply that there exists a planar embedding of @xmath433 whose vertices @xmath437 lie in the plane @xmath438 , and are contained in the grid @xmath439 and all edges of the graph @xmath433 are straight line segments .
next we make an identical copy @xmath440 of @xmath433 on the plane @xmath441 with @xmath442 , and vertex set @xmath443 .
we now consider the polyhedron @xmath444 which is the convex hull of @xmath445 and @xmath446 .
it is a triangular prism , because the outside face of @xmath433 is a triangle .
we add @xmath447 vertical edges connecting each vertex @xmath448 to its copy @xmath449 .
let @xmath450 denote these edges , together with all the edges in @xmath433 and @xmath440 . using these edges
, the polyhedron @xmath444 decomposes into @xmath435 triangular prisms @xmath451 , with the top and bottom faces of each being congruent triangular faces of @xmath433 and @xmath440 .
we next triangulate @xmath444 by dissecting each of the @xmath435 triangular prisms into 14 tetrahedra , as follows .
we subdivide each vertical rectangular face into four triangles using its diagonals .
then we cone each rectangular face to the centroid of @xmath444 , and note that the centroid lies in the plane @xmath426 .
we add 4 new vertices to each prism , one on each rectangular face and one in the center .
the point of this subdivision is that the triangulations of adjacent prisms are compatible .
note also that all new vertices added lie in the lattice @xmath452 .
let @xmath453 denote all the edges in the union of these triangulated prisms .
we identify the link diagram @xmath427 with the graph @xmath427 embedded in @xmath433 .
we next observe that there is a polygonal link @xmath21 embedded in the edge set @xmath453 whose projection is the link diagram @xmath419 .
we insist that any edge in @xmath395 that runs to an undercrossing have its vertex which is an undercrossing be in the plane @xmath438 , while any edge that runs to an overcrossing have the overcrossing vertex in the plane @xmath454 .
each such edge goes from one of the @xmath3 original vertices of @xmath419 to a special vertex adjacent to that vertex .
edges that do not meet vertices labeled overcrossing or undercrossing are assigned to the @xmath438 plane .
the edge corresponding to an edge running from the @xmath438 to the @xmath454 plane is contained in one of the diagonals added to a prism .
the resulting embedding @xmath21 in @xmath444 has a regular orthogonal projection onto the @xmath455-plane , since no edges are vertical and only transverse double points occur as singularities in the projection .
note that no edge of @xmath21 touches the outer triangle of @xmath433 because all edges project to the graph @xmath431 contained strictly inside @xmath433 .
this will be important for property ( iii ) .
the knot edges lie in the upper and lower boundary of @xmath444 , but do not touch its sides . to have them
not touch the boundary we take two additional copies of @xmath444 and glue one to its top , along the plane @xmath456 , and one to its bottom , along the plane @xmath457 .
the knot now lies entirely in the interior of the resulting polyhedron @xmath458 .
all the vertices added in this construction lie in @xmath452 .
the total number of tetrahedra used in triangulating @xmath458 is @xmath459 , which is at most @xmath460 , or at most @xmath421 for @xmath15 .
now triple all coordinates @xmath461 to obtain a triangulated polytope @xmath462 with integer vertices , with a triangulation that satisfies ( i)(iii ) .
@xmath76 + [ [ remark.-4 ] ] remark
. + + + + + + + ( 1 ) .
the construction in theorem [ triangulate ] can be effectively computed in time @xmath463 using the hopcroft - tarjan planarity algorithms @xcite and the embedding algorithm of de frajsseix , pach and pollack ( * ? ? ?
* section 4 ) .
one can ask whether a polygonal knot @xmath4 in @xmath12 that consists of @xmath3 line segments can be embedded in the 1-skeleton of some triangulated polyhedron that uses @xmath385 tetrahedra .
we do not know if this can always be done .
there exist polygonal knots whose planar projections in any direction have at least @xmath464 crossings .
avis and el gindy @xcite showed that a set of @xmath3 points in general position in @xmath12 can be embedded as vertices of a triangulation of their convex hull using @xmath385 tetrahedra , but that in the general case @xmath465 tetrahedra are sometimes needed .
we combine the earlier results to prove theorem [ thm101 ] with an explicit bound on the number of reidemeister moves required to isotop an unknot to the trivial projection . applying theorem [ triangulate ] to the knot diagram @xmath395 yields a triangulated convex polyhedron @xmath6 embedded in @xmath12 with @xmath9 tetrahedra , where , @xmath466 and with a knot @xmath4 embedded in the 1-skeleton of the interior of @xmath6 that projects to @xmath395 .
the hypotheses of theorem [ thm102 ] are satisfied for @xmath6 , hence there is a sequence of at most @xmath10 elementary moves transforming @xmath4 to the trivial knot in @xmath6 , with @xmath20 .
now theorem [ elementary.to.reidemeister ] gives a bound of at most @xmath467 reidemeister moves , since @xmath4 contains at most @xmath468 line segments .
the bound ( [ eqn92 ] ) yields the desired bound of the form @xmath2 for an explicit constant @xmath0 . using the value @xmath20 from the proof of theorem [ thm102 ]
, one finds that @xmath469 suffices .
@xmath148 99 c. adams , _ the knot book . an elementary introduction to the mathematical theory of knots _ , w. h. freeman , new york , 1994 . j. w. alexander and g. b. briggs , _ on types of knotted curves _ , ann . math .
, * 28 * ( 1926/27 ) , 562586 .
d. avis and h. el gindy , _ triangulating point sets in space _
, disc . & comp .
geom . , * 2 * ( 1987 ) ,
99111 . j. birman and m.d .
hirsch , _ recognizing the unknot , _ preprint , 1997 . j. birman and w. menasco , _ studying links via closed braids v : the unlink , _ trans .
ams , * 329 * , 1992 , 585606 .
g. burde and h. zieschang , _ knots _ , de gruyter 1985 . h. de frajsseix , j. pach and r. pollack , _ how to draw a planar graph on a grid , _ combinatorica , * 10 * ( 1990 ) , 4151 .
s. galatolo , _ on a problem in effective knot theory , _ rendiconti dell accademia dei lincei ( to appear )
. l. goeritz , _ bemerkungen zur knotentheorie , _ abh .
hamburg , * 10 * ( 1934 ) , 201210 .
w. haken , _ theorie der normalflachen , ein isotopie kriterium fr ein kreis , _ acta math . , * 105 * ( 1961 ) , 245375 .
j. hass , _ algorithms for recognizing knots and 3-manifolds , _ to appear in chaos , fractals and solitons .
j. hass , j. c. lagarias and n. pippenger , _ the computational complexity of knot and link problems , _ preprint , 1997 .
j. hass , j. c. lagarias and n. pippenger , _ the computational complexity of knot and link problems , preliminary report , _ proc .
38th annual symposium on foundations of computer science , ( 1997 ) , 172181 .
j. hass and g. p. scott , _ intersections of curves on surfaces , _ israel math .
j. , * 51 * ( 1985 ) , 90120 . g. hempel , _
3-manifolds , _ princeton university press , princeton nj , 1976 . j. e. hopcroft and r. e. tarjan , _ efficient planarity testing , _ j. assoc . comp .
mach . , * 21 * ( 1974 ) , 549568 .
hudson , _ piecewise - linear topology _ , w.a .
benjamin inc . 1969 .
w. jaco and h. rubinstein , _
@xmath5 equivariant surgery and invariant decompositions of 3-manifolds , _ advances in math . ,
* 73 * ( 1989 ) , 149191 .
w. jaco and j. l. tollefson , _ algorithms for the complete decomposition of a closed 3-manifold , _ ill .
, * 39 * ( 1995 ) , 358406 . h. kneser , _ geschlossene flachen in dreidimensionalen mannigfaltigkeiten , _ jahresbericht math . verein . ,
* 28 * ( 1929 ) , 248260 .
k. murasugi , _ knot theory and its applications _ , birkhauser , boston ma , 1996 .
h. reidemeister , _ knoten und gruppen _ ,
math . sem .
hamburg , * 5 * ( 1926 ) , 723 .
h. reidemeister , _ elementare begrndang der knotentheorie , _ abh .
hamburg , * 5 * ( 1926 ) , 2432 .
h. reidemeister , _ knotentheorie _ , springer : berlin 1932 ; chelsea ; new york , 1948 .
( english translation : _ knot theory _
, bcs associates , moscow , i d 1984 . ) d. rolfsen , _ knot and links _ , publish or perish inc .
, berkeley ca , 1976 .
rourke and b.j .
sanderson , _ introduction to piecewise - linear topology _ , springer - verlag , berlin , heidelberg , new york , 1982 .
a. schrijver , _ theory of linear and integer programming _ , john wiley and sons , new york , 1986 .
h. schubert , _ bestimmung der primfactor zerlegung von verkettungen , _ math . z. , * 76 * ( 1961 ) , 116148 .
a. seb , hilbert bases , _
caraetheodory s theorem and combinatorial optimization , _ in integer programming and combinatorial optimization , r. kannan and w. r. pulleybank , eds .
, u. waterloo press , 1990 , 431455 .
d. j. a. welsh , _ complexity : knots , colourings and counting _ , cambridge university press , cambridge 1993 .
d. j. a. welsh , _ the complexity of knots _ , in quo vadis , graph theory ? , j. gimbel , j. kennedy and l. v. quintoo , eds . , north - holland , amsterdam , 1993 , 159173 .
( also : annals disc . math .
, * 55 * ( 1993 ) , 159173 . ) d. j. a. welsh , _ knots and braids : some algorithmic questions , _ in graph structure theory , seattle , wa 1991 , contemporary math . vol . 147 , ams , providence ri , 1993 , 109123 .
f. y. wu , _ knot theory and statistical mechanics , _ rev .
, * 64 * ( 1992 ) , 10991131 . | there is a positive constant @xmath0 such that for any diagram @xmath1 representing the unknot , there is a sequence of at most @xmath2 reidemeister moves that will convert it to a trivial knot diagram , where @xmath3 is the number of crossings in @xmath1 .
a similar result holds for elementary moves on a polygonal knot @xmath4 embedded in the 1-skeleton of the interior of a compact , orientable , triangulated @xmath5 3-manifold @xmath6 .
there is a positive constant @xmath7 such that for each @xmath8 , if @xmath6 consists of @xmath9 tetrahedra , and @xmath4 is unknotted , then there is a sequence of at most @xmath10 elementary moves in @xmath6 which transforms @xmath4 to a triangle contained inside one tetrahedron of @xmath6 .
we obtain explicit values for @xmath0 and @xmath7 .
_ keywords _ : knot theory , knot diagram , reidemeister move , normal surfaces , computational complexity * the number of reidemeister moves + needed for unknotting * + + university of california , davis + davis , ca 95616 + + at&t labs research + florham park , nj 07932 + ( ) + |
evidence for absorption features in the x - ray spectra of bl lac objects at @xmath3 kev has existed for many years .
this is a tantalizing situation : bl lac objects were originally defined by their lack of the sharp spectroscopic features that provide the normal tools for understanding astrophyisical objects . without this tool , only large multisatellite ,
multitelescope campaigns spanning 20 decades of frequency have been able to constrain the bl lac emission mechanism .
absorption features can provide a unique tool to study the presence and the dynamical properties of any gas surrounding the nucleus of the otherwise ` featureless ' bl lac objects .
the extremely large extreme ultraviolet explorer ( _ euve _ ) satellite database on offers a so far a unique opportunity to search for and characterize any sharp spectroscopic feature .
is one of the brightest , best known and closest bl lac objects ( @xmath4=0.0308 , ulrich et al .
was in fact the first bl lac object found to have x - ray emission ( richetts et al .
1976 ) and its low redshift makes this object one of the few extragalactic sources detected at @xmath5-rays above 300 gev up to tev energies ( punch et al .
1992 ) . has been extensively observed at radio ( e.g. zhang & baath 1990 ) , uv / optical ( maza , martin & angel 1978 ; mufson et al . , 1990 ) and x - ray frequencies ( e.g. mushotzky et al .
, 1979 ; george , warwick & bromage 1988 , takahashi 1996 , guainazzi 1999 ) . shows optical polarization , a flat radio spectrum and significant time variability , characteristics of the blazar class ( makino et al . , 1987 ) and
its parent galaxy has been identified as a giant elliptical ( ulrich et al .
1975 ; mufson , hutter & kondo 1989 ) . in the euv band ( @xmath6 ) was detected during the _ rosat _ wfc survey ( pounds et al .
1993 , pye et al . 1995 ) and the _ euve _ all - sky survey ( marshall , fruscione & carone 1995 , fruscione 1996 ) .
it is one of the strongest euv extragalactic sources and was immediately recognized as a prime target for subsequent euv spectroscopic studies . for 4 almost consecutive years ( 1994 , 1995 , 1996 and 1998 ) was the target of multiwavelength campaigns covering radio to tev energies ( macomb et al . , 1995 ,
kerrick et al .
1995 , buckley et al .
1996 , takahashi et al . 1999 ) and the _ euve _ satellite ( which gives simultaneous photometric and spectroscopic data ) was always part of the campaigns .
in addition , in 1995 and in 1997 was observed serendipitously with the _ euve _ photometers ( the `` scanners '' ) .
+ the large amount of data accumulated for by the _ euve _ satellite to date ( @xmath7 ks of public imaging data and @xmath0 ks of public spectroscopic data ) represents the best coverage for any bl lac object at high energies .
less than half of this data has been analysed to date ( fruscione et al . 1996 , kartje et al .
1997 , hereafter k97 ) revealing that ( i ) the source exhibits a flare - like behavior ( ii ) the strong euv variability is correlated with soft x - ray and tev - energy emission , and ( iii ) the spectrum may have strong absorption features between 65 and 75 ( 0.16 - 0.19 kev ) ( k97 ) . however in the euv , all the previous analysis of bl lacs can only infer the presence of absorption from the extrapolation of the x - ray spectrum ( e.g. konigl et al 1995 for pks2155 - 304 and k97 for mrk 421 ) .
the purpose of this paper is to analyze , in an homogeneous way , all of the 950 ks of _ euve _ public spectroscopic data for .
the results of this paper will also provide a useful reference for the calibration of other satellites observing in this energy range .
in particular is a calibration target for both the chandra and xmm satellites , whose low energy transmission grating ( letg ) and reflection grating spectrometer ( rgs ) partially overlap with the _ euve _ energy band .
+ in section 2 we present all the _ euve _ spectra of taken from 1994 to 1997 ; in section 3 we discuss the results and in section 4 we give a brief summary and our conclusions .
all the photometric data , the corresponding lightcurves and a detailed variability analysis are presented in a companion paper ( cagnoni , papadakis & fruscione , 2000 , hereafter paper ii ) .
the _ euve _ satellite was launched in june 1992 and is still acquiring astronomical data in the wavelength range from @xmath8 to 750 . onboard _ euve _
there are 4 telescopes : three of them are co - aligned photometers observing in 4 different euv bandpasses , while the fourth is the deep survey / spectrometer ( ds / s , see e.g. welsh et al .
1990 ) mounted orthogonally to the scanners .
( figure 1 of paper ii shows the ds effective area compared to the scanner effective area ) .
in addition to a photometer ( 10 % bandpass 68 - 178 ) the ds / s telescope is also equipped with three spectrometers ( hettrick & bowyer , 1983 ; abbott et al .
1996 ) covering the `` short '' ( sw : @xmath9 ) , `` medium '' ( mw : @xmath10 ) and `` long '' ( lw : @xmath11 ) euv wavelengths .
this configuration allows simultaneous imaging and spectroscopy with a spatial resolution of @xmath12 and a spectral resolution of @xmath13 at the short wavelengths , i.e. 0.35 at 70 .
+ further details on the science instrumentation and performances can be found in malina ( 1994 ) , sirk et al .
( 1997 ) and abbott et al .
( 1996 ) .
the scanners were used to carry out the _ euve _ all - sky survey ( july 21 1992 - january 23 1993 ) and are currently utilized in the right angle program ( rap ) , that is the observation of targets of scientific interest with the scanner telescopes while the ds / s telescope is conducting the primary science observation .
the ds / s has been used for a deep euv survey along the ecliptic during the all - sky survey and for pointed spectroscopic and imaging observations since then .
we use throughout this paper all of the publicly available ds / s data on mrk 421 : from 1994 to 1997 , a total of 950 ks .
we are not considering the most recent observation ( april 1998 ) since the data is presently being analysed by the original proposers ( marshall in preparation ) .
was observed by _ euve _ several times from 1994 to 1997 , four times with the ds / s and twice with the scanners ; table 1 summarizes the observations .
+ because of strong absorption by the interstellar medium ( ism ) along the line of sight ( even at the low @xmath14 @xmath15 elvis , wilkes & lockman 1989 ) , was detected only at the shortest wavelengths : in the lex / b filter of the scanners ( 10% bandpass 58 - 174 ) , in the lex / b filter of the ds instrument ( 10% bandpass 67 - 178 ) and , relevant for this paper , in the sw spectrometer ( 70 - 190 ) .
] after rejecting the time intervals with high particle background , such as the satellite passages over the south atlantic anomaly , and correcting the data for instrumental deadtime and telemetry saturation , we measured a total effective exposure of 315858 s , 506780 s , and 354806 s for the 1994 , 1995 and 1996 observations respectively ( note that the first 1995 and the 1997 observations were performed with the scanners , instrument with imaging capabilities only ) . + we extracted all the spectra in a homogeneous way from the two - dimensional sw detector image , using the _ euve _ guest observer center software ( iraf / euv package ) and other standard spectroscopic iraf tasks .
the spectrum was extracted in a 14 pixel wide aperture and the background obtained by linearly fitting an average background in two regions ( @xmath16 pixel wide ) of the detector , one on each side of the spectrum .
the resulting wavelength - calibrated spectra ( figures 1 , 4 and 8) were convolved with the proper effective area and binned over 1 . for the 1994 on - axis observation ( figure 1 ) we used the on - axis effective area derived from in - orbit calibrations for the 75 - 110 range .
the 1995 and 1996 observations were performed off - axis ( a technique that slightly increases the short - wavelength coverage , down to @xmath17 ) .
this results in a different effective area ( marshall et al .
1999 ) that we applied here for the first time .
the 1@xmath18 error bars on the flux measurements shown in the figures of all the spectra were computed using the formula @xmath19}}{a_{eff}(\lambda)}\ ] ] where s is the signal from the target , b the averaged background , both in counts / s , @xmath20 the sw effective area in @xmath21 and @xmath22 and @xmath23 are respectively the width in pixels of the extraction region for spectrum ( 14 pixels ) and background ( 160 pixels ) in the direction perpendicular to the dispersion .
we modeled the euv spectrum of using version 2.0 of the sherpa modelling and fitting software ( recently developed at the chandra x - ray center ) .
we initially used an absorbed power law of the form @xmath24 where f(@xmath25 ) is a normalization factor .
we used @xmath26 ( e@xmath27=0.155 kev ) .
@xmath28 is the energy index and the absorption is characterized by a column density n(x ) and an absorption cross section @xmath29 for each element .
we included in the model h , he i and he ii ( rumph , bowyer and vennes , 1994 ) and heavier elements ( morrison and mccammon , 1983 ) .
we fixed the galactic hydrogen column density at @xmath30 @xmath15 ( elvis , wilkes & lockman 1989 ) and the ratios @xmath31/@xmath32=0.1 and @xmath33/@xmath32=0.01 .
the best fit values for the simple absorbed power law model are summarized in table 2 together with their @xmath34 uncertainties for 1 interesting parameter .
two - parameter @xmath35 contour confidence levels at @xmath36 + 1 , + 4.61 and + 5.99 corresponding respectively to the 1@xmath18 uncertainty for each single parameter and to the 90% and 95% joint confidence levels are shown in figure 2 , 5 and 9 for the 1994 , 1995 and 1996 data sets respectively .
our best fit flux in 1994 is slightly higher than the previous analysis of fruscione et al .
, 1996 ( f@xmath37 photons @xmath15
s@xmath38 @xmath38 compared to their f@xmath39 @xmath40jy@xmath41 photons @xmath15 s@xmath38 @xmath38 ) but our slope is within the errors ( @xmath42 compared to their @xmath43 ) .
the 1@xmath18 error bars reported in fruscione et al .
, 1996 were not corrected for the effective area ; the small difference in the value of the normalization can be possibly explained by the difference in the errors and a possible difference in the wavelength range used to model the spectrum .
this is the only spectrum for which the addition of an absorbed interved gaussian line does not improve the fit ( tab.3 ) .
the 1995 ds / s observation was part of a large multiwavelength campaign during which a powerful tev flare occured ( buckley 1996 ) . simultaneously an x - ray flare developed ( takahashi 1996 ) , and the euv followed with a short delay ( buckley 1996 ) . as illustrated in figure 3 ,
the observation spanned the flare ( @xmath44 ks ) , the decay ( @xmath45 ks ) and the following quiescent state ( @xmath46 ks ) the observation was performed @xmath47 off - axis and as expected the signal reaches down @xmath48 .
this spectral region is crucial since it covers the location of the wide euv absorption feature proposed by k97 using less than half ( the flare decay ) of the 1995 data . in our analysis
we corrected the data with the @xmath47 off - axis effective area ( marshall 1999 ) and the resulting spectrum , binned over 1 , is presented in figure 4 . in order to compare our data analysis to that of k97 , and in an attempt to look for changes during the flare , we divided the 1995 observation into three parts ( figure 3 ) : + flare : april 2528 ( @xmath44ks ) , the time interval of the flare ; + decay : april 29 may 6 ( @xmath45 ks ) , the time interval analysed by k97 ; + quiescence : may 713 ( @xmath46 ks ) , the remaining data .
the _ euve _
spectra for the 3 parts are shown in figiure 6 a , b and c. table 1 lists the exposures times and table 2 summarizes the best - fit parameters for each one of the three parts . in figure 7
we plot the two - parameter @xmath35 contour confidence levels .
the ` decay ' section of the 1995 spectrum deserves comparison with the analysis of k97 , given the importance of the absorption line evident in their figure 2 .
we show the effect of using the improper @xmath47 off - axis effective area in figure 6b ; we compared the spectrum obtained using the on - axis effective area ( open cirles ) as in k97 , and the @xmath49 off - axis effective area ( filled squares ) .
the effect of using the appropriate calibration is to increase the flux below @xmath16 and so weaken the absorption feature ( @xmath50% ) .
another effect to consider is that k97 evaluated the level of the continuum in the euv range by adopting an @xmath51 ( @xmath52 ) slope derived from the 0.15 kev euv flux and the 1.5 kev x - ray ( ) flux , which resulted in a strong absorption feature ( see also 3 for further discussion on this point ) . our best fit slope @xmath53 ( table 2 ) is flatter than the @xmath54 in k97 .
if we consider the same wavelength range ( 70 - 95 instead of 70 - 110 ) we find @xmath55 , consistent with k97 result . a simultaneous fit of an absorbed power law plus an absorbed ( inverted ) gaussian should provide a more reliable result .
if the position of the gaussian is a free parameter then the reduced @xmath56 goes to 1.17 and a narrow ( fwhm=@xmath57 ) absorption feature is found at @xmath58 while the slope becomes significanlty flatter ( @xmath59 ) ( see tab.3 ) .
this is the first time that an absorption feature in the euv range is derived from euv data alone . both for pks 2155 - 304 ( konigl 1995 ) and for ( k97 ) the presence of absorption features
was previously derived on the basis of an extrapolation of the x - ray power law and not directly from _ euve _ data .
however there is strong evidence that the x - ray spectrum of the high - energy peaked bl lac objects , including , is concave ( sambruna et al .
1994 , 1997 ; tashiro et al . 1995 ; takahashi et al .
1996a ; giommi et al .
1998 ; guainazzi et al .
1999 ) and a gradual steepening with energy agrees with the `` synchrotron self compton '' scenario , which has often been used to explain the spectral energy distribution of bl lac objects ( ghisellini , maraschi & treves 1985 , ghisellini 1989 ) .
therefore , a simple power law model is an unreliable extrapolation from x - rays to the euv . in order to verify this assumption for
, we reanalyze , as an example , one of the many one - hour - long 1995 _ asca _ observations ( the one on 25 - 26 april ) undertaken simultaneously with the _ euve _ observation ( 1995 ` flare ' ) .
this in order to investigate the spectral slope below 1.5 kev ( takahashi 1996b and 1999 only published data for 1.5 kev and above ) .
first we fitted the spectra with an absorbed broken power law and the best fit requires a break position at @xmath60 kev and the energy slopes varies from @xmath61 above the break to @xmath62 below the break ( consistent with what found by takahashi et al . ) .
the broken power law model however does not give a good fit of the spectrum below 1.5 kev ; a better result is found if we assume a `` curved spectrum '' parametrized as @xmath63 where @xmath64 , @xmath65 and @xmath66 are the low and high energy asymptotic slopes and @xmath67 is the curvature radius in the energy space .
the best fit values for the curved model with fixed @xmath68 ( e.g. guainazzi 1999 ) are @xmath69 , @xmath70 and @xmath71 ( see fig .
10 ) ; these values show that the slope flattens toward lower energies and that an extrapolation with a simple power law is not an accurate way to predict the euv continuum level ; the low energy asymptotic slope ( @xmath72 ) is generally in agreement with the flat slope derived from the euv spectrum ( @xmath73 from tab.3 ) .
+ note that in the case of pks 2155 - 304 there is an additional problem : the _ rosat _ spectral slope used for the extrapolation was not measured simulteneously to the euv spectrum and the spectral slope has been observed to harden when the source intensity increases ( giommi et al .
1998 ) . in 1996
was again observed @xmath47 off - axis with a total effective exposure of 354806 s. the 1 binned spectrum and the best fit parameters model are plotted in figure 8 ; the corresponding contour confidence levels are shown in figure 9 .
we performed several fits using an absorbed power law model plus gaussian for all the observations .
the results are listed in table 3 . in every case , except for the 1994 observation , the addition of a gaussian absorption line improves the fit .
the power law slope is in the range @xmath74 $ ] which is not too distant from the slope measured from x - ray data , ( see e.g. guainazzi with sax data which gives @xmath75 depending on the spectral model ) .
the best - fit center of the gaussian absorption lines is in the 71 - 75 range with full width half maximum varying from 0.2 to 6 .
we note however that the 6 fwhm is relative to the global 1995 observation and it is probably due to the superposition of narrower slightly variable lines .
ever since _ einstein
_ pioneering observations suggested the existence of an ubiquitous absorption feature around @xmath76 kev in the x - ray spectra of bl lacertae objects ( canizares & kruper 1984 , madejski et al .
1991 ) , much x - ray and euv data has been collected with the goal of confirming the existence of such absorption features . the first object to show evidence of an x - ray absorption feature was pks 2155 - 304 ( @xmath77 ) in a 1980 _ einstein _ grating spectrum ( canizares & kruper 1984 ) .
the feature was detected at @xmath3 kev ( @xmath78 ) and extended for 50 - 100 ev ( @xmath79 ) ; it was interpreted by the authors as o viii ly@xmath28 emitted by high velocity material .
looking at a 1979 _ einstein _ solid state spectrometer observation of the same object madejski et al .
1991 found evidence for absorption at @xmath76 kev , consistent with the grating result .
+ krolik et al .
1985 modelled the absorption as due to an outflowing subrelativistic wind beamed in our direction .
+ pks 2155 - 304 was subsequently observed by _ bbxrt _ in 1990 and the absorption feature at @xmath76 kev was confirmed in those data ( madejski et al .
1994 ) , although instrumental effects cast some doubt on this result ( sambruna 1997 , weaver 1995 ) .
+ in a long exposure ( @xmath80 ks ) _ beppo - sax _ spectrum of pks 2155 - 304 in the soft x - rays ( giommi et al .
1998 ) the best fit model is improved ( at 99% confidence ) when a notch at @xmath81 kev is added to the model , although the magnitude of the feature is close to the instrument calibration limits .
+ pks 2155 - 304 was also observed by _
euve_. a 1992 30 ks observation gives hints of an absorption feature at @xmath16 ( fruscione et al .
two longer observations in june ( 130 ks ) and in july ( 150 ks ) 1993 suggested the existence of 2 to 5 absorption features at a @xmath82 level in the range 75 to 110 ( konigl et al .
1995 ) . extrapolating the _ rosat _ x - ray spectral slope into the euv the authors found evidence of a wide absorption feature from @xmath83 to @xmath84 and interpreted it as the superposition of doppler - smeared absorption lines ( l - shell transitions of mg and ne as well as m - shell transitions of fe ) that originate in a high - velocity clumped outflow from the nucleus . according to this model the x - ray line spectrum should be dominated by an o vii k@xmath28 broadened and blueshifted feature .
this corresponds to the feature observed in by _
einstein _ and in 1990 by _ bbxrt _ , but the predictions are an order of magnitude greater than the observations ; to match _ bbxrt _ measurement the oxygen must be present in subcosmic abundance ( konigl 1995 ) . another bl lac object to show an x - ray absorption feature is h1426@xmath85428 ( @xmath86 ) .
sambruna et al .
( 1997 ) reanalyzed a _
bbxrt _ 1990 observation and found a broad feature around @xmath87 kev confirmed by an _
asca _ 1994 spectrum , but not by a _ rosat
_ 1993 spectrum .
a third bl lac , pks 0548 - 322 ( @xmath88 ) ( sambruna & mushotzky , 1998 ) confirmed the early discovery of an absorption feature from the _ einstein _ spectrum ( madejski et al .
1991 ) ; this feature is present in _ asca _
data and can be modeled with either an edge at @xmath89 kev and optical depth @xmath90 or with a notch at @xmath91 kev and fixed width of @xmath92 kev and covering fraction @xmath93 .
the _ asca _
absorpton feature is at energies significantly higher than in previous x - ray observations , when the continuum was in a slightly higher state ( by a factor @xmath94 ) ; this provides tentative evidence for variability in the absorption feature .
k97 interpret the absorption feature in the context of the konigl ( 1995 ) model for pks 2155 - 304 as a superposition of mg viii and mg ix absorption lines .
the implied ionization state of the gas would require stronger o vii absorption in the x - ray range than is allowed by observations and the authors invoke subcosmic oxygen abundance to explain its absence .
our analysis of the 1994 , 1995 and 1996 _ euve _ spectra of shows an absorption feature inferred directly from the data , ( i.e. by modelling the euv continuum and the absorption line simultaneously ) . the simplest model for this feature
is a reverted gaussian at about 71 - 75 a maximum fwhm of the order of few . in order to investigate likely absorption features in a bl lac environment , we ran photoionization equilibrium models ( nicastro 1999 , nicastro 2000 ) which include the strongest 300 emission and resonant absorption lines down to 50 ev ( oscillator strength @xmath95 ) .
we find that the predicted wavelengths and strengths of the absorption lines , both absolute and relative , are highly sensitive to the properties of the photoionizing continuum and the absorbing gas ( e.g. the ionization state and the assumed outflowing velocity ) .
the number of transitions in the model also has a strong effect .
however assuming the observed euv - to - x - ray continuum and adjusting the gas paramters to maximize the relative abundance of ovii ( @xmath96 ) both an ovii k@xmath28 absorption line ( ew=0.25 ) and a feix l ( @xmath97 , ew=0.2 ) absorption line are expected respectively around 0.6 kev and 73 ( assuming an outflow velocity of 0.15c ) .
the equivalent width of the lines is too small to give significant detections , however , especially in the euv regime , the presence of many fainter satellite lines ( not included in the model ) could explain the observed feature for example between the si vi line ( @xmath98 ) and the mg vi complex ( @xmath99 ) , both included in the model , at least 15 additional lines from ionized fe , al , be and ne are not included ( oscillator strength 0.01 - 0.1 ) and could be important especially if the elements are abundant . even though no strong absorption line seem to be visible in the x - ray spectrum , we note that both in the 1979 _ einstein _ data ( madejski et al .
1991 ) and in the 1997 and 1998 _ beppo - sax _ spectrum ( guainazzi et al . 1999 , fossati et al .
2000 ) the fit improves when adding an absorption feature centered at @xmath76 kev . in order to investigate any possible correlation between spectral changes and source flux , we plotted in fig.11 the energy index as a function of total euv flux for the the power law plus gaussian models .
however the large error bars prevent any statististically significant conclusion .
from _ euve _ data , we have found a clear absorption feature at @xmath100 in the spectrum of mrk 421 .
the absorption features detected in the euv / x - ray spectra of bl lac objects are often poorly determined statistically , as in the _ beppo - sax _ observations of pks 2155 - 304 and mrk 421 ( giommi et al . 1998 and guainazzi et al .
1999 ) , but it is interesting to notice that different instruments on different satellites point to the same possible @xmath101 kev feature and to a ( probably linked ) absorption feature in the euv spectra of the same objects .
we do not know if x - rays and euv absorption features are present in all or many bl lac objects , as initially suggested by madejski et al .
1991 , but to - date absorption features at @xmath102 kev have been found or at least statistically improves the spectral fitting in four of the brightest bl lacs objects ( pks 2155 - 304 , h1426@xmath85428 , pks 0548 - 322 and mrk 421 ) and euv absorption features has been found in two of them ( pks 2155 - 304 and mrk 421 ) .
future observations covering the euv to x - ray energy bands with much higher resolution and collecting area , as with _
chandra _ and _ xmm _ , will soon be able to answer this question and give more detailed observational data on which to develop new models .
we would like to thank dr .
herman marshall for providing the off - axis effective areas for the ds instrument , and drs .
i. papadakis , a. celotti , r. sambruna , m. elvis for the useful discussions .
we are particularly grateful to f. nicastro and g. matt for making their models available to us .
we thank the science team and in particular dr .
roger malina for their logistical support for our continuing analysis effort on data .
this research has made extensive use of the high energy astrophysics science archive research center online service ( heasarc ) , provided by the nasa - goddard space flight center , of the nasa / ipac extragalactic database ( ned ) operated by the jet propulsion laboratory , caltech , under contract with nasa and of nasa s astrophysics data system abstract service .
a.f . and i.c
are grateful to the brera observatory for the ospitality .
this work was supported by axaf science center nasa contract nas 8 - 39073 and by nasa grants nag 5 - 3174 and nag 5 - 3191 .
abbott , m. et al .
1996 , apjs , 107 , 451 bowyer , s. & malina , r. f. , 1991 , in extreme ultraviolet astronomy , ed .
malina & s. bowyer ( new york : pergamon press ) , 397 buckley , j. h. et al .
1996 , apj , 472 , 9 cagnoni , i. , fruscione , a. & papadakis , i. e. , ( paper ii ) 1999 , this volume canizares , c. r. & kruper , j. , 1984 , apj , 278 , 99 elvis , m. , wilkes , b. j. & lockman , f. j. , 1989 , aj , 97 , 777 fossati , g. et al .
, 2000 , apj in press fruscione , a. 1996 , apj , 459 , 509 fruscione , a. , bowyer , s. , konigl , a. & kahn , s. m. , 1994 , apj , 422 , 55 fruscione , a. , bruhweiler , f. , cheng , k.p . ,
hall , c.r .
, kafatos , m. , ramos , e. & kondo , y. , 1996 , in uv and x - ray spectroscopy of astrophysical and laboratory plasmas , ed . k. yamashita & t. watanabe ( tokyo : universal academic press ) , 351 george , i.m . ,
warwick , r. s. & bromage , g. e. , 1988 mnras 232 , 793 george , i. m. , warwick , r. s. & mchardy , i. m. , 1988 , mnras , 235 , 787 ghisellini , g. , 1989 , mnras , 236 , 341 ghisellini , g. maraschi , l. & treves , a. , 1985 , a&a 146 , 204 giommi , p. , barr , p. , garilli , b. , maccagni , d. & pollock , t. , 1990 , apj , 356 , 432 giommi , p. et al .
1998 , a&a , 333 , l5 guainazzi , m. , vacanti , g. , malizia , a. , oflaherty , k. s. , palazzi , e. & parmar , a. n. , 1999 , a&a , 342 , 124 hettrick , m.c . & bowyer , s. , 1983 , appl .
, 3921 kartje , j.f . ,
konigl , a. , hwang , c.y . & bowyer , s. , 1997 , apj , 474 , 630 ( k97 ) kerrick , a. d. et al .
1995 , apj , 438 , 59 konigl , a. , kartje , j. f. , bowyer , s. , kahn , s. m. & hwang , c .- y . , 1995 , apj , 446 , 598 krolik , j. h. , kallman , t. r. , fabian , a. c. & rees , m. j. , 1985 , apj , 295 , 104 macomb , d. j. et al .
1995 , apj , 449 , 99 madejski , g. m. , mushotzky , r. f. , weaver , k. a. , arnaud , k. a. & urry , c. m. , 1991 , apj , 370 , 198 madejski , g. m. et al .
1994 , in `` bbxrt , a preview to astronomical x - ray spectroscopy in the 90 s '' , ed .
p. serlemitsos & s. j. shrader , greenbelt : nasa , 63 marshall , h. l. , fruscione , a. & carone , t. e. , 1995 , apj , 439 , 90 marshall , h. l. et al .
1999 in preparation makino , f. et al .
1987 , apj , 313 , 662 malina , r. et al .
1994 , aj , 107 , 751 maza , j. , martin , p. g. & angel , j.r.p .
, 1978 , apj , 224 , 368 morrison , r. & mccammon , d. , 1983 , apj , 270 , 119 - 122 mufson , s.l . , hutter , d.j . &
kondo , y. , 1989 , in bl lac objects : proceedings of a workshop held in como , italy , september 20 - 23 , 1988 .
editors , l. maraschi , t. maccacaro , m .- h .
ulrich ; publisher , springer - verlag , berlin , new york , ny , 1989 , p.341 mufson , s.l . ,
hutter , d.j . , kondo , y. , urry , c. m. & wisniewski , w. z. , 1990 apj , 354 , 116 mushotzky , r. f. , boldt , e. a. , holt , s. s. & serlemitsos , p. j. , 1979 , apj , 232 , 17 nicastro f. , fiore f. and matt g. , 1999 , apj , 517 , 108 nicastro f. , elvis m. , fiore f. , matt g. and savaglio s. , 2000 , proceedings of `` x - ray astronomy 99 '' , 1999 , september 6 - 10 , bologna , italy , astro - ph/0007050 pounds , k. a. et al .
1993 , mnras , 260 , 77 punch , m. et al .
1992 , nature , 358 , 477 pye , j. p. et al .
1995 , mnras , 274 , 1165 richetts et al . 1976 ,
nature , 259,546 rumph , t. , bowyer , s. & vennes , s. , 1994 , aj , 107 , no .
6 , 2108 - 2114 sambruna , r. m. , barr , p. , giommi , p. , maraschi , l. , tagliaferri , g. & treves , a. , 1994 , apjs , 95 , 371 sambruna , r. m. et al .
1997 , apj , 483 , 774 sambruna , r. m. & mushotzky , r. f. , 1998 , apj , 502 , 630 sirk , m.m .
, internal euve memo mms / euve/0084/94 sirk , m.m . ,
vallerga , j.v . ,
finley , d.s . , jelinsky , p. & malina , r.f .
, 1997 , apjs , 110 , 347 takahashi , t et al .
1996a , 1996 , apj , 470 , 89 takahashi , t. et al .
1996b , mem .
67 , 533 t takahashi , t. , madejski , g. & kubo , h. , 1999 , apj , 11 , 177 tashiro , m. , makishima , k. , ohashi , t. , inda - koide , m. , yamashita , a. , mihara , t. & kohmura , y. , 1995 , pasj , 47 , 131 ulrich , m .- h .
, kinman , t. d. , lynds , c. r. , rieke , g. h. , & ekers , r. d 1975 , apj , 198 , 261 welsh , b.y . ,
vallerga , j.v . , jelinsky , p. , vedder , p.w . ,
bowyer , s. & malina , r.f .
, 1990 , opt .
eng . , 29 , 752 zhang , f. j. & baath , l. b. , 1990 , a&a , 236 , 47 1994 & ds / s & april 2 & april 12 & 280 & 0.18 & yes + 1995 & scanner & feb .
4 & feb . 7 & 68 & 0.45 & no + 1995 & ds / s & april 25 & may 13 & 355 & 0.29 & yes + 1995 ( flare ) & ds / s & april 25 & april 28 & 86 & 0.37 & yes + 1995 ( decay ) & ds / s & april 29 & may 6 & 154 & 0.29&yes + 1995 ( quiescence ) & ds / s & may 7 & may 13 & 115 & 0.22 & yes + 1996 & ds / s & april 17 & april 30 & 299 & 0.30 & yes + 1996 & ds / s & may 10 & may 11 & 3.6 & 0.30 & no + 1997 & scanner & feb . 7 & feb . 11 & 108 & 0.27 & no + 1998 & ds / s & april 19 & may 1 & & & + l l c c c c 1994 & & 75 - 110 & @xmath103 & @xmath104 & 1.06 + 1995 & & 70 - 110 & @xmath105 & @xmath106 & 2.45 + _ flare _ & apr . 25 - 28 & 70 - 110 & @xmath107 & @xmath108 & 1.57 + _ decay _ & apr .
29-may 6 & 70 - 110 & @xmath109 & @xmath110 & 1.24 + _ quiescence _ & may 7 - 13 & 70 - 110 & @xmath111 & @xmath112 & 1.47 + 1996 & & 70 - 110 & @xmath113 & @xmath114 & 1.61 + 1994 & & @xmath115 & @xmath116 & @xmath117 & @xmath118 & @xmath119 & @xmath120 & 5.06 + 1995 & & @xmath121 & @xmath122 & @xmath123 & @xmath124 & @xmath125 & 1.17 & 8.86 + flare&apr .
25 - 28 & @xmath126 & @xmath127 & @xmath128 & @xmath129 & @xmath130 & 0.57 & 14.11 + decay&apr .
29-may 6 & @xmath131 & @xmath132 & @xmath133 & @xmath134 & @xmath135&1.17 & 9.33 + quiescence&may 7 - 13 & @xmath136 & @xmath137 & @xmath138 & @xmath139 & @xmath140 & 1.29 & 4.83 + 1996 & & @xmath141 & @xmath142&@xmath143 & @xmath144 & @xmath145 & 1.10 & 12.02 + | we analyzed the @xmath0 ks of spectroscopic data accumulated by the _ extreme ultraviolet explorer _
( _ euve _ ) satellite between 1994 and 1997 for the bl lacertae object markarian 421 .
the euv spectrum is well detected in the 70110 ( 112177 ev ) range and can be fitted by a power law model plus an absorption feature in the @xmath1 range .
previous studies of euv absorption features in and in the other euv bright bl lac object , pks2155@xmath2304 , explain this absorption feature as a superposition of doppler - smeared absorption lines ( mainly l- and m - shell transitions of mg and ne ) originating in high - velocity gas clouds ionized by the beamed continuum of the associated relativistic jet .
we show that , for example , fe ix l could also be a possibility consistent with the marginal detection of oxygen absorption lines in the x - ray range .
however physical models are highly sensitive to the assumptions on the photoionizing continuum and the surrounding gas . |
observationally , quasars show very similar uv - optical spectra @xcite , which can be characterized by a featureless continuum and a series of broad emission lines @xcite . at wavelengths
longer than the ly@xmath8 line , from @xmath9 to @xmath10 , the continuum of quasars can be well described by a power - law @xmath11 , where @xmath12 characterizes the continuum slope with a typical value @xmath13 @xcite . in terms of @xmath14 ,
more negative @xmath14 means a steeper ( or redder , softer ) continuum .
there is a general agreement that the uv continuum of higher luminosity quasars is harder ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* however , see also @xcite ) . on the other hand ,
there has been a long running debate on the existence of the redshift dependence ( evolution ) of the quasar continuum .
some studies suggest the slight hardening of the continuum toward higher redshift @xcite , while others find hardly any evolution @xcite or even reverse @xcite .
several mechanisms could cause the redshift dependence of the quasar continuum , e.g. , the evolution of the accretion process @xcite and/or the evolution of the dust component . among them , one of the interesting mechanism is the intergalactic ( cosmic ) dust extinction , which makes the quasar spectra statistically redder at higher redshifts .
indeed , @xcite attributed the redder spectra toward higher redshift they detected to the effect of the cosmic dust , and concluded that the average line of sight has cosmic extinction of @xmath15 out to z=3 ( however , see also @xcite ) . compared to other attempts on probing the cosmic dust
extinction ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , quasars take advantage of their high luminosity and large sample size ; therefore , they can be explored to higher redshifts . in this letter , we take a large quasar sample from sloan digital sky survey ( sdss , @xcite ) and use the continuum slope of the quasar composite spectrum to probe its redshift dependence , then further explore its implications for the cosmic dust . this letter is organized as follows . in section 2
, we describe the data and method of building the composite spectrum . in section 3 ,
we report the finding of a significant dependence of @xmath12 on the redshift : redder spectra at higher redshifts . in section 4
, we build a simple cosmic dust extinction model to quantify the reddening effects . in section 5
, we discuss other possibilities . finally , we present our conclusions in section 6 . throughout this letter , we adopt the cosmological parameters @xmath16 = 0.7 , @xmath17= 0.3 , and @xmath18 .
the sdss legacy survey provides a database of quasar samples that includes photometries in _ u , g , r , i _ , and _ z _ bands and spectra from 3800 to 9200 @xmath19 with resolutions of about 2000 @xcite . in the sdss dr7 quasar catalog , @xcite compiled 105,783 quasars that are more luminous than @xmath20 and have at least one emission line with fwhm larger than 1000 km s@xmath21 or have interesting / complex absorption features .
this catalog provides the basic database for our study .
accompanying the dr7 quasar catalog , there are other value - added catalogs .
@xcite recalculated the redshift of quasars and reduced the systematic effects of the redshift to the level of 30 km s@xmath21 per unit redshift .
@xcite improved the sky subtraction in the red wavelength band .
after measuring the spectroscopic features , @xcite calculated the implied physical parameters ( e.g. , the bolometric luminosity , black hole mass , and eddington ratio ) for each quasar . in this study , we take the spectra from @xcite and use the improved redshifts from @xcite .
we exclude the quasars with broad absorption lines from our analysis according to the flags in @xcite .
the radio - loud quasars show significant redder continua than radio - quiet ones @xcite and the fraction of them changes with both redshift and luminosity @xcite . for these reasons
, we also remove them .
therefore , our final sample with calculated bolometric luminosity includes 91,131 objects .
we first correct the galactic reddening for each quasar using the sfd map @xcite and the reddening curve of @xcite .
then , we mask bad pixels and shift the spectra to their rest frame according to the redshifts of @xcite .
next , we follow @xcite to make the quasar composite and calculate the statistical error for the composite spectrum by dividing 68@xmath22 semi - quantile to the square of the contributing number at each wavelength . the composite spectrum of all 91,131 quasars is shown in fig . [ fig1 ] as the blue line . for comparison ,
the composite spectrum of @xcite is shown as the red line .
as can be seen , these two composites show good consistency .
because of a much larger sample size , our composite spectrum has a much higher signal - to - noise ratio ( shown in the sub - panel of fig . [ fig1 ] ) . in the uv - opt band
, the quasar continuum can be very nicely characterized by a single power law ( @xmath11 ) and is usually fitted from continuum windows . in @xcite ,
two continuum windows , 1350 - 1365 @xmath19 and 4200 - 4230 @xmath19 , have been adopted , which avoid the @xmath23 line at shorter wavelengths and contamination from the host galaxy at longer wavelengths . however , this range is too long to be covered by any single sdss spectrum at any given redshift .
similar to @xcite , we add a new continuum window 2210 - 2230 @xmath19 . with this configuration of continuum windows and the 3800 - 9200 @xmath19 wavelength range of sdss quasars , the far - uv slope @xmath0 and the near - uv slope @xmath2 can be fitted for quasars in the redshift intervals @xmath24 ( @xmath25 ) and @xmath26 ( @xmath27 ) , respectively .
we bin the quasars into four bolometric luminosity ( @xmath28 ) bins with a bin size of 1 dex from @xmath29 to @xmath30 and then make composites respectively . to show the possible variance of the composites more clearly , each composite spectrum
is further divided by the global composite spectrum ( blue line in fig .
[ fig1 ] ) and normalized at 3000 @xmath19 .
the results are shown in fig .
[ fig2 ] . the quasar composite spectra change with @xmath28 significantly , where the higher @xmath28 quasars have systematically harder continua .
we calculate the continuum slopes ( @xmath0 and @xmath2 ) for each composite spectrum and plot them as a function of @xmath28 in the sub - panel of the figure .
the @xmath0 and @xmath2 values are close at high @xmath28 and show similar trends with luminosity . at the low luminosity side ( @xmath31 @xmath32 )
, the change of the continuum slope with luminosity is very significant , while at the high luminosity side ( @xmath31 @xmath33 ) , quasar continua roughly keep a constant slope ( @xmath34 ) .
the redder spectra of lower luminosity quasars can hardly be explained by the accretion disk model , where the peak of the big blue bump moves toward lower frequency ( longer wavelength ) with increasing luminosity @xcite .
as discussed in @xcite , this discrepancy could be alleviated by assuming a higher intrinsic dust extinction in low luminosity quasars .
indeed , there is observational evidence that the low luminosity agns have higher intrinsic dust extinction ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
such a luminosity dependent intrinsic dust extinction scenario is also consistent with a receding dust torus model @xcite .
however , even if the dust extinction is independent of the quasar luminosity , quasars with higher extinction will also show lower luminosity and redder uv spectra .
such a selection effect may also partly explain the @xmath35 relation we observed . in this section
, we test whether the quasar continuum slope shows systematical variance with redshift . as a first step , we divide the sample into five redshift bins with bin sizes of 1 from @xmath36 to 5 and make composites . similar to fig .
[ fig2 ] , we calculate the ratios of the composites to the global one for different redshift bins and show the results in fig .
consistent with the finding of @xcite , we find little dependence of quasar spectra on redshift .
however , the sdss quasar sample is roughly a magnitude limit sample and thus high redshift quasars are biased to high luminosity ones .
according to the results in section 3.1 , if there are no evolutionary effects of quasar spectra , high redshift quasars should also have harder uv spectra .
therefore , the apparent non - evolutionary quasar spectra implies another redshift dependent mechanism that could compensate for the luminosity bias . to recover this hidden effect , we make composite quasar spectra at different redshift bins after @xmath28 controlled .
the quasars are grouped into two - dimensional bins of @xmath28 and @xmath37 .
the bin width of @xmath28 is set to 0.5 dex in logarithm while the width of @xmath37 is set to either 0.1 or 0.2 according to the number of the quasars available for composite .
the minimum number of quasars to make a composite spectrum is set to 20 .
the results are shown in fig .
[ fig4 ] , where the spectral slopes are plotted against redshift with diamonds ( @xmath0 ) and triangles ( @xmath2 ) .
the continuum slopes within the same @xmath28 bin are connected with lines , while different @xmath28 bins are represented by different colors .
we find that in any given @xmath28 bin , quasars at higher redshifts have systematically redder uv continuum slopes ( for both @xmath0 and @xmath2 ) .
this result clearly identifies the existence of the redshift dependence of quasar spectral slopes and implies that the previous debates are probably driven by redshift selection bias .
as we have mentioned in the introduction , the cosmic dust extinction is a very interesting mechanism that can make the quasar continua systematically redder at high redshift .
therefore , it is quite natural to introduce a cosmic dust extinction model to explain the results we have obtained in fig .
other possibilities will be discussed in section 5 .
we start from a toy model , which assumes that dust is uniformly distributed along the line of sight .
considering the intervening dust has a comoving number density of @xmath38 , the optical depth of the cosmic dust to the photons with wavelength @xmath39 emitted from a quasar at redshift @xmath40 is @xmath41 where @xmath42 , @xmath43 , and @xmath44 is the cross section of dust absorbers at wavelength @xmath45 @xcite .
we assume that the properties of the absorbers do not evolve with redshift and @xmath44 is scaled by the cross section of the absorber at 5500@xmath19 , @xmath46 : @xmath47 where el(@xmath45 ) stands for the extinction law .
we also assume that the intrinsic continuum of quasars in the uv - opt band can be characterized by a pure power - law form @xmath48 .
quasars are put to their redshifts @xmath40 , and then , their observational spectral continuum can be rebuilt and slopes @xmath49 can be measured after considering the cosmic dust extinction from @xmath50 to @xmath36 . to mimic the observation , we also fit the continuum slopes in different wavelength ranges for quasars at different redshifts ( @xmath0 for @xmath51 and @xmath2 for @xmath52 ) . in this toy model , besides the extinction law , we have two free parameters , one is the combined parameter @xmath53 , which represents the effective density of the cosmic dust particles , and the other is @xmath54 , the intrinsic quasar continuum slope .
we have little prior knowledge of the template extinction law of the cosmic dust . as the 2175 @xmath19 extinction bump
is rarely seen in other galaxies , an smc - type extinction curve is preferred . in this study
, we test the featureless smc extinction law first .
we choose sets of representative values of @xmath55 and show the predicted @xmath49 as a function of @xmath40 , which are represented by the dashed parallel lines in fig .
[ fig4 ] . here , @xmath56 is tuned to match the observed @xmath57 relation . as can be seen
, our simple cosmic dust extinction model can reproduce the observed @xmath57 relation at the low redshift range @xmath58 very well . at high redshift ,
our model ( smc extinction curve and @xmath56 ) predicts a much steeper @xmath59 relation than that is observed .
this discrepancy can be easily solved by adopting either a lower @xmath60 at @xmath61 or a shallower extinction curve at a far - uv wavelength . to keep our model simple , we test the later possibility . indeed , there are studies suggesting that the extinction curves in quasars are flatter in the uv @xcite .
we take the flat reddening curve from ( * ? ? ?
* hereafter , gb07 ) , which is quite similar to the milky way extinction law , except for the lack of a 2175 @xmath19 bump .
we show the gb07 dust extinction models with the same @xmath56 as the dotted lines in fig .
as expected , the gb07 extinction curve matches the observation much better than the smc extinction curve at @xmath61 . in order to have a more intuitive impression of the cosmic dust extinction , we take the gb07 dust extinction model and calculate the observer - frame extinction @xmath62 and reddening @xmath63 as a function of source redshift .
the @xmath62 and @xmath63 are plotted in the upper and lower panels of fig .
[ fig5 ] respectively . in this plot
, we also show the cosmic extinction constrained from a few other measurements for comparison .
the dashed line shows the constant comoving dust extinction model of @xcite , which is constrained from the statistical detection of dust reddening around galaxies up to large scales .
the other observational constraints are also extracted from fig .
9 of @xcite , which includes the results from measuring the excess scatter seen in higher redshift quasar colors @xcite , using the tolman test @xcite , combining constraints from luminosity distances and @xmath64 @xcite , and using mgii absorbers ( * ? ? ?
* see appendix c of @xcite for details ) .
further study about the impact of cosmic dust on supernova cosmology can be found in @xcite and @xcite .
in this section , we discuss whether there are any other explanations to the redshift dependence of the quasar continuum slopes other than the cosmic dust . in section 3.1
, we have argued that the internal dust extinction might be the cause of the @xmath12-luminosity trend that we observed .
could it also be responsible for the @xmath65 relation ?
in general , we know that metal and dust are accumulated through various processes in the galaxy evolution , e.g. , dust formation in type ii supernova and asymptotic giant branch stars @xcite .
therefore , galaxies at lower redshifts will typically have more dust .
this global trend is opposite to the @xmath65 relation we observed for quasars .
however , we know that quasar host galaxies are atypical .
many observations have shown the presence of large dust mass in high redshift @xmath66 quasars ; though , the origin of the large amount of dust in such early epochs is unclear ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and references therein ) .
moreover , there are also clues that the extinction curves of redshift @xmath66 quasars might be different from their lower redshift counterparts ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
despite the variations , we have not yet found any statistical evidence that quasars at higher redshifts show systematically higher intrinsic extinctions , at least for the redshift range we probed ( @xmath67 ) .
nevertheless , we emphasize that such a possibility still opens .
besides the possible evolution of the intrinsic dust , could other physical properties ( e.g. , the black hole accretion process ) of quasars evolve ?
since the black hole accretion is a local physical process , we do not expect that they show significant redshift evolution effects once their basic physical parameters are fixed ( e.g. , black hole mass or accretion rate ) . in this work ,
we group quasars into @xmath68 bins and do not further consider the variation of other physical parameters .
however , we have tested that whether , even after constraining quasars to small black hole mass @xmath69 and eddington ratio @xmath70 ranges ( 0.1 dex ) in the original 2d @xmath68 bins , the anti - correlation between @xmath12 and redshift still exists .
quasars are hosted by galaxies , but they outshine them in most cases
. however , for the low luminosity quasars at low redshift , the contamination of starlight might still be significant , especially at the near - uv band , i.e. , @xmath2 . for a @xmath31 @xmath71 quasar ,
a bright star - forming host galaxy with @xmath72 contributes @xmath73% of the flux in the @xmath74 range .
more than that , the redshift range of the near - uv band we probed is from @xmath75 to @xmath76 .
this is the time when the cosmic star formation rate ( sfr ) starts to decrease .
therefore , it is possible that there are larger contributions of the starlight at higher redshift , so that makes the quasar spectra look redder . to test this possibility
, we generate a series of model galaxy spectra at different redshifts according to the cosmic star formation history given in and their luminosities are assumed to be @xmath72 .
we then combine the galaxy spectra at different redshifts with the composite spectrum of @xmath31 @xmath77 quasars .
again , we measure the @xmath2 for the synthetic quasar+galaxy spectra at different redshifts . we find that , globally , @xmath2 becomes smaller ( redder ) when the galaxy spectra is combined into the quasar continuum .
however , on the other hand , we do not find that the combined spectra get redder toward high redshift .
the reason is that , while going to the higher redshifts from @xmath78 to @xmath79 , because of the higher sfr , not only does the contribution of the starlight in the near - uv band become larger , but the near - uv band spectrum of the galaxy itself also becomes bluer .
sdss quasars are selected using a complicated color - based algorithm @xcite .
quasar candidates at high and low redshift are selected based on different color - color diagrams .
therefore , the color - selection criteria might introduce biases to the intrinsic colors of quasars at different redshifts .
for example , is it possible that the redder quasar continua at higher redshifts are caused by the missing of red quasars at low redshift or vice versa ? to test this possibility , we take the reddest ( the last magenta diamond in fig .
[ fig4 ] ) composite spectrum and put it at different redshifts ( from @xmath80 to 0 ) and calculate their observed colors .
we do the same thing for the bluest one ( the first green triangle ) .
we find that the tracks of the colors of these two composites along the redshift still assemble in the quasar color - color selection area . from this test
, we argue that there is no significant selection bias in the colors of quasars at different redshifts in sdss .
in this letter , we use a large non - bal radio - quiet quasar sample from sdss dr7 to study the redshift and bolometric luminosity dependence of quasar continuum slope by means of the composite spectrum .
consistent with previous studies , we find that the uv - opt spectrum becomes harder toward higher bolometric luminosity , and we suggest that the internal dust reddening might be the main reason for this luminosity dependence . when the bolometric luminosity is controlled , we find that the quasar uv continuum slope is anti - correlated with redshift ( redder at higher redshift ) .
we propose that the redshift dependence of the quasar uv slope might be caused by the reddening of the cosmic dust .
we build a simple cosmic dust extinction model to quantify the reddening of the quasars . based on our model , we find that an effective dust density @xmath6 at @xmath7 can reasonably reproduce the @xmath57 relation .
the @xmath59 relation at high redshift ( @xmath61 ) can also be easily reproduced by using a shallow extinction curve or adopting a smaller effective density of the cosmic dust .
the cosmic dust extinction constrained from our model is globally consistent with the constrains from other independent measurements , and has reached to an unprecedented high redshift ( @xmath81 ) .
we thank the referee for a careful reading and highly appreciate the comments and suggestions .
x.x.y . thanks peng jiang in ustc for help with our code writing , hengxiao guo and fangting yuan in shao for helpful discussions , and brice mnard in jhu for providing their model data .
this work was supported by the strategic priority research program `` the emergence of cosmological structures '' of the chinese academy of sciences ( cas ; grant xdb09030200 ) , the national natural science foundation of china ( nsfc ) with the project numbers 11433003 , 11390373 , and 11103058 , and the `` 973 program '' 2014 cb845705 .
asano , r. s. , takeuchi , t. t. , hirashita , h. , & nozawa , t. 2014 , , 440 , 134 avgoustidis , a. , verde , l. , & jimenez , r. 2009 , jcap , 06 , 012 carballo , r. , gonzlez - serrano , j. i. , benn , c. r. , snchez , s. f. , & vigotti , m. 1999 , , 306 , 137 cardelli , j. a. , clayton , g. c. , & mathis , j. s. 1989 , , 345 , 245 cheng , f. h. , gaskell , c. m. , & koratkar , a. p. 1991
, , 370 , 487 corasaniti , p. s. 2006 , , 372 , 191 czerny , b. , li , j. , loska , z. , & szczerba , r. 2004 , , 348 , l54 davis , s. w. , woo , j .- h .
, & blaes , o. m. 2007 , , 668 , 682 francis , p. j. , hewett , p. c. , foltz , c. b. , & chaffee , f. h. 1992 , , 398 , 476 gaskell , c. m. , goosmann , r. w. , antonucci , r. r. j. , & whysong , d. h. 2004 , , 616 , 147 gaskell , c. m. , & benker , a. j. 2007 , arxiv:0711.1013 gu , m. 2013 , , 773 , 176 hewett , p. c. , & wild , v. 2010 , , 405 , 2302 hubeny , i. , agol , e. , blaes , o. , & krolik , j. h. 2000 , , 533 , 710 jiang , l. , fan , x. , ivezi , . , et al .
2007 , , 656 , 680 kennefick , j. , & bursick , s. 2008 , , 136 , 1799 kuhn , o. , elvis , m. , bechtold , j. , & elston , r. 2001 , , 136 , 225 labita , m. , treves , a. , & falomo , r. 2008 , , 383 , 1513 lawrence , a. 1991 , , 252 , 586 madau , p. , & dickinson , m. 2014 , , 52 , 415 mnard , b. , nestor , d. , turnshek , d. , et al .
2008 , , 385 , 1053 mnard , b. , scranton , r. , fukugita , m. , & richards , g. 2010a , , 405 , 1025 mnard , b. , kilbinger , m. , & scranton , r. 2010b , , 406 , 1815 mrtsell , e. , & goobar , a. 2003 , jcap , 09 , 009 more , s. , bovy , j. , & hogg , d. w. 2009 , , 696 , 1727 nozawa , t. , asano , r. s. , hirashita , h. , & takeuchi , t. t. 2015 , , 447 , l16 pentericci , l. , rix , h .- w .
, prada , f. , et al .
2003 , , 410 , 75 peterson , b. m. 1997 , an introduction to active galactic nuclei , publisher : cambridge , new york cambridge university press , 1997 physical description xvi , 238 p. isbn 0521473489 richards , g. t. , fan , x. , newberg , h. j. , et al .
2002 , , 123 , 2945 schlegel , d. j. , finkbeiner , d. p. , & davis , m. 1998 , , 500 , 525 schneider , d. p. , richards , g. t. , hall , p. b. , et al .
2010 , , 139 , 2360 shen , y. , richards , g. t. , strauss , m. a. , et al .
2011 , , 194 , 45 simpson , c. 2005 , , 360 , 565 stoughton , c. , lupton , r. h. , bernardi , m. , et al .
2002 , , 123 , 485 telfer , r. c. , zheng , w. , kriss , g. a. , & davidsen , a. f. 2002 , , 565 , 773 valiante , r. , schneider , r. , salvadori , s. , & bianchi , s. 2011 , , 416 , 1916 vanden berk , d. e. , richards , g. t. , bauer , a. , et al .
2001 , , 122 , 549 wild , v. , & hewett , p. c. 2010 , arxiv:1010.2500 wright , e. l. 1981 , , 250 , 1 york , d. g. , adelman , j. , anderson , j. e. , jr .
, et al .
2000 , , 120 , 1579 | we investigate the luminosity and redshift dependence of the quasar continuum by means of the composite spectrum using a large non - bal radio - quiet quasar sample drawn from the sloan digital sky survey .
quasar continuum slopes in the uv - opt band are measured at two different wavelength ranges , i.e. , @xmath0 ( @xmath1 ) and @xmath2 ( @xmath3 ) derived from a power - law fitting .
generally , the uv spectra slope becomes harder ( higher @xmath4 ) toward higher bolometric luminosity . on the other hand , when quasars are further grouped into luminosity bins , we find that both @xmath0 and @xmath2 show significant anti - correlations with redshift (
i.e. , the quasar continuum becomes redder toward higher redshift ) .
we suggest that the cosmic dust extinction is very likely the cause of this observed @xmath5 relation .
we build a simple cosmic dust extinction model to quantify the observed reddening tendency and find an effective dust density @xmath6 at @xmath7 .
the other possibilities that could produce such a reddening effect have also been discussed . |
elements heavier than iron are mainly synthesized by the rapid neutron capture process ( @xmath0-process ) as well as the slow neutron capture and proton capture processes .
more than 90% of the number of elements such as europium ( eu ) , gold ( au ) , and platinum ( pt)-process elements ] in the solar system are synthesized by @xmath0-process @xcite .
sufficiently neutron - rich environment is required in order to synthesize @xmath0-process elements with mass number ( @xmath6 ) over 110 .
+ the observed @xmath0-process elemental abundance ratios , such as [ eu / fe ] = @xmath7 , where @xmath8 and @xmath9 are number densities of elements a and b , respectively . ] , in extremely metal - poor ( emp ) stars ( [ fe / h ] @xmath103 ) show large star - to - star scatters .
these scatters are seen in emp stars in the milky way ( mw ) halo as well as the local group ( lg ) dwarf spheroidal galaxies ( dsphs ) ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the mechanism of star - to - star scatters in the abundance studies should be clarified simultaneously with the astrophysical site(s ) of @xmath0-process .
+ neutrino - driven winds from proto - neutron stars of core - collapse supernovae ( ccsne ) have long been regarded as one of the possible sites of @xmath0-process ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
previous chemical evolution studies suggest that the observed [ eu / fe ] scatter is well reproduced by models assuming that ccsne of low - mass ( 810 @xmath11 ) progenitors produce @xmath0-process elements ( e.g. , * ? ? ?
* ; * ? ? ?
? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . however , recent hydrodynamical simulations of ccsne , which include neutrino transport in sophisticated manner , suggest that the pns winds of ccsne do not necessarily produce a neutron - rich condition suitable for @xmath0-process ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
nucleosynthesis calculations suggest that heavy elements with @xmath12 110 , are difficult to be synthesized in ccsne due to such too weak neutron - rich environments ( e.g. , * ? ? ?
* ; * ? ? ?
+ binary neutron star mergers ( nsms ) are also suggested to be a promising site of @xmath0-process @xcite .
recent detailed nucleosynthesis calculations show that heavy @xmath0-process elements are successfully synthesized in nsms @xcite .
in addition , near infrared afterglow of the _ swift _ short gamma - ray burst , grb130603b @xcite is detected .
this is suggested to be a piece of evidence that progenitors of short gamma - ray bursts are compact binary mergers and @xmath0-process nucleosynthesis occurs there @xcite .
+ nsms have the long merger time and the low rate .
the neutron star ( ns ) binaries lose their energy very slowly due to the gravitational emissions . the merger time is estimated @xmath13 100 myr from observed binary pulsars @xcite .
recent predictions from population synthesis models also suggest that most of ns binaries merge @xmath13 100 myr after their formation @xcite . on the other hand ,
the nsm rate is estimated to be @xmath14
@xmath15 yr@xmath5 for a mw size galaxy from the observed binary pulsars @xcite .
+ however , early studies of galactic chemical evolution pointed out that it is difficult to reproduce the observed trend of [ eu / fe ] of emp stars by nsms due to their long merger time and the low rate @xcite .
most of other recent studies also conclude that short merger time ( @xmath3 10 myr ) or a second site of @xmath0-process such as jet - like explosions of magnetorotational ccsne ( e.g. , * ? ? ?
* ; * ? ? ?
* ) is required to account for large star - to - star scatters in emp stars @xcite . on the other hand
, detailed population synthesis calculations suggest that production of nsms with short - merger time ( @xmath2 10 myr ) highly depends on the treatment of common envelop phase which is not well understood ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
@xcite suggest that nsms with merger time of @xmath3 10 myr can not be produced in their most pessimistic model assuming each common envelop with an hertzsprung gap donor causes a merger ( submodel b in * ? ? ?
in addition , there are no observational clues that exist binary pulsars which merge within @xmath2 10 myr so far @xcite .
+ this discrepancy may be solved , if the galactic halo is formed via mergers of sub - halos within the framework of hierarchical structure formation scenario @xcite .
@xcite calculate the enrichment of @xmath0-process elements by nsms with merger time of 100 myr ( 95 % of nsms ) and 1 myr ( 5 % of nsms ) in their one - zone chemical evolution model for each sub - halo .
they suggest that [ eu / fe ] increases at [ fe / h ] @xmath16 if the star formation efficiencies are lower in less massive sub - halos . according to their calculations , the observed scatters in [ eu / fe ] in metal - poor stars are possibly explained by nsms with merger time of 100 myr .
key factors of chemical evolution such as the time variation of the star formation rate ( sfr ) , outflow , and inflow strongly depend on thermodynamical feedback from sne .
detailed chemo - dynamical evolution studies in low - mass galaxies like sub - halos are highly desirable to justify their assumptions to describe the enrichment history of @xmath0-process elements in a self - consistent manner between dynamical and chemical evolution of galaxies . + recent hydrodynamical studies have performed a series of simulations of galaxy formation assuming that the nsms are a major site of @xmath0-process @xcite .
@xcite suggest that gas mixing processes such as galactic winds and hydrodynamic flows play the important roles to explain the observed galactic @xmath0-process ratio .
their high - resolution model with @xmath17 of the mass of one gas particle ( @xmath18 ) is , however , difficult to reproduce the observed @xmath0-process abundance ratios .
they imply that additional metal mixing is required to explain the observation .
@xcite also suggest that the observed @xmath0-process abundance ratios are possibly taken into account in their nsm models if the metal mixing in star - forming region is implemented . however , the mass resolution ( @xmath19 ) in their models is as low as the fiducial low - resolution model of @xcite ( @xmath20 ) .
it is therefore important to demonstrate that if the nsm models account for the observations in independent simulations of much higher resolution .
+ in this paper , we calculate enrichment of @xmath0-process elements ejected by nsms in low mass galaxies with high - resolution ( @xmath21 ) chemo - dynamical evolution models .
we perform @xmath1-body / smoothed particle hydrodynamics ( sph ) simulations of dsph models using asura code @xcite .
we discuss effects of metal mixing in star - forming region as well as the dependence on the sfr , the merger time , and the rate of nsms .
+ in 2 , we introduce our code and models . in 3 , we compare model predictions and observed properties of dsphs generated by our models . in 4 ,
we discuss enrichment of @xmath0-process elements in dsphs .
finally in 5 , we summarize our main results .
we perform a series of simulations using an @xmath1-body / sph code , asura @xcite .
we adopt three different kinds of particles in our simulations : dark matter , gas , and star particles .
we treat dark matter and star particles as collisionless particles .
dark matter particles contribute to the dynamical evolution of our model galaxies .
star particles mainly contribute to feedback of energy and heavy elements produced by ccsne .
we solve the hydrodynamical evolution of gas particles using sph .
+ here we describe implementation of gravity and hydrodynamics in asura .
gravity is calculated by a treecode @xcite with grape method @xcite using the phantom - grape library @xcite .
hydrodynamics is computed using a standard sph method . for integration of self - gravitating fluid system
, we adopt the fully asynchronous split time - integrator ( fast ) algorithm in order to reduce calculation cost .
we use the time - step limiter , which forces the timestep difference among neighbor particles to be less than four times long , in order to follow the evolution of strong shock regions such as supernova ( sn ) remnants .
we use a metallicity dependent cooling / heating function generated by cloudy @xcite .
the cooling / heating function covers the temperature range from 10 k to 10@xmath22 k. + we allow star formation when gas particles satisfy three conditions .
( 1 ) gas particles are conversing ( @xmath23 0 ) . ( 2 ) the density is higher than threshold density , @xmath24 . ( 3 ) the temperature is lower than threshold temperature , @xmath25 .
we adopt @xmath24 = 100 @xmath26 for our fiducial model , which is the mean density of giant molecular clouds ( gmcs ) .
we adopt @xmath25 = 1000 k for our fiducial model .
the value of @xmath25 is insensitive to the final structure of galaxies @xcite .
+ when a gas particle satisfies the three conditions above , it becomes eligible to form new collisionless star particles .
star particles are produced by the following probability according to the prescription of @xcite : @xmath27 where @xmath28 and @xmath18 are the mass of star and gas particles , respectively , and @xmath29 is the dimensionless star formation efficiency ( sfe ) parameter .
we set @xmath30 following @xcite .
the mass of a gas particle of our fiducial model is initially assumed as @xmath31 , while it is reduced by star formation . when the mass of a gas particle becomes lower than one - third of initial mass , the particle is converted into a collisionless particle .
the dimensionless sfe parameter of our fiducial model ( @xmath29 = 0.033 ) is chosen based on the slow star - formation model @xcite .
@xcite suggest that when @xmath24 = 100 @xmath26 is adopted , the final results are fairly insensitive to the adopted value of @xmath29 in their mw model . in the appendix , we also confirm this result in our dsph models .
+ each star particle is treated as a single stellar population ( ssp ) , i.e. each star particle is assumed to be an assembly of stars with the same age and the same metallicity .
the initial mass function ( imf ) of star particles is the salpeter imf : @xmath32 , where @xmath33 @xcite with mass range of 0.1100@xmath11 .
we set the progenitor mass of ccsne to be 840 @xmath11 . in this model , stars more massive than 40 @xmath11 end their lives as black holes .
star particles , which explode in a time interval of @xmath34 are selected by the following probability ( @xmath35 ) , @xmath36 where @xmath37 is the turn off mass at age @xmath38 .
each ccsn explosion distributes thermal energy of @xmath39 erg to the surrounding sph particles .
the mass of one star particle is @xmath40 .
when a particle explode as ccsne , the number of ccsne inside each star particle corresponds to @xmath41 .
this method is also adopted in other studies ( e.g. , * ? ? ?
* ; * ? ? ?
in addition to the sn feedback , we implement heating by h@xmath42 region formed around young stars .
the number of the ly@xmath43 photons is evaluated using pgase @xcite .
the parameters of these baryonic physics are listed in table [ baryon ] .
llrr[htbp ] dimensionless sfe parameter&@xmath29&0.033&0.033 , 0.5 + threshold density for star formation&@xmath24&100 @xmath26&0.1100 @xmath26 + threshold temperature for star formation&@xmath25&1@xmath4410@xmath45 k&1@xmath4410@xmath45 3@xmath4410@xmath46 k + sn explosion energy&@xmath47&1 @xmath4410@xmath48 erg & ( 0.03 1 ) @xmath4410@xmath48erg we take into account both ccsne and nsms in our models .
we set initial gas metallicity equals to zero .
ccsne produce fe and nsms produce eu , which is regarded as a representative element of @xmath0-process .
binary black hole - neutron star mergers are also expected to eject @xmath0-process elements ( e.g. , * ? ? ?
* ; * ? ? ?
. however , they affect the rate of production of @xmath0-process elements by several factors which are much smaller than the uncertainty of the rate of nsms .
we therefore only implement nsms for simplicity .
we assume that gas particles around a star particle are enriched with metals when a ccsn or nsm occurs in a star particle .
metals are distributed in 32 nearest neighbor particles using weights of sph kernel .
mass of element x in the @xmath49th neighbor particle ejected by @xmath50th star particle , @xmath51 , is given by @xmath52 where @xmath53 is the distance between particle @xmath50 and @xmath49 , @xmath54 is the smoothing length , and @xmath55 is the sph kernel given by a cubic spline function ( e.g. , * ? ? ? * ) and the density of the gas particles is given as @xmath56 smoothing length is the scale of containing @xmath57 nearest neighbor particles .
following @xcite , we set @xmath57 = 32@xmath582 as a fiducial value .
+ the iron yield of ccsne are taken from @xcite .
the scope of this paper is to discuss the abundance ratio of @xmath0-process elements in emp stars before type ia supernovae ( sne ia ) contribute .
we do not implement sne ia in our simulation .
+ the nsm rate and the merger time ( @xmath59 ) are highly uncertain .
we therefore vary them @xmath2 2 dex in our simulations .
we regard a number fraction of nsms to the total number of neutron stars , @xmath60 , as a parameter , which determines the nsm rate . in this model , we assume the mass range of ns progenitor mass as 820 @xmath11 .
we set the upper mass of nsm progenitor stars as 20 @xmath11 from the lower limit of the mass of a black hole formation @xcite .
we set @xmath60 = 0.01 as a fiducial value .
the corresponding nsm rate in a milky - way size galaxy is @xmath2 10@xmath61 yr@xmath5 .
it is within the values of the galactic disk @xmath210@xmath6210@xmath63 yr@xmath5 , estimated from observed compact binaries @xcite .
+ yields of nsms are related to the rate of nsms .
[ eu / fe ] at [ fe / h ] = 0 is expected to be @xmath2 0.5 without sne ia because solar fe is estimated to be produced @xmath26065% by sne ia , and @xmath23540% by ccsne ( e.g. , * ? ? ?
* ; * ? ? ?
we thus simply set the yield of @xmath0-process elements to be [ eu / fe ] = 0.5 at [ fe / h ] = 0 .
+ observed @xmath0-process elemental abundance ratio such as [ eu / fe ] indicates that the production of @xmath0-process elements should have occurred before sne ia start to contribute ( @xmath13 1 gyr ) to galactic chemical evolution @xcite .
minimum merger time of nsms needs to be shorter than the typical delay time of sne ia .
as already mentioned , the most plausible merger time of nsms is regarded as @xmath2 100 myr ( e.g. , * ? ? ?
* ; * ? ? ?
we thus set the merger time of nsms is 100 myr as a fiducial value .
the abundance of a star must be identical to the abundance of the gas , which formed the star .
the abundance of a newly formed star particle inherits ( 1 ) the abundance of the star - forming gas particle ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) ; ( 2 ) the abundance of the average of gas particles within a sph kernel ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
method ( 1 ) does not have a metal mixing process except for hydrodynamical mixing process such as stellar winds , outflows and inflows due to the sn explosion .
[ eu / fe ] produced in method ( 1 ) is discussed in [ mix1 ] . on the other hand , in method ( 2 ) , a metal mixing process is taken into account .
we use a metallicity averaged over 32 neighbor gas particles in a sph kernel to a newly born star particle .
the region can be regarded as star - forming region , which corresponds to @xmath64 .
this mass corresponds to the typical size of giant molecular clouds ( gmcs ) ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
we discuss results inferred from method ( 2 ) in [ mix2 ] .
+ massive stars tend to be born in clusters and associations @xcite .
clusters and ob associations form from gmcs .
observations of stars in open clusters suggest that their metallicity is homogeneous @xcite . @xcite
theoretically show that turbulent mixing in star - forming regions causes this homogeneity .
the timescale of metal mixing is determined by the local dynamical time of star - forming regions ( @xmath3 1 myr ) .
this timescale is much shorter than the typical timescale of star formation ( @xmath1310 myr ) in slow star formation model @xcite .
we thus assume that the metals are instantaneously mixed in star - forming regions .
we discuss the effect of different implementation on abundance ratios in galaxies in
[ mix1 ] and [ mix2 ] .
we follow initial conditions of dsph models adopted in @xcite and @xcite .
we assume the density profile of dark matter , @xmath65 , as follows : @xmath66 where @xmath67 is the central density and @xmath68 is the core radius .
gas particles are also distributed along with the profile .
we set @xmath69 = 1.0 kpc , @xmath70 = 7.1 kpc , and @xmath71 = 7 @xmath72 , according to @xcite .
we adopt 0.15 of mass ratio of gas to dark matter particles ( baryon fraction , @xmath73 ) .
the value of @xmath74 is taken from @xcite .
+ following @xcite , we assume an isotropic velocity dispersion of dark matter particles , @xmath75 , for spherical distribution @xcite , @xmath76 where @xmath77 is the gravitational potential . for gas particles , we set a velocity equal to zero and an initial temperature of @xmath78 k. for both dark matter and gas particles , we adopt gravitational softening length ( @xmath79 ) of 28 pc for runs with the initial total number ( @xmath1 ) of @xmath80 , @xmath79 = 14 pc for runs with @xmath81 , @xmath82 , and @xmath83 , and @xmath79 = 7 pc for runs with @xmath1 = @xmath84 .
we run our simulations over 14 gyr .
the parameters of our model galaxies are listed in table [ model ] .
table [ list ] summarizes all runs discussed in this paper .
lll[htbp ] total mass&@xmath71&7@xmath4410@xmath85 + core radius&@xmath69&1 kpc + initial outer radius&@xmath70&7.1 kpc + baryon fraction&@xmath74&0.15 llrrrcrllcrl s000&@xmath84&2.3&0.4&7&100&1&1&32&no&100&0.01 + m000&@xmath84&2.3&0.4&7&100&1&1&32&yes&100&0.01 + mn16&@xmath84&2.3&0.4&7&100&1&1&16&yes&100&0.01 + mn64&@xmath84&2.3&0.4&7&100&1&1&64&yes&100&0.01 + m014&@xmath80&72.6&12.8&28&100&1&1&32&yes&100&0.01 + m016&@xmath86&18.2&3.2&14&100&1&1&32&yes&100&0.01 + m017&@xmath82&9.1&1.6&14&100&1&1&32&yes&100&0.01 + m018&@xmath83&4.5&0.8&14&100&1&1&32&yes&100&0.01 + mext&@xmath84&2.3&0.4&7&0.1&30&0.03&32&yes&100&0.01 + mt10&@xmath84&2.3&0.4&7&100&1&1&32&yes&10&0.01 + mt500&@xmath84&2.3&0.4&7&100&1&1&32&yes&500&0.01 + mr0.001&@xmath84&2.3&0.4&7&100&1&1&32&yes&100&0.001 + mr0.1&@xmath84&2.3&0.4&7&100&1&1&32&yes&100&0.1
we discuss chemo - dynamical evolution of model s000 to confirm that the parameter set of this model is appropriate for the case of dsphs .
parameter dependence is discussed in the appendix .
+ figure [ gas ] shows the evolution of the spatial distribution of gas and stars of model s000 .
upper panels of figure [ gas ] show the gas density maps of model s000 at 0 gyr , 1 gyr , 5 gyr , and 10 gyr from the beginning of the simulation .
the gas monotonically collapses during the first 1 gyr
. then , gas density reduces by star formation and gas outflow by energy feedback of sne .
red colored area in upper panels of figure [ gas ] corresponds to the star - forming region where the number density of gas is larger than 100 @xmath26 . as shown in this figure , star - forming region
is strongly confined at the center of the galaxy .
the red area is largest at 1 gyr .
this indicates that star formation is most active after @xmath2 1 gyr from the beginning of the simulation .
+ lower panels of figure [ gas ] show stellar density maps at 0 gyr , 1 gyr , 5 gyr , and 10 gyr .
the distribution of stars at 1 gyr is associated with the high - density region of gas ( red region in figure [ gas ] ) .
as shown in these figures , stars continuously form in the inner region of this model galaxy for over 10 gyr and stellar density distribution expands with time from the center to the outer region . at 10 gyr ,
the morphology of the galaxy becomes spherical symmetry .
+ in order to quantitatively discuss the structural and dynamical properties of models , we investigate the radial profiles .
figure [ rp ] shows the radial profiles of model s000 .
we define the galactic center using potential minimum .
the values in each point are calculated in each bin from the center to the outer region .
+ figure [ rp ] ( a ) shows time variation of the dark matter density profile . at 1 gyr ,
the dark matter follows the initial density profile given in eq .
( [ pseudo ] ) . after the collapse in the first 1 gyr ,
the shape of the dark matter profile does not change over 10 gyr .
+ figure [ rp](b ) shows time variation of the gas density profile .
inner region ( @xmath3 1 kpc ) of the gas density profile follows the evolution of the dark matter density profile . outer region of gas is blown away due to the outflow induced by sne .
in addition , the total amount of gas reduces because of star formation . however , gas still remains even at 10 gyr .
as in the appendix , all of our models have gas at 10 gyr .
the observed lg dsphs in contrast have no or little gas ( e.g. * ? ? ?
this result suggests that physical processes such as ram pressure and tidal stripping are required to remove all gas away from dsphs @xcite . +
as shown in figure [ rp ] ( c ) , the stellar density profile of our simulation well reproduces observations . in figure [ rp ] ( c ) , we present the stellar density profiles .
stars distribute within @xmath2 1 kpc , which is consistent with the observed tidal radii ( @xmath20.5 3 kpc ) of dsphs in the lg @xcite .
the density profile of stars is basically associated with the dark matter density profile .
+ figure [ rp ] ( d ) shows the stellar velocity dispersion profile .
the observed stellar velocity dispersion of dsphs is almost constant within @xmath2 1 kpc from the center @xcite .
model s000 has similar properties with the observed radial stellar velocity dispersion profiles inside 1 kpc from the center in the lg dsphs .
figure [ sfr ] shows the time variation of the sfr in model s000 ( red curve ) and the observed values of the fornax and the sculptor dsphs @xcite .
the sfr of model s000 is peaked at @xmath2 2 gyr .
gas density increases with accretion ( see figure [ rp ] ( b ) ) and finally reaches the threshold density for star formation . on the other hand , sn feedback drives gas away from the inner region to the outer region @xcite . because of the shallow gravitational potential and high threshold density for star formation ( @xmath24 = 100 @xmath26 ) , sn feedback significantly affects the timescale of gas accretion .
it therefore takes long time ( @xmath2 1 gyr ) to reach the peak of the sfr .
the sfr of model s000 ( @xmath87 yr@xmath5 ) is consistent with the observed value of the fornax and the sculptor dsphs inferred from color - magnitude diagram analysis ( @xmath87 yr@xmath5 , * ? ? ?
* ; * ? ? ?
the sfr of model mext ( magenta curve ) is discussed in [ p_m ] .
metallicity distribution is one of the best properties to test reliability of chemical evolution models .
figure [ md ] compares metallicity distribution between model s000 and observation .
all data of the fornax @xcite and the sculptor dsphs @xcite .
metallicity distribution of model s000 is almost consistent with the observed value of the sculptor dsph .
the metallicity at the peak of the distribution of model s000 is [ fe / h ] = @xmath881.57 , which is lower than that of the fornax dsph , [ fe / h ] = @xmath881.06 @xcite .
this is because we do not implement sne ia in this model , while the fornax dsph must be significantly affected by the metal ejection of sne ia ( e.g. , * ? ? ?
if we take account of the products of sne ia , the peak metallicity is expected to shift by @xmath2 0.5 dex to the higher metallicity , which is closer value to that of the fornax dsph . +
in this section , we discuss [ eu / fe ] as a function of [ fe / h ] predicted in model s000 . in model s000
, the metallicity of a star particle inherits that of the gas particle from which the star particle was formed , according to the method ( 1 ) in [ metal - stars ] .
figure [ rnomix ] shows [ eu / fe ] as a function of [ fe / h ] predicted in model s000 .
we also put the observed data of the galactic halo and several dsphs , i.e. , carina , draco , leo i , sculptor , and ursa minor dsphs ( saga database , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , excluding carbon - enhanced stars , which are possibly affected by gas transfer in binaries .
we also exclude stars in the fornax dsph because some of them have extremely high [ eu / fe ] ( @xmath89 0.5 dex ) due to significant contamination of @xmath90-process @xcite .
as shown in figure [ rnomix ] , stars of highly @xmath0-process enhanced stars ; [ eu / fe ] @xmath89 1 ( so - called @xmath0-ii stars ) , are over - abundant . in addition , @xmath0-deficient stars in @xmath91 [ fe / h ] @xmath92 are predicted
. such low [ eu / fe ] stars are not seen in the observation .
these stars are not simply caused by delayed production of eu by nsms .
in fact , the average value of [ eu / fe ] does not increase with metallicity at around [ fe / h ] @xmath93 .
the significant dispersions of chemical components among gas particles seem to be rather essential .
the relations between the galactic age and abundances of fe and eu show the reason why such large dispersions are seen in low metallicity region .
+ figure [ agefe ] ( a ) shows [ fe / h ] as a function of time .
metallicity obviously increases with time , as ccsne produce iron .
the metallicity however has more scatters at the earlier time , especially during the first few gyrs .
then , later formed stars are enriched by more numerous ccsne , and as a result , the dispersion of stellar metallicity decreases with time .
we denote the stars , which are formed from the gas enriched only by a single ccsn by black circles in figure [ agefe ] ( a ) .
their metallicity widely distributes over @xmath2 3 dex .
these stars concentrate only in @xmath3 2 gyr .
+ figure [ agefe ] ( b ) shows [ eu / h ] as a function of time .
in contrast to figure [ agefe ] ( a ) , large star - to - star scatters in [ eu / h ] remain over the whole evolution of the galaxy .
as shown in black circles in figure [ agefe ] ( b ) , gas particles affected by one nsm remain over 10 gyr .
one of the reasons must be the low rate of nsms .
the rate of nsms is one hundred times lower than that of ccsne in this model
. the total number of nsms may not be enough to converge the [ eu / h ] in this model .
+ in addition to the low nsm rate , the efficiency of gas mixing in this model seems to cause unnaturally large scatters in [ eu / h ] .
if a star particle contains products from a single nsm , the value of [ eu / h ] must be determined by the distance from the nsm , which enriched the star - forming gas particle . however , as shown in figure [ agefe ] ( b ) , dispersion of such stars drags longer than 5 10 gyrs , which is much longer than the merger time of nsms .
it implies that gas particles never change the abundance of eu , unless other nsms enrich them again , though in actual , gas clouds are expected to interact with others .
+ in fact , observations of open clusters show that stellar metallicity is quite homogeneous in each cluster ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the gas in star - forming region is possibly homogenized by hydro - dynamical effects , such as turbulent mixing @xcite .
+ in addition , previous studies suggest that the standard sph simulations without metal mixing tend to predict a lower amount of gas with low metallicity and higher metallicity of inter galactic medium @xcite .
@xcite suggest that it is difficult to reproduce the observed [ eu / fe ] as a function of [ fe / h ] without metal mixing in star - forming region .
+ on the other hand , the fiducial model of @xcite reproduces the observed [ eu / fe ] of metal - poor stars , although they adopt the same definition of metallicity as our model s000 .
they suggest that large - scale metal mixing such as galactic winds and hydrodynamical flows is important to reproduce the observed [ eu / fe ] as a function of [ fe / h ] . in their model , nsms eject metals in the region of @xmath94 , which is much larger than the swept - up mass of nsms ( @xmath95 ) . the treatment of @xcite is identical to implement metal mixing .
as discussed in the previous section , the effect of mixing of enriched gas must be essential to account for the observed values of [ eu / fe ] in metal - poor stars .
therefore , we take account of the effect of metal mixing according to the method ( 2 ) in [ metal - stars ] , and we adopt the average metallicity of gas particles in the sph kernel of the progenitor gas particle for the metallicity of newly formed stars .
figure [ rmix ] shows [ eu / fe ] as a function of [ fe / h ] with the metal mixing model , m000 .
model m000 has the same parameter set for s000 except for the effect of metal mixing in star - forming region ( see table [ list ] ) . as shown in this figure ,
@xmath0-deficient stars in [ fe / h ] of @xmath96 to @xmath97 seen in figure [ rnomix ] disappear due to metal mixing .
the fraction of @xmath0-ii stars is also reduced in this model due to adopted averaged metallicity in star - forming region .
model m000 apparently reproduces the observational tendency of [ eu / fe ] in metal - poor stars much better than the model s000 .
our model does not require the assumption of short merger time ( @xmath98 10 myr ) , which is required to reproduce observations in previous studies ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
+ figure [ eu_fe_mix ] shows [ eu / fe ] distributions in stars of [ fe / h ] @xmath99 predicted in s000 ( without mixing model ) and m000 ( with mixing model ) .
observational values of the mw ( red histogram ) and dsphs ( blue histogram ) are provided by the saga database @xcite .
while model s000 overproduces stars with [ eu / fe ] @xmath100 , model m000 significantly reduces the fraction of @xmath0-deficient stars .
in addition , the fraction of @xmath0-ii stars also reduces in model m000 .
this result therefore suggests that metal mixing in star - forming region is fairly important physical process to reproduce the observed [ eu / fe ] as a function of [ fe / h ] .
+ the predicted distribution of [ eu / fe ] must be affected by the mass of the mixed gas ( @xmath101 ) , and the initial total number of particles ( @xmath1 ) .
we define @xmath102 .
the mass of one gas particle ( @xmath103 ) is proportional to @xmath104 .
table [ mixing_region ] lists the examined values of parameters and the corresponding mixing mass ( @xmath101 ) .
figure [ convergence ] ( a ) shows the median value and the dispersion of [ eu / fe ] of models , which have different @xmath105 .
the dispersion of [ eu / fe ] becomes smaller with larger @xmath57 as shown in this figure .
when we increase the @xmath57 , the value of @xmath101 increases and larger mass fraction of gas in the galaxy is mixed . due to the effects of increasing @xmath101 ,
the dispersion becomes smaller in the models with larger value of @xmath57 .
+ figure [ convergence ] ( b ) shows the median value of [ eu / fe ] as a function of [ fe / h ] of models , which have different @xmath1 . as shown in figure [ convergence ] ( b ) , the dispersion of [ eu / fe ] shows the similar tendency irrespective of @xmath1 , i.e. , it decreases with increasing of metallicity . when we adopt the larger value of @xmath1 , more star particles are produced , i.e. , the number of events of metal mixing increases .
this means that metals are more mixed in the star - forming region if @xmath106 is constant .
however , @xmath101 is defined to be proportional to @xmath104 .
models with larger value of @xmath1 have smaller value of @xmath106 .
the effect of increasing @xmath1 offsets the effect made by decreasing @xmath106 .
thus , the dispersion is not affected by @xmath1 .
lrrr m000&2@xmath107&32&1.3 + mn16&2@xmath107&16&0.6 + mn64&2@xmath107&64&2.7 + m018&2@xmath108&32&2.6 + m017&2@xmath109&32&5.1 + m016&2@xmath110&32&10.3 + m014&2@xmath111&32&41.0 + the value and scatters in the abundance ratio of [ eu / fe ] must be affected by sfr , especially for metal - poor stars . in order to examine the effect of sfr
, we discuss [ eu / fe ] as a function of [ fe / h ] with the model mext , in which extremely weak sn feedback energy ( @xmath112 erg ) and low threshold density for star formation ( @xmath24 = 0.1 @xmath26 ) are assumed ( table [ list ] ) .
according to appendix [ app ] , the weaker feedback model produces higher sfr and model with smaller value of @xmath24 shifts the peak of sfr to the earlier time .
since mext has lower values of @xmath47 and @xmath24 , stars are more easily produced comparing to the model m000 .
the sfr of model mext during the first 1 gyr rises rapidly and is larger than @xmath113yr@xmath5 .
the peak of sfr reaches @xmath114yr@xmath5 , while that of m000 is @xmath115yr@xmath5 .
the sfr of model mext at @xmath3 1 gyr is much higher than the observational values ( figure [ sfr ] ) .
+ figure [ fhrm ] shows [ eu / fe ] as a function of [ fe / h ] in model mext .
this model predicts significantly different [ eu / fe ] from m000 ( figure [ rmix ] ) although both models adopt the same initial distribution of gas particles .
model mext produces a distribution of @xmath0-deficient stars around [ fe / h ] @xmath1162 .
in addition , no stars are produced below [ fe / h ] @xmath1163 .
+ this difference is related to the timescale of chemical evolution in the early phase of the galaxy formation . for model m000 ,
the median metallicity of gas particles at 1 gyr is [ fe / h ] = @xmath883.32 . on the other hand ,
the median metallicity of gas particles in mext is [ fe / h ] @xmath1170.91 at 1 gyr .
the sfe of model mext is estimated to be @xmath1181 gyr@xmath5 .
the value of sfe is comparable to some other inhomogeneous chemical evolution studies @xcite .
model mext proceeds chemical evolution much faster than m000 due to the high sfr of mext ( @xmath119yr@xmath5 ) .
the [ eu / fe ] as a function of [ fe / h ] in model mext is inconsistent with the observation due to fast chemical evolution by the high sfr .
this result suggests that the sfr and sfe in the early phase of dsphs are @xmath120yr@xmath5 and @xmath3 0.10 gyr@xmath5 , respectively , if the @xmath0-process elements are ejected by nsms with a long merger time ( @xmath2100 myr ) . as shown in figure
[ sfr ] , the sfr of @xmath87yr@xmath5 is a reasonable agreement with the observed value of the fornax dsph @xcite .
this sfr is also consistent with sub - halo models of @xcite ( case 1 in their model suggests that the appropriate value of star formation efficiency for a sub - halo with a stellar mass of @xmath121 should be 0.10 gyr@xmath5 , which corresponds to the order of sfr as @xmath87yr@xmath5 ) .
they also suggest that the observed [ eu / fe ] scatter in metal - poor stars by nsms with a long merger time . in this section ,
we discuss the effect of merger time of nsms .
figure [ delay ] shows resultant [ eu / fe ] as a function of [ fe / h ] by nsms with different merger time ( @xmath59 ) .
eu in figure [ delay ] ( a ) and ( b ) are produced by nsms with @xmath59 = 10 myr ( mt10 ) and 500 myr ( mt500 ) , respectively . although mt10 has a slightly smaller fraction of stars in @xmath122 [ fe / h ] @xmath123 than model m000 , the global relative abundance ratio is similar to m000 ( @xmath59 = 100 myr ) .
contrary to the models m000 and mt10 , the model with much longer merger time such as 500 myr in mt500 shows large scatters in [ eu / fe ] at higher metallicity and can not account for the observed scatters in [ fe / h]@xmath124 . + figure [ age_log ] shows [ fe / h ] as a function of the substantial galactic age , i.e. , the elapsed time from the rise of the major star formation . as shown in figure 3
, we can regard that the major star formation arises from 600 myr from the beginning of the calculation .
the average metallicity of stars is almost constant during the first @xmath2 300 myr . due to low star formation efficiency of the galaxy , spatial distribution of metallicity
is highly inhomogeneous in @xmath3 300 myr . in this epoch ,
since most of gas particles are enriched only by a single sn , metallicity of stars is mainly determined simply by the distance from each sn to the gas particles which formed the stars .
therefore , nsms with @xmath125 myr can account for the observation of emp stars , as well as those with @xmath126 myr .
in contrast , metallicity is well correlated with the galactic age after @xmath2 300 myr , irrespective of the distance from each sn to the gas particles .
because sn products have already been well mixed in a galaxy , the stellar metallicity is determined by the number of the sne , which enriched the stellar ingredients .
therefore , if the merger time of nsms is much longer than @xmath2 300 myr , it is too long to reproduce observations .
+ the yields of @xmath0-process elements in our models are related to the nsm rate as already mentioned in
[ chem ] , though the galactic rate of nsms is highly uncertain .
the estimated galactic nsm rate is 10@xmath62 to 10@xmath63 yr@xmath5 based on three observed binary pulsars @xcite .
table [ yield ] lists yields of models discussed here .
figure [ rate ] shows predicted [ eu / fe ] as a function of [ fe / h ] assuming different nsm rate .
figure [ rate ] ( a ) and ( b ) represent models with the nsm fractions @xmath60 = 0.001 ( mr0.001 ) and @xmath60 = 0.1 ( mr0.1 ) , respectively .
the corresponding nsm rate in a mw - like galaxy is @xmath127 yr@xmath5 ( mr0.001 ) and @xmath128 yr@xmath5 ( mr0.1 ) .
model mr0.001 predicts larger scatter and a smaller number of stars at [ fe / h ] @xmath1293 than m000 .
model mr0.001 has [ eu / fe ] dispersion by more than 3 dex at [ fe / h ] = @xmath882 .
in addition , there remains @xmath2 1 dex dispersion even for stars with [ fe / h ] @xmath1302 .
in contrast , model mr0.1 predicts smaller scatter than m000 , though it does not seem to be inconsistent with observations .
such tendencies are also seen in @xcite , @xcite and @xcite . + our fiducial model , m000 , reproduces the observed @xmath0-process ratio as discussed in [ mix2 ] .
the nsm rate of m000 for a mw - like galaxy is @xmath4 yr@xmath5 .
the total mass of @xmath0-process elements produced by each nsm corresponds to @xmath119 .
the value is consistent with recent nucleosynthesis calculations : @xmath131 to @xmath113 ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
+ @xcite construct an inhomogeneous chemical evolution model of the mw halo .
their model is difficult to reproduce [ eu / fe ] by nsms with the galactic nsm rate of @xmath132 yr@xmath5 due to high star formation efficiency .
[ eu / fe ] produced in their model is similar to that of mext ( figure [ fhrm ] ) .
+ from the discussion above , nsm rate of @xmath2 10@xmath61 yr@xmath5 in a mw size galaxy is preferred to reproduce the observed [ eu / fe ] .
this rate is consistent with the estimated galactic nsm rate from the observed binary pulsars @xcite . near future gravitational detectors , kagra , advanced ligo , and advanced virgo @xcite
are expected to detect 10 100 events per year of gravitational wave from nsms .
+ lll mr0.001&0.001&10@xmath5 + m000&0.01&10@xmath133 + mr0.1&0.1&10@xmath63
we have carried out numerical simulations of chemo - dynamical evolution of dsphs using @xmath1-body / sph code , asura to investigate the enrichment history of @xmath0-process elements .
this study suggests that nsms with the merger time of @xmath2 100 myr and the galactic event rate of @xmath4 yr@xmath5 can explain the dispersion of [ eu / fe ] in reasonable agreement with observations in emp stars .
this study supports that nsms are the major astrophysical site of the @xmath0-process .
our isolated dsph models reproduce basic properties of the observed lg dsphs such as radial profiles , time variations of the sfr as well as metallicity distribution . here
, we summarize the main results . *
the abundance ratio of [ eu / fe ] produced in our models without metal mixing in star - forming regions has too large dispersion .
this is because metals in a gas particle do not diffuse out to mix into the other particles throughout the evolution of galaxies .
* models with metal mixing in star - forming region reproduce the observed [ eu / fe ] distribution and its scatter as a function of [ fe / h ] .
our model shows good convergence of the resolution .
we show that nsms with @xmath59 = 100 myr is favorable for reproducing the observed [ eu / fe ] as a function of metallicity .
this result implies that the metal mixing process is critical to reproduce the [ eu / fe ] distribution .
in addition , this study suggests that the sfr of dsphs in the early epoch of their evolution @xmath41gyr is @xmath120yr@xmath5 . * the nsms with merger time of @xmath3 300 myr is acceptable to account for the observed abundance of emp stars .
this is because metallicity is not correlated with time up to @xmath2 300 myr from the start of the simulation due to low star formation efficiency of the model galaxy .
* this study suggests that the galactic nsm rate to account for the observed @xmath0-process abundance scatters is @xmath4 yr@xmath5 .
the total mass of @xmath0-process elements ejected by one nsm is @xmath119 , which is consistent with the value suggested by nucleosynthesis studies ( @xmath15@xmath113 ) .
next generation gravitational detectors kagra , advanced ligo and advanced virgo are expected to detect gravitational wave from nsms and their event rate would be over 10 per year
. their detections will give us more reliable galactic nsm rate .
in this study , we have focused on the enrichment history of @xmath0-process elements in isolated dsphs with fixed mass . to fully understand the enrichment history of @xmath0-process elements in the lg galaxies , it is important to show how the mass and size of galaxies affect the enrichment of @xmath0-process elements .
recent observations suggest that the low abundance of @xmath0-process elements ( [ ba / fe]-process element in [ fe / h ] @xmath134 @xmath100 ) in stars with [ fe / h ] @xmath135 .
these stars provide clues to understand the astrophysical site(s ) of @xmath0-process elements and the metal enrichment in the first galaxies .
some studies suggest that the @xmath0-process abundance of these stars can be explained by short merger time channel ( @xmath2 1 myr ) of nsms @xcite or accretion of materials from inter stellar medium to the population iii stars @xcite . since we have only focused on the star - to - star scatters of [ eu / fe ] in stars with [ fe / h ] @xmath1163 assuming that nsms are the major site of @xmath0-process in this paper , we have not discussed the origin of these stars .
it is profitable to discuss the origin of these stars by the detailed simulation of galaxies . to address all of these issues , it is required to understand how the mw halo formed .
we need to clarify the relation between the building block galaxies of the mw and the present lg dsphs .
it is now underway to extend our numerical simulations of chemodynamical evolution of dsphs to larger scale simulations of the mw in order to fully understand the enrichment of @xmath0-process elements in the mw and confirm the validity of the scenario of hierarchical structure formation .
we appreciate the referee for giving us profitable comments to improve our paper .
we are grateful to masaomi tanaka , shinya wanajo , grant j. mathews , and yuichiro sekiguchi for fruitful discussion .
we also appreciate for takuma suda for providing us the latest version of saga database including the dataset of the lg dsphs .
this work is supported in part by mext spire and jicfus .
yh is financially supported by grant - in - aid for jsps fellows ( 15j00548 ) .
the authors are financially supported by jsps grants - in - aid for scientific research ( yi : 26400232 , trs : 26707007 , msf : 26800108 , tk : 26105517 , 24340060 , 15h03665 ) . numerical computations and analysis were in part carried out on cray xc30 and computers at center for computational astrophysics , national astronomical observatory of japan .
we compare models with different threshold densities for star formation ( @xmath24 ) , dimensionless sfe parameters ( @xmath29 ) , and sn feedback energies ( @xmath47 ) .
table [ list_ap ] lists all models discussed here .
figure [ rp_appendix ] shows the radial profiles of our models .
horizontal axis of figure [ rp_appendix ] is the distance from the galactic center .
figure [ rp_appendix ] ( a ) shows the dark matter density profile . we find that all models have similar dark matter profiles .
the dark matter profile is not affected by physical parameters such as threshold density for star formation , dimensionless sfe parameters , and sn feedback energy .
+ gas density , stellar density , and stellar velocity dispersion profiles have variations among these models . the gas density of sn10 and se01 is lower than that of s000 ( figure [ rp_appendix ] ( b ) ) . in these models ,
most of gas is consumed in the early phase of their evolution .
the gas density of se01 is truncated at 0.3 kpc while the gas profiles of the other models continue over 10 kpc .
this is because the feedback energy in se01 is too weak to blow the gas away to the outer region of the galaxy .
+ stellar density profiles for all models ( figure [ rp_appendix ] ( c ) ) are truncated within a few kpc , which is consistent with the observed truncation radius ( @xmath20.5 3 kpc ) of dsphs in the lg @xcite . in sn10
, stars distribute to a larger radius than the other models .
when a low value of @xmath24 is used , stars can form in the outer region of the galaxy .
if the sn feedback is weak ( model se01 ) , stellar distribution concentrates on the central region of the galaxy .
+ figure [ rp_appendix ] ( d ) shows the velocity dispersion profile .
all models except for se01 are consistent with the observed radial velocity dispersion profiles inside 1 kpc from the center in the lg dsphs @xcite .
model se01 has higher velocity dispersion at the center of our model galaxy due to the high central concentration of stars .
+ in contrast to @xmath24 and @xmath47 , the dimensionless sfe parameter ( @xmath29 ) does not greatly affect the radial profiles .
figure [ rp_appendix ] shows all profiles of s000 ( red curve ) and sc50 ( blue curve ) are similar although they have different value of the dimensionless sfe parameter .
these features suggest that radial properties of galaxies are insensitive to the value of @xmath29 when we adopt a reasonable value of @xmath24 (= 100 @xmath26 ) @xcite .
+ figure [ sfr_appendix ] and [ md_appendix ] show sfr as a function of time and metallicity distribution , respectively .
sfr as a function of time and metallicity distribution are characterized by the threshold density for star formation and sn feedback energy .
model sn10 has lower @xmath24 (= 10 @xmath26 ) than that of s000 ( @xmath24 = 100 @xmath26 ) .
the second peak of sfr of sn10 is earlier than s000 .
this reflects time required to reach @xmath24 is shorter than s000 because of using low @xmath24 value in sn10 .
in addition , the first peak of sfr of sn10 is higher than s000 .
gas is consumed by star formation and removed by outflow at @xmath136 0.1 gyr in sn10 .
its sfr is therefore @xmath22 dex lower than s000 .
the early bursty star formation of sn10 produces more metal - poor stars than s000 ( see green dashed and red curves in figure [ md_appendix ] ) .
higher sfr of sn10 @xmath2 0.1 gyr produces more ccsne in this phase .
ccsne produce outflow .
model sn10 thus loses larger amount of gas around @xmath2 0.1 gyr .
chemical evolution of sn10 , therefore , quenches at @xmath137 0.1 gyr and produces only few metal - rich stars .
+ sn feedback energy also significantly affects the time variation of sfr and metallicity distribution .
the sn feedback energy of se01 ( @xmath138 erg ) is ten times smaller than that of s000 ( @xmath139 erg ) .
the peak of the sfr of se01 is over 1 dex higher than that of s000 .
sn feedback energy gives thermal energy to gas particles .
it prevents collapse of gas particles . as a result
, star formation is suppressed due to sn feedback energy .
sn feedback energy of se01 is too weak to suppress star formation .
the peak of sfr of se01 is thus @xmath2 1 dex higher than s000 .
gas is consumed for star formation due to high sfr around 2 gyr of se01 , and the sfr at @xmath140 4 gyr is eventually suppressed . due to the low sn feedback energy in model se01 ,
the peak of metallicity distribution of se01 is @xmath2 0.5 dex higher metallicity than s000 .
+ on the other hand , the value of the dimensionless sfe parameter does not significantly affect the results . models s000 and sc50 have different value of @xmath29 = 0.033 and 0.5 , respectively .
the time variation of the sfr and metallicity distribution of sc50 is however similar to that of s000 .
this result suggests that the time variation of the sfr and metallicity distribution are insensitive to the value of @xmath29 if we use a reasonable value of @xmath24 (= 100 @xmath26 ) .
slightly lower metallicity of sc50 than s000 is due to slightly lower sfr of sc50 than s000 .
this result suggests that the value of @xmath29 does not affect the metallicity distribution as well as radial profiles and the sfr .
when we adopt @xmath24 = 100 @xmath26 , it takes much longer local dynamical time to flow from the reservoir ( @xmath141 1 @xmath26 ) to the star - forming regions ( @xmath142 100 @xmath26 ) .
this timescale does not depend on @xmath29 @xcite .
our results are , thus , independent of @xmath29 .
+ these results suggest that the time variation of the sfr and metallicity distribution is significantly affected by the threshold density for star formation and sn feedback energy .
low @xmath24 model ( sn10 ) produces too many emp stars .
in contrast , low @xmath47 model ( se01 ) has too many metal - rich stars .
these differences in metallicity distribution are due to the difference of the time variation of the sfr among models . on the other hand ,
model s000 reproduces observation of metallicity distribution as well as dynamical properties .
parameters of s000 are taken from the observed values .
threshold density of s000 ( @xmath24 = 100 @xmath26 ) is taken from mean density of gmcs .
sn feedback energy of s000 ( @xmath47 = @xmath39 erg ) is taken from the canonical explosion energy of ccsne ( e.g. , * ? ? ?
we thus treat s000 as a model that has fiducial parameter sets .
abadie , j. , abbott , b. p. , abbott , r. et al .
2010a , cqgra , 27 , 173001 abadie , j. , abbott , b. p. , abbott , r. , et al .
2010b , nimpa , 624 , 223 accadia , t. , acernese , f. , antonucci , f. , et al .
2011 , cqgra , 28 , 114002 aoki , w. , suda , t. , boyd , r. n. , kajino , t. , & famiano , m. a. 2013 , , 766 , l13 argast , d. , samland , m. , gerhard , o. e. , & thielemann , f .- k .
2000 , , 356 , 873 argast , d. , samland , m. , thielemann , f .- k . , & qian , y .- z .
2004 , , 416 , 997 barnes , j. & hut , p. 1986
, , 324 , 446 bauswein , a. , goriely , s. , & janka , h .- t .
2013 , , 773 , 78 berger , e. , fong , w. , & chornock , r. 2013 , , 774 , l23 binney , j. , & tremaine , s. 2008 , galactic dynamics ( 2nd ed . , princeton , nj : princeton university press ) bubar , e. j. , & king , j. r. , 2010 , , 140 , 293 burris , d. l. , pilachowski , c. a. , armandroff , t. e. , et al .
2000 , , 544 , 302 cescutti , g. , romano , d. , matteucci , f. , chiappini , c. , & hirschi , r. 2015 , , 577 , a139 cioffi , d. f. , mckee , c. f. , & bertschinger , e. 1988 , , 334 , 252 de boer , t. j. l. , tolstoy , e. , hill , v. , et al .
2012a , , 539 , a103 de boer , t. j. l. , tolstoy , e. , hill , v. , et al .
2012b , , 544 , a73 de silva , g. m. , freeman , k. c. , asplund , m. , et al .
2007a , , 133 , 1161 de silva , g. m. , freeman , k. c. , bland - hawthorn , j. , asplund , m. , & bessell , m. s. 2007b , , 133 , 694 de silva , g. m. , freeman , k. c. , bland - hawthorn , j. , et al .
2011 , , 415 , 563 de silva , g. m. , dorazi , v. , melo , c. , et al .
2013 , , 431 , 1005 dominik , m. , belczynski , k. , fryer , c. , et al .
2012 , , 759 , 52 eichler , d. , livio , m. , piran , t. , & schramm , d. n. 1989 , , 340 , 126 feng , y. & krumholz , m. r. 2014 , , 513 , 523 ferland , g. j. , korista , k. t. , verner , d. a. , et al . 1998 , , 110 , 761 ferland , g. j. , porter , r. l. , van hoof , p. a. m , et al .
2013 , rmxaa , 49 , 137 fioc , m. , & rocca - volmerange , b. 1997 , , 326 , 950 franois , p. , depagne , e. , hill , v. et al .
2003 , , 403 , 1105 franois , p. , depagne , e. , hill , v. et al .
2007 , , 476 , 935 frebel , a. , kirby , e. n. , & simon , j. d. 2010a , , 464 , 72 frebel , a. , simon , j. d. , geha , m. , & willman , b. 2010b , , 708 , 560 frebel , a. , & norris , j. e. 2015 , , 53 , 631 freiburghaus , c. , rosswog , s. , & thielemann , f .- k .
1999 , , 525 , l121 fulbright , j. p. 2000 , , 120 , 1841 gingold , r. a. , & monaghan , j. j. 1977 , , 181 , 375 goriely , s. , bauswein , a. , & janka , h .- t .
2011 , , 738 , l32 goswami , a. & prantzos , n. 2000 , , 359 , 191 harris , w. e. , & pudritz , r. e. 1994 , , 429 , 177 heyer , m. , krawczyk , c. , duval , j. , & jackson , j. m. 2009 , , 699 , 1092 hernquist , l. 1993 , , 86 , 389 honda , s. , aoki , w. , kajino , t. , et al .
2004 , , 607 , 474 horowitz , c. j. , shen , g. , oconnor , e. , & ott , c. d. 2012 , , 86 , 065806 hotokezaka , k. , kiuchi , k. , kyutoku , k. , et al .
2013 , , 87 , 024001 hopkins , p. f. , quataert , e. , & murray , n. 2011 , , 417 , 950 ishigaki , m. n. , aoki , w. , & chiba , m. 2013 , , 771 , 67 ishimaru , y. , & wanajo , s. 1999 , , 511 , l33 ishimaru , y. , wanajo , s. , aoki , w. , & ryan , s. g. 2004 , , 600 , l47 ishimaru , y. , wanajo , s. , & prantzos , n. 2015 , , 804 , l35 irwin , m. , & hatzidimitriou , d. 1995 , , 277 , 1354 johnson , j. a. 2002 , , 139 , 219 katz , n. 1992 , , 391 , 502 katz , n. , weinberg , d. h. , & hernquist , l. 1996 , , 105 , 19 kauffmann , g. , white , s. d. m. , & guiderdoni , b. 1993 , , 264 , 201 kawata , d. 2001 , apj , 558 , 598 kinugawa , t. , inayoshi , k. , hotokezaka , k. , nakauchi , d. , & nakamura , t. , 2014 , , 442 , 2963 kirby , e. n. , & cohen , j. g. 2012 , , 144 , 168 kirby , e. n. , cohen , j. g. , guhathakurta , p. , et al .
2013 , , 779 , 102 kirby , e. n. , guhathakurta , p. , bolte , m. , sneden , c. , & geha , m. c. 2009 , , 705 , 328 kirby , e. n. , guhathakurta , p. , simon , j. d. , et al .
2010 , , 191 , 352 kobayashi , c. , & nakasato , n. 2011 , , 729 , 16 komiya , y. , yamada , s. , suda , t. , & fujimoto , m. y. 2014 , , 783 , 132 korobkin , o. , rosswog , s. , arcones , a. , & winteler , c. 2012 , , 426 , 1940 krumholz , m. r. , & tan , j. c. 2007 , , 654 , 304 kuroda , k. , & lcgt collaboration 2010 , cqgra , 27 , 084004 lada , c. j. , & lada , e. a. 2003 , ar , 41 , 57 larson , r. b. 1981 , , 194 , 809 lattimer , j. m. , & schramm , d. n. 1974 , , 192 , l145 lattimer , j. m. & schramm , d. n. 1976 , , 210 , 549 lattimer , j. m. , mackie , f. , ravenhall , d. g. , & schramm , d. n. 1977 , , 213 , 225 lemasle , b. , hill , v. , tolstoy , e. , et al . 2012 , , 538 , a100 letarte , b. , hill , v. , tolstoy , e. , et al .
2010 , , 523 , a17 ligo scientific collaboration 2013 , arxiv : 1304.0670 liszt , h. s. , delin , x. , & burton , w. b. 1981 , , 249 , 532 lorimer , d. r. 2008 , lrr , 11 , 8 lucy , l. b. 1977 , , 82 , 1013 makino , j. 1991 , , 43 , 621 maoz , d. , mannucci , f. 2012 , , 29 , 447 mennekens , n. , & venbeveren , d. 2014 , , 564 , a134 meyer , b. s. 1989 , , 343 , 254 mayer , l. , mastropietro , c. , wadsley , j. , stadel , j. , & moore , b. 2006 , , 369 , 1021 martnez - pinedo , g. , fischer , t. , lohs , a. , & huther , l. 2012 , , 109 , 251104 mathews , g. j. , & cowan , j. j. 1990 , , 345 , 491 mathews , g. j. , bazan , g. , & cowan , j. j. 1992 , , 391 , 719 matteucci , f. , romano , d. , arcones , a. , korobkin , o. , & rosswog , s. 2014 , , 438 , 2177 mcconnachie , a. w. 2012 , , 144 , 4 mcwilliam , a. , preston , g. w. , sneden , c. , & leonard , s. 1995 , , 109 , 2757 mcwilliam , a. 1998 , , 115 , 1640 meyer , b. s. , mathews , g. j. , howard , w. m. , woosley , s. e. , & hoffman , r. d. 1992 , , 399 , 656 monaghan , j. j. 1992 , ar , 30 , 543 monaghan , j. j. , & lattanzio , j. c. 1985 , , 149 , 135 moore , b. , ghigna , s. , governato , f. , et al . 1999 , , 524 , l19 navarro , j. f. , & white , s. d. m. 1993 , , 265 , 271 nichols , m. , revaz , y. , & jablonka , p. 2014
, , 564 , a112 nishimura , n. , takiwaki , t. , & thielemann , f .- k .
2015 , , 810 , 109 nomoto , k. , tominaga , n. , umeda , h. , kobayashi , c. , & maeda , k. 2006 , , 777 , 424 norris , j. e. , ryan , s. g. , & beers , t. c. 2001 , , 561 , 1034 okamoto , t. , eke , v. r. , frenk , c. s. , & jenkins , a. 2005 , , 363 , 1299 okamoto , t. , jenkins , a. , eke , v. r. , quilis , v. , & frenk , c. s. 2003 , , 345 , 429 okamoto , t. , nemmen , r. s. , & bower , r. g. 2008 , , 385 , 161 pancino , e. , carrera , r. , rossetti , e. & gallart , c. 2010 , , 511 , a56 planck collaboration xvi 2014 , , 571 , a16 portegies zwart , s. f. , & yungelson , l. , r. , , 332 , 173 prantzos , n. 2008 , , 489 , 525 qian , y .- z .
2000 , , 534 , l67 qian , y .- z . , &
woosley , s. e. 1996 , , 471 , 331 raiteri , c. m. , villata , m. , gallino , r. , busso , m. , & cravanzola , a. 1999 , , 518 , l91 reddy , a. b. s. , giridhar , s. & lambert , d. l. 2012 , , 419 , 1350 reddy , a. b. s. , giridhar , s. & lambert , d. l. 2013 , , 431 , 3338 reddy , s. , prakash , m. , & lattimer , j. m. 1998 , , 58 , 013009 revaz , y. , & jablonka , p. 2012
, , 538 , a82 revaz , y. , jablonka , p. , sawala , t. , et al .
2009 , , 501 , 189 roberts , l. f. 2012 , , 755 , 126 roberts , l. f. , reddy , s. , & shen , g. 2012 , , 86 , 065803 rosswog , s. , korobkin , o. , arcones , a. , thielemann , f .- k . , & piran , t. 2014 , , 439 , 744 ryan , s. g. , norris , j. e. , & beers , t. c. 1996 , , 471 , 254 saitoh , t. r. , daisaka , h. , kokubo , e. , et al .
2008 , , 60 , 667 saitoh , t. r. , daisaka , h. , kokubo , e. , et al .
2009 , , 61 , 481 saitoh , t. r. , & makino , j. 2009 , , 697 , l99 saitoh , t. r. , & makino , j. 2010 , , 62 , 301 salpeter , e. e. 1955 , , 121 , 161 sanders , d. b. , scoville , n. z. , & solomon , p. m. 1985 , , 289 , 373 shen , s. , cooke , r. , ramirez - ruiz , e. , madau , p. , mayer , l. , & guedes , j. 2015 , , 807 , 115 shen , s. , wadsley , j. , & stinson , g. 2010 , , 407 , 1581 shetrone , m. d. 1996 , , 112 , 1517 sneden c. , cowan , j. j. , & gallino , r. 2008 , , 46 , 241 solomon , p. m. , rivolo , a. r. , barrett , j. , & yahil , a. 1987 , , 319 , 730 steinmetz , m. , & mller , e. , 1994 , , 281 , l97 stinson , g. , seth , a. , katz , n.,wadsley , j. , governato , f. , & quinn , t. 2006 , , 373 , 1074 suda , t. , katsuta , y. , yamada , s. , et al .
2008 , , 60 , 1159 suda , t. , yamada , s. , katsuta , y. , et al .
2011 , , 412 , 843 suda , t. , hidaka , j. , ishigaki , m. , et al .
2014 , , 85 , 600 symbalisty , e. , & schramm , d. n. 1982 , apl , 22 , 143 tanaka , m. , & hotokezaka , k. 2013 , , 775 , 113 tanikawa , a. , yoshikawa , k. , okamoto , t. , & nitadori , k. 2012 , , 17 , 82 tanvir , n. r. , levan , a. j. , fruchter , a. s. , et al .
2013 , , 500 , 547 thornton , k. , gaudlitz , m. , janka , h. -th . , & steinmetz , m. 1998 , , 500 , 95 ting , y. -s . ,
de silva , g. m. , freeman , k. c. , & parker , s. j. 2012 , , 427 , 882 tolstoy , e. , hill , v. , & tosi , m. 2009 , , 47 , 371 tsujimoto , t. , & shigeyama , t. 2014 , , 565 , l5 tsujimoto , t. , shigeyama , t. , & yoshii , y. 2000 , , 531 , l33 travaglio , c. , galli , d. , gallino , r. , busso , m. , ferrini , f. , straniero , o. 1999 , , 521 , 691 travaglio , c. , galli , d. , & burkert , a. 2001 , , 547 , 217 van de voort , f. , quataert , e. , hopkins , p. f. , kere , d. , & faucher - gigure , c .- a .
2015 , , 447 , 140 walker , m. g. , mateo , m. , olszewski , e. w. , et al .
2007 , , 667 , l53 walker , m. g. , mateo , m. , olszewski , e. w. , et al .
2009 , , 704 , 1274 wanajo , s. 2013 , , 770 , l22 wanajo , s. , janka , h - t . , & mller , b. 2011 , , 726 , l15 wanajo , s. , kajino , t. , mathews , g. j. , & otsuki , k. 2001 , , 554 , 578 wanajo , s. , sekiguchi , y. , nishimura , n. , et al .
2014 , , 789 , l39 wehmeyer , b. , pignatari , m. , & thielemann , f .- k . , 2015 , , 452 , 1970 westin , j. , sneden , c. , gustafsson , b. , & cowan , j. j. 2000 , , 530 , 783 wiersma , r. p. c. , schaye , j. , theuns , t. , vecchia , c. d. , & tornatore , l. 2009 , , 399 , 574 winteler , c. , kppeli , r. , perego , a. et al .
2012 , , 750 , l22 woolf , v. m. , tomkin , j. , & lambert , d. l. 1995 , , 453 , 660 woosley , s. e. , wilson , j. r. , mathews , g. j. , hoffman , r. d. , & meyer , b. s. 1994 , , 433 , 229 yamada , s. , suda , t. , komiya , y. , aoki , w. , & fujimoto , m. y. , 2013 , , 436 , 1362 zuckerman , b. , & evans , n. j. , ii .
1974 , , 192 , l149 | the rapid neutron - capture process ( @xmath0-process ) is a major process to synthesize elements heavier than iron , but the astrophysical site(s ) of @xmath0-process is not identified yet .
neutron star mergers ( nsms ) are suggested to be a major @xmath0-process site from nucleosynthesis studies .
previous chemical evolution studies however require unlikely short merger time of nsms to reproduce the observed large star - to - star scatters in the abundance ratios of @xmath0-process elements relative to iron , [ eu / fe ] , of extremely metal - poor stars in the milky way ( mw ) halo .
this problem can be solved by considering chemical evolution in dwarf spheroidal galaxies ( dsphs ) which would be building blocks of the mw and have lower star formation efficiencies than the mw halo .
we demonstrate that enrichment of @xmath0-process elements in dsphs by nsms using an @xmath1-body / smoothed particle hydrodynamics code .
our high - resolution model reproduces the observed [ eu / fe ] by nsms with a merger time of 100 myr when the effect of metal mixing is taken into account .
this is because metallicity is not correlated with time up to @xmath2 300 myr from the start of the simulation due to low star formation efficiency in dsphs .
we also confirm that this model is consistent with observed properties of dsphs such as radial profiles and metallicity distribution . the merger time and the galactic rate of nsms
are suggested to be @xmath3 300 myr and @xmath4 yr@xmath5 , which are consistent with the values suggested by population synthesis and nucleosynthesis studies .
this study supports that nsms are the major astrophysical site of @xmath0-process . |
controlling the size , shape and placement of nanoparticles is a main concern in many processes of nanostructures fabrication . for instance
, the incorporation of plasmonic nanoparticles into photovoltaic devices can strongly increase their efficiency @xcite , since the surface plasmon resonance between metallic nanostructures depends on the size and spacing @xcite of the particles .
other applications include biodiagnostics and sensing , where au nanoparticles bind to specific dna markers , permitting their detection @xcite .
the potential applications for organized metallic nanostructures are wide ranging and include raman spectroscopy @xcite , catalysis @xcite , photonics @xcite , and spintronics @xcite . one strategy to create and organize structures at the nanoscale is to make use of the inherent self
assembly mechanisms of a material .
the combination of natural instabilities related to the physical properties of liquid metals , such as low viscosity and high surface energy , with the technological ability to lithographically pattern nanoscale features , has created an open field of research .
these techniques require the study of the liquid state dewetting dynamics @xcite leading to liquid instabilities @xcite , with the goal of directing the assembly to precise , coordinated nanostructures in one @xcite and two @xcite dimensions . within this context , the phenomena of wetting and dewetting of a shaped liquid film ( metallic or not ) covering a solid substrate are at the basis of all these processes and applications . however , even if these phenomena have been studied for the last two decades @xcite , several of the main issues related to them are still under discussion .
therefore , one of the purposes of this paper is to provide a clear description of the wetting and dewetting mechanisms , including effects of contact angle hysteresis , and their implications in the formation of liquid structures .
we consider in detail the dynamics of a long liquid filament sitting on a horizontal substrate under partial wetting conditions .
in particular , we focus our attention on the bulged region at its ends that is formed due to the filament retraction along its longitudinal direction .
this process not only involves a dewetting motion , but also a wetting one in the transverse direction . due to these
advancing and receding motions at the ends , the contact line becomes curved in the region connecting the bulge ( head ) with the rest of the filament .
this perturbation triggers an instability that leads to the appearance of a neck with rapidly decreasing width @xcite .
the breakup of this neck leads to two fluid portions which evolve separately .
on one side , the head separated from the filament recedes until a final drop shape is achieved .
the main features of this drop , such as its non circular footprint , was thoroughly studied in previous works @xcite . on the other side
, the remaining filament has a new shaped end as a result of the breakup process .
the following receding longitudinal motion of this new end of the filament leads to another bulged region and a neck formation in a successive fashion .
a group of pictures of this process , as obtained in this work for a silicon oil on a coated glass , is shown in fig .
[ fig : sequence](a ) . for comparison
, we also include in fig .
[ fig : sequence](b ) the analogous process observed at nanoscale @xcite for a filament of width @xmath0 nm and thickness @xmath1 nm , which was generated in the context of pulsed laser induce dewetting ( plid ) ( see e.g. @xcite and references cited there ) . the main goal of this work is to describe and understand the physical mechanisms involved in the evolution of the head and its retraction .
the flow before and after the breakup is studied experimentally by means of optical techniques , which allow to measure contact line positions , contact angles , thickness profiles and contact line contours .
these data are then compared with a simple heuristic model as well as numerical simulations , which solve the navier stokes equations with appropriate boundary conditions at the contact line .
note that the modeling of this flow implies the description of the contact line movement , and the subsequent pressure and velocity fields in a fluid domain confined by the substrate and the fluid free surface .
thus , the main issue in this modeling is how to determine the contact line motion , where both the stress and viscous dissipation rate diverge in the continuum description if slip is not allowed . here
, we consider a small amount of slip within the navier boundary condition , and consistently add a relationship between the velocity and the contact angle . since these relations must account for wetting and dewetting , we choose the combined approach proposed in @xcite , which accounts for both the hydrodynamic of the macroscopic region @xcite and the molecular kinetic @xcite for the contact line motion .
the relevant physical parameters in this approach , which leads to a constitutive relationship for the contact line dynamics , are determined here by an additional study of the hysteretic behavior of a sessile drop . in section [ sec : wet ]
, we obtain the main wettability properties of the substrate and liquid ( silicon oil ) system to be used in the filament experiments described in section [ sec : exp ] .
the experimental data are compared both with a simple heuristic model ( section [ sec : model ] ) and numerical simulations of the navier stokes equations ( section [ sec : num ] ) .
section [ sec : conclu ] is devoted to a summary and main conclusions of the work .
in order to perform the experiments , we employ a similar setup to that reported in @xcite , i.e. we use a substrate that is partially wetted by our working fluid , namely a silicon oil ( polydimethylsiloxane , pdms ) .
the substrate is a microscope slide ( glass ) which is coated with a fluorinated solution ( egc-1700 of 3 m ) using a chemat dip coater under controlled speed .
the pdms partially wets the substrate , since the coating lowers the surface energy of the glass .
in order to have reproducible wetting properties , we found that it is necessary to follow a protocol to get rid of the remaining solvent in the coating .
thus , the coated substrates are heated in an oven at @xmath2c for @xmath3 minutes , and left on a close and dry vessel for about @xmath4 days .
both the surface tension , @xmath5 , and density , @xmath6 , of the pdms are measured with a krss k11 tensiometer , while its viscosity , @xmath7 , is determined with a haake vt550 rotating viscometer .
the values obtained for these parameters are : @xmath8 dyn / cm , @xmath9 g/@xmath10 , and @xmath11 poise at temperature @xmath12c . in order to characterize the wettability of the pdms
, we employ a technique which advances or recedes the contact line by changing the volume of a sessile drop initially placed on the substrate @xcite .
this is accomplished by injecting or withdrawing liquid through of a needle ( connected to a syringe ) in contact with the apex of the drop ( see inset in fig .
[ fig : hyst_static](a ) ) .
the syringe volume is controlled by an automated dispensing system of a rame
hart model 250 goniometer , and the measurements are based on the analysis of the axisymmetric drop shape profile ( adsa p technique ) .
we distinguish here two scenarios : one in which the injection / withdrawal process is produced by pulses ( between which the drop is allowed to reach a static shape ) , and another one where this process is practically continuous . thus , in the first case we focus on the measurement of the static contact angle , @xmath13 , while in the second case we are concerned with the dynamic angle , @xmath14 , and its relationship with the drop contact line velocity .
we start with a drop of volume @xmath15l and use injection / withdrawal pulses of @xmath16 s duration at a flow rate @xmath17l / s .
this is done by means of an automated control of the piston motion in the syringe .
the pulses are separated by time intervals of about @xmath18 s , so that the drop has time enough to relax to its static shape after every volume variation , @xmath19l .
initially at @xmath20 , we have a sessile drop of volume @xmath21 with @xmath22 and null contact line displacement ( @xmath23 ) .
as the volume is increased , it spreads and reaches intermediate rest states on a dry surface ( not previously covered by the pdms ) till the maximum volume , @xmath24l , is achieved .
this is shown by the blue symbols in fig .
[ fig : hyst_static](a ) for @xmath13 ( full circles ) and @xmath25 ( hollow circles ) , respectively .
note that @xmath13 remains practically the same for all these volumes , while @xmath25 increases .
this stage is not part of the hysteresis cycle , since only contact line motions on a previously wetted surface are considered in its determination .
after this preliminary stage , the needle withdraws liquid until @xmath26 diminishes to @xmath27 @xmath28l ( see the portion of the cycle from a to b in fig .
[ fig : hyst_static](a ) : clockwise for @xmath14 , and counterclockwise for @xmath25 , as indicated by the arrows ) . finally , the increasing volume stage ( from b to a ) starts at @xmath29 till @xmath30 is reached again .
this process can be repeated indefinitely , but only the first hysteresis cycle is shown in fig . [
fig : hyst_static](a ) for brevity ( the full black circles stand for @xmath13 , while the hollow red ones correspond to @xmath25 ) . in order to fully understand the meaning of these curves in terms of the local behavior at the contact line , we plot @xmath13 versus @xmath25 in fig .
[ fig : hyst_static](b ) , and ignore @xmath26 since it is not a relevant parameter for this analysis . here
, we see that the front starts to have a forward displacement ( advance ) at @xmath31 to achieve a new static position , so that this angle is called the ( static ) advancing contact angle .
similarly , we define the ( static ) receding contact angle , @xmath32 , as the value of @xmath13 at which the front must have a backwards displacement ( recede ) to achieve equilibrium . as it will be shown in the following sections , these two limiting angles are related to the dynamic contact angle behavior .
another relevant couple of angles is given by the maximum and minimum values of @xmath13 , between which static drop states are possible .
this range is @xmath33 .
now , we consider the relationship between the dynamic contact angle , @xmath14 , and the contact line velocity , @xmath34 . here , we inject / withdraw the liquid at the same flow rate , @xmath35 , as before , but with longer and larger pulses . except for the first pulse ( at @xmath20 ) , which has a duration of @xmath36 s with a volume increase of @xmath37 @xmath28l , all the other ones last @xmath38 s with a volume variation of @xmath39 @xmath28l ( see dashed lines in fig . [
fig : hyst_dyn ] ) .
unlike the static case , we continuously measure the dynamic contact angle , @xmath40 , and the corresponding contact line displacement,@xmath41 , as the drop spreads / contracts and relaxes to equilibrium for about @xmath42 s. the values of @xmath14 and @xmath43 as a function of time are plotted in fig . [
fig : hyst_dyn ] .
note that during the injection and withdrawal stages @xmath14 varies in phase ( without delay ) with @xmath26 , while the effect of @xmath44 on the contact line displacement , @xmath43 , is delayed .
this is due to the fact that during these stages the drop volume variation is quickly absorbed by a modification of the contact angle , while the contact line does not move till @xmath14 reaches the critical contact angle ( @xmath45 for advancing , and @xmath46 for receding ) .
we derive numerically the data of @xmath43 versus @xmath47 to obtain the contact line velocity , @xmath48 .
the hollow circles in fig .
[ fig : theta_vcl ] show @xmath14 as a function of @xmath34 , using @xmath47 as a parameter .
interestingly , the @xmath14 range for which @xmath49 is practically coincident with the hysteresis range of @xmath13 , namely @xmath50 shown in fig .
[ fig : hyst_static](b ) .
this agreement confirms that the methodology used to determine both types of angles , namely static and dynamic , is appropriate to describe the wettability of the substrate by the pdms .
, as a function of the contact line velocity of the drop , @xmath34 ( hollow circles ) .
these data correspond to four cycles of the injection / withdrawal continuous flow at the drop apex .
the arrow points to the maximum dewetting velocity , @xmath51 , which corresponds to @xmath52 . ] in order to fully describe the dewetting stage , we must enlarge the velocity range for @xmath53 .
in fact , it is known that the dewetting velocity has a maximum absolute value at which the dynamic contact angle is zero @xcite , and our results in fig .
[ fig : theta_vcl ] for the dynamic cycle of the sessile drop can not show this phenomenon . with this goal
, we perform a different experiment which allows to obtain the maximum velocity of dewetting , referred here to as @xmath51 ( @xmath54 ) . in this new setup ,
we plunge a coated substrate into a deep pool of pdms , and withdraw it at a speed , @xmath55 , greater than @xmath51 .
this is achieved when , for a certain @xmath55 , the liquid is entrained off the pool , and forms a film on the substrate . once the substrate has stopped , the contact line of this film starts to slide down the surface due to gravity .
this dewetting motion proceeds at the maximum possible velocity , @xmath51 , with zero contact angle @xcite .
we perform experiments with different withdrawal speeds ( @xmath56 cm / s ) , and confirmed that the value of @xmath51 is independent of @xmath55 , as expected .
the contact lines on both sides of the coated glass are seen on the digital image due to the transparency of the substrate .
they are not exactly coincident , thus introducing some error in the measurement , which is at most @xmath57 and increases as the front dewets .
finally , by measuring the contact line position as a function of time , we are able to obtain the linear regression expression @xmath58 this value of @xmath51 is shown in fig .
[ fig : theta_vcl ] for @xmath52 , and it is of great importance to determine the parameters of the physical mechanism that describes the @xmath14@xmath34 relationship , as will be shown in the following section . in order to describe the motion of the contact line over the substrate ( @xmath59-plane ) , we employ two constitutive relations to account for the liquid solid interaction .
firstly , we overcome the stress singularity by relaxing the no slip boundary condition at the substrate through the navier formulation ( see e.g. @xcite ) , @xmath60 where @xmath61 is the slip length .
this parameter embraces somehow the main features related with the intermolecular forces in that region .
secondly , we use a relationship between the dynamic contact angle , @xmath14 , and the contact line velocity , @xmath34 , which is normal to the contact line itself .
since we need to describe not only advancing motions ( wetting , @xmath62 ) , but also receding ( dewetting , @xmath53 ) ones , we define the dynamic contact angle from a combination of the hydrodynamic and molecular kinetics models @xcite as first proposed in @xcite . for the former
, we have the cox - voinov relationship @xcite @xmath63 where @xmath64 is the microscopic contact angle , and @xmath65 , @xmath61 are macroscopic and microscopic length scales , respectively .
usually , the former is assumed to be a constant given by young s law , and independent of @xmath34 . here , we take @xmath65 as the capillary length , i.e. @xmath66 where @xmath67 is the gravity .
thus , we leave @xmath61 as a parameter to be determined by adjusting the experimental data from the final modeling . note , however , that eq .
( [ eq : theta_hyd ] ) with @xmath68 is not appropriate to adjust well the experimental data in fig .
[ fig : theta_vcl ] , which show a sharp jump at @xmath49 due to hysteresis effects .
then , we must resort to an alternative approach .
in fact , several authors @xcite admit the possibility of a non hydrodynamic velocity dependence of @xmath64 . here , we consider that it is given by the blake molecular kinetics model @xcite in the form , @xmath69 where @xmath70 is the ( microscopic ) equilibrium contact angle ( @xmath71 ) .
the other parameters , which are of molecular origin , are given by @xcite @xmath72 where @xmath73 is the frequency of molecular displacement at equilibrium , @xmath74 is the average length of each molecular displacement ( or distance between adsortion sites ) , @xmath75 is the temperature , and @xmath76 is the boltzmann s constant . since we do not focus our attention in the kinetic process itself , we consider @xmath77 and @xmath78 as fitting parameters of our modeling . on the other hand ,
since the surface energies are usually associated with the cosine of a contact angle , as in young s law , we consider the cosine of the equilibrium angle , @xmath70 , as an average of the cosines of two characteristic angles of the hysteresis range .
if we take the static advancing and static receding contact angles , @xmath45 and @xmath46 , respectively , we have @xmath79 which yields @xmath80 .
we could also have taken the pair @xmath81 instead of @xmath82 , in which case the angle would be @xmath83 . however , as we will discuss later , this choice does not lead to any improvement in the comparison with the experimental results . note that both eqs .
( [ eq : theta_hyd ] ) and ( [ eq : theta_bl ] ) account for the hysteresis of the contact line , since they provide the dynamic contact angle , @xmath14 , as a function of @xmath34 . in order to determine the parameters @xmath77 , @xmath78 , and @xmath61
, we fit the data from the dynamic cycling contact angle measurements ( circles in fig . [
fig : theta_vcl ] ) , and the value @xmath84 ( rhombic symbol in fig .
[ fig : theta_vcl ] , and eq . ( [ eq : vmax ] ) ) with the approximating function @xmath85 + 9 \frac{\mu v_{cl}}{\gamma } \ln \left ( \frac{a_c}{\ell } \right ) .
\label{eq : theta_fit}\ ] ] this combined cox - voinov - blake model ( for brevity _
cv+b _ , or hybrid model ) is solved with the following iterative procedure .
for a given pair @xmath86 we obtain @xmath87 such that @xmath88 ; then , with this value of @xmath87 we calculate a new pair @xmath89 as given by a square minimum method using eq .
( [ eq : theta_fit ] ) with the dynamic cycling measurements . by using this pair ,
we determine a new value of @xmath90 such as @xmath88 , and so on .
after some tens of iterations , we obtain the converged values @xmath91 the best fitting curve given by eq .
( [ eq : theta_fit ] ) with these parameters is shown in fig .
[ fig : theta_vcl ] ( see blue line ) .
the red line is the corresponding curve using @xmath92 instead of @xmath70 .
even if the difference between them is small and is mainly noticeable in the dewetting region , we will use in what follows the parameters in eq .
( [ eq : param ] ) for the blue curve in fig .
[ fig : theta_vcl ] .
these parameters allow us to estimate the molecular kinetic parameters for our system coated glass / pdms / air as given by eq .
( [ eq : v0_gamma ] ) .
thus , we find @xmath93 interestingly , @xmath74 is of the same order as those reported in @xcite , but @xmath73 is much smaller than for those systems with glycerol , likely due to the higher viscosity of pdms . at the same time , @xmath61 is at least ten times larger here than in @xcite .
the main dynamical process that we analyze here is the axial dewetting motion of a liquid filament placed on a horizontal and previously coated glass substrate @xcite .
we generate the filament from a vertical jet of pdms flowing out from a small nozzle at the bottom of a vessel filled with pdms .
it is captured on the substrate by horizontally flipping the substrate in a rigid frame . quickly afterwards
, the frame is rotated @xmath94 degrees and a second horizontal flip is performed so that a new ( auxiliary ) filament , perpendicular to the first one , is also captured on the substrate .
this other filament crosses the first one near one of its ends , so that the latter adopts a reproducible and characteristic rounded shape , whose axial dewetting will be observed and measured in detail . finally , the substrate is rapidly placed on a horizontal position .
all these movements take about @xmath95 seconds , which is a very short time interval compared to the time scale of the experiment itself due to the high viscosity of the pdms .
thus , we obtain a fluid filament of uniform width , @xmath96 , with parallel and straight contact lines , so that the initial configuration has a constant cross sectional area along its axis .
we calibrate the system by relating the fluid height in the vessel and the jet diameter for a given nozzle .
both the jet diameter and the corresponding filament widths ( for a contact angle equal to @xmath45 ) could be varied from @xmath97 to @xmath98 mm and from @xmath99 to @xmath100 mm , respectively .
a typical dewetting process of the filament end is shown in fig .
[ fig : sequence ] .
initially , it is rounded and has a contact angle of around @xmath101 so that the tip certainly recedes according to the wettability framework described above ( see e.g. fig .
[ fig : theta_vcl ] ) .
this motion induces the development of a head , whose width and thickness grow with time .
the evolution proceeds until the filament tip stops and , after a while , a narrower region ( neck ) starts forming in the filament somewhere beyond the bulged region . finally , the neck breaks up , so that the portion of fluid between the tip and the neck gives place to a static drop .
the characteristic shape of this sessile drop with non - circular footprint was previously studied in @xcite , but here we will focus on the description of the dynamical process previous to the rupture . this dewetting and breakup mechanism repeats itself starting from a new end of the filament , and finishes with the formation of a series of similar drops . here , we show results for the tracking of the first four heads at the left end of this filament . the experimental setup allows to observe , though not simultaneously , both top and side views of the filament evolution . in fig .
[ fig : side_top ] , we show the filament profiles for two similar , albeit not strictly equal , widths . from this type of profiles , we are able to extract the position of the tip , the contact angle there , as well as the thickness and width of the head .
these parameters are later compared with the results from an heuristic model and numerical simulations of the full hydrodynamic equations .
it should be noted that the first drop formed from the end of the filament is a consequence of a different breakup process ( numbered here as head @xmath102 ) than the following ones .
this is due to the fact that its preceding breakup occurs at crossings with a transversal auxiliary filament , unlike the rest of the drops .
since the initial conditions of the formation process determines important features of the dynamics , such as the distance travelled by the retracing tip till the following breakup , we shall restrict ourselves to the most repetitive case .
therefore , we exclude it in our analysis . in fig .
[ fig : xf - thex - h0_17 ] we show with symbols the time evolution of the axial position of the tip ( front ) , @xmath103 , the dynamic contact angle there , @xmath104 , and the maximum thickness of the head for the first four heads ( @xmath16 to @xmath105 ) formed at the left end of the filament of width @xmath106 ( see fig . [
fig : side_top](a ) ) .
the origin of @xmath103coordinates for successive heads is the point where each breakup took place .
the almost perfect superposition of the data for all three quantities ( @xmath103 , @xmath104 , and @xmath107 ) indicates that the dewetting process is repeatable after each breakup . from the data presented in fig .
[ fig : xf - thex - h0_17 ] , we generate the corresponding @xmath104@xmath34 relation , as shown by symbols in fig . [
fig : the_vcl - fil ] .
interestingly , these spontaneous filament dewetting data are in good agreement with the previously obtained relationship @xmath108 for the drop under the cyclic procedure of forced wetting and dewetting motions ( solid line in fig . [
fig : the_vcl - fil ] ) .
this fact confirms that the previously obtained relationship certainly describes a local phenomenon , and is independent of the geometry of the flow .
, versus the contact line velocity , @xmath34 , as obtained from the data in fig .
[ fig : xf - thex - h0_17 ] .
the solid line corresponds to the _ cv+b _
relationship , eq .
( [ eq : theta_fit ] ) , with the parameters given in eq .
( [ eq : param ] ) . ] the measured evolution of the footprint of a filament can be seen in fig .
[ fig : conos ] for two different filament widths ( see also fig . [
fig : side_top](b ) ) .
the results show that all footprints are inside an angular sector whose borders are always tangent at some point @xmath109 of the contact line .
the aperture of this sector is fixed , so that the slopes of the straight lines are constant during the axial retraction process , except when the neck behind the head becomes very narrow .
the straight lines satisfy the expression @xmath110 cm , which means that they intersect at @xmath111 cm .
we take this point as the origin of coordinates for the following discussions , so that this envelope can be described as straight lines with slopes @xmath112 .
the physical meaning of this result will become apparent from the analysis of the full simulations in section [ sec : num ] . in fig .
[ fig : xf - w_20 ] we show the tip position , @xmath103 , the maximum width at the head , @xmath113 , and the minimum width at the neck , @xmath114 , as a function of time for the filament width @xmath115 cm .
as expected , the curve for @xmath103 is very similar to that in fig .
[ fig : xf - thex - h0_17](a ) .
in particular , that for early times , say @xmath116 s , the data for both widths are practically coincident , the main difference being for later times when @xmath103 approaches the corresponding limiting value .
on the other hand , fig .
[ fig : xf - w_20](b ) shows @xmath113 and @xmath114 , which are obtained as the maximum and minimum values of the width along the filament .
note that the growth rate of the head slightly diminishes when the neck starts developing ( @xmath117 s ) .
in order to develop a simple model that accounts for the main features of the dynamic process of dewetting , we consider that the shape of the head at the end of the filament can be approximated as a part of an ellipsoidal cap ( see fig . [
fig : model ] ) .
this simple geometrical configuration continuously connects to the rest of the filament of circular cross section .
diameters along @xmath118 , @xmath119 and @xmath120 axes are @xmath121 , @xmath122 and @xmath123 , respectively , which we consider as constants during the whole evolution .
the distance between the center of the ellipsoid and the substrate in the vertical direction is @xmath124 , and the cap has length @xmath125 along @xmath118-axis and width @xmath126 in the @xmath119 direction , as shown in fig .
[ fig : model](b ) . and
@xmath127 , while the bottom line shows the top view for the same times . ]
the key idea is that the head can be emulated with the part of this ellipsoid that is above the substrate level .
this cap volume , @xmath26 , varies as the dewetting motion of the tip proceeds .
the dynamics can be described as the superposition of both a vertical and a horizontal motion of the ellipsoid , with velocities @xmath128 and @xmath129 , respectively .
since the ellipsoid is not only emerging but also moving horizontally , we describe it in terms of the coordinates of its center , @xmath130 , and thus , it is described by the equation @xmath131 then , the volume of the ellipsoidal cap is given by @xmath132 due to the horizontal retraction of the tip , there is a mass accretion in the cap from the filament that is being swept by the receding head .
therefore , the cap volume @xmath133 increase can be calculated in first approximation by @xmath134 where @xmath135 , the area of the cross section of the filament of width @xmath96 and transverse contact angle @xmath45 , is given by @xmath136 here , we define @xmath137 as the distance between the tip and the center position , @xmath138 , i.e. @xmath139 .
note that the vertical motion of the ellipsoid yields a variation of @xmath137 .
therefore , the contact line velocity along @xmath118axis is the result of the superposition of the bulk horizontal displacement and the variation of @xmath137 , @xmath140 where @xmath141 and @xmath142 is the front ( tip ) velocity as a function of the dynamic contact angle at the tip , @xmath104 , whose functional form is given implicitly by eq .
( [ eq : theta_fit ] ) .
the vertical motion of the ellipsoid and the horizontal displacement are not independent since @xmath143 so that there is an increase of the volume in the head as well as a variation of the contact angle , @xmath104 . upon writing @xmath144 , eqs .
( [ eq : dvdt ] ) and ( [ eq : vel ] ) lead to @xmath145 whose integration allows a comparison with the time evolution of most of the parameters measured in the experiments . in order to find the values of the constants @xmath121 and @xmath123 of the ellipsoid , we consider eq .
( [ eq : elipsoid ] ) at the tip position ( @xmath146 ) , @xmath147 and its slope ( contact angle ) @xmath148 where @xmath149 for given values of @xmath137 , @xmath104 , and @xmath107 ( @xmath150 ) , eqs .
( [ eq : eqtip])([eq : x0 ] ) yield , @xmath151 where @xmath152 must satisfy the condition @xmath153 . here
, we consider @xmath137 , @xmath104 , and @xmath107 at @xmath20 . on the other hand , to calculate @xmath122 we consider eq .
( [ eq : elipsoid ] ) at the point of the contact line where the width of the head is maximum ( @xmath154 ) , @xmath155 and its slope ( contact angle ) @xmath156 where @xmath157 by using eqs .
( [ eq : eqtip])([eq : x0 ] ) and ( [ eq : eqwmax])([eq : y0 ] ) , we define the following ratios : @xmath158 therefore , in order to determine @xmath122 in terms of @xmath121 , any of these equalities could be used . since @xmath104 and @xmath159 can not be measured simultaneously , we discard the ratio of their tangents . moreover ,
even if one could attempt to produce an @xmath160plane view , any measurement of @xmath159 would be blocked by the drops of the previous ruptures .
instead , we consider the ratio @xmath161 , which can be measured from top view observations . from experiments with three different widths ( @xmath162 , @xmath163 and @xmath164 cm ) and four ruptures for each width , we obtain an average value of the ratio @xmath165 , as @xmath166 .
finally , we obtain the function @xmath167 inverting @xmath108 evaluated at the tip , see eq .
( [ eq : theta_fit ] ) , and solve eq .
( [ eq : dx0dt ] ) numerically in time .
thus , we have @xmath168 where @xmath169 is given by eq .
( [ eq : thex ] ) , and @xmath170 .
once @xmath137 versus @xmath47 is found , we can calculate the contact angle @xmath171 , the maximum thickness @xmath172 , and the tip position @xmath173 where the integral @xmath174 is @xmath175_{x_{0,i}}^{x_0(t)}. \label{eq : it}\ ] ] we show in fig . [
fig : xf - thex - h0_17 ] a comparison of the time evolution as predicted by the model with the experimental data . as a first attempt
, we determine the constants @xmath121 , @xmath122 , and @xmath123 by taking the values of @xmath137 , @xmath107 , and @xmath104 corresponding to the first frame capture after the breakup ( @xmath20 ) .
thus , we obtain the curves shown as dashed magenta lines in fig . [
fig : xf - thex - h0_17 ] ( @xmath176 ) . in spite of the simplicity of the model , based on a rough approximation of the free surface shape ( assumed as an ellipsoidal cap ) ,
the agreement with experiments is remarkably good , except for the maximum thickness of the head , @xmath172 .
note , however , that immediately after the breakup the shape of the head is strongly changing from a spike to a rounded contour ( see fig . [
fig : conos ] ) .
thus , during a short time of the order of a few seconds , the head does not achieve an ellipsoidal shape , and then , a too early selection of the initial values of @xmath137 , @xmath107 , and @xmath104 can lead to an inaccurate determination of @xmath121 , @xmath122 , and @xmath123 .
moreover , since these initial profiles are quite flat at the top , with @xmath107 being close to the filament thickness , it is difficult to determine @xmath137 with sufficient accuracy .
better agreement with experiments could be expected if the initial values are taken from a later head whose shape has settled closer to that of an ellipsoidal cap .
therefore , we use instead the second frame ( @xmath177 s later ) as starting time , and obtain new results from the model ( see solid magenta lines in fig .
[ fig : xf - thex - h0_17 ] , @xmath178 ) .
we observe that the agreement with experiments is greatly improved for @xmath179 and @xmath180 , and even the @xmath172 curve better approximates the experimental data .
another interesting feature of the flow that can be extracted from the model is the maximum displacement of the receding tip , @xmath181 .
in fact , according eq .
( [ eq : theta_fit ] ) , the tip stops when @xmath104 reaches the value @xmath70 .
thus , the maximum value of @xmath137 is given by eq .
( [ eq : thex ] ) in the form @xmath182 which yields a maximum value for the tip position , @xmath181 , as given by eqs .
( [ eq : xft ] ) and ( [ eq : it ] ) for @xmath183 .
the values of @xmath181 for the first four heads in fig .
[ fig : xf - thex - h0_17 ] are shown in table [ tab : xfmax ] .
we note that the ratio @xmath184 is practically constant in experiments .
the corresponding theoretical @xmath185 as obtained from our simple model is @xmath186 , which is in good agreement with the average experimental value .
this result is in accordance with one of the features of the model , namely that all the distance parameters in eq .
( [ eq : xft ] ) , such as @xmath137 , @xmath107 , @xmath121 , @xmath122 and @xmath123 , are proportional to @xmath96 .
.experimental values of the ratio @xmath184 for the four heads in fig .
[ fig : xf - thex - h0_17 ] with @xmath187 cm .
the experimental average value @xmath188 , while the model yields @xmath189 [ cols="^,^,>",options="header " , ] the use of the model for top view experiments is a bit more involved due to the lack of information about the contact angle at the tip , which is essential to the dynamics of the system .
however , taking into account the proportionality of the coefficients with @xmath96 ( see table [ tab : xfmax ] ) , it is possible to infer the values of the semi axes of the ellipsoid by using the information from a single side view experiment , such as the case above .
thus , by multiplying the previous values of @xmath121 , @xmath122 , and @xmath123 by the ratio @xmath190 , we find the model results shown in fig .
[ fig : xf - w_20 ] .
even with this rather indirect method to determine the parameters , the model is able to predict reasonably well the evolution of the tip position , @xmath179 .
instead , it fails to properly describe @xmath191 , and yields higher values than expected .
this result is consistent with the fact that @xmath172 is underestimated in the side view case ( see fig [ fig : xf - thex - h0_17](b ) ) , since it implies an increase of @xmath192 to balance the mass flow swept by the receding tip .
note that the top view experiments do not provide any out of plane information , and therefore the input to the model dynamics is really scarce , so that the fact that it can pretty well describe at least @xmath179 is remarkable .
these results suggest that the shape of the head can be roughly considered an ellipsoidal cap during most the receding stage .
however , our model does not include any information about the transversal wetting , so it does not have the necessary ingredients to accurately predict the evolution of its width , @xmath191 , and maximum thickness , @xmath172 .
on the other hand , the experimental evidence shows that by considering @xmath137 , @xmath107 , and @xmath104 at successive times , the corresponding values of @xmath121 , @xmath122 , and @xmath123 vary with time .
in fact , while @xmath121 slightly oscillates around a typical value ( so that one could assume an average constant ) , @xmath122 and @xmath123 steadily increase with time .
thus , the differences in @xmath172 observed between the model of constant semi - axes and the experiments can be explained in terms of the increasing length of transverse semi - axes of the actual approximating ellipsoid .
we numerically simulate the evolution of one end of the filament , and assume that the other one is so far away that it does not affect the other .
thus , we consider that at @xmath20 the filament starts at @xmath193 with the shape of a cylindrical cap of length @xmath194 , width @xmath96 ( @xmath195 ) , and with transverse equilibrium contact angle , @xmath45 , along both parallel contact lines .
the fluid region around the origin ( @xmath196 , @xmath197 ) remains practically at rest , while the main flow develops far away from there , close to the end region ( @xmath198 ) .
therefore , the boundary condition at @xmath193 is that for a no - slip wall . in order to emulate the actual filament , which ends at a rounded shape due to the breakup process that took place before ( @xmath199 ) , we here assume that this shaped end can be approximated by an additional cap of length @xmath137 at the end of the cylinder ( @xmath200 ) .
thus , the fluid domain is composed by a cylindrical cap of length @xmath194 plus an ellipsoidal cap of length @xmath137 .
the length @xmath137 is chosen in such a way that the filament tip has a given contact angle @xmath201 , resulting from the breakup stage ( see fig .
[ fig:3d](a ) ) .
the time evolution of this liquid filament of total length @xmath202 is obtained by numerically solving the dimensionless navier - stokes equation @xmath203 = - \vec { \nabla } p + \nabla ^{2 } \vec { v } -\vec{z } , \label{eq : ns}\ ] ] where the last term stands for the gravity force . here , the scales for the position @xmath204 , time @xmath47 , velocity @xmath205 , and pressure @xmath206 are the capillary length @xmath207 , @xmath208,@xmath209 , and @xmath210 , respectively .
therefore , the laplace number is @xmath211 . in our experiments we have @xmath212 mm and @xmath213 , so that inertial effects are practically irrelevant .
the @xmath118 and @xmath119axes are assigned along and across the original filament , respectively .
besides , the normal stress at the free surface accounts for the laplace pressure in the form @xmath214 where @xmath215 and @xmath216 are the versors standing for the normal and tangential directions to the free surface . since , the surrounding fluid ( e.g. , air ) is passive , we assume that the tangential stress is zero , i.e. @xmath217 . as regards to the boundary condition at the contact line , the dynamic contact angle , @xmath14 , is given by the dimensionless contact line velocity , @xmath218 , according to the cv+b model in the form ( see eq .
( [ eq : theta_hyd ] ) ) @xmath219 + 9 ca \ln \left ( \frac{1}{\hat \ell } \right ) , \label{eq : theta_fit_num}\ ] ] where @xmath220 , and @xmath221 .
the contact line velocity is calculated from the velocity field as @xmath222 , where @xmath223 is the versor normal to the contact line .
note that this condition introduces a high nonlinearity to the problem , since the solution itself , namely the velocity field at @xmath224 , yields the corresponding contact angle .
we use a finite element technique in a domain which deforms with the moving fluid interface by using the arbitrary lagrangian - eulerian ( ale ) formulation @xcite . the interface displacement is smoothly propagated throughout the domain mesh using the winslow smoothing algorithm @xcite .
the main advantage of this technique compared to others such as the level set or phase field techniques is that the fluid interface is and remains sharp @xcite .
the main drawback , on the other hand , is that the mesh connectivity must remain the same , which precludes the modeling of situations for which the topology might change ( rupture of the filament ) .
the default mesh used throughout is unstructured , and typically has @xmath225 triangular elements ( linear elements for both velocity and pressure ) .
the mesh nodes are constrained to the plane of the boundary they belong to for all but the free surface . in order to compare the numerical results with the experiments
, we perform simulations with the wettability parameters as given by eq .
( [ eq : param ] ) .
although the calculation can not be continued beyond the breakup moment , the simulation is useful enough to account for the dewetting and wetting processes between breakups .
thus , the initial condition used here tries to emulate the scenario just after an actual breakup , and the resulting fluid motion is simulated till just before the next breakup .
[ fig:3d ] shows the time evolution of the filament of width @xmath106 . here
, we use @xmath226 to define the cylinder length ( we have checked that for @xmath227 , the results do not depend on @xmath65 at all ) , and @xmath228 at the tip , which determines the ellipsoidal shape that emulates the filament end ( or nose ) just after the rupture . the solid lines in fig .
[ fig : xf - h0-thex_17_num ] show the time dependence of the tip position , @xmath103 , tip contact angle , @xmath104 , and maximum head thickness , @xmath107 , as given by the numerical simulations using the values of the constants @xmath70 and @xmath51 from eqs .
( [ eq : theta0 ] ) and ( [ eq : vmax ] ) , respectively . regarding fig .
[ fig : xf - h0-thex_17_num](a ) , the simulations yield smaller values of both the slope of @xmath179 for early times and the maximum tip position for late times respect to the experimental data . note also , that the numerical values of @xmath104 and @xmath107 lie below the experimental points ( see figs .
[ fig : xf - h0-thex_17_num](b ) and ( c ) ) . clearly , the degree of agreement between simulations and experiments is highly dependent on the parameters in eq .
( [ eq : theta_fit_num ] ) .
the weakest part of this fitting process is the inclusion of a single point to account for an important part of the dewetting region of this constitutive relation , namely @xmath51 , which is obtained by performing a quite different experiment than that used to generate the rest of the data ( see fig .
[ fig : theta_vcl ] ) .
since there is an uncertainty in the value of @xmath51 due to the experimental error ( of the order of @xmath57 ) , we explore the possibility that its inaccuracy can be a source of these differences .
moreover , although there is consensus on the hypothesis that the contact angle is zero at @xmath51 @xcite , one can not guarantee that this is actually the case .
if a different contact angle corresponds to this velocity , then the resulting @xmath108 curve allows both higher angles and dewetting velocities .
the simulations show that considering a @xmath57 increase in @xmath51 strongly improves the comparison of the slope of @xmath179 for early times , while the saturation value is not practically not affected . on the other hand
, there is an additional uncertainty on how the value of @xmath70 is determined , since the only requirement is that it falls inside the hysteresis range , namely @xmath229 . in this case , the numerical results show that the saturation value of @xmath179 favorably compares with the experiments if @xmath70 is increased a few degrees with respect to the value given in eq .
( [ eq : theta0 ] ) .
thus , in order to better approximate the experimental data , we perform a simulation by slightly modifying both parameters to the values @xmath230 and @xmath231 , and using consistent modified values of the coefficients in eq .
( [ eq : param ] ) to fit the experimental data in fig .
[ fig : theta_vcl ] : @xmath232 , @xmath233 cm , @xmath234 cm / s .
the results of the simulation are shown by the dashed lines in fig .
[ fig : xf - h0-thex_17_num ] .
we observe a clear improvement of the comparison for all three quantities shown in that figure , in particular for the evolution of the contact angle at the tip , @xmath104 . a similar tendency to approach the experimental data is evident in the thickness profiles when using the alternative values @xmath235 and @xmath236 ( see fig .
[ fig : hx_comp ] ) . in order to focus our attention on the shape of the profiles , both the experimental and numerical ones have been shifted so that the position of the tip is always at @xmath193 . although , there are some departures both near the maximum and the neck , the latter are the most significant ones . + since the differences between experiments and numerics are mainly quantitative and reasonable small , the latter are still very useful to learn about the hydrodynamics of the dewetting process at the filament end .
in fact , the numerical results in fig . [ fig : hvp_num](a ) show the behaviour of the velocity and pressure fields in the head and neck regions for times close to breakup at the intersection of the free surface with the plane @xmath146 . at the tip , the pressure is relatively high and induces the dewetting motion ( @xmath237 ) . inside the head ,
the pressure is practically uniform and increases towards the neck , where it reaches a maximum value .
this peak drives a flow out of the neck region towards the head ( @xmath237 ) and the filament ( @xmath238 ) .
[ fig : hvp_num](b ) shows the flow field in a horizontal plane close to the substrate .
this vectorial field velocity complements the picture already described and indicates that the change of behaviour from receding to advancing occurs at different values of @xmath118 whether one observes the problem at the top of the free surface or near the contact line . in order to describe the evolution of the shape of the head , we resort now to an experiment observed from the top , see fig .
[ fig : side_top](b ) , which allows us to register the evolution of the whole contact line . in fig .
[ fig : xf - w_20_num ] , we compare the tip position , @xmath179 , the head width , @xmath113 , and the neck width , @xmath114 , with the experimental data .
the comparison of @xmath179 yield similar conclusions with respect to the effects of using the modified values @xmath239 to those of the side view case in fig .
[ fig : xf - h0-thex_17_num](a ) .
however , unlike the previous case we observe here that the slope of @xmath179 for early times is a bit smaller than the experimental results . also , fig .
[ fig : xf - w_20_num](b ) shows a greater departure of the widths of both head and neck from the experiments , as well as very slight improvements when using the primed parameters .
we believe that all these differences can be attributed to the fact that the ellipsoidal shape assumed at @xmath20 for the filament end after breakup might not be the best choice to describe the longitudinal thickness profile .
unfortunately we are not able to observe simultaneously both the thickness profile and the shape of the footprint to completely elucidate this issue . in fig .
[ fig : foot_comp ] , we compare the contours of the footprints at different times by using @xmath240 ( solid lines ) and @xmath239 ( dashed lines ) , as we have done for the thickness profiles in fig .
[ fig : hx_comp ] . here , we also focus our attention on the shape of these footprints , so that both the experimental and numerical contours have been shifted so that the position of the tip is always at @xmath193 . unlike the comparison made for the thickness profiles , the modified parameters yield here a slight improvement only in the prediction of the neck shape , while the shape of the rest of the contact line is barely affected .
+ interestingly , the simulations also show the existence of an angular sector that embraces the evolution of the head contact line , similarly to what is observed in the experiments ( see fig .
[ fig : the_vcl - fil](b ) ) .
this feature is shown in fig .
[ fig : env_sim](a ) .
in fact , the red thick lines in that figure correspond to the envelope of the parametric family of curves @xmath241 representing the contact line in the simulations . here
, the envelope is calculated by numerically solving the condition : @xmath242 note this curve can be approximated by a straight line , as the one observed in the experiments and whose slope turned out to be independent of @xmath96 .
additional simulations show that this is actually the case , though the constant @xmath243 is @xmath244 here . in accordance with fig .
[ fig : foot_comp ] , this angle is not modified by using the set of parameters @xmath239 .
the difference with the experimental value of @xmath243 is consistent with the departures of @xmath103 and @xmath113 mentioned above . on the other hand ,
we calculate the curve joining the contact line points where the local contact angle is equal to @xmath70 , i.e. where the contact line velocity is zero .
the interesting fact is that this other curve turns out to be identical to the above envelope .
thus , we conclude that the existence of the envelope is related to the points of the contact line which do not have normal velocity , but only tangential one .
this effect is illustrated by the black points in fig .
[ fig : env_sim](b ) , where the contact line velocity field is shown by vectors .
therefore , these points divide the portion of the contact line that is receding ( dewetting ) from that portion which is advancing ( wetting ) in the normal direction to the contact line .
this result allows to understand the nature of the envelope that embraces the evolving footprint of the head and provides the physical underlying reason of its existence .
in this paper we are concerned with the dynamics of the axial dewetting of a liquid filament laying on a horizontal substrate under partial wetting conditions .
we observe that the retraction motion as well as the consequent transverse spreading strongly depend on the features of the contact angle hysteresis .
for this reason , we study in detail the hysteresis cycles both under static and dynamic conditions . in the first case , we are able to determine the hysteresis range as well as the critical angles at which the drop contact line position must change in order to obtain a new static shape as its volume is varied ( see section [ sec : st_wet ] ) . for the dynamic case
, we obtain the constitutive relationship between the contact line velocity and the dynamic contact angle , and include this relation within a hybrid model that takes into account both a hydrodynamic approach and a molecular kinetic description ( see section [ sec : dyn_wet ] ) .
these results play a fundamental role in understanding the dynamics of the retracting filament , which is the main goal of this work .
the measurements of different characteristics of the evolving filament , such as tip position , contact angle at the tip , head thickness , head width , and neck width as well as contact line shape and vertical thickness profile , allow to perform a complete description of the whole evolution , so that the proposed wettability model can be thoroughly tested . in order to understand the main features of this flow
, we develop a heuristic model to account for the time evolution of these parameters .
the main outcome of this model is a remarkably good prediction of the retraction motion of the tip and the contact angle there .
however , it is unable to properly describe the time evolution of the width of the head , since it does not consider the properties of the contact angle hysteresis in the transverse direction .
therefore , we also perform a comparison of the experimental data with the results of the numerical solution of the full navier stokes equations , which include the contact line velocity versus the contact angle as given by the hybrid constitutive relationship resulting from eqs .
( [ eq : theta_hyd ] ) and ( [ eq : theta_bl ] ) .
the simulations are able to properly describe the flow when using this hybrid law with the coefficients as measured in the characterization of the hysteresis cycle .
it should be pointed out that the use of this function @xmath108 is an essential ingredient to adequately account for the experimental data .
naturally , one could be tempted to use a more usual and simpler relation as given by eq .
( [ eq : theta_hyd ] ) with @xmath245 , known as cox voinov model .
however , the simulations show that using this law instead of the hybrid one leads to results that do not compare well with the experiments ( see gray lines in figs .
[ fig : xf - h0-thex_17_num ] and [ fig : xf - w_20_num ] ) .
this fact strongly emphasizes the need to properly model the hysteretic effects when describing the dynamics of a receding filament , which simultaneously involves wetting and dewetting motions of the contact line . as discussed in section [ sec : num ] , the main differences with the simulations observed in some experimentally measured quantities for the retracting filament end can be attributed to an incomplete knowledge of its shape just after the breakup .
one of the most interesting facts revealed by the simulations is the physical property of the contact line points which belong to the envelope that encloses the footprints at different times .
it is shown that these points have no normal velocity of the contact line , and correspond to the characteristic contact angle of the hysteresis curve , @xmath70 .
this envelope is also observed in the experiments as two straight lines , in agreement with the simulations .
our analysis has focused on the retraction mechanism of the filament prior to its breakup .
however , the final stages of this process can be affected by the rupture dynamics , whose understanding certainly merits future work .
the authors acknowledge support from consejo nacional de investigaciones cientficas y tcnicas ( conicet , argentina ) with grant pip 844/2012 and agencia nacional de promocin cientfica y tecnolgica ( anpcyt , argentina ) with grant pict 931/2012 .
we also thank dr .
mathieu sellier ( canterbury univ . ,
nz ) for his contributions in the numerical simulations .
j. l. wu , f. c. chen , y. s. hsiao , f. c. chien , c. h. kuo p. l. chen , m. h. huang , and c. s. hsu .
surface plasmonic effects of metallic nanoparticles on the performance of polymer bulk heterojunction solar cells .
, 5:959967 , 2011 .
f. le , d. w. brandl , y. a. urzhumov , h. wang , j. kundu ; n. j. halas , j. aizpurua , and p. nordlander .
metallic nanoparticle arrays : a common substrate for both surface - enhanced raman scattering and surface - enhanced infrared absorption .
, 2:707718 , 2008 .
s. a. wolf , d. d. awschalom , r. a. buhrman , j. m. daughton , s. von molnar , m. l. roukes , a. y. chtchelkanova , and d. m. treger .
spintronics : a spin - based electronics vision for the future .
, 294:14881495 , 2001 .
y. wu , j. d. fowlkes , n. a. roberts , j. a. diez , l. kondic , a. g. gonzlez , and p. d. rack .
competing liquid phase instabilities during pulsed laser induced self - assembly of copper rings into ordered nanoparticle arrays on sio2 .
, 27:13314 , 2011 .
j. d. fowlkes , l. kondic , j. a. diez , a. g. gonzlez , y. wu , n. a. roberts , c. e. mccold , and p. d. rack .
parallel assembly of particles and wires on substrates by dictating instability evolution in liquid metal films .
, 4:7376 , 2012 .
j. d. fowlkes , n. a. roberts , y. wu , j. a. diez , a. g. gonzlez , c. hartnett , k. mahady , s. afkhami , l. kondic , and p. d. rack .
hierarchical nanoparticle ensembles synthesized by liquid phase directed self - assembly .
, 14:774782 , 2014 .
n. a. roberts , j. d. fowlkes , k. mahady , s.afkhami , l. kondic , and p. d. rack . directed assembly of one- and two - dimensional nanoparticle arrays from pulsed laser induced dewetting of square waveforms . , 5:44504456 , 2013 . | we study the hydrodynamic mechanisms involved in the motion of the contact line formed at the end region of a liquid filament laying on a planar and horizontal substrate .
since the flow develops under partially wetting conditions , the tip of the filament recedes and forms a bulged region ( head ) , that subsequently develops a neck region behind it .
later on , the neck breaks up leading to a separated drop , while the rest of the filament restarts the sequence .
one main feature of this flow is that the whole dynamics and final drop shapes are strongly influenced by the hysteresis of the contact angle typical in most of the liquid / substrate systems .
the time evolution till breakup is studied experimentally and pictured in terms of a hybrid wettability theory which involves the cox
voinov hydrodynamic approach combined with the molecular kinetic theory developed by blake .
the parameters of this theory are determined for our liquid / substrate system ( silicon oil / coated glass ) .
the experimental results of the retracting filament are described in terms of a simple heuristic model , and also compared with numerical simulations of the full navier stokes equations .
this study is of special interest in the context of pulsed laser induce dewetting ( plid ) . |
the introduction of an entirely new concept to explain a class of observed phenomena in any field of science would seem to be a redundant activity , if the phenomena can reasonably be accounted for with use of existing theories and models .
however , in the cases of certain astronomical observations of h(@xmath4 ) emission , standard explanations appear to be inadequate .
this letter focusses specifically on recorded observations of intense h(@xmath0 ) emission in reddened early - type stars .
it will be shown below that enough spectral information exists for such stars to suggest strongly that a new paradigm is here needed .
one that appears to be quite suitable for explaining the h(@xmath0 ) emission data was recently developed in the field of quantum electronics and is termed ` laser without inversion ( lwi ) ' .
the lwi concept is itself inseparably tied to another recently developed quantum electronics concept , that of ` electromagnetically induced transparency ( eit ) ' . in the present letter a novel astrophysical model based upon lwi / eit
is introduced that appears to be fully capable of explaining the most intense and spectrally narrowest h(@xmath0 ) emissions seen in reddened early - type stars .
in rawlings et al . ( @xcite ) new optical spectroscopy for 45% of an older catalog ( stephenson @xcite ) of @xmath5440 reddened stars was performed in an attempt to isolate reddened stars which are also early - type .
the spectra between 6100 and 6900 of all the stars studied are shown in rawlings et al .
( @xcite ) .
only fifteen of the stars surveyed could be classified as early - type stars . with regard to h(@xmath0 ) , the spectra of these early - type stars show tremendous variations .
some ( fig .
[ fig1]a ) show h(@xmath4 ) only in absorption .
others ( figs .
[ fig1]b , [ fig1]c , [ fig1]d ) show h(@xmath0 ) strongly in emission , with an intensity completely disproportionate to all other absorption features present .
yet the spectral types of these stars are all apparently roughly comparable . in the spectra of all the cooler stars surveyed by rawlings et al .
( @xcite ) , h(@xmath0 ) , if present at all , appears only in absorption .
much spectral data exists for the star , hd 44179 , that powers the fascinating red rectangle nebula .
we here consider an apertured line - of - sight that includes just the star , not the nebula .
in such a line - of - sight , one sees ( cohen et al .
@xcite ) that all balmer lines , with the notable exception of h(@xmath0 ) , appear as broad absorptions against the star s blackbody continuum .
no emissions are present in the blue - green spectral region
. however , h(@xmath0 ) appears as a strong , narrow , emission line centrally placed in a broad absorption .
a recent stis scan showing h(@xmath4 ) emission in hd 44179 is shown in fig .
[ fig2 ] . in view of the enormous variations in h(@xmath0 ) spectral width and intensity observed in lines - of - sight to reddened stars of roughly the same spectral type , a non - stellar origin for this emission is strongly suggested .
hence - at least in the most singular cases - this emission must originate from h atoms located outside the photosphere of the illuminating star .
if the h(@xmath0 ) emissions resulted from electron impact excitation occurring in compact h ii regions about the stars , one would expect also to detect emissions at the wavelengths of h(@xmath3 ) , h(@xmath6 ) , h(@xmath7 ) , etc .. at least in the case of hd 44179 , these are not observed .
h(@xmath0 ) emission could reasonably be produced via linear photoexcitation of h atoms located in neutral regions bordering h ii regions .
however , as discussed below , this would generally result in broadened h(@xmath0 ) emission - line profiles , and can not account for the extreme @xmath7-function - like profiles observed in some lines - of - sight .
in essence , then , the nonlinear photoexcitation model introduced here will attempt to explain why some reddened early - type stars show h(@xmath0 ) emissions that are comparatively broad , particularly at the base , while others ( e.g. fig .
[ fig1]b ) display ones that are much sharper , more intense , and more singular - looking ( i.e. unaccompanied by a significant pedestal ) .
to understand the concept of lwi , one must first understand that of eit , since the former is built upon the latter .
a good introductory review describing some of the recent exciting developments in eit is harris ( @xcite ) .
the particular lwi / eit model being considered in the present letter involves the so - called three - level @xmath1 scheme shown in fig .
transitions 1@xmath83 and 2@xmath93 are allowed ; the total radiative decay rate of quantum level @xmath10 is @xmath11 .
transition 1@xmath92 is assumed to be basically forbidden . for h atoms
, @xmath12 decays radiatively to @xmath13 via spontaneous two - photon emission at a rate @xmath14 8 sec@xmath15 .
this degree of metastability qualifies the h - atom 2s level to be level @xmath12 in a @xmath1-type eit scheme .
when a nearly monochromatic laser beam ( the eit ` coupling beam ' ) , tuned to the center frequency @xmath16 of the 2@xmath93 transition , is applied to a gas of three - level atoms , a sharp dip in absorption in the vicinity of the 1@xmath93 transition occurs ( fig . [ fig4 ] ) .
this cancellation in the imaginary part of the linear susceptibility ( absorption is proportional to i m @xmath17 ) can be shown to arise from quantum interference .
for the ideal case @xmath18=0 , there is perfect transparency at the minimum .
the width of the transparency hole varies as @xmath19 , the so - called rabi frequency , which is itself proportional to the square root of the laser power ( @xmath20= @xmath21 , where @xmath22 is the dipole matrix element and @xmath23 is the optical field ) .
if another nearly monochromatic laser beam ( the eit ` probe beam ' ) , tuned to the center frequency @xmath24 of the 1@xmath93 transition , is applied to the same gas of atoms , the system will quickly adjust itself so that there is also no absorption about the line center of the 2@xmath93 transition .
this occurs independently of the ratio of rabi frequencies @xmath25/@xmath19 .
the two laser beams rapidly drive all atoms of the system into a stable ` coherently trapped population state ' . in this state ,
the wavefunction of all the atoms in the gas becomes an antisymmetric linear combination of the unperturbed wavefunctions of levels @xmath26 and @xmath12 . in this state
, no atoms are present in level @xmath10 .
one of the earliest experiments in which coherent trapping of atomic populations was realized is described in gray et al .
( @xcite ) .
atoms coherently trapped by this method are also commonly referred to as ` coherently phased ' atoms .
the coherently trapped state is sometimes denoted by nc , standing for ` non - coupled ' , because i m @xmath17 @xmath27 at the line centers of the two transitions 1@xmath93 and 2@xmath93 . once all the atoms of a gas have been driven to the nc state , light at the two applied laser frequencies can continue to pass over the atoms without being attenuated .
there is a state orthogonal to the nc state in which the wavefunction of the atoms is a symmetric linear combination of the unperturbed wavefunctions of levels @xmath26 and @xmath28 . because i m @xmath29 @xmath30 at the two allowed transitions for atoms in this state , it is sometimes referred to as the c ( ` coupled ' ) state .
consider a @xmath1-type system with all atoms maintained in the nc state via the eit process described above .
if , through some separate means , some of these atoms can be excited to level @xmath10 , then these atoms can fluoresce at both 1@xmath93 and 2@xmath83 line centers via allowed transitions to the two c states shown in fig .
since the eit mechanism rapidly and continually transfers all atoms from the c state to the nc state , true population inversions can exist between level @xmath10 and both c states .
gain can therefore be present on these transitions , and stimulated emission ( i.e. laser action ) can result . this scenario is termed a ` laser without inversion ' , because there are many more atoms in each of the two nc levels than in level @xmath10 .
an lwi demonstrated by the group of m.o .
scully ( padmabandu et al .
@xcite ) works by combining optical pumping and eit . in considering mechanisms that could pump an h - atom lwi in space ,
one first should analyze simple linear photoexcitation of nc atoms via absorption of far - ultraviolet ( fuv ) continuum light from the star in the spectral vicinity of ly-@xmath3 .
typical h - atom densities in cold , neutral clouds surrounding bright stars are 10@xmath31 - 10@xmath32 @xmath33 , implying collision rates @xmath3410@xmath35 - 10@xmath36 sec@xmath15 .
such media should therefore be regarded as being virtually collisionless . in a collisionless regime ,
fuv light in the vicinity of ly-@xmath3 - but offset enough from it to avoid the @xmath37 minimum ( fig .
[ fig4 ] ) - will either be elastically scattered by nc atoms or will undergo spontaneous resonant raman scattering .
in the latter process , an atom becomes excited to the nc @xmath12 level , and a photon slightly offset from h(@xmath0 ) is simultaneously emitted .
however , no excitation of level @xmath10 occurs ( loudon @xcite ) .
linear photoexcitation would therefore be an ineffective pumping mechanism for lwi in space , but it could perhaps account for the presence of spectrally broad h(@xmath0 ) emissions seen from some reddened early - type stars .
harris et al .
( @xcite ) were the first to show that , while eit makes a material system transparent to resonant laser radiation , it nonetheless allows a large nonlinear resonant process to occur with high probability .
we propose that one such process , resonant hyper - raman scattering ( hrs ) , might be a viable mechanism for pumping an h - atom lwi in space . in the unit hrs pumping process ( fig . [ fig5 ] ) , a pair of fuv photons , with frequencies lying very close to ly-@xmath3 but offset from it by equal amounts to high and low energies , are absorbed from the star s blackbody continuum , a photon at ly-@xmath3 line center is emitted , and an nc atom is excited to level @xmath38 - with all events in this energy conserving process occurring simultaneously .
this hrs mechanism would be an efficient h - atom lwi pump , since , in the unit process , both a ly-@xmath3 photon at line center is generated , and an nc atom is excited to the desired 3p level .
an important feature of lwi excitation occurring via this mechanism is that the rate of pumping is itself proportional to the intensity of the radiation present at ly-@xmath3 line center .
one would expect an h - atom lwi associated with a reddened early - type star to occur in a thin shell of neutral gas located just outside an ionized hydrogen ( h ii ) region . for a b0 iii star , the radius @xmath39 of such a region would typically be @xmath50.01 pc . at 1000 , the fuv continuum flux at @xmath39 propagating outward from the star would be @xmath53 x 10@xmath40 photons per @xmath41 per sec per wavenumber .
this flux could be additionally enhanced several hundred times around the wavelength of ly-@xmath2 via photoprocesses occurring in the h ii region .
it is also possible that nonlinear elastic scattering by h atoms of the two lwi narrow - band emissions generated would lead to additional intensity enhancement of these emissions in the active region via the effect of decreased spatial diffusion .
this , in turn , would increase the rate of pumping . from fig .
[ fig4 ] , one predicts a telling signature of lwi / eit to be extremely narrow - band emission observed at both the h(@xmath0 ) and ly-@xmath3 line centers . the hrs pumping process should be manifest by symmetrical absorption of continuum light around ly-@xmath3 in a pattern that might superficially resemble the one shown in fig .
( in fig .
[ fig2 ] , however , the apparent absorption dip has a stellar origin , as was noted above . )
an lwi / eit mechanism can reasonably explain certain other intense , extremely narrow emission lines recorded by astronomers , examples being the sodium d - line doublet appearing in hd 44179 , or the powerful o vi ( 1032 , 1037 ) emission doublet that dominates the fuv spectra of symbiotic stars such as rr tel . in both these cases , lwi / eit would have to occur via the so - called ` v ' scheme , which incidentally is the easiest one to excite in the presence of significant thermal doppler broadening ( boon et al .
@xcite ) . however , the environments existing in cold , collisionless neutral clouds surrounding h ii regions are generally favorable for all lwi / eit schemes .
scully has coined the word @xmath42 to describe an ensemble of coherently phased atoms .
an lwi based upon such a medium , he suggests , might appropriately be termed a @xmath43 .
should the occurrence of lwi / eit in space ever become substantiated , widespread usage of these terms in astronomy might follow .
boon j. r. , zekou e. , mccloin d. et al .
, 1999 , phys .
a59 , 4675 cohen m. , anderson c. m. , cowley a. et al . , 1975 , apj 196 ,
179 gray h. r. , whitley r. m. , stroud c. r. 1978 , opt . lett . 3 , 218 harris s. e. , field j. e. , imamoglu a. 1990 , phys . rev .
64 , 1107 harris s. e. 1997 , physics today 50 , 36 loudon r. 2000 , the quantum theory of light , 3rd .
, oxford univ .
press , oxford padmabandu g. g. , welch g. r. , shubin i. n. et al .
, 1996 , phys .
76 , 2053 rawlings m. g. , adamson a. j. , whittet d. c. b. 2000 , apjs 131 , 531 stephenson c. b. 1992 , aj 103 , 263 | a model is suggested to explain the frequently observed presence of intense , narrow - band , h(@xmath0 ) emission lines in the optical spectra of reddened , early - type stars ( e.g. hd 44179 , iras 18179 - 1346 , iras 20298 + 4011 ) .
it is proposed that hydrogen atoms surrounding compact h ii regions enveloping such stars become coherently phased via a nonlinear photonic mechanism that leads to ` electromagnetically induced transparency ( eit ) ' .
eit is a powerful technique that can be used to make a material system transparent to resonant laser radiation , while still allowing large nonlinear resonant processes to occur with high probability . in eit terminology ,
a ` @xmath1 ' configuration , involving h - atom levels ( 1s , 3p , and 2s ) , is here assumed to be operative .
the eit ` coupling beam ' is the narrow - band h(@xmath0 ) radiation predicted to be coherently generated via a standard ` laser without inversion ( lwi ) ' scenario when coherently phased atoms are excited to the 3p level by means of a separate nonlinear excitation process known as resonant hyper - raman scattering ( hrs ) . in the unit hrs pumping process , a pair of far - ultraviolet ( fuv ) photons , with frequencies lying very close to ly-@xmath2 but offset from it by equal amounts to high and low energies ,
are absorbed from the star s blackbody continuum , a photon at ly-@xmath3 line center is emitted , and an atom is excited to the 3p level - with all events in this energy conserving process occurring simultaneously .
the eit ` probe beam ' is the light predicted to be coherently generated at ly-@xmath3 line center , which - as a result of the complete linear transparency afforded at this frequency by the coherently phased h atoms - can propagate completely unattenuated through the optically thick h - atom cloud surrounding the star . |
superconducting qubits @xcite consist of electronic nanocircuits embedding josephson junctions whose dynamics can , in certain parameter regimes , be restricted to a two - dimensional manifold .
these qubits can be used as test - beds for studying quantum mechanics at its most fundamental level @xcite , and are also potential candidates for the practical implementation of quantum information processors @xcite .
an interesting and challenging aspect of these endeavours is the process of decoherence , whereby qubits lose their quantum - mechanical nature and are rendered dynamically equivalent to their classical two - level counterparts . in the case of a superconducting charge qubit , such as the cooper - pair box ( cpb ) , dephasing ( phase decoherence ) is dominated by low - frequency noise thought to be caused by interactions with two - level fluctuators @xcite ( tlfs ) in the local environment .
these tlfs may be charge traps caused by defects / impurities in the josephson junction or in the substrate .
there has been a substantial amount of research on tlfs causing single qubit decoherence in josephson junction systems .
theoretical works have concerned larger ensembles of tlfs , both incoherent @xcite and coherent @xcite , that randomly switch between two configurations , producing low - frequency fluctuations in the relevant qubit parameters . as discussed in detail in all these works , whenever a large number of tlfs are very weakly coupled to the qubit
, their effect can be described by a conventional boson bath with a suitable chosen spectral density .
this is however not always the case there may be situations when only one or a few impurities are important .
indeed , neeley , @xcite have demonstrated the existence of _ coherent _ tlfs .
lupacu , @xcite provide evidence that these tlfs are in fact genuine two - level systems .
in neeley s experiment @xcite a single tlf that was coupled to the qubit led to an avoided crossing in the qubit energy spectrum .
the tlf was used as a proof - of - principle memory qubit , but such tlfs will in general be detrimental to the operation of superconducting qubits .
it is therefore desirable to understand the behaviour of tlfs in the vicinity of superconducting qubits in order to better - equip quantum information scientists to manage the challenge of decoherence . in view of its importance for understanding decoherence in superconducting nanocircuits
we study in this paper the effect that a few underdamped , coherent tlfs have on the quantum dynamics of a charge qubit . a single qubit coupled to a single coherent tlf , that was in turn ( under-)damped by a bosonic bath
was already considered in @xcite .
we find this type of non - equilibrium environment @xcite to be an appealing model of the environment of a josephson junction qubit because tlfs may be damped by phonons in the substrate , for example .
we refer to this as a composite _ spin - boson environment_. the case of a few tlfs , which will be considered here , is rather complex and therefore we resort to extensive numerical calculations .
our approach represents quite a general strategy for probing a composite spin - boson environment , whereby we derive a markovian master equation for the average dynamics of the probe+tlfs after tracing out the external baths . to allow for the random nature of tlf formation
, we select the tlf hamiltonian parameters according to probability distributions designated in @xcite . of the many environmental properties that could be studied in this scenario , we focus on inferring the presence or absence of coherent coupling between the tlfs .
connectivity _ of the tlfs has been identified as important in the decoherence of a bipartite qubit system @xcite , with further - reaching importance for quantum computing . performing a more thorough treatment of this situation is increasingly important in light of recent experiments @xcite involving superconducting qubits coupling predominantly to only a small number of tlfs . in these experiments , measurements on the qubit were used to infer properties of the tlf ( although that was nt the focus in @xcite ) .
that is , the qubit was used as a probe of the environment @xcite .
probing properties of a small collection of tlfs , such as whether or not they interact with each other , necessitates consideration of the full dynamics of the probe plus tlfs under the influence of external baths . from the theory point of view , identifying structural properties in the environment could be done using some form of noise correlation measurements . here
we focus on inferring environmental features via the analysis of the entanglement that will build up in a probe consisting of two non-(directly ) interacting qubits whose remote coupling is mediated by the tlfs present in the surroundings .
when the probing is `` local '' , so that each probe qubit couples to just one single tlf , the absence of quantum correlation generation would immediately signal a non - connected environment , given that the probe qubits can only become entangled if the fluctators would couple to each other .
entanglement swapping in those circumstances has been discussed in the literature @xcite .
when probe qubits are subject to the action of a _ few _ tlfs , as it happens in qubit realizations in the solid state , we will show that the remote entanglement in the probe bears signatures that can be linked to the connectivity in the environment and can in some cases be related to monogamy constraints @xcite .
given that bipartite entanglement has been shown @xcite to be lower - bounded by combinations of pseudo - spin observables , we also analyze what information can be extracted from magnetization measurements along a given direction ( in the case considered here the magnetization along the z - direction corresponds to the average charge ) and study the power spectra of magnetization observables using both single and bipartite probes .
we find that a double - qubit probe generally outperforms a single - qubit probe , a result that could perhaps be expected given the extra degrees of freedom available in the composite system .
we supplement our analysis of correlation measurements by investigating the decoherence of composite probes initially prepared in a certain maximally entangled state when subject to a composite spin - boson environment , as well as the performance of entangling gate operations when performed in the presence of this type of noise .
the paper is organized as follows .
section [ sec : system ] sets the scene by describing our model for the double - qubit probe and spin - boson environment .
a detailed derivation of the proposed master equation as well as a discussion of its validity domain are presented in the appendix .
numerical results for probing the connectivity of the spin - boson environment are presented in section [ sec : connectivity ] , using both probe entanglement and estimated power - spectrum analyses .
we summarize and discuss these results in section [ sec : discussion ] , as well as compare the double - qubit probe to a single - qubit probe . in section
[ sec : entdecay ] we investigate the decoherence of maximally - entangled bell states induced by a composite spin - boson environment .
the performance of bipartite entangling gates in the presence of this form of noise is analyzed and discussed in section [ sec : gates ] .
section [ sec : conclusion ] concludes .
the system we consider is illustrated in .
it consists of two charge qubits ( blue spheres ) acting as probes of an environment containing tlfs ( grey spheres ) .
each qubit is coupled to a few tlfs ( black lines in the figure ) but probe qubits are assumed to not directly couple to each other . in the numerical calculations we will consider the case in which there are four tlfs .
four tlfs is a balance between generating the desired spectral features ( requiring an ensemble of tlfs @xcite ) and maintaining reasonable computation time ( smaller hilbert space ) .
also , it may be the case that only a few tlfs will couple strongly to a josephson junction qubit , as in recent experiments @xcite .
we should stress that the conclusions of our work do not depend on this choice . in the charge basis , each qubit /
tlf has a local free hamiltonian consisting of both longitudinal ( @xmath0 ) and transverse ( @xmath1 ) components : @xmath2 . in the eigenbasis
the corresponding pseudo - spin hamiltonians are @xmath3 , where the spin frequency is @xmath4 .
( throughout this article we denote pauli operators in the charge basis by @xmath5 , and in the pseudo - spin basis by @xmath6 . ) for simplicity , we `` engineer '' the probe qubit hamiltonians to have only longitudinal components ( @xmath7 ) . [ [ probe ] ] probe : + + + + + + we label the identical probe qubits a and b. choosing uncoupled probe qubits for reasons that will become clear later , the total hamiltonian for the probe is the sum @xmath8 .
[ [ impurities ] ] impurities : + + + + + + + + + + + we label the four tlf impurities with @xmath9 .
the total hamiltonian for the tlfs is then @xmath10 , where @xmath11 describes coherent couplings between the tlfs , if they exist ( defined below ) .
note that the pseudo - spin basis for the impurities is different to the probe ( the @xmath12 axis is rotated relative to @xmath13 ) because the probe and tlf energies will differ , in general .
recent theoretical work @xcite suggested specific distributions of these tlf energies in order to account for both low- and high - frequency noise observed in superconducting quantum systems . in our numerical study , we have adopted these distributions to determine the tlf bias energies @xmath14 ( linear distribution ) and tunnel amplitudes @xmath15 ( log - uniform distribution ) . throughout the paper we will refer to @xmath15 as the _ local field_. again ,
note that our results are independent of the specific choice of frequency distribution and the same qualitative results can be derived when using a different functional form , e.g. , a linear or a uniform distribution in a selected interval around the qubit frequency .
double - qubit probe schematic .
blue spheres are the probe qubits , alice and bob .
grey spheres are the tlfs .
probe tlf couplings ( @xmath16 ) are depicted as black lines .
tlf - tlf couplings ( @xmath17 ) are specified by red lines .
each fluctuator is also subject to the action of a bosonic bath at a temperature @xmath18 .
interactions between a and b are mediated by the tlfs , which lead to remote ( tlf mediated ) entanglement generation in the probe.,scaledwidth=50.0% ] [ [ interactions ] ] interactions : + + + + + + + + + + + + + it is sensible to expect that the dominant interaction in a system of coherent two - level charges is an electrostatic one @xcite . that is , charge - charge interactions . we therefore assume bipartite zz interactions ( @xmath19 ) between subsystems . within the tlfs
we assume nearest - neighbour zz interactions of strength @xmath17 , where @xmath20 .
we assume no direct interaction between the probe qubits a and b.
so , @xmath21 . for coupling strength @xmath16 between the @xmath22th impurity and the probe qubits
, we have @xmath23 .
we define @xmath24 . in the numerical simulations , noting that we expect distant tlfs to have very little impact , we have assumed all couplings @xmath25 and all @xmath26 to avoid unnecessarily cumbersome results .
our conclusions are valid even when small variations in the parameters , of the order of @xmath27 , are considered .
the impurities are coupled to independent reservoirs of bosons ( e.g. , phonons in the substrate ) , leading to dissipation ( damping ) as in the spin - boson model @xcite . under appropriate weak - coupling assumptions ( see [ sec : mederive ] for our derivation ) , the dissipative dynamics of the composite system ( qubits plus damped tlfs ) can be expressed in the following born - markov master equation for the joint state @xmath28 of the probe plus impurities : @xmath29 + \sum_j ( \mathcal{d}_\mathrm{z}^{(j ) } + \mathcal{d}_+^{(j ) } + \mathcal{d}_-^{(j)})\rho(t ) , \label{eq : me}\end{aligned}\ ] ] where the total hamiltonian is @xmath30 the @xmath31 superoperators @xcite represent decoherence in the tlfs due to coupling with the bosonic baths , which are at temperature @xmath18 .
the decoherence consists of dephasing @xmath32 $ ] , emission into the baths @xmath33 $ ] , and absorption from the baths @xmath34 $ ] .
the tlf decoherence rates @xmath35 are proportional to the respective dephasing , emission and absorption rates @xmath36 ( see [ sec : mederive ] ) , which are functions of the temperature and spectral properties of the @xmath22th bosonic bath ( with absorption dramatically reduced at low temperatures ) .
further , the tlf decoherence rates are also functions of the ratio of local field to bias , @xmath37 @xcite . to get a feel for the influence of this ratio ,
if we assume dissipation - limited dephasing ( @xmath38 ) and sufficiently low temperature ( @xmath39 ) , then @xmath40 dictates the dominance of pure dephasing or relaxation in each tlf . specifically , @xmath41 so that pure dephasing dominates the tlf decoherence for weak local fields , and relaxation dominates for strong local fields . following @xcite , we distribute the random tlf parameters @xmath14 , @xmath15 , and @xmath42 as per the distributions @xmath43 , @xmath44 , @xmath45 .
we choose these parameters to take values within the following moderate ranges : @xmath14 @xmath46 @xmath47 ; @xmath15 @xmath46 @xmath48 ; @xmath42 @xmath46 @xmath49 $ ] , where @xmath50 is the minimum spin frequency amongst the tlfs .
we are free to select sensible values for the overbar quantities @xmath51 and @xmath52 , which we will reference to the tunable probe frequency @xmath53 .
importantly , the tlfs are underdamped ( therefore requiring a quantum - mechanical description ) , so @xmath54 .
we take the probe - tlf coupling to be uniform ( @xmath26 ) and weak compared with all of the tlf frequencies : @xmath55 .
our assumption of weak probe - tlf coupling simplifies the master equation derivation significantly ( see equation [ approxexp ] ; a detailed discussion of an analogous situation can be found in @xcite ) , and is in addition to the standard born - markov approximation of weak tlf - bath coupling . weak probe - tlf coupling is in accord with the recent experiments of @xcite , as well as the experiment of @xcite ( see @xcite ) where @xmath56 mhz and @xmath57 ghz .
the tlf - tlf coupling is also assumed to be uniform @xmath25 .
note that the authors of @xcite point out that @xmath58 must also be randomly distributed in order to realize @xmath59 noise in the probe .
generating the correct statistics is not within the scope of this paper as we are interested in describing effects when the environment is dominated by only a few fluctuators .
we restrict our knowledge to the probe subsystem ( as would be the case in an experiment ) .
a notable observable quantity on the probe is the magnetization , which is related to a simple sum of pauli operators : @xmath60 .
the appeal of considering the probe magnetization is that it requires only tractable , local measurements on each probe qubit .
that is , our results may be easily tested in an experiment .
it is worth noting that a result of audenaert and plenio @xcite shows that measuring correlations @xmath61 along the xx and zz ` directions ' suffice to give a lower bound on the probe entanglement .
this can remove the requirement for full tomographic ( entanglement ) measurements when verifying or quantifying entanglement in the probe .
the time series resulting from measuring the probe s x - magnetization is @xmath62 .
the mean - square power spectrum of @xmath63 is given by @xmath64 where the reduced auto - correlation function is @xmath65 .
the angle brackets here denote the expectation value of an operator @xmath66 $ ] . in reality
, @xmath67 is a discrete quantity ( the data is a time series ) , and so the power spectrum obtained is an estimate ( the fourier transform of the reduced autocorrelation of the time series @xmath67 ) .
this estimate of average power as a function of frequency can theoretically be improved ( by increasing the duration of the experiment , for example ) , but this may not be practical in reality . strictly speaking , @xmath67 must be a wide - sense stationary process ( time - independent first and second moments ) for the power spectrum to exist .
can probe observables reveal the degree of connectivity of a composite spin - boson environment ? in this section we present numerical results showing that this is indeed the case .
observable quantities we consider are the probe magnetization , its estimated power spectrum , and the remote entanglement between the probe qubits . entanglement generated between probe qubits that are initially in a separable state is primarily due to the structure of the spin - boson environment ( e.g. , the presence or absence of tlf - tlf interactions in the surroundings of the probe ) .
initially we set the probe to be in a state orthogonal to the @xmath13 eigen - axis : @xmath68 .
the tlfs are assumed to be initially in a zero - temperature thermal state , i.e. , the ground state of @xmath69 , which we denote as @xmath70 .
so , @xmath71 . we plot observable quantities as a function of the ratio of tlf - tlf coupling strength to probe - tlf coupling strength , @xmath72 , which is often believed to be small ( see @xcite , for example ) .
considers the estimated power spectrum of @xmath73 . considers the build up of entanglement in the probe .
consider the case of `` weak '' local fields in the spin - boson environment where @xmath75 ( specifically @xmath76 , where tlf dephasing dominates relaxation : @xmath77 ) . in our probing of tlf connectivity ,
we tune the ratio of probe
tlf splitting @xmath78 to 10 , 3 and 1 , corresponding to figures [ fig2a ] , [ fig2b ] and [ fig2c ] , respectively .
we observe two effects : 1 . for weaker tlf interconnectivity @xmath79
, a decrease is observed in the height of the single dominant peak in the spectrum ; 2 . for stronger tlf interconnectivity @xmath80 ,
the spectrum splits into multiple peaks , the most dominant of which is shifted in frequency relative to @xmath81 .
this is visible in where the gold traces ( @xmath82 ) are the most qualitatively different from the blue traces ( @xmath83 ) .
the power in the signal @xmath73 redistributes from one dominant frequency for unconnected tlfs ( @xmath83 ) , to multiple frequencies as @xmath72 approaches unity ( highly connected tlfs ) .
remarkably , the most dominant peak for highly connected tlfs @xmath82 is qualitatively similar to the case of an isolated probe ( ) a single peak at @xmath84 although the peak visibility ( height ) is noticeably less than the isolated probe .
[ fig2a ] estimated power spectrum of @xmath85 for weak local fields @xmath86 with the initial probe state @xmath87 .
sampling parameters ( units of @xmath88 ) : @xmath89 , @xmath90 , @xmath91 .
figure [ fig2d ] is the power spectrum for an isolated probe .
the insets show the smaller peaks magnified .
the relative heights and positions of the peaks can be used to distinguish unconnected tlfs ( @xmath83 , thick solid blue line ) from highly connected tlfs ( @xmath92 , thick dashed gold line ) . , title="fig:",scaledwidth=50.0% ] [ fig2b ] estimated power spectrum of @xmath85 for weak local fields @xmath86 with the initial probe state @xmath87 .
sampling parameters ( units of @xmath88 ) : @xmath89 , @xmath90 , @xmath91 .
figure [ fig2d ] is the power spectrum for an isolated probe .
the insets show the smaller peaks magnified .
the relative heights and positions of the peaks can be used to distinguish unconnected tlfs ( @xmath83 , thick solid blue line ) from highly connected tlfs ( @xmath92 , thick dashed gold line ) . , title="fig:",scaledwidth=50.0% ] [ fig2c ] estimated power spectrum of @xmath85 for weak local fields @xmath86 with the initial probe state @xmath87 . sampling parameters ( units of @xmath88 ) : @xmath89 , @xmath90 , @xmath91 .
figure [ fig2d ] is the power spectrum for an isolated probe .
the insets show the smaller peaks magnified .
the relative heights and positions of the peaks can be used to distinguish unconnected tlfs ( @xmath83 , thick solid blue line ) from highly connected tlfs ( @xmath92 , thick dashed gold line ) . , title="fig:",scaledwidth=50.0% ] [ fig2d ] estimated power spectrum of @xmath85 for weak local fields @xmath86 with the initial probe state @xmath87 .
sampling parameters ( units of @xmath88 ) : @xmath89 , @xmath90 , @xmath91 .
figure [ fig2d ] is the power spectrum for an isolated probe .
the insets show the smaller peaks magnified .
the relative heights and positions of the peaks can be used to distinguish unconnected tlfs ( @xmath83 , thick solid blue line ) from highly connected tlfs ( @xmath92 , thick dashed gold line ) .
, title="fig:",scaledwidth=50.0% ] above we have established that we can distinguish between highly connected tlfs ( @xmath93 ) and unconnected tlfs ( @xmath83 ) for weak local field strength @xmath94 .
it is interesting to ask how stronger local fields affect our ability to distinguish these two values of @xmath95 . to explore this
, we increase the ratio of tlf field strength to splitting : @xmath96 , which was less than 1 in .
this varies the range for @xmath15 , which ( we remind the reader ) we have taken to be @xmath97 .
this range ensures a sensible variation in the tlf local field strengths of no greater than one - half of the probe frequency .
( we assume that the tlf energies are distributed over a relatively small range as might be expected for systematically formed impurities / defects . ) as the local field increases @xmath98 , two different effects occur : 1 .
larger local fields cause relaxation to dominate over dephasing in the tlf decoherence ; 2 .
the tlf eigenstates increasingly align towards the @xmath1 axis , and seem to have a decreasing effect on the probe , perhaps because the interaction is of the @xmath19 type . evidence to support this
is shown in , which shows a peak visibility ( height ) reduction of about one order of magnitude as @xmath99 is increased by one order of magnitude from 1/3 to 3 [ [ fig3a ] to [ fig3b ] ] .
so , although extra features appear in the power spectrum , they become increasingly difficult to observe . despite this , there is at least one plot in each column for which @xmath83 and @xmath93 are distinguishable .
thus it is apparent that tuning the probe frequency ( selecting a row in figure [ fig3 ] ) allows these tlf connectivities to be distinguished for a wide range of values of the tlf local field strength ( we obtained similar results for @xmath100 ) .
[ fig3a ] estimated power spectrum of @xmath85 for @xmath101 ( top row ) , @xmath102 ( second row ) and @xmath103 ( third row ) .
as the local field strength @xmath104 increases from [ fig3a][fig3b ] , tlf relaxation dominates tlf dephasing .
see text for discussion .
, title="fig : " ] + estimated power spectrum of @xmath85 for @xmath101 ( top row ) , @xmath102 ( second row ) and @xmath103 ( third row ) .
as the local field strength @xmath104 increases from [ fig3a][fig3b ] , tlf relaxation dominates tlf dephasing
. see text for discussion .
, title="fig : " ] + estimated power spectrum of @xmath85 for @xmath101 ( top row ) , @xmath102 ( second row ) and @xmath103 ( third row ) .
as the local field strength @xmath104 increases from [ fig3a][fig3b ] , tlf relaxation dominates tlf dephasing .
see text for discussion .
, title="fig : " ] [ fig3b ] estimated power spectrum of @xmath85 for @xmath101 ( top row ) , @xmath102 ( second row ) and @xmath103 ( third row ) .
as the local field strength @xmath104 increases from [ fig3a][fig3b ] , tlf relaxation dominates tlf dephasing .
see text for discussion .
, title="fig : " ] + estimated power spectrum of @xmath85 for @xmath101 ( top row ) , @xmath102 ( second row ) and @xmath103 ( third row ) . as the local field strength @xmath104 increases from [ fig3a][fig3b ] , tlf relaxation dominates tlf dephasing .
see text for discussion .
, title="fig : " ] + estimated power spectrum of @xmath85 for @xmath101 ( top row ) , @xmath102 ( second row ) and @xmath103 ( third row ) . as the local field strength @xmath104 increases from [ fig3a][fig3b ] , tlf relaxation dominates tlf dephasing .
see text for discussion .
, title="fig : " ] interactions between probe qubits are mediated by the tlfs and entanglement between the probe qubits can be generated in this indirect way .
we now consider using the probe entanglement to distinguish between connected and unconnected tlfs .
we use the logarithmic negativity as a measure of bipartite entanglement between the probe qubits , defined as @xcite @xmath105 where @xmath106 denotes the trace norm , and @xmath107 is the partial transpose of @xmath108 .
figure [ fig4 ] shows the logarithmic negativity of @xmath109 for the same data sets as in ( weak local fields @xmath86 ) .
we make two observations .
firstly , tuning the probe frequency provides only one benefit for distinguishing @xmath83 from @xmath110 .
this is evidenced by the remarkable qualitative similarity between figures [ fig4a ] , [ fig4b ] , and [ fig4c ] .
the benefit is the number of probe qubit cycles required to distinguish the two cases ( time is shown in units of probe qubit cycles ) .
secondly , it is clear that strongly coupled tlfs ( the gold line ) cause less entanglement to generate within the probe .
this can be explained by invoking the concept of entanglement monogamy @xcite . as the tlf - tlf connectivity increases , the indirect link between the two probe qubits is weakened , and so remote entanglement generation slows .
it is important to emphasize which partitions one should consider when invoking monogamy arguments .
the relevant quantity is the entanglement shared between each probe qubit and a given tlf ; this is the quantity that is sensitive to entanglement sharing in two ways : ( 1 ) it decreases as the entanglement in the probe builds up , independently of the connectivity in the tlf environment ; ( 2 ) it `` feels '' the tlf - tlf coupling in the sense that , within a selected time interval , the stronger the fluctuators couple , the smaller the entanglement between probe qubit and tlf becomes . as a result , remote entanglement builds up more slowly when fluctuators couple so that it quickly degrades in a decohering environment , as illustrated by results in .
[ fig4a ] entanglement ( logarithmic negativity ) in the probe as a function of probe cycles , for small tlf local field strength @xmath86 .
note the distinct difference between strongly interacting tlfs ( @xmath111 ) and non - interacting tlfs ( @xmath83).,title="fig:",scaledwidth=33.0% ] [ fig4b ] entanglement ( logarithmic negativity ) in the probe as a function of probe cycles , for small tlf local field strength @xmath86 .
note the distinct difference between strongly interacting tlfs ( @xmath111 ) and non - interacting tlfs ( @xmath83).,title="fig:",scaledwidth=33.0% ] [ fig4c ] entanglement ( logarithmic negativity ) in the probe as a function of probe cycles , for small tlf local field strength @xmath86 .
note the distinct difference between strongly interacting tlfs ( @xmath111 ) and non - interacting tlfs ( @xmath83).,title="fig:",scaledwidth=33.0% ] + unlike the power spectrum ( ) , the probe entanglement @xmath112 remains useful for distinguishing @xmath83 from @xmath113 for strong local fields in the tlfs where @xmath98 .
figure [ fig5 ] shows the probe entanglement as a function of time for the same data sets as in .
it is clear that entanglement is generated within the probe for some time , before the tlf decoherence causes it to dissipate .
as argued before , this loss of generated probe entanglement occurs faster for highly connected tlfs with @xmath82 , as one might expect , even without no explicit mention of monogamy constraints , since these tlf - tlf connections provide more links between the probe qubits and the tlf decoherence channels .
this faster dissipation of generated entanglement for highly - connected tlfs allows us to distinguish between @xmath81 and @xmath82 after 10 to 50 probe qubit cycles , depending on the tlf parameters .
similar conclusions can be drawn for @xmath114 .
[ fig5a ] entanglement ( logarithmic negativity ) between the probe qubits for @xmath101 ( top row ) , @xmath102 ( second row ) and @xmath103 ( third row ) .
the cases @xmath83 ( solid blue line ) and @xmath93 ( dashed gold line ) are qualitatively distinguishable unless both @xmath115 and @xmath98 . note that the upper limits of the axes change between plots .
same data sets as in .
, title="fig : " ] + entanglement ( logarithmic negativity ) between the probe qubits for @xmath101 ( top row ) , @xmath102 ( second row ) and @xmath103 ( third row ) . the cases @xmath83 ( solid blue line ) and @xmath93 ( dashed gold line ) are qualitatively distinguishable unless both @xmath115 and @xmath98 .
note that the upper limits of the axes change between plots .
same data sets as in . ,
title="fig : " ] + entanglement ( logarithmic negativity ) between the probe qubits for @xmath101 ( top row ) , @xmath102 ( second row ) and @xmath103 ( third row ) . the cases @xmath83 ( solid blue line ) and @xmath93 ( dashed gold line ) are qualitatively distinguishable unless both @xmath115 and @xmath98 .
note that the upper limits of the axes change between plots .
same data sets as in .
, title="fig : " ] [ fig5b ] entanglement ( logarithmic negativity ) between the probe qubits for @xmath101 ( top row ) , @xmath102 ( second row ) and @xmath103 ( third row ) .
the cases @xmath83 ( solid blue line ) and @xmath93 ( dashed gold line ) are qualitatively distinguishable unless both @xmath115 and @xmath98 .
note that the upper limits of the axes change between plots .
same data sets as in .
, title="fig : " ] + entanglement ( logarithmic negativity ) between the probe qubits for @xmath101 ( top row ) , @xmath102 ( second row ) and @xmath103 ( third row ) . the cases @xmath83 ( solid blue line ) and @xmath93 ( dashed gold line ) are qualitatively distinguishable unless both @xmath115 and @xmath98 .
note that the upper limits of the axes change between plots .
same data sets as in .
, title="fig : " ] + entanglement ( logarithmic negativity ) between the probe qubits for @xmath101 ( top row ) , @xmath102 ( second row ) and @xmath103 ( third row ) . the cases @xmath83 ( solid blue line ) and @xmath93 ( dashed gold line ) are qualitatively distinguishable unless both @xmath115 and @xmath98 .
note that the upper limits of the axes change between plots .
same data sets as in .
, title="fig : " ] since we are performing local measurements on each probe qubit , one might ask if there are any advantages of using a double - qubit probe .
an obvious advantage is that entanglement within the probe becomes an accessible quantity that is not possible in a single - qubit probe .
this is important for detecting the tlf - tlf connectivity , as we have seen that this task was achievable over a wider range of tlf parameters using the probe entanglement than the power spectra of the probe magnetization ( which yielded qualitatively similar results for both types of probes ) . in section [ sec : connectivity ]
we were able to distinguish between composite spin - boson environments with high connectivity from those with low connectivity .
the parameter ranges for which our findings were robust are : @xmath116 ( tunable ) , @xmath117 .
other parameters ( @xmath118 , etc ) are restricted to lie within the ranges that ensure validity of the master equation see [ sec : mederive ] . for all numerical calculations in this paper
we have assumed effectively zero - temperature bosonic baths where @xmath119^{-1 } \ll 1 $ ] .
the relevant frequencies for experiments with josephson qubits are in the vicinity of 10 ghz @xcite , with cryostat temperatures of the order of 30 mk @xcite .
these values give @xmath120 , so the low - temperature approximation is good . in general , determining the probe entanglement would require full quantum - state tomography .
here we consider estimating the probe entanglement from measurements less costly than full quantum state tomography , as described in @xcite ( and references within ) .
the result is lower bounds on the entanglement we refer the reader to @xcite for details .
for all parameter regimes considered in this paper we found that one of the lower bounds given in @xcite the optimal one given in ( [ eq : cprime ] ) provided a remarkably good approximation to the probe entanglement for all times .
the other lower bounds @xmath121 in @xcite did nt approximate the probe entanglement well for any time .
an example is shown in figure [ fig6 ] .
the solid line is the entanglement within the probe ( logarithmic negativity ) , and the dotted line shows the lower bound given by @xmath122 , \label{eq : cprime}\ ] ] where @xmath123 are the eigenvalues of the matrix @xmath124 the matrix @xmath125 is formed from probe observables @xmath126 $ ] ( @xmath127 ) .
note that @xmath128 due to the symmetry of the problem .
lower bounds on the probe entanglement as described in @xcite . only the optimal bound @xmath129 ( dotted red line )
provides a good approximation to the probe entanglement .
parameters are @xmath86 , @xmath130 , and @xmath131.,title="fig:",scaledwidth=45.0% ] lower bounds on the probe entanglement as described in @xcite . only the optimal bound @xmath129 ( dotted red line )
provides a good approximation to the probe entanglement .
parameters are @xmath86 , @xmath130 , and @xmath131.,title="fig:",scaledwidth=45.0% ]
entanglement has been identified as a key resource for quantum information processing .
it is therefore important to study the loss of entanglement induced by coupling with an environment , as this coupling is generally unavoidable . in this section
we reconsider our double - qubit probe as a double - qubit _ register _ ( dqr ) interacting with the same spin - boson environment ( four damped tlfs ) as above .
starting the dqr in an entangled pure - state , and the tlfs in their ground state ( as before ) with weak local fields @xmath132 , we numerically investigate the behaviour of the logarithmic negativity as a function of time .
specifically , we consider the lifetime of distillable entanglement ( for which the logarithmic negativity is an upper bound ) .
we compare our results to previous studies of entanglement decay @xcite , all of which used rather less sophisticated models for the environment .
nevertheless , we find some qualitative similarities between our results and previous work .
reference @xcite consider multiqubit states whereby each qubit is damped by independent baths .
we refer to this as ` direct ' damping , by an equilibrium environment ( a reservoir ) . in our non - equilibrium spin - boson environment ,
the damping is mediated by the tlfs and we refer to this as ` indirect ' damping of the dqr .
reference @xcite parameterizes time via the probability for a qubit to exchange a quantum of energy with its bath ( in the absence of pure dephasing ) , @xmath133 $ ] . here
@xmath134 is the zero - temperature damping rate , and @xmath135 is the mean number of excitations in the bath ( @xmath136 is zero temperature ) .
when the dqr logarithmic negativity falls below an arbitrarily small fraction of its initial value , @xmath137 , the distillable entanglement can be considered zero .
the time at which this occurs is @xmath138 . for generalized ghz states ( requiring @xmath139 qubits ) , and for three different types of direct damping
, @xcite found that @xmath140 ( although they were looking at it from a slightly different perspective , as we comment later ) .
remarkably , we find numerical evidence of the same qualitative behaviour for the decay of the @xmath141 bell states in a dqr ( see figure [ fig7 ] ) . for low temperature @xmath142mk ( as we have considered throughout this paper unless otherwise noted ) , weak tlf local field @xmath143 , and tlfs with relatively small charge : @xmath144 , we consider the dqr to be initially in each of the four bell states in turn : @xmath145 , @xmath146 .
figure [ fig7 ] shows the dqr logarithmic negativity as a function of time , as well as @xmath147 for a dqr initially in the states @xmath141 .
it is evident that @xmath148 for @xmath141 .
further , we can see that interacting tlfs ( the gold traces ) tend to reduce the dqr entanglement faster .
this is expected since interacting tlfs provide more connections between the dqr and the baths . for the chosen set of parameters ( particularly @xmath149 for the dqr ) , the states @xmath150 commute with @xmath151 and so do nt evolve , nor couple to the tlfs .
we found the same qualitative behaviour for @xmath152 and @xmath153 when the tlf local fields were not strong : @xmath154 .
for strong local fields @xmath155 , the linear relationship @xmath148 did not hold in general .
this was because the probe entanglement tended to exhibit quite erratic behaviour , such as multiple collapses and revivals .
decay of entanglement ( logarithmic negativity ) in a double - qubit register initially in the bell states @xmath156 ( top row ) and @xmath157 ( second row ) for uncoupled tlfs ( @xmath83 , solid blue line ) and coupled tlfs ( @xmath93 , dashed gold line ) .
other parameters are @xmath86 and @xmath158
. see text for discussion.,title="fig:",scaledwidth=45.0% ] decay of entanglement ( logarithmic negativity ) in a double - qubit register initially in the bell states @xmath156 ( top row ) and @xmath157 ( second row ) for uncoupled tlfs ( @xmath83 , solid blue line ) and coupled tlfs ( @xmath93 , dashed gold line ) .
other parameters are @xmath86 and @xmath158
. see text for discussion.,title="fig:",scaledwidth=45.0% ] + decay of entanglement ( logarithmic negativity ) in a double - qubit register initially in the bell states @xmath156 ( top row ) and @xmath157 ( second row ) for uncoupled tlfs ( @xmath83 , solid blue line ) and coupled tlfs ( @xmath93 , dashed gold line ) .
other parameters are @xmath86 and @xmath158
. see text for discussion.,title="fig:",scaledwidth=45.0% ] decay of entanglement ( logarithmic negativity ) in a double - qubit register initially in the bell states @xmath156 ( top row ) and @xmath157 ( second row ) for uncoupled tlfs ( @xmath83 , solid blue line ) and coupled tlfs ( @xmath93 , dashed gold line ) .
other parameters are @xmath86 and @xmath158
. see text for discussion.,title="fig:",scaledwidth=45.0% ] a comment on the previous work in @xcite is appropriate here . in those works ,
the decay of @xmath159-particle entanglement was considered as a function of @xmath159 . in @xcite
it was found that @xmath160 .
here we have fixed @xmath161 and found that @xmath162 .
our focus is slightly different , but it is interesting that the loss of distillable entanglement ( logarithmic negativity ) is qualitatively the same for direct and indirect damping of bipartite qubit states ( within the parameter regimes discussed in the previous paragraph ) .
it is important to remark that the coincidence with the predictions for the decoherence of multipartite states subject to independent reservoirs should not be considered as a general result given that we analyzed a very special case , which is the one of two entangled qubits in selected parameter regimes .
what is relevant for our purposes is the fact that the agreement with the analytical prediction in @xcite for direct decoherence points out a sharp asymmetry in the processes of entanglement `` destruction '' and ( remote ) entanglement generation in a composite environment , so that there are circumstances where the tlf systems may be essentially invisible when analyzing the decoherence of initially entangled probe states , while the presence of the tlfs would be revealed when monitoring entanglement creation in the probe .
in this section we investigate the effects of the composite spin - boson environment on the performance of entangling gates . starting the dqr in the separable state @xmath163 , we consider two entangling gates : a zz gate ( @xmath164 ) ; and an xx+yy gate [ @xmath165 . in the ideal case
there are no tlfs and bipartite entanglement ( quantified again by the logarithmic negativity ) is generated between the isolated register qubits in an oscillatory fashion as shown by the solid red curves in . in the presence of four tlfs , the entangling gate performance is clearly reduced , and further modified depending on the strength of the local fields in the tlfs .
this may be understood as follows .
we have argued that the presence of a coherent coupling between the tlfs leads to a decrease in the effective interaction strength between the qubits in the probe , i.e. , the `` effectiveness '' of the indirect link between probe qubits is diminished , a result that can be interpreted in terms of monogamy constraints leading to a slow down in the process of remote entanglement creation . in the case of the entangling gates , one could perhaps argue similarly , considering now a bipartition separating the probe qubits . the higher the connectivity in the environment , the slower the entangling gate can operate , as illustrated in . [ fig8a ] entanglement ( logarithmic negativity ) between the register qubits for @xmath131 , @xmath152 generated by xx+yy ( top row ) , and zz ( bottom row ) gates .
the presence of tlfs ( blue dashed and green dotted lines ) diminishes the performance of the entangling gates , as would be expected .
, title="fig : " ] + entanglement ( logarithmic negativity ) between the register qubits for @xmath131 , @xmath152 generated by xx+yy ( top row ) , and zz ( bottom row ) gates .
the presence of tlfs ( blue dashed and green dotted lines ) diminishes the performance of the entangling gates , as would be expected .
, title="fig : " ] [ fig8b ] entanglement ( logarithmic negativity ) between the register qubits for @xmath131 , @xmath152 generated by xx+yy ( top row ) , and zz ( bottom row ) gates .
the presence of tlfs ( blue dashed and green dotted lines ) diminishes the performance of the entangling gates , as would be expected .
, title="fig : " ] + entanglement ( logarithmic negativity ) between the register qubits for @xmath131 , @xmath152 generated by xx+yy ( top row ) , and zz ( bottom row ) gates .
the presence of tlfs ( blue dashed and green dotted lines ) diminishes the performance of the entangling gates , as would be expected .
, title="fig : " ] at longer times ( the order of 100 register - qubit cycles and greater ) , the dqr appears to approach an entangled steady state for weak local fields @xmath86 .
this is shown in for the zz gate ( the same qualitative behaviour occurred for the xx+yy gate ) .
we were unable to obtain analytical results to verify the presence of steady - state entanglement in the dqr generated by either gate ( zz or xx+yy ) , when coupled to four tlfs .
our numerical study found that entanglement generated in the dqr was dissipated more rapidly ( in less probe qubit cycles ) for strong local fields in the tlfs .
long - time entanglement ( logarithmic negativity ) between the register qubits generated by a zz gate , for @xmath131 , @xmath152 and @xmath166.,scaledwidth=65.0% ]
we have considered superconducting qubits which are subject to decoherence dominated by low - frequency noise thought to be produced by interactions with a small number of defects / impurities .
we model these impurities as coherent two - level fluctuators ( tlfs ) that are under - damped by baths of bosonic modes ( e.g. phonons ) .
we probed such a composite spin - boson environment by making measurements on a pair of noninteracting qubits that each interact directly with the tlfs .
our extensive numerical study revealed that the presence or absence of coherent coupling ( _ connectivity _ ) between the tlfs can be discriminated in two ways : from the estimated power spectrum of the probe magnetization ( requiring relatively long - time measurements ) and from entanglement generated within the probe , mediated by the spin - boson environment .
we argue that entanglement monogamy considerations @xcite allow interpretation of our results in terms of an effective decrease in the remote interaction strength when the environment is connected , which yields to a creation of quantum correlations on a much larger time scale as compared with the uncoupled fluctuator case .
we also showed that this remotely generated entanglement can be well - estimated by a lower bound @xcite that requires less experimental effort than the full quantum state tomography required to evaluate the entanglement .
the upshot is that this connectivity of the tlfs should be discernible using tractable measurements in a real experiment .
this result is important for studies of quantum - mechanical phenomena in josephson devices ( including quantum computing ) where it is desirable to minimize the effects of decoherence , for which the tlf - tlf connectivity may play a significant role @xcite .
when considering the effects of the spin - boson environment on a double - qubit register initially prepared in a maximally - entangled ( bell ) state , we showed that the presence of tlfs may be unnoticeable in certain parameter regimes , in the sense that entanglement degradation there is well approximated by the same decrease law as for direct decoherence .
this fact emphasizes the possible usefulness of monitoring the reverse process of entanglement generation for environmental probing .
the presence of tlf - tlf coupling also showed a reduction in the performance of entangling gate operations performed on the register .
our extensive numerical study can be supplemented by an analytical treatment of a simpler situation valid for short times , where tlf decoherence can be ignored .
these results will be presented elsewhere @xcite .
we are very grateful to simone montangero for discussions on several aspects of the simulations and to shash virmani for discussions on monogamy constraints .
this work was supported by the eu through the strep project corner , the integrated project on _ qubit applications _ qap , and the integrated project eurosqip .
ar acknowledges support from a university of hertfordshire fellowship .
consider a single charge qubit `` system '' coupled to a finite number of independent charged impurities that fluctuate coherently between two configurations .
these coherent two - level fluctuators ( tlfs ) are coupled to independent bosonic baths ( phonons in the substrate , for example ) that produce damping .
the total hamiltonian is the sum of free hamiltonians for the single qubit , the tlfs , and the baths , and the interaction hamiltonians : @xmath167 the free hamiltonians are @xmath168 , @xmath169 , @xmath170 ( the baths can be assumed to be identical so @xmath171 ) .
the coupling hamiltonians are @xmath172 ( coulomb interactions ) , @xmath173 ( similarly to @xmath174 , the couplings are assumed to be independent of the baths : @xmath175 ) . we denote the charge - basis pauli operators by @xmath176 . at this stage
the tlfs are not interacting .
the tlf - related energies @xmath14 , @xmath15 and @xmath16 are randomly distributed following independent distributions discussed in @xcite .
the details of these distributions are critical for realizing the experimentally observed @xmath59 noise spectrum of the qubit voltage / bias .
we refer to the eigenbases of @xmath177 and @xmath178 as the pseudo - spin bases , and denote the corresponding pauli operators as @xmath179 .
the hamiltonians in the pseudo - spin basis are @xmath180 where @xmath37 . in order to obtain the master equation we move to an interaction picture with respect to the hamiltonian @xmath181 in this picture
the evolution equation for the total system is : @xmath182\ ] ] where @xmath183 by iterating once the above equation as usual @xcite we obtain @xmath184\ ] ] where @xmath185 denote the trace over all of the baths . by assuming a factorized initial state of the form @xmath186 , with @xmath187 the initial state of the `` qubit + tlfs '' and @xmath188 a thermal state of the @xmath22th bath
, we may make a born approximation in the coupling constants @xmath189 .
the evolution equation becomes @xmath190 .\ ] ] since the baths are independent and they are all in a thermal state ( diagonal in the number basis ) , by inserting the expression for @xmath191 it is easy to check that @xmath192\bigg ] .\end{aligned}\ ] ] this is basically a sum of the expressions obtained in the standard derivation @xcite for the one qubit case .
now we make the next crucial assumption in the derivation . assuming weak qubit - fluctuator coupling whereby @xmath193 ( as in the experiments of @xcite see below ) , we may approximate the heisenberg operators of the tlfs in the interaction picture as : @xmath194 where @xmath195 is the sum of the free hamiltonians .
for example , we find that @xmath196
. one may question the validity of our assumption of weak qubit - fluctuator coupling . for guidance
we consider the experiments of @xcite ( charge qubit ) , @xcite ( phase qubit ) and @xcite ( flux qubit ) , where @xmath197 and @xmath198 , which falls within this weak - coupling regime ( the qubit - fluctuator coupling strength in the experiment of @xcite is estimated in @xcite ) . in this regime
we can insert the result of the one qubit case for every fluctuator @xmath22 : @xmath199/2\right\}\\ & & + \gamma_+^{(j)}\left\{\tilde{s}^{(j)}_+(t)\tilde{\rho}(t)\tilde{s}^{(j)}_-(t ) - [ \tilde{s}^{(j)}_-(t)\tilde{s}^{(j)}_+(t)\tilde{\rho}(t ) + \tilde{\rho}(t)\tilde{s}^{(j)}_-(t)\tilde{s}^{(j)}_+(t ) ] /2\right\ } , \end{aligned}\ ] ] where the tlf ladder operators are @xmath200 , and the decoherence rates are @xmath201 , @xmath202 and @xmath203 . here
@xmath204 , @xmath205 and @xmath206 are the dephasing rate , the spontaneous emission rate and stimulated emission rate respectively , which can be calculated by knowing the spectral properties and temperature of the bath .
now we return to the schrdinger picture . for consistency
the heisenberg operators need to be sent back into the schrdinger picture with the same @xmath207 used above , so that in fact we just need to remove the @xmath208 argument everywhere above because all the oscillating phase factors accrued actually cancel .
therefore the final result is : @xmath209 + \sum_j \gamma_\mathrm{z}^{(j)}\left[{\hat{s}_\mathrm{z}}^{(j)}\rho(t){\hat{s}_\mathrm{z}}^{(j)}-\rho(t)\right ] \\ & & + \gamma_-^{(j)}\left[{\hat{s}}^{(j)}_-\rho(t){\hat{s}}^{(j)}_+ - ( { \hat{s}}^{(j)}_+{\hat{s}}^{(j)}_-\rho(t)+\rho(t){\hat{s}}^{(j)}_+{\hat{s}}^{(j)}_-)/2\right ] \\ & & + \gamma^{(j)}_+\left[{\hat{s}}^{(j)}_+\rho(t){\hat{s}}^{(j)}_- - ( { \hat{s}}^{(j)}_-{\hat{s}}^{(j)}_+\rho(t)-\rho(t){\hat{s}}^{(j)}_-{\hat{s}}^{(j)}_+)/2\right ] .\end{aligned}\ ] ] this master equation is valid when @xmath210 and @xmath211 , @xmath212 . the first inequality is needed in the interaction picture ( [ approxexp ] ) and the second inequality is the standard requirement for the born - markov approximation . within the same framework we can analyze other situations , such as interacting tlfs , and additional qubits coupled to the tlfs .
the master equation for interacting tlfs is the same as above , but with the additional hamiltonian @xmath213 and the requirement that @xmath214 .
additional qubits can be included under similar conditions .
r j , clerk a a , girvin s m , lehnert k w , and devoret m h 2003 noise and measurement backaction in superconducting circuits : qubits as spectrometers of quantum noise _ noise and information in nanoelectronics , sensors and standards _
vol 5115 ) ed l b kish ( bellingham , wa , usa ) pp 356376 . | we consider non - interacting multi - qubit systems as controllable probes of an environment of defects / impurities modelled as a composite spin - boson environment .
the spin - boson environment consists of a small number of quantum - coherent two - level fluctuators ( tlfs ) damped by independent bosonic baths .
a master equation of the lindblad form is derived for the probe - plus - tlf system .
we discuss how correlation measurements in the probe system encode information about the environment structure and could be exploited to efficiently discriminate between different experimental preparation techniques , with particular focus on the quantum correlations ( entanglement ) that build up in the probe as a result of the tlf - mediated interaction .
we also investigate the harmful effects of the composite spin - boson environment on initially prepared entangled bipartite qubit states of the probe and on entangling gate operations .
our results offer insights in the area of quantum computation using superconducting devices , where defects / impurities are believed to be a major source of decoherence . |
the discovery of nucleon spin puzzle created the field of high - energy spin physics .
so far , the studies have been focused on spin-1/2 nucleon .
it is known that there exist additional structure functions @xmath0 for a spin-1 hadron in charged - lepton deep inelastic scattering due to the existence of tensor structure @xcite .
there are some theoretical studies on the tensor - polarized structure functions , and the first measurement of @xmath1 was reported by the hermes collaboration @xcite .
a useful parametrization was proposed for the tensor - polarized parton distribution functions ( pdfs ) from the hermes data @xcite by using a constraint for the @xmath1 sum proposed for the tensor - polarized valence - quark distributions @xcite .
although the @xmath1 data are not still accurate , the measurement indicates interesting features different from conventional theoretical calculations . as the deviation from the naive - quark - model prediction created the topic of nucleon spin puzzle , it is important to show a conventional theoretical estimate with the standard deuteron model since a possible experimental deviation may indicate a
tensor - polarization puzzle " .
as such a standard " model of the deuteron for describing the twist - two structure function @xmath1 , we use a convolution picture .
namely , the tensor - polarized structure function @xmath1 is given by unpolarized structure functions convoluted with tensor - polarized lightcone momentum distributions of the nucleon .
there are such studies for @xmath1 in refs .
we try to test their results by independent ways .
we use two theoretical descriptions .
one is a basic convolution model and another is the virtual - nucleon - approximation model .
such convolution models have been used also as a baseline calculation for the nuclear structure function @xmath3 at medium and large @xmath2 in which nuclear effects were taken into account through the binding and nucleon fermi motion in a nucleus together with short - range correlations .
these physics ingredients are contained in the spectral function , which is the nucleon s four - momentum distribution in the nucleus .
we directly apply such descriptions for calculating the structure function @xmath1 .
if the calculated @xmath1 distribution is much different from the hermes measurement , a new hadron physics mechanism could be considered beyond the standard model of the deuteron in describing the tensor structure in terms of quark and gluon degrees of freedom @xcite .
the purpose of our research is to show the convolution result on @xmath1 as the standard theoretical estimate and to discuss its comparison with the hermes data @xcite .
in particular , a new accurate measurement of @xmath1 will start in a few years at jlab @xcite and a possible drell - yan experiment is considered at fermilab with a tensor - polarized deuteron @xcite .
furthermore , other experiments are possible in principle for the tensor - polarized structure functions at bnl , eic , j - parc , gsi - fair , and ihep@russia .
the cross section for deep inelastic scattering of charged lepton from a spin-1 hadron is given by a hadron tensor multiplied by a lepton tensor , and the hadron tensor is expressed in terms of eight structure functions as @xcite @xmath5 \ , | \ , p , \lambda \
, \rangle \nonumber \\[-0.15 cm ] = & -f_1 \hat{g}_{\mu \nu } + \frac{f_2}{m \nu } \hat{p}_\mu \hat{p}_\nu + \frac{ig_1}{\nu}\epsilon_{\mu \nu \lambda \sigma } q^\lambda s^\sigma \nonumber \\[-0.15 cm ] & + \frac{i g_2}{m \nu ^2}\epsilon_{\mu \nu \lambda \sigma } q^\lambda ( p \cdot q \
, s^\sigma - s \cdot q \ , p^\sigma ) \notag \\[-0.15 cm ] & -b_1 r_{\mu \nu } + \frac{1}{6 } b_2 ( s_{\mu \nu } + t_{\mu \nu } + u_{\mu \nu } ) \notag \\[-0.15 cm ] & + \frac{1}{2 } b_3 ( s_{\mu \nu } -u_{\mu \nu } ) + \frac{1}{2 } b_4 ( s_{\mu \nu } -t_{\mu \nu } ) .
\label{eqn : w-1}\end{aligned}\ ] ] the tensor - polarized structure functions are @xmath0 , which do not exist for the spin-1/2 nucleon . here , @xmath6 and @xmath7 are defined as @xmath8 and @xmath9 to satisfy the current conservation @xmath10 .
the coefficients @xmath11 , @xmath12 , @xmath13 , and @xmath14 are defined by the spin-1 polarization vector , hadron and virtual - photon momenta ( @xmath15 , @xmath16 ) , and initial and final spin states ( @xmath17 , @xmath18 ) , and their actual expressions should be found in refs.@xcite .
the structure functions @xmath19 and @xmath20 are twist-4 , and @xmath1 and @xmath21 are leading - twist functions which are related to each other by the callan - gross type relation @xmath22 in the bjorken scaling limit .
we may first investigate the leading function @xmath1 ( or @xmath21 ) .
the structure function @xmath1 is expressed in terms of the tensor - polarized pdfs @xmath23 in the parton model as @xmath24 , \nonumber \\ \ ! \ !
\delta_{_t } f ( x , q^2 ) & \equiv f^0 ( x , q^2 ) - \frac{f^{+1 } ( x , q^2 ) + f^{-1 } ( x , q^2)}{2}. \label{eqn : b1-parton}\end{aligned}\ ] ] here , @xmath25 is an unpolarized parton distribution in the hadron spin state @xmath17 and @xmath26 is the charge of the quark flavor @xmath27 .
the bjorken scaling variable @xmath2 is defined as @xmath28 with the nucleon mass @xmath29 and @xmath30 , and its range is given by @xmath31 for the deuteron .
there is a sum rule for @xmath1 @xcite in the similar way with the gottfried sum rule @xcite : @xmath32 , \nonumber \\
\int \frac{dx}{x } \ , [ f_2^p & ( x ) - f_2^n ( x ) ] = \frac{1}{3 } + \frac{2}{3 } \int dx \ , [ \bar u(x ) - \bar d(x ) ] , \label{eqn : b1-sum - gottfried}\end{aligned}\ ] ] where @xmath33 is the electric quadrupole form factor for the spin-1 hadron and the first term , which comes from tensor - polarized valence - quark distributions , vanishes : @xmath34 .
it could be used as a guideline in investigating @xmath1 . as the gottfried - sum - rule violation initiated the studies of @xmath35 @xcite
, a finite @xmath1 sum could indicate tensor - polarized antiquark distributions .
in fact , a finite @xmath1 sum was suggested in the hermes experiment , and it is interesting to measure the tensor - polarized antiquark distributions in the fermilab - e1039 drell - yan experiment @xcite . for calculating @xmath1 in the standard deuteron model with d - state admixture , we introduce two convolution models .
( a ) one is a basic convolution description and ( b ) another is a virtual nucleon approximation which includes higher - twist contributions . in the convolution description of nuclear structure functions
@xmath36 , the hadron tensor is given by the nucleonic one @xmath37 convoluted with the momentum distribution of the nucleon , so called spectral function @xmath38 , as @xmath39 where @xmath40 and @xmath41 are momenta for the nucleon and nucleus , and @xmath42 is the momentum - space wave function for the @xmath27-th nucleon .
it explains major features of nucleon modifications at medium and large @xmath2 ( @xmath43 ) by the mechanisms of nuclear binding , fermi motion , and short - range correlations contained in the spectral function .
the hadron tensor for the deuteron can be expressed by their helicity amplitudes of the virtual photon as @xcite @xmath44 and the corresponding one @xmath45 for the nucleon . here
, @xmath46 is the photon polarization vector .
the structure function @xmath1 of the deuteron and @xmath47 of the nucleon ( @xmath48 ) are expressed as @xcite @xmath49 in the bjorken scaling limit . using these equations
, we obtain a convolution expression for @xmath1 defined by the one per nucleon as @xmath50 the lightcone momentum distribution is expressed by the deuteron wave function @xmath51 as @xmath52 the variable @xmath53 is the momentum fraction defined by @xmath54 where the light - cone momentum is defined by @xmath55 by taking @xmath56-axis along the virtual - photon momentum direction . using the deuteron wave function with the d - state admixture , we obtain @xmath57 \nonumber \\ & \times ( 3 \cos^2 \theta -1 ) \ , \delta \left ( y - \frac{p\cdot q}{m_n \nu } \right ) , \label{eqn : delta - t - f}\end{aligned}\ ] ] where @xmath58 and @xmath59 are s- and d - state wave functions . according to this basic convolution model , the structure function @xmath1 arises due to the d - state admixture . for calculating eq.([eqn : b1-convolution ] ) , we need the structure function @xmath48 . in our work ,
we use the leading - order ( lo ) expression with the longitudinal - transverse ratio @xmath60 as @xmath61 } \ , f_2^n ( x , q^2 ) , \nonumber \\
f_2^n ( x , q^2)_{\text{lo } } & = x \sum_i e_i^2 \left [ q_i ( x , q^2 ) + \bar q_i ( x , q^2 ) \right ] _ { \text{lo } } .
\label{eqn : f1-lo}\end{aligned}\ ] ] the cross section for the charged - lepton dis from the polarized deuteron is expressed by the polarization factors and structure functions : @xmath62\ , , \label{eq : cross } \vspace{-0.15cm}\end{aligned}\ ] ] where @xmath63 is the degree of the longitudinal polarization of the virtual photon , the details on the longitudinal and transverse polarization factors ( @xmath64 , @xmath65 , @xmath66 ) and the angles ( @xmath67 , @xmath68 ) should be found in ref.@xcite . among the structure functions , @xmath69 and @xmath70
are related to @xmath1 and they are expressed by the helicity amplitudes and tensor - polarized structure functions @xmath71 as @xmath72 & = -\frac{1}{x}\sqrt{\frac{2}{3}}\left[2(1+\gamma^2)xb_1-\gamma^2\left(\frac{1}{6 } b_2-\frac{1}{2}b_3\right)\right ] , \nonumber \\
f_{ut_{tt}}^{\cos(2\phi_{t_\perp } ) } & = -\sqrt{\frac{2}{3 } } \re e a_{+-,-+ } = -\sqrt{\frac{2}{3}}\frac{\gamma^2}{x } \left(\frac{1}{6}b_2-\frac{1}{2}b_3 \right ) , \label{eqn : futll - lt } \vspace{-0.00cm}\end{aligned}\ ] ] where @xmath73 . from these equations
, @xmath1 is expressed by these two structure functions as @xmath74 .
\label{eq : b1_f } \vspace{-0.00cm}\ ] ] now , we explain the virtual nucleon approximation ( vna ) for calculating @xmath1 .
it considers the @xmath75 component of the light - front deuteron wave function .
the virtual photon interacts with one nucleon which is off the mass shell , while the second non - interacting `` spectator '' is assumed to be on its mass shell .
then , the structure functions are obtained by integrating over all possible spectator momenta @xmath76 : @xmath77 where @xmath78 is the phase space for the spectator nucleon .
the factor @xmath79 comes from the definition of deuteron lightcone wave function , and the factor @xmath80 exists because the hadron tensor @xmath37 is for the nucleon with momentum @xmath81 instead of the nucleon at rest @xcite .
the lightcone momentum fractions are defined for the interacting ( @xmath27 ) and spectator ( @xmath82 ) nucleons as @xmath83 and @xmath84 . the deuteron density @xmath85
is defined by the deuteron wave function @xmath86 as @xmath87^\dagger \psi^{d}_\lambda(\vec k,\lambda_{n}',\,\lambda_{n})}{\alpha_{_n } \ , \alpha_i } .
\label{eqn : rho - d } \vspace{-0.2cm}\ ] ] the wave function @xmath88 is then expressed by the s- and d - wave components @xmath89 and @xmath90 . calculating the structure functions in eq.([eqn : futll - lt ] ) and using the relation of eq.([eq : b1_f ] ) ,
we finally obtain @xmath1 in the vna model : @xmath91 & \times \bigg [ f_{1}^n(x_i , q^2 ) \left(6\cos^2\theta_k-2\right ) \nonumber \\[-0.18 cm ] & \ \ \ \ \ -\frac{t^2 } { 2 \ , p_i \cdot q } f _ { 2 } ^n ( x_i , q^2 ) \left(5\cos^2\theta_k-1\right ) \bigg ] \nonumber \\[-0.05 cm ] & \times \left [ - \frac{\phi_0 ( k ) \phi_2(k)}{\sqrt{2}}+\frac{\phi_2 ( k)^2}{4}\right ] , \label{eq : b1_vna}\end{aligned}\ ] ] where @xmath92 is the angle between @xmath93 and @xmath94 , and @xmath95 is defined by @xmath96 with @xmath97 . as obvious from the above derivation , the @xmath1 of the vna model includes higher - twist contributions , whereas the first basic model of eq.([eqn : b1-convolution ] ) was obtained by using the relation ( [ eqn : a - b1f1 ] ) in the scaling limit .
we show results on @xmath1 by integrating the expressions in eqs.([eqn : b1-convolution ] ) and ( [ eq : b1_vna ] ) . for this numerical evaluation
, we choose the pdfs for calculating @xmath98 , the longitudinal - transverse ratio @xmath60 , and the deuteron wave function .
they are taken as mstw2008 ( martin - stirling - thorne - watt , 2008 ) leading - order ( lo ) parametrization , the slac - r1998 parametrization , and the cd - bonn wave function , respectively . since the average scale of the hermes measurement is @xmath99 gev@xmath100
, we show our result at this @xmath101 . in eqs.([eqn : b1-convolution ] ) and ( [ eq : b1_vna ] ) , there are two components , @xmath102 and @xmath103 , which are called sd and dd terms .
these contributions are shown in fig.[fig : xb1-sd - dd ] with total @xmath1 curves for the two theoretical descriptions .
the order of magnitude is rather small and the distributions are less than @xmath104 .
we notice that these results are very different from previous convolution estimates of refs.@xcite in the following points . *
although theoretical formalisms are similar , our distributions , namely magnitude and @xmath2 dependence , are very different .
especially , the sd curves have opposite sign to the one in ref.@xcite . *
the finite distributions exist at large @xmath2 , even at @xmath105 although there is no distribution in ref.@xcite .
there are also relatively large differences between two theory results .
we checked that both results become similar in the scaling limit , which indicates that higher - twist effects are the major sources for the differences .
in addition , there are effects coming from slightly different normalizations for the lightcone wave functions @xcite .
calculated by the two convolution descriptions at @xmath101=2.5 gev@xmath100 .
the contributions from the sd term , dd term , and their summation are shown @xcite.,width=226 ] gev@xmath100 are compared with hermes experimental data @xcite.,width=226 ] next , our results are compared with the hermes data in fig.[fig : xb1-hermes ] .
it is obvious that theoretical curves are much different from the data . in the measured @xmath2 range ( @xmath106 ) ,
the experimental magnitude is one - order larger than both theoretical estimates .
furthermore , there are relatively large distributions even at large @xmath2 ( @xmath107 ) . because the hermes errors are large , we can not draw a solid conclusion from this comparison . however , the large differences indicate that possibly a new hadron physics mechanism could be needed for their interpretation , although there are still some rooms to improve , for example , by considering higher - twist effects in extracting @xmath1 experimentally from the spin asymmetry @xmath108 as pointed out in ref.@xcite
. it is puzzling to find the large differences between the data and our standard convolution descriptions . in future ,
the jlab experiment will start in a few years to measure accurately @xmath1 at medium @xmath2 ( @xmath109 ) , and there is a possibly to measure the proton - deuteron drell - yan process in the fermilab - e1039 drell - yan experiment @xcite .
therefore , such a puzzle should be clarified by future studies ; however , further theoretical studies are needed to clarify the situation and to consider a new mechanism to explain the hermes data .
possibly , a new hadron spin field could be explored by such studies .
we calculated the tensor - polarized structure function @xmath1 by using the standard deuteron model with d - state admixture and the two convolution models .
we found that our @xmath1 values are much smaller in magnitude than the hermes data in the range @xmath106 .
it could indicate possible existence of a new hadron mechanism for interpreting the large differences , although other contributions such as higher - twists should be investigated .
00 l. l. frankfurt and m. i. strikman , nucl .
a * 405 * , 557 ( 1983 ) .
p. hoodbhoy , r. l. jaffe , and a. manohar , nucl .
b * 312 * , 571 ( 1989 ) ; r. l. jaffe and a. manohar , nucl . phys . *
b321 * , 343 ( 1989 ) .
a. airapetian _ et al . _
( hermes collaboration ) , phys .
lett . * 95 * , 242001 ( 2005 ) .
s. kumano , phys . rev .
d * 82 * , 017501 ( 2010 ) .
f. e. close and s. kumano , phys . rev .
d * 42 * , 2377 ( 1990 ) .
h. khan and p. hoodbhoy , phys .
c * 44 * , 1219 ( 1991 ) .
g. a. miller , phys .
c * 89 * , 045203 ( 2014 ) .
w. cosyn , yu - bing dong , s. kumano and m. sargsian , arxiv:1702.05337 .
+ proposal to jefferson lab pac-38 ( pr12 - 11 - 110 ) , j .-
s. hino and s. kumano , phys .
d * 59 * , 094026 ( 1999 ) ; d * 60 * , 054018 ( 1999 ) .
s. kumano and q. t. song , phys .
d * 94 * , 054022 ( 2016 ) .
s. kumano , j. phys .
ser . * 543 * , 012001 ( 2014 ) .
s. kumano , phys . rept . * 303 * , 183 ( 1998 ) ; j .- c . peng and j .- w .
qiu , prog . part .
. phys . * 76 * , 43 ( 2014 ) .
kimura and s. kumano , phys .
d * 78 * , 117505 ( 2008 ) .
w. cosyn , m. sargsian , and c. weiss , to be submitted for publication . | there are polarized structure functions @xmath0 for the spin-1 deuteron .
we calculated the leading - twist tensor structure function @xmath1 by using convolution description for the deuteron .
we found large differences between our theoretical functions and hermes experimental data on @xmath1 .
although higher - twist effects should be considered in obtaining experimental @xmath1 , it suggests a possible existence of new hadron physics mechanism for spin-1 hadrons .
furthermore , we found that there are significant distributions at large bjorken @xmath2 . in future ,
an experimental measurement is planned at jlab for @xmath1 and there is a possibility of a proton - deuteron drell - yan experiment at fermilab with the tensor - polarized deuteron , so that further theoretical studies are needed for clarifying the physics origin of tensor structure in terms of quark and gluon degrees of freedom . |
black holes in general relativity have shown us many fertile aspects of spacetime such as the event horizon , the singularity , hawking radiation , and the connection with thermodynamics . among these phenomena
, hawking radiation is particularly interesting because it gives a strong support to the thermodynamical interpretation of black holes .
moreover , it reveals an aspect of quantum gravity .
usually , hawking radiation is explained by quantizing a free field in a fixed black hole background @xcite . to be specific ,
we consider a real scalar field @xmath0 in the schwarzschild black hole background @xmath1 where @xmath2 is a constant of integration .
when the radius coordinate @xmath3 becomes @xmath2 , the spacetime seems to be singular .
of course , we know it is just a coordinate singularity and the position @xmath4 is the event horizon of the spacetime .
the action for gravity with a scalar field reads @xmath5 where @xmath6 , @xmath7 and @xmath8 are the newton constant , the determinant of the metric @xmath9 and the scalar curvature .
classically , nothing can come out from the inside of the horizon .
quantum mechanically , however , it has been shown that black holes can emit radiation , the so - called hawking radiation @xcite .
remarkably , it turns out that hawking radiation has the planck spectrum .
this notable feature conforms to the black hole thermodynamics .
however , taking look at the above action , we notice that the metric should be also quantized .
then , the event horizon becomes fuzzy , that is , black holes are fluctuating . the purpose of this paper is to study hawking radiation from fluctuating black holes .
technically , we consider the interaction between the scalar field and the metric fluctuations . in particular , it is interesting to see if the planck spectrum is modified by the interaction @xcite whatever small it is .
apparently , this effect becomes significant if the horizon radius of black holes @xmath10 is close to the planck length @xmath11 .
recently , various planck scale black holes and hawking radiation from them have been considered @xcite .
these black holes can be created at the lhc or in the early universe .
indeed , black holes would be created at the lhc if the dimension of space - time is more than six and braneworld picture is correct @xcite .
this is because the higher dimensional planck mass becomes @xmath12tev which can be reached at the lhc . in other words ,
the mass of a black hole created at the lhc is of the order of the higher dimensional planck mass . on the other hand , density fluctuations in the early universe
could produce black holes with any mass .
black holes with @xmath13 g created in the early universe lose their masses by hawking radiation and become planck scale just now .
the spectrum of hawking radiation from these black holes might be distorted due to the nonlinear interaction between quantum fields and the metric . here
, we stress that , even for black holes much larger than the planck length , the fluctuations of black holes might be relevant to hawking radiation .
the point is that the planck spectrum of hawking radiation stems from the exponential elongation of short wavelength quantum fluctuations which can exceed the planck scale .
hence , the interaction between the scalar field and quantum mechanical fluctuations of the metric would be relevant even for the large black holes . in this paper
, we calculate the effect of nonlinear interaction on hawking radiation of a real scalar field from a 4-dimensional schwarzschild black hole . in order to achieve this aim ,
the canonically normalized quadratic action for metric perturbations and the interaction hamiltonian are needed .
therefore , we first construct these actions by perturbing the einstein - hilbert action with a real scalar field .
after that , we give a formalism to study hawking radiation from fluctuating black holes .
the organization of this paper is as follows . in section
[ 3 ] , we construct the quadratic action for metric perturbations by perturbing einstein - hilbert action in 4-dimensional schwarzschild background . in section [ 2 ] , we quantize the field and derive hawking radiation . in section
[ 4 ] , we present a method for treating interaction in spherical symmetric space - time and then calculate the cubic interaction hamiltonian from the action of a real scalar field . using these results , we calculate the effect of nonlinear interaction on hawking radiation of a real scalar field . section [ 5 ] is devoted to the conclusion .
technical details can be found in appendixes . in this paper
, we use the convention @xmath14 and write the schwarzschild metric as @xmath15 where @xmath16 represents the metric of the sphere @xmath17 and @xmath18 corresponds to the mass of the black hole .
in this section , we derive the quadratic action for metric perturbations @xmath19 in the schwarzschild black hole background @xmath20 .
note that our derivation is slightly different from the original one by moncrief @xcite .
now , the metric is given by @xmath21 here , we can use the general coordinate invariance to reduce the number of variables . under infinitesimal coordinate transformations
@xmath22 , @xmath19 transforms as @xmath23 which shows that there are 4 gauge degrees of freedom .
hence , we can fix 4 variables .
moreover , we can classify the remaining 6 variables into 4 scalar type and 2 vector type ones .
once the gauge is completely fixed , we can obtain the quadratic action . after eliminating unphysical variables using constraint equations , we finally obtain the reduced action for physical variables . it is easy to extend the analysis in this section to more general cases . in appendix
[ appendixa ] , we present the analysis for topological schwarzschild black holes with a cosmological constant . first , we consider the scalar type perturbations @xmath24 where @xmath25 denotes a covariant derivative with respect to @xmath16 . under the scalar type gauge transformations
@xmath26 with @xmath27 the perturbed variables change as @xmath28 where @xmath29 and @xmath30 denote derivatives with respect to @xmath3 and @xmath31 , respectively . then ,
choosing @xmath32 , @xmath33 and @xmath34 , we get the following metric perturbations @xmath35 using the above gauge , we obtain the quadratic action @xmath36+(c.c ) \biggr\}\nonumber\\ & & \hspace{1 cm } + \sum_{\ell } \int dr dt\biggl[\frac{\gamma_s}{2}h_{1\ell 0}{}^2 -\gamma_s{\dot w}_{\ell 0}h_{1\ell 0}+ \frac{\gamma_s}{2}{\dot w}_{\ell
0}^2 + 2r\dot{h}_{\ell 0}h_1{}_{\ell 0 } \nonumber\\ & \ & \hspace{4.5cm}-\gamma_s\frac{2f+rf^{'}}{2r}w_{\ell 0}h_{\ell 0}+\frac{(fr)^{'}}{2}h_{\ell
0}^2+\gamma_s\frac{f}{r^2}w_{\ell 0}^2 -\frac{1}{2}{\bar h}_{\ell 0}q_2{}_{\ell 0}\biggr ] \ , \label{even_ac}\end{aligned}\ ] ] where @xmath37 means complex conjugate and we have defined @xmath38 it is easy to solve as @xmath39 where we have defined @xmath40 . in eqs.([even_ac ] ) , ( [ q ] ) , and ( [ wh ] ) , we expanded metric perturbations in terms of spherical harmonics @xmath41 like as @xmath42 . here , @xmath43 .
since spherical harmonics satisfy the relation @xmath44 , metric perturbations @xmath45 satisfy reality conditions @xmath46 .
this allows us to relate negative @xmath47 modes to positive @xmath47 modes .
hence , we can consider only @xmath48 modes in ( [ even_ac ] ) .
note that @xmath49 mode is real but @xmath50 modes are complex . from now on ,
we concentrate on @xmath49 mode and omit the indices .
we can get the same result for @xmath50 modes with similar calculations . taking the variation of the quadratic action ( [ even_ac ] ) with respect to @xmath51
, we obtain the momentum constraint equation @xmath52 this equation enables us to delete @xmath51 from the action ( [ even_ac ] ) . substituting this result ( [ even_constraint ] ) into the action ( [ even_ac ] )
, we obtain the action @xmath53\ .\label{action2}\end{aligned}\ ] ] taking the variation of the action ( [ action2 ] ) with respect to the lagrange multiplier @xmath54 , we can derive the hamiltonian constraint @xmath55 then , substituting @xmath56 into the action ( [ action2 ] ) and using the relations ( [ wh ] ) , we can express the action by @xmath57 . with a new variable @xmath58 defined by @xmath59 the quadratic action takes the canonical form
@xmath60 \label{}\end{aligned}\ ] ] where @xmath61 represents a tortoise coordinate and we have defined the potential function as @xmath62\nonumber\\ & = & \frac{2f}{r^3}\frac{\lambda^2(\lambda+1)r^3 + 3\lambda^2mr^2 + 9\lambda rm^2 + 9m^3}{(r\lambda+3m)^2 } \ .\label{zerilli_potential}\end{aligned}\ ] ] we can also calculate the quadratic actions for @xmath50 modes .
thus , we have @xmath63\right\ } + \frac{1}{2}\int dr^ { * } dt \left[(\partial_t \psi^{z}_{\ell 0})^2-(\partial_{r^ { * } } \psi^{z}_{\ell 0})^2-v_{z}(r ) \psi^{z}_{\ell 0}{}^2\right ] \label{even_action } \ , \end{aligned}\ ] ] which gives the famous zerilli equation@xcite @xmath64 for later purposes , we deduce the relation between @xmath65 and metric perturbations . from eqs.([q ] ) , ( [ wh ] ) , ( [ even_constraint ] ) and ( [ psiz ] ) , it is easy to get the relation between @xmath65 and metric perturbations except for @xmath54 .
the results are @xmath66 in order to get the relation between @xmath67 and @xmath68 , we need to look at the equation of motion for metric perturbations . taking the variation of ( [ action2 ] ) with respect to @xmath69 and @xmath70
, we can get two equations @xmath71 eliminating @xmath72 from the above equations , we obtain the equation for @xmath68 @xmath73\ . \end{aligned}\ ] ] using this equation , the relation ( [ even_master1 ] ) and zerilli equation , we can show that @xmath68 is related to @xmath45 and @xmath74 as @xmath75 since @xmath45 and @xmath74 are related to @xmath76 by eqs.([even_master1 ] ) , this equation implies a relation between @xmath68 and @xmath76 .
next , we consider the quadratic action for vector type perturbations @xmath77}\\
\end{array } \right ) \ , \label{odd_pert}\end{aligned}\ ] ] where @xmath78 , @xmath79 and @xmath80 satisfy @xmath81 . for vector type perturbations ( [ odd_pert ] ) , the gauge transformations with @xmath82 read @xmath83 where @xmath84 satisfies the transverse condition .
then , we can completely fix the gauge by choosing @xmath85 , which is called regge - wheeler gauge . in the regge - wheeler gauge ,
metric perturbations are written as @xmath86 using the metric perturbations ( [ odd_p ] ) , we can calculate the quadratic action @xmath87+(c.c)\right\}\nonumber\\ & \ & \hspace{1cm}+(2\lambda+2 ) \int drdt \bigl[\frac{1}{2}\left(\dot{w}_{\ell 0}+\frac{2}{r}v_{\ell 0}-v^{'}_{\ell 0}\right)^2+\frac{\lambda}{fr^2}v^2_{\ell 0}-\frac{\lambda f}{r^2}w^2_{\ell 0}\bigr]\ , \label{odd_ac}\end{aligned}\ ] ] where metric perturbations are expanded by vector harmonics @xmath88 like as @xmath89 . since vector harmonics satisfy relations @xmath90 , @xmath74 satisfy reality conditions @xmath91 .
hence , we can consider only @xmath48 modes in ( [ odd_ac ] ) .
for a while , we concentrate on @xmath49 mode and omit the indices . taking the variation of the action ( [ odd_ac ] ) with respect to @xmath92
, we can derive momentum constraint equation @xmath93^{'}+\frac{2\lambda}{f}v=0 \ .\label{odd_const}\end{aligned}\ ] ] we have to solve this constraint equation .
however , it is not easy to do so . in order to circumvent this difficulty
, we move on to hamiltonian formalism . from the quadratic action ( [ odd_ac ] )
, we can read off the lagrangian @xmath94\ , \label{}\end{aligned}\ ] ] and also define the momentum conjugate to @xmath70 as @xmath95 then , the hamiltonian @xmath69 is given by @xmath96 \ . \label{odd_ham}\end{aligned}\ ] ] now , the constraint equation @xmath97 gives the relation @xmath98 substituting this relation into the hamiltonian ( [ odd_ham ] ) , we obtain the reduced hamiltonian @xmath99 \ . \label{hamiltonian}\end{aligned}\ ] ] we further perform the canonical transformation @xmath100 to interchange the role of @xmath70 and @xmath101 @xmath102 which leads to a new hamiltonian @xmath103 \ . \nonumber\end{aligned}\ ] ] using the hamilton equation for @xmath104 @xmath105 we can go back to the lagrangian formalism @xmath106
= \frac{1}{2\lambda+2}\int dr dt\biggl[\frac{r^2}{4\lambda f}\dot{q}^2-\left(\frac{1}{2}-\frac{(fr)^{'}}{2\lambda}+\frac{f}{\lambda}\right)q^2-\frac{fr^2}{4\lambda}q^{'}{}^{2}\biggr ] \ .\label{odd_act}\end{aligned}\ ] ] with a change of normalization @xmath107 the quadratic action ( [ odd_act ] ) becomes @xmath108 \ , \end{aligned}\ ] ] where @xmath61 represents a tortoise coordinate and we have defined the potential function @xmath109 so far we have only considered the @xmath49 mode in ( [ odd_ac ] ) . of course , we can also calculate the action for @xmath50 modes .
the result reads @xmath110\biggr\ } \nonumber\\ & & \qquad + \frac{1}{2}\int dr^{*}dt\left[\left(\partial_t \psi^{rw}_{\ell 0}\right)^2-\left(\partial_{r^ * } \psi^{rw}_{\ell 0}\right)^2-v_{rw}(r)\psi^{rw}_{\ell 0}{}^2\right ] \ , \label{eq : rwaction}\end{aligned}\ ] ] which gives the regge - wheeler equation @xcite @xmath111 finally , we clarify the relation between @xmath112 and metric perturbations . from eqs.([sol_con ] ) , ( [ qp ] ) , ( [ dotq ] ) and ( [ psirw ] ) , the relations are given by @xmath113
in this section , we review hawking radiation using a spherically symmetric collapsing model@xcite . in this model , it is assumed that outside of the collapsing matter can be described by the schwarzschild metric @xmath114 where @xmath115 . here ,
@xmath18 and @xmath16 are the adm mass and the metric of @xmath17 , respectively .
the action for the real massless scalar field @xmath116 can be written as @xmath117\nonumber\\ & \ & + \frac{1}{2}\int dr^*dt\left [ \left(\partial_t \phi_{\ell 0}\right)^2 - \left(\partial_{r^*}\phi_{\ell 0}\right)^{2}-f\left(\frac{f^{'}}{r}+\frac{\gamma_s}{r^2}\right)\phi_{\ell 0}^2\right]\ , \label{action_phi}\end{aligned}\ ] ] where @xmath61 is a tortoise coordinate . here , we used the harmonic expansion @xmath118 . note that @xmath119 is real . hence , complex coefficients @xmath120 satisfy reality conditions @xmath121 .
then , the equation of motion outside the matter @xmath122 can be written as @xmath123 it should be noted that the quantization of metric perturbations can be performed in the same way as that of the scalar field explained below .
let us quantize the scalar field @xmath119 by promoting @xmath119 to the operator : @xmath124 \ . \label{decompose_var}\end{aligned}\ ] ] in the above formula , @xmath125 are complete orthonormal basis satisfying klein - gordon equation @xmath126 note that this equation is independent of @xmath47 , so we used @xmath127 instead of @xmath128 . here
, the norm is defined by the inner product @xmath129 where @xmath130 is a cauchy surface and @xmath131 means area element times a unit normal vector .
now , the canonical commutation relation implies the following commutation relations ; @xmath132 = \delta ( \omega - \omega^ { ' } ) \
\delta_{\ell,\ell^{'}}\delta_{m , m^ { ' } } \ . \label{com}\end{aligned}\ ] ] using the property of spherical harmonics @xmath44 , we can rewrite ( [ decompose_var ] ) as @xmath133\frac{y_{\ell m}}{r } \ .\label{decompose2}\end{aligned}\ ] ] let us define @xmath120 by @xmath134 .
then , we have @xmath135 \ .\end{aligned}\ ] ] from eq.([decompose2 ] ) , we see the following relations @xmath136 these relations are equivalent to @xmath137 which is nothing but the reality condition for @xmath119 .
and this reality condition suggests that we can only consider @xmath48 modes . from ( [ com ] ) and ( [ abd ] )
, we can deduce the commutation relations for @xmath138 and @xmath139 for @xmath48 modes .
the results are as follows .
* @xmath49 modes + we can define @xmath140 from the relations ( [ abd ] ) . using the commutation relations ( [ com ] )
, we can show that @xmath141 = \delta ( \omega - \omega^ { ' } ) \ \delta_{\ell,\ell^ { ' } } \ . \end{aligned}\ ] ] the commutation relations show that @xmath142 are real scalar fields whose annihilation operators are @xmath143 .
note that these scalar fields @xmath142 have the action ( [ action_phi ] ) . * @xmath50 modes + from the commutation relations ( [ com ] ) , it is easy to show @xmath144 = \delta ( \omega - \omega^{'})\ \delta_{\ell,\ell^{'}}\delta_{m , m^ { ' } } \ , \quad [ d_{\omega \ell m},d_{\omega^ { ' } \ell^ { ' } m^{'}}^{\dagger } ] = \delta ( \omega - \omega^ { ' } ) \
\delta_{\ell,\ell^{'}}\delta_{m , m^ { ' } } \ . \end{aligned}\ ] ] thus , @xmath120 @xmath145 are complex scalar fields whose annihilation operators are @xmath146 and @xmath147 .
note that complex scalar fields @xmath120 have the action ( [ action_phi ] ) . to introduce the bogolubov transformation ,
we define complete orthonormal basis @xmath148 suppose that there are two kind of complete orthonormal basis of @xmath119 , @xmath149 and @xmath150 . as both @xmath151 and @xmath152 are complete basis , we can expand @xmath119 in two ways @xmath153 where @xmath154 and @xmath155 are annihilation operators for @xmath156 and @xmath157 , respectively . using these annihilation operators , we can define two kind of vacua as @xmath158 due to the completeness of @xmath151 and @xmath152
, we can expand @xmath159 as @xmath160 or conversely @xmath161 where @xmath162 and @xmath163 are called bogolubov coefficients . from the above formula
, we can deduce the bogolubov coefficients as @xmath164 since both complete basis are proportional to the spherical harmonics @xmath41 , bogolubov coefficients are proportional to @xmath165 and @xmath166 , that is @xmath167 substituting ( [ v_barv ] ) and ( [ barv_v ] ) into ( [ tennkai ] ) and using ( [ angular ] ) , we can get bogolubov transformations between @xmath154 and @xmath155 as follows ; @xmath168 furthermore , substituting the second relation in ( [ a_bara ] ) into the first relation in ( [ a_bara ] ) , we see that bogolubov coefficients satisfy @xmath169 so far , we have considered bogolubov transformation for @xmath119 .
we can also describe bogolubov transformation in terms of @xmath170 . in order to get bogolubov transformation for @xmath171 which is defined by @xmath172
, we set @xmath49 in ( [ a_bara ] ) .
then , we get @xmath173 similarly , setting @xmath50 in ( [ a_bara ] ) , we have bogolubov transformation for @xmath146 which is defined in ( [ abd ] ) ; @xmath174 and we obtain bogolubov transformation for @xmath147 by considering @xmath175 in ( [ a_bara ] ) @xmath176 finally , we calculate particles created due to the time evolution of space - time . it is assumed that @xmath150 is a natural base at the initial time ( in region " ) and @xmath151 is natural at the final time ( out region " ) .
furthermore , we assume that the initial state is @xmath177 . note that this state does not change because we use a heisenberg picture .
under these assumptions , the number spectrum of @xmath119 particles at the final time can be expressed by @xmath178 \ . \end{aligned}\ ] ] then , using bogolubov transformations for @xmath146 , @xmath171 and @xmath147 , we can calculate the number spectrum @xmath179 \ . \label{harm_beta}\end{aligned}\ ] ] this shows that we need @xmath180 defined in ( [ beta ] ) and ( [ angular ] ) in order to know the number spectrum .
let us consider hawking radiation by using spherically symmetric collapsing model described by the penrose diagram in fig.[fig1 ] . in this figure
, @xmath181 corresponds to in region " and @xmath182 represents out region " .
means an affine parameter , width=226 ] in order to calculate @xmath183 , we must define natural basis in in region " and out region " .
we assume that outside of the collapsing matter is described by the schwarzschild metric .
then , @xmath120 satisfy @xmath184 in this region . here
, @xmath185 is ignored because @xmath185 dumps as @xmath186 . noticing ( [ asymptotic ] )
, we choose the natural basis @xmath187 and @xmath188 defined by @xmath189 given the complete basis , we can calculate @xmath180 defined in ( [ beta ] ) and ( [ angular ] ) .
we have to choose @xmath130 to calculate the inner product in the formula ( [ beta ] ) . here
, we take @xmath181 .
hence , it is necessary to know the behavior of @xmath190 near @xmath181 . for that purpose ,
we use geometric optics approximation where the worldline of a particle is a null ray of a constant phase and trace this null line backward to @xmath181 .
this approximation is quite good near the horizon @xmath191 because the frequency effectively becomes very large in this region .
hereafter , we write @xmath192 and @xmath193 . furthermore ,
we define @xmath194 as an affine parameter with respect to @xmath195 which satisfies @xmath196 on @xmath197 . near @xmath198
, this parameter can be expressed as @xmath199 where @xmath200 is the surface gravity and @xmath201 and @xmath202 are some constants .
because @xmath203 line is also null line , geometric optics approximation suggests that @xmath203 surface is equivalent to a constant phase surface .
then , the constant phase surface @xmath204 satisfies @xmath205 and so @xmath206 this means that @xmath190 approximately behaves as @xmath207 near @xmath181 .
then , we can calculate the inner products in ( [ beta ] ) using ( [ uout ] ) .
note that ( [ uout ] ) is a good approximation in @xmath208 but the effect from this region is dominant because the phase of @xmath190 diverges .
therefore it is also a good approximation to use the expression ( [ uout ] ) on @xmath181 in calculating inner products .
the results of calculations are @xmath209 and @xmath210 this gives @xmath211 derived from ( [ property ] ) , we obtain bogolubov coefficients @xmath212 @xmath213 substituting this result into ( [ harm_beta ] ) , we find that the observer in the asymptotic region see the thermal spectrum of @xmath119 particles
. actually , eq.([mondai ] ) diverges and is proportional to the delta function @xmath214 .
we interpret this as @xmath215 and introduce the regularization @xmath216 hereafter .
namely , what we obtained is the number of particles created per frequency and per unit time . in the above formula
, we have discarded this factor .
however , we should keep in mind this fact in the following calculations .
in this section , we consider interaction between the scalar field @xmath0 and gravitons @xmath217 .
first , we introduce a method for treating interaction in the spherically symmetric space - time @xcite .
next , we calculate cubic interaction terms from the action ( [ scalar_action ] ) . finally , we evaluate the effect of interaction on hawking radiation . here ,
we follow the notations in sec.[2 ] , namely , we use @xmath120 defined by @xmath134 and annihilation operators @xmath146 , @xmath171 , @xmath147 . let us consider the lagrangian @xmath218 where @xmath219 represents the free field part of the action and @xmath220 is the interaction part .
furthermore , we assume that @xmath220 vanish both in the remote past ( or in region " ) and in the remote future ( or out region " ) . under these assumptions
, we can define asymptotic fields @xmath221 where @xmath222 and @xmath223 satisfy equation of motion derived from @xmath224 .
therefore , using basis @xmath225 in in region " and @xmath226 in out region " , we can expand asymptotic fields @xmath222 as @xmath227 and @xmath223 as @xmath228 from the above expansion , we can define four kind of vacua @xmath229 here , @xmath230 is a natural vacuum in in region " , while @xmath231 is natural in out region " .
other vacua are introduced merely for the computational purpose .
now we calculate the number of particles created due to the time evolution of space - time in the presence of interaction . given the natural vacuum @xmath230 , the number spectrum of @xmath119 becomes @xmath232\ , \label{keisanshitai}\end{aligned}\ ] ] first , we evaluate @xmath233
. the bogolubov transformations ( [ b_bogo ] ) for @xmath234 read @xmath235 then , @xmath236 can be expressed by @xmath237 \ . \end{aligned}\ ] ] the expectation value of this number operator in the state @xmath230 is given by @xmath238 \ .\label{effect_from_b}\end{aligned}\ ] ] the first term of the right hand side of this equation is the same as ( [ harm_beta ] ) and so this term represents the conventional hawking radiation .
hence , other terms in ( [ effect_from_b ] ) are induced by interaction .
in fact , these terms vanish if there is no interaction .
then , in order to know the effect of interaction on hawking radiation , what we have to do is to calculate these terms . for later purpose
, we decompose these terms into @xmath239 cases and @xmath240 cases . using @xmath241 , for example , @xmath242
$ ] in ( [ effect_from_b ] ) can be written as @xmath243\nonumber\\ & = & \frac{1}{l}\int d\omega^{'}\left[\left|\alpha_{\omega\ell,\omega^{'}\ell}\right|^2\langle 0\ { \rm in}|b_{\omega^{'}\ell m}^{out\dagger } b_{\omega^{'}\ell m}^{out}|0\ { \rm in}\rangle \right]\nonumber\\ & \ &
\ + \int d\omega^ { ' } d\omega^{''}\alpha_{\omega\ell,\omega^{''}\ell}\alpha_{\omega\ell,\omega^{'}\ell}^{*}\biggl[\langle 0\ { \rm in}|b_{\omega^{''}\ell m}^{out\dagger } b_{\omega^{'}\ell m}^{out}|0\ { \rm in}\rangle-\frac{\delta ( \omega^{'}-\omega^{''})}{l}\langle 0\ { \rm in}|b_{\omega^{'}\ell m}^{out\dagger } b_{\omega^{'}\ell m}^{out}|0\ { \rm in}\rangle\biggr]\ .
\label{dec_equal_neq}\end{aligned}\ ] ] the second term in the above equation reads @xmath244 therefore , eq.([dec_equal_neq ] ) can be rewritten as @xmath245 + \int_{\omega^{'}\neq\omega^ { '' } } d\omega^ { ' } d\omega^{''}\left[\alpha_{\omega\ell,\omega^{''}\ell}\alpha_{\omega\ell,\omega^{'}\ell}^{*}\langle 0\ { \rm in}|b_{\omega^{''}\ell m}^{out\dagger } b_{\omega^{'}\ell m}^{out}|0\ { \rm in}\rangle\right]\ .
\label{effect_b_int}\end{aligned}\ ] ] similarly , we can evaluate expectation values @xmath246 and @xmath247 in ( [ keisanshitai ] ) as @xmath248 \
, \nonumber\\ \langle 0\ { \rm in}|{\bar c}_{\omega \ell } ^{out\dagger}{\bar c}^{out}_{\omega \ell } |0\ { \rm in}\rangle & = & \int d\omega ' \left|\beta_{\omega\ell,\omega^{'}\ell}\right|^2 + \int d\omega ' \int d\omega '' \biggl[(\alpha_{\omega\ell,\omega^{''}\ell}\alpha_{\omega\ell,\omega^{'}\ell}^{*}+\beta_{\omega\ell,\omega^{'}\ell}\beta_{\omega\ell,\omega^{''}\ell}^{*})\langle 0\ { \rm in}|c_{\omega^ { '' } \ell}^{out\dagger } c_{\omega^ { ' } \ell}^{out}|0\ { \rm in}\rangle \nonumber\\ & \ & \hspace{4cm}-2\re\left\{\alpha_{\omega\ell,\omega^{''}\ell}\beta_{\omega\ell,\omega^{'}\ell}^{*}\langle 0\ { \rm in}|c_{\omega^{''}\ell } ^{out\dagger } c_{\omega^{'}\ell } ^{out\dagger}|0\ { \rm in}\rangle \right\}\biggr ] \ . \label{jyuuyou}\end{aligned}\ ] ] these results are almost the same as @xmath233 and the terms other than the first term of the above formula have information of interaction . substituting ( [ effect_from_b ] ) and ( [ jyuuyou ] ) into ( [ keisanshitai ] ) , we find the deviation from the planck spectrum due to the interaction as @xmath249 \ . \label{keisanshitai2}\end{aligned}\ ] ] in order to calculate @xmath250 etc .
, it is useful to define the basis @xmath251 which satisfy the completeness relation @xmath252 substituting ( [ completenes ] ) into just after @xmath253 and just before @xmath230 , for example , we can calculate @xmath250 as follows : * if @xmath254 , @xmath255 where @xmath241 . * if @xmath240 , @xmath256\ .
\label{hituyou2}\end{aligned}\ ] ] similar rules apply to @xmath257 and @xmath258 . in the same way , @xmath259 becomes * if @xmath239 , @xmath260 * if @xmath240 , @xmath261\ .
\label{hituyou4}\end{aligned}\ ] ] and also @xmath262 is given by @xmath263 \ .
\label{hituyou5}\end{aligned}\ ] ] these results will be used in the calculation of s - matrix . before moving on to calculations of s - matrix elements and clarify the effect of interaction on hawking radiation , we need to calculate the interaction hamiltonian . in this subsection ,
we derive cubic interaction terms in the following action @xmath264 we put the metric @xmath265 where @xmath20 is the background schwarzschild metric . in the regge - wheeler and the convenient gauge , @xmath19 reads @xmath266 using this gauge , we obtain the cubic action @xmath267\nonumber\\ & & \qquad - \frac{1}{2}\int r^2\sqrt{\gamma}\left[\frac{{\bar h}}{f}{\dot \varphi}^2 + 2h_1\varphi^{'}{\dot \varphi}+\frac{2}{r^2f}v^{a}{\dot \varphi}\varphi_{|a}-fh\varphi^{'2}-\frac{2f}{r^2}\left(w^{|a}+w^{a}\right)\varphi^{'}\varphi_{|a}\right ] \label{all_interaction}\end{aligned}\ ] ] we see that scalar and vector perturbations are decoupled , so we will calculate them separately . in the action ( [ all_interaction ] ) , terms including vector perturbations are @xmath268 \ .\end{aligned}\ ] ] we expand these variables by harmonics ; for example , @xmath269 where @xmath88 are vector harmonics and @xmath41 are spherical harmonics .
the result is @xmath270\ , \label{odd_d}\end{aligned}\ ] ] where @xmath271 it is easy to show the relation @xmath272 using the property @xmath273 .
we also see that @xmath274 are pure imaginary numbers proportional to @xmath275 because both vector harmonics and spherical harmonics are proportional to @xmath276 . furthermore , because @xmath274 are pure imaginary and both harmonics have property such as @xmath277 , @xmath274
, we can derive the following formula , @xmath278 .
note that we omitted @xmath279 in ( [ odd_d ] ) .
thus , using ( [ odd_master ] ) and @xmath280 , we can get cubic interaction terms @xmath281\ . \label{odd_interaction}\end{aligned}\ ] ] terms including scalar perturbations in the action ( [ all_interaction ] ) read @xmath282\nonumber\\ & \ & \hspace{0.5cm}-\frac{1}{2}\int r^2\sqrt{\gamma}\left[\frac{{\bar h}}{f}{\dot \varphi}^2 + 2h_1\varphi^{'}{\dot \varphi}-fh\varphi^{'2}-\frac{2f}{r^2}w^{|a}\varphi^{'}\varphi_{|a}\right]\ .\end{aligned}\ ] ] using expansion by spherical harmonics like @xmath283 we can write the action in the following form @xmath284\ , \label{even_interaction1}\end{aligned}\ ] ] where we have defined @xmath285 these coefficients @xmath286 have almost the same property as @xmath274 .
it is easy to derive the relation @xmath287 .
because of @xmath288 , @xmath286 are real numbers proportional to @xmath275 .
the reality of @xmath286 implies that @xmath289 . using ( [ even_master1 ] ) , ( [ even_master2 ] ) and @xmath280
, we can derive cubic interaction terms for scalar perturbations @xmath290\ , \nonumber\\ \label{even_interaction}\end{aligned}\ ] ] where @xmath291 these function @xmath292 etc .
are regular outside the horizon .
however , the other coupling functions in ( [ even_interaction ] ) are singular at the horizon because these terms are proportional to @xmath293 .
we also have a relation @xmath294 note that we ignored the terms which is related to equation of motion for @xmath119 in ( [ even_interaction ] ) .
this is because these terms do not affect our calculation .
we notice that the cubic interaction hamiltonian have the following structure @xmath295 from the interaction hamiltonian ( [ 7_interaction ] ) , the first term of ( [ odd_interaction ] ) can be reproduced by the replacement @xmath296 note that @xmath47 takes the integer in the range @xmath297 ; @xmath298 and @xmath299 take the value in the same range . as is shown in sec.[2 ] , @xmath120 ( @xmath50 )
are complex scalar fields and @xmath142 are real scalar field . by using reality conditions
, we can rewrite the interaction hamiltonian ( [ 7_interaction ] ) using only @xmath48 modes . from now on , @xmath47 moves in the range @xmath300 unless we specify its range . with this notation , each term for scalar perturbation in ( [ 7_interaction ] )
can be rewritten as @xmath301 in this decomposition , we use the property @xmath286 and reality conditions for @xmath65 and @xmath120 . note that there exist other terms in ( [ c_expand ] ) such as @xmath302 .
however , such terms must vanish due to the fact @xmath303 .
similarly , the terms related to vector perturbations read @xmath304 in this subsection , we calculate the s - matrix elements which constitute the expectation values ( [ hituyou1 ] ) , ( [ hituyou2 ] ) , ( [ hituyou3 ] ) , ( [ hituyou4 ] ) and ( [ hituyou5 ] ) by using the cubic interaction ( [ 7_interaction ] ) .
we here only consider the first order interaction and do not deal with loop diagram like fig.[fig3 ] . under these assumptions
, we consider the s - matrix elements like @xmath305 then , we only consider ( [ hituyou1 ] ) and ( [ hituyou2 ] ) because the others do not include such s - matrix elements .
.,width=188 ] now , we derive the feynman rules . recall that @xmath120 and @xmath142 are written as @xmath306 where @xmath307 satisfy eq.([klein ] ) and the boundary condition in in - region " ( [ in_condition ] )
then , we can deduce the feynman rules as follows : @xmath308 we can also obtain the feynman rules for @xmath309 and @xmath310 . note that @xmath67 and @xmath311 can be expanded as @xmath312 where @xmath313 satisfy the zerilli equation and the boundary condition in in - region " ( [ in_condition ] ) .
then , the feynman rules for @xmath309 become @xmath314 the same rules apply to @xmath310 .
using such rules and eq.([c_expand ] ) , we can perform the following calculations @xmath315 note that only one term in ( [ c_expand ] ) is related to this calculation .
consider another state @xmath316 .
it is easy to obtain @xmath317 where the two terms in ( [ c_expand ] ) are relevant . in the same manner
, we can consider the terms related to @xmath310 .
for the state @xmath318 , we have @xmath319 note that additional sign comes from the antisymmetric property @xmath272 .
these results suggest that it is useful to define @xmath320 in fact , using these notations , all s - matrix elements can be written as ( coefficient of ( [ c_expand]))@xmath321 or ( coefficient of ( [ d_expand]))@xmath322 .
noticing that the effect of the first term of ( [ odd_interaction ] ) on @xmath323 is given by @xmath324 using these notations , we calculate ( [ hituyou1 ] ) and ( [ hituyou2 ] ) in the following .
first , we consider @xmath67 . in the formula ( [ hituyou2 ] ) , we only need the cases @xmath325 and one graviton " or @xmath326 and one graviton " . from ( [ c_expand ] ) , for both cases
, this graviton should be d - particle " .
in the former case , ( [ hituyou2 ] ) becomes @xmath327^{*}\left[\frac{1}{\sqrt{2}}c^{\ell ^{'''}m^{'''}}_{\ell m;\ell m}k^{\omega^{'''}\ell ^{'''}}_{\omega^{'}\ell ; \omega^{'}\ell } \right]\nonumber\\ & = & \frac{1}{l}\int d\omega ' '' \sum_{\ell ^{'''}}\left|c^{\ell ^{'''}2m}_{\ell m;\ell m}\right|^2k^{\omega^{'''}\ell ^{'''}}_{\omega^{'}\ell ; \omega^{''}\ell } { } ^{*}k^{\omega^{'''}\ell ^{'''}}_{\omega^{'}\ell ; \omega^{'}\ell } \ , \end{aligned}\ ] ] where we have introduced the regularization @xmath328 .
here , @xmath329 has the dimension of the length .
note that we used @xmath303 in the last equality . in the above calculation
, we used the formula @xmath330 .
we can get the same result as this for the latter case and adding these results gives @xmath331 we only calculated the case that @xmath120 is b - particle " so far .
the same calculation leads to the result for c - particle " @xmath332 and that for d - particle " @xmath333 taking the summation of @xmath334 and @xmath47 in ( [ keisanshitai2 ] ) , the effect of ( [ hituyou2 ] ) on ( [ keisanshitai2 ] ) is @xmath335 next , we consider @xmath310 .
same calculation leads to almost the same result as ( [ 7_12 ] ) @xmath336 however , because of @xmath337 , this vanishes .
first , we consider @xmath309 . in this case
, we must consider @xmath338 . if @xmath339 , the same calculation as before yields @xmath340 next , we consider @xmath341 . in this case , we must consider another @xmath0-particle and one graviton " .
furthermore , if this additional @xmath0-particle is b - particle " with @xmath342 , we must consider a different mode @xmath343 .
then , we have to calculate the followings @xmath344 note that the rule @xmath345 should be used when we evaluate these quantities . taking look at ( [ c_expand ] ) , we see that the graviton in the first equation of ( [ 128 ] ) must be d - particle " .
the result is given by @xmath346 we can also calculate the second equation of ( [ 128 ] ) .
in this case , the graviton must be d - particle " .
the answer becomes @xmath347 in the last equation of ( [ 128 ] ) , we have to consider all types of gravitons .
more precisely , the sign of @xmath348 determines particle " species .
thus , we obtain @xmath349\nonumber\\ & & \quad = l \int d\omega ' '' \sum_{{\bar \omega},{\bar \ell } , { \bar m } } \sum_{\ell ^{'''}}\biggl [ \sum_{m^{'''}}\biggl\ { \left|c^{\ell ^{'''}m^{'''}}_{{\bar \ell } { \bar m};\ell -m}\right|^2\left|k^{\omega^{'''}\ell ^{'''}}_{\omega^{'}\ell ; { \bar \omega}{\bar \ell } } \right|^2 + \left|c^{\ell ^{'''}m^{'''}}_{{\bar \ell } -{\bar m};\ell m}\right|^2\left|k^{\omega^{'''}\ell ^{'''}}_{\omega^{'}\ell ; { \bar \omega}{\bar \ell } } \right|^2 \biggr\ } + \left|c^{\ell 0}_{{\bar \ell } { \bar m};\ell -m}\right|^2\left|k^{\omega^{'''}\ell ^{'''}}_{\omega^{'}\ell ; { \bar \omega}{\bar \ell } } \right|^2 \biggr]\end{aligned}\ ] ] this equation is divided into three parts , but only one has nonzero value for fixed @xmath47 and @xmath350 because @xmath351 . then
, after taking the summation of @xmath352 , we have @xmath353 summing up eqs.([127 ] ) , ( [ 129 ] ) , ( [ 130 ] ) and ( [ 132 ] ) , we see that the contribution from the tree diagrams to ( [ hituyou1 ] ) becomes @xmath354 so far , we have only considered the case there are at least one b - particle " for @xmath0 . calculating c - particle " and d - particle " and taking the summation of @xmath355 and @xmath47 in ( [ keisanshitai2 ] ) , we obtain @xmath356 next , we consider @xmath310 . with similar calculations with ( [ hituyou2 ] )
, we can get the result by replacing c with d and k with h in ( [ 7_20 ] ) .
the resultant expression becomes @xmath357 to summarize , substituting these s - matrix into ( [ keisanshitai2 ] ) , we can write down the effect of interaction on hawking radiation from the tree diagrams as @xmath358\ , \label{conclusion}\end{aligned}\ ] ] where @xmath359 and @xmath212 are calculated in ( [ hawking_beta ] ) . recall that we should use the rule @xmath360 when we evaluate the second term in the above formula ( [ conclusion ] ) and @xmath361 in the second term should be @xmath362 as understood from ( [ effect_b_int ] ) .
there are coupling function @xmath363 which is proportional to @xmath293 in the definition of @xmath364 , so it seems that @xmath364 would not converge .
however , as you can see , the divergences cancel out each other . in the interaction ( [ even_interaction ] ) , terms proportional to @xmath293 are @xmath365 \nonumber\end{aligned}\ ] ] or @xmath366\ .
\label{interaction_f}\end{aligned}\ ] ] then , the contribution from these interactions to @xmath367 becomes @xmath368\ , \label{k_f}\end{aligned}\ ] ] where @xmath369 and @xmath370 satisfy zerilli equation @xmath371 and equation of motion for @xmath0 @xmath372 in these equation , @xmath373 and @xmath185 dump sufficiently first as @xmath374 ( or near horizon ) .
therefore , @xmath369 and @xmath370 behave as @xmath375 substituting these expressions into ( [ k_f ] ) , we see that the contribution from near horizon vanishes . indeed , the contribution from the pattern @xmath376 is clearly zero and that from the pattern @xmath377 is cancelled out by the pattern @xmath378 .
thus , it turns out that @xmath367 converge by taking into account the equations of motion for @xmath309 and @xmath0 .
now , we estimate the effect of interaction ( [ conclusion ] ) . to begin with , we estimate @xmath367 and @xmath323 defined by @xmath379 where @xmath380 satisfy the boundary condition of in region " ( [ in_condition ] ) . using the geometric optics approximation
, we can deduce the asymptotic solution @xmath381 where @xmath192 and @xmath193 .
then , in order to estimate the effect of interaction , we approximate the solution in two ranges as @xmath382 and we call the former range outer area " and the latter range inner area " ( see fig.[fig : estimate ] ) . and the other area .
we put this boundary at @xmath383 .
, width=226 ] we consider the case that all @xmath384 in @xmath385 and @xmath386 are @xmath387 . for both
inner area " and outer area " , this case does not contribute to the integral because this pattern is proportional to the delta function @xmath388 since we only consider the positive frequency mode , the delta function gives zero .
hence , we need to consider only the mixed cases
. we can show that the contributions from outer area " always vanish .
we have already considered the case that all @xmath380 are @xmath387 .
let us consider the cases that one of @xmath389 is @xmath390 . if we consider only the outer area " , this effect is negligible because the integral becomes @xmath391 similarly , we can show that the contributions from outer area " for other cases are also zero .
therefore , we can concentrate on the contribution from inner area " .
we consider only the dominant terms in interactions ( [ odd_interaction ] ) and ( [ even_interaction ] ) .
first , we check the interaction terms proportional to @xmath293
. however , these terms do not affect the integral . because we assumed ( [ u_assume ] ) , the same calculation as the last subsection shows that these terms do not affect .
secondly , we consider the terms which are proportional to @xmath392 in ( [ even_interaction ] ) .
these terms are @xmath393 again these terms do not affect the integral from the same consideration as the last subsection .
there are @xmath394 in ( [ even_interaction ] ) ; @xmath395 and also exist in ( [ odd_interaction ] ) ; @xmath396 in ( [ even_phi_0 ] ) , we can show that the first three terms becomes zero by using the same method as the last subsection and ( [ ace ] ) .
furthermore , the other terms in ( [ even_phi_0 ] ) behave @xmath397 and @xmath398 as @xmath399 ; and coupling function in ( [ odd_phi_0 ] ) dumps as @xmath399 .
then , these terms do not affect in the asymptotic region .
this means that we can estimate the effect of these terms using not coupling functions but their value at the horizon , that is @xmath400 , @xmath401 and @xmath402 .
other terms can be neglected because their coupling functions are proportional to @xmath403 .
in fact , these terms do not affect both near the horizon and in the asymptotic region . therefore , summarizing these , the dominant interactions in inner area " are given by @xmath404 and @xmath405 thus , @xmath367 can be estimated as @xmath406 \ .\end{aligned}\ ] ] note that we ignored the constant factor .
and @xmath323 is @xmath407 \ .\end{aligned}\ ] ] however , from ( [ u_assume ] ) , we see this vanishes . then , we consider @xmath367 hereafter .
we assume @xmath408 . in the above ,
the pattern @xmath409 has been already considered .
hence , at least , one of @xmath410 should be @xmath411 . among these cases ,
the most dominant one is @xmath412 .
this is because @xmath413 implies that @xmath414 thus , the case with no derivative of @xmath411 is dominant . from these consideration , @xmath367 can be estimated as follows ; @xmath415 the integral in the above equation diverges .
however , this divergence occurs because we only considered outside of the horizon .
in fact , since the horizon fluctuates due to metric perturbations , we must extend the range of integral to inside the horizon .
we assume the range can be extended by @xmath416 .
then , the integral converges as @xmath417 so , @xmath367 can be estimated as @xmath418 finally , using this result , we can evaluate the nonlinear effect .
assuming @xmath419 , ( [ 7_12 ] ) can be deduced as @xmath420 where @xmath421 is a cut - off frequency . in the same manner , ( [ 7_20 ] ) can be estimated as @xmath422 here , we note the bogolubov coefficients are explicitly given by @xmath423 from which we obtain @xmath424 now , we can calculate the contribution from ( [ 7_20 ] ) to ( [ keisanshitai2 ] ) as @xmath425 \sim\frac{\epsilon^2\omega_{cut}^3}{\kappa^3m^2 l}\frac{e^{2\pi\omega/\kappa}+1 } { e^{2\pi\omega/\kappa}-1 } \ .\end{aligned}\ ] ] and , the contribution from ( [ 7_12 ] ) becomes @xmath426\nonumber\\ & \ & \sim \frac{\epsilon^2\omega_{cut}^3}{\kappa^3m^2l}\frac{e^{\pi\omega/\kappa}+e^{-\pi\omega/\kappa}}{e^{\pi\omega/\kappa}-e^{-\pi\omega/\kappa}}=\frac{\epsilon^2\omega_{cut}^3}{\kappa^3m^2l}\frac{e^{2\pi\omega/\kappa}+1}{e^{2\pi\omega/\kappa}-1 } \ .\end{aligned}\ ] ] taking into account that @xmath427 has dimension of the length , we can recover @xmath11 .
thus , it turns out that the cubic interaction modifies hawking radiation by the term proportional to @xmath428 this effect is suppressed by the planck scale .
therefore , the deviation from planck distribution is very small but always exists .
in this paper , we have studied hawking radiation from fluctuating black holes . for this purpose
, we constructed the canonical quadratic action for metric perturbations and also obtained the cubic interaction terms in 4-dimensional schwarzschild background .
taking into account cubic interaction terms in the action for a real scalar field , we derived the deviation from the planck spectrum by considering only tree level diagrams .
it turned out that the deviation is proportional to @xmath429 .
this result may have implications for black holes produced at the lhc and in the early universe .
as is mentioned in the introduction , our analysis is also related to the trans - planckian problem in hawking radiation .
there are many works to be done .
first of all , it is intriguing to investigate higher order corrections for metric perturbations , i.e. self - interaction terms of @xmath217 . as the amplitude of quantum fluctuations of the metric
is estimated as @xmath430 , these higher order corrections must be important if black holes become as small as planck scale . considering these terms
, we can investigate non - planckianity " of hawking radiation from black holes like non - gaussianity in inflation @xcite .
furthermore , from the ads / cft correspondence point of view , these interaction terms will be important for three point correlation functions in cft .
it is also interesting to generalize the present analysis to higher dimensional schwarzschild black holes .
this generalization is important because higher dimensional black holes might be created at the lhc if dimension of space - time is more than six .
the calculations would be more complicated , but the linear equations of motion have been already derived in @xcite .
therefore , it would be possible to compute the effect of interaction even for higher dimensional black holes .
it is also possible to consider gauge fields and fermions instead of a scalar field . finally , since einstein theory may not be valid near the planck scale
, it would be important to change the gravitational theory from einstein theory to other theories motivated by quantum gravity .
in fact , string theory predicts higher curvature corrections in addition to einstein - hilbert action @xcite .
again , the linear analysis has been already done in @xcite .
hence , we can extend our analysis to these cases .
this work is supported by the grant - in - aid for scientific research fund of the ministry of education , science and culture of japan no.22540274 , the grant - in - aid for scientific research ( a ) ( no .
22244030 ) , the grant - in - aid for scientific research on innovative area no.21111006 and the grant - in - aid for the global coe program the next generation of physics , spun from universality and emergence " .
we classify metric perturbations into scalar and vector perturbations in terms of the symmetry in 2-dimensional maximally symmetric space with the line element @xmath431 and the curvature @xmath432
. they can be expanded by scalar harmonics and vector harmonics . in this appendix
, we summarize our conventions .
scalar harmonics @xmath433 are defined by the eigenvalue equations @xmath434 where the eigenvalue @xmath435 depends on the index @xmath436 .
for example , if the 2-dimensional curvature is positive @xmath437 , scalar harmonics are nothing but spherical harmonics @xmath41 and the eigenvalue is given by @xmath438 . in this case
, @xmath436 corresponds to @xmath439 .
we impose normalization conditions @xmath440 using scalar harmonics , we can define vector harmonics @xmath441 where @xmath442 and @xmath443 .
the vector harmonics satisfy the transverse conditions @xmath444 and eigenvalue equations @xmath445 note that the vector harmonics satisfy the normalization conditions @xmath446
in this appendix , we calculate the quadratic action for metric perturbations using 3 + 1 decomposition .
the background spacetime we consider is a ( a)ds - schwarzschild black hole @xmath447 where @xmath16 denotes the metric of 2-dimensional maximally symmetric space with the curvature @xmath448 . using this metric ansatz , einstein equations with a cosmological constant @xmath449 @xmath450 read @xmath451
this can be easily solved as @xmath452 . here , @xmath18 is the constant of integration .
the lapse function @xmath453 and shift vector @xmath454 of this space - time are given by @xmath455 the induced metric is @xmath456 apparently , the extrinsic curvature of a @xmath457 surface vanishes @xmath458 the spatial curvatures are calculated as @xmath459 according to 3 + 1 decomposition , einstein - hilbert action becomes @xmath460 where @xmath7 is the determinant of 3 dimensional metric and @xmath461 is the trace part of extrinsic curvature . from now on ,
we call @xmath462 potential term " and @xmath463 kinetic term " .
let us define @xmath464 and @xmath465 . in order to calculate the quadratic action
, we must expand @xmath466 and @xmath329 ; @xmath467 and @xmath468 , where @xmath469 means the n - th order perturbation . substituting these expressions into ( [ ehaction ] )
, we obtain @xmath470\ .
\label{expand}\end{aligned}\ ] ] in the above action ( [ expand ] ) , we notice @xmath471 from ( [ back ] ) and ( [ back2 ] ) .
we also have @xmath472 because this term is proportional to background equations of motion . furthermore , because the extrinsic curvature of background vanishes , @xmath473 is given by the first order perturbation of @xmath474 , @xmath475 , and @xmath476 can be calculated as @xmath477 hence , the action becomes @xmath478 this tells us that we only need the first order lapse function , shift vector and extrinsic curvature .
hereafter , we omit @xmath479 " , that is , @xmath480 means first order perturbation of 3-dimensional ricci tensor , for example .
finally , we list up formulae for calculating perturbed quantities . the 3-dimensional background and perturbed metric are @xmath481 and @xmath482 , respectively . with this notation ,
the first and second order perturbations of the inverse metric are @xmath483 we can calculate christoffel symbols from the following formulae @xmath484 and , ricci tensor are given by @xmath485 note that @xmath486 means covariant derivative with respect to the 3-dimensional background metric @xmath481 .
we treat scalar perturbations using zerilli gauge .
later , we will change the gauge to a more convenient gauge . in the zerilli gauge ,
metric perturbations are expressed by @xmath487 then , up to the first order , the lapse function and the shift vector are given by @xmath488 and the induced metric becomes @xmath489 first , we calculate the second order perturbation of potential term " in the einstein - hilbert action . using formulae ( [ formula1])@xmath490([formula2 ] ) , in the zerilli gauge ,
perturbed quantities are calculated as follows : * inverse metric @xmath491 @xmath492 * christoffel symbols @xmath493 @xmath494 * ricci tensor @xmath495_{|a}\ , \nonumber\\ r^{(1)}{}_{ab}&=&-\frac{1}{2}h_{|ab}+\bigl[f\left(rh-\frac{1}{2}(r^2k)^{'}\right)^{'}+\frac{f^{'}}{2}\left(rh-\frac{1}{2}(r^2k)^{'}\right ) -\frac{1}{2}rfh^{'}-\frac{1}{2}k^{|a}{}_{|a}\bigr]\gamma_{ab}\ .\end{aligned}\ ] ] * scalar curvature @xmath496 * @xmath497 @xmath498 using these results , we can obtain potential terms " of the action .
the second order scalar curvature perturbation can be calculated as @xmath499 .
the first two term can be easily calculated by using the previous results .
using eq.([formula2 ] ) , the last term can be written as @xmath500 in the above eq.([r2last ] ) , the last two terms must be calculated using the formulae ( [ b19 ] ) , while the first two terms can be calculated by using the integration by parts @xmath501 thus , we can show that the second order potential terms " become @xmath502\ .\end{aligned}\ ] ] next , we calculate the kinetic terms " of the einstein - hilbert action . here
, we define @xmath503 . at the first order
, we have @xmath504 then , the trace part of the extrinsic curvature is given by @xmath505 and the quadratic part reads @xmath506 therefore , the second order perturbation of kinetic terms " can be deduced as @xmath507 \ .\end{aligned}\ ] ] then , we obtain the total quadratic action for perturbed metric @xmath508\ .
\label{zerilli_second_order}\end{aligned}\ ] ] we can derive the constraint equations taking the variation of this second order action with respect to @xmath51 and @xmath54 . however , we can not easily solve resultant constraint equations .
therefore , we need to change the gauge from zerilli gauge to the convenient gauge defined by ( [ fs_gauge ] )
. this gauge transformation can be obtained from ( [ even_gauge_transport ] ) .
the gauge transformation we use is @xmath509 where the index @xmath510 means zerilli gauge .
this gauge transformation can be obtained by setting @xmath511 .
then , from ( [ even_gauge_transport ] ) , the relation between these variables can be read off as @xmath512 substituting the transformation rules ( [ eq : zerilli2fs ] ) into the action ( [ zerilli_second_order ] ) , we get the quadratic action for metric perturbations in the convenient gauge as @xmath513 \ .\end{aligned}\ ] ] now , we expand metric perturbations by scalar harmonics and only consider the real mode ( for example , if @xmath437 , @xmath49 modes are real ) . of course , other modes can be calculated and we can get almost the same result as real modes . after the expansion
, the quadratic action becomes @xmath514 \ , \label{even_fs_action}\end{aligned}\ ] ] where @xmath515 or @xmath516 therefore , the hamiltonian constraint equation obtained by varying the action with respect to @xmath54 gives @xmath517 and the momentum constraint equation obtained by varying @xmath51 is @xmath518 substituting ( [ app_wh ] ) , ( [ app_q2 ] ) and ( [ app_h1 ] ) into ( [ even_fs_action ] ) , we get the quadratic action which is a functional of @xmath57 . using a new variable @xmath519 instead of @xmath57
, we obtain the quadratic action @xmath520 \ , \label{}\end{aligned}\ ] ] where we defined @xmath521 and @xmath522 is given by @xmath523\nonumber\\ & = & \frac{2f}{r^3}\frac{\lambda^2(\lambda+ k ) r^3 + 3\lambda^2mr^2 + ( 9\lambda r-3r^3\lambda)m^2 + 9m^3}{(r\lambda+3m)^2 } \ .
\end{aligned}\ ] ] finally , defining @xmath524 and using the tortoise coordinate , we obtain the canonical quadratic action for scalar type perturbations @xmath525\ .\label{app_even}\end{aligned}\ ] ] next , we derive the quadratic action for vector perturbations using regge - wheeler gauge @xmath526 where @xmath527 and @xmath79 satisfy transverse conditions @xmath528 . in this gauge ,
the zeroth and first order lapse function and shift vector are given by @xmath529 and induced metric becomes @xmath530 first , we consider the second order potential term " in the einstein - hilbert action . in our gauge , perturbed quantities are calculated as follows ; * inverse metric @xmath531 @xmath532 * christoffel symbols @xmath533 @xmath534 *
ricci tensor @xmath535 this proves that the first order scalar curvature @xmath536 vanishes .
therefore , in ( [ nijiaction ] ) , we only calculate @xmath537 for the potential term " . this term is easily calculated using the technique ( [ tric ] ) and the result is @xmath538 next , we calculate the kinetic term " .
the first order @xmath503 are @xmath539 these lead @xmath540 and @xmath541 then , the second order kinetic term " is given by @xmath542 \ .\end{aligned}\ ] ] note that we ignored total derivative terms .
thus , we have obtained the total quadratic action for vector type perturbations @xmath543 \ .\end{aligned}\ ] ] let us expand metric perturbations by vector harmonics @xmath544 and only consider the real modes .
the quadratic action becomes @xmath545 \ , \end{aligned}\ ] ] where @xmath521 and the overall factor comes from @xmath546 .
taking the variation of this action with respect to @xmath92 , we obtain the constraint equation which can not be solved easily .
however , this difficulty can be circumvented by working in phase space . from the lagrangian @xmath547 \ , \nonumber\end{aligned}\ ] ] we can define conjugate momentum @xmath101 as @xmath548 which leads to @xmath549 the hamiltonian is @xmath550 \ .
\nonumber\end{aligned}\ ] ] in phase space , constraint equation can be written as @xmath551 it is easy to solve the above constraint as @xmath552 substituting this result into the hamiltonian , we have reduced hamiltonian @xmath553 \ , \label{hamiltonian}\end{aligned}\ ] ] which is a functional of @xmath70 and @xmath101 .
let us make the following simple canonical transformation @xmath554 where @xmath555 is a new hamiltonian .
then , we have @xmath556 \ . \end{aligned}\ ] ] now , we define @xmath557 thus , the action becomes @xmath558 \ .\end{aligned}\ ] ] finally , changing normalization by @xmath559 and using the tortoise coordinate , we obtain the canonical quadratic action for vector perturbation @xmath560 \ , \label{app_odd}\end{aligned}\ ] ] where we have defined @xmath561
in this appendix , we identify the symmetry in the quadratic action .
as shown in @xcite , we can rewrite @xmath522 and @xmath562 as @xmath563 where @xmath564 now we consider the quadratic action ( [ app_even ] ) and ( [ app_odd ] ) . using @xmath565 ,
this action can be rewritten as @xmath566 \ , \label{double}\end{aligned}\ ] ] where @xmath567 represents @xmath568 . here , the operator @xmath569 is defined by @xmath570 where @xmath571 is a pauli matrix .
then , if there exist operators which commutate with @xmath569 , the canonical action ( [ double ] ) has symmetry .
in fact , there are two such operators defined by @xmath572 where @xmath573 and @xmath574 are also pauli matrices .
these operators satisfy anti - commutation relation @xmath575=0 .
then we can conclude that the quadratic action in the ( a)ds - schwarzschild background has n=2 supersymmetry " .
s. w. hawking , commun .
phys . * 43 * , 199 ( 1975 ) [ erratum - ibid . *
46 * , 206 ( 1976 ) ] .
w. g. unruh , phys .
d * 14 * , 870 ( 1976 ) .
w. g. unruh and n. weiss , phys .
d * 29 * , 1656 ( 1984 )
. n. d. birrell and p. c. w. davies , phys .
d * 18 * , 4408 ( 1978 ) .
d. a. leahy and w. g. unruh , phys .
d * 28 * , 694 ( 1983 ) .
p. kanti , lect .
notes phys .
* 769 * , 387 ( 2009 ) [ arxiv:0802.2218 [ hep - th ] ] .
d. ida , k. y. oda and s. c. park , phys .
d * 67 * , 064025 ( 2003 ) [ erratum - ibid .
d * 69 * , 049901 ( 2004 ) ] [ arxiv : hep - th/0212108 ] .
k. murata , j. soda and s. kanno , phys .
d * 75 * , 104017 ( 2007 ) [ arxiv : gr - qc/0703029 ] .
h. ishihara and j. soda , phys .
d * 76 * , 064022 ( 2007 ) [ arxiv : hep - th/0702180 ] . s. b. giddings and s. d. thomas , phys . rev .
d * 65 * , 056010 ( 2002 ) [ arxiv : hep - ph/0106219 ] . v. moncrief , annals phys .
* 88 * , 323 ( 1974 ) .
f. j. zerilli , phys .
* 24 * , 737 ( 1970 ) .
t. regge and j. a. wheeler , phys .
108 * , 1063 ( 1957 ) .
n. d. birrell and s. d. davies quantum fields in curved space " cambridge university press .
j. m. maldacena , jhep * 0305 * , 013 ( 2003 ) [ arxiv : astro - ph/0210603 ] .
h. kodama and a. ishibashi , prog .
phys . * 110 * , 701 ( 2003 ) [ arxiv : hep - th/0305147 ] ; a. ishibashi and h. kodama , prog .
phys . * 110 * , 901 ( 2003 ) [ arxiv : hep - th/0305185 ] ; h. kodama and a. ishibashi , prog .
phys . * 111 * , 29 ( 2004 ) [ arxiv : hep - th/0308128 ] .
d. g. boulware and s. deser , phys . rev
. lett . * 55 * , 2656 ( 1985 ) .
g. dotti and r. j. gleiser , class .
* 22 * , l1 ( 2005 ) [ arxiv : gr - qc/0409005 ] ; + g. dotti and r. j. gleiser , phys .
d * 72 * , 044018 ( 2005 ) [ arxiv : gr - qc/0503117 ] .
t. takahashi and j. soda , phys .
d * 79 * , 104025 ( 2009 ) [ arxiv:0902.2921 [ gr - qc ] ] ; + t. takahashi and j. soda , phys .
d * 80 * , 104021 ( 2009 ) [ arxiv:0907.0556 [ gr - qc ] ] .
i. bakas , jhep * 0901 * , 003 ( 2009 ) [ arxiv:0809.4852 [ hep - th ] ] . | classically , black holes have the rigid event horizon .
however , quantum mechanically , the event horizon of black holes becomes fuzzy due to quantum fluctuations .
we study hawking radiation of a real scalar field from a fluctuating black hole .
to quantize metric perturbations , we derive the quadratic action for those in the black hole background .
then , we calculate the cubic interaction terms in the action for the scalar field . using these results , we obtain the spectrum of hawking radiation in the presence of interaction between the scalar field and the metric .
it turns out that the spectrum deviates from the planck spectrum due to quantum fluctuations of the metric . |
in recent years remarkable progress has been made towards formulations of lattice fermions which have no doublers and possess an exact chiral symmetry without giving up desirable features like flavor symmetry , locality , unitarity or gauge - invariance .
the ginsparg - wilson relation @xcite @xmath1 has been recognized @xcite as the fundamental condition in this context , and a dirac operator @xmath2 which satisfies this relation can indeed describe chiral fermions on the lattice .
one of the possible solutions of the ginsparg - wilson relation has been found by neuberger @xcite starting from the overlap formalism @xcite . in the massless case
the overlap - dirac operator is @xmath3 , \label{eq : dn}\ ] ] where @xmath4 in terms of the usual wilson - dirac operator @xmath5 , \ ] ] @xmath6 .\ ] ] in the range @xmath7 the right spectrum of massless fermions is obtained @xcite . for a quark of bare mass @xmath8
the overlap - dirac operator becomes @xmath9 since additive mass renormalization is forbidden , one avoids altogether a source of systematic errors that is always present with wilson fermions @xcite .
lscher has shown @xcite that a fermion obeying the ginsparg - wilson relation possesses an exact chiral symmetry of the general form @xmath10 a fully legitimate form of global chiral symmetry is thus preserved for finite lattice spacing .
the expected value of the global anomaly @xcite ( which comes from the non - invariance of the fermionic integration measure under the transformations above ) , the atiyah - singer index theorem on the lattice @xcite and the chiral ward identities are then fully attained before taking the continuum limit .
thus , the central point in this framework is not to insist on the lattice form of the canonical chiral transformations ( in fact actions satisfying eq . ( [ eq : gw ] ) are not chirally invariant in the canonical sense ) , but rather be satisfied with a modified form of the chiral symmetry .
the nielsen - ninomyia theorem @xcite is then circumvented altogether , and it is possible to define chiral fermions without doublers or breaking of flavor symmetry or other unpleasant drawbacks .
the chiral transformations ( [ eq : symmetry ] ) depend also on the interaction , but this does not forbid the non - perturbative construction of chiral gauge theories on the lattice with exact gauge invariance @xcite . for recent reviews on the interesting properties of fermions satisfying the ginsparg - wilson relation , see refs .
. we will be interested in the following in performing chiral - invariant computations by using the overlap - dirac operator ( [ eq : dn ] ) .
this has been proven to be local @xcite , and although simulations with the neuberger operator look computationally very demanding , progress is under way @xcite .
there has been activity also on the analytic side , and some 1-loop calculations with overlap fermions have been already carried out @xcite .
the most recent and advanced calculations have featured the relation between the @xmath11 parameter in the lattice scheme defined by the overlap operator and in the @xmath12 scheme @xcite , and the renormalization factors of the quark bilinears @xmath13
@xcite .
in this paper we present the calculation in lattice qcd of the renormalization factors of a few operators which measure the lowest two moments of non - singlet quark distributions .
the generic operators which measure their @xmath14-th moments are @xmath15 and @xmath16 for unpolarized and polarized structure functions respectively .
contrary to what happens with wilson fermions , in the present case , thanks to the exact chiral symmetry that we maintain , the renormalization constants for every pair of corresponding unpolarized and polarized operators which differ by a @xmath0 matrix are the same .
this paper is organized as follows : in sect .
2 we introduce the various operators that we have studied , in sect
. 3 their renormalization on the lattice is discussed , and in sect .
4 some details about the perturbative calculations are given . finally , in sect .
5 we present our results . in the appendices one can find the feynman rules that we have used as well as some analytic results .
the hadronic physics in deep inelastic scattering is contained in the matrix elements of the t - product of the hadronic currents @xmath17 @xmath18 being the momentum transfer and @xmath19 the target momentum .
an operator product expansion on the light cone of the kind @xmath20 describes the physics in the bjorken limit , in which scaling is reached and the structure functions depend only on the bjorken variable @xmath21 .
the moments of the structure functions , which measure the moments of quark distributions , are directly connected to the forward matrix elements of the local operators appearing in the light - cone expansion . since the divergence of the wilson coefficients is governed by the twist ( dimension minus spin ) of the corresponding operators , infinite towers of operators of increasing dimension and spin but with the same twist ( dimension minus spin ) appear at a given order in @xmath22 .
the dominant contribution is given by twist two , which in the flavor non - singlet case means the symmetric traceless operators @xmath23 where @xmath24 are flavor matrices ( which in the following will be omitted and implicitly understood ) .
these operators measure the moment of unpolarized and polarized structure functions respectively , that is the moments @xmath25 of quark momentum distributions and the moments @xmath26 of quark helicity distributions , at leading twist .
higher - twist operators give the sub - dominant contributions , and in particular twist-4 operators ( which include also 4-fermion operators ) measure the @xmath22 corrections to these moments @xcite .
since we are considering flavor non - singlet operators , there can be no mixing with operators like @xmath27 and @xmath28 which measure the gluon distributions . mixing with these operators
is prohibited even when one considers unquenched calculations .
singlet quark operators instead do mix with gluon operators . for more detailed discussions of deep inelastic scattering on the lattice , see refs .
@xcite .
the operators in the light - cone expansion ( [ eq : ope ] ) of which we compute the renormalization factors are the following : @xmath29 which measure the first moment of quark momentum distributions , @xmath30 which measure the first moment of quark helicity distributions , @xmath31 which measure the second moment of quark momentum distributions , and @xmath32 which measure the second moment of quark helicity distributions .
symmetrization in all lorentz indices is understood .
we use @xmath33 , with the following lattice discretizations : @xmath34 \\ \bar{\psi } ( x ) \stackrel{\leftarrow}{d_\mu } & = & \frac{1}{2a } \big [ \bar{\psi } ( x+a\hat{\mu } ) u^\dagger ( x,\mu ) - \bar{\psi}(x - a\hat{\mu } ) u(x - a\hat{\mu},\mu ) \big ] .\end{aligned}\ ] ] we have chosen the lorentz indices of each operator appearing in the continuum expansion in two different ways , so that they fall in two different representations of the hypercubic group ( the symmetry group of the lattice , remnant of the lorentz symmetry ) , and on the lattice they will then renormalize in a different way .
the representations where the indices are all different from each other are least likely to mix with other operators , however one needs more components of the hadron momentum different from zero when simulating the corresponding matrix elements , and this can lead to more lattice artifacts .
sometimes one of the representations can be easier to handle in practice , depending on the trade - off between the amount of mixing and the number of non - zero momentum components .
for example , in the case of the second moment the operator ( [ eq:2ud ] ) is multiplicatively renormalized , but it can be more advantageous to use the other representation ( operator ( [ eq:2ue ] ) ) when it is important that fewer components of the momenta are different from zero , although in this case one has then to deal with a mixing .
the operator ( [ eq:2ue ] ) ( together with the corresponding polarized ( [ eq:2pe ] ) ) is the only one in the list above that is not multiplicatively renormalized on the lattice .
what happens is that that the two operators @xmath35 which are not symmetrized in their indices , do not receive the same 1-loop corrections on the lattice @xcite , and so the operator ( [ eq:2ue ] ) , @xmath36 does not go into itself under lattice renormalization , because the symmetric combination is lost . in the polarized case ,
an analogous situation occurs for @xmath37 a detailed discussion of these effects can be found in sect .
we mention that for the second moment there exists also a third independent lattice representation , in which all indices are equal .
however , it mixes with a lower dimensional operator with a power - divergent coefficient ( @xmath38 ) , and so we do not consider it here .
to relate the numbers obtained from monte carlo simulations to physical continuum quantities , a lattice renormalization of the relevant matrix elements is necessary .
the connection is given by @xmath39 where @xmath40 are the continuum and lattice 1loop expressions respectively , and the tree - level matrix element is the same in both cases .
the differences @xmath41 enter then in the renormalization factors @xmath42 which connect the lattice to the continuum .
these renormalization factors are independent of the state , depend only on the scale @xmath43 and are gauge - invariant . using them
, a matrix element obtained from monte carlo simulations can be renormalized to a continuum scheme .
although there are in principle also non - perturbative methods with which one can determine the renormalization factors , perturbation theory is still important .
in fact , it can happen that for some operators a window for the non - perturbative signal is difficult to obtain , and then perturbation theory remains the only possibility of computing the relevant renormalization factors . for the neuberger operator , extracting the non - perturbative renormalization factors in addition to simulating the bare matrix elements could turn out to be computationally very expensive .
in general , perturbative lattice renormalization is important as a hint and a guide for non perturbative renormalization studies , and even more when mixings are present , which are generally more intricate than in the continuum case , and more transparent when looked at in perturbation theory , especially if some amounts of mixings are small .
perturbative renormalization can in any case be very useful in checking and understanding results obtained with non - perturbative methods .
in the continuum we renormalize the operators in the @xmath44 scheme . as perturbative renormalization condition on the lattice
we impose that the 1-loop amputated matrix elements at a certain reference scale @xmath45 are equal to the corresponding bare tree level quantities . for lattice matrix elements of multiplicatively renormalized operators computed between one - quark states
this condition means @xmath46 where @xmath47 is the wave - function renormalization , computed from the quark self - energy .
[ fig : diagrams ] the 1-loop lattice matrix elements of the operators we consider here , which are at most logarithmically divergent , will have the form @xmath48 where @xmath49 is the finite contribution of the vertex and sails diagrams ( a , b and c in fig . 1 ) , @xmath50 refers to the tadpole arising from the operator ( d in fig . 1 )
, s is the finite contribution ( proportional to @xmath51 ) of the quark self - energy of one leg , including the leg tadpole ( e and g , or f and h , in fig . 1 ) , and @xmath52 . the @xmath53 factor for an operator @xmath54 will then be given by @xmath55 with @xmath56 we will call `` proper '' contributions the ones that exclude the self - energy diagrams .
they correspond to the diagrams a - d in fig . 1 .
although the theory defined by the overlap - dirac operator ( [ eq : dn ] ) has an exact chiral symmetry and no lattice artifacts of order @xmath57 are present in the action , matrix elements of operators still possess corrections of order @xmath57 and therefore they need to be improved .
the improved operator corresponding to @xmath58 is @xcite @xmath59 and the big bonus with overlap fermions is that the renormalization factors for the improved operator and for the unimproved operator without the rotations @xmath60 are the same .
what happens is that in 1-loop amplitudes a factor @xmath61 can combine with a quark propagator , but since it has an @xmath57 in front and ( contrary to the wilson case ) there is no @xmath62 factor in the propagator , as additive mass renormalization is forbidden by chiral symmetry , the contribution of @xmath61 to the renormalization factors is zero @xcite .
thus , we can simulate the improved operator ( [ eq : opimp ] ) and renormalize it as if it were the unimproved operator , which saves a lot in the perturbative calculations .
chiral symmetry , in addition to avoiding the extrapolations to zero quark mass in the simulations , gives also useful constraints on the mixing patterns and on the values of the renormalization factors .
in general , mixings that in the wilson case were allowed by the breaking of chirality are now forbidden , although this has less relevance here than in other problems like in weak interactions @xcite .
an important consequence here is that , similarly to what happens in the case of the bilinears where we have @xmath63 and @xmath64 @xcite , chiral symmetry forces pairs of corresponding unpolarized and polarized operators , like @xmath65 and @xmath66 , to have the same renormalization constants ) , which transform like chiral multiplets , but since they have the same renormalization constants as the unimproved operators , these symmetry relations are preserved . ] .
the interaction vertices and the propagator of the overlap - dirac operator are much more complicated than the ones in the wilson formulation , and this causes the computations to be rather cumbersome ( see appendices ) .
computer programs need then to be introduced , and we have performed the calculations using an ensemble of routines written in the symbolic manipulation language form .
these routines are an extension of the ones used to do calculations with the wilson action in various occasions @xcite .
the outputs of the form codes , which correspond to the results of the analytic calculations , are fed to fortran programs which perform the numerical integrations . to treat the divergent integrals
we use the kawai method @xcite , in which they are split in integrals independent of external momenta plus continuum integrals .
divergent lattice integrals which contain both gluon and overlap - quark propagators are reduced to divergent integrals containing only gluon propagators , which are computed in a exact way as in @xcite , plus a finite rest .
many checks have been performed on our results .
we did the computations in a general covariant gauge ] ] , and the cancellations of the gauge - dependent parts between the continuum scheme and the lattice results is a strong check on the good behavior of the form codes , as well as of the integration routines .
we have used the codes to compute also the renormalization constants for all quark bilinears , and we agree with the results given in feynman - gauge by alexandrou et al .
@xcite , as well as with their self - energy . in the case of the leg tadpole and of the vertex graph of the scalar current @xmath67
we have also made the computations entirely by hand ( although only in the feynman gauge ) , and successfully checked them against the form output expressions .
these analytic results are reported in appendix b. we tried to do a calculation by hand for a first moment operator and for the rest of the leg self - energy , but they involved a huge amount of manipulations .
we rely also on the vast amount of results which have been produced in different occasions using these form codes limited to the wilson formulation @xcite , and these results have by now a certain number of cross - checks . with overlap fermions
we need to introduce a new propagator and new vertices , but the operators are unaltered , and the same wilson expressions for their expansion in @xmath57 can be used here . many of the routines ( for example the gamma algebra reduction ) are also exactly the same as in the wilson case .
we checked in all cases that the polarized operators have the same @xmath68s as the corresponding unpolarized operators .
this is a rather strong check , as the analytic results are widely different for the two cases , and it is only after performing the numerical integrations and seeing that the difference of the two results is much smaller than the precision of our integrals that we able to say that the renormalization factors are indeed equal .
finally , since our codes are able to do calculations both in dimensional regularization and using a mass regularization , we also checked that using two different intermediate regularizations while treating the divergences with the kawai method @xcite we get the same results .
we obtain the @xmath69 constants with 5 digits after the decimal points , by doing the numerical integrations on a @xmath70 regular grid and using the method proposed by lscher and weisz in @xcite to accelerate convergence .
we have checked that these digits do not vary when a @xmath71 integration grid is used . in any case , to get more significant digits , grids of @xmath72 or @xmath73 points would still not be enough , and they would require an enormous computational effort , as a typical form output for the second moment operators already contains thousands of terms .
comparing our results for the self - energy and the bilinears with the numbers in ref .
@xcite , we can see that a few times a difference of one unit on the fifth digit can be noticed , and this agrees with our estimate of errors . in that paper
the results are given in terms of the constants @xmath74 , and this explains the presence of two more significant digits after the decimal point .
we give in this section the renormalization factors of the operators considered in this work . that these factors are identical for pairs of corresponding unpolarized and polarized operators
has been explicitly verified for all operators .
we give then the numerical results only for the unpolarized operators , for @xmath75 .
we first consider the 1-loop contributions of the proper diagrams ( a - d in fig . 1 ) , which for the operators that do not mix are @xmath76 o_{v_2,d}^{\rm tree } , \nonumber \\
o_{v_2,e}^{\rm proper } = \frac{g_0 ^ 2}{16\pi^2 } c_f & & \bigg [ \big ( \frac{5}{3 } + ( 1-\alpha ) \big ) \log a^2 p^2 \\ & & \qquad + v^{\ \alpha=1}_{v_2,e } - ( 1-\alpha)\ , 6.850272 + t_{v_2,e } \bigg ] o_{v_2,e}^{\rm tree } , \nonumber \\ o_{v_3,d}^{\rm proper } = \frac{g_0 ^ 2}{16\pi^2 } c_f & & \bigg [ \big ( \frac{19}{6 } + ( 1-\alpha ) \big ) \log a^2 p^2 \nonumber \\ & & \qquad + v^{\ \alpha=1}_{v_3,d } - ( 1-\alpha)\ , 7.369693 + t_{v_3,d } \bigg ] o_{v_3,d}^{\rm tree } \nonumber .\end{aligned}\ ] ] the contributions of the sails and vertices , @xmath49 , have been for convenience separated in the feynman gauge results ( @xmath77 , @xmath78 and @xmath79 ) , tabulated in table 1 for various values of the parameter @xmath80 , and the remaining parts proportional to @xmath81 , which are instead independent of @xmath80 .
the analytic expressions of the latter are very complicated functions of @xmath80 containing thousands of terms , and the numerical cancellation of this dependence is a rather non - trivial check of our computations .
these numbers are also independent of the lattice representation , and furthermore they have the same value as with the wilson action @xcite , so they seem to be to a certain extent independent of the particular fermion action chosen .
.the feynman - gauge constants @xmath82 for the multiplicatively renormalized operators . [ cols="^,>,>,>",options="header " , ] the contributions @xmath50 of the operator tadpoles ( d in fig . 1 ) are shown in table 2 , which contains also their results for the other operator , @xmath83 .
however , since this operator is not multiplicatively renormalized on the lattice , we consider in its place the two operators @xmath84 and @xmath85 introduced in sect .
2 . they mix with each other and we can write their renormalization as and an operator of mixed symmetry were considered instead of @xmath84 and @xmath85 .
] @xmath86 in terms of the bare operators @xmath84 and @xmath85 , the renormalized operator @xmath87 appearing in the dis light - cone expansion will be @xmath88 and since on the lattice @xmath89 , the result is that @xmath83 is not multiplicatively renormalized .
in fact , the 1-loop contributions coming from the proper diagrams are in this case @xmath90 o_a^{\rm tree } \nonumber \\ + \frac{g_0 ^ 2}{16\pi^2 } c_f & & \bigg [ \log a^2
p^2 + v_{ab}^{\ \alpha=1 } + \frac{1}{3 } ( 1-\alpha ) + t_{ab } \bigg ] o_b^{\rm tree } \nonumber \\
o_b^{\rm proper } = \frac{g_0 ^ 2}{16\pi^2 } c_f & & \bigg [ 2\log a^2 p^2 + v_{ba}^{\ \alpha=1 } + \frac{2}{3 } ( 1-\alpha ) + t_{ba } \bigg ] o_a^{\rm tree } \\
+ \frac{g_0 ^ 2}{16\pi^2 } c_f & & \bigg [ \big ( \frac{13}{6 } + ( 1-\alpha )
\big ) \log a^2 p^2 \nonumber \\ & & \qquad + v_{bb}^{\ \alpha=1 } - ( 1-\alpha ) \ , 7.703026 + t_{bb } \bigg ] o_b^{\rm tree } \nonumber,\end{aligned}\ ] ] where the finite contributions of vertex and sails in the feynman gauge are shown in table 3 .
we stress again that this mixing is a pure lattice effect , deriving from the breaking of the lorentz group to the hypercubic group . in the continuum
we do have @xmath91 , and thus the operator @xmath83 is multiplicatively renormalized , and has the same @xmath68 as the operator with different indices @xmath92 , as they both belong to the same representation of the lorentz group .
we remind also that in the polarized case one encounters exactly the same situation with the same numerical results , i.e. @xmath93 , @xmath94 etc .
, as we have explicitly verified .
to complete the computation of the renormalization factors , we have now to add to the proper diagrams the 1-loop amplitudes of the self - energy and tadpole of one leg which are proportional to @xmath51 , @xmath95,\ ] ] where the feynman - gauge finite results @xmath96 are given in table 4 . putting everything together
, we get the expressions of the renormalized operators on the lattice for overlap fermions , which for @xmath97 are : @xmath98 o_{v_2,d}^{\rm tree } \nonumber \\
\widehat{o}_{v_2,e } & = & \bigg [ 1 -\frac{g_0 ^ 2}{16\pi^2 } c_f \bigg(\frac{8}{3 } \log a^2 \mu^2 -52.82171 + ( 1-\alpha ) \bigg ) \bigg ] o_{v_2,e}^{\rm tree } \nonumber \\
\widehat{o}_{v_3,d } & = & \bigg [ 1 -\frac{g_0 ^ 2}{16\pi^2 } c_f \bigg(\frac{25}{6 } \log a^2 \mu^2 -68.99341 + \frac{3}{2 } \ , ( 1-\alpha ) \bigg ) \bigg ] o_{v_3,d}^{\rm tree } \nonumber \\
\widehat{o}_{v_3,e } & = & \frac{1}{3 } \bigg [ 1 -\frac{g_0 ^ 2}{16\pi^2 } c_f \bigg(\frac{25}{6 } \log a^2 \mu^2 -73.29777 + \frac{3}{2 } \ , ( 1-\alpha ) \bigg ) \bigg ] o_a^{\rm tree } \\ & & + \frac{1}{3 } \bigg [ 1 -\frac{g_0 ^ 2}{16\pi^2 } c_f \bigg(\frac{25}{6 } \log a^2 \mu^2 -67.83586 + \frac{3}{2 } \ , ( 1-\alpha ) \bigg ) \bigg ] o_b^{\rm tree } .
\nonumber \end{aligned}\ ] ] we give here for comparison the numerical values obtained with the usual wilson action , without any improvement .
we have computed them again and checked them with the results in the literature @xcite . in this case
however chiral invariance is broken and the renormalization factors for the unpolarized operators @xmath99 o_{v_2,d}^{\rm tree } \nonumber \\ \widehat{o}^{\rm wilson}_{v_2,e } & = & \bigg [ 1 -\frac{g_0 ^ 2}{16\pi^2 } c_f \bigg(\frac{8}{3 } \log a^2 \mu^2 -1.88259 + ( 1-\alpha ) \bigg ) \bigg ] o_{v_2,e}^{\rm tree } \nonumber \\
\widehat{o}^{\rm wilson}_{v_3,d } & = & \bigg [ 1 -\frac{g_0 ^ 2}{16\pi^2 } c_f \bigg ( \frac{25}{6 } \log a^2
\mu^2 -19.00763 + \frac{3}{2 } \ , ( 1-\alpha ) \bigg ) \bigg ] o_{v_3,d}^{\rm tree } \nonumber \\ \widehat{o}^{\rm wilson}_{v_3,e } & = & \frac{1}{3 } \bigg [ 1 -\frac{g_0 ^ 2}{16\pi^2 } c_f \bigg ( \frac{25}{6 } \log a^2
\mu^2 -21.78271 + \frac{3}{2 } \ , ( 1-\alpha ) \bigg ) \bigg ] o_a^{\rm tree } \\ & & + \frac{1}{3 } \bigg [ 1 - \frac{g_0 ^ 2}{16\pi^2 } c_f \bigg ( \frac{25}{6 } \log a^2 \mu^2 -18.46640 + \frac{3}{2 } \ , ( 1-\alpha ) \bigg ) \bigg ] o_b^{\rm tree } \nonumber\end{aligned}\ ] ] differ from the ones of the polarized operators @xmath100 o_{a_2,d}^{\rm tree } \nonumber \\ \widehat{o}^{\rm wilson}_{a_2,e } & = & \bigg [ 1 -\frac{g_0 ^ 2}{16\pi^2 } c_f \bigg(\frac{8}{3 } \log a^2 \mu^2 -4.27705 + ( 1-\alpha ) \bigg ) \bigg ] o_{a_2,e}^{\rm tree } \nonumber \\
\widehat{o}^{\rm wilson}_{a_3,d } & = & \bigg [ 1 -\frac{g_0 ^ 2}{16\pi^2 } c_f \bigg ( \frac{25}{6 } \log a^2 \mu^2 -19.56159 + \frac{3}{2 } \ , ( 1-\alpha ) \bigg ) \bigg ] o_{a_3,d}^{\rm tree } \nonumber \\ \widehat{o}^{\rm wilson}_{a_3,e } & = & \frac{1}{3 } \bigg [ 1 -\frac{g_0 ^ 2}{16\pi^2 } c_f \bigg ( \frac{25}{6 } \log a^2 \mu^2 -22.39940 + \frac{3}{2 } \ , ( 1-\alpha ) \bigg ) \bigg ] o_{a^5}^{\rm tree } \\ & & + \frac{1}{3 } \bigg [ 1 - \frac{g_0 ^ 2}{16\pi^2 } c_f \bigg ( \frac{25}{6 } \log a^2 \mu^2 -19.25837 + \frac{3}{2 } \ , ( 1-\alpha ) \bigg)\bigg ] o_{b^5}^{\rm tree}. \nonumber \end{aligned}\ ] ] what one can notice is that the values of the renormalization factors for overlap fermions are in general much larger than for wilson fermions .
this seems to be true for most values of @xmath80 , and is to trace largely to the quark self - energy and to the operators tadpoles .
it can be observed that when in the wilson formulation one improves the theory by adding the clover term to the action ( with @xmath101 ) and by canceling @xmath102 effects on the operators at tree level , the renormalization factors become also somewhat larger , as in the examples below @xcite : @xmath103 o_{v_2,d}^{\rm tree } \nonumber \\
\widehat{o}^{\rm wilson,~imp}_{v_3,d } & = & \bigg [ 1 -\frac{g_0 ^ 2}{16\pi^2 } c_f \bigg ( \frac{25}{6 } \log a^2 \mu^2 -29.815 + \frac{3}{2 } \ , ( 1-\alpha ) \bigg ) \bigg ] o_{v_3,d}^{\rm tree } \nonumber \\ \widehat{o}^{\rm wilson,~imp}_{v_3,e } & = & \frac{1}{3 } \bigg [ 1 -\frac{g_0 ^ 2}{16\pi^2 } c_f \bigg ( \frac{25}{6 } \log a^2 \mu^2 -39.192 + \frac{3}{2 } \ , ( 1-\alpha ) \bigg ) \bigg ] o_a^{\rm tree } \\ & & + \frac{1}{3 } \bigg [ 1 - \frac{g_0 ^ 2}{16\pi^2 } c_f \bigg ( \frac{25}{6 } \log a^2 \mu^2 -22.141 + \frac{3}{2 } \ , ( 1-\alpha ) \bigg ) \bigg ] o_b^{\rm tree } \nonumber .\end{aligned}\ ] ] one could then speculate whether the large renormalization factors that we have obtained for overlap fermions are related to the fact that the neuberger action is improved and the @xmath68s correspond to the ones of improved operators .
a calculation with wilson fermions with which to compare in this sense our overlap results would be one in which the operators are improved beyond tree level , but although we know in some cases the contributions of the operator counterterms that are needed for the full improvement , unfortunately not all their coefficients are yet determined @xcite .
we give finally the 1-loop results for the various matrix elements in the @xmath44 scheme , so that one can complete the connection with the continuum as in eq .
( [ eq : matching ] ) . in the continuum
there is only one renormalization constant for all possible operators ( unpolarized and polarized ) measuring the first moment , and only one for all possible operators measuring the second moment : @xmath104 o_{v_2}^{\rm tree } \label{eq : msbar}\\ \widehat{o}^{\overline{\rm{ms}}}_{v_3 } & = & \bigg [ 1 - \frac{g_0 ^ 2}{16\pi^2 } c_f \bigg ( \frac{25}{6 } \log \frac{p^2}{\mu^2 } -\frac{67}{9 } + \frac{3}{2 } \ , ( 1-\alpha ) \bigg ) \bigg ] o_{v_3}^{\rm tree } . \nonumber \end{aligned}\ ] ] the connection of overlap lattice fermions with the continuum @xmath44 is then given by the gauge - invariant factors @xmath105 o^{\rm lat}_{v_2,d } \nonumber \\
\widehat{o}^{\overline{\rm{ms}}}_{v_2,e } & = & \bigg [ 1 -\frac{g_0 ^ 2}{16\pi^2 } c_f \bigg(\frac{8}{3 } \log a^2 \mu^2 -48.37727 \bigg ) \bigg ] o^{\rm lat}_{v_2,e } \nonumber \\
\widehat{o}^{\overline{\rm{ms}}}_{v_3,d } & = & \bigg [ 1 - \frac{g_0 ^ 2}{16\pi^2 } c_f \bigg(\frac{25}{6 } \log a^2 \mu^2 -61.54897 \bigg ) \bigg ] o^{\rm lat}_{v_3,d } \nonumber \\
\widehat{o}^{\overline{\rm{ms}}}_{v_3,e } & = & \frac{1}{3 } \bigg [ 1 - \frac{g_0 ^ 2}{16\pi^2 } c_f \bigg(\frac{25}{6 } \log a^2 \mu^2 -65.85333 \bigg ) \bigg ] o^{\rm lat}_a \\ & & + \frac{1}{3 } \bigg [ 1 - \frac{g_0 ^ 2}{16\pi^2 } c_f \bigg ( \frac{25}{6 } \log a^2 \mu^2 -60.39142 \bigg ) \bigg ] o^{\rm lat}_b .
\nonumber \end{aligned}\ ] ] it looks increasingly difficult to go to higher moments , as the number of terms that are present in the form outputs and that need to be numerically integrated becomes very large .
this has at the moment limited our calculations to second moment operators , but we hope to be able to compute the renormalization of third - moment operators in the near future .
finally , we remark that all the renormalization constants presented here can be considered as computed in the unquenched case , because we limit ourselves to 1-loop computations with non - singlet quark operators , where internal quark loops never have the chance to come to play .
i have enjoyed stimulating discussions with robert edwards and leonardo giusti .
i would also like to thank mark alford for critically reading the manuscript . both the form and fortran computations
have been done at mit on a few pentium iii pcs running on linux .
this work has been supported in part by the u.s .
department of energy ( doe ) under cooperative research agreement de - fc02 - 94er40818 .
we give here the feynman rules which are needed to perform 1-loop calculations using the overlap - dirac operator .
an explicit derivations of these rules is given in @xcite .
the quark propagator is @xmath106 where @xmath107 the vertices needed for 1-loop calculations can be entirely given in terms of the vertices of the wilson action @xmath108 ( where @xmath109 and @xmath110 are the quark momenta flowing in and out of the vertices ) and of the quantity @xmath111 the quark - quark - gluon vertex in the overlap theory has the expression @xmath112 , \end{aligned}\ ] ] and the quark - quark - gluon - gluon vertex is @xmath113 \nonumber \\ & & + \frac{1}{2}\rho \ , \frac{1}{\omega(p_1 ) + \omega(p_2 ) }
\frac{1}{\omega(p_1 ) + \omega(k ) } \frac{1}{\omega(k ) + \omega(p_2 ) } \times \nonumber \\ & & \qquad \bigg[x_0(p_2 ) w_{1\mu}^\dagger ( p_2,k ) w_{1\nu } ( k , p_1 ) + w_{1\mu } ( p_2,k ) x_0^\dagger ( k ) w_{1\nu } ( k , p_1 ) \nonumber \\ & & \qquad \qquad + w_{1\mu } ( p_2,k ) w_{1\nu}^\dagger ( k , p_1 ) x_0 ( p_1 ) \nonumber \\ & & \qquad -\frac{\omega(p_1)+\omega(k)+\omega(p_2)}{\omega(p_1 ) \omega(k)\omega(p_2 ) } x_0 ( p_2 ) w_{1\mu}^\dagger ( p_2,k ) x_0 ( k ) w_{1\nu}^\dagger ( k , p_1 ) x_0 ( p_1 ) \bigg ] .
\nonumber\end{aligned}\ ] ]
we give here the analytic results for the leg tadpole and for the vertex of the scalar current , in the feynman gauge for @xmath75 . in order to be able to write them in a compact form
it is convenient to introduce the abbreviations @xmath114 @xmath115 @xmath116 @xmath117 the result for the 1-loop leg tadpole is then given by @xmath118 \nonumber \\ & & + g_0 ^ 2 \int \frac{d^4k}{2\pi^4 } g^2(k ) \frac{1}{\rho^2 ( 1+\sqrt{a})^2 } \times \nonumber \\ & & \qquad \sum_\lambda \bigg [ -2 s_\mu^2 n_\lambda^2 + \bigg(1+\frac{1}{\sqrt{a } } \bigg ) \bigg ( 2 s_\mu^2 ( b + 2 m_\mu^2 - m_\lambda^2 ) \bigg ) \bigg ] .
\nonumber \end{aligned}\ ] ] the first term comes from the part of the @xmath119 vertex ( [ eq : v2 ] ) containing @xmath120 and @xmath121 , and its value for @xmath97 is @xmath122 while the 1-loop result for the whole leg tadpole is smaller : @xmath123 adding now the value @xmath124 for the diagram e in fig .
1 , which is much harder to compute by hand and would be a very lengthy analytic expression anyway , gives the result @xmath125 for the complete self - energy in the feynman gauge for @xmath97 .
the results for the 1-loop vertex diagram of the scalar operator ( the sails are not present in this case ) can be written in the form @xmath126 , \ ] ] for small @xmath19 , where @xmath127 \nonumber \\ y & = & \sum_\lambda \bigg [ s^2 + \frac{1}{\rho \sqrt{a } } 2s^2 ( m_\lambda^2 + n_\lambda^2 ) \nonumber \\ & & \qquad + \frac{1}{\rho^2 a } s^2 \bigg ( -2 b ( m_\lambda^2 -n_\lambda^2 ) + s^2 -b^2 \bigg ) \bigg ] .\end{aligned}\ ] ] h. neuberger , phys .
* d57 * ( 1998 ) 5417 ; h. neuberger , phys .
rev . lett . * 81 * ( 1998 ) 4060 ; r. g. edwards , u. m. heller and r. narayanan , nucl . phys .
* b540 * ( 1999 ) 457 ; r. g. edwards , u. m. heller and r. narayanan , phys . rev .
* d59 * ( 1999 ) 094510 ; a. bode , u. m. heller , r. g. edwards and r. narayanan , `` first experiences with hmc for dynamical overlap fermions '' , hep - lat/9912043 ; p. hernndez , k. jansen and l. lellouch , `` a numerical treatment of neuberger s lattice dirac operator '' , hep - lat/0001008 ; r. narayanan and h. neuberger , `` an alternative to domain wall fermions '' , hep - lat/0005004 ; l. giusti , c. hoelbling and c. rebbi , in preparation .
s. capitani et al . , `` local bilinear operators on the lattice and their perturbative renormalisation including o(a ) effects '' , hep - lat/9711007 , in : `` deep inelastic scattering off polarized targets : theory meets experiment '' , proceedings of the spin97 topical workshop , j. blmlein , a. de roeck , t. gehrmann and w .- d .
nowak eds .
( desy , 1997 ) , p. 266 ; s. capitani et al . ,
* b * ( proc . suppl . )
* 63 * ( 1998 ) 874 .
s. capitani , m. gckeler , r. horsley , p. e. l. rakow and g. schierholz , phys . lett .
* b468 * ( 1999 ) 150 ; s. capitani , m. gckeler , r. horsley , p. e. l. rakow and g. schierholz , nucl . phys .
* b * ( proc . suppl . )
* 83 - 84 * ( 2000 ) 893 . | using the overlap - dirac operator proposed by neuberger , we have computed in lattice qcd the one - loop renormalization factors of ten operators which measure the lowest two moments of unpolarized and polarized non - singlet quark distributions .
these factors are necessary to extract physical numbers from monte carlo simulations made with overlap fermions .
an exact chiral symmetry is maintained in all our results , and the renormalization constants of corresponding unpolarized and polarized operators which differ by a @xmath0 matrix have the same value .
we have considered two lattice representations for each continuum operator .
the computations have been carried out using the symbolic language form , in a general covariant gauge . in some simple cases they have also been checked by hand .
mit - ctp-2978 + may 2000 |
the richness is an important defining attribute of a galaxy cluster , yet has always proven to be difficult to quantify rigorously .
abell ( 1958 ) , in creating the first comprehensive catalog of rich galaxy clusters , defined the richness classification which now bears his name based on the number of galaxies within 2 mag of the third ranking galaxy in an angular area of radius 1.7/@xmath5 , or about 3 mpc ( for @xmath6 km s @xmath7 mpc @xmath7 ) .
the procedure that abell used in counting the number of galaxies from palomar sky survey prints must be considered approximate , and difficult to extend to higher redshift .
the accuracy of the abell classification for abell clusters is accurate probably to only about 1 class ( e.g. , see dressler 1978 ) .
uncertainties can arise from many sources : inaccurate estimate of photometry , poor star - galaxy classification , incorrect background correction , contamination from foreground or background clusters , and large uncertainty in the estimated redshift .
in addition , the relatively large area used for the counting increases the chance of contaminating projections . at higher redshift , most of these problems would only be magnified , making a similar richness classification impractical without additional data such as accurate photometry and redshift .
furthermore , at redshifts higher than @xmath80.2 , it becomes increasingly more difficult , if not impossible , to specify , for example , the third brightest cluster member , due to the large number of foreground galaxies .
a robust richness measurement for galaxy clusters is an important observational quantity . beside being one of the primary attributes of a cluster , it can be used as a baseline for comparing other properties of clusters in a meaningful way .
this is especially important as advances in technology have allowed us to study galaxy clusters at different epochs and at increasingly higher redshifts , up to and above 1 ( e.g. , dickinson 1997 ) .
ultimately , a more fundamental parameter is the mass of the cluster .
however , there is evidence that the richness of a cluster , properly determined , can be a good predictor of the mass of the cluster ( carlberg et al .
1996 , and yee et al .
1999 ) based on results from the cnoc1 cluster redshift survey .
various attempts have been carried out to produce a more accurate measure of the richness of abell clusters .
these include the @xmath9 of bahcall ( 1981 ) , and the revised abell count numbers , @xmath0 , of abell , corwin , & olowin ( 1989 , hereafter aco ) .
to arrive at a robust and rigorous quantitative measure of cluster richness , three observational issues must be taken into account when counting the number of galaxies in a cluster : the angular coverage , the depth of the observation , and the background correction .
the availability of wide field digital imaging due to advances in large format ccds has enabled us to address these issues quantitatively much better .
the wide field allows us to sample the bulk of even very low redshift clusters , and digital imaging provides us with accurate photometry , which is vital for determining the depth of the galaxy counts , and correcting for background .
in addition , a useful quantitative richness measure is one that can be applied uniformly to different samples at vastly different redshifts .
the cluster - center galaxy correlation amplitude ( ) , first used by longair & seldner ( 1979 ) to quantify the environment of radio galaxies , is one parameter that appears to be well - suited as a robust richness parameter . by assuming an universal luminosity function ( lf ) and spatial profile ,
the excess galaxy counts around any reference point can be properly corrected to create an unbiased estimator of the richness of the environment .
yee & green ( 1987 ) extended the method to study the environment of quasars by using a `` universal '' lf with evolution that is self - consistent with the background galaxy counts .
the ability to track the evolution of the lf at different epochs is extremely important in comparing richness of clusters at different redshifts in a meaningful way .
however , although such a method has been used in many studies of the environments of active objects , fully digital photometric data have not been applied to the quantification and calibration of the richness of ordinary galaxy clusters , such as the abell clusters , and to test the robustness of the parameter . in this paper , we present the richness measurements , in the form of , for a sample of 47 abell clusters based on digital photometric data .
these measurements can be used as a calibration between the parameter and the more traditional abell richness class ( hereafter , arc ) .
we show that is a robust richness estimator , provided that a galaxy lf well matched to that found in clusters is used for normalization .
we find that there is a good correlation between the @xmath0 and for clusters at @xmath1 with a dispersion of about 1 richness class .
however , @xmath10 , @xmath0 often over - estimates the true richness of abell clusters . in 2 we present the sample and the data set .
the method for deriving is briefly discussed in 3 , along with the lfs used in the computation . in
4 we test the robustness of the parameter by investigating the dependence of the parameter on various quantities such as the lf , magnitude limit , and sampling area .
section 5 discusses the results and compare the arc with the parameter .
a summary is presented in 6 . throughout this paper , unless otherwise specified , we use @xmath6 km s@xmath7 mpc @xmath7 and @xmath11 .
the data used in this work are from a sample of abell clusters originally chosen for a study of correlations between x - ray and optical properties ( the low - redshift cluster optical survey ( locos ) , see lpez - cruz & yee 1999 ) .
the detailed description of the sample , observation , and data reduction can be found in lpez - cruz ( 1997 , hereafter lc97 ) . here
, we present a brief summary .
the clusters were selected from a compilation of bright x - ray clusters made by jones & foreman ( 1998 ) with the following criteria : ( 1 ) galactic latitude @xmath12 ; ( 2 ) @xmath13 ; ( 3 ) arc@xmath31 ; and ( 4 ) the declination @xmath14 .
although there is an imposed criterion of arc@xmath31 , data for a number of arc=0 clusters were also obtained , often for the purpose of filling right ascension gaps during the observing runs .
the arc=0 clusters are not considered as part of the fair sample .
this is because they were not picked randomly ; ones that appear to be rich or have well - defined x - ray emission were preferentially chosen . nevertheless , they are analysed along with the other clusters , but are used only when such a selection effect does not affect the particular conclusion . as a comparison , the central region of coma ( a1656 , @xmath15 )
was also observed .
the total number of abell clusters in the sample is 47 , with 36 satisfying the criteria listed above .
the cluster names , along with their redshifts and abell richness classes , are listed in table 1 .
direct images in @xmath16 , @xmath17 , and @xmath18 were obtained using the t2ka ccd camera at the 0.9 m telescope at kpno over five runs in 1992 and 1993 .
photometry was calibrated to the johnson - kron - cousin system of landolt ( 1992 ) .
the ccd has 2048@xmath192048 pixels with a scale of 0.68@xmath20 pixel@xmath7 , giving a field of view of [email protected]@xmath22 .
the integration times varied between 900 to 2500 second , depending on the filter used and the redshift of the cluster .
nine control fields at least @xmath23 away from any cluster fields were also observed for the purpose of background count correction .
these fields were observed and reduced in an identical manner as the cluster fields .
photometric catalogs of all objects in each field were created using the program ppp ( yee 1991 ) to perform automatic object finding , photometry , and galaxy - star classification .
the typical 100% completeness magnitude in @xmath17 is @xmath24 mag .
this provides a range of completeness in absolute @xmath17 magnitudes at the redshift of the clusters from 15.5 to 18.5 .
the cluster richness parameter is defined as the amplitude of the cluster center
galaxy correlation function : @xmath25 this parameter was first used by longair & seldner ( 1979 ) to quantify the richness of radio galaxy environment .
it was subsequently adopted by yee & green ( 1984 ) to study the environment of quasars using a self - consistent lf estimated from the same data .
further applications of the parameter to the study of the environments of quasars and radio galaxies can be found in yee & green ( 1987 ; hereafter yg87 ) ; ellingson , yee , & green ( 1991 ) ; prestage & peacock ( 1988 ) ; and yates , miller , & peacock ( 1989 ) .
this method has also been used to quantify the environment of bl lacertae objects ( smith , odea , & baum 1995 ; wurtz et al .
1997 ) , seyfert galaxies ( de robertis , yee , & hayhoe 1998 ) , and the richness of a small sample of abell clusters ( anderson & owen 1994 ) . detailed descriptions of the method adopted for this study can be found in yg87 , and also andersen & owen ( 1994 ) , among others . here , we provide a brief outline .
we observe galaxies in projection on the sky , which allows us to measure the angular two - point correlation function of galaxies , @xmath26 , as a function of the angle @xmath27 on the sky .
the function @xmath26 can be approximated by a power - law of the form ( e.g. , davis & peebles 1983 ) : @xmath28 where @xmath29 is the galaxy - galaxy angular correlation amplitude . for our purpose
, we can interpret @xmath26 as the distribution of galaxies projected on the sky around a reference point , such as the center of a galaxy cluster . in this case
, the amplitude , now notated as @xmath30 , can be measured directly from an image by counting the excess ( i.e. , background corrected ) galaxy counts , @xmath31 , up to a certain apparent magnitude around this reference point within some @xmath27 ; i.e. , if a fixed @xmath32 is assumed , then @xmath33 , where @xmath34 is the background counts within @xmath27 .
the amplitude can then be estimated via a deprojection of the angular correlation function into the spatial correlation function by assuming spherical symmetry , giving the relation between and @xmath30 ( see longair & seldner 1979 ) : @xmath35}},\eqno(3)\ ] ] where @xmath34 is the background galaxy counts to apparent magnitude @xmath36 ; @xmath37 $ ] is the integrated lf of galaxies up to the absolute magnitude @xmath38 corresponding to @xmath36 at the redshift of the cluster @xmath5 .
@xmath39 is an integration constant arising from the deprojection , ( with @xmath40 for @xmath41 ) , and @xmath42/q_0 ^ 2(1+z)^2\}$ ] is the angular diameter distance to @xmath5 .
as eqn ( 3 ) indicates , the computation of requires a knowledge of the luminosity function ( which provides a normalization ) and a background count correction .
these are described in the next two subsections .
the background counts used were obtained from the same cluster imaging program using the same telescope - ccd - filter combination and the same reduction procedure , which are described in detail in lc97 . briefly
, 9 fields with integration times similar to those of the cluster fields were obtained at least 5@xmath43 from any of the program clusters .
the control field counts were combined using the technique described in yee , green , & stockman ( 1986 ) .
figure 1 shows the counts .
these counts are in good agreement with the red band counts from other investigations ( e.g. , driver et al .
1994 , metcalfe et al .
1991 , and yee et al .
1986 ) . using background counts obtained in an identical manner as the cluster data removes one of the major systematic uncertainties in the determination of .
the choice of the lf used is a crucial one in the determination of .
the sensitivity of the value to the lf will be discussed in detail in 4.1 .
the assumption of a universal lf is not strictly correct , as field and cluster galaxies have different population mixes and evolution histories .
however , for the computation of using galaxies brighter than @xmath44 mag , the varying faint - end slope of cluster galaxy lf ( lpez - cruz et al .
1997 ) has little or no effect .
we choose to test three lfs derived from very different data sources and methods .
they are all represented functionally by the schechter function ( schechter 1976 ) .
the three functions are ( 1 ) the lf derived from the clusters themselves ( lpez - cruz & yee 1999 , hereafter the lcy lf ) ; ( 2 ) the low - redshift field galaxy lf from king & ellis ( 1985 , hereafter the ke lf ) ; and ( 3 ) the field galaxy lf from the cnoc2 redshift survey of intermediate redshift galaxies ( lin et al . 1999 , hereafter the cnoc2 lf ) .
the parameters of the lfs are listed in table 2 and the lfs are described in the following .
we will first discuss primarily the shape of the lfs used and defer the description of the normalization of the lfs to later in the section .
we derive the shape of the lcy lf using the cluster data themselves .
the derivation of the lf for this sample of clusters is described in detail in lc97 , lpez - cruz et al .
( 1997 ) , and lpez - cruz & yee ( 1999 ) .
the bright end is fitted to a schechter function with @xmath45 . the result ( lc97
, lpez - cruz & yee 1999 ) shows that the sample has @xmath46 , where the uncertainty is the dispersion of the distribution . however , cd clusters appear to have a larger range of @xmath47 which is apparently correlated with richness ( lc97 , lpez - cruz & yee 1999 ) . removing the cd clusters produces a @xmath48 .
we adopt an `` universal '' lf for the clusters with @xmath49 ( at a median redshift of 0.07 ) and @xmath45 . for deriving richness measurements for clusters at significant redshifts ,
the evolution of galaxies comes into play .
counting galaxies to a fixed absolute magnitude will in general over estimate the richness for high redshift clusters , as even mild passive evolution will mean effectively counting deeper into the lf . hence , following yg87 , we add a mild luminosity evolutionary term parametrized as @xmath50 mag , where @xmath51 , to @xmath47 to compensate for this effect .
( yg87 used a second order polynominal for @xmath47 evolution , we have simplified this to a single term in @xmath5 , since any such measurement is currently extremely approximate . ) there is good evidence that the lf in clusters evolves , to first order , in such a manner from galaxies associated with quasars ( yg87 ) , and from the cnoc1 cluster redshift survey ( yee et al .
1999 ) . since the median redshift for the lc97 clusters is 0.07 , we correct the @xmath47 to redshift 0 to @xmath52 . for a consistency check with previous applications of the @xmath16 parameters in analyses of the environments of quasars ( e.g. , yg97 ) , radio galaxies ( yates et al .
1989 ) , bl lacs ( wurtz et al . 1997 ) , and seyfert galaxies ( de robertis et al .
1998 ) , we also compare values using the field galaxy ke lf with its separate components for different morphological types .
the ke lf was derived as a preliminary result using a small number of low - redshift galaxies from the aat / durham redshift survey . here
, the @xmath47 values are transformed into the cousin @xmath17 band from the gunn @xmath53 band as listed in yg87 .
the same luminosity evolution as that for the lcy lf is incorporated into the ke lf .
the final lf used is the interim lf derived from the cnoc2 field galaxy redshift survey ( yee et al .
1997 , lin et al . 1999 ) .
the lf is determined based on about 2300 galaxies with redshifts between 0.1 and 0.55 , covering two patches on the sky .
the galaxies are separated into early , intermediate , and late spectral types based on 5-color photometry .
the lf for each component is then fitted to a schechter function with two evolution parameters , @xmath54 and @xmath55 , in the form of @xmath56 ( see lin et al .
1999 for details ) .
hence , the evolution of this lf is derived explicitly with @xmath55 measuring the luminosity evolution , and @xmath54 , the density evolution ( see table 2 ) .
an important parameter in the lf for the determination of is the normalization constant @xmath57 .
the value of is inversely proportional to it . as discussed in yg87 , the best way to minimize systematic effects in
is to determine a self - consistent lf with a normalization constant that reproduces the background counts as a function of apparent magnitude .
we hence renormalized @xmath57 by multiplying it by @xmath58 which is determined by fitting the modeled counts to the background counts .
we determine @xmath57 for the lcy lf directly by minimizing the @xmath59 between model counts generated by using the lf and the background count data . for the ke and cnoc2 lfs , the relative @xmath57 s for each component
are retained , but an overall normalization is applied based on the background count fitting .
we found that the renormalization of the cnoc2 lf , which is derived using the same photometry band , is small , less than 10% .
the modeled counts for the various lfs are shown in figure 1 .
the values obtained using the lcy lf are listed in table 1 .
the values obtained using the other two lfs are similar , with variations of less than @xmath810% .
a more detailed comparison of the dependence of on the lf is left to section 4.1 . for each cluster , we count galaxies down to a limit of @xmath60 in a circle of 0.5 mpc radius nominally using the brightest cluster galaxy ( bcg ) as the center of the cluster . however , we note that a few clusters have their bcg not sitting in the obvious center of the distribution of galaxies .
the most significant deviation occurs in a168 , where the bcg is almost isolated from the cluster . in this case
, we have chosen the brightest galaxy in the high density clump ( with the correct color ) to be the center of the cluster . choosing the bcg as the center for a168 would produce a almost 3 times smaller .
the uncertainty for the parameter is computed using the formula : @xmath61 where @xmath31 is the net counts of galaxies over the background of @xmath34 .
this is a conservatively large error estimate as it includes the expected counting statistics in @xmath31 and the expected dispersion in background counts .
the factor @xmath62 is included to account approximately for the additional fluctuation from the clustered ( and hence non - poissonian ) nature of the background counts ( see lc97 and yg87 ) .
this is a fair estimate of the uncertainty of the richness of the cluster , as it includes the poissonian uncertainty in the net counts , which is equivalent to the uncertainty in the number of galaxies drawn from a schechter function with a fixed normalization ( hence of fixed richness ) .
however , we note that the inverse of this error estimate is an underestimate of the statistical significance of the excess galaxy counts in the sample , which would be better represented by @xmath63 . hence , at these low redshifts , while the average uncertainty in the parameter is about 25% , the significance of the excess counts over background is considerably better at about the 20@xmath64 level .
in this section , we examine the results of exhaustive tests on the dependence of on various quantities , such as the lf shape and parameters , sampling and spatial limits , and photometric errors .
these tests not only provide a clear indication of the robustness of the parameter , but also the pitfalls in applying the method . a richness measurement based on the number of galaxies in a cluster is invariably tied to the assumption of the luminosity function of the galaxies .
if clusters have greatly different galaxy lfs , then any such richness measurement would be rendered meaningless . in this section ,
we test the sensitivity of to the lf used in converting @xmath30 to @xmath65 . for these tests
we vary the lf parameters but fix the sampling limit at @xmath60 mag .
this sampling limit was chosen with the expectation that it minimizes the effect of having an incorrect lf .
varying the lf with different sampling limits causes more complex behaviors ; these effects are tested in 4.2 .
we have computed using three lfs derived using very different data and method .
the lcy lf is based on the cluster lf measured from the same set of data , and describes cluster galaxies the best ; while the ke and cnoc2 lf are based on field galaxies at different redshifts .
the three give remarkably similar results with the mean of the ratio of ( lcy):(cnoc2):(ke ) being 1.00:1.10:0.92 , with the mean value for ( lcy ) being 1246@xmath66417 , where the uncertainty is the dispersion of the values in the sample .
this result demonstrates that the systematic uncertainty arising from the not - strictly - correct assumption of a `` universal '' lf for both cluster and field galaxies is of the order of 10% .
the ratios indicate that the cnoc2 lf is slightly fainter overall compared to the lcy lf , while the ke lf is somewhat brighter .
this indeed is the case when comparing the @xmath47 s of early type galaxies in the two multi - component lfs with that of the lcy lf . for the remainder of the paper
, we will use the values derived using the lcy lf as the fiducial set , and compare the results from the various tests to it .
first , we test the effect of an `` universal '' lf with incorrect parameters . in varying the @xmath47 and @xmath67 of the lf
, we also adjust the normalization constant of the lf , so that it reproduces the control field galaxy counts up to the magnitude of sampling . note
that this is an important step to make sure that the s are obtained with a self - consistent lf model .
this is different from testing the effects of statistical fluctuations of the lf of individual clusters from the mean for which one would not perform the renormalization ( see 4.1.3 ) . in these tests
we first adjust the lcy lf by altering @xmath47 in steps of @xmath660.25 mag .
each time we recompute the normalization constant for the lf by fitting the background counts .
we note that the fit to the background counts deteriorates with each step , but remains reasonable up to a @xmath68 of @xmath660.5 mag .
the results are summarized in table 3 .
the results for @xmath47 being off by @xmath660.25 mag are quite acceptable , causing variations of about 10% , similar to the effects of using the three different lfs .
this is not surprising , as the uncertainties in the three lfs are of the same order .
even when the error is 0.5 mag , the effect is about 20% , comparable to about 1/2 an abell richness class ( equivalent to @xmath69 of about 200 ; see section 5 ) .
we next alter the faint - end slope @xmath67 of the lf , again , with renormalization based on fitting the background . here
, we change @xmath67 by steps of 0.15 .
the results are tabulated in table 4 .
again , one sees that changing the slope by as much as @xmath70 alters the values by less than 20% .
these tests , along with the results from using three different lfs , demonstrate that a lf which self - consistently reproduces the background counts can be substantially incorrect and still produces results that are accurate to better than 1/2 an abell class . in this section ,
we test the effect of statistical fluctuations from the mean in the lf of the cluster galaxies . here , we vary @xmath71 and @xmath67 , but do not adjust the normalization constant of the lf .
note that in the case of @xmath71 , it has the same effect as a photometric error of the same magnitude in the cluster data .
we use the same step sizes for @xmath38 and @xmath67 as in 4.1.2 , and the results are listed in tables 3 and 4 . here , we see that the effects are significantly larger .
specifically , a fluctuation in @xmath47 ( or equivalently , photometric error in the data ) of about 0.4 mag will produce a 20% ( 1/2 abell class ) effect .
however , a fluctuation of 0.2 mag would still be quite acceptable , producing uncertainties of only about 10% .
since cluster @xmath47 s have a dispersion of about 0.2 mag , this indicates that the instrinsic dispersion of the lf from cluster to cluster does not affect significantly .
similar fluctuations due to variations in @xmath67 of up to 0.3 are also seen .
the independence of values on galaxy counting limiting magnitude is an important issue , especially when different fields are observed to different depths .
if the correct lf is chosen for the computation of and the lf is truly universal , then should subject only to counting statistical variations when different limiting magnitudes are used . a substantial systematic variation in with limiting magnitude
can be taken as an indication that a lf with the incorrect shape has been chosen .
moreover , for the purpose of producing robust results , there is an optimal range of absolute magnitudes for galaxy counting .
counting to too bright a limit will produce a large uncertainty from counting statistics .
furthermore , one is at the mercy of small intrinsic variations of @xmath47 in individual clusters , or equivalently , small photometric zero point error in the galaxy photometry . on the other hand
, a larger uncertainty may also result from counting too deep into the lf , as the background counts rise more rapidly than the flat part of the cluster lf .
in addition , in counting beyond @xmath72 one may encounter large variations in the faint end slopes of the cluster lf ( lpez - cruz et al .
1997 and lcy ) , producing additional fluctuations . ideally , one would like the richness parameter of a galaxy cluster to be representative of the total mass of the cluster , which , under the light - traces - mass scenario ( e.g. , tyson & fischer 1996 ; carlberg , yee , & ellingson 1997 ) , should be proportional to the total light .
hence , the richness of a cluster is best represented by the number of @xmath73-like galaxies , rather than by the number of dwarf galaxies ; consequently , a robust measure of is best obtained using a magnitude limit corresponding to the flat part of the lf : from about 1 mag to 3 mag past @xmath47 .
the data in our sample of abell clusters sample to at least @xmath74 mag for all the clusters ; hence , they provide a good test on the robustness of versus counting magnitude , and the effects of varying the lf . for these tests , we compute the parameters using @xmath75 , 21.0 , -20.0 , 19.0 , and 18.0 for comparison with the standard limit of 20 ( i.e. , about 2 mag past @xmath47 ) . the five derivations provide ratios of average of : 1.04:1.05:1.00:1.00:1.07 .
a comparison of the individual s using sampling limits of 21.0 and 20.00 is shown in figure 2 .
this test shows very strongly that , as expected , when a lf closely resembling that of the cluster galaxy lf is used , the sampling magnitude limit has little or no systematic effect on the derived richness .
the variations in individual values are the order of the computed uncertainties , and are due to counting statistics .
we note that the largest deviation occurs with the deepest sampling limit .
this is expected as the steepening of the galaxy lf for some clusters begins at about @xmath76 , which would produce a larger count relative to the flat ( @xmath77 ) lf used .
next , we consider the effects of the combination of lf variations and more extreme sampling limits .
we focus on testing the effect of incorrect lf and photometric errors when sampling to only about @xmath47 , a situation that may arise often when data which are not quite deep enough are pressed into use .
the results are shown in table 3 . in general , at a sampling limit of @xmath78 mag ( about @xmath47 ) , the errors committed when @xmath47 is varied with renormalization of the lf ( equivalent to using an incorrect lf ) are about twice as large as those when the sampling is to @xmath82 mag past @xmath47 ( about 20 mag ) . however
, when no renormalization is performed , which is also equivalent to photometric uncertainty , the errors incurred when counting to 22 mag can be as large as 4 times those to 20 mag .
for example , a + 0.5 mag change in @xmath47 produces values 2.2 times larger when counting to 22 mag , compared to 1.3 times larger when counting is done to 20 mag .
we note that the results for changing @xmath67 are not nearly as drastic , as that affects mainly the faint end of the lf ( see table 4 ) .
we hence conclude that counting to about @xmath47 produces results that are extremely sensitive to errors in the assumed lf , photometry , and in general statistical fluctuation , with results that could be off by a factor of 1.5 to over 2 when the error in magnitude is 0.25 to 0.5 mag .
this may explain the large discrepancies noted in the values of bl lac objects between smith , odea , & baum ( 1995 ) and wurtz et al .
( 1997 ) , as noted by wurtz et al .. we also note that this may also be the possible explanation of the discrepancies of between the richness derived for abell clusters by andersen & owen ( 1994 ) and this work ( see 5.1 ) .
ideally , a robust richness parameter is one such that the counting area used should not have a significant effect on the richness derived .
such a parameter would allow comparison of the richness parameter from data taken with a wide variety of areal coverage .
this is strictly possible only if the spatial distribution function of cluster galaxies has a similar form for all clusters .
if an incorrect slope ( @xmath32 ) for the correlation function is used , the value will be dependent on the sampling area . in the computation of it
is assumed that the galaxy distribution drops off in the same fashion as the spatial correlation function for field galaxies .
this assumption is adopted so that comparisons over a wide range of richness and environments , including that of field galaxies , can be made .
however , this assumption is most likely incorrect for rich clusters , as various investigations have shown that the power - law slope of their galaxy density distribution is probably greater than 2 ( e.g. , lilje & efstathiou 1988 , pebbles 1993 ) , somewhat steeper than that of the clustering of field galaxies .
the effect of using an inexact @xmath32 can be examined by comparing the values for our sample using different limiting radii .
we have computed values ( with the lcy lf ) using sampling radii of 0.25 , 0.5 , 1.0 mpc , all with a @xmath79 sampling limit of 20 mag .
the ratio of the mean s for the 3 radii are : 1.03:1.00:0.94 .
the results for individual clusters with sampling radii of 0.5 and 1.0 mpc are shown in figure 3 .
the results indicate that a change of a factor of 2 in the sampling radius causes a scattering of about 5% in the richness parameter .
this small effect is reassuring , and provides confidence when comparing data sampled over metric field sizes of a factor of several , as in those used by de robertis et al .
( 1998 ) for nearby seyfert galaxies .
the decreasing trend in the with larger radii is consistent with existing evidence that cluster galaxy distribution is likely to be steeper than the slope of the field galaxy correlation function .
however , since the effect is very small , we have chosen to retain the canonical @xmath80 so that direct comparisons with field and group clustering amplitudes can be made .
the various tests in section 4 have established the robustness of the parameter as a measure of cluster richness . in this section ,
we compare the parameter with the abell richness class , establishing a calibration between the two .
we also discuss the correlation between and the abell count number , which allows us to examine the accuracy of the abell count number as a measure of richness . based on these comparisons
, we suggest a revised richness classification scheme based on the calibration between the parameter and arc .
finally , we illustrate the advantage of using a robust richness parameter in the investigation of cluster properties by examining the correlation between richness and velocity dispersion .
one of the main goal of this investigation is to establish a calibration between a more quantitative richness parameter , such as , and the traditional abell richness class . in figure 4 , we plot the parameters versus arc .
we see in general a very broad correlation between and arc .
we note that because we have specifically chosen to image arc 0 clusters which appear to be the richest , they should not be considered in the correlation .
andersen & owen ( 1994 ) produced a qualitatively very similar plot of the two quantities using a different sample and very different type of data .
however , quantitatively , there is a significant discrepancy in the scaling of the two quantities : their values are close to a factor of two lower than ours ( after correcting for @xmath81 ) , with their average for arc 1 clusters being @xmath82 mpc@xmath83 , compared to about 1000 mpc@xmath83 in our work .
this discrepancy is perhaps not surprising , given that : ( 1 ) andersen & owen use poss photographic data which may not be well calibrated , ( 2 ) their object counting goes down to only @xmath84 , and ( 3 ) they did not use a lf that has been renormalized to the background .
all these can contribute substantially to the systematic uncertainty , up to a factor of 2 , as demonstrated by our tests in 4 .
since the arc is an approximate and quantized measure , we will next look at the comparison with actual abell counts ( @xmath0 ) before revisiting the correlation between and arc in 5.3 .
a more detailed comparison between the arc and the measurement can be obtained by using the abell number counts ( @xmath0 ) as published by aco .
we plot @xmath0 vs in figure 5 . taken as a whole ,
again we see a definite correlation between and @xmath0 but with a large dispersion .
specifically , it is noted that at large @xmath0 the correlation breaks down in that their corresponding values are smaller than indicated by @xmath0 .
we also note that all the clusters in our sample with large @xmath0s are those at higher redshifts .
we hence break the sample into two groups , dividing them at @xmath85 . to increase the @xmath86 sample
, we also add 3 clusters from the cnoc1 survey ( yee et al .
1999 ) which are abell clusters at @xmath87 .
these are a2390 , ms0451 + 02 ( a520 ) , and ms0906 + 11 ( a750 ) .
removing the high redshift sample provides a much improved correlation bewteen and @xmath0 .
for the lower @xmath5 sample , though the scatter is about one abell class , there are few or no outliers . for the higher @xmath5 sample ,
there is almost no correlation between and @xmath0 .
the discrepancies come entirely in the high @xmath0 clusters which all have too large a @xmath0 for their .
it is interesting to note that a665 , which is the only abell cluster classified as arc 5 , with @xmath88 , has a value which indicates that its richness is overestimated by a factor of 3 by @xmath0 .
a number of abell clusters in fact have similar richness to a665 , including a2390 and a401 .
hence , we conclude that at @xmath1 , the aco counts provide reasonable estimates of the richness of the clusters ; however , at @xmath2 , and/or for @xmath89 , the @xmath0 counts almost always over predict the cluster s richness .
this is not a surprising result , as many authors have concluded that the abell catalog becomes incomplete at @xmath2 ( e.g. , southerland 1988 ; struble & rood 1991b ) . using figures 4 or 5
, we can relate the more robust measurements directly to the abell richness class , in essence , calibrating the classical richness class designations .
the mediam value of arc 1 clusters in our sample is very close to 1000 17 .
hence , we propose to use 1000 17 as the fiducial value for the average arc 1 counts .
abell ( 1958 ) and aco used a non - linear relationship between the abell counts and the richness class with the higher arc having disproportionately higher counts , presumably to account for the few apparently very high @xmath0 clusters .
however , 5.2 demonstrates that the very high @xmath0 clusters are probably much less rich .
if we scale the values corresponding to each arc counts using the fiducial of 1000 17 as arc 1 , we see that there is a significant mismatch at arc@xmath33 , as illustrated by the dotted line in figure 4 , with arc 5 being equivalent to @xmath90 .
yee et al .
( 1999 ) also found that none of the 16 clusters from the cnoc1 survey , which contains the most x - ray luminous clusters from the emss survey ( see gioia & luppino 1994 ) , have larger than @xmath91 mpc @xmath83 .
hence , there appears to be no known clusters that approach the original definition of arc 5 . in view of this
, we propose a linear richness scale where adjacent richness classes change by the same increment of 400 17 in .
this scale would cover most of the range of known cluster richness ( including very rich , x - ray luminous , higher redshift clusters , such as those in the cnoc1 sample ) , with the first 3 classes ( 0 , 1 , and 2 ) matching the original arc definition reasonably well .
this revised classification definition is shown graphically in figure 4 . for @xmath6
km s@xmath7 mpc @xmath7 , the calibration for arc 0 to 5 is then : 600@xmath66200 , 1000@xmath66200 , 1400@xmath66200 , 1800@xmath66200 , 2200@xmath66200 , and @xmath92 17 , respectively .
we note that this calibration is substantially different from those provided by prestage & peacock ( 1988 ) based on a very small number of 4c radio galaxies situated in abell clusters .
their calibration of arc 0 and 1 having of @xmath8380 and 850 17 is about 1/3 to 1/2 of an abell class too high .
hence , previous descriptions of the abell class equivalence for the environments of radio galaxies and quasars ( e.g. , yg87 ) need to be revised correspondingly downward .
so far we have shown that a richness parameter such as , which uses photometric data and normalized by the lf and spatial distribution of galaxies , produces a robust measurement of the richness of a cluster .
one would then expect that such a robust parameter will provide a better description of the clusters , and hence produce more meaningful correlations with other cluster properties .
we will show , as a demonstration of the improvement expected , an example of such a correlation .
one fundamental cluster property is the mass of a cluster , which can be estimated by the velocity dispersion ( @xmath64 ) . in figure 6
we plot the correlation of @xmath64 versus .
the velocity dispersion measurements are taken from the compilation of struble & rood ( 1991a ) , and fadda et al .
we include only those clusters with 10 or more velocity measurements .
there is a clear correlation between @xmath64 and , and the correlation is much tighter than using @xmath0 as the variable for richness which is shown in figure 7 .
girardi et al .
( 1993 ) showed a similar improvement in the correlation between velocity dispersion and richness when bahcall s ( 1981 ) @xmath9 parameter is used instead of the abell counts .
this improved correlation lends support that is a more robust richness measurement .
the @xmath64- correlation derived from the cnoc1 cluster redshift survey ( yee et al .
1999 ) provides an even tighter correlation between the two quantities .
this is almost certainly due to the fact that the velocity dispersion measurements from cnoc1 are derived from a homogeneous set of velocity data with a large number of velocities for each cluster .
we note that the @xmath64- relationships derived from the two different samples ( this work and cnoc1 ) are identical within the uncertainty .
the @xmath64 relationship from cnoc1 is plotted as a dotted line in figure 6 .
given the different photometric systems used ( kron - cousin @xmath17 vs gunn @xmath53 ) , the different redshift range of the sample , and the different instrumentations used for these two samples , this again points to the robustness of the measurements as a richness estimator .
we have used a sample of 47 abell clusters with photometric catalogs from ccd images to test the robustness of as a parameter for measuring the richness of galaxy clusters .
we first tested the effect of the choice of galaxy lf , which is used to normalize the excess galaxy counts . using three lfs derived from different sources , including both cluster and field galaxies
, we found that as long as we adjust the density normalization constant of the lf so that it reproduces the background counts used to derive the excess counts , different choices of the lf produce variations in of approximately 10% .
furthermore , the parameters for the lf can be substantially incorrect ( e.g. , with @xmath47 being off by 0.5 mag ) , errors in can also be minimized by renormalizing the lf to fit the background counts .
we also found that if a correct lf is used , counting to different absolute magnitude limits produces systematic variations of only a few percent
. however , to minimize uncertainties introduced by lf discrepancy and photometry errors , the results are most reliable when the counting is done to about 1 to 2 mag past @xmath47 .
similarly , the canonical @xmath80 for the spatial correlation function of field galaxies provides reasonably stable values as one varies the radius of the counting aperture . altering the counting radius from 0.25 to 1.5
mpc produces systematic changes in of less than 10% .
we have compared our richness parameters with the abell richness class and the abell count number ( @xmath0 ) . limiting the sample to either arc@xmath932 or @xmath1
, we found a good correlation between @xmath0 and .
however , clusters designated as arc@xmath33 appear often to have their richness over estimated , sometimes by as much as a factor of 3 .
since most of the arc@xmath33 clusters are at @xmath10 , this suggests that arc classifications for clusters with @xmath2 is not a good indicator of their true richness .
we suggest a revised richness classification which is based on a linear scale in .
finally , we demonstrate the improvement in the correlation between richness and velocity dispersion when a robust richness parameter , like , is used .
we wish to thank kpno for the generous allotment of telescope time and assistance in carrying out the low - redshift cluster optical survey ( locos ) .
hy acknowledges nserc for financial support in the form of an operating grant .
olc was supported in part by an overseas scholarship by conacyt - mxico and inaoe .
olc also acknowledges financial support from the department of astronomy and the school of graudate studies of the university of toronto , nserc through hy s operating grant , and conacyt - mxico through a `` catedra de repatriacin '' and a `` proyecto de investigacin inicial . ''
abell , g. 1958 , apjs , 3 , 211 abell , g. , corwin , h.g . , & olowin , r.p .
1989 , apjs , 70 , 1 ( aco ) anderson , v. & owen , f. 1994 , aj , 108 , 361 bahcall , n. 1981 , apj , 247 , 787 carlberg , r.g . ,
yee , h.k.c .
, & ellingson , e. 1997 , apj , 478 , 462 carlberg , r.g . ,
yee , h.k.c . ,
ellingson , e. , abraham , r.g . ,
gravel , p. , morris , s.l . , & pritchet , c.j .
1996 , apj , 462 , 32 davis , m. & peebles , p.j.e . 1983 , apj , 294 , 70 de robertis , m.m . ,
yee , h.k.c .
, & hayhoe , k. 1998 , apj , 496 , 93 dickinson , m.e .
1997 , in the hubble space telescope and the high redshift universe , eds n. tanvir et al . , ( world scientific : singapore ) , page 207 dressler , a. 1978 , apj , 226 , 55 driver , s.p . , phillipps , s. , davies , j.i . ,
morgan , i. , & disney , m.j .
1994 , mnras , 266 , 155 .
ellingson , e. , yee , h.k.c .
, & green , r.f . 1991 , apj , 371 , 49 fadda , d. , girardi , m. , giuricin , g. , mardirossian , f. , & mezzetti , m. 1996 , apj , 473 , 670 gioia , i.m .
& luppino , g.a .
1994 , apjs , 94 , 583 girardi , m. , biviano , a. , giuricin , g. , mardirossian , f. , & mezzetti , m. 1993 , apj , 404 , 38 jones , c. & forman w. 1998 , preprint king , c. , & ellis , r.s . 1984 , apj , 288 , 456 ( ke ) landolt , a. 1992 , aj , 104 , 340 lilje , p.b . &
efstathiou , g. 1988 , mnras , 231 , 635 lin , h. , yee , h.k.c . ,
carlberg , r.g .
, morris , s. , sawicki , m. , patton , d. , wirth , g. , & shepherd , c.w . 1999 ,
apj , in press longair , m.s . , & seldner , m. 1979 , mnras , 189 , 433 lpez - cruz , o. 1997 , ph . d. thesis , university of toronto ( lc97 ) lpez - cruz , o. , & yee , h.k.c . 1999 , to be submitted to apj ( lcy ) lpez - cruz , o. , yee , h.k.c . , brown , j.p . , jones , c. , & forman , w. 1996 , apj , 476 , l97 metcalfe , n. , shanks , t. , fong , r. & jones , l. r. 1991 , mnras , 249 , 498 peebles , p.j.e . 1993 , _ principles of physical cosmology _ , ( princeton university press : princeton ) , page 471 prestage , r.m . & peacock , j.a .
1988 , mnras , 230 , 131 smith , e.p .
, odea , c.d . , & baum , s. 1995 , apj , 441 , 113 schechter , p.l .
1976 , apj , 203 , 297 southerland , w. 1988 , mnras , 234 , 159 struble , m.f . &
rood , h.j .
1991a , apjs , 77 , 363 struble , m.f . & rood , h.j
. 1991b , apj , 374 , 398 tyson , j.a .
& fischer , p. 1996 , apj , 446 , l55 wurtz , r. , stocke , j.t . , ellingson , e. , yee , h.k.c .
1997 , apj , yates , m.g . , miller , l. , & peacock , j.a .
1989 , mnras , 240 , 129 yee , h.k.c .
1991 , pasp , 103 , 396 yee , h.k.c . , ellingson , e. , morris , s. , & carlberg , r.g .
1999 , preprint to be submitted to apj yee , h.k.c . , &
green , r.f .
1984 , apj , 280 , 79 yee , h.k.c . , & green , r.f .
1987 , apj , 319 , 28 ( yg87 ) yee , h.k.c . ,
green , r.f . , &
stockman , h.s .
1986 , apjs , 62 , 681 yee , h.k.c .
1997 , in the proceedings of the iau joint discussion 11 : redshift surveys in the 21st century , ed .
a. fairall , in press ( astro - ph/9710356 ) a21 & 0.0946 & 1 & 1509 & 241 a84 & 0.1030 & 1 & 1000 & 204 a85 & 0.0518 & 1 & 757 & 174 a98 & 0.1043 & 3 & 1921 & 268 a154 & 0.0638 & 1 & 1393 & 230 a168 & 0.0452 & 2 & 973 & 192 a399 & 0.0715 & 1 & 1449 & 235 a401 & 0.0748 & 2 & 2242 & 286 a407 & 0.0472 & 0 & 1368 & 225 a415 & 0.0788 & 1 & 534 & 159 a514 & 0.0731 & 1 & 946 & 196 a629 & 0.1380 & 1 & 1205 & 224 a646 & 0.1303 & 0 & 858 & 196 a665 & 0.1816 & 5 & 2186 & 290 a671 & 0.0491 & 0 & 1266 & 217 a690 & 0.0788 & 1 & 701 & 175 a957 & 0.0437 & 1 & 1075 & 201 a1213 & 0.0469 & 1 & 931 & 189 a1291 & 0.0530 & 1 & 1122 & 207 a1413 & 0.1427 & 3 & 1686 & 257 a1569 & 0.0784 & 0 & 768 & 181 a1650 & 0.0845 & 2 & 1861 & 263 a1656 & 0.0232 & 2 & 1242 & 282 a1775 & 0.0700 & 2 & 1018 & 202 a1795 & 0.0621 & 2 & 1430 & 233 a1913 & 0.0530 & 1 & 954 & 192 a1983 & 0.0430 & 1 & 974 & 192 a2029 & 0.0768 & 2 & 1736 & 255 a2244 & 0.0997 & 2 & 1698 & 254 a2255 & 0.0800 & 2 & 2296 & 289 a2256 & 0.0601 & 2 & 2174 & 281 a2271 & 0.0568 & 0 & 645 & 164 a2328 & 0.1470 & 2 & 1941 & 273 a2356 & 0.1161 & 2 & 944 & 201 a2384 & 0.0943 & 1 & 1575 & 245 a2399 & 0.0587 & 1 & 906 & 190 a2410 & 0.0806 & 1 & 699 & 175 a2415 & 0.0597 & 0 & 902 & 190 a2420 & 0.0838 & 2 & 1230 & 220 a2440 & 0.0904 & 0 & 1053 & 207 a2554 & 0.1108 & 3 & 1218 & 222 a2556 & 0.0865 & 0 & 828 & 187 a2593 & 0.0421 & 0 & 1211 & 212 a2597 & 0.0825 & 0 & 665 & 172 a2626 & 0.0573 & 0 & 945 & 193 a2657 & 0.0414 & 1 & 740 & 170 a2670 & 0.0761 & 3 & 1771 & 257 lcy & cluster & 0.00226 & 1.0 & 1.00 & 22.20 & 1.4 & 0.0 & & & & & & & ke & e / s0 & 0.000346 & 1.79 & 1.00 & 22.45 & 1.4 & 0.0 & sab & 0.000346 & 1.87 & 1.00 & 22.14 & 1.4 & 0.0 & sbc & 0.000346 & 1.87 & 1.00 & 21.81 & 1.4 & 0.0 & scd / sdm & 0.000346 & 3.24 & 1.00 & 21.29 & 0.7 & 0.0 & & & & & & & cnoc2 & early & 0.92 & 0.00212 & 0.09 & 21.52 & 1.94 & 2.02 & intermediate & 0.92 & 0.00111 & 0.69 & 21.84 & 0.62 & 0.59 & late & 0.92 & 0.00053 & 1.45 & 21.44 & 0.88 & 2.68 | using photometric catalogs from wide - field ccd images , we derive the cluster - galaxy correlation amplitudes , , for 47 low - redshift abell clusters .
we apply a number of tests to establish the robustness of the parameter as a quantitative measure of richness of galaxy clusters . these include using different galaxy luminosity functions to normalize the excess galaxy counts , counting galaxies to different absolute magnitude limits , and counting galaxies to different cluster - centric radii .
these tests show that with a properly normalized luminosity function , the parameter is relatively insensitive ( at better than 1/2 of an abell richness class ) to magnitude limit , areal coverage , and photometric errors of up to about 0.25 mag .
we compare the values to both the abell richness class ( arc ) and abell count numbers ( @xmath0 ) .
it is found that there is a good correlation between and @xmath0 for abell clusters with @xmath1 , with a dispersion of about one arc ; whereas the abell richness classifications for abell clusters at @xmath2 are much less well correlated with the true cluster richness .
we also find evidence that the richness of arc@xmath33 clusters has the tendency of being over estimated in the abell catalog . _ to appear in the astronomical journal , may 1999
_ = cmr7 = cmr10 17mpc@xmath4 |
most of modern equity exchanges are organized as _ order driven _ markets . in such type of markets , the price formation exclusively results from operating a _ limit order book _ ( lob ) , an order crossing mechanism where _ limit orders _ are accumulated while waiting to be matched with incoming _ market orders_. any market participant is able to interact with the lob by posting either market orders or limit orders is an order to buy ( sell ) @xmath0 units of the asset being traded at the lowest ( highest ) available price in the market , its execution is immediate ; a limit order of size @xmath1 at price @xmath2 is an order to buy ( sell ) @xmath1 units of the asset being traded at the specified price @xmath2 , its execution is uncertain and achieved only when it meets a counterpart market order . given a security , the _ best bid _ ( resp . _
ask _ ) price is the highest ( resp .
lowest ) price among limit orders to buy ( resp . to sell ) that are active in the lob .
the _ spread _ is the difference , expressed in numraire per share , of the best ask price and the best bid price , positive during the continuous trading session ( see @xcite ) . ] . in this context , _ market making _ is a class of strategies that consists in simultaneously posting limit orders to buy and sell during the continuous trading session . by doing so , market makers provide counterpart to any incoming market orders : suppose that an investor @xmath3 wants to sell one share of a given security at time @xmath4 and that an investor @xmath5 wants to buy one share of this security at time @xmath6 ; if both use market orders , the economic role of the market maker @xmath7 is to buy the stock as the counterpart of @xmath3 at time @xmath4 , and carry until date @xmath8 when she will sell the stock as a counterpart of @xmath5 .
the revenue that @xmath7 obtains for providing this service to final investors is the difference between the two quoted prices at ask ( limit order to sell ) and bid ( limit order to buy ) , also called the market maker s spread .
this role was traditionally fulfilled by specialist firms , but , due to widespread adoption of electronic trading systems , any market participant is now able to compete for providing liquidity . moreover , as pointed out by empirical studies ( e.g. @xcite,@xcite ) and in a recent review @xcite from amf , the french regulator , this renewed competition among liquidity providers causes reduced effective market spreads , and therefore reduced indirect costs for final investors .
empirical studies ( e.g. @xcite ) also described stylized features of market making strategies . first , market making is typically not directional , in the sense that it does not profit from security price going up or down .
second , market makers keep almost no overnight position , and are unwilling to hold any risky asset at the end of the trading day . finally , they manage to maintain their _ inventory _ , i.e. their position on the risky asset close to zero during the trading day , and often equilibrate their position on several distinct marketplaces , thanks to the use of high - frequency order sending algorithms .
estimations of total annual profit for this class of strategy over all u.s .
equity market were around @xmath9 g$ in 2009 @xcite .
popular models of market making strategies were set up using a risk - reward approach .
three distinct sources of risk are usually identified : the inventory risk , the adverse selection risk and the execution risk . the _ inventory risk _
@xcite is comparable to the market risk , i.e. the risk of holding a long or short position on a risky asset .
moreover , due to the uncertain execution of limit orders , market makers only have partial control on their inventory , and therefore the inventory has a stochastic behavior .
the _ adverse selection risk _ , popular in economic and econometric litterature , is the risk that market price unfavourably deviates , from the market maker point of view , after their quote was taken .
this type of risk appears naturally in models where the market orders flow contains information about the fundamental asset value ( e.g. @xcite ) . finally , the _ execution risk _ is the risk that limit orders may not be executed , or be partially executed @xcite .
indeed , given an incoming market order , the matching algorithm of lob determines which limit orders are to be executed according to a price / time priority , and this structure fundamentally impacts the dynamics of executions .
some of these risks were studied in previous works . the seminal work @xcite provided a framework to manage inventory risk in a stylized lob .
the market maker objective is to maximize the expected utility of her terminal profit , in the context of limit orders executions occurring at jump times of poisson processes .
this model shows its efficiency to reduce inventory risk , measured via the variance of terminal wealth , against the symmetric strategy .
several extensions and refinement of this setup can be found in recent litterature : @xcite provides simplified solution to the backward optimization , an in - depth discussion of its characteristics and an application to the liquidation problem . in @xcite ,
the authors develop a closely related model to solve a liquidation problem , and study continuous limit case .
the paper @xcite provides a way to include more precise empirical features to this framework by embedding a hidden markov model for high frequency dynamics of lob .
some aspects of the execution risk were also studied previously , mainly by considering the trade - off between passive and aggressive execution strategies . in @xcite ,
the authors solve the merton s portfolio optimization problem in the case where the investor can choose between market orders or limit orders ; in @xcite , the possibility to use market orders in addition to limit orders is also taken into account , in the context of market making in the foreign exchange market .
yet the relation between execution risk and the microstructure of the lob , and especially the price / time priority is , so far , poorly investigated . in this paper
we develop a new model to address these three sources of risk .
the stock mid - price is driven by a general markov process , and we model the market spread as a discrete markov chain that jumps according to a stochastic clock . therefore , the spread takes discrete values in the price grid , multiple of the tick size . we allow the market maker to trade both via limit orders , which execution is uncertain , and via market orders , which execution is immediate but costly .
the market maker can post limit orders at best quote or improve this quote by one tick . in this last case , she hopes to capture market order flow of agents who are not yet ready to trade at the best bid / ask quote . therefore
, she faces a trade off between waiting to be executed at the current best price , or improve this best price , and then be more rapidly executed but at a less favorable price .
we model the limit orders strategy as continuous controls , due to the fact that these orders can be updated at high frequency at no cost . on the contrary , we model the market orders strategy as impulse controls that can only occur at discrete dates .
we also include fixed , per share or proportional fees or rebates coming with each execution .
execution processes , counting the number of executed limit orders , are modelled as cox processes with intensity depending both on the market maker s controls and on the market spread . in this context , we optimize the expected utility from profit over a finite time horizon , by choosing optimally between limit and market orders , while controlling the inventory position
. we study in detail classical frameworks including mean - variance criterion and exponential utility criterion .
the outline of this paper is as follows . in section 2 ,
we detail the model , and comment its features .
we also provide direct calibration methods for all quantities involved in our model .
we formulate in section 3 the optimal market making control problem and derive the associated hamilton - jacobi - bellman quasi variational inequality ( hjbqvi ) from dynamic programming principle .
we show how one can reduce the number of state variables to the inventory and spread for the resolution to this qvi in two standard cases . in the last section 4 , we provide some numerical results and empirical performance analysis .
let us fix a probability space @xmath10 equipped with a filtration @xmath11 @xmath12 @xmath13 satisfying the usual conditions .
it is assumed that all random variables and stochastic processes are defined on the stochastic basis @xmath14 .
the mid - price of the stock is an exogenous markov process @xmath15 , with infinitesimal generator @xmath16 and state space @xmath17 .
for example , @xmath15 is a lvy process ( e.g. an arithmetic brownian motion ) , or an exponential of lvy process ( e.g. geometric brownian motion ) . in the limit order book ( lob ) for this stock
, we consider a stochastic bid - ask spread resulting from the behaviour of market participants , taking discrete values , which are finite multiple of the tick size @xmath18 @xmath19 @xmath20 , and jumping at random times .
this is modelled as follows : we first consider the tick time clock associated to a poisson process @xmath21 with deterministic intensity @xmath22 , and representing the random times where the buy and sell orders of participants in the market affect the bid - ask spread .
we next define a discrete - time stationary markov chain @xmath23 , valued in the finite state space @xmath24 @xmath12 @xmath25 , @xmath26 @xmath27 @xmath28 , @xmath0 @xmath29 @xmath30 , with probability transition matrix @xmath31 , i.e. @xmath32 $ ] @xmath12 @xmath33 , s.t . @xmath34 @xmath12 @xmath20 , independent of @xmath35 , and representing the random spread in tick time . the spread process @xmath36 in calendar time
is then defined as the time - change of @xmath37 by @xmath35 , i.e. [ dyns ] s_t & = & s_n_t , t 0 .
hence , @xmath36 is a continuous time ( inhomogeneous ) markov chain with intensity matrix @xmath38 @xmath12 @xmath39 , where @xmath40 @xmath12 @xmath41 for @xmath42 , and @xmath43 @xmath12 @xmath44 .
we assume that @xmath45 and @xmath15 are independent .
the best - bid and best - ask prices are defined by : @xmath46 @xmath12 @xmath47 , @xmath48 @xmath12 @xmath49 .
we consider an agent ( market maker ) , who trades the stock using either limit orders or market orders .
she may submit limit buy ( resp .
sell ) orders specifying the quantity and the price she is willing to pay ( resp .
receive ) per share , but will be executed only when an incoming sell ( resp .
buy ) market order is matching her limit order .
otherwise , she can post market buy ( resp .
sell ) orders for an immediate execution , but , in this case obtain the opposite best quote , i.e. trades at the best - ask ( resp . best bid ) price , which is less favorable . _ limit orders strategies .
_ the agent may submit at any time limit buy / sell orders at the current best bid / ask prices ( and then has to wait an incoming counterpart market order matching her limit ) , but also control her own bid and ask price quotes by placing buy ( resp .
sell ) orders at a marginal higher ( resp .
lower ) price than the current best bid ( resp .
ask ) , i.e. at @xmath50 @xmath27 @xmath51 ( resp .
@xmath52 @xmath27 @xmath53 ) .
such an alternative choice is used in practice by a market maker to capture market orders flow of undecided traders at the best quotes , hence to get priority in the order execution w.r.t .
limit order at current best / ask quotes , and can be taken into account in our modelling with discrete spread of tick size @xmath18 .
there is then a tradeoff between a larger performance for a quote at the current best bid ( resp .
ask ) price , and a smaller performance for a quote at a higher bid price , but with faster execution .
the submission and cancellation of limit orders are for free , as they provide liquidity to the market , and are thus stimulated . actually , market makers receive some fixed rebate once their limit orders are executed .
the agent is assumed to be small in the sense that she does not influence the bid - ask spread .
the limit order strategies are then modelled by a continuous time predictable control process : _ t^make & = & ( q_t^b , q_t^a , l_t^b , l_t^a ) , t 0 , where @xmath54 @xmath12 @xmath55 valued in @xmath56 ^ 2 $ ] , @xmath57 @xmath19 @xmath20 , represents the size of the limit buy / sell order , and @xmath58 @xmath12 @xmath59 represent the possible choices of the bid / ask quotes either at best or at marginally improved prices , and valued in @xmath60 @xmath12 @xmath61 , with @xmath62 @xmath12 @xmath63 , @xmath64 @xmath12 @xmath65 : notice that when the spread is equal to one tick @xmath18 , a bid quote at best price plus the tick is actually equal to the best ask , and will then be considered as a buy market order .
similarly , an ask quote at best price minus the tick becomes a best bid , and is then viewed as a sell market order .
in other words , the limit orders @xmath70 @xmath12 @xmath71 take values in @xmath72 , where @xmath73 @xmath12 @xmath74 when @xmath75 @xmath19 @xmath18 , @xmath73 @xmath12 @xmath76 when
@xmath75 @xmath12 @xmath18
. we shall denote by @xmath77 @xmath12 @xmath62 for @xmath78 @xmath19 @xmath79 , and @xmath77 @xmath12 @xmath80 for @xmath78 @xmath12 @xmath79 , and similarly for @xmath81 for @xmath78 @xmath29 @xmath26 .
we denote at any time @xmath4 by @xmath82 and @xmath83 the bid and ask prices of the market maker , where the functions @xmath84 ( resp .
@xmath85 ) are defined on @xmath86 ( resp .
@xmath87 ) by : ^b(q^b , p , s ) & = & \ { one can take into account proportional rebates received by the market makers , by considering ; @xmath89 @xmath12 @xmath93 , @xmath91 @xmath12 @xmath94 , for some @xmath95 @xmath29 @xmath96 , or per share rebates with : @xmath89 @xmath12 @xmath97 , @xmath91 @xmath12 @xmath98 , for some @xmath95 @xmath19 @xmath20 .
the limit orders of the agent are executed when they meet incoming counterpart market orders .
let us then consider the arrivals of market buy and market sell orders , which completely match the limit sell and limit buy orders of the small agent , and modelled by independent cox processes @xmath99 and @xmath100 .
the intensity rate of @xmath101 is given by @xmath102 where @xmath103 is a deterministic function of the limit quote sell order , and of the spread , satisfying @xmath104 @xmath105 @xmath106 .
this natural condition conveys the price / priority in the order execution in the sense that an agent quoting a limit sell order at ask price @xmath107 will be executed before traders at the higher ask price @xmath108 , and hence receive more often market buy orders .
typically , one would also expect that @xmath103 is nonincreasing w.r.t .
the spread , which means that the larger is the spread , the less often the market buy orders arrive .
likewise , we assume that the intensity rate of @xmath109 is given by @xmath110 where @xmath111 is a deterministic function of the spread , and @xmath112 @xmath105 @xmath113 .
we shall denote by @xmath114 @xmath12 @xmath115 , @xmath116 @xmath12 @xmath117 for @xmath75 @xmath12 @xmath92 , @xmath78 @xmath29 @xmath26 .
for a limit order strategy @xmath118 @xmath12 @xmath119 , the cash holdings @xmath120 and the number of shares @xmath121 hold by the agent ( also called inventory ) follow the dynamics dy_t & = & l_t^b dn_t^b - l_t^a dn_t^a , [ dyny ] + dx_t & = & - ^b(q_t^b , p_t^-,s_t^- ) l_t^b dn_t^b + ^a(q_t^a , p_t^-,s_t^- ) l_t^a dn_t^a .
[ dynx ] _ market order strategies .
_ in addition to market making strategies , the investor may place market orders for an immediate execution reducing her inventory . the submissions of market orders , in contrast to limit orders , take liquidity in the market , and are usually subject to fees .
we model market order strategies by an impulse control : ^take & = & ( _ n,_n)_n0 , where @xmath122 is an increasing sequence of stopping times representing the market order decision times of the investor , and @xmath123 , @xmath124 @xmath125 @xmath79 , are @xmath126-measurable random variables valued in @xmath127 $ ] , @xmath128 @xmath19 @xmath20 , and giving the number of stocks purchased at the best - ask price if @xmath123 @xmath125 @xmath20 , or selled at the best - bid price if @xmath123 @xmath105 @xmath20 at these times .
again , we assumed that the agent is small so that her total market order will be executed immediately at the best bid or best ask price . in other words , we only consider a linear market impact , which does not depend on the order size . when posting a market order strategy , the cash holdings and the inventory jump at times @xmath129 by : y__n & = & y__n^- + _ n , [ sauty ] + x__n & = & x__n^- - c(_n , p__n , s__n ) [ sautx ] where c(e , p , s ) & = & ep + |e| + represents the ( algebraic ) cost function indicating the amount to be paid immediately when passing a market order of size @xmath130 , given the mid price @xmath131 , a spread @xmath75 , and a fixed fee @xmath132 @xmath19 @xmath20 .
we shall denote by @xmath133 @xmath12 @xmath134 for @xmath75 @xmath12 @xmath92 ,
@xmath78 @xmath29 @xmath26
. one can also include proportional fees @xmath95 @xmath29 @xmath96 paid at each market order trading by considering a cost function in the form : @xmath134 @xmath12 @xmath135 , or fixed fees per share with @xmath134 @xmath12 @xmath136 . in most order - driven markets ,
available data are made up of _ level 1 data _ that contain transaction prices and quantities at best quotes , and of _ level 2 data _ containing the volume updates for the liquidity offered at the @xmath54 first order book slices ( @xmath54 usually ranges from 5 to 10 ) . in this section ,
we propose some direct methods for estimating the intensity of the spread markov chain , and of the execution point processes , based only on the observation of _ level 1 data_. this has the advantage of low computational cost , since we do not have to deal with the whole volume of _ level 2 data_. yet , we mention some recent work on parameters estimation from the whole order book data @xcite , but involving heavier computations based on integral transforms . * _ estimation of spread parameters .
_ * assuming that the spread @xmath45 is observable , let us define the jump times of the spread process : _ 0 = 0 , _
n+1 & = & t > _ n : s_t s_t- , n 1 . from these observable quantities , one can reconstruct the processes : n_t & = & # _ j > 0 : _ j t , t 0 , + _ n & = & s__n , n 0 . then , our goal is to estimate the deterministic intensity of the poisson process @xmath21 , and the transition matrix of the markov chain @xmath140 from a path realization with high frequency data of the tick - time clock and spread in tick time over a finite trading time horizon @xmath141 , typically of one day .
from the observations of @xmath142 samples of @xmath143 , @xmath124 @xmath12 @xmath144 , and since the markov chain @xmath145 is stationary , we have a consistent estimator
( when @xmath142 goes to infinity ) for the transition probability @xmath33 @xmath27 @xmath32 $ ] @xmath12 @xmath146/\pb[\hat s_n = i\delta]$ ] given by : [ estimrho ] _ ij & = & for the estimation of the deterministic intensity function @xmath22 of the ( non)homogeneous poisson process @xmath147 , we shall assume in a first approximation a simple natural parametric form .
for example , we may assume that @xmath148 is constant over a trading day , and more realistically for taking into account intra - day seasonality effects , we consider that the tick time clock intensity jumps e.g. every hour of a trading day .
we then assume that @xmath148 is in the form : ( t ) & = & _ k 1_t_kt < t_k+1 where @xmath149 is a fixed and known increasing finite sequence of @xmath150 with @xmath151 , and @xmath152 is an unknown finite sequence of @xmath153 . in other words ,
the intensity is constant equal to @xmath154 over each period @xmath155 $ ] , and by assuming that the interval length @xmath156 is large w.r.t .
the intensity @xmath154 ( which is the case for high frequency data ) , we have a consistent estimator of @xmath154 , for all @xmath157 , and then of @xmath22 given by : [ estimlambda ] _ k & = & .
we performed these two estimation procedures and on sogn.pa stock on april 18 , 2011 between 9:30 and 16:30 in paris local time .
we used tick - by - tick level 1 data provided by quanthouse , and fed via a onetick timeseries database . in table
[ transitionmatrix ] , we display the estimated transition matrix : first row and column indicate the spread value @xmath158 and the the cell @xmath159 shows @xmath160 . for this stock and at this date , the tick size was @xmath161 euros , and we restricted our analysis to the first @xmath162 values of the spread @xmath163 due to the small number of data outside this range .
we display both the gross " information ratio @xmath165 and the net " information ratio @xmath166 to have more precise interpretation of the results .
indeed , @xmath167 seems largely overestimated in this simulated data backtest compared to what would be real - world performance , for all @xmath168 .
then , to ease interpretation , we assume that @xmath169 has zero mean performance in real - world conditions , and therefore offset the mean performance @xmath170 by the constant @xmath171 when computing the nir .
this has simple visual interpretation as shown in figure .
observe that highest ( net ) information ratio is reached for @xmath172 for this set of parameters . at this point @xmath172 , the annualized value of the nir ( obtained by simple extrapolation ) is @xmath173 , but this simulated data backtest must be completed by a backtest on real data .
qualitatively speaking , the effect of increasing the inventory penalization parameter @xmath174 is to increase the zone @xmath175 where we trade at market .
this induces smaller inventory risk , due to the fact that we unwind our position when reaching relatively small values for @xmath176 .
this feature can be used to enforce a soft maximum inventory constraint directly by choosing @xmath174 .
frey s. and grammig ( 2008 ) : liquidity supply and adverse selection in a pure limit order book market " _ high frequency financial econometrics , physica - verlag hd _ , 83 - 109 , j. bauwens l. , pohlmeier w. and d. veredas ( eds . ) grillet - aubert l. ( 2010 ) : ngociation dactions : une revue de la littrature lusage des rgulateurs de march " , _ amf _ , available at : http://www.amf-france.org/documents/general/9530$_$1.pdf[http://www.amf-france.org/documents/general/9530_1.pdf ] . | we propose a framework for studying optimal market making policies in a limit order book ( lob ) .
the bid - ask spread of the lob is modelled by a markov chain with finite values , multiple of the tick size , and subordinated by the poisson process of the tick - time clock .
we consider a small agent who continuously submits limit buy / sell orders at best bid / ask quotes , and may also set limit orders at best bid ( resp .
ask ) plus ( resp .
minus ) a tick for getting the execution order priority , which is a crucial issue in high frequency trading . by trading with limit orders ,
the agent faces an execution risk since her orders are executed only when they meet counterpart market orders , which are modelled by cox processes with intensities depending on the spread and on her limit prices . by holding non - zero positions on the risky asset ,
the agent is also subject to the inventory risk related to price volatility .
then the agent can also choose to trade with market orders , and therefore get immediate execution , but at a least favorable price because she has to cross the bid - ask spread .
the objective of the market maker is to maximize her expected utility from revenue over a short term horizon by a tradeoff between limit and market orders , while controlling her inventory position .
this is formulated as a mixed regime switching regular / impulse control problem that we characterize in terms of quasi - variational system by dynamic programming methods . in the case of a mean - variance criterion with martingale reference price or when the asset price follows a levy process and with exponential utility criterion , the dynamic programming system can be reduced to a system of simple equations involving only the inventory and spread variables .
calibration procedures are derived for estimating the transition matrix and intensity parameters for the spread and for cox processes modelling the execution of limit orders .
several computational tests are performed both on simulated and real data , and illustrate the impact and profit when considering execution priority in limit orders and market orders .
* keywords : * market making , limit order book , inventory risk , point process , stochastic control . |
the young ( @xmath1 old ) supernova remnant ( snr ) cassiopeia a ( cas a ) is a unique astrophysical laboratory for studying high - energy phenomena in snrs because of its brightness across the entire electromagnetic spectrum from radio ( e.g. , * ? ? ?
* ) to very - high - energy gamma - rays @xcite . while the x - ray emission of cas a below 3 kev is dominated by line emissions from the thin thermal plasma of the shocked ejecta , as demonstrated by observations with
@xcite , a prominent nonthermal tail is known to be present and dominant above 10 kev @xcite .
nonthermal x - rays appear to be emitted mainly from fragmented filaments or knots in reverse - shocked regions .
the bulk of the hard x - ray emission in 1012 kev from cas a comes from such components ; the total emission in the outer peripheral filaments found by _
@xcite is only a small fraction of the total 46 kev continuum flux @xcite .
the interpretation of the nonthermal component in reverse - shocked ejecta remains an open issue ( e.g. , * ? ? ?
* ) . in this letter
, we present the analysis of multi - epoch _ chandra _ imaging and spectrometric data of cas a. this work was motivated by our recent discovery of time variability in the synchrotron x - ray emission of snr rx j1713.7@xmath03946 @xcite , where the year - scale time variability requires , most likely , the presence of a largely amplified magnetic field of 1 mg in compact filaments .
the strength of magnetic field in the compact radio knots in cas a is estimated to be 13 mg @xcite , just sufficient to produce year - scale variability through synchrotron cooling and shock acceleration of multi - tev electrons .
indeed , here we report significant time variations in nonthermal x - ray emission from many compact regions .
the distance to cas a is assumed to be 3.4 kpc @xcite so the angular scale of @xmath2 corresponds to 1 pc .
we analyzed the x - ray data obtained with the acis - s3 chip on 2000 january 30 @xcite , 2002 february 6 @xcite , and 2004 february 8 @xcite , each with @xmath3 ksec exposure .
table [ tbl : log ] gives the log of the _ chandra _ observations .
the observations were made with acis - s3 in the graded mode , with which telemetry size can be reduced .
the three observations have almost identical configuration such as the aim point and the roll angle , allowing us to directly compare the three images without taking account of the varying point - spread functions ( psfs ) .
the aim point is located @xmath4 southeast of the central point source , the compact remnant left behind after the supernova explosion .
the x - ray analysis was done using the ciao software package ( version 3.4 ) .
we reprocessed the event data , applying standard data reduction with the latest calibration files .
we find that the sky coordinates of the central point source in the three epochs are consistent with each other with an accuracy of @xmath5 .
the point source itself is estimated to be moving at an angular speed of @xmath6 @xcite , thus @xmath7 in the 4-yr span .
ccccccl p1 & 114 & 2000 01 30 & 350.9176 & 58.7934 & 323.4 & 49.9 + p2 & 1952 & 2002 02 06 & 350.9172 & 58.7938 & 323.4 & 49.6 + p3 & 5196 & 2004 02 08 & 350.9188 & 58.7949 & 325.5 & 49.5 figure [ fig : whole ] shows the _ chandra _ image ( with @xmath8 pixels ) of cas a in the 46 kev band averaged over the three epochs .
this energy band corresponds to the continuum emission between ca and fe k lines , comprised of thermal bremsstrahlung and nonthermal component(s ) . assuming that the hard x - ray above 10 kev is dominated by nonthermal radiation processes , we estimate about @xmath9 of 46 kev photons should be of nonthermal origin .
filamentary structures having high surface brightness in the continuum band are found only for radii less than @xmath10 ( referred as inner filaments ) , implying their association with the reverse shock as is shown to be the case by @xcite .
we are interested here in the nonthermal x - ray components .
we selected the two boxes shown in fig .
[ fig : whole ] based on the following criteria .
first , we identified regions sufficiently bright in the 46 kev band , exceeding @xmath11 at least in one epoch , in order to obtain statistics enough for a variability search .
note that this condition is met only by the inner filaments , associated with the reverse shock or its secondary shocks .
the second requirement is that the level of the continuum emission is high relative to the silicon line strength . specifically , we calculated the ratio of 46 kev counts to 1.72.2 kev counts at each pixel , and selected those regions having the ratio exceeding 0.5 . in this way
, we found that the _ inner filaments _ satisfying these criteria are distributed inside the two boxes only .
since we do not aim at constructing a statistical sample in this work , we adopted these simple criteria , accounting for neither position - dependent psfs nor effective area .
figure [ fig : filaments ] displays a sequence of three - epoch 46 kev images of the central box ( left ) and the western box ( right ) , each with dimensions of @xmath12 .
morphologically distinctive components that satisfy the criteria mentioned above are labeled as a , b , c , e , g , and h. also we tagged an additional two features , d and f. the 46 kev images over a 4-yr time interval reveal a dozen time - variable structures .
region c has decreased by a factor of 2 from 2000 to 2004 .
region h underwent a flux brightening by 50% over the time interval of 4 yr . ( by measuring the silicon line flux from the compact ejecta features , we estimated a systematic error in the flux measurements to be @xmath13 . ) the variable filaments are spatially extended beyond _ chanda _
s psfs , typically @xmath14 in length and @xmath15 in width .
also , some filaments that we did not label , such as those in the south of region c , exhibit time variability .
year - scale time variability appears to be the prevailing nature in the bright filaments or knots in the 46 kev band . before investigating the epoch - dependence of the spectral parameters , we first constructed x - ray spectra averaged over the three epochs .
regions a and b were combined into a single zone , and regions d , e , f , and g were merged . the background spectrum taken from a nearby ,
low brightness region was subtracted from each spectrum .
since the choice of a background region largely affects the low - energy part of the x - ray spectrum of region h , photons below 3 kev were discarded for this region .
the four x - ray spectra ( a@xmath16b , c , d@xmath16e@xmath16f@xmath16 g , and h ) in the 18 kev band are shown in figure [ fig : xspec ] .
they all appear to be line - less .
to characterize the shape of the line - less x - ray spectrum , we adopted a simple power - law model attenuated by interstellar absorption , which yielded statistically good fits .
we obtained the best - fit parameters in ranges of @xmath17 and @xmath18 as summarized in table [ tbl : parameter ] .
the absorbing column density is broadly consistent with those for thermal components @xcite .
we also performed spectral fitting using a thermal bremsstrahlung model .
a statistically acceptable fit was obtained in each case .
the best - fit electron temperature is in a range of 4.06.6 kev , much higher than the electron temperatures , 12 kev , derived for the x - ray knots showing thermal spectra ( e.g. , * ? ? ?
moreover , by combining all the spectral data , the equivalent width of the iron k lines is constrained to be @xmath19 ev .
the absence of the emission lines disfavors any bremsstrahlung interpretation of the featureless x - ray spectra , including bremsstrahlung by suprathermal electrons @xcite .
ccccl ab & @xmath20 & @xmath21 & @xmath22 & 0.85 ( 274 ) + c & @xmath23 & @xmath24 & @xmath25 & 1.08 ( 221 ) + defg & @xmath26 & @xmath27 & @xmath28 & 0.98 ( 365 ) + h & 1.80 ( fix ) & @xmath29 & @xmath30 & 0.84 ( 206 ) in order to track the time variations in terms of both intensity and spectral shape , the x - ray spectrum extracted from each labeled region at each epoch was fit by an absorbed power - law model .
the absorbing column density was fixed at a relevant value in table [ tbl : parameter ] .
figure [ fig : evolution ] shows the time sequence of the unabsorbed x - ray flux normalized by that in the year 2000 .
four out of the eight regions exhibit large time variations ( @xmath31 ) .
the time sequence of photon index @xmath32 for each region is also presented in figure [ fig : evolution ] .
we found notable variations of @xmath32 only in two cases , regions f and h. region h gives the hardest index of @xmath33 in 2000 , suggesting a direct link between flux increase and spectral hardening .
the change in @xmath32 found in region f should be treated with care , because region f faded after 2000 and became almost invisible in 2002 and 2004 .
photon index measured in 2002 and 2004 would not be physically connected with that in 2000 .
we have found that a half ( @xmath34 ) of the filamentary regions that were selected simply by their brightness and hardness , showed a year - scale variability in x - ray flux .
the x - ray spectra of the selected regions in the 18 kev range are featureless and well fit by an absorbed power law with a photon index @xmath35 . given the absence of emission lines , we prefer to attribute the nonthermal x - ray emission in regions a h to synchrotron radiation by very high - energy electrons .
synchrotron x - ray filaments or knots would originate in slightly overdense ejecta , say by a factor of @xmath36 , into which a reverse shock ( or transmitted shocks ) with a typical shock speed of @xmath37 has been driven ( see * ? ? ?
recently , @xcite have reported x - ray variability in compact thermal knots , resulting from the passage of reverse shock over the ejecta clumps . in the entire face of the remnant
, they identified six time - varying structures ; all but one region ( their r4 , corresponding to our region c ) were modeled by a
thermal " spectrum , specifically by a non - equilibrium - ionization plasma with an electron temperature of @xmath38 kev . on the other hand
, we found a comparable number of variable nonthermal " features just inside the two small boxes , by making use of 46 kev maps which are sensitive to nonthermal variability .
morevoer , if we apply our criteria of significant variability , @xmath31 , to the @xcite sample , only two features remain .
we conclude that variable nonthermal features are much more prevailing than the thermal ones .
cas a is a very bright nonthermal radio source which indicates the presence in the remnant of relativistic electrons with a lorentz factor @xmath39 .
the same electrons produce also high energy gamma - rays through relativistic bremsstrahlung .
since the gas density in cas a is known quite well , the gamma - ray flux is robustly controlled by the strength of the magnetic field .
therefore , high energy gamma - ray observations , combined with radio data , can provide reliable measurements of the magnetic field @xcite . a simple homogeneous model based on the radio observations and the upper limits of the @xmath40 mev flux reported by the egret collaboration @xcite yields a lower limit on the magnetic field of 0.4 mg @xcite . in a more sophisticated two - zone model of @xcite , the magnetic field in the compact radio knots , where particle acceleration is supposed to be taking place ,
is constrained to be @xmath41 .
these values would reasonably be applicable to the x - ray filaments , which have a similar size with the radio features , and are also presumed to be the sites of efficient particle acceleration by reverse ( or secondary ) shocks .
it is interesting to ask whether the x - ray flux changes are attributable solely to the changes in magnetic field strength .
if so , it is expected that a substantial flux change in synchrotron x - rays accompanies a marked change in radio intensity .
while the radio flux is proportional to @xmath42 , the x - ray flux has weaker dependence on @xmath43 , because x - rays are emitted in a quasi - saturated regime
. therefore one would expect stronger variations in the radio than in the x - ray band . according to @xcite , however , among about 70 bright radio compact knots exceeding @xmath44
, none showed a brightness increase larger than @xmath45 in the archive very large array data at 5 ghz .
a typical rate of brightening is about @xmath46 , corresponding to an @xmath47-folding time of @xmath48 , which is much longer than the x - ray variability timescale of @xmath49 .
therefore it is more likely that the observed time variations are caused mainly by a rapid increase or decrease in the density of x - ray - emitting electrons .
( the increase in the radio - electron density above the sea " level is expected to be less significant , provided that fresh electrons have a harder energy spectrum than old electrons . ) indeed , as we argue below , the known constraint on @xmath43 provides the synchrotron cooling time being compatible with the observed time variability .
we have reported a similar x - ray variability in snr rx j1713.7@xmath03946 , and attributed it to fast synchrotron cooling and shock acceleration of electrons @xcite .
let us examine the observed variability timescale , using the equations presented in @xcite .
synchrotron cooling time is given by @xmath50 , where @xmath51 is the magnetic field strength in units of mg and @xmath52 is the mean synchrotron photon energy in units of kev . for @xmath53 ,
the synchrotron cooling timescale becomes comparable to the variability timescale , @xmath54 yr .
therefore , even a magnetic field of 0.4 mg is able to explain the observed variability . for @xmath55 , if the injection of relativistic electrons responsible for the synchrotron x - rays has suddenly ceased in a filament , the x - ray flux can be reduced on a timescale of @xmath56 .
if the electron injection rate ( i.e. , the number of particles picked up and injected into the acceleration process in unit time ) decreases noticeably in 4 yr , the number of x - ray - emitting electrons and consequently x - ray intensity can drop as observed .
we note that plasma characteristics may change on a timescale of a few years given the arcsecond - scale size of the thermal ejecta filaments .
rising flux is a signature of ongoing production ( acceleration ) of synchrotron - emitting electrons with energies of @xmath57 .
it implies that electron energy can increase to @xmath57 in 4 yr or less at the shock . assuming that electrons are accelerated via diffusive shock acceleration , the acceleration time of the x - ray - emitting electrons is given by @xmath58 , where @xmath59 is a gyrofactor , and @xmath60 is the velocity of the reverse shock relative to the unshocked ejecta ahead of it in units of @xmath61 .
therefore , reverse shocks with a typical speed of @xmath37 @xcite and a magnetic field of 2 mg can accelerate x - ray - emitting electrons within a timescale of @xmath62 yr , provided that the acceleration proceeds close to the bohm diffusion limit ( @xmath63 ) .
the presence of time variability together with the absence of the fe - k lines strongly prefer a synchrotron origin of the hard x - ray emission in the selected filaments , providing new evidence of acceleration to multi - tev energies at the reverse ( and transmitted ) shocks in ejecta .
though the regions we studied here contain only a portion of the reverse - shocked gas , the bulk of the hard x - rays above 10 kev is also likely of a synchrotron origin .
if so , the steepness of the continuum spectrum above 10 kev , @xmath64 @xcite , indicates that the synchrotron spectrum rolls off at around 1 kev , being consistent with the theory of diffusive shock acceleration ( see * ? ? ?
the maximum energy of the electrons is thus constrained to be @xmath57 .
therefore the tev gamma - rays measured by hegra @xcite and magic @xcite groups are difficult to be explained by electron bremsstrahlung in the reverse - shocked ejecta , and are likely to be of a hadronic origin .
the interpretation of the x - ray variability has to be critically assessed through future theoretical and observational works .
regardless of what is behind the nonthermal x - ray variability , it offers us a new diagnostic tool to understand particle acceleration in supernova shocks .
aharonian , f. , et al .
( hegra collaboration ) 2001 , , 370 , 112 albert , j. et al.(magic collaboration ) 2007 , , 474 , 937 allen , g. e. , et al .
1997 , , 487 , l97 anderson , m. c. , & rudnick , l. 1995 , , 441 , 307 atoyan , a. m. , aharonian , f. a. , tuffs , r. j. , vlk , h. j. 2000 , , 355 , 211 baars , j. w. m. , genzel , r. , pauliny - toth , i. i. k. , & witzel , a. 1977 , , 61 , 99 bleeker , j. a. m. , willingale , r. , van der heyden , k. , dennerl , k. , kaastra , j. s. , aschenbach , b. , & vink , j. 2001 , , 365 , l225 cowsik , r. , & sarkar , s. 1980 , mnras , 191 , 855 delaney , t. , & rudnick , l. 2003 , , 589 , 818 esposito , j. a. , hunter , s. d. , kanbach , g. , & sreekumar , p. 1996
, , 461 , 820 favata , f. , et al .
1997 , , 324 , l49 gotthelf , e. v. , koralesky , b. , rudnick , l. , jones , t. w. , hwang , u. , & petre , r. 2001 , , 552 , l39 helder , e. a. , & vink , j. 2008 , , submitted holt , s. s. , gotthelf , e. v. , tsunemi , h. , & negoro , h. 1994 , , 46 , l151 hughes , j. p. , rakowski , c. e. , burrows , d. n. , & slane , p. o. 2000 , , 528 , l109 hwang , u. , holt , s. s. , & petre , r. 2000 , , 537 , l119 hwang , u. , et al .
2004 , , 615 , l117 laming , j. m. 2001 , , 546 , 1149 laming , j. m. , & hwang , u. 2003 , , 597 , 347 patnaude , d. j. , & fesen , r. a.2007 , , 133 , 147 reed , j. e. , hester , j. j. , fabian , a. c. , & winkler , p. f. 1995 , , 440 , 706 thorstensen , j. r. , fesen , r. a. , & van den bergh , s. 2001 , , 122 , 297 uchiyama , y. , aharonian , f. a. , tanaka , t. , takahashi , t. , & maeda , y. 2007 , , 449 , 576 vink , j. , & laming , j. m. 2003 , , 584 , 758 willingale , r. , bleeker , j. a. m. , van der heyden , k. j. , kaastra , j. s. , & vink , j. 2002 , , 381 , 1039 | recent discovery of the year - scale variability in the synchrotron x - ray emission of the supernova remnant ( snr ) rx j1713.7@xmath03946 has initiated our study of multi - epoch x - ray images and spectra of the young snr cassiopeia a based on the _ chandra _ archive data taken in 2000 , 2002 , and 2004 .
we have found year - scale time variations in the x - ray intensity for a number of x - ray filaments or knots associated with the reverse - shocked regions .
the x - ray spectra of the variable filaments are characterized by a featureless continuum , and described by a power law with a photon index within 1.92.3 .
the upper limits on the iron k - line equivalent width are 110 ev , which favors a synchrotron origin of the x - ray emission .
the characteristic variability timescale of 4 yr can be explained by the effects of fast synchrotron cooling and diffusive shock acceleration with a plausible magnetic field of 1 mg .
the x - ray variability provides a new effective way of studying particle acceleration at supernova shocks . |
as is well known , near - infrared ( nir ) photons are up to 10 times less affected by dust extinction than their visible counterparts .
hence , cosmological surveys based upon nir - selected galaxy samples can probe the universe almost completely through the zone of avoidance in the galactic plane , and moreover have a clearer view of galaxies , virtually unobscured by dust ( especially important for the central regions of spirals and for edge - on spirals ) .
the other main advantage of nir - selected galaxy samples is that they are not biased towards recent star formation , contrary to optically - selected samples ( especially in blue light ) , and hence the nir - selected galaxies are more weighed by their mass , or at least their stellar mass .
however , the iras galaxy samples , heavily weighted towards galaxies with recent star formation , appear to be better tracers of the mass in the universe than are early - type galaxy samples .
the first spectroscopic followups of nir selected galaxy samples show fairly small fractions of galaxies with emission lines in their spectra : 37% in @xmath7 samples
( mamon et al . 1999 ) and 30% in @xmath8 samples ( huchra , 1999 ) , in comparison with 60% in similar depth @xmath9-selected samples .
hence , denis and 2mass will be more weighted towards early - type galaxies and may lead to stronger biases relative to the dark matter than the iras samples , but this remains to be seen ( especially because , contrary to the iras samples , they will have a mix of early- and late - type galaxies ) .
the denis ( deep near infrared southern sky survey ) consortium is undertaking a complete imaging survey of the southern sky ( @xmath10 ) in the gunn @xmath11 , johnson @xmath12 ) , and @xmath13 bands ( epchtein et al .
1997 ) , and at this writing , 62% of the southern sky has been observed ( see http://www.iap.fr/users/gam/denis/slots.html ) .
the completion date is planned for january 2001 .
the denis team has built an ir camera and mounted it on the eso 1 m telescope ( which had previously only done optical aperture photometry ) .
the @xmath14 , @xmath15 , and @xmath16 images are obtained simultaneously through the use of dichroics with integration times of 9 sec . the pixel sizes are @xmath17 in @xmath14 and @xmath18 in @xmath15 and @xmath16 , and the latter two are dithered in 9 sub - exposures of 1 sec each to yield images with a pseudo - resolution of @xmath17 .
star / galaxy separation is performed with classical estimators , and except in the zone of avoidance ( @xmath19 ) , the much higher sensitivity and spatial resolution of the @xmath14 band allow us to base our star / galaxy separation on the @xmath14 images for objects detected in the @xmath15 or @xmath16 images .
we have performed detailed cross - identifications of the denis extractions with the cosmos extractions from @xmath20 photographic plates ( obtained over the world wide web ) , over 8 denis strips , corresponding to @xmath21 .
these strips were randomly selected at typically high galactic latitudes @xmath22 ) from the great majority of strips that are not flagged as poor quality . to compare the quality of the two surveys , we performed two cross - identifications : 1 ) relative to all denis objects ; and 2 ) relative to cosmos objects ( @xmath23 ) in the exact geometry of the 8 denis strips .
we assumed that all objects , on which both denis and cosmos agreed were galaxies , were indeed galaxies .
we systematically inspected visually the denis @xmath14-band images for all cases with @xmath1 ( our estimated limit for 95% reliable visual classification ) where one of the surveys identified a galaxy while the other called it a star or did not detect it altogether .
objects that both denis and cosmos classify as galaxies have @xmath24 and 95% have @xmath25 .
hence , our estimates of cosmos completeness are limited to @xmath26 .
figure 1 shows the resulting denis ( _ left _ ) and cosmos ( _ right _ ) completeness ( _ dashed _ ) and reliability ( _ solid _ ) as a function of magnitude . while denis achieves better than 95% completeness and reliability over the full range @xmath27 , cosmos has worse than 80% completeness and reliability at @xmath28 , and the numbers get even worse at brighter magnitudes : in particular the completeness falls to below 10% at @xmath29 . at bright magnitudes
, cosmos suffers from the non - linearity of the plate photometry .
its reliability suffers from the blending of low non - linear peaks of double stars , while its strong incompleteness is caused by systematic over - compensation of the photometric non - linearity , thus bright galaxies are classified as even brighter stars .
the strong case for spectroscopic followups of near - ir selected galaxy samples , coupled with the realization that schmidt telescopes offer wide enough fields to sample most efficiently the sky for shallow samples of 510 galaxies deg@xmath30 , has led the aao to replace its existing manually - configured flair ii 92-fiber system on the ukst with a robotized , 150-fiber system , called 6df ( for six degree field ) .
the 6df instrument , currently under construction uses an @xmath31 positioner , adapted to the curved focal surface of the ukst .
the aao has suggested granting 300 nights of dark / grey time during 20012003 to a scientific advisory group ( 6dfsag , see acknowledgments , below ) , to conduct a redshift survey based upon near - ir selected galaxy samples .
the 6dfsag is envisioning a complete spectroscopic followup of @xmath32,@xmath33 denis/2mass selected southern galaxies .
the primary sample should be a 2mass-@xmath16 or denis-@xmath15 selected sample .
moreover , we will add galaxies to form complete subsamples in denis @xmath14 and 2mass @xmath34 , as well as supercosmos @xmath9 and @xmath35 .
because fibers can not be placed closer than @xmath36 to one another , we envision an adaptive tiling scheme , with typically 2 exposures per effective field to minimize the losses due to crowding . at a galaxy density of 7 deg@xmath30 , our 27 deg@xmath37 field will have an average of 189 galaxies .
if 10 fibers are reserved for sky , we will have @xmath38 fibers available for objects per field , hence an average of @xmath39 spare fibers , which we intend to use for other classes of objects such as quasars and/or iras galaxies , as well as allow for cosmic variance of the field density .
the principal motivation of the 6dfsag is to perform a peculiar velocity survey in the full ( @xmath40 ) southern hemisphere . assuming 3/4 dark or grey ukst telescope time and two 4 hours exposures per night , we expect to measure @xmath41 resolution spectra for roughly 12,000 ellipticals and lenticulars at @xmath42,@xmath43 . to limit the survey duration , only the densest half of the fields will be observed .
the line - widths will be coupled with surface photometry ( using denis @xmath14 or 2mass @xmath16 images ) , to yield distances ( with the @xmath44 relation ) and hence peculiar velocities .
we hope to be awarded the time to pursue the peculiar velocity survey during the period 20032005 . in the meanwhile , the denis - hi consortium has just begun a spectroscopic followup of denis southern galaxies with the nanay radio - telescope .
we estimate that given one - sixth the nanay time , we can reach the estimated 5000 inclined spiral galaxies at @xmath45 with @xmath46 and @xmath47,@xmath48 in 20002004 , thanks to the 5-fold increase in sensitivity and velocity coverage that nanay should obtain in the fall of 1999 .
so far , a hundred spectra have been obtained .
the denis - hi and 6df peculiar velocity surveys should increase the coverage of the southern hemisphere from 1500 to 17,000 galaxies , and within @xmath49 , the space density of objects will be 4 times higher than today . therefore , the problems related to the sparseness of the current peculiar velocity samples should disappear with these two surveys , leading to more accurate estimates of bulk flow , @xmath5bias , @xmath6 itself , and especially the primordial density fluctuation spectrum .
i thank members of the denis team , especially n. epchtein , p. fouqu , and j. borsenberger .
thanks also to f. giraud who participated in the comparison of denis and cosmos , and e. bertin for useful discussions on the matter .
i also acknowledge my colleagues on the 6df science advisory group : m. colless , j. huchra , o. lahav , j. lucey , q. parker , e. sadler , w. saunders , and f. watson and on the denis - hi team : g. theureau , i. vauglin , g. paturel , l. bottinelli , j .- m .
martin , and w. van driel . | the completeness and reliability of the denis @xmath0 survey and the edsgc ( derived from the cosmos scans of uskt plates ) are obtained by detailed cross - identifications and systematic visual inspections of conflictual classifications .
the denis galaxy extraction turns out to be over 95% complete and reliable out to @xmath1 , while the cosmos / edsgc galaxy catalogue is less than 80% complete and reliable at @xmath2 , and even less than 10% complete at @xmath3 .
spectroscopic followups of denis and the similar near - ir 2mass survey are described : 1 ) a redshift survey of 120,000 galaxies using the 6df robotic multi - fiber spectrograph , currently under construction at the ukst , and for which a total of 300 nights are guaranteed for 2001 - 2003 , 2 ) a peculiar velocity survey of 12,000 early - type galaxies with the 6df , and 3 ) the denis - hi peculiar velocity survey of 5000 inclined spirals visible from nanay ( @xmath4 ) , which has just begun .
the denis - hi and 6df peculiar velocity samples will have the strong advantage of covering entire regions of the southern sky , and combined , will multiply by 10 and 4 respectively the projected and space number densities of objects in the southern sky .
these two surveys should thus provide considerably more accurate estimates of the bulk flow , @xmath5bias , @xmath6 itself , and the primordial density fluctuation spectrum .
-2 cm _ contribution to workshop on cosmic flows , held in victoria , bc , canada , july 1317 , 1999 , ed .
s. courteau , m. strauss & j. willick , asp series _ |
the collapse - driven supernova is supposed to be an outcome of gravitational collapse of a massive star @xmath3 at the end of its evolution .
although more than 60 years have passed since the original idea by baade and zwicky@xcite , and the detection of neutrinos from sn1987a@xcite confirmed that our scenario is correct on the whole , we have not yet figured out how this phenomenon occurs .
these days many researchers of the supernova consider that neutrinos diffusing out of the proto neutron star play a crucial role in heating the material behind the shock wave and expelling the outer layer of the star@xcite .
it turned out , however , that this mechanism is quite sensitive to the neutrino luminosity@xcite .
in fact , janka and mller@xcite showed by numerical simulations that only about 20% of increase in neutrino luminosity could lead to a successful explosion of a model which otherwise failed to explode .
hence mechanisms to boost the neutrino luminosity have been quested .
although many authors have been devoted in the study of convection in the core and it has been shown that the convection does help neutrinos heat matter , it is still controversial if the convection alone is sufficient for successful explosion or not@xcite .
it is also found that the sophistication of the numerical treatment of neutrino transport could increase the neutrino luminosity although the quantitative assessment of the difference it makes in the realistic context remains to be done@xcite . on the other hand ,
our knowledge of the neutrino reaction rates in the hot dense medium is rather poor .
in fact , even in the most elaborate simulations of supernovae it has been assumed that the reactions of neutrinos with nucleons are the same as in vacuum and the effect of surrounding matter is ignored@xcite .
however , the wavelength of a 30mev neutrino , for example , is longer than the mean separation of nucleons for the density @xmath4g/@xmath2 , and the time scale corresponding to the same energy is roughly of the same order as the mean free time of nucleon between collisions .
hence we have to study the spatial and temporal correlations of the matter , and it could be possible that the many body effects change the opacity for neutrinos considerably and we get the desired enhancement of neutrino luminosity and/or energy .
in fact , efforts to find a possible modification of rates of neutrino reactions with nucleons have been made by several authors@xcite . in these studies
neutrino - nucleon scatterings are one of the targets , since it is one of the major sources of opacity for neutrinos .
reddy et al.@xcite pointed out that taking a correct effective mass of nucleon into account changes the scattering rate in the high density regime @xmath5 which is the saturation density .
horowitz et al.@xcite , burrows et al.@xcite , reddy et al.@xcite and yamada et al.@xcite discussed the correlation effects due to the particle - hole excitation using a random phase approximation ( rpa ) . in these studies
the nucleons were assumed to be quasi - particles with the vanishing width of the spectral functions . on the other hand , raffelt and
his collaborators@xcite claimed that the neutrino scattering rate could be also reduced by losing the temporal correlations of the spins of nucleon , thus broadening the width of the structure function due to collisions of the nucleon with other nucleons surrounding it ( see also @xcite ) .
this is a counter part for scattering of the so - called landau - pomeranchuk - migdal effect@xcite for the bremsstrahlung .
it was also pointed out that this broadening of the structure function could enhance the inelasticity of the scattering and affect the energy spectra of neutrinos@xcite , since the width of the structure function is of the order of the matter temperature at a relatively low density @xmath1g/@xmath2 according to their estimate . in these studies they assumed that the structure function of nucleon takes the lorentzian form and inferred its width from the typical collision rate of nucleon with the overall normalization determined by the sum rule and the detailed balance imposed by hand .
in addition , they only discussed long wave length limits . in this paper
we study this collisional effect on the neutrino
nucleon scattering rates on the field theoretical basis .
the width of the spectral function of nucleon is evaluated by solving the bethe - salpeter equation in medium and calculating the imaginary part of the nucleon self - energy@xcite , and then it is used to calculate the structure functions of nucleon as a function of the transfered four momentum@xcite . in so doing the separable yamaguchi potential@xcite
was used for simplicity .
the paper is organized as follows . in the next section
we formulate the basic equations .
the results are presented in section 3 .
discussion is given in the last section .
the scattering rates of neutrinos with nucleons are quite generally formulated as : @xmath6 \quad , \end{aligned}\ ] ] where @xmath7 and @xmath8 are the incident and outgoing neutrino energy , respectively , @xmath9 is the scattering angle , and @xmath10 is the four momentum transferred from neutrino to nucleon . in the right hand side of eq .
( [ eq : decom ] ) , the third term is usually much smaller than the other two , so that we ignore it in the following .
the so - called dynamical structure functions @xmath11 and @xmath12 are given in the non - relativistic limit which we assume in the following , as @xmath13 here @xmath14 and @xmath15 are the weak coupling constants of nucleons , @xmath16 stands for the thermal ensemble average of the argument , and @xmath17 and @xmath18 are the density and the spin density of nucleon . thus it is obvious that @xmath19 and @xmath20 are nothing but their correlations in the matter . as is clear from the factor of the @xmath20 in eq .
( [ eq : decom ] ) , @xmath20 is more important than @xmath19 .
thus we will discuss in this paper the modification of @xmath20 , or the spin - density correlation function , due to the nucleon
nucleon collisions in the supernova core . just by the same argument
as made by knoll et al.@xcite for the momentum correlation , the spin - density correlation of the nucleon which is successively scattered by other nucleons with a mean collision rate , @xmath21 , changing the direction of its spin via spin - dependent interactions , is given by ignoring the spatial non - uniformity as @xmath22 in the above equation , @xmath23 is the number of the scatterings during the time interval of @xmath24 .
@xmath25 is the spin - density at the m - th collision , and @xmath26 means taking the average of the argument with respect to the ensemble of the sequence of @xmath23 scatterings .
taking the fourier transform of @xmath27 , we obtain @xmath28 it was pointed out in the paper by knoll et al.@xcite that the above summation of the lorentz functions correspond to the summation of the feynman diagrams depicted in fig .
[ fig1 ] in the closed time path formalism .
if we further assume that the spin - density auto - correlation decreases by a constant rate , @xmath29 , after each scattering , @xmath30 , then we obtain @xmath31 this implies that the higher order terms account for the difference between the collision rate , @xmath21 , and the relaxation time of spin - density , @xmath32 . if the spin flip is fast enough , @xmath33 , these two rates become almost identical . in this case , the spin correlation function of eq .
( [ eq : r2a ] ) is approximated by the first diagram in fig .
note that the propagator in that diagram is not a free propagator but a dressed one in the medium .
the self - energy included in the dressed propagator comes from the nucleon collisions with other nucleons . in this paper
this self - energy is evaluated by solving the bethe - salpeter equation for two nucleons as drawn schematically in fig .
[ fig2]@xcite . in so doing
, we assumed a separable yamaguchi potential@xcite in order to facilitate solving the bethe - salpeter equation .
although this is an oversimplification of the nuclear interactions , this is not a bad approximation as long as the low density regimes , @xmath34 g/@xmath2 is concerned , which is the case in this study , and gives some insight into what the collisional effects are like and how important they could be .
since the scatterings are suppressed due to the fermi blocking in the high density regime , it is expected the scattering effects are most important in the low density regime .
in addition , the resulting broadening of the structure functions of nucleon will affect the neutrino energy spectra which are formed in this low density regime@xcite .
some remarks are necessary .
the above approximation is not applicable to the vector current contribution to the neutrino scattering rates , since it does not take into account the conservation law , and , as a result , the long wave length limit is not properly reproduced . in the case of the axial vector current
, the spin density is not conserved if one takes into account the nuclear tensor force .
what is more important , the nuclear axial vector current is nearly isovector .
thus unless the matter is entirely composed of neutrons or protons , the axial vector component of the nuclear weak current is not a conserved one as long as the nuclear force is spin - dependent . here
we summarize the formulation outlined above .
the spin correlation functions are given by the dressed green function as @xmath35 where the nucleon green functions are defined as @xmath36 \\ i \ , g_{n}^{<}(p ) \ & \ = \ & \ - \ , \rho(p ) \
f_{n}(p ) \qquad .\end{aligned}\ ] ] @xmath37 is the fermi - dirac distribution of nucleon .
@xmath38 is the spectral function of nucleon given by the self - energy @xmath39 as @xmath40^{2 } \ + \ \left [ \ , \displaystyle{\frac{im \sigma(p)}{2}}\ , \right ] ^{2 } } \qquad .\ ] ] @xmath41 is the nucleon mass .
the quasi - particle approximation , which is usually assumed , is obtained as a limit @xmath42 . in the hartree or hartree - fock approximation ,
the self - energy is real and the nucleon can be treated as a quasi - particle . beyond those approximations
, the self - energy has an imaginary part in general .
it is obvious that the imaginary part of the self - energy is a width of the spectral function of the transfer energy @xmath43 .
hence the spectral function becomes a @xmath44 function in the quasi - particle approximation . in this paper ,
following alm et al .
@xcite , we evaluate the imaginary part of the nucleon self - energy from the two - particle @xmath45-matrix which is given as a solution of the bethe - salpeter equation in the ladder approximation , @xmath46 here @xmath47 is defined as the product of two single - particle green functions and calculated in the quasi - particle approximation as @xmath48 @xmath49 is the on - shell energy determined from the real part of the denominator of eq .
( [ eq : rho ] ) and approximated as explained later .
( [ eq : bs ] ) is solved for the separable yamaguchi potential , @xmath50 , with @xmath51 , where the coupling constant and the effective range are @xmath52 ( mev @xmath53)@xmath54 for @xmath55 and @xmath56 ( mev @xmath53)@xmath54 for @xmath57 and @xmath58 @xmath53 , respectively .
then the imaginary part of the self - energy is given by @xmath59 \
i m t_{ex}(p , p ' , p , p ' , e(p ' ) + \omega + i \varepsilon ) \quad .\ ] ] here again the quasi - particle approximation is used for a single - particle green function .
@xmath60 - 1 \}^{-1}$ ] is a bose - einstein distribution function with the chemical potential of the nucleon pair .
the subscript @xmath61 denotes the inclusion of the exchange term .
the real part of the self - energy is also derived from the @xmath45-matrix in a similar way .
since it depends on the on - shell energy which is in turn determined by the real part of the self - energy , iterations are required to obtain the self - consistent value of the on - shell energy . for simplification
we abandon this consistency in this paper and set the on - shell energy as @xmath62 , where the effective mass @xmath63 and the potential @xmath64 are taken from the relativistic mean field theory@xcite . using this on - shell energy and the corresponding imaginary part of the self - energy calculated by eq .
( [ eq : imsig ] ) , we approximate eq .
( [ eq : rho ] ) as @xmath65^{2 } + [ \ , i m \sigma(p ) / 2 \ , ] ^{2 } \}$ ] .
finally we evaluate eq .
( [ eq : r2a ] ) with this approximate spectral function of nucleon .
in fig . [ fig3 ] , we show the imaginary part of the neutron self - energy as a function of the transfer momentum and energy for the density @xmath66g/@xmath2 and temperature @xmath67mev and electron fraction @xmath68 .
the on - shell momenta roughly correspond to the ridge of the hill in this figure .
the imaginary part of the self - energy is essentially a scattering rate of a neutron with another nucleon .
it is evident from the figure that this is energy- and momentum - dependent and becomes as large as the average kinetic energy of the nucleon @xmath69 for small momentum transfers .
@xmath70 decreases as one goes to higher energies along the on - shell , since the scattering rate becomes smaller for higher energy incident nucleons .
the self - energy of proton is almost the same as that of neutron in this case . for more asymmetric matter with smaller @xmath71 ,
@xmath72 becomes larger while @xmath73 gets smaller as seen in fig .
this is simply because the scattering rate for the neutron
proton pair is larger than that for the neutron neutron or the proton
proton pair .
the greater degeneracy for neutron also contributes to the reduction of the neutron self - energy .
the density- , temperature- and @xmath71-dependences of @xmath74 can be understood essentially in the same way . since we solved the bethe - salpeter equation , the contribution from the deuteron pole , if any , is automatically taken into account .
in fact , the parameter values of the yamaguchi potential used in this paper are chosen so that the deuteron binding energy is reproduced in the vacuum . for the above cases , we confirmed no bound state can exist due to the medium effect as shown by alm et al.@xcite for the lower density @xmath75g/@xmath2 , however , the deuteron pole does exist and give tiny contributions to the imaginary part of the self - energy as shown in fig .
[ fig5 ] , where only the contribution from the deuteron pole was shown .
the existence of the bound state is , thus , not relevant for the density , temperature and @xmath71 of our interest .
the spectral function is shown in fig . [ fig6 ] as a function of the transfer energy for different transfer momenta the on - shell momentum is located again at the peak of the spectrum . as mentioned above , this function would be a @xmath44-function of the transfer energy if there were no interactions between nucleons . with collisions of two nucleons the spectral function
is broadened , as expected , and the width of the spectral function is roughly the scattering rate for the given transfer momentum .
it is seen that the width of the spectral function or the scattering rate of nucleon is dependent on the transfer momentum and the spectral function becomes more narrow - peaked as the transfer momentum gets greater , reflecting the decrease of scattering rates .
it should be again emphasized that the width of the spectral function is of the same order for the low momentum transfer as the average kinetic energy of nucleon .
hence it is expected that the kinetics of neutrino nucleon scattering will be also affected .
the density- , temperature- and @xmath71-dependences of the spectral function can be inferred from that of the imaginary part of the nucleon self - energy .
as the density increases , the scattering rate of nucleon with another one is increased , thus leading to larger widths of the spectral function .
note that the greater degeneracy of nucleon counteracts via the fermi blocking to reduce the scattering rate .
the scattering rate is determined by this competition . as the temperature increases , the blocking effect is diminished .
this enhances the scattering rate .
however , the larger kinetic energy of nucleon lowers the scattering rates . again
the width is determined by the balance of these factors . as the matter becomes neutron rich ,
the width of the spectral function for neutron is decreased partly because the scattering rate for neutron
neutron pair is smaller than that for neutron proton and partly because the neutron becomes more degenerate and the fermi blocking reduces the scattering rate more effectively . using the spectral functions obtained above , we calculated the nucleon structure functions with eq .
( [ eq : r2a ] ) .
the typical result is shown in fig .
[ fig7 ] for @xmath76 as a function of the transfer energy @xmath77 for three different transfer momenta @xmath10 . in this figure ,
three cases are compared .
the thin dotted lines represent the structure functions for the free neutron , while the thin solid lines present the case for which only the effective mass of nucleon was taken into account in the spectral function , eq .
( [ eq : rho ] ) .
the smaller effective mass lowers the peak of the spectral function as pointed out by some authors@xcite , although this effect is small for this low density ( @xmath78 ) .
it also tends to broaden the structure functions , since the recoil of the nucleon becomes larger as the effective mass of nucleon gets smaller .
again this is negligible in the low density regime considered here . on the other hand , the structure function
is considerably broadened when the collisions among nucleons were taken into account , as shown with the thick solid lines in fig .
the width is essentially given by the scattering rate of nucleons and taken over from the corresponding spectral function .
the widths are not different for this low momentum transfers , which is qualitatively different from the results for the free nucleons and those for the quasi - particle approximation . as the density increases ,
the width becomes larger in general .
as the temperature decreases , the part of the negative energy transfer is reduced , since the extraction of energy from the medium becomes more and more difficult as the nucleons become more degenerate .
this is nothing but the detailed balance relation @xmath79 which is automatically satisfied in the formulation used in this paper .
the width of the structure function becomes smaller for neutron and greater for proton as @xmath71 is decreased , as expected from the above discussions for the spectral function .
the total scattering rate of the neutrino with the incident energy @xmath7 is given by the integration of the structure function with respect to the transferred momentum @xmath10 and energy @xmath77 : @xmath80\nonumber \\ & = & \frac{1}{(2 \pi ) ^{3 } } \ , \int ^{\infty}_{0 } \ !
2 \pi k \ , dk \
, \int _ { -k}^{\omega ^{max } } \ ! \ ! \ ! d \omega \ \frac{e_{\nu } ^{in } - \omega } { e_{\nu } ^{in } } \frac{1}{2 e_{\nu } ^{in } \ , 2 e_{\nu } ^{out } } \
r(e_{\nu } ^{in } , \ , e_{\nu } ^{out } , \ , \cos \theta ) \
\left [ \ , 1 \ , - \ , f_{\nu } ( e_{\nu } ^{out } ) \ ,
\right ] , \end{aligned}\ ] ] with @xmath81 .
the scattering angle @xmath9 and the energy of the outgoing neutrino @xmath8 are functions of @xmath7 and @xmath10 .
the collisions of nucleons reduce in two ways the above scattering rates from the quasi - particle approximation .
not only the amplitude of the structure function is lowered , but the broadening of the structure function tends to put some fraction of the structure function outside the integration region . ignoring the fermi blocking of the outgoing neutrino in eq .
( [ eq : toteq ] ) , we integrated the axial vector part @xmath82 for @xmath83 and compared the result with the collision taken into account and the result for the free nucleon . for the model shown in fig .
[ fig7 ] , for example , we found that the scattering rate is suppressed by the factor of 0.41 for neutron and 0.38 for proton , respectively . note that the contribution of the effective mass is only a few percent in this case . for @xmath75g/@xmath2 , @xmath84mev and @xmath68
, we still found the reduction of @xmath85 .
the larger width of the structure function implies that the neutrino can exchange energy with nucleons more efficiently . in the supernova simulations performed so far , the neutrino
nucleon scattering was commonly approximated as the iso - energetic scattering , which means that this process does not contribute to the thermalization of the neutrino spectra .
the above results stand by the claim by raffelt and his company@xcite that this may not be the case .
in fact , the mean square root of energy transfer increased from @xmath86mev to @xmath87mev for the model of fig .
although this is still not a large value , it should be born in my mind that this process has about ten times larger cross section than the scattering with electrons .
hence it is expected the energy spectra of neutrinos , particularly muon- and tau - neutrinos , will be affected by this effect .
thus we need to somehow implement this effect in a numerical code and to study quantitatively the difference it will make in the realistic settings .
we studied in this paper the possible effects of collisions among nucleons on the reaction rates of the neutrino nucleon scatterings . by solving the bethe - salpeter equation for two nucleons , we estimated the imaginary parts of the nucleon self - energy , which in turn give the spectral function of the nucleon in medium . using this spectral function ,
we calculated the structure functions of nucleon , which are directly related to the neutrino nucleon scattering rates . in the limit of no collision
, the spectral function reduces to the delta function of the transfer energy and the structure function gives the reaction rates which have been commonly used in the literature thus far .
it was shown that the collisional broadening of the structure functions might be the main factor in the low density regime considered in this paper to modify not only the neutrino reaction rates with the nucleon but also the neutrino spectra .
it is obvious that the further study is needed on this issue .
although we used yamaguchi potential to facilitate calculations , this is clearly an oversimplification of the dynamics of two nucleons .
the self - consistency was also sacrificed for simplicity .
thus our results should be regarded as of qualitative nature .
more importantly , we have to assess the corrections of the terms we ignored in estimating the the structure functions , eq .
( [ eq : sik ] ) .
the in - medium vertex corrections should be also taken into account and are inferred from the fermi liquid theory , for example@xcite . though we neglected in this paper these corrections in order to make clear the effect of the width of the spectral function of nucleon , they could be as important and should be studied on the same basis in the future work .
it should be also mentioned that the formulation in this paper is not conserving@xcite and can not be applied to the vector current , whose long wave length limit is dictated by the baryon number conservation .
the accomodation of this limit should be another subject of the future work . at present the neutrino transport
is thought to play a crucial role in the collapse - driven supernovae .
it is , thus , important to study the microphysics involved and , at the same time , to investigate its consequences to the supernova simulations , which we are planning to do with the numerical code recently developed by us to solve the boltzmann equations for neutrinos@xcite .
i gratefully acknowledge critical discussions with h .- th .
janka , g. raffelt and h. toki .
i would highly appreciate various support by e. mller during the stay in max planck institut fr astrophysik .
this work is partially supported by the japanese society for the promotion of science ( jsps ) , postdoctoral fellowships for research abroad and by the grants - in - aid for the center - of - excellence ( coe ) research of the ministry of education , science , sports and culture of japan to resceu ( no.07ce2002 ) .
w. baade and f. zwicky , phys . rev . *
45 * , 138 ( 1934 ) w. baade and f. zwicky , phys
. rev . * 46 * , 76 ( 1934 ) r. m. bionta et al .
lett . * 58 * , 1494 ( 1987 ) k. hirata et al .
lett . * 58 * , 1490 ( 1987 ) j. r. wilson , proc .
illinois meeting on numerical astrophysics ( 1982 ) h. a. bethe and j. r. wilson , astrophys .
j. * 295 * , 14 ( 1985 ) a. burrow and j. goshy , astrophys .
j. * 416 * , 75 ( 1993 ) h .- th .
janka and e. mller , frontiers of neutrino astrophysics , proc . of the international symposium on neutrino astrophysics , takayama / kamioka , japan , oct .
19.-22 . , 1992 , edited by y. suzuki and k. nakamura ( universal academy press , tokyo , 1993 ) , p203 h
janka and e. mller , astron . astrophys . * 306 * , 167 ( 1996 ) m. herant , w. benz , j. hix , c. freyer and s. a. colgate , astrophys . j. * 435 * , 339 ( 1994 ) a. burrows , j. hayes and b. a. fryxell , astrophys . j. * 450 * , 830 ( 1995 ) t. shimizu , s. yamada and k. sato , astrophys . j. lett .
* 432 * , l119 ( 1994 ) o. e. b. messer , a. mezzacappa , s. w. bruenn and m. w. guidry , astrophys . j. * 507 * , 353 ( 1998 ) s. yamada , h .- th .
janka and h. suzuki , astron .
astrophys . * 344 * , 533 ( 1999 ) a. burrows , t. young , p. a. pinto , r. eastman and t. thompson , submitted to astrophys .
j. ( 1999 ) s. w. bruenn , astrophys .
* 58 * , 771 ( 1985 ) a. mezzacappa and s. w. bruenn , astrophys .
j. * 405 * , 637 ( 1993 ) a. mezzacappa and s. w. bruenn , astrophys .
j. * 405 * , 669 ( 1993 ) a. mezzacappa and s. w. bruenn , astrophys .
j. * 410 * , 740 ( 1993 ) n. iwamoto and c. j. pethick , phys .
d * 25 * , 313 ( 1982 ) r. f. sawyer , phys .
c * 40 * , 865 ( 1989 ) c. j. horowitz and k. wehrberger , phys .
b * 266 * , 236 ( 1991 ) r. f. sawyer , phys .
lett . * 75 * , 2260 ( 1995 ) g , raffelt and d. seckel , prd . *
52 * , 1780 ( 1995 ) g , raffelt , d. seckel and g. sigl , prd . * 54 * , 2784 ( 1996 ) g , raffelt and g. sigl , sumitted prd . ( 1998 ) a. burrow and r. f. sawyer , prc .
* 58 * , 554 ( 1998 ) a. burrow and r. f. sawyer , prc . *
59 * , 510 ( 1999 ) s. reddy , m. prakash , j. m. lattimer and j. a. pons , preprint astro - ph/9811294 ( 1998 ) s. reddy , m. prakash and j. m. lattimer , proc .
second oak ridge symposium on atomic and nuclear astrophysics , 1998 s. yamada and h. toki , submittded to phys . rev .
c ( 1999 ) l. d. landau and i. pomeranchuk , dokl .
sssr * 92 * , 535 ( 1953 ) l. d. landau and i. pomeranchuk , dokl .
sssr * 92 * , 735 ( 1953 ) a. b. migdal , phys . rev . * 103 * , 1811 ( 1956 ) s. hannestad and g. raffelt , astrophys . j. * 507 * , 339 ( 1998 ) m. schmidt , g. rpke and h. schulz , ann . phys . * 202 * , 57 ( 1990 ) t. alm , a. schnell , n. h. kwong and h. s. khler , phys . rev .
c * 53 * , 2181 ( 1996 ) g. rpke and a. schnell , prog .
. phys . * 42 * , 53 ( 1999 ) j. knoll and d. n. voskresensky , phys .
b * 351 * , 43 ( 1995 ) y. yamaguchi , phys . rev . *
95 * , 1628 ( 1954 ) k. sumiyoshi and h. toki , astropphys .
j. * 422 * , 700 ( 1994 ) y. sugahara and h. toki , nucl .
a. * 579 * , 557 ( 1994 ) d. n. voskresensky and a. v. senatorov , sov .
* 45 * , 411 ( 1987 ) a. sedrakian and a. dieperink , preprint nucl - th/9905039 ( 1999 ) g. baym and l. p. kadanoff , phys
. rev . * 124 * , 287 ( 1961 ) g. baym , phys . rev . * 127 * , 1391 ( 1962 ) | we studied possible modifications of the neutrino
nucleon scattering rate due to the nucleon
nucleon collisions in the hot dense matter which we find in the supernova core .
we show that the finite width of the nucleon spectral function induced by the nucleon collisions leads to broadening of the dynamical spin structure function of the nucleon , resulting in the reduction of the rate of neutrino nucleon scattering via the axial vector current and making the energy exchange between neutrinos and nucleons easier .
the reduction rate is relatively large @xmath00.6 even at density @xmath1g/@xmath2 and could have a significant impact on the dynamics of the collapse - driven supernova as well as the cooling of the proto neutron star . |
the characterization of the globular cluster luminosity function ( gclf ) remains a high priority as a means to better understand both the formation of globular clusters and the buildup of galaxies .
the differences in the breadth and peak position of gclfs in late - type galaxies in the local group and early - type galaxies in other clusters has been known for some time ( harris et al . 1991 ) .
efforts to use gclfs as distance indicators or tracers of galaxy mergers would clearly benefit from more complete coverage over luminosity and galaxy type . with its 01 resolution , low sky background , and consequent ability to distinguish very faint point - sources , the _ hubble space telescope _ ( _ hst _ ) is an ideal platform from which to carry out such a study .
the first surveys of globular clusters in the fornax cluster were carried out photographically ( dawe & dickens 1976 ; hanes & harris 1986 ; harris & hanes 1987 ) .
later ccd studies ( see ashman & zepf 1997 for a list of references ) detected bimodal color distributions ( zepf & ashman 1993 ) and color gradients ( ostrov , geisler , & forte 1993 ) in the cluster system of ngc 1399 .
_ hst _ studies have now confirmed the existence of multi - modal color distributions in several other giant ellipticals ( e.g. m 87 , elson & santiago 1996b ; whitmore et al .
1995 ; ngc 5846 , forbes et al .
what follows is the first systematic _ hst _ investigation of the globular cluster systems in the fornax cluster .
the results should help us to understand how cd galaxies and their large associated globular cluster populations formed ( ashman & zepf 1992 ; forbes , brodie , & grillmair 1997b ) .
one aim of this study is to search for variations in the gc populations among a sample of early - type galaxies in the same cluster . in this paper
we examine the globular cluster populations of ngc 1399 , ngc 1404 , and ngc 1316 . in companion papers we study the metallicity and spatial distributions of globular clusters in ngc 1399 and ngc 1404 ( forbes et al .
1998 ) and examine the globular cluster system of the lower luminosity elliptical ngc 1379 ( elson et al . 1998 ) .
ngc 1399 is the cd elliptical at the center of the fornax cluster and , like m 87 , has an extraordinarily large number of globular clusters for its luminosity ( see references in ashman & zepf 1997 ) .
ngc 1404 is an e1 elliptical about half a magnitude fainter and lying only 10 arcminutes from ngc 1399 in projection .
attempts to study its globular cluster system ( hanes & harris 1986 ; richtler et al .
1992 ) have been complicated by its location within the extended envelope of ngc 1399 .
ngc 1316 is another giant elliptical in the fornax cluster , more than a magnitude brighter than ngc 1399 , with a double - lobed radio source ( fornax a ) , an x - ray halo , and dust and other irregularities ( schweizer 1980 ) indicating a recent merger .
ngc 1316 is situated more than three degrees away from ngc 1399 , clearly dominating its corner of the fornax cluster , and has been the subject of a recent wfpc1 study by shaya et al .
( 1996 ) .
if the high specific frequency of globular clusters around ngc 1399 is to be attributed to tidal stripping or accretion of existing globular clusters in other cluster galaxies ( which would be consistent with the kinematic data of grillmair et al .
1994a and kissler - patig et al .
1997b ) , we might expect mean gc metallicities to be very similar to those of other cluster galaxies .
alternatively , if much of ngc 1399 s globular cluster system formed more recently from enriched gas , its globulars may be overly metal - rich compared to other galaxies of the same magnitude . if new globular clusters are formed in the normal course of mergers ( ashman & zepf 1992 ) , then we might expect to see a a number of young , metal - rich globulars surrounding ngc 1316 .
we describe the observations and their analysis in section [ sec : obs ] .
the color distribution of detected globular clusters is discussed in section [ sec : color ] .
we analyze and compare the luminosity functions in section [ sec : lf ] .
section [ sec : n1316 ] more closely examines the characteristics of objects around ngc 1316 , and section [ sec : inter ] briefly discusses the possibility of intergalactic globular clusters in our sample .
we summarize our conclusions in section [ sec : summary ] .
the metallicity and spatial distributions of globular clusters in these galaxies and their implications for formation scenarios are discussed by forbes et al .
an observing log for program go # 5990 is given in table 1 .
while ngc 1379 was observed as an example of a `` normal '' elliptical , telescope acquisition problems prevented useful f814w data from being taken by the time of writing .
the ngc 1379 observations in the f450w filter are discussed by elson et al . (
field 0338 ( f0338 ) is situated in the cd envelope of ngc 1399 , at approximately the same projected distance from ngc 1399 as ngc 1404 , but on the opposite side of the galaxy .
field 0336 ( f0336 ) is situated approximately 1.4 degrees south of ngc 1399 in a blank region of sky and serves to measure the surface density of background sources .
the relative positions and orientations of these fields are shown in figure [ fig : plate1 ] . filters f450w ( @xmath4 ) and f814w ( @xmath5 ) were chosen to increase metal sensitivity by about a factor of two over that attainable using @xmath6 ( brodie 1981 ; geisler , lee , & kim 1996 ) .
the images were adc - corrected , bias - subtracted , and flat - fielded in the course of standard pipeline processing .
lcccccc ngc 1399 & 03 38 29.0 & -35 27 00.5 & 1996 jun 2 & f814w & 3 @xmath7 600 & 242.0 ngc 1399 & & & 1996 jun 2 & f450w & 4 @xmath7 1300 & 242.0 ngc 1379 & 03 36 03.9 & -35 26 26.0 & 1996 mar 11 & f814w & 3 @xmath7 0.5 & 148.3 ngc 1379 & & & 1996 mar 11 & f450w & 2 @xmath7 1300 + 2 @xmath7 500 + 1400 & 148.3 ngc 1404 & 03 38 51.7 & -35 35 36.0 & 1996 apr 3 & f814w & 260 + 600 + 1000 & 171.8 ngc 1404 & & & 1996 apr 3 & f450w & 2 @xmath7 500 + 2 @xmath7 1300 + 1400 & 171.8 ngc 1316 & 03 22 41.8 & -37 12 29.8 & 1996 apr 7 & f814w & 260 + 600 + 1000 & 181.3 ngc 1316 & & & 1996 apr 7 & f450w & 2 @xmath7 500 + 2 @xmath7 1300 + 1400 & 181.3 field 0338 & 03 37 57.0 & -35 21 54.5 & 1996 apr 6 & f814w & 3 @xmath7 600 & 175.3 field 0338 & & & 1996 apr 6 & f450w & 4 @xmath7 1300 & 175.3 field 0336 & 03 36 02.7 & -36 45 54.0 & 1996 apr 11 & f814w & 3 @xmath7 600 & 182.4 field 0336 & & & 1996 apr 11 & f450w & 4 @xmath7 1300 & 182.4 each set of between 3 and 5 exposures of a given field was split into two pointings offset by 05 from one another .
this corresponds approximately to integer pixel shifts in both the pc ( 00455 pix@xmath3 ) and the wf chips ( 00996 pix@xmath3 ) ( the actual shifts are generally within 0.1 pixels of integer values ) . aligning and median - combining the images consequently allowed us to eliminate the majority of cosmic rays and hot pixels at the same time . to prevent as much as possible the contamination of the faint end of the gclf , known warm / hot pixels , charge traps , and bad columns were flagged and subsequently ignored in the median procedure .
the distribution of unflagged pixels was further tested for the presence of cosmic ray events before computing the median value from the remaining pixels .
the results were found to be cleaner both statistically and visually when compared with the results of such standard cosmic - ray removal procedures as are available , for example , within stsdas / iraf .
the pc images of the sample galaxies were analyzed using the vista surface - brightness profile - fitting routine snuc , which fits logarithmically - spaced , concentric ellipses to unmasked portions of the galaxy profile .
the resulting model profiles were then subtracted from each image in an effort to flatten the background as much as possible . for the wf chips ,
similar model surface brightness distributions were generated using both large - scale window - medianing , and convolution with a two - dimensional gaussian having @xmath8 .
luminosity fluctuations in f0336 and f0338 were found to be on scales sufficiently large that model subtraction was deemed unnecessary .
source detection and aperture photometry were carried out using daophot ii ( stetson 1987 ) .
owing to the high resolution inherent in these images , overlap of sources was negligible and aperture photometry yielded a slightly narrower color distribution than could be obtained using psf - fitting photometry .
the detection threshold was fixed at three times the rms expected _ locally _ from photon statistics and readout noise , the former computed by taking into account the subtracted model surface - brightness distributions . given the degree of undersampling inherent to the wfpc2 detectors , a @xmath9 detection threshold predictably yielded a substantial number of spurious detections .
however , the final photometry table was generated using only those sources which were detected in both passbands and having colors in the range @xmath10 .
consequently , the number of spurious detections remaining in the final sample ( as determined by visual inspection ) was found to be negligible .
ngc 1316 is something of a special case in our sample owing to the presence of large amounts of dust obscuration . for this galaxy
, we masked large areas of the wfpc2 field of view in which there was visible evidence of dust in f450w .
the extent of the dust and the regions we chose to exclude from analysis are shown in figure [ fig : plate2 ] .
the `` unobscured '' detections attributed to ngc 1316 thus come from approximately the outer halves of each of the wf chips ; detections in the pc are not considered in the subsequent analysis owing to the strong likelihood that all detections are obscured and reddened to a significant extent .
faint background galaxies and bright foreground stars often produced multiple detections along spiral arms or diffraction spikes , and several small regions of each image were masked accordingly .
in addition , we disregarded all detections with @xmath11 ( the difference in magnitude computed for apertures of radius 1 and 2 pixels , respectively ) @xmath12 magnitudes ( @xmath13 magnitudes for the pc ) .
visual examination showed that this effectively removed almost all obvious faint galaxies from the sample .
a more extensive discussion of the contamination issue is presented by elson et al .
( 1998 ) , and as we show presently , the number of background objects in our sample to @xmath14 is @xmath15 . at the distance of the fornax cluster ,
typical globular clusters are marginally resolved by wfpc2 . treating them as point sources from the standpoint of photometry would therefore underestimate to varying degrees their total luminosities ( see holtzman et al .
1996 for a discussion of this problem ) . to estimate this shortfall
, we convolved an array of king models ( king 1966 ) with wfpc2 psfs generated using tiny tim ( krist 1995 ) for each globular s position on the detector .
we then sampled these models using wfpc2-sized pixels and a @xmath16 grid of pixel - centerings .
each model realization was then compared on a pixel - by - pixel basis with the real data in the ngc 1399 images for @xmath17 pixels to compute @xmath18 . while this procedure yielded a range of best - fit core radii not unlike that found in galactic globular clusters , it failed sufficiently often that we do not regard the computed core radii as reliable .
the reasons for this may include the fact that we did not take into account image dithering and subsequent integer - pixel shifting used to combat hot pixels , as well as inaccuracies in the tiny tim model psfs .
however , aperture corrections derived from king model fits with acceptable values of @xmath18 agreed very well with those determined empirically for the brightest , most isolated clusters in the field .
we therefore adopted mean 05 aperture corrections to our 2-pixel - radius measurements determined directly from the images in each frame , ranging from @xmath19 magnitudes for the wf chips to @xmath20 magnitudes for the pc . based on our simulations ,
the use of mean aperture corrections applied to all clusters in a given wf chip is likely to introduce a random error of @xmath21 magnitudes to the photometry of any individual cluster . to convert to @xmath22 and @xmath23 magnitudes we used the gain ratios and zeropoints given by holtzman et al .
( 1995 ; 1997 ) .
the final magnitudes and colors of all detected objects are plotted in figure [ fig : cm ] , and are available in electronic form from cjg . = 4.0 in completeness tests were carried out by adding a total of @xmath24 artificial stars to each image and processing the results in a manner identical to that used for the original data .
artificial globulars ( generated from a composite psf derived from real images in each frame ) were added in batches of 100 with successively fainter @xmath22 magnitudes , and with colors @xmath25 and 2.3 magnitudes , respectively .
the results of these tests are shown for ngc 1399 in figure [ fig : compfig ] and figure [ fig : returned ] .
the 50% completeness level for the bulk of the detected objects occurs at @xmath26 in the wf frames , and @xmath27 in the pc images .
for f0336 , the 50% completeness level for the same color occurs at @xmath28 in the wf frames , and @xmath29 in the pc .
the completeness tests for ngc 1399 yield an estimate for the photometric uncertainty for @xmath30 of @xmath31 mag rms at @xmath22 = 25 , and @xmath32 mag rms at @xmath22 = 26 .
= 4.0 in = 4.0 in
in figure [ fig : color ] we show the numbers of globular clusters found per 0.1 mag color interval brighter than @xmath33 in each of the five fields .
the corresponding surface densities for the three galaxies are tabulated in table 2 .
the shaded regions in figure [ fig : color ] show the color distributions for objects brighter than @xmath34 . assuming that the color distribution of any intergalactic globular clusters roaming the fornax cluster should be similar to that seen in the three galaxies in our sample , the distribution apparent in f0336 indicates that such clusters are rare , and that the apparent distribution of colors is primarily representative of unresolved background galaxies .
completeness fractions were computed by spline - interpolation of our color - magnitude completeness grid . correcting each bin by simply dividing by the computed completeness fraction and subtracting the f0336 distribution yields the distributions indicated by the dotted lines in figure [ fig : color ] .
= 4.0 in cccccccc -0.1 0.0 & 0 & 0.0 & 2 & 0.4 & 0.1 & 1 & 3.8 0.0 0.1 & 1 & 0.0 & 0 & 0.0 & 0.0 & 1 & 0.6 0.1 0.2 & 1 & -0.2 & 1 & -0.2 & 0.2 & 1 & 0.3 0.2 0.3 & 1 & 0.2 & 0 & 0.0 & 0.0 & 1 & 1.7 0.3 0.4 & 0 & 0.0 & 0 & 0.0 & 0.0 & 0 & 0.0 0.4 0.5 & 0 & -0.4 & 1 & -0.2 & -0.4 & 2 & 0.8 0.5 0.6 & 1 & 0.0 & 1 & 0.0 & -0.4 & 2 & 1.4 0.6 0.7 & 0 & -0.2 & 2 & 0.3 & -0.1 & 3 & 1.5 0.7 0.8 & 2 & 0.5 & 1 & 0.2 & -0.4 & 1 & 0.6 0.8 0.9 & 3 & -0.1 & 2 & -0.5 & -0.2 & 4 & 1.7 0.9 1.0 & 1 & 0.0 & 0 & -0.2 & -0.2 & 6 & 3.0 1.0 1.1 & 3 & 0.7 & 4 & 1.0 & 1.0 & 4 & 2.2 1.1 1.2 & 0 & 0.0 & 3 & 0.7 & 0.3 & 2 & 1.0 1.2 1.3 & 1 & 0.2 & 0 & 0.0 & -0.6 & 4 & 2.1 1.3 1.4 & 4 & 0.7 & 2 & 0.2 & 0.0 & 10 & 4.3 1.4 1.5 & 15 & 2.8 & 7 & 0.9 & 1.1 & 7 & 2.7 1.5 1.6 & 46 & 9.7 & 14 & 2.7 & 0.6 & 11 & 4.8 1.6 1.7 & 50 & 10.8 & 24 & 5.1 & 4.5 & 18 & 8.0 1.7 1.8 & 46 & 10.4 & 23 & 4.8 & 4.1 & 21 & 9.2 1.8 1.9 & 48 & 10.2 & 17 & 3.6 & 3.0 & 17 & 7.5 1.9 2.0 & 59 & 12.6 & 21 & 4.1 & 3.6 & 16 & 6.6 2.0 2.1 & 88 & 18.7 & 29 & 6.1 & 5.9 & 12 & 5.1 2.1 2.2 & 67 & 14.2 & 26 & 5.5 & 5.1 & 7 & 3.1 2.2 2.3 & 70 & 14.7 & 16 & 3.4 & 3.4 & 1 & 0.4 2.3 2.4 & 19 & 4.0 & 8 & 1.7 & 1.5 & 1 & 0.4 2.4 2.5 & 3 & 0.6 & 2 & 0.4 & 0.4 & 1 & 0.4 2.5 2.6 & 2 & 0.5 & 0 & 0.0 & 0.0 & 0 & 0.0 2.6 2.7 & 2 & 0.4 & 0 & 0.0 & 0.0 & 0 & 0.0 2.7 2.8 & 0 & 0.0 & 0 & 0.0 & 0.0 & 0 & 0.0 the color distributions of globular clusters in both ngc 1399 and ngc 1404 are evidently bimodal .
moreover , the colors of the peaks and the overall width of the color distributions are almost identical , though ngc 1399 may have a slightly larger proportion of red clusters .
the mean @xmath35 colors of the two samples are @xmath36 and @xmath37 , and a kolmogorov - smirnov test indicates that the hypothesis that the two populations were drawn from the same parent population can not be rejected at the 63% confidence level .
based on the ratios of the numbers of globular clusters counted to @xmath38 and @xmath39 , it appears that the bluer clusters are on average somewhat brighter than the red clusters .
we discuss this finding further in section [ sec : lf ] and in forbes et al .
( 1998 ) .
the higher surface density of globular clusters around ngc 1404 , as well as the difference in mean colors when compared with a similar field on the opposite side of ngc 1399 ( f0338 ) , indicate that these globular clusters indeed belong to ngc 1404 as opposed to the much more populous and extended population of ngc 1399 .
taking into account the orientations and the area - weighted centers of the wfpc2 field - of - view in ngc 1404 and f0338 , we find that the f0338 field is 0.39 closer in projection to ngc 1399 than the ngc 1404 field . assuming a radial surface density profile of ngc 1399 globular clusters which goes as @xmath40 ( kissler - patig et al .
1997a ) , we apply a correction factor of 0.93 to the surface density of globulars detected in f0338 to match the surface density of ngc 1399 globulars expected at the radius of ngc 1404 . subtracting the resulting , completeness - corrected distribution of globular clusters in f0338 from that of ngc 1404 yields the dotted line in figure [ fig : color ] .
the total number of globular clusters and the mean color are minimally affected , supporting the notion that the majority of these globular clusters are indigenous to ngc 1404 .
furthermore , the similarity in the color distributions of ngc 1399 and ngc 1404 are consistent with the idea that ngc 1399 s overabundance of globular clusters may have come partly at the expense of ngc 1404 and other galaxies in the fornax cluster ( forbes et al .
1997b ; kissler - patig et al .
this does not , of course , rule out the possibility that the red and blue clusters originated in different galaxies or under different circumstances .
interestingly , the color distribution of compact sources in f0338 appears to be significantly different from those seen in the galaxy - centered fields . even after subtraction of the color distribution in f0336 ( the background field ) , the colors of objects in f0338 peak at @xmath41 , which is 0.1 magnitudes bluer than the centroid of the bluest peak in either ngc 1399 or ngc 1404 .
there is also a group of objects with @xmath42 which appears to significantly exceed the number of such objects found in the f0336 .
these findings are consistent with the presence of a significant color gradient , as first reported by ostrov et al .
these observations and their consequences are discussed in detail by forbes et al .
( 1998 ) . in spite of the possibility of unseen , distributed dust , the `` unobscured '' objects in ngc 1316 are on average 0.45 magnitudes bluer in @xmath35 than either those in ngc 1399 or ngc 1404 .
the peak in the number counts occurs just redward of the position of the bluest peak in ngc 1399 and ngc 1404 , with a significant tail of objects with @xmath43 .
the bluer objects are much less luminous than the bright , blue clusters found by holtzman et al .
( 1992 ) in ngc 1275 .
the lack of bright , blue globulars , while perhaps surprising given the ample evidence for a recent merger event , agrees with the findings of shaya et al .
however , this may not be entirely surprising given the much smaller quantity of dust visible in ngc 1316 than in ngc 1275 . and
despite the arguments of shaya et al .
( 1996 ) , we can not rule out bright young clusters in the obscured , highly reddened central regions of the galaxy .
very young ( @xmath44 5 myrs ) _ stellar _ populations do produce stars as bright as @xmath45
. however , such stars would mostly be far hotter and bluer than the objects we detect in ngc 1316 .
shown in figure [ fig : cm ] is the @xmath0 color - magnitude sequence of the turnoff and horizontal branch expected for a stellar population of age @xmath46 yrs and [ fe / h ] = -0.4 ( worthey , private communication ) . while bearing a general resemblance to the distribution of the faintest objects in ngc 1316 , the luminosity function for such a stellar population requires a very much larger number of hot , blue objects ( @xmath47 ) than we actually find .
many of the objects surrounding ngc 1316 are similar in color and magnitude to the old open clusters or the faint , metal - poor , outlying globular clusters in our own galaxy such as pal 1 and eridanus . based on the models of worthey ( 1994 ) , the objects with @xmath48 and @xmath49 are consistent with a population of old , metal - poor clusters with masses of order @xmath50 m@xmath51 , or metal - rich clusters with ages as young as 1 gyr and masses of order @xmath52 m@xmath51 .
we discuss the nature of these objects further in section [ sec : n1316 ] .
both whitmore et al .
( 1995 ) and elson & santiago ( 1996a , 1996b ) used _ hst _ data to confirm the presence of color dichotomies in the gc system of m 87 first suggested by the ground - based data of lee & geisler ( 1993 ) .
the color differences between the blue and red peaks in the color histograms of globular clusters in m 87 are @xmath53 magnitudes .
however , the @xmath0 color is about twice as sensitive to metallicity as @xmath54 .
if we assume that age is not an important contributor to the spread in color , and if we use couture , harris , & allwright s ( 1990 ) color - metallicity calibrations for milky way globular clusters , we find that the full - width - at - half - maxima for the color distributions of elson & santiago ( 1996 ) and whitmore et al .
( 1995 ) ( @xmath55 mag ) correspond to a spread in [ fe / h ] of about 2.3 dex .
ngc 1399 and ngc 1404 have @xmath56 in both ngc 1399 and ngc 1404 , yielding a metallicity spread @xmath57 [ fe / h ] @xmath58 dex .
these values are in good agreement with the metallicity spread inferred from @xmath59 measurements of the brighter globulars in ngc 1399 by ostrov et al .
( 1993 ) . using only the color differences between peaks in m 87 ,
ngc 1399 , and ngc 1404 , we find @xmath57 [ fe / h ] @xmath60 and 1.1 dex , respectively . as found in previous ground - based studies ( see geisler et al .
1996 for a tabulation ) , the ranges of globular cluster metallicities to be found in m 87 and ngc 1399 ( and other bright ellipticals ) are similar , spanning an interval which extends from the metal - poorest globulars in our galaxy to a metallicity significantly higher than that of the sun .
the luminosity functions ( lfs ) for our three target galaxies are given in table 3 and are plotted in figures [ fig : lf ] and [ fig : lfi ] .
the plotted lfs for ngc 1399 and ngc 1316 have been corrected for completeness , and background - corrected by subtracting the observed , completeness - corrected lf of f0336 .
similarly , the gclf for ngc 1404 has been adjusted using the completeness - corrected ( but not background subtracted ) lf in f0338 after scaling by a factor of 0.93 as described earlier to account for the fact that the center of the f0338 field is 04 closer to ngc 1399 than is ngc 1404 .
completeness corrections were made for each globular cluster counted based on the wfpc2 chip on which it was found , but no account was taken of the cluster s position on the chip .
= 4.0 in = 4.0 in ccccccc 20.120.3 & 0 & 0.0 & 0 & 0.0 & 0 & 0.0 20.320.5 & 0 & 0.0 & 0 & 0.0 & 1 & 0.4 20.520.7 & 0 & 0.0 & 1 & 0.0 & 0 & 0.0 20.720.9 & 0 & 0.0 & 1 & 0.2 & 0 & 0.0 20.921.1 & 0 & 0.0 & 0 & 0.0 & 0 & 0.0 21.121.3 & 2 & 0.4 & 0 & 0.0 & 0 & 0.0 21.321.5 & 0 & 0.0 & 0 & 0.0 & 0 & 0.0 21.521.7 & 0 & 0.0 & 0 & 0.0 & 0 & 0.0 21.721.9 & 3 & 0.6 & 1 & 0.0 & 0 & 0.0 21.922.1 & 2 & 0.4 & 1 & 0.2 & 1 & 0.4 22.122.3 & 8 & 1.7 & 0 & 0.0 & 1 & 0.4 22.322.5 & 10 & 1.9 & 3 & 0.6 & 2 & 0.6 22.522.7 & 15 & 2.5 & 4 & 0.4 & 1 & -0.2 22.722.9 & 14 & 2.9 & 4 & 0.8 & 2 & 0.8 22.923.1 & 15 & 3.2 & 5 & 0.7 & 3 & 1.3 23.123.3 & 25 & 5.3 & 4 & 0.4 & 5 & 2.1 23.323.5 & 26 & 5.5 & 8 & 1.3 & 4 & 1.7 23.523.7 & 28 & 5.9 & 15 & 2.9 & 2 & 0.8 23.723.9 & 29 & 6.1 & 6 & 0.9 & 3 & 1.3 23.924.1 & 33 & 6.7 & 22 & 4.2 & 4 & 1.5 24.124.3 & 46 & 9.7 & 6 & 1.1 & 4 & 1.7 24.324.5 & 41 & 8.7 & 22 & 3.5 & 5 & 2.1 24.524.7 & 39 & 8.3 & 16 & 2.0 & 9 & 3.8 24.724.9 & 35 & 7.2 & 8 & 0.7 & 14 & 5.9 24.925.1 & 35 & 7.5 & 18 & 3.5 & 8 & 3.5 25.125.3 & 39 & 7.8 & 15 & 1.9 & 18 & 7.5 25.325.5 & 30 & 6.4 & 16 & 2.3 & 17 & 8.2 25.525.7 & 28 & 7.4 & 14 & 3.8 & 25 & 11.8 25.725.9 & 24 & 4.5 & 15 & 3.7 & 22 & 13.9 25.926.1 & 20 & 4.0 & 13 & 1.3 & 15 & 6.1 26.126.3 & 17 & 1.6 & 6 & 0.1 & 21 & 8.2 26.326.5 & 19 & 3.9 & 5 & -2.0 & 19 & 14.5 the lf for objects in ngc 1316 is clearly different from those of ngc 1399 and ngc 1404 .
in addition to a bluer mean color , the number of objects in ngc 1316 increases rapidly faintwards and does not show a log - normal distribution of the kind normally seen in globular cluster systems .
we discuss ngc 1316 s cluster system in more detail below .
the gclfs for ngc 1399 and ngc 1404 appear to be very similar .
a straightforward comparison using the kolmogorov - smirnov test reveals that the null hypothesis ( that the two populations were drawn from the same parent population ) can not be rejected at the 90% confidence level .
if the magnitudes of ngc 1404 globular clusters are offset by @xmath61 magnitudes ( see below ) , the k - s test yields a probability of 74% in favor of the null hypothesis .
after binning , completeness correction , and background subtraction , a @xmath18 test over the range @xmath62 gives an 86% probability in favor of the null hypothesis .
we have used version 2.0 of the peak - finding code described by secker & harris ( 1993 ) ( and kindly provided by j. secker ) to compute the maximum - likelihood peaks and dispersions of both gaussian and student s @xmath63 functions from our gclf sample .
owing to the current limitations of the code we have used only the globular clusters detected in the wf frames , for which the completeness functions are reasonably similar . as we are neglecting only 10% ( 20% ) of the total number of clusters detected around ngc 1399 ( ngc 1404 ) ,
the statistics are not seriously affected .
for the present purposes we use only clusters with colors in the range @xmath64 .
the results are given in table 4 .
we also include maximum - likelihood fits to the milky way and m 31 cluster data described below , and we quote the values given by whitmore et al .
( 1995 ) for the _ hst _ @xmath65-band data of m 87 .
we assume reddenings of e(b - v ) = 0.11 and 0.01 for m 31 and fornax , respectively ( burstein & heiles 1982 ; schlegel , finkbeiner , & davis 1998 ) . figure [ fig : contours ] shows the @xmath18 contours for the ngc 1399 @xmath22-band gclf gaussian fit in the peak - dispersion plane .
for each entry in table 4 the listed uncertainties correspond to the projections of the @xmath66 contours onto the peak magnitude and dispersion axes .
the gaussian which best fits the combined ngc 1399 + ngc
1404 data is shown in figure [ fig : lg ] .
the value of @xmath67 we find for ngc 1399 is in good agreement with the value of @xmath68 ( after extinction correction using our adopted reddening ) determined by kohle et al .
( 1996 ) from ground - based data .
= 3.0 in lcccccc milky way & -6.49 @xmath69 0.19 & 1.54 @xmath69 0.15 & -6.61 @xmath69 0.17 & 1.28 @xmath69 0.15 & & m 31 & 17.38 @xmath69 0.11 & 1.05 @xmath69 0.08 & 17.41 @xmath69 0.11 & 0.90 @xmath69 0.09 & & ngc 1399 & 24.55 @xmath69 0.08 & 1.19 @xmath69 0.05 & 24.56 @xmath69 0.08 & 1.07 @xmath69 0.06 & 23.73 @xmath69 0.08 & 1.19 @xmath69 0.05 ngc 1404 & 24.82 @xmath69 0.20 & 1.41 @xmath69 0.15 & 24.84 @xmath69 0.21 & 1.20 @xmath69 0.15 & & ngc 1399 ( @xmath70 ) & 24.71 @xmath69 0.08 & 1.16 @xmath69 0.06 & 24.70 @xmath69 0.08 & 1.05 @xmath69 0.07 & & ngc 1399 ( @xmath71 ) & 24.32 @xmath69 0.12 & 1.24 @xmath69 0.09 & 24.34 @xmath69 0.09 & 1.10 @xmath69 0.08 & & ngc 1399 + ngc 1404 & 24.62 @xmath69 0.07 & 1.23 @xmath69 0.05 & 24.62 @xmath69 0.08 & 1.09 @xmath69 0.05 & 23.77 @xmath69 0.06 & 1.22 @xmath69 0.05 m 87 & & & & & 23.72 @xmath69 0.06 & 1.40 @xmath69 0.06 the peak magnitudes computed for ngc 1399 and ngc 1404 differ by @xmath72 magnitudes in the sense that ngc 1404 may lie some 2.1 mpc beyond ngc 1399 , though the uncertainty is large .
the gclf width for ngc 1404 is evidently slightly broader than that for ngc 1399 , though again the uncertainties on the ngc 1404 result do not rule out identical widths .
as this is the first time we have been able to examine the lf of fornax globular clusters to this depth , it is interesting to compare it with the lfs of local group globular clusters . in figure
[ fig : lg ] we compare the background - corrected , combined gclf for ngc 1399 and ngc 1404 ( after shifting the latter by @xmath73 magnitudes ) to the lfs of globular clusters in our own galaxy and m 31 .
the milky way gclf derives from the halo globular cluster compilation of harris ( 1996 ) on the mcmaster university www page .
the data for globular clusters in m 31 comes from the halo sample tabulated by reed et al .
( 1994 ) , and we assume ( m m)@xmath74 = 24.43 for m 31 ( ajhar et al .
the local group gclfs are compared with the completeness- and background - corrected gclf of ngc 1399 + ngc 1404 , shifted assumed ( m m)@xmath74 = 31.05 ( see below ) and @xmath75 ( schlegel , finkbeiner , & davis 1998 ) .
the local group gclfs have been normalized to the total number of gcs counted in ngc 1399 and ngc 1404 with @xmath76 .
= 4.0 in while the milky way gclf appears in most respects to be fairly similar in form to the gclf for ngc 1399 + 1404 , the gclf for m 31 globular clusters appears to be somewhat more peaked .
the computed gclf widths determined for the galaxy and m 31 are broader and narrower , respectively , than we find for the early - type galaxies .
the uncertainty on the gclf width of galactic globulars is large enough to accommodate the fornax and m 87 results , but the m 31 distribution is clearly at odds with all others .
the mean magnitudes and widths we find for the m 31 gclf are consistent with those determined for the outer - halo m 31 globulars by kavelaars & hanes ( 1997 ) .
incompleteness of the m 31 sample at the faint end , though difficult to quantify , must be at least partially responsible for this discrepancy . in any event , the numbers of local group globular clusters are too small to allow strong conclusions to be made .
a @xmath18 test comparing the ngc 1399 + 1404 gclf with the milky way and m 31 gclfs indicates that the null hypothesis remains valid at the 66% and 63% confidence levels , respectively . similarly comparing the ngc 1399 + 1404 gclf with the combined milky way and m 31
gclfs yields a probability for the null hypothesis of greater than 64% .
ashman , conti , & zepf ( 1995 ) have shown that good agreement between the measured gclf peaks of early - type and late - type galaxies could be achieved by accounting for the different opacities and increased line - blanketing expected in higher metallicity systems . for ngc 1399 in particular , they predict an offset @xmath77 mag with respect to the peak observed in the milky way .
using this offset and the values in table 4 , we find @xmath78 .
ashman et al.s ( 1995 ) models were computed using only gaussian dispersions . however , since their model distributions were symmetric , the @xmath79 for @xmath63 distribution should be very similar .
the peak magnitudes computed for the milky way and ngc 1399 using the @xmath63 function then yield @xmath80 = 30.91 @xmath69 0.19 .
the agreement between these distance moduli and those determined using other distance - measuring techniques ( see below ) underscores the remarkable uniformity in the underlying mass functions of globular clusters in very different galaxy types .
also shown in figure [ fig : lg ] are the lfs for the 230 bluer ( @xmath81 ) and the 348 redder ( @xmath70 ) globular clusters in ngc 1399 . according to table 4 there is no significant difference between the widths of these two distributions , and it appears from figure [ fig : lg ] that a simple faintward shift of the blue distribution could bring it into line with the red one .
this is reflected in the mean and peak magnitudes in table 4 , which differ by @xmath82 mags ( @xmath63 ) between the red and blue samples . after shifting the red lf by -0.36 magnitudes ,
a @xmath18 test gives a probability for the null hypothesis of 51% .
current data for two positions in m 87 ( whitmore et al .
1995 ; elson & santiago 1996b ) and ngc 5846 ( forbes et al .
1997a ) are consistent with the blue peak being from 0.0 to 0.3 magnitudes brighter in @xmath65 than the red peak . if attributed entirely to metallicity effects , the 0.36 @xmath22-band magnitude difference we find for ngc 1399 clusters would correspond to @xmath83 \approx 1 $ ] dex ( couture et al .
1990 ) , and would produce a @xmath65-band difference between the two peaks of @xmath84 magnitudes . in figure [
fig : m87 ] we compare the gclf for ngc 1399 with the @xmath65-band , _ hst _ gclf for m 87 of whitmore et al .
we have converted our @xmath22-band measurements to @xmath65-band for the comparison using @xmath85 which is a least - squares fit to the least - reddened ( e(b - v ) @xmath86 ) galactic globular clusters as compiled on the mcmaster www page .
simply comparing values of @xmath87 in table 4 reveals that the peak of the ngc 1399 gclf is 0.01 @xmath69 0.10 magnitudes fainter than that of m 87 .
this is in good agreement with offsets of @xmath88 magnitudes found by kohle et al .
( 1996 ) and @xmath89 magnitudes found by blakeslee & tonry ( 1996 ) ) from deep , ground - based gclf studies , after correction for our adopted reddening .
the ngc 1399 gclf was recomputed after applying a shift of @xmath90 to the computed @xmath65 magnitudes to bring the peak into line with that of m 87 as per table 4 .
the resulting @xmath65-band luminosity function is compared with a scaled version of whitmore et al.s ( 1995 ) m 87 gclf in the upper panel of figure [ fig : m87 ] .
= 4.0 in there is clearly very little difference between the gclfs of the two cd galaxies . in the lower panel of figure [ fig : m87 ]
we show the bin - by - bin ratio of the normalized m 87 luminosity function to that of ngc 1399 .
the gclfs are evidently indistinguishable over most of the range of comparison , though there may be proportionately more very bright globulars in m 87 than there are in ngc 1399 .
a @xmath18 test indicates that the null hypothesis can not be rejected at the 69% confidence level .
the gclf width listed in table 4 for ngc 1399 is significantly narrower ( @xmath91 ) than that found by whitmore et al .
1995 for m 87 .
this is due in part to the very bright m 87 globulars noted above , and in part to differences in the methods used to measure the widths ; the method of secker & harris ( 1993 ) we have adopted here takes into account the broadening of the lf due to photometric errors .
the gclf width we find for ngc 1399 is very similar to those found for ngc 4278 , ngc 4494 ( forbes 1996b ) and ngc 1404 , but is significantly narrower than those of either ngc 4365 gclf ( forbes 1996a ) or ngc 5846 ( forbes et al .
1996 ) , all measured using the code of secker & harris ( 1993 ) . thus , ngc 1399 does not support kissler - patig et al.s ( 1997a ) suggestion that gclf properties vary little among the lesser galaxies , be they early or late - type , but that they vary significantly among cluster - dominating ellipticals . adopting for m 87 a distance of 16.1 @xmath69 1.3 mpc ( ferrarese et al .
1997 ; yasuda et al .
1997 ) , then the relative shift of 0.01 @xmath69 0.1 magnitudes we find for the gclf of ngc 1399 gives a distance to fornax of 16.2 @xmath69 1.5 mpc .
this is in good agreement with a fornax distance of 16.9 @xmath69 1.1 mpc determined by mcmillan , ciardullo , & jacoby ( 1993 ) using planetary nebula luminosity functions , though in somewhat poorer agreement with the distance of 18.4 @xmath69 1.8 mpc determined by madore et al .
( 1996 ) using _ hst _ observations of cepheids in ngc 1365 .
while cepheids are believed to be more accurate distance indicators than gclfs , it may also be that ngc 1365 lies well out at the periphery of the fornax cluster , as spiral galaxies are wo nt to do .
since ngc 1399 sits almost exactly at the center of the cluster s
potential well , we can use our inferred distance to compute a value for the hubble constant without incurring the additional uncertainty introduced by the expanse of the fornax cluster itself .
we use a heliocentric velocity for fornax of 1450 @xmath69 34 km s@xmath3 ( held & mould 1994 ) . correcting for the solar motion with respect to the local group ( 91 km s@xmath3 ; yahil , tammann , & sandage 1977 ) , and a virgocentric infall component along the line of sight to fornax of 36 km s@xmath3 ( tammann & sandage 1985 )
, we obtain an expansion - induced velocity of 1323 @xmath92 km s@xmath3 .
this then yields a hubble constant @xmath93 km s@xmath3 mpc@xmath3 , where the uncertainty reflects random measurement errors only .
this agrees reasonably well with a value of 73 @xmath69 6 ( random ) @xmath69 8 ( systematic ) km s@xmath3 mpc@xmath3 found by madore et al .
( 1996 ) using , among other things , _
hst_-measured cepheid distances to several nearby galaxy groups including virgo and fornax .
why is the lf of clusters in ngc 1316 not log - normal in form , as are those of almost all globular cluster systems studied to date ?
one might postulate that the lf of objects surrounding ngc 1316 reflects a young system which has yet to undergo depletion by the disruption of its faintest members .
removal of the majority of objects fainter than @xmath94 could conceivably leave behind a log - normal lf .
however , the number of objects at @xmath95 is small , and such a depletion would leave ngc 1316 with roughly as many clusters as ngc 1404 .
the specific frequency of globular clusters in ngc 1404 is already quite low ( @xmath96 , forbes et al .
1997 ) , and given the 1.4 magnitude difference in total luminosity between these two galaxies , we would be left with a remarkably low @xmath97 for ngc 1316 .
while this would be highly unusual for any early - type galaxy , it would be even more surprising in view of ngc 1316 s evident merger history and apparent dominance of its corner of the fornax cluster .
if ngc 1316 has undergone a significant starburst and is destined to fade by at least a factor of four in the time required to disrupt the fainter clusters , it is conceivable that @xmath98 might eventually approach a value more typical of early - type galaxies .
however , a direct comparison between the wf3 images of ngc 1316 and ngc 1404 reveals that the ratio in the total luminosities of these two galaxies is largely preserved in surface brightness measurements at the larger radii where ngc 1316 appears to be free of dust .
it seems unlikely that significant recent star formation has taken place at these large radii , and that we must therefore be looking primarily at old stars .
thus , even if there is substantial fading of a younger stellar population residing at the center of ngc 1316 , the total magnitude ( and therefore @xmath98 ) will be largely unaffected .
the luminosity function of the objects we see in ngc 1316 has been seen in other galaxies as well , ngc 4038/4039 ( whitmore & schweizer 1995 ) and ngc 3597 ( carlson et al .
1998 ) being two well - studied cases .
superposed on the ngc 1316 luminosity function in figure [ fig : lf ] is the `` completeness - corrected '' luminosity function of galactic open clusters tabulated by van den bergh & lafontaine ( 1984 ) , scaled upwards by a factor of 10 . while there may be relatively fewer bright objects in ngc 1316 than there are open clusters in the galaxy , the luminosity functions are reasonably similar over the range for which we have good statistics . in figure
[ fig : m1m2 ] we show the distribution of clusters over @xmath11 ( the difference between magnitudes measured for one- and two - pixel - radius apertures ) and magnitude for ngc 1316 s unobscured clusters and for ngc 1399 globular clusters which fall in the same , outer portions of the wfpc2 detectors .
( by limiting our comparison to the same regions of the detectors we minimize the effects that position - dependent changes in the shape of the psf might have on object centering . )
the mean measured @xmath11 for objects with @xmath99 are identical , but the overall distribution of @xmath11 is about 40% broader for ngc 1316 than it is for ngc 1399 . whereas a kolmogorov - smirnov test shows that the distributions of @xmath11 of clusters in ngc 1399 and ngc 1404 can not be distinguished at the 50% confidence level , comparing ngc 1399 and ngc 1316 using the same test indicates that we can reject the null hypothesis at @xmath100 confidence level .
= 4.0 in the distribution of colors , magnitudes , and @xmath11 in ngc 1316 is in some respects similar to that of compact field objects in f0336 , but the total surface density of objects is some 15 times larger . disregarding the possibility that there may be an undiscovered , populous cluster of galaxies situated beyond ngc 1316 which mimics the color and luminosity distribution of field galaxies ,
it is exceedingly unlikely that the observed number of objects in ngc 1316 could be a chance fluctuation in the density of background galaxies .
moreover , over the limited , dust - free region available in the wfpc2 field of view , the radial distribution of the unobscured clusters with @xmath101 follows a powerlaw form with @xmath102 .
this is not significantly different from ngc 1316 s surface brightness profile , which shaya et al .
( 1996 ) found to have an asymptotic slope of @xmath103 .
this is consistent with the radial distributions of many other cluster systems and supports our view that these objects must be physically associated with ngc 1316 .
given the recent merger which has clearly taken place in ngc 1316 , one might have expected significant numbers of metal - rich _ globular _ clusters to have formed as prescribed by the model of ashman & zepf ( 1992 ) . however , neither we nor shaya et al .
( 1996 ) ( who also studied objects in the dust - obscured central regions ) have found evidence for presumed young globular clusters of the type found by holtzman et al .
( 1992 ) in ngc 1275 and whitmore et al .
( 1993 ) in ngc 7252 .
this may be due to the much smaller amount of gas and dust in ngc 1316 compared to these other galaxies .
alternatively , perhaps the active nucleus and twin jets of ngc 1316 are preventing the condensation of gas clouds sufficiently large to produce globular clusters .
inspection of the northeastern quadrant of ngc 1316 in figure [ fig : plate2 ] reveals structures in the dust which are reminiscent of the photoevaporating columns of m 16 ( hester et al . 1996 ) , though on a much larger scale .
the radially - oriented columns in the dust could be a result of erosion by a radiation or particle flux originating in the central regions of the galaxy .
alternatively , we may be seeing rayleigh - taylor instabilities in action .
nicmos observations of the tips of these columns may be useful in determining whether proto - open clusters are indeed condensing out of the material .
given that the only large open cluster populations we know of are in spiral galaxies , would the existence of large numbers of open clusters in ngc 1316 imply that the merger remnant we see today must at one time have been a sizable spiral ?
this would certainly be consistent with the presence of dust , though the relatively small amount of dust compared with the amount seen in ngc 1275 suggests that ngc 1316 s intruder may have been a smaller or an earlier type of spiral .
could it be that many early - type galaxies have exponential cluster luminosity functions ( or at least significant exponential components ) , but that they have only really become detectable with the the advent of the _ hst _ ?
if the objects in ngc 1316 really are old open clusters , where are the globular clusters ?
if ngc 1316 had itself also been a spiral galaxy before the merger , then a globular cluster specific frequency of @xmath104 might not seem so unreasonable .
however , a merger between two spiral galaxies generating virtually no new globular clusters would be at odds with the model predictions of ashman & zepf ( 1992 ) .
could it be that the clusters we see in ngc 1316 started out as all cluster systems do , but that they have somehow avoided the tidal shocking and disruption which may be responsible for the log - normal gclf seen in most other galaxies ( vesperini 1998 ) ? given the anomaly that ngc 1316 s cluster system represents , these questions are well worth pursuing and may provide us with an important key to understanding both star cluster and galaxy formation .
theuns & warren ( 1996 ) , arnaboldi et al . ( 1996 ) , and ferguson et al .
( 1997 ) recently detected significant numbers of stars and planetary nebulae at rather large distances from the cd galaxies ngc 1399 and m 87 .
grillmair et al .
( 1994a ) found that globular clusters around ngc 1399 seem to be kinematically related more to the whole fornax cluster than to the central galaxy itself .
this has prompted us to examine our data for evidence of such intergalactic nomads among the globular clusters in our sample .
the various components of the fornax cluster ( stars , globular clusters , and cluster galaxies ) all seem to fall off with distance from ngc 1399 as @xmath40 ( grillmair et al . 1994a ; kissler - patig et al .
1997a ) . in the wf frames only , we find 517 , 74 , and 48 objects with @xmath105 in fields n1399 , f0338 , and f0336 , respectively .
the radial surface density distributions of globular clusters in cd galaxies generally have a fairly extended core " ( grillmair et al .
1986 ; lauer & kormendy 1986 ; grillmair et al .
1994b ) where the surface density becomes nearly constant .
if we thus ignore our central pointing and assume that in the outer parts the surface density falls off as @xmath106 , then solving @xmath107 for @xmath108 and using only the outer fields at 0.14 and 1.4 degrees , we find @xmath109 arcmin@xmath110 . thus the surface density of unresolved objects ( 10.6 arcmin@xmath110 ) at 14 from ngc 1399 would be entirely consistent with a roughly constant surface density of compact background galaxies .
put another way , under the assumption that the globular cluster surface density profile maintains a power - law form with @xmath108 at all radii , we would predict only @xmath111 globular clusters in our outermost field .
[ as an aside , this powerlaw would predict @xmath112 globular clusters in our ngc 1399 field , or about 60% more than we actually see , which is consistent with the degree of core - flattening in the surface density distribution found by kissler - patig et al .
( 1997a ) ] .
if we adopt the flatter slope with @xmath113 found by forbes et al .
( 1998 ) , we would predict @xmath114 globular clusters in our outermost field . clearly , the data are consistent with no intergalactic globular clusters at 14 , but are insufficient to rule them out . despite the relatively large color / metallicity spread in ngc 1399 s globular cluster system , figure [ fig : cm ]
demonstrates how well defined their distribution is over color and magnitude compared to the background field .
indeed , one could draw a box with @xmath115 and @xmath116 and capture more than 95% of the cluster sample .
if we then place such a box over the color - magnitude distribution of objects in f0336 , we find a total of 7 objects crowding the very faint end of the box .
close visual inspection of these objects reveals that two of the seven objects are probably galaxies , but we can not rule out that the remaining five objects may be globular clusters in the envelope of ngc 1399 .
however , if this is so , then the color - magnitude distribution of these outlying clusters must be very different from those near the center of the galaxy ; the odds of obtaining the observed color - magnitude distribution by sampling the gclf in figure [ fig : lf ] are vanishingly small .
we have analyzed wfpc2 images of three early - type galaxies and two background fields in the fornax cluster . from an investigation of the global distribution over color and magnitude , we conclude the following : * the color distributions of globular clusters in the central regions of ngc 1399 and its nearby neighbor ngc 1404 are bimodal , and statistically very similar to one another . * with respect to the color distributions seen in ngc 1399 and ngc 1404 ,
objects in a field 8.5 from ngc 1399 have a significant blueward bias .
the metallicity gradient found by ostrov et al .
( 1993 ) in ngc 1399 appears to be present even among the faintest clusters . *
the luminosity functions of globular clusters in ngc 1399 and ngc 1404 are essentially identical to one another and very similar to the luminosity function of globulars in m 87 . * the luminosity function for the blue globulars in ngc 1399 peaks @xmath82 magnitudes brighter in @xmath22 than the luminosity function of the red clusters . *
the luminosity function of objects surrounding ngc 1316 is unlike that found for globular clusters in the other galaxies .
based on the bluer color distribution , the exponential shape of the luminosity function , and the relatively large spread in cluster sizes , we tentatively identify these objects with an extensive population of old _ open _ clusters . * the measured peak of ngc 1399 s luminosity function ( converted to @xmath65-band magnitudes ) is 0.01 @xmath69 0.1 magnitudes fainter than that of m 87 , yielding @xmath117 = 82 @xmath69 8 km s@xmath3 mpc@xmath3 , where the uncertainty reflects only the effects of random errors .
* we see no evidence for intergalactic globular clusters , though our field of view is too small to strongly constrain this finding .
cjg is grateful to s. faber and j. kormendy for many interesting and useful discussions .
we also thank m. kissler - patig for a critical reading of the manuscript .
finally , we are grateful to an anonymous referee for a very helpful report . this research was supported in part by stsci grant no .
go-05990.01 - 94a , and by faculty research funds from the university of california , santa cruz .
holtzman , j. a. , faber , s. m. , shaya , e. j. , lauer , t. r. , groth , e. j. , hunter , d. a. , baum , w. a. , ewald , s. p. , hester , j. f. , light , r. m. , lynds , c. r. , oneil , e. j. jr .
, westphal , j. a. 1992 , , 103 , 691 | we examine the luminosity and @xmath0 color distribution of globular clusters for three early - type galaxies in the fornax cluster using imaging data from the wide field / planetary camera 2 on the _ hubble space telescope_. the luminosity functions we derive are in most cases better than 50% complete down to @xmath1 .
we find that the color distributions of globular clusters in the central region of ngc 1399 and its nearby neighbor ngc 1404 are bimodal and statistically indistinguishable .
the metallicity spread , as inferred from the color distributions in these two galaxies , is very similar to that of m 87 .
ngc 1399 s luminosity function is also very similar to that of m 87 , and comparing their respective peak magnitudes indicates that the fornax cluster is at very nearly the same distance from the local group as is the virgo cluster . from this
we derive @xmath2 km s@xmath3 mpc@xmath3 , where the uncertainty reflects only the effects of random errors .
the number of unresolved objects we find at a projected distance of 440 kpc from ngc 1399 is consistent with nothing more than compact background galaxies , though the small field of view of the wfpc2 does not allow us to put strong constraints on the number of intergalactic globular clusters .
the luminosity function of objects detected around ngc 1316 is more nearly exponential than log - normal , and both the color and size distribution of these objects distinguishes them from the clusters surrounding ngc 1399 .
we suggest that these objects are more akin to old open clusters in the galaxy than they are to globular clusters in typical early - type galaxies . |
the harmonic oscillator is the simplest approximation to a physical oscillator and , when perturbation terms are taken into account , the resulting _ anharmonic oscillator _ is governed by the nonlinear differential equation @xmath1 where @xmath2 denotes the derivative with respect to the independent time or space variable @xmath3 , @xmath4 a damping factor , @xmath5 a time - dependent frequency coefficient , @xmath6 the simplest possible anharmonic term , @xmath7 a forcing term . as to the anharmonicity exponent @xmath8
, it can be either real if @xmath9 is real positive , which is the case for lane - emden @xcite gas dynamics equilibria , rational , like @xmath10 in the thomas and fermi @xcite atomic model , or more usually integer : @xmath11 for the ermakov @xcite or pinney @xcite equation , @xmath12 for the duffing oscillator @xcite . for generic values of the coefficients ,
this equation is equivalent to a third order autonomous dynamical system , which generically admits no closed form general solution .
the purpose of this article is to review all the nongeneric situations for which there exist exact analytic results , such as a first integral or a closed form solution , either particular or general .
this can only happen when the coefficients satisfy some constraints .
the paper is organized as follows . in section [ sectionlh ]
, we give a lagrangian and a hamiltonian formulation for any value of the coefficients @xmath13 .
this generalizes all the previous particular results , obtained for values of @xmath14 equal to : @xmath15 @xcite , @xmath16 ( * ? ? ?
* eq . ( 3.7 ) ) , @xmath17 @xcite , @xmath18 @xcite , @xmath19 @xcite , @xmath20 @xcite , ( * ? ? ?
* section 6.74 , vol . 1 ) .
in section [ sectionparticularfirstintegral ] , we provide two conditions on @xmath13 which are sufficient to ensure the existence of a first integral . in section [ sectioninterpretation ] , we give a natural interpretation of these two conditions .
finally , in section [ sectionpa ] , we perform the painlev analysis of ( [ eq1 ] ) . most of this work has already been done by painlev and gambier @xcite . indeed , the ordinary differential equation ( ode ) ( [ eq1 ] ) belongs , at least for specific values of @xmath8 and maybe after a change @xmath21 of the dependent variable @xmath22 in case @xmath8 is not an integer , to the class of second order odes which they studied and classified .
however , as opposed to these classical authors , we do not request the full painlev integrability of the ode , only some partial integrability , and this requires some additional work .
in particular , we compute the condition for the absence of any infinite movable branching , a multivaluedness which occurs at a location depending on the initial conditions . such a condition , like for linear odes ,
arises from any integer value of the difference of the two fuchs indices , whether positive or negative , and we check that this condition is a differential consequence of the two conditions for the existence of a particular first integral .
this detailed painlev analysis of equation ( [ eq1 ] ) happens to be an excellent example for several features of painlev analysis which are most of the time overlooked . for convenience
, we use the notation @xmath23 and the convention that function @xmath24 implicitly contains an arbitrary multiplicative constant ; letter @xmath25 , with or without subscript , denotes an arbitrary constant .
function @xmath24 frequently occurs , for the way to suppress term @xmath26 in ( [ eq1 ] ) is to perform the change of function @xmath27 .
for every value of @xmath13 , including the logarithmic case @xmath28 , the anharmonic oscillator can be put in lagrangian form @xmath29 or in hamiltonian form @xmath30 as shown by the explicit expressions for @xmath31 @xmath32 + \frac{1}{2 } \left ( h u^2 \right ) ' , \label{eq5 } \\ & & h(q , p , x ) = g_1 \left[u'^2 + 2 g_3 \int_0^u u^n \d u + g_2 u^2 + 2 g_4 u \right ] - \frac{1}{2 } h ' u^2 , \label{eq6 } \\ & & q = u,\ p=2 g_1 u ' + h u , \label{eq7}\end{aligned}\ ] ] in which @xmath33 is an arbitrary gauge function of @xmath3 .
according to noether theorem , one can find first integrals by looking at the infinitesimal symmetries of the lagrangian . for a detailed review of this
lie symmetries approach to the anharmonic oscillator , the interested reader can refer to @xcite .
since the dependence of ode ( [ eq1 ] ) in @xmath22 is rather simple , let us determine under which conditions on parameters @xmath13 there exists a particular first integral containing the same kind of terms than the hamiltonian @xmath34 in which the six functions @xmath35 of @xmath3 are to be determined .
eliminating @xmath36 between @xmath37 and @xmath38 , we obtain @xmath39 out of the nine monomials @xmath40 , only eight are linearly independent since @xmath41 , thus generating eight linear homogeneous differential equations in six unknowns , hence generically two conditions on @xmath13 .
note that , even in the logarithmic case @xmath28 , the first generated equation is @xmath42 .
functions @xmath43 to @xmath44 are given by @xmath45 and function @xmath46 must be a nonzero solution common to the three linear equations @xmath47 f_1 + ( n+3 ) g_3 f_1 ' = 0 , \label{eq11a } \\ & & { \hskip -1.0truecm } ( -2 g_1 g_2 + 2 g_1 g_1 ' -g''_1 + g_2 ' ) f_1 + ( g_1 ^ 2 + 2 g_2 - \frac{5}{2 } g_1 ' ) f_1 ' - \frac{3}{2 } g_1 f_1 '' + \frac{1}{2 } f_1 ' '' = 0 , \label{eq11b } \\ & & { \hskip -1.0truecm } ( -2 g_1 g_4 + 2 g'_4 ) f_1 + 3 g_4 f_1 ' = 0 .
\label{eq11c}\end{aligned}\ ] ] each above equation can be integrated once , @xmath48 \d x , \label{eq12b } \\ & & k_3=f_1 ^ 3 g_1^{-2 } g_4 ^ 2
. \label{eq12c}\end{aligned}\ ] ] whatever be @xmath13 , the function @xmath46 can always be computed from ( [ eq11b ] ) ; depending on @xmath49 , it is also given by @xmath50'=0 \label{eq13c}\end{aligned}\ ] ] where the constants @xmath51 and @xmath52 have been absorbed in the definition of @xmath24 .
the only case in which equation ( [ eq13c ] ) needs to be considered is @xmath53 , and its solution can be found in @xcite @xcite @xcite ( * ? ? ?
* eq . e12 ) @xcite .
once @xmath46 is determined , @xmath43 and @xmath54 are given by ( [ eq10a ] ) , ( [ eq10d ] ) , and @xmath55 by the three following expressions , corresponding to cases ( [ eq13a ] ) , ( [ eq13b ] ) , ( [ eq13c ] ) respectively , @xmath56 parameters @xmath13 must satisfy the conditions , polynomial in @xmath57 , resulting from the elimination of @xmath46 between the three linear equations ( [ eq11a])-([eq11c ] ) .
there are two such conditions when @xmath58 is nonzero , and only one when it is zero .
the simplest choice of these two conditions is ( the labelling refers to the contributing @xmath59 s ) : @xmath60 uniquely defined as , respectively , the condition independent of @xmath61 and the one independent of @xmath62 . by elimination ,
one obtains the condition independent of @xmath63 and the one independent of @xmath64 , @xmath65 = 0 , \label{eq15c } \\
g_4 \not= 0 : c_{234 } \equiv & & { \hskip -0.4 truecm }
n^3 g_2 ' + n^2 ( -g_2 \gamma_3'- \gamma_3'''+\frac{3}{2 } \gamma_3 '' \gamma_4 ' + \frac{1}{2 } \gamma_3 ' \gamma_4 '' -2 \gamma_4 ' \gamma_4 '' + \gamma_4 ' '' ) \nonumber \\ & & { \hskip -0.4 truecm } -\frac{3}{2 } \gamma_3'^2 \gamma_4 ' + \frac{1}{2 } \gamma_3'^3 -n^2(n-1 ) g_2 \gamma_4 ' \nonumber \\ & & { \hskip -0.4 truecm } -\frac{1}{2 } ( n^2 - 3 ) \gamma_3 ' \gamma_4'^2 + \frac{1}{2 } ( n^2 - 1 ) \gamma_4'^3 = 0 . \label{eq15d}\end{aligned}\ ] ] for @xmath66 , any two of the above four conditions are functionally independent . for @xmath67 ,
one has @xmath68 and independent conditions are @xmath69 and @xmath70 .
all above conditions admit an integrating factor , a natural consequence of the integrated forms ( [ eq12a])([eq12c ] ) .
this is evident for @xmath69 ; for each of the three others , it is sufficient to integrate it as a first order linear inhomogeneous ode in @xmath61 , @xmath71 g_1^{-8/3 } g_4^{-4/3 } , \label{eq16b } \\
k_{123 } \equiv & & { \hskip -0.5truecm } [ ( n+3)^2 g_2 - ( n+3 ) ( 2 g_1 ' + \gamma''_3 ) -2(n+1)g_1 ^ 2 \nonumber \\ & & { \hskip -0.5truecm } -(n-1 ) g_1 \gamma'_3 + \gamma'^2_3]^{n+3 } g_1^{2(n-1 ) } g_3^{-4 } , \label{eq16c } \\
g_4 \not= 0 : k_{234 } \equiv & & { \hskip -0.5truecm } \left [ g_2 + \frac { ( n+2 ) \gamma_3 ' \gamma_4 ' -\gamma_3'^2
-(n+1 ) \gamma_4'^2 + 2 n ( \gamma_3 '' - \gamma_4'')}{2 n^2 } \right ] \times \nonumber \\ & & { \hskip -0.5truecm } g_3^{-4/3 } g_4^{4/n}. \label{eq16d}\end{aligned}\ ] ] in the duffing case @xmath72 , condition @xmath73 has already been given @xcite , together with its integrated form @xmath74 @xcite .
a very simple interpretation can be given for the two conditions .
indeed , the form of equation ( [ eq1 ] ) is invariant under the simultaneous change of dependent and independent variables @xmath75 where @xmath76 and @xmath77 are two arbitrary gauge functions .
the transformed ode reads @xmath78 u ' + \frac{1}{\xi'^2 } \left[g_2+g_1\frac{\alpha'}{\alpha}+\frac{\alpha''}{\alpha } \right ] u \nonumber \\ & & + \frac{\alpha^{n-1}}{\xi'^2 } g_3 u^n + \frac{1}{\alpha \xi'^2}g_4 = 0 . \label{eq18}\end{aligned}\ ] ] let us adjust the two functions @xmath79 so as to make two of the four new coefficients as simple as possible .
one of the three possible ways is to cancel the damping term by the choice @xmath80 , which reduces ode ( [ eq18 ] ) to @xmath81 u + \alpha^{n+3 } g_1 ^ 2 g_3 u^n + \alpha^3 g_1 ^ 2 g_4 = 0 . \label{eq19}\end{aligned}\ ] ] canceling the new @xmath61 coefficient amounts to solving the general linear second order ode for @xmath76 , which is possible ( from the point of view of painlev , adopted here ) but does not lead to an explicit value of @xmath76 .
this reduced form is then @xmath82 and this means that one can freely set @xmath83 in ( [ eq1 ] ) without altering its global properties ( existence of first integrals , painlev property , etc ) . instead of that , one can make constant either the reduced @xmath84 coefficient iff @xmath85 by choosing @xmath86 , or the reduced @xmath63 coefficient iff @xmath87 by the choice @xmath88 ( let us recall that @xmath24 implicitly contains an arbitrary multiplicative constant ) .
we are thus led to the reduced forms @xmath89 \times \nonumber \\ & & { \hskip -0.4truecm } \phantom{g_2 \mapsto } g_1^{-8/3 } g_4^{-4/3},\ \nonumber \\ & & { \hskip -0.4truecm } g_3 \mapsto g_3 g_1^{-2n/3 } g_4^{-(n+3)/3 } , \nonumber \\ n\not=-3 : & & { \hskip -0.4truecm } g_1 \mapsto 0,\ g_3 \mapsto 1,\ \label{eq20b } \\ & & { \hskip -0.4truecm } g_2 \mapsto \big [ g_2 -\frac{1}{n+3 } ( 2 g_1 ' + \gamma_3 '' ) + \frac{1}{(n+3)^2 } ( -2(n+1)g_1 ^ 2 \nonumber \\ & & { \hskip -0.4truecm } \phantom{g_2 \mapsto } -(n-1 ) g_1 \gamma_3 ' + \gamma_3'^2 ) \big ] g_1^{2(n-1)/(n+3 ) } g_3^{-4/(n+3)},\ \nonumber \\ & & { \hskip -0.4truecm } g_4 \mapsto g_4 g_1^{2n/(n+3 ) } g_3^{-3/(n+3 ) } , \nonumber \\ n=-3 ,
g_4=0 : & & { \hskip -0.4truecm } g_1 \mapsto 0,\ g_3 \mapsto g_3 g_1 ^ 2,\ \nonumber \\ & & { \hskip -0.4truecm } g_2 \mapsto \left[g_2+g_1\frac{\alpha'}{\alpha}+\frac{\alpha''}{\alpha } \right ] \alpha^4 g_1^{2 } \mapsto 0 .
\label{eq20c}\end{aligned}\ ] ] then the interpretation is obvious : any reduced coefficient distinct from @xmath90 or @xmath91 is the of one of the integrated conditions ( [ eq16a])([eq16d ] ) .
conversely , any integrated condition is one of the remaining coefficients when two coefficients have been made constant by a choice of gauge .
for instance , @xmath92 is the reduced @xmath61 coefficient associated to reduced coefficients @xmath84 and @xmath63 equal to unity .
this can also be seen in a more elementary way . in a gauge @xmath93
such that @xmath94 , an expression for the first integral is @xmath95 and , from the relation @xmath96 one deduces that the two other coefficients @xmath61 and @xmath84 or @xmath63 must be constant .
the hamiltonian ( [ eq6 ] ) is a first integral if and only if @xmath97 and all other @xmath59 s are constant .
painlev set up the problem of finding nonlinear differential equations able to define functions , just like the first order elliptic equation @xmath98 defines the elliptic function of weierstrass @xmath99 , a doubly periodic function which includes as particular cases the well known trigonometric and hyperbolic functions . for a tutorial introduction , see the books @xcite .
a by - product of this quest for new functions has been the construction of exhaustive lists of nonlinear differential equations , the general solution of which can be made singlevalued ( in more technical terms , without movable critical singularities , this is the so - called _ painlev property _ ( pp ) ) , which implies that their general solution is known in closed form .
in particular , the list of second order first degree algebraic equations , i.e. @xmath100 with @xmath101 rational in @xmath102 , algebraic in @xmath22 , analytic in @xmath3 , which possess the pp has been established by painlev and gambier @xcite .
these classical results apply to our problem only for those values of @xmath8 for which eq .
( [ eq1 ] ) , maybe after a monomial change of the dependent variable @xmath103 , belongs to the class ( [ eqgambierclass ] ) . these values , which include at least all the integers , are determined below .
then , the way those classical results can be applied is twofold . 1 .
require the pp for our equation or its transform under @xmath104 .
2 . restricting to the values of @xmath13 for which the first integral ( [ eq8 ] ) exists , check that the two conditions for the existence of this first integral imply the identical satisfaction of the necessary condition that eq .
( [ eqgambierclass ] ) have no movable logarithmic branch points .
indeed , this is a classical result of poincar that the movable singularities ( those which depend on the initial conditions ) of first order algebraic odes can only be algebraic , @xmath105 , and never logarithmic , with some @xmath106 term .
let us do that without too many technical considerations .
the above mentioned necessary condition that eq .
( [ eqgambierclass ] ) have no movable logarithmic branch points can only be computed after performing the following steps ( for the unabribged procedure , see @xcite ) .
_ step 1_. for each _ family of movable singularities _
@xmath107 determine the _ leading behaviour _ @xmath108 .
this is achieved by balancing the highest derivative @xmath36 with a nonlinear term
. therefore , there exist two leading behaviours , denoted `` family @xmath84 '' ( balancing of @xmath36 and @xmath109 ) and `` family @xmath63 '' ( balancing of @xmath36 and @xmath63 ) @xmath110^{\displaystyle{\frac{1}{n-1}}},\ n\not=-1,\ \\
( g_4 ) \ : \ & & p=2,\ u_0=-\frac{1}{2 } g_4,\ g_4 \not=0.\end{aligned}\ ] ] _ step 2_. for each family , compute the fuchs indices , the roots @xmath111 of the indicial equation of the linear equation obtained by linearizing ( [ eq1 ] ) near its leading behaviour @xmath112 , and require every fuchs index to be integer .
this linearized equation is @xmath113 and the fuchs indices are obtained by requiring the solution @xmath114 , @xmath115 the diophantine condition that @xmath116 be integer has a countable number of solutions since we have not yet put restrictions on @xmath8 . _
step 3_. for each family , compute all the necessary conditions for the absence of movable logarithms ( in short , no - log conditions ) , which might occur when one computes the successive coefficients @xmath117 of ( [ eqlaurentu ] ) .
one can check that the family @xmath63 can never generate such no - log conditions .
these conditions need not be computed on the original equation ( [ eq1 ] ) , they can be computed on any algebraic transform if this proves more convenient ( indeed , movable logarithms are not affected by an algebraic transform on @xmath22 ) , such as @xmath118 the transformed powers @xmath119 are @xmath120 , and the fuchs indices are unchanged .
the computation of the no - log conditions is impossible unless there exists a @xmath121 making all the powers of @xmath122 in ( [ eq125 ] ) at least rational . in order to avoid the technical complications of dealing with rational values of the leading exponent @xmath119
, we restrict to those values of @xmath8 for which there exists a @xmath121 making @xmath123 and , if @xmath63 is nonzero , @xmath124 integer .
the useful transforms are @xmath125 which are polynomial if and only if @xmath126 the original ode ( [ eq1 ] ) is identical to ( [ eq126a ] ) for @xmath72 and to ( [ eq126b ] ) for @xmath127 . to summarize , let us compute the no - log condition @xmath128 on the ode for @xmath129 ( [ eq126a ] ) .
unfortunately , one does not know how to obtain the dependence of @xmath130 on @xmath8 , since @xmath8 must first be given a numerical value before @xmath130 is computed ; this makes uneasy the comparison with conditions ( [ eq15a])([eq15b ] ) , which depend on @xmath8 .
to fix the ideas , a list of useful values of @xmath131 is displayed in table [ table1 ] .
= 1.5truemm = 0.5truemm = 0.8truemm .values of @xmath132 for @xmath111 integer @xmath133 $ ] . [ cols="^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^",options="header " , ] [ table1 ]
the computation of @xmath130 for positive values of @xmath111 is classical @xcite . denoting for shortness @xmath134 ,
one finds the following expressions @xmath130 for the indicated values of @xmath49 , @xmath135^{-1/2 } [ ( \gamma_3 ' - g_1 ) c_1 - c_1 ' ] + \frac{1}{6 } c_2 , \label{eq26.4 } \\ ( \frac{7}{3},0 ) : q_5= & & { \hskip -0.4 truecm } ( .
g_1 ^ 2 + .
g_1 \gamma'_3 + .
\gamma_3'^2 ) c_1 + ( .
\gamma_3 ' ) c'_1 + .
c''_1 , \label{eq26.5 } \\ ( 2 , g_4 ) : q_6= & & { \hskip -0.4 truecm } .
c_1 ^ 2 + .
c'_2 , \label{eq26.6 } \\
( \frac{9}{5},0 ) : q_7= & & { \hskip -0.4 truecm } .
c_1 + \dots + .
c_1^{(4 ) } , \label{eq26.7 } \\ ( \frac{5}{3},0 ) : q_8= & & { \hskip -0.4 truecm } .
c_1 + \dots + .
c_1^{(5)}. \label{eq26.8}\end{aligned}\ ] ] where dots stand for rational numbers when @xmath136 and polynomials of @xmath64 , @xmath61 , @xmath137 , @xmath138 when @xmath139 .
similar relations have been checked for @xmath140 and @xmath141 ( thomas - fermi case ) but are not reproduced here .
condition @xmath142 contains a @xmath143 sign arising from the two possible choices for @xmath144 and is equivalent to the two conditions @xmath145 we therefore check the property that each @xmath130 is indeed a differential consequence of the two conditions @xmath146 for the existence of a first integral ( [ eq8 ] ) @xmath147 for negative @xcite values of the fuchs index @xmath111 , the results @xcite are the following : the family @xmath63 never generates any no - log condition , and , for the family @xmath84 , a no - log condition arises from the fuchs index @xmath148 , and this condition is a differential consequence of conditions ( [ eq15a])([eq15b ] ) , at least for the examples handled @xmath149 .
this is also an experimental verification of @xmath150 and this relation can not be reversed , as proven by painlev and gambier .
for instance , in the case of the duffing oscillator @xmath151 , condition @xmath142 implies the reducibility of @xmath129 to the second painlev transcendent whereas the stronger conditions @xmath152 imply the reducibility of @xmath129 to an elliptic function . _
remark_. when one includes the contribution of the schwarzian in the definition of the gradient of the expansion variable @xmath153 , as done in the invariant painlev analysis @xcite , @xmath154 all the computed no - log conditions @xmath128 , equations ( [ eq26.1])([eq26.8 ] ) , are independent of this schwarzian @xmath155 , as opposed e.g. to the lorenz model @xcite .
this certainly indicates some hierarchy between the level of nonintegrability of these two dynamical systems . _
remark_. for some small values of @xmath156 , there is equivalence between the no - log condition and ( [ eq15a])([eq15b ] ) .
this nongeneric situation occurs only for the following values of @xmath157 , @xmath158 , i.e. the ermakov - pinney equation @xcite , @xmath159 , i.e. an equation considered by lane and emden @xcite , chandrasekhar @xcite and logan @xcite , @xmath160 , an equation which could deserve more study .
this work generalizes all previous results on the partial integrability of the anharmonic oscillator .
it gives a natural interpretation of the two conditions for the existence of a particular first integral , in terms of reduced coefficients .
finally , this system is an excellent example to study several features of painlev analysis . a good
, recent bibliography can be found in ref . @xcite .
the author wishes to thank m. musette for fruitful discussions during the completion of this work and for her encouragement to publish these lecture notes , a first draft of which was delivered at a meeting in dijon @xcite .
r. conte , unification of pde and ode versions of painlev analysis into a single invariant version , _ painlev transcendents , their asymptotics and physical applications _
, 125144 , eds .
d. levi and p. winternitz ( plenum , new york , 1992 ) .
r. conte , the painlev approach to nonlinear ordinary differential equations , _ the painlev property , one century later _
, 77180 , ed .
r. conte , crm series in mathematical physics ( springer , new york , 1999 ) .
solv - int/9710020 .
r. conte and m. musette , a simple method to obtain first integrals of dynamical systems , _ solitons and chaos _ ( research reports in physics
nonlinear dynamics ) 125128 , eds .
i. a. antoniou and f. j. lambert ( springer , berlin , 1991 ) .
e. l. ince , _ ordinary differential equations _ ( longmans , green , and co. , london and new york , 1926 ) . reprinted ( dover , new york , 1956 ) .
see errata in cosgrove 1993 , preprint .
russian translation ( gtiu , kharkov , 1939 ) .
e. kamke , _ differentialgleichungen : lsungsmethoden und lsungen _ ,
vol . 1 , 243 pages ; vol . 2 , 668 pages .
akademische verlagsgesellschaft , geest & portig k .-
g . , leipzig 1947 . reprinted ( chelsea , new york , 1948 ) . | we consider the anharmonic oscillator with an arbitrary - degree anharmonicity , a damping term and a forcing term , all coefficients being time - dependent : @xmath0 its physical applications range from the atomic thomas - fermi model to emden gas dynamics equilibria , the duffing oscillator and numerous dynamical systems .
the present work is an overview which includes and generalizes all previously known results of partial integrability of this oscillator .
we give the most general two conditions on the coefficients under which a first integral of a particular type exists .
a natural interpretation is given for the two conditions .
we compare these two conditions with those provided by the painlev analysis .
[ firstpage ] |
to explore the formation and the evolution of the present - day `` normal '' galaxies like our galaxy , it is important to characterize where and how star formation occurs in a galaxy . in the local universe , it is well known that star formation rate ( sfr ) correlates with dust opacity ( @xcite ; @xcite ) , where intensely star - forming regions and galaxies are obscured by a huge amount of dust , and become optically thick ( @xcite ; @xcite ) .
therefore , the uv continuum emission and hydrogen recombination lines in the optical ranges such as h@xmath3 at 0.6563 @xmath2 m and h@xmath8 at 0.4861 @xmath2 m emitted from those dusty star - forming regions are easily attenuated , and then it would be complicated to correctly understand the distribution of star - forming regions in the whole galaxy , especially when the galaxy has a patchy distribution of dust ( @xcite ) . on the other hand , paschen-@xmath3 ( pa@xmath3 ) at 1.8571 @xmath2 m ( in air ; 1.8756 @xmath2 m in vacuum ) , the strongest emission line in the near - infrared ( nir , @xmath9 @xmath6 0.82.5 @xmath2 m ) wavelength range , is less affected by dust thanks to its longer wavelength than the optical lines ( @xcite ) , and becomes a powerful and direct indicator of sfr , especially in dusty regions .
in particular , the observed pa@xmath3 emission becomes stronger than h@xmath3 at @xmath10 1.2 ( @xmath11 3.7 ) and than br@xmath12 at @xmath13 28.0 ( @xmath14 86.2 ) with an assumption of a milky - way - like extinction curve and @xmath15 3.08 ( @xcite ) , while the intrinsic pa@xmath3 luminosity is 0.12 of h@xmath3 , 2.14 of pa@xmath8 ( 1.2818 @xmath2 m in air ) , and 12.4 of br@xmath12 ( 2.1655 @xmath2 m in air ) for case b recombination with an electron temperature of @xmath16 k and density of @xmath16 @xmath17 ( @xcite ) .
( 80mm,80mm)figure01.eps however , pa@xmath3 observations from the ground have been quite difficult so far due to deep atmospheric absorption features around their wavelength , mostly caused by water vapor , and hence most pa@xmath3 observational studies have been limited to those by near infrared camera and multi - object spectrometer ( nicmos , @xcite ) on board the _ hubble _ space telescope ( _ hst _ ) . for example , @xcite use the nicmos pa@xmath3 narrow - band ( f187n and f190n ) imaging data to study star formation properties of dusty star - forming galaxies in the local universe with a high spatial resolution , and find the compact distributions of pa@xmath3 emission along various galaxy structures such as nucleus and spiral arm .
they also establish a linear empirical relation between pa@xmath3 and 24 @xmath2 m luminosity as well as the total infrared luminosity as an indicator of sfr for those galaxies .
in addition , a wide - field pa@xmath3 imaging survey of the galactic center ( @xcite ; @xcite ) produces a high spatial - resolution map of stars and an ionized diffuse gas with a possible new class of massive stars not associated with any known star clusters .
recently , a nir camera and spectrograph , flitecam ( @xcite ) , for nasa s stratospheric observatory for infrared astronomy ( sofia ) has been developed , and will support pa@xmath3 imaging observations with two narrow - band filters centered at 1.875 and 1.900 @xmath2 m .
as described above , pa@xmath3 observations are quite useful for a variety of targets obscured by dust and can provide a new insight into those studies , yet the amount of observing time and facilities available for that purpose are still very limited .
therefore , we have developed a new instrument at an appropriate astronomical site for _ ground - based _ pa@xmath3 observations .
ll ' '' '' wavelength & 0.952.4 @xmath2 m + ' '' '' detector & pace hawaii-2 + ' '' '' pixel format & 1024 @xmath0 1024 + ' '' '' pixel pitch & 18.5 @xmath2 m + ' '' '' field of view & @xmath0 + ' '' '' pixel scale & pixel@xmath1 + ' '' '' broad - band filters & @xmath18 , @xmath19 , @xmath20 , @xmath21 + ' '' '' narrow - band filters & @xmath5128 , @xmath51875 , @xmath5191 , @xmath22 + ( 170mm,170mm)figure02.eps the university of tokyo atacama observatory ( tao , @xcite ) 1.0-m telescope , called minitao , is an optical / infrared telescope installed in 2009 , located at the summit of cerro chajnantor ( an altitude of 5,640 m or 18,500 ft above sea level ) in northern chile as the world s highest astronomical observatory ( @xcite ; @xcite ) .
the site was expected to be very dry from satellite data reported by @xcite ( precipitable water vapor , pwv , of 0.5 mm at 25th percentile ) and from radiosonde measurements by @xcite ( 0.5 mm at a median above an altitude of 5,750 m ) .
figure [ fig : atran_nir ] shows a comparison of simulated atmospheric transmittances between the tao site and the other sites at lower altitudes , in which there are remarkable improvements of the transmittance at gaps between conventional broad atmospheric windows , especially around 1.9 @xmath2 m , i.e. , pa@xmath3 wavelength .
the atacama nir camera ( anir ) is one of the instruments for the cassegrain focus of the minitao telescope .
it is capable of wide - field pa@xmath3 narrow - band ( nb ) imaging observations as well as ordinary broad - band imaging one .
it also has a capability of optical imaging and slitless spectroscopic observations simultaneously with the nir imaging using a retractable dichroic mirror .
we have achieved first - light observations using nb filters targeted on the pa@xmath3 emission line ( figure [ fig : atran_paa ] ) in 2009 , and now anir is in operation mainly for pa@xmath3 observations of the galactic center / plane and nearby starburst galaxies in order to understand the star formation activities hidden in thick dust clouds ( e.g. , @xcite ; @xcite ) .
( 170mm,170mm)figure03.eps in this paper , we describe the overall design of anir in section [ sect : instrument ] , its performance evaluated through laboratory tests and actual observations at the tao site in section [ sect : performance ] , and a site evaluation through the pa@xmath3 imaging data in section [ sect : pwv ] .
we refer the reader to @xcite and @xcite for a detailed description of data reduction processes and quantitative analysis of pa@xmath3 data .
here we describe the design of the nir unit providing broad - band and nb imaging functions in the nir .
a hardware design of the optical unit is described in section [ sect : instrument_opt ] .
figure [ fig : anir_2d ] shows a cross - sectional view of anir , and a block diagram of the hardware is shown in figure [ fig : hardware_diagram ] .
the specifications of the nir unit are summarized in table [ tab : spec_nir ] .
the unit covers a field of view ( fov ) of @xmath0 with a spatial resolution of pixel@xmath1 using an engineering grade producible alternative to cdte for epitaxy ( pace ) hawaii-2 array detector , which is a 2048 @xmath0 2048 pixel hgcdte nir focal plane array ( fpa ) manufactured by teledyne scientific & imaging llc .
note that only a single quadrant with 1024 @xmath0 1024 pixels is used .
the filter set consists of four standard broad - band filters ( @xmath18 , @xmath19 , @xmath20 , and @xmath21 ) and four nb filters ( @xmath5128 , @xmath51875 , @xmath5191 , and @xmath5207 ) as listed in table [ tab : spec_nir_nb ] .
the @xmath51875 filter has a central wavelength of @xmath23 1.8759 @xmath2 m and a bandwidth of @xmath24 0.0079 @xmath2 m , which covers the pa@xmath3 ( @xmath41.8751 @xmath2 m ) with radial velocities within @xmath6 @xmath25580 to @xmath26680 km s@xmath1 . on the other hand
, the @xmath5191 filter ( @xmath27 = 1.911 @xmath2 m , @xmath24 0.033 @xmath2 m ) corresponds to redshifted pa@xmath3 lines with recession velocities of c@xmath28 @xmath6 29008200 km s@xmath1 .
the @xmath5191 filter is also used for taking off - band ( continuum ) data for the @xmath51875 data , and vice versa .
figure [ fig : atran_paa ] shows details of the atmospheric transmittances shown in the top panel of figure [ fig : atran_nir ] at the wavelength range of interest for the pa@xmath3 nb filters . while there are absorption features mostly by water vapor even at the altitude of 5,640 m , several windows with a high transmittance exist .
the average atmospheric transmittances at the pwv of 0.5 mm are approximately 0.48 and 0.64 within the bandpasses of the @xmath51875 and @xmath5191 filters , respectively .
lccccl ' '' '' name & & targeted line + ' '' '' & & & + ' '' '' & center ( @xmath27 ) & width ( @xmath29 ) & center & width + ' '' '' @xmath5128 & 1.2818 & 0.0210 & 1.2814 & 0.0217 & pa@xmath8 ( 1.2818 @xmath2 m ) + ' '' '' @xmath51875 & 1.8751 & 0.0080 & 1.8759 & 0.0079 & pa@xmath3 ( 1.8751 @xmath2 m ) + ' '' '' @xmath5191 & 1.9010 & 0.0306 & 1.9105 & 0.0329 & pa@xmath3 off - band + ' '' '' @xmath5207 & 2.0750 & 0.0400 & 2.0742 & 0.0391 & c ( 2.078 @xmath2 m ) + ( 85mm,85mm)figure04a.eps ( 85mm,85mm)figure04b.eps the cryostat is a compact cube with approximately 260 mm on a side , in which offner relay optics ( section [ sect : optics ] ) , a filter box with two wheels , and the hawaii-2 fpa are housed , as shown in figure [ fig : anir_2d ] .
all the components in the cryostat are cooled down to 70 k to reduce thermal radiation , especially for observations in the @xmath21-band . a single - stage closed - cycle
mechanical cooler with a cooling capacity of 25 w at 77 k is equipped .
a cold head anti - vibration mount and bellows are inserted between the cold head and the cryostat to reduce vibration caused by the cold head . in addition , a teflon plate and a sapphire plate are inserted to electrically insulate the cryostat from the cold head .
it takes approximately 24 hours to cool down and stabilize the cryostat ( @xmath30 k at the detector box ) from the ambient temperature down to 70 k. the two filter wheels are installed just after the focal plane of the telescope .
each filter wheel has five slots with @xmath31 34 mm . a stepping motor controls its rotation , and neodymium magnets and a hall - effect sensor sense the position of the slots . ( 85mm,85mm)figure05a.eps ( 85mm,85mm)figure05b.eps reflective offner relay optics are employed for re - imaging , consisting of two ( concave primary and convex secondary ) spherical mirrors .
the primary mirror forms an image of the telescope pupil on the secondary mirror , which works as a cold lyot stop .
the specifications of the mirrors are summarized in table [ tab : offnerspec ] .
these mirrors are gold - coated to achieve high reflectivity .
figure [ fig : nir_sd ] shows spot diagrams at @xmath9 = 1.65 @xmath2 m across the fov .
sharp image quality is achieved at any positions within the fov . as shown in the right - hand panel of figure [ fig : nir_sd ] ,
even with a dichroic mirror inserted for simultaneous optical - nir observations ( section [ sect : instrument_opt ] ) , spot sizes are still smaller than or comparable to the size of the airy disk .
figure [ fig : nir_ece_wdm ] shows encircled energy distributions , the fraction of energy in a given radius to the total energy from a point source , for the center and off - center spots with the dichroic mirror inserted , compared to that of a diffraction - limited spot .
we confirm that the diameter encircling the 80% energy is less than 1.7 pixels ( @xmath6 31.5 @xmath2 m ) or across the fov .
( 80mm,80mm)figure06.eps llr ' '' '' primary mirror & radius of curvature & 140 mm + ' '' '' & effective diameter & 90 mm + ' '' '' secondary mirror & radius of curvature & 70 mm + ' '' '' & effective diameter & 9 mm + ' '' '' offset of optical axis & & 24 mm + ( 170mm,170mm)figure07.eps anir is operated by two linux pcs ( named _ uni _ and _ uni2 _ ) ; one is dedicated to control hawaii-2 fpa ( tao array controller , tac ) , and the other handles all the other tasks , including operation of the filter wheels and the optical unit ( see section [ sect : instrument_opt ] ) , acquisition of house keeping information such as temperature and vacuum pressure , and management of various instrument status .
the tac system ( see @xcite for more details ) is a high - speed and flexible array controller using a real - time operating system instead of conventional dedicated processor devices such as digital signal processors ( dsps ) .
one core of a dual - core cpu in _ uni _ is assigned to the real - time data processing , and engages in clock pattern generation and processing of acquired frame data .
this allows us to control the fpa in real time without affecting the performance of the operating system and other software .
a rate to read a pixel ( pixel rate ) can be set to 3 to 8 @xmath2s . considering increasing readout noise with faster readout ( slower pixel rate ) , the rate of 4 @xmath2s is usually used , corresponding to a readout time of 1024 @xmath0 1024 pixels of 4.2 s , which is the shortest exposure time of the nir unit .
figure [ fig : software_diagram ] summarizes the functions and gives a command / data flow diagram of the control system . to efficiently handle status data increasing with time
, we adopt a relational database management system , mysql .
an successful example of the application of mysql in astronomical instrumentation is described in @xcite .
several database tables are prepared in mysql , and information concerning instrument status ( filters used , insertion of the dichroic mirror , temperatures / pressures ) is stored into the respective tables every minute .
when the exposure command is executed , a fits ( flexible image transport system ) header is constructed by collecting the latest information from those tables . to benefit from the advantages of the site , such as good seeing condition with at @xmath32-band at a median ( @xcite
; see also @xcite ) , anir is capable of optical and nir simultaneous imaging observations by inserting a dichroic mirror in front of the entrance window of the cryostat of the nir unit ( the upper part of figure [ fig : anir_2d ] ) .
the specifications of the optical unit are summarized in table [ tab : spec_opt ] .
the unit covers a fov of @xmath0 with a spatial resolution of pixel@xmath1 using a commercial , peltier - cooling ccd camera unit incorporating a e2v backside - illuminated ccd , proline pl4710 - 1-mb , fabricated by finger lakes instrumentation .
four johnson - system broad - band filters ( @xmath33 , @xmath32 , @xmath34 , and @xmath35 ) for imaging and a low - resolution transmission grating ( grism ) with 75 lines mm@xmath1 and a blaze angle of 4.3@xmath36 for slitless spectroscopy
are available .
the dichroic mirror has a dimension of 50 @xmath0 71 @xmath0 10 mm with a wedge angle of 0.78@xmath37 to minimize astigmatism in the nir image .
its reflectance at 0.40.86 @xmath2 m is higher than 0.9 with an incident angle of 45@xmath36 , while its transmittance at 0.952.4 @xmath2 m is kept higher than 0.9 .
figure [ fig : opt_sd ] shows spot diagrams of the optical unit at the final focal plane . due to chromatic aberration of the re - imaging refractive optics , the focal position is offset by @xmath6 0.2 mm in the @xmath35-band , which can be effectively compensated as shown in the right - hand panel of figure [ fig : opt_sd ] by shifting the camera optics mounted on a linear stage .
the dichroic mirror is mounted on a linear stage , and can be retracted to derive better - quality ( higher efficiency , lower thermal background ) nir data when no simultaneous optical observation is required .
in this section , we describe the imaging performance of the nir and optical units obtained through observations carried out in 20092011 .
( 85mm,85mm)figure08a.eps ( 85mm,85mm)figure08b.eps we measure the conversion factor of the hawaii-2 fpa readout system using flat frames with increasing exposure time in the laboratory , and obtain 3.2 @xmath38 adu@xmath1 from the calculated mean and dispersion of pixel values .
the linearity of the fpa is evaluated from other flat images taken by increasing exposure time .
we ensure that the linearity is kept within @xmath6 1% up to 18,000 adu or 57,600 @xmath38 .
the readout noise is measured on the telescope in dark frames with the shortest exposure ( i.e. , 4.2 s ) which are images with the @xmath19-band and @xmath51875 filter inserted , where no photons fall on the fpa because those bandpasses do not overlap each other .
we obtain 16.6 @xmath38 as the rms of their median values with the correlated double sampling ( cds ) readout method .
a multiple readout ( `` up - the - ramp '' sampling ) method is also examined , and we obtain 13.9 @xmath38 with 16 readouts . the dark current derived from dark frames with 300500 s exposures is 0.28 @xmath38 s@xmath1 pixel@xmath1 . we should mention that the dark current of hawaii and hawaii-2 fpas have _ persistence _ and that the level of the _ persistence _ tends to be dependent on incident flux ( @xcite ; @xcite ; @xcite ) .
for example , the _ persistence _ of a dark current pattern would appear strongly on frames with a nb filter taken after exposures of high background , such as ones with a broad - band filter .
the tac system continues to reset the fpa with an interval of @xmath6 0.1 s in order to reduce the _ persistence _ effect even if no exposure command is issued .
ll ' '' '' wavelength & 40008600 + ' '' '' ccd unit & fli proline pl4710 - 1-mb + ' '' '' & ( e2v backside - illuminated ccd ) + ' '' '' pixel format & 1056 @xmath0 1027 + ' '' '' pixel pitch & 13.5 @xmath2 m + ' '' '' field of view & @xmath0 + ' '' '' pixel scale & pixel@xmath1 + ' '' '' filters & @xmath33 , @xmath32 , @xmath34 , @xmath35 ( johnson system ) + ' '' '' grism & 75 lines mm@xmath1 , blaze angle @xmath39 4.3@xmath36 + the performance of the optical ccd are also evaluated in a similar manner to that described above .
we take flat images with increasing exposure time in the laboratory , and obtain the conversion factor of 1.818 @xmath38 adu@xmath1 from the calculated mean and dispersion of pixel values .
the readout noise is measured from bias frames to be 14.5 @xmath38 rms .
we obtain the dark current of 0.011 @xmath38 s@xmath1 pixel@xmath1 at the ccd temperature of @xmath6 220 k. we evaluate the image quality of the nir unit using bright ( but unsaturated ) and isolated field stars in images taken under good seeing condition . by matching positions on the image ( @xmath40 , @xmath41 ) of about 20 point sources located over the fov with coordinates ( _ _ @xmath3__@xmath42 , _ _
@xmath43__@xmath42 ) obtained from the two micron all sky survey ( 2mass ) point source catalog ( psc , @xcite ) , we obtain a pixel scale of pixel@xmath1 with a fitting uncertainty of @xmath6 pixel@xmath1 .
image distortion is confirmed to be negligible ( @xmath44 1 pixel ) over the fov , as designed .
we derive a typical fwhm size of 2.41 @xmath7 0.24 pixels or @xmath7 in the @xmath21-band where ellipticities of the stars are small and uniform ( 0.08 @xmath7 0.04 ) .
the image quality of the optical unit is evaluated in a similar manner .
raw images have non - negligible image distortion , especially at their corners .
we evaluate and correct for the distortion in the similar manner for the nir unit as described above , by matching the positions of stars with those in the _ hst _ guide star catalog ( gsc , @xcite ) version 2.3 and fitting them to the latter using fifth - order polynomials to take into account the image distortion .
the mapping uncertainties are @xmath6 0.3 pixels .
we obtain a pixel scale of pixel@xmath1 at the center of the fov for all the broad - band filters ( @xmath33 , @xmath32 , @xmath34 , and @xmath35 ) , and a typical fwhm size of 3.04 @xmath7 0.22 pixels or @xmath7 with a homogeneous ellipticity distribution ( 0.04 @xmath7 0.08 ) in the @xmath32-band .
( 85mm,85mm)figure09a.eps ( 85mm,85mm)figure09b.eps figure [ fig : seeing ] shows the seeing distribution for images taken in the latter half of the observation in 2010 and the former half in 2011 . the seeing size ( fwhm )
is measured using the iraf ` psfmeasure ` task .
the median seeing size of @xmath6 or less is obtained for all the bands .
when we consider the diffraction - limited size and hartmann constant ( ) of the telescope optics ( @xcite ) and suppose that some data may be undersampled ( fwhm @xmath45 2 pixel ) due to good seeing condition , the actual seeing distributions at the tao site are expected to have a peak at a smaller size .
lccccr ' '' '' & throughput & & sky brightness & @xmath46 + ' '' '' & [ % ] & @ 60 s & @ 600 s & [ mag arcsec@xmath47 & + ' '' '' @xmath33 & 17.5 & 21.3 & 23.5 & 22.0 & 2600 + ' '' '' @xmath32 & 29.2 & 21.4 & 23.2 & 20.8 & 800 + ' '' '' @xmath34 & 29.4 & 21.5 & 23.2 & 20.3 & 450 + ' '' '' @xmath35 & 20.1 & 20.6 & 22.0 & 18.1 & 70 + ' '' '' @xmath18 & 14.3 & 18.8 & 20.1 & 16.8 & 130 + ' '' '' @xmath19 & 18.9 & 19.5 & 20.9 & 16.4 & 50 + ' '' '' @xmath20 & 29.1 & 18.4 & 19.7 & 14.6 & 10 + ' '' '' @xmath21 & 30.3 & 18.7 & 20.0 & 15.3 & 20 + ' '' '' @xmath5128 & 17.0 & 16.9 & 18.3 & 14.7 & 80 + ' '' '' @xmath51875 & 18.0 & 15.6 & 16.9 & 13.5 & 100 + ' '' '' @xmath5191 & 17.7 & 16.9 & 18.3 & 15.0 & 120 + ' '' '' @xmath5207 & 27.7 & 17.6 & 18.9 & 15.5 & 100 + throughput is defined as the ratio of the number of photons coming from an object into the earth atmosphere of an aperture of the telescope to that of electrons detected , where atmospheric extinction is included . the former is calculated with the effective collecting area of the telescope ( @xmath6 0.78 m@xmath48 ) and the bandwidth of a filter used .
the latter is simply obtained by multiplying a count rate of an object ( adu s@xmath1 ) by the detector conversion factor .
we use optical standard stars ( @xcite ) for the optical unit and 2mass psc stars within 1117 mag and with a median photometric error of @xmath6 0.1 mag for the nir unit for the measurements .
the results are shown in figure [ fig : efficiency ] and summarized in table [ tab : performance ] .
since the 2mass catalog has only the @xmath19 , @xmath20 , and @xmath21-band data , we calculate the magnitudes of the psc stars for the @xmath18-band and nb filters as follows . for the @xmath18-band , we linearly ( in logarithmic scale ) extrapolate the @xmath19-band ( 1.275 @xmath2 m ) magnitude to the @xmath18-band ( 1.038 @xmath2 m ) with a slope between the @xmath19 and @xmath20-bands ( 1.673 @xmath2 m ) for each star . for the @xmath5128 filter
, we use the @xmath19-band magnitude .
the lower throughput of the @xmath5128 filter compared to the @xmath19-band is attributed to the combination use of the @xmath19-band filter as a thermal blocker for observations with the @xmath5128 filter .
we calculate magnitudes at the @xmath51875 , @xmath5191 and @xmath5207 filters by linearly ( in logarithmic scale ) interpolating their @xmath20 and @xmath21-band ( 2.149 @xmath2 m ) magnitudes .
we examine the influence of stellar photospheric features within the wavelength ranges of interest on the interpolation . at first
, we use @xmath49 and @xmath50 color - color diagram to explore which kind of stars dominate the 2mass psc , and derive median colors of @xmath51 and @xmath52 , which correspond to intrinsic colors of k- to m - type stars ( @xcite ; @xcite ) or would be those of g- to k - type stars with dust extinctions of @xmath53 12 mag for data in low galactic latitudes such as galactic plane and large / small magellanic clouds ( @xcite ; @xcite ) .
then , we evaluate the influence of spectral features of those medium - mass stars by using a stellar atmosphere model of @xcite containing o- to m - type stars . we find no significant differences ( @xmath45 0.01 mag ) between magnitudes derived by interpolation and those derived by direct integration of the model spectrum convolved with the transmission curve of the nb filter , although those stars have hydrogen absorption lines , of course including pa@xmath3 .
therefore , the magnitudes at those nb filters we use hereafter are thought to be reliable enough for the performance verification of those nb filters and the site evaluation ( section [ sect : pwv ] ) .
the resulting throughputs of the @xmath51875 and @xmath5191 filters are significantly lower than those of the @xmath20 and @xmath21-bands , and have larger dispersion by a factor of two , which are thought to be due to the lower atmospheric transmittance combined with the airmasses of the observations and temporal fluctuations in the atmospheric absorption mainly from water vapor ( see figure [ fig : atran_paa ] for the transmittance with a pwv of 0.5 mm at an airmass of 1.0 ) .
we evaluate the pwv content above the tao site by using the throughputs of those filters in section [ sect : pwv ] .
the throughput of the @xmath5207 filter is consistent with that of the @xmath21-band when taking into account the difference in their average filter transmittances : @xmath6 78% ( @xmath5207 ) and @xmath6 92% ( @xmath21 ) .
it should be noted that , although there is several night - sky oh emission lines within the bandpasses of the nb filters ( e.g. , @xcite ) , we consider their influence on our result to be small because we perform aperture photometry with _
local _ sky subtraction for each source .
( 80mm,80mm)figure10.eps the limiting magnitudes for the optical and nir units are estimated using the dark current , readout noise , throughputs obtained above , and sky brightness ( _ photons _
@xmath54 s@xmath1 @xmath2m@xmath1 arcsec@xmath55 ) . the sky brightness is measured on a `` sky '' frame made by combining unregistered subsequent frames with objects masked .
table [ tab : performance ] gives the estimated limiting magnitudes for a point source with s@xmath56n @xmath39 5 , @xmath31 aperture , and two different exposure times ( 60 and 600 s ) .
also listed is an approximate exposure time ( @xmath46 ) required to achieve background - limited performance ( blip ) that means the s@xmath56n of data is dominated by the background noise .
we define blip as when the poisson noise of the sky background becomes twice as large as the readout noise .
( 170mm,170mm)figure11.eps here we briefly demonstrate the performance of the minitao / anir _ ground - based _ pa@xmath3 observations . the reduced pa@xmath3 emission image of the nearby starburst galaxy , ic5179 taken with the @xmath5191 filter , is shown in figure [ fig : lirg_paa ] , demonstrating that the tao site indeed provides access to the pa@xmath3 emission line from the ground .
star - forming regions associated with the spiral arm and the nucleus are clearly seen . for a comparison of its sensitivity , the same galaxy taken by the _
f190n filter is also shown in the right - hand panel of figure [ fig : lirg_paa ] , from which the f187n image is subtracted as a continuum emission .
a quantitative comparison confirms the consistency of pa@xmath3 fluxes obtained with anir and nicmos within an accuracy of 10% ( @xcite ) .
we refer the reader to @xcite and @xcite for the data reduction processes , verification , quantitative analyses , and comparison of the performance with nicmos , of pa@xmath3 emission line data taken by the @xmath51875 and @xmath5191 filters .
finally , we evaluate how the site is suitable for the infrared astronomy in terms of pwv by using the @xmath51875 and @xmath5191 data taken between 2009 and 2011 .
we note that our approach using the nb filters around 1.9 @xmath2 m which are sensitive to pwv is independent of any previous studies carried out at ( the vicinity of ) the chajnantor plateau ( e.g. , @xcite ; @xcite ; @xcite ; @xcite ) .
the throughputs derived in section [ sect : throughput ] ( hereafter @xmath57 ) consist of the following factors : ( i ) the reflectivity of the telescope mirrors ( @xmath58 ) , ( ii ) the system efficiency of anir excluding the filter transmittance ( @xmath59 ) , ( iii ) the filter transmittance ( @xmath60 ) , ( iv ) atmospheric transmittance ( @xmath61 ) at the zenith which is dependent on pwv .
then , the throughput is described as : @xmath62 where the superscript `` band '' denotes that the factor depends on the wavelength measured and @xmath63 is the airmass which is necessary for considering the optical path length of the atmosphere during the observation .
for the sake of simplicity , we assume a homogeneous atmosphere in terms of airmass to obtain the atmospheric transmittance at a given airmass by scaling that at the airmass of 1.0 .
in fact , such a simple scaling tends to overestimate the atmospheric extinction at @xmath63 @xmath64 1 due to the non - linear dependence of the amount of extinction on airmass ( @xcite ; @xcite ) .
thus , the resultant quantities , @xmath65 and pwv , are considered to be lower and upper limits , respectively .
@xmath60 is measured at 77 k in the laboratory ( table [ tab : spec_nir_nb ] ) .
we calculate @xmath61 using the atran model . as the atmospheric transmittances at the @xmath20 and @xmath21-bands
are largely independent regardless of pwv ( see figure [ fig : atran_nir ] ) , the factor @xmath66 can be calculated , for example , for the @xmath21-band as the following : @xmath67 in the same way , we calculate the factor for the @xmath20-band ( @xmath68 ) .
since the factor for the @xmath20-band is much the same as that for the @xmath21-band with a dispersion of a few percent , indicating that the factor has small dependence on wavelength , we then interpolate @xmath68 and @xmath69 to estimate the same factor for a nb filter ( @xmath51875 or @xmath5191 ) , @xmath70 . finally , we obtain pwv by iteratively calculating the left - hand integral in equation ( [ eq : integral ] ) within the bandpass [ @xmath71 , @xmath72 of the nb filter with changing pwv to make it equal to the right - hand value .
@xmath73 ( 85mm,85mm)figure12a.eps ( 85mm,85mm)figure12b.eps the evaluated pwv values and the corresponding atmospheric transmittances at the @xmath51875 ( @xmath5191 ) filter are shown in the left- and right - hand ordinates of figure [ fig : pwv_comp]a ( [ fig : pwv_comp]b ) , respectively .
each data point is calculated by using a reduced and combined nb image of several dithering frames which has a total integration time of @xmath6 1200 s. the vertical error bars include in quadrature ( i ) the 1@xmath74 uncertainty of the throughput of the nb data and ( ii ) frame - by - frame fluctuations of the throughput in the same dithering sequence .
we evaluate the latter factor by processing individual frames in the same manner as described above for two objects : one with a higher ( @xmath6 0.8 mm ) and another with a lower ( @xmath6 0.2 mm ) pwv evaluated from their stacked image .
we find a frame - by - frame fluctuation of about 0.14 mm for both objects , and we thus introduce a constant factor of 0.14 mm as the frame - by - frame fluctuation for all the data in figure [ fig : pwv_comp ] .
our result tends to show lower pwv values ( the median and its 1@xmath74 dispersion are 0.40 @xmath7 0.30 for @xmath51875 and 0.37 @xmath7 0.21 mm for @xmath5191 ) than those reported previously ( i.e. , @xcite ; @xcite ) .
this may be partly because our observations have been carried out mostly in may and october , when the climate is expected to be drier than the yearly mean .
note that the effect of the airmass has been corrected in figure [ fig : pwv_comp ] , so that the dispersion of each measurement of pwv is likely to be caused by the temporal fluctuation in the atmospheric absorption . for a quantitative comparison of the result ,
we use archival pwv data measured by a radiometer of atacama pathfinder experiment ( apex , @xcite ) located near the base of cerro chajnantor at an altitude of 5,100 m. we extract the apex pwv data during our @xmath51875 ( @xmath5191 ) observations , which are plotted along the abscissa in figure [ fig : pwv_comp]a ( [ fig : pwv_comp]b ) .
the error bars represent 1@xmath74 dispersion of the extracted apex pwv during our observations ( @xmath6 1200 s per object ) .
we clearly see that the pwv evaluated at the tao site is remarkably lower ( the median ratio and its 1@xmath74 dispersion are 49% @xmath7 38% for @xmath51875 and 59% @xmath7 26% for @xmath5191 ) than those measured at the apex site . a simple linear fit through the origin to the distribution yields a slope of 0.47 for @xmath51875 and 0.60 for @xmath5191 ( the _ dotted _ line in figure [ fig : pwv_comp ] ) .
let us now consider the vertical distribution of the water vapor to discuss possible causes of the difference in the pwv between the apex ( pwv@xmath75 ) and the tao ( pwv@xmath76 ) sites .
when the water vapor is assumed to be distributed exponentially , pwv is derived by integrating an exponential profile , @xmath77@xmath78@xmath79 $ ] , over altitude @xmath80 , where @xmath81 is the water vapor density ( in kg m@xmath82 ) at a given altitude @xmath80 above sea level ( in km ) , @xmath83 the density at a reference altitude @xmath84 , @xmath85 the scale height at which the water vapor density decreases by a factor of @xmath86 .
@xcite have measured the vertical distribution of the water vapor above the chajnantor plateau by a combination of radiometric ( at both 183 and 225 ghz ) and radiosonde measurements , and derived the median scale height @xmath85 @xmath6 1.13 km by fitting the individual distributions to the exponential profile with @xmath84 of 5.0 km ( i.e. , chajnantor plateau ) . by using a scale height of 1.13 km ,
the pwv at the tao site is calculated for a given pwv at the apex site , as follows ( see also , ) .
the density @xmath83 is found by integrating the above equation above 5,100 m since the integral equals to the pwv at the apex site .
for example , the pwv at the apex site of 1.0 mm leads to @xmath87 1.44 g @xmath17 .
then , the pwv at the tao site is derived , to be @xmath6 0.62 mm in the case of the example , by integrating the equation above 5,640 m with @xmath83 substituted . using the measurements of the exponential scale height at night time ( the peak - to - peak values in @xmath85 of @xmath6 0.31.9 km ) by @xcite
, we derive the corresponding pwv values at the tao site as a function of the pwv at the apex site , which are shown in figure [ fig : pwv_comp ] with a _ shaded _ region .
we find that almost all of the data points are in excellent agreement , within the uncertainties , with the expectation of such exponential distributions of the water vapor having the measured scale heights .
note that there are a few outliers above the region , which might be caused by liquid phase of water ( fog or clouds ) in the atmosphere which is not detected by the apex radiometer operating at 183 ghz , as suggested by previous studies ( @xcite ; @xcite ) .
we should also note the possible influence of the presence of temperature inversion layers on the water vapor distribution .
@xcite find that their radiosonde data often show temperature inversions which make the distribution far from the exponential shape , so that much of the water vapor would be trapped below the inversion layers . to explore
the influence requires a large sample of radiosonde data ( or any equivalent data on the vertical distributions of the water vapor ) and it is far beyond the scope of this paper , but we consider that our conclusion of the advantage of the site remains unrevised , because temperature inversions often take place below the altitude of 5,500 m at night time ( @xcite ) , indicating systematically lower pwv content at the tao site than at the apex site , which is qualitatively the same trend as derived from figure [ fig : pwv_comp ] . while there is relatively large uncertainties in our analysis , it is encouraging to see that the low pwv content at the tao site is confirmed by independent methods ( i.e. , nb imaging and radiosonde ) , which suggests that the tao site at the summit of cerro chajnantor is suitable for infrared astronomy , and in particular that minitao / anir has an excellent capability for pa@xmath3 observations at around @xmath4 1.9 @xmath2 m .
we have developed a near - infrared camera , anir ( atacama nir camera ) , for the 1.0-m minitao telescope installed at the summit of cerro chajnantor ( an altitude of 5,640 m ) in northern chile .
the camera covers a field of view of @xmath0 with a spatial resolution of pixel@xmath1 in the wavelength range of 0.95 to 2.4 @xmath2 m .
the unique feature of the camera coupled with advantages of the site is a capability of narrow - band imaging observations of a strong hydrogen emission line , paschen-@xmath3 ( pa@xmath3 ) , at @xmath4 1.8751 @xmath2 m from the ground , at which wavelength it has been quite difficult to conduct ground - based observations so far due to deep atmospheric absorption mostly from water vapor .
we have been successfully obtaining pa@xmath3 images of galactic objects and nearby galaxies since the first light observation with anir in 2009 .
the throughputs at the narrow - band filters ( @xmath51875 and @xmath5191 ) show larger dispersion ( @xmath6 10% ) than those of the broad - band filters ( a few percent ) , indicating that they are affected by temporal fluctuations in precipitable water vapor ( pwv ) above the site . combining the atmospheric transmission model with the throughputs
, we evaluated the atmospheric transmittance at the narrow - band filters and the pwv content at the site .
we find that the median and the dispersion of the pwv are 0.40 @xmath7 0.30 for @xmath51875 and 0.37 @xmath7 0.21 mm for @xmath5191 data .
comparing those data with the radiometer data taken by apex , we find that the site has remarkably lower ( 49% @xmath7 38% for @xmath51875 and 59% @xmath7 26% for @xmath5191 ) pwv values than the apex site ( 5,100 m ) .
the differences in pwv between those sites are found to be in excellent agreement with those expected from the exponential distribution of the water vapor with scale height within 0.31.9 km which has been measured using the radiosonde at night time ( @xcite ) .
although temperature inversions ( which mostly take place below the summit at night - time ) make the water vapor distribution far from exponential , it leads to lower pwv content above the inversion layers , which is qualitatively consistent with our findings described above . taken all together
, we conclude that minitao / anir and the site , the summit of cerro chajnantor , provides us with an excellent capability for _ ground - based _ pa@xmath3 observations .
we are extremely grateful to the anonymous referee for useful comments and suggestions that helped improve the quality of the paper .
we would like to acknowledge mara teresa ruiz gonzlez , leonardo bronfman , mario hamuy , and ren alejandro mndez bussard at the university of chile for their support for the tao project .
operation of anir on the minitao telescope is supported by ministry of education , culture , sports , science and technology of japan , grant - in - aid for scientific research ( 17104002 , 20040003 , 20041003 , 21018003 , 21018005 , 21684006 , 22253002 , 22540258 , and 23540261 ) from japan society for the promotion of science ( jsps ) .
part of this work has been supported by the institutional program for young researcher overseas visits operated by jsps .
part of this work has been supported by the advanced technology center , national astronomical observatory of japan ( naoj ) .
part of this work has been supported by naoj research grant for universities , and by optical & near - infrared astronomy inter - university cooperation program , supported by the mext of japan .
part of the anir development was supported by the advanced technology center , naoj .
the pace hawaii-2 fpa array detector has been generously leased by subaru telescope , naoj .
the image reduction and analysis facility ( iraf ) used in this paper is distributed by the national optical astronomy observatories , which are operated by the association of universities for research in astronomy , inc .
, under cooperative agreement with the national science foundation .
this publication makes use of data products from the two micron all sky survey , which is a joint project of the university of massachusetts and the infrared processing and analysis center / california institute of technology , funded by the national aeronautics and space administration and the national science foundation .
some / all of the data presented in this paper were obtained from the mikulski archive for space telescopes ( 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 nnx09af08 g and by other grants and contracts .
we would like to thank apex for providing their radiometer data on the web .
apex is a collaboration between the max planck institute for radio astronomy , the european southern observatory , and the onsala space observatory .
alonso - herrero , a. , rieke , g. h. , rieke , m. j. , colina , l. , prez - gonzlez , p. g. , & ryder , s. d. 2006b , 650 , 835 bessell , m. s. & brett , j. m. , 1988 , , 100 , 1134 bothwell , m. s. , 2011 , , 415 , 1815 dobashi , k. , bernard , j .- p . , kawamura , a. , egusa , f. , hughes , a. , paradis , d. , bot , c. , & reach , w. t. 2009 , , 137 , 5099 dong , h. , 2011 , , 417 , 114 erasmus , d. , & sarazin , m. 2002 , aspc , 266 , 310e finger , g. , biereichel , p. , mehrgan , h. , meyer , m. , moorwood , a. f. , nicolini , g. , & stegmeier , j. 1998 , proc .
spie , 3354 , 87 garca - marn , m. , colina , l. , arribas , s. , alonso - herrero , a. , & mediavilla , e. 2006 , , 650 , 850 giovanelli , r. , 2001a , , 113 , 789 giovanelli , r. , 2001b , , 113 , 803 gsten , r. , nyman , l. . , schilke , p. , menten , k. , cesarsky , c. , & booth , r. 2006 , , 454 , l13 hodapp , k .- w .
, 1996 , new astron . , 1 , 177 imara , n. & blitz , l. 2007 , , 662 , 969 kerber , f. , 2010 , proc .
spie , 7733 , 77331 m kennicutt , jr . ,
r. c. 1998 , , 36 , 189 komugi , s. , 2012 , , 757 , 138 kurucz , r. , 1993a , atlas9 stellar atmosphere programs and 2 km / s grid .
kurucz cd - rom no .
cambridge , mass . : smithsonian astrophysical observatory , 1993 .
, 13 landolt , a. u. 1992 , , 104 , 340 lasker , b. m. , 2008 , , 136 , 735 lord , s. d. 1992 , nasa technical memorandum 103957 , a. & bell , r. a. 1979 , , 91 , 848 matsushita , s. & matsuo , h. 2003 , , 55 , 325 mclean , i. s. , smith , e. c. , becklin , e. e. , dunham , e. w. , milburn , j. w. , & savage , m. l. 2012 , proc .
spie , 8446 , 844619 minezaki , t. , 2010 , proc .
spie , 7733 , 773356 motohara , k. , 2002 , pasj , 54 , 315 motohara , k. , 2008 , proc .
spie , 7012 , 701244 osterbrock , d. e. 1989 , astrophysics of gaseous nebulae and active galactic nuclei ( mill valley , ca : univ .
science books ) , a. , 2010 , , 122 , 470 otrola , a. , querel , r. , & kerber , f. 2011 , arxiv:1103.3025 , j. b. , radford , s. j. e. , ade , p. a. r. , chamberlin , r. a. , okelly , m. j. , peterson , k. m. & schartman , e. 2003 , , 115 , 383 piqueras lpez , j. , colina , l. , arribas , s. , & alonso - herrero , a. 2013 , , 553 , a85 pei , y. c. 1992 , , 395 , 130 rousselot , p. , lidman , c. , cuby , j .- g . , moreels , g. & monnet , g. 2000 , 354 , 1134 sako , s. , 2008a , proc .
spie , 7012 , 70122 t sako , s. , 2008b , proc .
spie , 7021 , 702128 simons , d. a. & tokunaga , a. 2002 , , 114 , 169 skrutskie , m. f. , 2006 , , 131 , 1163 stead , j. j. & hoare , m. g. 2011 , , 418 , 2219 takeuchi , t. t. , buat , v. , heinis , s. , giovannoli , e. , yuan , f .-
, iglesias - pramo , j. , murata , k. l. , & burgarella , d. 2010 , , 514 , a4 tamura , y. , 2011 , , 63 , 347 tateuchi , k. , 2012a , proc .
spie , 8446 , 84467d tateuchi , k. , 2012b , publ .
korean astron .
, 27 , 297 tateuchi , k. , 2014 , , accepted ( arxiv:1412.3899 ) thompson , r. i. , rieke , m. , schneider , g. , hines , d. c. , & corbin , m. r. 1998 , apjl , 492 , l95 tokunaga , a. t. , simons , d. a. , & vacca , w. d. 2002 , , 114 , 180 tokunaga , a. t & vacca , w. d. 2005 , , 117 , 421 wang , q. d. , 2010 , , 402 , 895 yoshii , y. , 2010 , proc .
spie , 7733 , 773308 yoshikawa , t. , omata , k. , konishi , m. , ichikawa , t. , suzuki , r. , tokoku , c. , uchimoto , y. k. , & nishimura , t. 2006 , proc .
spie , 6274 , 62740y | we have developed a near - infrared camera called anir ( atacama near - infrared camera ) for the university of tokyo atacama observatory 1.0-m telescope ( minitao ) installed at the summit of cerro chajnantor ( 5,640 m above sea level ) in northern chile .
the camera provides a field of view of @xmath0 with a spatial resolution of pixel@xmath1 in the wavelength range from 0.95 to 2.4 @xmath2 m , using offner relay optics and a pace hawaii-2 focal plane array .
taking advantage of the dry site , the camera is capable of hydrogen paschen-@xmath3 ( pa@xmath3 , @xmath4 1.8751 @xmath2 m in air ) narrow - band imaging observations , at which wavelength ground - based observations have been quite difficult due to deep atmospheric absorption mainly from water vapor .
we have been successfully obtaining pa@xmath3 images of galactic objects and nearby galaxies since the first - light observation in 2009 with anir .
the throughputs at the narrow - band filters ( @xmath51875 , @xmath5191 ) including the atmospheric absorption show larger dispersion ( @xmath6 10% ) than those at broad - band filters ( a few percent ) , indicating that they are affected by temporal fluctuations in precipitable water vapor ( pwv ) above the site .
we evaluate the pwv content via the atmospheric transmittance at the narrow - band filters , and derive the median and the dispersion of the distribution of the pwv of 0.40 @xmath7 0.30 and 0.37 @xmath7 0.21 mm for the @xmath51875 and @xmath5191 data , respectively , which are remarkably smaller ( 49 @xmath7 38% for @xmath51875 and 59 @xmath7 26% for @xmath5191 ) than radiometry measurements at the base of cerro chajnantor ( an altitude of 5,100 m ) . the decrease in pwv
can be explained by the altitude of the site when we assume that the vertical distribution of the water vapor is approximated at an exponential profile with scale height within 0.31.9 km ( previously observed values at night ) .
we thus conclude that minitao / anir at the summit of cerro chajnantor indeed provides us an excellent capability for a _ ground - based _
pa@xmath3 observation . |
one dimensional single chirp signal , defined as @xmath0 for @xmath1 , is frequently used in different field of sciences , for example , sonar , radar , communications systems , as well as in oceanography and geology .
one may see abatzoglou ( 1986 ) , kumaresan and verma ( 1987 ) , djuric and kay(1990 ) , gini et al .
( 2000 ) , lin and djuric ( 2000 ) , lahiri et al .
( 2012 , 2014 ) and the references cited therein for details .
recently various types of parameter estimation techniques and their various properties have been studied for the signal ( [ eq1:model 1 ] ) , for example see kumaresan and verma ( 1987 ) , djuric and kay ( 1990 ) , gini et al .
( 2000 ) , nandi and kundu ( 2004 ) , kundu and nandi(2008 ) , lahiri et al .
( 2014 ) , saha and kay ( 2002 ) and references cited therein .
kumerasan and verma ( 1987 ) used rank reduction technique for estimating parameters of the model .
djuric and kay ( 1990 ) proposed a linear regression technique after phase unwrapping .
gini et al .
( 2000 ) used maximum likelihood ( ml ) technique as one of their estimation technique .
saha and kay ( 2002 ) used ml technique on superimposed chirp signals .
they have used mcmc importance sampling for find maximum likelihood estimates .
lin et al .
( 2004 ) has found the maximum likelihood estimates of the parameters of chirp signal using simulated annealing technique .
it is seen that most of the methods concentrated on ml technique in recent past .
recently some other techniques have drawn attention to the statistics community .
for example , nandi and kundu ( 2004 ) first provided the asymptotic properties of least square estimates ( lse ) of the parameters involved in one dimensional chirp signal with i. i. d. error structure .
kundu and nandi ( 2008 ) extended those result in case of linear stationary errors with known auto covariance function .
lahiri et al .
( 2014 ) has used the least absolute deviation ( lad ) technique to find the estimates of the parameters involved in the model .
they also gave the asymptotic properties of lad estimates under i. i. d. error structure .
although , similar to kundu and nandi ( 2008 ) , lahiri et al .
( 2014 ) assumed that the error variance is known .
therefore , it is seen that considerable amount of classical estimation techniques have been used for estimating the parameters of the chirp signal and their theoretical properties have studied in different circumstances for a while .
some bayesian analysis of the chirp signal are found in the literature also .
lin and djuric ( 2000 ) has done estimation of parameters of multiple of chirp signal using mcmc technique .
however , they have only taken i. i. d. error structure into account . moreover , it is important to mention that none of the methods , proposed so far , has taken the prediction issue into consideration .
here we have analysed the one dimensional single chirp signal for forecasting in bayesian paradigm . to be precise , our main aim , in this paper ,
is to predict a future observation through the bayesian analysis of one dimensional single chirp signal .
the advantage of using the bayesian analysis for purpose of prediction is that it gives not only a single value or an interval , but also a complete density , which is known as posterior predictive density .
it is also well known that the posterior predictive density is used for checking whether the model and the prior give a reasonable clarification of the uncertainty in a study ( see box e. p. george and tiao c. george ( 1973 ) and bickel j. peter and doksum a. kjella ( 2007 ) ) .
to achieve posterior predictive density we have used mcmc technique suitably and in the path of getting posterior predictive density , posterior densities of the parameters involved in the model have been found as by product . using these posterior densities one may perform the bayesian inference of the parameters involved in the model , when required .
the first part of the work mainly focuses on the i. i. d. error structure where we have simulated four different samples from the model ( [ eq1:model 1 ] ) and have illustrated the mcmc based bayesian analysis of the model .
moreover , this mcmc based bayesian method is applied on three different real data sets , obtained from http://archive.ics.uci.edu/ml , to see how our method is performing in practice , and in particular one of these three data set is used for multiple step forecasting . in second part , we deal with the dependent error structure though mcmc based methodologies with the same goal of forecasting .
kundu and nandi ( 2008 ) has dealt with the model ( 1 ) assuming stationary error structure in great detail using classical inference , focusing estimation of the model parameters . however , in their numerical studies they have assumed that the auto covariance function ( acf ) is completely known . in our discussion
it is assumed that the covariance structure of the error is exponentially decaying but unknown . in discrete time
, it is known that exponentially decaying acf corresponds to stationary auto regressive process of order one ( ar(1 ) ) and kundu and nandi ( 2008 ) has presented the ar(1 ) example in their paper in numerical studies . with the same choice of the parameter values ,
as done in kundu and nandi ( 2008 ) , a simulation study is done in our paper for purpose of illustration .
the remaining part of the paper is designed as follows . in section 2
we describe the parameter spaces and give a overview of mcmc based bayesian methodology . in subsection ( 2.1.1 )
we provide the required details for gibbs sampling , used in getting sample from joint posterior of the parameters . in the next subsection ( 2.1.2 )
prior specifications for the parameters are made and the full conditional density functions , which are required for gibbs sampling , are evaluated for the cases where the closed form of the conditional densities are available . in all other cases random walk mcmc is proposed ( see gamerman and lopes ( 2006 ) and liu , s. jun ( 2008 ) ) to update the parameters . in section 3
we give the results of simulation studies based on our method , and in section 4 we show the performance of our method when applied to real data .
section 5 deals with the dependent error structure where we assume an exponentially decaying covariance function with respect to time .
finally we give conclusion and future work in section 6 .
one dimensional single chirp signal ( defined as equation ( [ eq1:model 1 ] ) ) , assuming @xmath2 be random with @xmath3 = 0 , and var(@xmath2 ) = @xmath4 for all @xmath5 , has 5 parameters , namely , @xmath6 , @xmath7 , @xmath8 , @xmath9 and @xmath10 .
following lahiri et .
al ( 2012 ) we assume the following conditions on the parameters @xmath6 , @xmath7 , @xmath8 and @xmath9 : 1 .
@xmath11 + @xmath12 @xmath13 , for some known real number @xmath14 .
2 . @xmath8 , @xmath9 @xmath15 . for purpose of ease in computation , we further reparametrize the above structure as follows .
we take @xmath6 = @xmath16 and @xmath7 = @xmath17 , with @xmath18 and @xmath19 @xmath20 @xmath21 $ ] . then obviously , @xmath22 = @xmath23 @xmath24 @xmath25
therefore , we rewrite the model ( [ eq1:model 1 ] ) as , for @xmath26 = @xmath27 , @xmath28 with parameters @xmath29 , @xmath19 , @xmath8 , @xmath9 and @xmath10 along with their parameter spaces : * @xmath29 @xmath20 @xmath30 . * @xmath19 @xmath20 @xmath21 $ ] . *
@xmath8 , @xmath9 @xmath15 . *
@xmath10 @xmath20 @xmath31 .
it needs to be noted that lahiri et .
al ( 2012 , 2014 ) assumed @xmath4 is known , unlike us .
although nandi and kundu ( 2004 ) provided an estimate of @xmath4 in their theoretical study , but for numerical studies they took @xmath4 to be known .
for purpose of bayesian analysis , we assume that the parameters are random and each having a prior distribution .
our main goal is to get the posterior predictive distribution of @xmath32 given the data @xmath33 in bayesian paradigm .
we assume that @xmath34 \sim n(0,\sigma^2_{\mbox{\scriptsize $ \epsilon$}})$ ] , for all @xmath5 .
given @xmath29 , @xmath19 ( i.e. , @xmath6 and @xmath7 ) , @xmath8 , @xmath9 and @xmath4 , @xmath33 = @xmath35 , follows multivariate normal distribution with e(@xmath36 = @xmath37 and cov(@xmath36 = @xmath4 @xmath38 , where @xmath37 = @xmath39 , with @xmath40 = @xmath41 , @xmath1 , and @xmath38 is an identity matrix of order @xmath42 .
we want to find @xmath43 $ ] , the conditional distribution of @xmath32 given the data @xmath33 = @xmath35 .
using augmentation technique @xmath43 $ ] can be written as @xmath44 = \int \int \int \int \int [ y_{t+1}|{\mbox{\boldmath{$ y $ } } } , r,\theta,\alpha,\beta,\sigma_{\mbox{\scriptsize $ \epsilon$ } } ] [ r,\theta,\alpha,\beta,\sigma_{\mbox{\scriptsize $ \epsilon$}}|{\mbox{\boldmath{$ y $ } } } ] dr\ , d\theta\ , d\alpha\ , d\beta\ , d\sigma_{\mbox{\scriptsize $ \epsilon$}}\end{aligned}\ ] ] it is not possible to get an analytical form to the above integration .
therefore , simulation technique has to be implemented .
details follow . given a sample from @xmath45 $ ] ,
@xmath46 $ ] will follow a normal distribution with mean @xmath47 and variance @xmath4 . hence ,
once a sample is available from the posterior @xmath45 $ ] , the corresponding samples drawn from @xmath46 $ ] are from the posterior predictive ( [ eq3:posterior predictive ] ) , using which required posterior summaries can be obtained . to get the samples from @xmath45 $ ] , gibbs sampler method is used . in the next subsection
we give a brief description how we apply gibbs sampler technique in the present situation .
we denote the prior densities of @xmath48 , @xmath29 and @xmath4 as @xmath49 $ ] , @xmath50 $ ] , @xmath51 $ ] , @xmath52 $ ] and @xmath53 $ ] , respectively . assuming the independence of prior distributions we get the joint posterior density of @xmath29 , @xmath19 , @xmath8 , @xmath9 and @xmath54 given @xmath33 as @xmath55 \propto [ r][\theta][\alpha][\beta][\sigma_{\mbox{\scriptsize $ \epsilon$}}^2 ] [ { \mbox{\boldmath{$ y $ } } } |r , \theta , \alpha , \beta , \sigma_{\mbox{\scriptsize $ \epsilon$}}^2],\end{aligned}\ ] ] for gibbs sampling the conditional distribution of each parameters given all others ( commonly known as full conditional distribution ) , denoted by @xmath56 $ ] , are needed and given by @xmath57 \propto [ r][{\mbox{\boldmath{$ y $ } } } |r , \theta , \alpha , \beta , \sigma_{\mbox{\scriptsize $ \epsilon$}}^2],\end{aligned}\ ] ] @xmath58 \propto [ \theta][{\mbox{\boldmath{$ y $ } } } |r , \theta , \alpha , \beta , \sigma_{\mbox{\scriptsize $ \epsilon$}}^2],\end{aligned}\ ] ] @xmath59 \propto [ \alpha][{\mbox{\boldmath{$ y $ } } } |r , \theta , \alpha , \beta , \sigma_{\mbox{\scriptsize $ \epsilon$}}^2],\end{aligned}\ ] ] @xmath60 \propto [ \beta][{\mbox{\boldmath{$ y $ } } } |r , \theta , \alpha , \beta , \sigma_{\mbox{\scriptsize $ \epsilon$}}^2],\end{aligned}\ ] ] @xmath61 \propto [ \sigma_{\mbox{\scriptsize $ \epsilon$}}^2][{\mbox{\boldmath{$ y $ } } } |r , \theta , \alpha , \beta , \sigma_{\mbox{\scriptsize $ \epsilon$}}^2].\end{aligned}\ ] ] we assume the following prior distributions on the parameters @xmath62 & \sim \mbox{uniform}(0,m)\\[1ex ] [ \theta ] & \sim \mbox{uniform}(0,2\pi)\\[1ex ] [ \alpha ] & \sim \mbox{vonmises}(\alpha_0,\alpha_1)\\[1ex ] [ \beta ] & \sim \mbox{vonmises}(\beta_0,\beta_1)\\[1ex ] [ \sigma^2_{\mbox{\scriptsize $ \epsilon$ } } ] & \sim \mbox{inverse gamma}(\sigma_0,\sigma_1)\end{aligned}\ ] ] the closed form of the full conditional densities of @xmath19 , @xmath8 , @xmath9 can not be obtained in closed form .
so , we have used random walk mcmc to update these parameters .
however , the conditional density of @xmath29 given all the others , i.e. , @xmath63 $ ] follows a truncated normal distribution with truncation between @xmath64 and with the mean parameter @xmath65 ) = \frac{\sum_{t=1}^{t } y_t \left(\cos \theta \cos(\alpha t+\beta t^2 ) + \sin \theta \sin(\alpha t+\beta
t^2)\right)}{\sum_{t=1}^{t } \left(\cos \theta \cos(\alpha t+\beta t^2 ) + \sin
\theta \sin(\alpha t+\beta t^2)\right)^2}\ ] ] and variance parameter @xmath66 ) = \frac{\sigma^2_{\mbox{\scriptsize $ \epsilon$}}}{\sum_{t=1}^{t } \left(\cos \theta \cos(\alpha t+\beta t^2 ) + \sin \theta \sin(\alpha t+\beta t^2)\right)^2}.\ ] ] ( the proof is given in the appendix )
. moreover , it is straightforward to see that @xmath4 given all the others , i.e. , @xmath67 $ ] follows inverse gamma distribution with the parameters @xmath68 and @xmath69 .
in this section we have done four simulation studies to illustrate our method .
we have given the true values of the parameters of simulated samples taken for our experiment in the table [ table1:simu details ] .
in each of four samples we keep the last observation for purpose of prediction .
so , we have basically 100 observations for first three samples and 19 observation for last sample .
we have applied the random walk mcmc algorithm for updating parameters @xmath19 , @xmath8 , and @xmath9 . for all practical purposes the true values of @xmath14
is not known so , we decide to take a sufficiently large value of @xmath14 to be in safe side .
we choose @xmath14 to be equal to @xmath70 .
to run mcmc simulations it is needed to choose the prior parameters appropriately . for choosing mean directions in the prior distributions of @xmath8 ,
@xmath9 special technique is used .
loglikelihood function is maximized using simulated annealing technique with respect to the parameters @xmath8 and @xmath9 ( for details see robert , p. and casella , g. ( 2004 ) and liu , s. ( 2008 ) ) , separately , and these values are used as initial values for mcmc iterations as well , for @xmath8 and @xmath9 , respectively . below we discuss about the choice of prior parameters for @xmath71 , @xmath8 and @xmath9 in details .
.description of parameters for simulated samples [ cols="^,^,^,^,^,^,^ " , ] with the above choice of hyper parameters , @xmath72 mcmc iteration have been done with burning period @xmath73 .
we use the normal random walk proposal with variance @xmath74 to update @xmath19 , @xmath8 , and @xmath9 , for all these four simulated samples . the choice of this variance is set based on a pilot run of mcmc iteration .
we mention here that once the sample observations are obtained from @xmath29 and @xmath19 , we transform the sample values to that of @xmath6 = @xmath75 and @xmath7 = @xmath76 .
details about the results of mcmc iteration for each of the sample are discussed here .
posterior densities along with the true values are provided in the figures ( [ fig : post of a , b , alpha , beta , sigma for sample1 ] ) , ( [ fig : post of a , b , alpha , beta , sigma for sample2 ] ) , ( [ fig : post of a , b , alpha , beta , for sample3 ] ) and ( [ fig : post of a , b , alpha , beta , for sample4 ] ) , for sample 1 , 2 , 3 and 4 , respectively . except for @xmath9 in the figure ( [ fig : post of a , b , alpha , beta , for sample4 ] ) , all other true values are well within the high probability region .
we have taken only 20 observations for sample 4 .
so , it is not unusual to notice such an incident , specially when the posteriors are not unimodal .
it can also be noted that as soon as the number of observations are increased to @xmath70 , the problem of @xmath9 is solved ( figure ( [ fig : post of a , b , alpha , beta , for sample3 ] ) ) .
true signals along with 95% credible intervals , obtained based on mcmc simulations , are provided in the figure ( [ fig : fit for sample1,2,3,4 ] ) .
it is seen that in all the cases the true signal falls well within the 95% credible intervals . finally , posterior predictive densities for @xmath77th observations of samples 1 , 2 , 3 and @xmath78th observation of sample 4 , are given in the figure ( [ fig : post predictive for sample1,2,3,4 ] ) .
it is seen that true future values are well within the 95% credible interval in each of the cases , which is our main aim for this paper .
three real data sets have been taken from http://archive.ics.uci.edu/ml of which two are of type sonar rocks and one is of type sonar mines .
each signal contains 60 observations .
bache , k. and lichman , m. ( 2013 ) mainly used the data for classification purpose .
they got sonar signals from two different substances one is mine , other is rocks . here
we use these data sets for showing performance of our method for purpose of one step and multiple step forecasting .
first we consider two different signals , the one from sonar mine and one of the two signals from sonar rock and keep the last observations for purpose of prediction .
therefore , we use @xmath79 observations for our analysis for the two above mentioned signals .
we analyse another sonar rock signal in a different mode , in the sense that we keep last 5 observations for purpose of multiple step forecasting .
that means that for this data set we only use @xmath80 observations for analysis . for first two signals ( the sonar mine and one of the sonar rock , for which @xmath79 observations are considered for analysis ) , we give the 95% credible interval based on sample observations obtained from mcmc simulations for purpose of fitting and the posterior predictive densities for purpose of prediction . for the last sonar rock signal five posterior predictive densities
are given to show how more than one true future values being captured by 95% credible intervals .
we follow the same path for choosing the prior parameter values for @xmath10 as we have done in the case of simulated samples . here in particular
we choose the value of @xmath81 ( prior mean ) to be @xmath82 , @xmath83 and @xmath84 for the three data sets respectively .
@xmath85 has set to be @xmath86 as earlier , so that the variance becomes half of the square of the mean , for all the data sets considered here .
accordingly we find the values of @xmath87 for each cases .
the above choice of means have been done after running a pilot mcmc iterations . for @xmath8 as well as for @xmath9 ,
the scale parameters for vonmises distributions have been chosen to be @xmath88 for each of the data sets .
the mean directions of vonmises for @xmath8 have been set to be @xmath89 and @xmath90 for the sonar mine signal and the first sonar rock signal , respectively .
similarly , for @xmath9 , we choose the mean directions of vonmises distributions to be @xmath91 and @xmath92 for the sonar mine signal and the first sonar rock signal , respectively .
these values are obtained based on a small iteration ( @xmath93 iterations ) of simulated annealing technique , separately , on @xmath8 and @xmath9 for each of the data sets .
finally , for the second sonar rock signal ( in which case @xmath80 observations are considered for analysis ) , the choice of the mean directions for @xmath8 and @xmath9 are taken to be @xmath94 and @xmath95 , obtained as a result of small number of iterations ( @xmath93 iterations ) of simulated annealing . for @xmath29 ,
the value of @xmath14 needs to be given however , the true value of @xmath14 is not known here so , we choose a large value of @xmath14 , @xmath70 , for all these real data sets . with these choices of the prior parameters we run @xmath72 mcmc iterations with burning period @xmath73 , and the following results are noted .
figure ( [ fig : sonar_data_mines_1_2_rocks_3_fit ] ) provides the 95% fit for the sonar mine signal , and the first sonar rock signal based on 59 observations .
there are 60 observations for each of these signals .
we have taken 59 observations for purpose of fitting and have kept 60th observation for prediction purpose .
figure ( [ fig : sonar_data_1_2_posterior_predictive_rocks_3_predictive ] ) gives the posterior predictive densities of 60th observations for the sonar mine signal and the first sonar rock signal .
we have noted from figure ( [ fig : sonar_data_mines_1_2_rocks_3_fit ] ) that 95% credible intervals mostly contain the true signals in both the two cases .
95% credible completely contains the true sonar rock signal .
however , three true observations ( 5th , 29th and 54th ) fall outside the 95% credible interval for the sonar mine signal ( first graph of figure ( [ fig : sonar_data_mines_1_2_rocks_3_fit ] ) ) . at the same time
it is noticed that the pattern of the signal has been best captured for the sonar mine signal . on the other hand , from figure ( [ fig : sonar_data_1_2_posterior_predictive_rocks_3_predictive ] ) it is observed that the true values of 60th observation fall well within the credible intervals for each of the two signals , the sonar mine signal and the first sonar rock signal .
the second sonar rock signal , consisting of 60 observations , is analysed as follows .
we keep first 55 observations as the known data and last 5 observations for purpose of multiple step prediction , as discussed earlier .
now , in figures ( [ fig : sonar rocks predictions_1st three ] ) and ( [ fig : sonar rocks predictions_last two ] ) the five posterior predictive densities are given for last five observations , respectively .
it is interesting to observe that true values of 56th , 57th , 58th , 59th and 60th observations fall well within the 95% credible region .
it is notable to see that even with only @xmath80 observations we can predict next @xmath96 observations in a reasonable way .
in this section we assume that @xmath97 = @xmath98 , given @xmath54 and @xmath99 has a multivariate normal distribution with mean @xmath100 and covariance matrix @xmath101 where @xmath102 = @xmath103 is the correlation matrix of order @xmath42 , with the following structure @xmath104 \exp{(-\rho |i - j| ) } & \mbox { otherwise } , \end{cases } \label{eq30:correlation of epsilon}\ ] ] with @xmath99 @xmath20 @xmath31 . under the above assumptions
@xmath105 $ ] follows a multivariate normal distribution with the mean parameter @xmath37 = @xmath106 ( @xmath40 is equal to @xmath41 , for @xmath1 ) and the covariance matrix @xmath107 of order @xmath42 .
given a data @xmath33 , our main aim is to get a posterior predictive density of @xmath32 for one step forecasting .
the density of @xmath43 $ ] can be written as @xmath108 = \int \int \int \int \int \int [ y_{t+1}|{\mbox{\boldmath{$ y $ } } } , r,\theta,\alpha,\beta,\sigma_{\mbox{\scriptsize $ \epsilon$}},\rho ] [ r,\theta,\alpha,\beta,\sigma_{\mbox{\scriptsize $ \epsilon$}},\rho|{\mbox{\boldmath{$ y $ } } } ] \ , dr\ , d\theta\ , d\alpha\ , d\beta\ , d\sigma_{\mbox{\scriptsize $ \epsilon$}}\ , d\rho,\ ] ] as done in equation ( [ eq3:posterior predictive ] ) for independent error structure .
it has to be noted that now the number of parameter increases to @xmath109 from @xmath96 ( the number of parameters present in the i. i. d. case ) .
as mentioned earlier in section ( [ 2.1:mcmc mth ] ) , it is not possible to get an analytical form of the above integration .
the same simulation technique , as done in ( [ 2.1:mcmc mth ] ) is implemented here .
it is easily seen that @xmath110 $ ] follows a normal distribution with mean @xmath111 and variance @xmath112 where @xmath47 = @xmath113 + @xmath114 and @xmath115 is a vector of order @xmath116 , containing the covariances between @xmath32 and @xmath117
. therefore , once a sample is available from the posterior @xmath118 $ ] , the corresponding samples drawn from @xmath46 $ ] are from the posterior predictive ( [ eq22:post pred depend ] ) .
it is good to mention here that samples can be generated from @xmath119 $ ] as well for multiple step forecasting , with a little generalization of augmentation technique , adding each simulated @xmath120 to the previous set of data @xmath121 to get @xmath122 , denoted by @xmath123 , for @xmath124 = @xmath125 .
then the above mcmc technique can be used . to be precise , at each augmentation stage
a single mcmc sample is required to draw from @xmath126 $ ] and once this sample is generated , it is easy to obtain a sample from @xmath127 $ ] , denoted as @xmath128 $ ] , because @xmath128 $ ] will follow a normal distribution with mean @xmath129 and variance @xmath130 where @xmath131 is the mean at time @xmath132 , @xmath133 is the expectation of @xmath123 , @xmath134 is the correlation matrix of @xmath123 , and @xmath135 is a vector of order @xmath136 containing the covariances between @xmath137 and @xmath123 .
we give details of simulations in one step forecasting here which can be easily generalized for multiple step forecasting . to get the sample from @xmath118 $ ] we use gibbs sampler technique as earlier .
assuming the independence of the prior distributions , @xmath118 $ ] can be written as @xmath118 $ ] @xmath138 @xmath52 $ ] @xmath51 $ ] @xmath49 $ ] @xmath50 $ ] @xmath139 $ ] @xmath140 $ ] @xmath141 $ ] .
the choice of the prior distributions for @xmath142 and @xmath10 is taken to be the same as done in section ( [ 2.1.2:priors ] ) . for @xmath99
, we assume that @xmath143 \sim \mbox{gamma}(\rho_0,\rho_1)\ ] ] and obtain the full conditional density of @xmath99 as @xmath144 \propto [ \rho ] [ { \mbox{\boldmath{$ y $ } } } |\ldots].\ ] ] the forms of the full conditional distributions of @xmath29 , @xmath19 , @xmath8 , @xmath9 and @xmath10 remain the same as equations ( [ eq7:posterior of r ] ) , ( [ eq8:posterior of theta ] ) , ( [ eq9:posterior of alpha ] ) , ( [ eq10:posterior of beta ] ) , ( [ eq11:posterior of sigma ] ) , respectively . in the current scenario
also , the closed form of the full conditional densities are available only for @xmath54 and @xmath29 . for rest of the parameters we use the normal random walk mcmc with variance @xmath74 , as earlier , for updating .
the full conditional distribution of @xmath29 , @xmath63 $ ] , turns out to be a truncated normal distribution , truncation between @xmath64 , with mean @xmath145 and variance @xmath146 where @xmath147 is such that @xmath37 = @xmath29 @xmath147 ( the proof is given in appendix ) .
it is easy to seen that the full conditional distribution of @xmath54 is the inverse gamma distribution with the parameters @xmath68 and @xmath148 . with
the above discussion a simulation study has been done here . for simulation of the data we choose the values of the parameters as @xmath6
= @xmath149 , @xmath7 = @xmath150 , @xmath8 = @xmath151 , @xmath9 = @xmath152 , @xmath54 = @xmath74 and @xmath99 = @xmath153 . the choice of the above parameter values are motivated from kundu and nandi ( 2008 ) .
the choice of the prior parameters are decided as before , that is , for @xmath8 and @xmath9 we run a small number of iterations ( @xmath93 ) of simulated annealing technique to maximize the loglikelihood , separately with respect to @xmath8 and @xmath9 and the values are found to be @xmath154 and @xmath155 for @xmath8 and @xmath9 , respectively .
we take these values of @xmath8 and @xmath9 as initial values for mcmc iterations as well . for @xmath54 we set the mean of the prior density , denoted by @xmath81 , to be @xmath156 and @xmath85 to be @xmath157 as done earlier ( in section ( [ 3:simu study ] ) ) and
accordingly we evaluate the values of the @xmath85 and @xmath87 . in case of @xmath99 we set the mean of prior density , @xmath158 , 1 and
the @xmath159 is chosen to be @xmath160 , so that the variance , @xmath161 , becomes one half of the mean .
accordingly we get the values of @xmath162 is calculated . using the above choice of prior parameters
@xmath72 mcmc iterations has been done with @xmath73 burning period .
here we discuss the outcomes of the experiment .
figure ( [ fig : post of a , b , alpha , beta , sigma , rho for dependent sample ] ) provides the posterior densities of @xmath6 , @xmath7 , @xmath8 , @xmath9 , @xmath10 and @xmath99 .
it is observed that the true values of these parameters fall in high probability region in each of the cases .
finally , the posterior predictive density for @xmath77th observation and the 95% credible interval obtained from simulated sample for the true signal are provided in the figure ( [ fig : posterior predictive and fit for dependent error ] ) .
it is seen that the true future value is well within the 95% credible interval of the posterior density , indicating the usefulness of our mcmc based method for forecasting .
it is also seen that the true signal is almost always contained in the 95% credible interval obtained based on samples from mcmc simulations , except for three values , namely , @xmath163st , @xmath164th and @xmath165nd observations . @xmath163st and @xmath164th true values fall below the 2.5% interval and @xmath165nd observation fall above the 97.5% interval .
however , the pattern of the true signal is very nicely described by the intervals .
in this paper we have shown that using appropriate mcmc simulation technique one can successfully forecast one or more future observations based on the available data on one dimensional single chirp signal in bayesian paradigm .
( in this regard we have considered both independent error covariance structure as well as dependent error covariance structure . for independent error covariance structure , ) we have seen that for simulated as well as for real data our method has performed very well for purpose of forecasting . in simulation studies
we have considered four different samples with different values of parameters .
we have kept the last observation for purpose of forecasting for each of these samples .
it is observed that true future values for different samples fall within the 95% credible intervals in all these cases .
moreover , in these simulation studies we have shown that true values of the parameters fall in high probability regions of the posterior densities in most of the cases .
for the case where we have used only @xmath78 observations , true value of @xmath9 has fallen in a very low probability region in posterior density of @xmath9 ( between the two modes of the posterior density of @xmath9 ) , which is clearly because of small sample size ( note that as soon as we increase the sample size to @xmath70 with the same set of values of parameters , posterior density of @xmath9 has rightly captured the true value of @xmath9 ) . here
we once again emphasise that we take @xmath10 to be unknown unlike lahiri et al .
( 2012 , 2014 ) .
this gives more freedom for using our method in practice .
mcmc simulation technique has been applied on three real data sets , taken from [ http://archive.ics.uci.edu/ml ] , to see how the method is performing in practice . in this
website data on two different types of signal are available , e.g. , sonar mine signal and sonar rock signal .
there are @xmath166 observations available corresponding to each signal . for our experiment
we choose one sonar mine signal and two sonar rock signal .
we apply mcmc iteration on @xmath79 observations for the sonar mine and one of the sonar rock signals , keeping the last observation for purpose of forecasting .
@xmath80 observations are taken for purpose of analysis in case other sonar rock signal .
we keep @xmath96 observations for purpose of prediction to show how the method is working for more than one future observations .
we have observed that in case of the sonar mine signal and the first sonar rock signal , posterior predictive densities have nicely captured the true values of @xmath166th observations and the fitting to the signals are extremely well , in the sense that 95% credible intervals , obtained from mcmc simulations , have nicely captured the true signal . for the second sonar rock data ,
we have observed that the all five true future observations have fallen within the @xmath167 credible region of the posterior predictive densities .
it is encouraging to note that with only @xmath80 observations , using mcmc iteration technique suitably , one can predict more than one observations in a significant manner , for one dimensional single chirp signal . through out these experiments ,
simulated annealing technique is used to get the initial values and mean directions of the prior distribution for the parameters @xmath8 and @xmath9 . for dependent error covariance
, we consider the covariance being exponentially decaying proportional to lag difference in this paper .
this special covariance structure corresponds to the case of auto regressive process with lag one on errors in discrete time domain ( chatfiled , 2003 ) .
kundu and nandi ( 2008 ) have done a theoretical and numerical study on the stationary error structure .
however , they have assumed that the auto covariance function is fully known , that is , they have taken the error variance and covariance to be known .
we , on the contrary , take @xmath54 and @xmath99 as unknowns and assuming prior densities on these , come up with posterior densities of @xmath168 and @xmath99 . in the numerical example
, we have shown that the true values fall in high probability region for both @xmath54 and @xmath99 , respectively .
this is clearly an improvement over the previous work . for future work
we will consider multiple chirp signal for purpose of bayesian analysis and forecasting .
one dimensional multiple chirp signal is defined as @xmath169 for details of multiple signal one may see saha and kay ( 2002 ) , kundu and nandi ( 2008 ) etc . in most of the cases
@xmath170 is assumed to be known .
we will consider @xmath170 as unknown and will perform ttmcmc ( das and bhattachariya , 2014 ) for purpose of estimation and forecasting in bayesian paradigm for our future work .
the author is very grateful to moumita das for her insightful comments and suggestions which improve the paper in every aspect .
the author is also grateful to the assistance obtained from uci machine learning data repository ( http://archive.ics.uci.edu/ml ) for the real data sets .
abatzoglou , t. ( 1986 ) . fast maximum likelihood joint estimation of frequency and frequency rate . _ ieee transactions on aerospace and electronic systems _ , * 22 * , 708 - 714 . bache , k. and lichman , m. ( 2013 ) .
uci machine learning repository [ http://archive.ics.uci.edu/ml ] .
irvine , ca : university of california , school of information and computer science .
bickel j. piter and doksum a. kjella .
( 2007 ) mathematical statistics , vol.i , 2nd ed . , pearson prentice hall
. box e. p. george and tiao c. george .
( 1973 ) bayesian inference in statistical analysis , addison wesley publishing co. chatfield , christopher .
( 2003 ) the analysis of time series : an introduction .
chapman & hall / crc .
das , m. and bhattacharya , s. ( 2014 ) .
transdimensional transformation based markov chain monte carlo . submitted .
djuric , p.m. and kay , s.m .
( 1990 ) , parameter estimation of chirp signals . _ ieee transactions on acoustics , speech and signal processing _ , * 38 * , 2118 - 2126 . gamerman , d. and lopes f. hedibert ( 2006 ) .
markov chain monte carlo : stochastic simulation for bayesian inference , 2nd ed .
chapman & hall / crc .
gini , f. , montanari , m. and verrazzani , l. ( 2000 ) .
estimation of chirp signals in compound gaussian clutter : a cyclostationary approach .
_ ieee transactions on acoustics , speech and signal processing _ , * 48 * , 1029 - 1039 . kundu , d. and nandi , s. ( 2008 ) .
parameter estimation of chirp signals in presence of stationary noise .
_ statistica sinica _ , * 18 * , 187 - 201 .
kumaresan , r. and verma , s. ( 1987 ) . on estimating the parameters of chirp signals using rank reduction techniques .
_ proceedings of 21 st asilomar conference _ , 555 - 558 , pacific grove ,
california .
lahiri , a. , kundu d. and mitra a. ( 2012 ) .
efficient algorithm for estimating the parameters of chirp signal . _ journal of multivariate analysis _ , * 108 * , 15 - 27 .
lahiri a. , kundu d. , mitra a. ( 2014 ) .
on least absolute deviation estimator of one dimensional chirp model .
_ statistics _ , * 48 * , no . 2 , 405 - 420 , 2014 .
lin c.c . and djuric p.m. ( 2000 ) .
estimation of chirp signals by mcmc .
ieee conf .
acoustics , speech and signal processing ( icassp ) _ , istambul , turkey , * 1 * , 265268 .
lin , y. , peng , y. and wang , x. maximum likelihood parameter estimation of multiple chirp signals by a new markov chain monte carlo approach .
_ proc . of the ieee , radar conf . _
* 1 * , 559 - 562 .
liu , j.s .
_ monte carlo strategies in scientific computing _ , springer .
nandi , s. and kundu , d. asymptotic properties of the least squares estimators of the parameters of the chirp signals . _
annals of the institute of statistical mathematics _ , * 56 * , no . 3 , 529 - 544 .
robert , c. p. and casella , g. ( 2004 ) .
_ monte carlo statistical methods _ , 2nd ed .
springer .
saha , s. and kay , s. m. ( 2002 ) . maximum likelihood parameter estimation of superimposed chirps using monte carlo importance sampling .
_ ieee trans . signal process _ , * 50 * , 224 - 230 .
[ thm1:full conditional of r ] let @xmath171 @xmath172 @xmath173 , where @xmath37 = @xmath39 , with @xmath40 = @xmath41 , @xmath1 , and @xmath38 be the identity matrix of order @xmath42 .
also we assume that @xmath29 @xmath20 @xmath64 , for some known real @xmath14 and @xmath52\sim \mbox { uniform } ( 0,m)$ ]
. then @xmath174 $ ] , denoted as @xmath63 $ ] , follows a truncated normal distribution with truncation between @xmath64 and with the mean parameter and variance parameter as specified in equation @xmath175 and @xmath176 , respectively .
we note that @xmath177 & \propto [ r][{\mbox{\boldmath{$ y $ } } } |\ldots ] \\ & \propto \exp \left[-\frac{1}{2\sigma^2_{\mbox{\scriptsize $ \epsilon$}}}\sum_{t=1}^{t}\left\{y_t - r(\cos \theta \cos(\alpha t+\beta t^2)+\sin \theta
\sin(\alpha t+\beta t^2))\right\}^2\right]\chi_{(0,m)}(r)\end{aligned}\ ] ] we simplify the exponent term ( without @xmath178 ) below .
therefore , @xmath177 & \propto \exp \left[-\frac{1}{\sigma_r^2 } \left\ { r-\frac{\sum_{t=1}^{t } y_t \left(\cos \theta \cos(\alpha t+\beta t^2)+\sin \theta \sin(\alpha t+\beta t^2)\right)}{\sum_{t=1}^{t } \left(\cos \theta \cos(\alpha t+\beta t^2)+\sin \theta \sin(\alpha t+\beta t^2)\right)^2}\right\}^2\right]\chi_{0,m}(r),\end{aligned}\ ] ] where @xmath180 hence the proof follows
. [ thm2:full conditional of r in dependent case ] let @xmath181 @xmath172 @xmath182 , where @xmath37 = @xmath39 , with @xmath40 = @xmath41 , @xmath1 , and @xmath102 be the correlation matrix or order @xmath42 , with the elements as specified in equation @xmath183 .
also we assume that @xmath29 @xmath20 @xmath64 , for some known real @xmath14 and @xmath52\sim \mbox { uniform } ( 0,m)$ ] .
then @xmath184 $ ] , denoted as @xmath63 $ ] , follows a truncated normal distribution with truncation between @xmath64 and with the mean parameter and variance parameter as specified in equation @xmath185 and @xmath186 , respectively .
we follow similar line of proof as done in theorem ( [ thm1:full conditional of r ] ) . writing @xmath37 = @xmath187 , we clearly observe that @xmath177 \propto \exp\left(-\frac{1}{2\sigma^2_{\mbox{\scriptsize $ \epsilon$}}}({\mbox{\boldmath{$ y $ } } } -r{\mbox{\boldmath{$ b $ } } } _ { t})'\delta_{t}^{-1}({\mbox{\boldmath{$ y $ } } } -r{\mbox{\boldmath{$ b $ } } } _ { t})\right ) \chi_{(0,m)}(r ) \end{aligned}\ ] ] after simplifying the exponent term it is readily seen that @xmath177 \propto \exp\left[-\frac{{\mbox{\boldmath{$ b $ } } } _ { t}'\delta_{t}^{-1}{\mbox{\boldmath{$ b $ } } } _ { t}}{2\sigma^2_{\mbox{\scriptsize $ \epsilon$}}}\left(r-\frac{{\mbox{\boldmath{$ y $ } } } ' \delta_{t}^{-1}{\mbox{\boldmath{$ b $ } } } _ { t}}{{\mbox{\boldmath{$ b $ } } } _ { t}'\delta_{t}^{-1}{\mbox{\boldmath{$ b $ } } } _ { t}}\right)^2\right]\chi_{(0,m)}(r)\end{aligned}\ ] ] hence the proof follows . , @xmath7 , @xmath8 , @xmath9 and @xmath188 for sample 1 of table 1 , where true values are indicated with vertical lines.,title="fig:",width=144,height=144 ] , @xmath7 , @xmath8 , @xmath9 and @xmath188 for sample 1 of table 1 , where true values are indicated with vertical lines.,title="fig:",width=144,height=144 ] , @xmath7 , @xmath8 , @xmath9 and @xmath188 for sample 1 of table 1 , where true values are indicated with vertical lines.,title="fig:",width=144,height=144 ] , @xmath7 , @xmath8 , @xmath9 and @xmath188 for sample 1 of table 1 , where true values are indicated with vertical lines.,title="fig:",width=144,height=144 ] , @xmath7 , @xmath8 , @xmath9 and @xmath188 for sample 1 of table 1 , where true values are indicated with vertical lines.,title="fig:",width=144,height=144 ] , @xmath7 , @xmath8 , @xmath9 and @xmath188 for sample 2 of table 1 , where true values are indicated with vertical lines.,title="fig:",width=144,height=144 ] , @xmath7 , @xmath8 , @xmath9 and @xmath188 for sample 2 of table 1 , where true values are indicated with vertical lines.,title="fig:",width=144,height=144 ] , @xmath7 , @xmath8 , @xmath9 and @xmath188 for sample 2 of table 1 , where true values are indicated with vertical lines.,title="fig:",width=144,height=144 ] , @xmath7 , @xmath8 , @xmath9 and @xmath188 for sample 2 of table 1 , where true values are indicated with vertical lines.,title="fig:",width=144,height=144 ] , @xmath7 , @xmath8 , @xmath9 and @xmath188 for sample 2 of table 1 , where true values are indicated with vertical lines.,title="fig:",width=144,height=144 ] , @xmath7 , @xmath8 , @xmath9 and @xmath188 for sample 3 of table 1 , where true values are indicated with vertical lines.,title="fig:",width=144,height=144 ] , @xmath7 , @xmath8 , @xmath9 and @xmath188 for sample 3 of table 1 , where true values are indicated with vertical lines.,title="fig:",width=144,height=144 ] , @xmath7 , @xmath8 , @xmath9 and @xmath188 for sample 3 of table 1 , where true values are indicated with vertical lines.,title="fig:",width=144,height=144 ] , @xmath7 , @xmath8 , @xmath9 and @xmath188 for sample 3 of table 1 , where true values are indicated with vertical lines.,title="fig:",width=144,height=144 ] , @xmath7 , @xmath8 , @xmath9 and @xmath188 for sample 3 of table 1 , where true values are indicated with vertical lines.,title="fig:",width=144,height=144 ] , @xmath7 , @xmath8 , @xmath9 and @xmath188 for sample 4 of table 1 , where true values are indicated with vertical lines.,title="fig:",width=144,height=144 ] , @xmath7 , @xmath8 , @xmath9 and @xmath188 for sample 4 of table 1 , where true values are indicated with vertical lines.,title="fig:",width=144,height=144 ] , @xmath7 , @xmath8 , @xmath9 and @xmath188 for sample 4 of table 1 , where true values are indicated with vertical lines.,title="fig:",width=144,height=144 ] , @xmath7 , @xmath8 , @xmath9 and @xmath188 for sample 4 of table 1 , where true values are indicated with vertical lines.,title="fig:",width=144,height=144 ] , @xmath7 , @xmath8 , @xmath9 and @xmath188 for sample 4 of table 1 , where true values are indicated with vertical lines.,title="fig:",width=144,height=144 ] th observations for sample 1,2,3 and that of @xmath78th observation for sample 4 of table 1 , where true values are indicated with long vertical lines and 95% credible intervals are shown with short vertical lines.,title="fig:",width=144,height=144 ] th observations for sample 1,2,3 and that of @xmath78th observation for sample 4 of table 1 , where true values are indicated with long vertical lines and 95% credible intervals are shown with short vertical lines.,title="fig:",width=144,height=144 ] th observations for sample 1,2,3 and that of @xmath78th observation for sample 4 of table 1 , where true values are indicated with long vertical lines and 95%
credible intervals are shown with short vertical lines.,title="fig:",width=144,height=144 ] th observations for sample 1,2,3 and that of @xmath78th observation for sample 4 of table 1 , where true values are indicated with long vertical lines and 95% credible intervals are shown with short vertical lines.,title="fig:",width=144,height=144 ] , @xmath7 , @xmath8 , @xmath9 , @xmath188 and @xmath99 for the numerical example of dependent error with exponentially decaying covariance structure , where true values are indicated with vertical lines.,title="fig:",width=144,height=144 ] , @xmath7 , @xmath8 , @xmath9 , @xmath188 and @xmath99 for the numerical example of dependent error with exponentially decaying covariance structure , where true values are indicated with vertical lines.,title="fig:",width=144,height=144 ] , @xmath7 , @xmath8 , @xmath9 , @xmath188 and @xmath99 for the numerical example of dependent error with exponentially decaying covariance structure , where true values are indicated with vertical lines.,title="fig:",width=144,height=144 ] , @xmath7 , @xmath8 , @xmath9 , @xmath188 and @xmath99 for the numerical example of dependent error with exponentially decaying covariance structure , where true values are indicated with vertical lines.,title="fig:",width=144,height=144 ] , @xmath7 , @xmath8 , @xmath9 , @xmath188 and @xmath99 for the numerical example of dependent error with exponentially decaying covariance structure , where true values are indicated with vertical lines.,title="fig:",width=144,height=144 ] , @xmath7 , @xmath8 , @xmath9 , @xmath188 and @xmath99 for the numerical example of dependent error with exponentially decaying covariance structure , where true values are indicated with vertical lines.,title="fig:",width=144,height=144 ] | chirp signals are frequently used in different areas of science and engineering .
mcmc based bayesian inference is done here for purpose of one step and multiple step prediction in case of one dimensional single chirp signal with i. i. d. error structure as well as dependent error structure with exponentially decaying covariances .
we use gibbs sampling technique and random walk mcmc to update the parameters .
we perform total five simulation studies for illustration purpose .
we also do some real data analysis to show how the method is working in practice . *
key words and phrases * : bayesian inference , chirp signal , gibbs sampling , posterior predictive density , random walk mcmc . |
the development of robust methods for the computation of nonequilibrium properties of quantum many - particle systems is a crucial issue in present - day condensed matter physics , with impact on topics ranging from nonequilibrium transport in nanostructures@xcite to pump - probe spectroscopy of bulk condensed matter systems@xcite and the wide range of new spectroscopies possible in cold atom systems.@xcite an important step forward occurred with the development of continuous - time quantum monte carlo ( ctqmc ) methods for impurity models .
these algorithms were first introduced as imaginary - time methods for obtaining equilibrium properties @xcite and soon afterwards were extended to real - time dynamics and nonequilibrium problems.@xcite the continuous - time methods are in essence stochastic samplings of diagrammatic expansions of the time evolution operator .
the mean perturbation order required in the calculation increases with the time ( or inverse temperature ) to be studied and the calculations are limited by the perturbation order which can be achieved with given computational resources . in the equilibrium case one
considers the imaginary - time evolution operator @xmath4 $ ] which is real and positive definite , so the computational task is to estimate a sum of real ( decaying ) exponentials and the only sign problem which arises is the fermion sign problem occurring in models complicated enough to sustain fermion loops . for these reasons
the ctqmc methods have proven to be very powerful in the equilibrium context.@xcite in the real time case , on the other hand , one must consider the intrinsically complex time evolution operator @xmath5 $ ] , and convergence comes from the cancellation of oscillations .
the theoretical task is therefore to estimate the sum of terms with oscillating signs or rotating phases and a severe dynamical " sign problem occurs even in the absence of fermion loops .
the average sign decreases exponentially with perturbation order , which limits the accessible range of interaction strengths and simulation times . because of these limitations , important questions such as the nonequilibrium kondo effect
could so far not be adequately addressed .
the equilibrium kondo effect in quantum dots,@xcite which involves the formation of a scattering resonance ( density of states peak ) at the fermi level , was experimentally confirmed in the zero bias limit.@xcite while at very low voltage biasses , the pinning of the kondo resonance to the fermi level leads to an unrenormalized conductance for symmetric dots , it is well known that a high voltage bias destroys the kondo effect .
the crossover from the low voltage universal regime ( linear response regime " ) to the higher bias coulomb blockade regime is presently not understood .
it has been proposed on the basis of the noncrossing approximation,@xcite real time diagrammatic methods,@xcite and perturbative calculations@xcite that the peak in the density of states splits into two in a certain parameter regime .
our previous investigation of the non - equilibrium anderson model @xcite produced no sign of this phenomenon , but the accuracy of the simulations in the low - bias region was not sufficient to settle the issue .
methodological improvements allowing a more accurate numerical study of the small - to - intermediate voltage regime are therefore needed .
the existing continuous - time quantum monte carlo approaches for nonequilibrium systems are more - or - less direct extensions of the imaginary time algorithms previously developed .
it appears worthwhile to attempt to optimize them , even though the dynamical sign problem inherent in these methods unavoidably limits what can be achieved . in this paper
we present an efficient implementation of the weak - coupling diagrammatic monte carlo method for non - equilibrium systems , describing ways to reorganize the expansion and to improve the measurement formulae in order to increase the accuracy of the numerical data for a given set of parameters .
the method introduced previously@xcite corresponds to the simulation of a system prepared in the nonequilibrium but non - interacting state , with the interaction turned on at time @xmath6 .
we refer to this simulation method as an interaction quench " .
since the real - time methods compute the time evolution of the system after the quench , an important consideration is the time needed for the system to evolve to the interacting steady state .
optimized preparation of the initial ensemble has the potential to reduce this relaxation time , therefore leading to simulations requiring a smaller total time interval for the measurement of a given property .
motivated by this idea we extend the formalism from two real - time branches to an
l - shaped " contour which includes an imaginary time branch .
evolution along the imaginary time branch may be thought of as preparing the system in a correlated equilibrium state , after which the voltage is turned on at time @xmath6 .
we refer to this simulation method as a voltage quench " .
one purpose of this paper is to compare interaction and voltage quenches .
we will show that at temperature @xmath7 interaction quenches are suitable for voltage biasses larger than the kondo temperature ( i. e. for voltage biasses large enough to suppress the kondo resonance in the many - body density of states ) .
the times which can be reached in the monte carlo simulation are long enough to observe convergence into a steady state even at large interaction strengths . on the other hand ,
if the voltage bias is small and the temperature is finite , the voltage quench is a suitable alternative , because it allows the important ground state correlations to be built up via the computationally less problematic imaginary time evolution .
we show that our optimized implementation allows the computation of accurate currents over a wide voltage range , even for interaction strengths which are clearly outside the reach of low - order perturbation theory .
we use the numerical results to test predictions based on fourth order perturbation theory in the self - energy.@xcite we determine the largest interaction strength for which the perturbation theory provides accurate results over the entire voltage range , and for larger interactions , the voltage window where deviations appear . the predicted splitting of the kondo resonance @xcite
is not evident in the numerical data .
the rest of this paper is organized as follows . in section [ model ]
we introduce the model to be solved and present the methods used to solve it , in particular defining the voltage and interaction quenches .
sections [ interactionquench ] and [ voltagequench ] present results for the interaction and voltage quenches respectively .
section [ iv ] gives results for the current - voltage characteristics of the model and section [ conclusions ] is a summary and conclusion .
we consider the one - orbital anderson impurity model , which describes a single spin - degenerate ( @xmath8 ) level with a hubbard interaction @xmath0 ( the dot " ) coupled by hybridization @xmath9 to two reservoirs ( leads " ) labeled by @xmath10 .
the hamiltonian @xmath11 of this model contains the terms @xmath12 in the following we will consider two sources of time dependence in @xmath13 . in the interaction
quench we take @xmath14 for times @xmath15 with an instantaneous step to a non - zero @xmath0 at @xmath6 ; in the voltage quench we take @xmath16 for time @xmath15 with an instantaneous step to a nonzero @xmath17 at @xmath6 .
we assume that the lead electrons equilibrate instantly to the new chemical potential so that the equal time correlators of lead operators are @xmath18 , with @xmath19 the fermi distribution function for temperature @xmath20 and @xmath21 the value of the chemical potential for lead @xmath22 at the appropriate time . in this paper we will consider only symmetric voltage biasses ( @xmath23 ) and half - filled dots ( @xmath24 ) .
the consequences of relaxing these assumptions will be briefly mentioned in the conclusions .
the energy scales of the model are set by the level broadenings @xmath25 associated with the leads @xmath22 . the total level broadening @xmath26 is used as the energy unit throughout the paper .
we consider flat bands centered at zero , with a high energy cutoff @xmath27 .
as we shall see in section [ voltagequench ] a sharp high frequency cutoff leads to oscillations in the time evolution of the current after a voltage quench .
a sufficiently smooth band cutoff damps the oscillations but does not affect the steady state current .
we adopt a fermi - function like smoothing with smoothing parameter " @xmath28 , @xmath29 we use the weak - coupling formulation of the real - time diagrammatic monte carlo approach as described in ref . .
this is a real - time implementation of the continuous - time auxiliary field algorithm @xcite which is based on the combination of a weak - coupling expansion and an auxiliary field decomposition . here
, we will briefly summarize the main aspects of this method and then discuss some relevant issues concerning its efficient implementation . in order to enable simulations starting from an interacting initial state ,
we formulate the method on the l - shaped contour which runs from @xmath30 to @xmath31 and back to @xmath30 along the keldysh real time axis , then to @xmath32 along the imaginary time axis . , the imaginary time branch of the contour is shifted to @xmath33 and need not be explicitly considered in the monte carlo simulation .
the red arrows represent auxiliary ising spin variables .
the top panel shows a monte carlo configuration corresponding to perturbation order @xmath34 , @xmath35 , and the bottom panel a configuration corresponding to @xmath36 , @xmath35 , @xmath37 . ]
the weak coupling algorithm may be taken to start from the following expression for the partition function @xmath38 : @xmath39 , \label{z}\end{aligned}\ ] ] with @xmath40 and @xmath41 some arbitrary ( non - zero ) constants .
the interaction and the chemical potentials need not be the same on the imaginary time branch as they are on the real - time branches . in the formalism as written the interaction and chemical potentials are taken to be time independent on the real - time branches , but it is straightforward to generalize the method to time - dependent @xmath0 and @xmath42 .
the notation @xmath43 indicates that on the real - time portion of the contour the two leads have different chemical potentials @xmath44 , whereas @xmath45 means that on the imaginary time portion of the contour the two leads have the same chemical potential @xmath46 .
henceforth we choose energies such that @xmath47 and consider a symmetrically applied bias voltage @xmath9 ( @xmath48 ) .
the time evolution along the real - time and imaginary - time contours is expanded in powers of @xmath49 and @xmath50 , respectively .
each interaction vertex is then decoupled using ising spin variables according to the formula@xcite ( @xmath51 or @xmath52 ) @xmath53 the resulting collection of ising spin variables on the contour represents the monte carlo configuration @xmath54 , with @xmath55 denoting the position of spin @xmath56 on the l - shaped contour ( see illustration in fig .
[ lcontour ] ) .
there are @xmath57 spins on the forward branch , @xmath58 spins on the backward branch and @xmath59 spins on the imaginary - time branch of the contour ( @xmath60 ) .
the weight of such a configuration is obtained by tracing over the dot and lead degrees of freedom and can be expressed in terms of two determinants of @xmath61 matrices @xmath62:@xcite @xmath63 here @xmath64 is the @xmath65 element of the @xmath61 matrix of non - interacting green functions @xmath66 computed using the possibly time - dependent chemical potentials and evaluated at the time arguments defined by the ising spins .
the quantity @xmath67 is a diagonal matrix depending on the spin variables ( with @xmath68 for spins located on the real - time branches and @xmath69 for spins on the imaginary time branch ) .
@xmath70 is the contour ordering operator , which exchanges the product @xmath71 of two operators if @xmath31 is earlier on the contour than @xmath72 ( a minus sign is added if the exchange involves an odd number of fermi operators ) . a monte carlo sampling of all possible spin configurations is then implemented based on the absolute value of the weights ( [ weight ] ) .
the contribution of a specific configuration @xmath73 to the current is given by @xcite @xmath74_{i , j}a_{0,\sigma}(t_{j } , t'),\label{eqa } \label{tildea}\end{aligned}\ ] ] with the first term on the right hand giving the contribution to the non - interacting current and the second term a correction due to the interactions .
( [ tildea ] ) @xmath75 denotes a dot - lead correlation function of the noninteracting model for the composite left lead operator @xmath76 .
the current expectation value is @xmath77,\ ] ] where @xmath78 denotes the monte carlo average and @xmath79 the phase of the weight of the configuration @xmath80 . in an interaction quench ,
the imaginary - time evolution is not explicitly considered in the monte carlo simulation and temperature appears only as a parameter in the noninteracting green functions ( see fig .
[ lcontour ] ) .
moreover , the latter depend only on time differences , and thus can be easily expressed in terms of their fourier transform . assuming a large band cutoff and neglecting the real part of the lead self - energy we find@xcite @xmath81 in the voltage quench , on the other hand , the interaction is non - vanishing on the imaginary time portion of the contour ( fig .
[ lcontour ] ) , while the chemical potential difference jumps instantaneously from zero ( on the imaginary branch ) to @xmath9 ( on the real branches ) .
because of the time dependence of the chemical potentials , the noninteracting green functions are not time translation invariant and we can not express @xmath82 and the dot - lead correlator @xmath83 in the form of a fourier transform . instead , those functions must be computed numerically from their equations of motion , as explained in the appendix .
the sign ( phase ) problem in the weak - coupling ctqmc method grows exponentially with the average perturbation order on the real - time branches , which in turn is proportional to the simulation time , while operators on the imaginary time branch do not add significantly to the sign problem . to reach long times or strong interactions
, it is therefore important to reduce the average perturbation order on the real - time branches as much as possible .
an essential point to note in this context is that in the particle - hole symmetric case , the parameters @xmath84 of the algorithm can be chosen such that only even perturbation orders appear in the expansion .
in fact , for @xmath85 the spin degree of freedom effectively disappears ( @xmath86 ) and the algorithm becomes the real - time version of rubtsov s weak - coupling method@xcite for the particle - hole symmetric interaction term @xmath87 .
( for a detailed discussion of the equivalence between the rubtsov and ctaux methods for the anderson impurity model , consult ref . ) .
the odd perturbation orders are continuously suppressed as @xmath84 approaches @xmath88 . for @xmath89 and sufficiently small @xmath90 ,
the average perturbation order can be reduced by about half compared to the @xmath91 used in the simulations presented in ref . .
this in turn allows us to reach times and interaction strengths which are a factor of two larger .
we note in passing that the suppression of odd perturbation orders was also essential in the nonequilibrium dynamical mean field calculations of ref . .
we next discuss some tricks to improve the efficiency of the current measurement .
first , we rewrite eq .
( [ eqa ] ) as @xmath92_{i , j}\delta_\mathcal{c}(s_2,t_{j})\big\rangle a_{0,\sigma}(s_2 , t ' ) , \label{x}\ ] ] where the variables @xmath93 and @xmath94 run over the entire contour and the contour delta function is defined by @xmath95 .
it is therefore sufficient to accumulate the quantity @xmath96_{i , j}\delta_\mathcal{c}(s_2,t_{j})\big\rangle.\ ] ] furthermore , it follows from eq .
( [ weight ] ) that the weight of a monte carlo configuration changes sign if the last spin ( corresponding to the largest time argument ) is shifted from the forward contour to the backward contour or vice versa . since the absolute value of the weight does not change , these two configurations will be generated with equal probability . as a result ,
all the terms in eq .
( [ x ] ) which do not involve the last operator on the contour will cancel .
it is therefore more efficient and accurate to accumulate @xmath97 \big\rangle,\end{aligned}\ ] ] with @xmath98_{i , j}\delta_\mathcal{c}(s_2,t)$ ] and @xmath99 if @xmath100 and 0 otherwise .
also , by comparing the contributions to the current of the original configuration and the one with the last operator shifted from the upper to the lower contour ( or vice versa ) , one finds that they almost ( but not completely ) cancel .
the errorbars on the current can thus be substantially reduced by appropriate symmetrizations of @xmath101 .
calculations based on an interaction quench from @xmath14 are particularly simple , because there are no interaction vertices ( or spins ) on the imaginary time branch , and only the real - time branches of the contour need to be considered in the simulation .
temperature enters only as a parameter in the lead correlators , making it possible to treat arbitrary temperatures , including @xmath7 . at time @xmath6 , the system is non - interacting but subject to an applied bias @xmath9 , so a current @xmath102 appropriate to the non - interacting model is flowing through the dot . at @xmath103
the interaction is turned on and the system relaxes into the steady - state configuration appropriate to the interacting model . figure [ i_t_v4 ] shows the time dependence of the current calculated for the large bias voltage @xmath104 and several interaction strengths .
we see that the transient behavior is such that the current initially decreases sharply , overshoots and eventually relaxes more slowly back up into the new steady state .
the interaction - dependence of the steady - state current is a consequence of the coulomb blockade physics , apparent even at the large voltages studied here . for intermediate and large voltage bias ( @xmath105 ) and not too large interaction ( @xmath106 )
the time required for convergence to the steady state is @xmath107 , essentially independent of interaction strength .
given the scaling of the perturbation order ( and hence the sign problem ) with @xmath0 and @xmath31 , interactions up to @xmath108 are accessible with the current implementation .
a comparison of fig .
[ i_t_v4 ] to fig .
13 of ref .
shows that the technical improvements introduced in this paper have substantially extended the range of applicability of the weak - coupling monte carlo method ( about a factor 2 - 3 in @xmath0 or @xmath31 ) and allow us to obtain accurate results in the intermediate - to - strong correlation regime . in fig .
[ i_t_u6 ] we plot the time evolution of the current for fixed @xmath109 and several voltage biasses . for voltages
@xmath110 , even though the transient behavior is clearly voltage - dependent , the current settles into the new steady state after a time @xmath111 .
however , as the voltage is decreased below @xmath112 the transient time increases . at @xmath113
the long time limit is attained only for @xmath114 and as @xmath9 is further decreased the approach to the asymptotic behavior becomes even slower . to better analyse the approach to steady state at small voltages we present in the upper panel of fig .
[ i_t_smallv ] the time dependence of the current for several smaller voltages and two interaction strengths . for better comparison ,
we plot here the ratio @xmath115 of the interacting current @xmath116 to the noninteracting current @xmath117 .
one sees that as @xmath9 is decreased or @xmath0 is increased the evolution of the current from the post - quench minimum to the long - time steady state value takes an increasingly long time .
since the longest accessible time is @xmath118 for @xmath119 and @xmath120 for @xmath109 , the accurate measurement of @xmath116 becomes impossible in the small voltage regime .
however the short - time transient behavior is accessible at all voltages .
while the ratio @xmath121 is clearly voltage dependent at higher biasses , the data seem to converge as @xmath9 is reduced to a non - trivial curve with a pronounced minimum near an only weakly @xmath0-dependent time @xmath122 .
we believe that the increasingly slow convergence as @xmath123 is a signature of the kondo effect , which is characterized by an energy scale which becomes exponentially small as @xmath0 increases . after the interaction
quench , the kondo resonance has to be built up as time progresses , and in the limit @xmath123 , @xmath124 this requires an increasingly large number of interaction vertices and hence an increasingly long simulation time . on physical grounds
one expects that the time needed to evolve into steady state is proportional to the inverse of the associated energy scale .
empirically , we find that the slow relaxation becomes an issue in the linear response regime , where the non - interacting and interacting currents are very similar .
for @xmath125 , where the interacting current is substantially smaller than @xmath117 , a useful estimate of @xmath116 seems possible , even though in the voltage window up to @xmath112 a small drift in the current may remain up to the longest accessible times .
this drift makes it difficult to define reliable error bars on @xmath116 , but it appears unlikely that the steady state value will differ from @xmath126 by more than the largest deviation in the window @xmath127 $ ] , which we use as error estimate . for @xmath128 , an accurate estimate is not possible from the interaction quench procedure , but the current is very close to the linear response value , so the uncertainty is in fact not that important .
this is illustrated in the bottom panel of fig .
[ i_t_smallv ] , which compares the monte carlo data to the noninteracting current and results from fourth order perturbation theory.@xcite the plot also indicates as hashed region the voltage range @xmath129 where accurate measurements of the long - time limit become prohibitively difficult at @xmath7 .
we will see below that this roughly corresponds to voltages smaller than the kondo temperature @xmath130 .
it is also of interest to examine the temperature dependence of the current .
the interplay between voltage and temperature as the kondo regime is approached presents an interesting problem .
one expects that as the temperature is increased , the kondo effect gets washed out and the simulations would therefore more readily converge even at small bias voltages .
the temperature dependence of the current calculated from the interaction quench for @xmath109 and several values of the voltage bias is plotted in fig .
[ i_beta ] . in the linear response regime ( @xmath131 , @xmath132 ) the ratio of the interacting current @xmath133 to the noninteracting current @xmath134 exhibits a strong temperature dependence , even at @xmath135 .
the temperature dependence arises because lowering the temperature strengthens the kondo resonance and leads to an increase in the interacting current .
the temperature dependence for small voltage bias ( @xmath131 ) approaches the analytical result for the temperature dependent zero - bias conductance in ref . and thus allows us to estimate ( from the temperature at which @xmath136 ) the kondo temperature as @xmath137 , in good agreement with the _ a priori _ estimate from the standard formula@xcite @xmath138 this formula is valid in the strong correlation regime and for @xmath109 yields @xmath139 .
as @xmath9 is increased the temperature dependence is weakened . at intermediate values of @xmath9 , in the coulomb blockade regime ( @xmath140 ) , the current has little temperature dependence at low @xmath20 . at large voltage bias ( @xmath104 ) , correlation effects are already weakened due to the voltage , as is evident from the increase in @xmath115 , and the almost perfect agreement with fourth order perturbation theory discussed in section [ sec : iv ] .
the current in this regime remains insensitive to temperature at low @xmath20 .
an alternative procedure to calculate the steady state current of interacting quantum dots is to start from an interacting state in equilibrium ( @xmath141 ) and turn on the voltage at @xmath103 . while this approach is computationally more expensive and is restricted to nonzero temperatures , because it involves operators on the imaginary - time branch , it has the advantage that the kondo resonance in the many - body density of states is present already in the initial state .
the kondo resonance is built up during the evolution along the imaginary - time branch , which does not add significantly to the sign problem .
one might expect that this @xmath9-quench is particularly suitable to study the small voltage regime , because turning on a small voltage will not change the spectral function dramatically .
since the voltage quench has not yet been discussed in the previous literature , we will now analyze the properties of the current after such a @xmath9-quench in some detail , and in particular its dependence on the band width ( @xmath27 ) and smoothness of the cutoff ( @xmath28 ) .
the top panel of fig .
[ nu_cut ] shows the time evolution of the current in a model with @xmath142 ( lines ) and @xmath119 ( symbols ) if the voltage is suddenly increased to @xmath143 . in the initial state ,
the system is in equilibrium , with no current flowing through the dot .
after the voltage bias is turned on , the current increases . in the model with hard band cutoff ( @xmath144 , red online ) oscillations in the current appear which make it difficult to estimate the steady state value . a smoother band cutoff
( @xmath145 , blue online ) almost completely eliminates these oscillations . we will thus in the rest of this subsection show results for the smoothing parameter " @xmath145 .
the lower panel of fig .
[ nu_cut ] illustrates the dependence of the current on the cutoff value @xmath27 . while the short time behavior of
the current depends strongly on the bandwidth , the steady state value shows little cutoff dependence as long as @xmath27 is substantially larger than the applied voltage bias .
this is consistent with the observation for the @xmath0-quench in ref . , where it was found that @xmath146 was enough to get accurate results up to @xmath147 .
hence , we will choose @xmath146 for the rest of this paper . in fig .
[ vu46 ] we plot the non - interacting and interacting currents for @xmath148 and several values of the voltage bias .
the top panel is for @xmath119 and the bottom panel for @xmath109 . in the small voltage regime ( @xmath149 )
the interacting current increases monotonically with time and eventually settles into a steady state within the accessible time window .
the @xmath9-quench therefore allows us to measure accurate steady state currents for finite temperature in the small voltage regime .
however , once the voltage becomes too big ( see @xmath150 in the top panel of fig .
[ vu46 ] ) , the interacting current overshoots and only slowly settles into the steady state , making it impossible to measure an accurate steady state value using this approach .
however , as shown in the previous section , the simulation based on the @xmath0-quench provides accurate results once @xmath125 .
the two simulation methods are therefore complementary in the sense that the @xmath9-quench works best for small voltage bias ( @xmath149 ) and the @xmath0-quench at larger voltage bias ( @xmath125 ) . the temperature dependence of the current after a @xmath9-quench in the low - voltage regime ( @xmath128 and 0.5 ) is shown in fig . [ beta ] , which plots results for @xmath142 ( lines ) and @xmath119 ( symbols ) for @xmath151 , @xmath1 and @xmath152 .
a rather strong temperature dependence is evident , in particular in the interacting current .
this is consistent with the @xmath0-quench data shown in fig .
[ i_beta ] and a consequence of the destruction of the kondo resonance by temperature .
remarkably , a strong temperature dependence is observed even for @xmath153 , which means that the applied voltage does not effectively raise the temperature to a value of order @xmath9 . in this voltage regime
the non - zero voltage state therefore is not simply equivalent to a thermal state .
the nature of the correlations which give rise to the temperature dependence are an interesting subject for further investigation . in the @xmath9-quench calculations , for @xmath109 , we can access temperatures down to @xmath154 . at even lower temperatures ,
the perturbation order on the imaginary time branch becomes so large and the individual monte carlo updates so expensive that it is increasingly difficult to reach the very high statistical accuracy required for simulations with average signs of the order @xmath155 .
since the problem of slow convergence in @xmath0-quench calculations at small bias is considerably alleviated by finite temperature , it turns out that the accuracy of the latter approach matches that of @xmath9-quench calculations even at very small voltage bias ( see @xmath109 data in fig .
[ currentsmall ] ) .
for finite temperature simulations in the experimentally relevant temperature range ( @xmath156 ) , the @xmath0-quench approach thus appears to be more powerful and sufficient to treat the entire voltage range .
the good agreement between the @xmath0-quench and @xmath9-quench results in fig .
[ currentsmall ] furthermore shows that the steady state results obtained by the diagrammatic monte carlo method do not depend on the initial preparation of the system .
we now apply the machinery described in the previous section to compute the current - voltage characteristics of the anderson impurity model at half - filling . the initial rise of the current at finite temperature ( @xmath148 ) is shown in fig . [ currentsmall ] .
the blue circles and black diamonds have been obtained using the @xmath9-quench .
the current - voltage characteristics in the @xmath123 limit becomes linear , although the slopes of the interacting and non - interacting models are not identical .
this is the temperature effect on the kondo resonance ( particularly pronounced for large @xmath0 ) which was discussed in the previous sections .
as the temperature is lowered to zero , the initial slope of the current approaches that of the non - interacting model .
figure [ currentlarge ] shows the @xmath7 result obtained using interaction quenches ( @xmath157 , essentially the wide band limit ) .
the black curve shows the monotonic increase of the non - interacting current with increasing applied bias voltage . the red , blue and pink lines show the interacting current for @xmath119 , @xmath158 and @xmath159 predicted by fourth order perturbation theory.@xcite consistent with analytical arguments,@xcite the interacting current initially rises with the same slope as the non - interacting current , and reaches the non - interacting value also in the large - voltage limit . at intermediate values of @xmath9
the effect of interactions is to suppress the current ( coulomb blockade ) . in fourth order perturbation theory
, a hump appears in the @xmath116-@xmath9 curve around @xmath140 for @xmath109 and @xmath159 . at even larger
@xmath0 ( clearly outside the range of applicability ) fourth order perturbation theory will presumably lead to a negative differential conductance at intermediate @xmath9 .
the appearance of this hump is related to the splitting of the kondo resonance as discussed in ref . .
the monte carlo results for @xmath119 , @xmath158 , @xmath159 , and 10 are shown by the red stars , blue circles , pink diamonds , and orange triangles , respectively . since these are @xmath0-quench results for @xmath7 , only @xmath125 data are shown . in the large voltage regime ( @xmath160 )
the numerical results agree with the prediction from fourth order perturbation theory .
apparently , the fast decay of the green functions for large voltage bias simplifies the diagram structure such that fourth order in @xmath161 is sufficient at @xmath160 . at intermediate voltages , @xmath162
, differences between the monte carlo data and fourth order perturbation theory appear .
the essentially exact numerical data show no prominent hump feature near @xmath140 , and hence no negative differential conductance in the intermediate to strong correlation regime .
the hump , and the associated splitting of the kondo resonance , must therefore be an artefact of fourth order perturbation theory .
this is consistent with the conclusion reached in ref .
on the basis of ( less accurate ) hybridization expansion results . the data in fig .
[ currentlarge ] show that fourth order perturbation theory yields correct results over the entire voltage range for @xmath163 . for larger interactions , and in particular around @xmath112 more complicated self energy diagrams become important .
we have discussed the implementation of the weak - coupling continuous - time monte carlo method on the l - shaped keldysh contour and the application of this formalism to the study of transport through a quantum dot .
calculations based on interaction quenches from the current carrying state at @xmath14 can be restricted to the real - time contours and provide accurate steady state currents for @xmath125 and interaction strenghts @xmath164 , for arbitrary temperature and bandwidth . at finite temperature ,
convergence into the steady state is considerably faster , which allows access to the small voltage regime . as an alternative method
, we have introduced calculations based on voltage quenches , which start from the interacting equilibrium state and which can be used to calculate the steady state current in the small voltage regime ( @xmath149 ) at finite temperature . since the sign problem turns out to be essentially independent of the number of operators on the imaginary time branch , temperatures of order @xmath148
can easily be dealt with .
the @xmath9-quench approach is however not more efficient than finite - temperature @xmath0-quench calculations .
we have used the methods to accurately compute the current - voltage characteristics of the half - filled anderson - impurity model in the intermediate - to - strong coupling regime .
comparison to fourth order perturbation theory showed that the latter fails at voltages around @xmath112 for @xmath165 , but becomes accurate for @xmath166 .
the splitting of the kondo resonance predicted by low order perturbation theory is an artefact not present in the numerical data .
the results presented in this paper show that diagrammatic monte carlo is one of the most powerful numerical tools for the study of non - equilibrium systems . the accuracy of the improved weak - coupling approach and its range of applicability rivals or surpasses other state - of - the - art numerical approaches such as time - dependent dmrg.@xcite for most practical purposes the numerical problem of calculating the steady - state current through a half - filled anderson impurity model with symmetrically applied voltage can be considered as solved
. an interesting , and presumably straight - forward extension of our work would be the study of asymmetrically applied bias voltages .
one of the optimizations of the monte carlo algorithm the suppression of the odd perturbation orders is however specific to the particle - hole symmetric model .
away from particle hole symmetry , odd perturbation orders contribute to the current and therefore must be considered in the simulation .
this leads to an increase in the average perturbation order and to a more severe sign problem , such that the accessible times and interaction strengths will be reduced .
the optimal choice of the @xmath84-parameters in the particle - hole asymmetric case is an open problem for future investigations .
another issue which should be considered is the optimal shape of the @xmath0- or @xmath9-quench . by slowly ramping up the interaction or voltage bias
, it may be possible to avoid overshooting and thus observe a faster relaxation of the current into the steady state .
pw and me are supported by the swiss national science foundation ( grant pp002 - 118866 ) , to by a grant - in - aid for young scientists ( b ) from mext , and ajm by the us national science foundation division of materials research under grant dmr-0705847 .
this work also benefitted from the academic guest program ( center for theoretical studies ) of eth zurich ( to ) and the hospitality of the aspen center for physics ( pw ) .
we thank n. tsuji , t. fujii and k. ueda for helpful discussions .
the simulations were performed on the brutus cluster at eth zurich using a code based on alps.@xcite
in this appendix we present the formalism needed for the voltage quench , evaluating noninteracting green functions on the l - shaped contour with lead chemical potential @xmath21 equal to the equilibrium value @xmath167 for times on the imaginary contour and with arbitrary time - dependence @xmath168 on the real time portions of the contour .
the results of the paper correspond to @xmath169 . because the voltage bias is time dependent , noninteracting green functions
can not be expressed in the form of a fourier transform , and instead they are computed numerically by the solution of their equations of motions in real ( imaginary ) time .
a closed set of equations is obtained if one considers the noninteracting dot green function [ eq .
( [ eqn : g0input ] ) ] , @xmath170 the hybridization of the dot to a single bath level @xmath171 and the dot - decoupled green function of a single bath state , @xmath172 here @xmath173 are arbitrary points on the real or imaginary portions of the contour , the time evolution is performed with @xmath14 but time - dependent voltage bias , and @xmath174 is the grand - canonical expectation value in the noninteracting initial state ( at @xmath175 ) . the contour - ordering operator @xmath176 exchanges the product @xmath71 of two operators if @xmath31 is earlier on the contour than @xmath72 ( a minus sign is added if the exchange involves an odd number of fermi operators ) .
equations of motions for the green functions ( [ g0-def ] ) to ( [ geps - def ] ) are obtained from taking time - derivatives and evaluation of the resulting commutators , @xmath177 g_{0,\sigma}(t , t ' ) = \sum_p g_{p,\sigma}^\alpha(t , t ' ) v_p^\alpha - \delta_{\mathcal{c}}(t , t ' ) , \\
\label{app - eom - f } & [ i\partial_{t ' } + \epsilon_{p\sigma}^\alpha-\mu_\alpha(t ' ) ] \,g_{p\sigma}^\alpha(t , t ' ) = g_{0,\sigma}(t , t')\,(v_p^\alpha)^ * , \\
\label{app - eom - geps } & [ i\partial_{t ' } + \epsilon_{p\sigma}^\alpha - \mu_\alpha(t ' ) ] \,g_{p\sigma}^\alpha(t , t ' ) = -\delta_{\mathcal{c}}(t , t').\end{aligned}\ ] ] note that when @xmath178 is on the imaginary branch of the contour , the time derivative is given by @xmath179 .
the contour delta function is given by @xmath180 , where @xmath181 if @xmath31 is later on @xmath182 than @xmath72 and zero otherwise .
( [ app - eom - g0 ] ) to ( [ app - eom - geps ] ) have a unique solution , provided that the contour green functions satisfy an antiperiodic boundary condition on the contour @xmath182 in both time arguments .
equation ( [ app - eom - geps ] ) can be solved explicitly , @xmath183
\nonumber\\ \times \exp\big ( & i\int_0^{t ' } \!\!d\bar t \,[\epsilon-\mu_\alpha(\bar t ) ] -i\int_0^{t } \!\!d\bar t\ , [ \epsilon-\mu_\alpha(\bar t ) ] \big),\end{aligned}\ ] ] where @xmath184 is along the contour .
furthermore , one can show from eqs .
( [ app - eom - f ] ) and ( [ app - eom - geps ] ) that the solution of eq .
( [ app - eom - f ] ) is given by @xmath185 where the integral runs over the whole contour .
this expression is inserted into eq .
( [ app - eom - g0 ] ) in order to derive a single closed equation for @xmath186 , @xmath187 g_{0,\sigma}(t , t')- \int \!ds \ , g_{0,\sigma}(t , s)\delta_\sigma(s , t ' ) = -\delta_{\mathcal{c}}(t , t').\ ] ] here the sum over bath states has been condensed into the integral over the hybridization function ( [ gamdef ] ) @xmath188 in practice , we determine @xmath189 from eqs .
( [ intgamma ] ) , ( [ geps ] ) and ( [ gamdef ] ) .
equation ( [ app - eom - g0 - 2 ] ) is an integrodifferential equation on the contour @xmath182 .
its solution is equivalent to a boundary value problem for the imaginary time component of the green function and initial value problems for the components involving real time - arguments .
the equation is solved numerically , using langreth rules for the decoupling of real and imaginary time components [ see ref . ] . the correlator ( [ defa0 ] ) which enters eq .
( [ eqa ] ) for the current is by definition given by @xmath190 using eqs .
( [ f ] ) and ( [ deltaalpha ] ) , this function can be obtained from the contour integral @xmath191 note that the equations of motion ( [ app - eom - f ] ) and ( [ app - eom - geps ] ) also hold in the interacting case , with @xmath186 replaced by @xmath192 .
hence eqs .
( [ a1 ] ) and ( [ afinal ] ) are still valid in the interacting case with the same replacement , and the interacting current can be obtained directly from the interacting dot green function .
this procedure is however equivalent to the approach which is used in the present paper , where the green function is not measured and the current is obtained instead from eq .
( [ eqa ] ) .
d. goldhaber - gordon , h. shtrikman , d. mahalu , d. abusch - magder , u. meirav and m. a. kastner , nature * 391 * , 156 ( 1998 ) .
s. iwai , m. ono , a. maeda , m. matsuzaki , h. kishida , h. okamoto , and y. tokura , phys .
lett . * 91 * , 057401 ( 2003 ) .
s. iwai and h. okamoto , j. phys .
* 75 * , 011007 ( 2006 ) .
u. schneider _
et al . _ ,
science * 322 * , 1520 ( 2008 ) .
s. m. a. rombouts , k. heyde , and n. jachowicz , phys .
rev . lett . * 82 * , 4155 ( 1999 ) .
a. n. rubtsov , v. v. savkin and a. i. lichtenstein , phys . rev . *
b 72 * , 035122 ( 2005 ) .
p. werner , a. comanac , l. de medici , m. troyer and a. j. millis , phys .
* 97 * , 076405 ( 2006 ) .
e. gull , p. werner , o. parcollet and m. troyer , europhys .
* 82 * 57003 ( 2008 ) .
l. mhlbacher and e. rabani , phys .
lett . * 100 * , 176403 ( 2008 ) .
t. schmidt , p. werner , l. mhlbacher , and a. komnik , phys .
b * 78 * , 235110 ( 2008 ) .
p. werner , t. oka , and a. j. millis , phys .
b 79 , 035320 ( 2009 ) .
m. schiro and m. fabrizio , phys .
b * 79 * , 153302 ( 2009 ) .
e. gull , p. werner , a. j. millis , and m. troyer , phys .
b * 76 * , 235123 ( 2007 ) . t. k. ng and p. a. lee , phys . rev
. lett . * 61 * , 1768 , ( 1988 ) .
l. i. glazman and m. e. raikh , zh .
* 47 * , 378 ( 1988 ) [ jetp lett . * 47 * ( 1988 ) 452 ] .
y. meir , n. s. wingreen and p. a. lee , phys .
lett . * 70 * , 2601 ( 1993 ) .
j. knig , j. schmid , h. schoeller , and g. schn , phys . rev .
b * 54 * , 16820 ( 1996 ) .
t. fujii and k. ueda , phys .
b * 68 * , 155310 ( 2003 ) . a .-
jauho , n. s. wingreen , and y. meir , phys .
b * 50 * , 5528 ( 1994 ) .
m. eckstein , m. kollar , and p. werner , phys .
rev . lett . * 103 * , 056403 ( 2009 ) ; arxiv:0910.5674 .
r. van leeuwen , n. e. dahlen , g. stefanucci , c .- o .
almbladh and u. von barth , arxiv : cond - mat/0506130v1 ( published in _ time - dependent density functional theory _ , lecture notes in physics * 706 * , springer 2006 ) . | we present optimized implementations of the weak - coupling continuous - time monte carlo method defined for nonequilibrium problems on the keldysh contour .
we describe and compare two methods of preparing the system before beginning the real - time calculation : the interaction quench " and the voltage quench " , which are found to be suitable for large and small voltage biasses , respectively .
we also discuss technical optimizations which increase the efficiency of the real - time measurements .
the methods allow the accurate simulation of transport through quantum dots over wider interaction ranges and longer times than have heretofore been possible .
the current - voltage characteristics of the particle - hole symmetric anderson impurity model is presented for interactions @xmath0 up to @xmath1 times the intrinsic level width @xmath2 .
we compare the monte carlo results to fourth order perturbation theory , finding that perturbation theory begins to fail at @xmath3 . within the parameter range studied we find no evidence for a splitting of the kondo resonance due to the applied voltage .
the interplay of voltage and temperature and the coulomb blockade conductance regime are studied . |
our sun is surrounded by a disk of warm ( @xmath0150 k ) `` zodiacal '' dust that radiates most of its thermal energy at 1030 microns .
this zodiacal dust is produced largely in the inner part of the solar system by collisions in the asteroid belt ( dermott et al .
1992 ) and cometary outgassing ( liou and zook 1996 ) .
zodiacal dust is interesting as a general feature of planetary systems , and as an indicator of the presence of larger bodies which supply it ; dust orbiting a few au from a star is quickly removed as it loses angular momentum to poynting - robertson drag ( robertson 1937 ) . understanding the extra - solar analogs of zodiacal dust
may also be crucial in the search for extra - solar planets ( beichman et al .
1996 ) since exo - zodiacal dust in a planetary system could easily outshine the planets and make them much harder to detect .
the best current upper limits for the existence of exo - zodiacal dust disks come from iras measurements of 12 and 25 micron excesses above photospheric emission . seen from a nearby star , solar system zodiacal dust would create only a @xmath1 excess over the sun s photospheric emission at 20 microns .
iras measurements , however , have typical measurement errors of 5 percent ( moshir et al .
1992 ) and display systematic offsets of a similar magnitude when they are compared to other photometry ( cohen et al .
if there were a solar - type zodiacal disk with 1000 times the density of the disk around the sun around tau ceti , the nearest g star , the excess infrared emisison would barely exceed the formal 68% confidence intervals of the iras photometry .
moreover , all photometric detection schemes of this sort are limited by how accurately the star s mid - infrared photospheric emission is known . for
farther , fainter stars than tau ceti , inferring the presence of dust from the iras data becomes still harder .
the detection of faint exo - zodiacal - dust emission is more feasible if one can resolve the dust emitting region .
the high resolution and dynamic range needed for these observations will generally require large interferometers like the keck interferometer , the large binocular telescope , and the very large telescope interferometer .
but it is already possible to resolve the zodiacal dust mid - infrared emitting regions of the nearest stars . a 10-meter telescope operating at 12 microns has a diffraction - limited resolution of 0.25 arc seconds , corresponding , for example , to a transverse distance of 2 au at 8 parsecs .
we have begun a search for zodiacal dust around the nearest stars using the mid - infrared imaging capabilities of the long wavelength spectrometer ( lws ) ( jones & peutter 1993 ) on the w. m. keck telescope .
the large aperture of the telescope allows us to make spatially resolved images of the zodiacal dust 11.6 micron emitting region around the stars so that we can look for dust emission above the wings of the point - spread function ( psf ) rather than as a tiny photometric excess against the photosphere .
we present here the results of two nights of observations , and compare them with a simple model of exo - zodiacal thermal emission to place upper limits on the amount of dust present in the systems we observed .
we observed six nearby stars with lws on the w. m. keck telescope on august 3rd and 4th , 1996 using standard mid - infrared imaging techniques .
the target stars were the nearest a
k main - sequence stars observable from mauna kea on those dates . with the object on - axis
, we took a series of frames lasting 0.8 ms each , chopping the secondary mirror between the object and blank sky 8 arcseconds to the north at a frequency of 10 hz .
then we nodded the primary mirror for the next series of frames so that the sky was on - axis and the object off - axis .
we repeated this process for 3 nods over a period of 5 minutes , for an on - source integration time of 1.1 minutes , and a typical noise of 2 mjy in one 0.11 by 0.11 arcsecond pixel due to the thermal background .
the seeing was poor both nights , up to 2 arc seconds in the visible . to measure the atmosphere - telescope transfer function
, we made similar observations of seven distant , luminous calibrator stars near our targets on the sky , alternating between target and calibrator every 510 minutes .
we increased our frame - rate for the second night of observations so that we could compensate for the seeing using speckle analysis .
figure 1 shows a cut through a single 84 ms exposure of altair on august 4 , compared to an airy function representing the diffraction - limited psf of a filled 10-meter aperture at 11.6 microns .
the cores of the images are diffraction - limited , but the wings are sensitive to the instantaneous seeing , making speckle analysis necessary .
table 1 provides a summary of our observations .
we flat - fielded the images by comparing the response of each pixel to the response of a reference pixel near the center of the detector .
first we plotted the data number ( dn ) recorded by a given pixel against the dn in the reference pixel for all the frames in each run .
since the response of each pixel is approximately linear over the dynamic range of our observations and most of the signal is sky background , which varies with time but is uniform across the chip , the plotted points for each pixel describe a straight line ; if all the pixels had the same response , the slope of each line would equal 1 .
we divided each pixel s dn by the actual slope of its response curve relative to the reference pixel , effectively matching all pixels to the reference pixel .
we then interpolated over bad pixels , frame by frame . to compensate for the differences in the thermal background between the two nod positions
, we averaged together all the on - axis sky frames to measure the on - axis thermal background and subtracted this average from all of the on - axis frames both object and sky .
we used the same procedure to correct the off - axis frames .
next , we chose subframes of 32 by 32 pixels on each image , centered on the star ( or for sky frames , the location of the star in an adjacent object frame ) , and processed these according to classical speckle analysis ( labeyrie 1970 ) .
we fourier transformed them , and summed the power spectra , yielding a sky power spectrum and an object power spectrum for each series .
then we azimuthally averaged the power spectra in the u - v plane that is , we averaged over all the frequency vectors of a given magnitude , @xmath2 .
this azimuthal averaging corrects for the rotation of the focal plane of the alt - az - mounted keck telescope with respect to the sky .
we then subtracted from every object power spectrum the corresponding sky power spectrum and divided each corrected target power spectrum by the corrected power spectrum of a calibrator star observed in the same manner as the target star immediately before or after the target star .
figure 2 shows an azimuthally - averaged power spectrum of altair and the corresponding sky power spectrum , compared with a power spectrum of calibrator gamma aquila and its corresponding sky power .
we then averaged all the calibrated power spectra for a given target . if the object and calibrator are both unresolved , the average calibrated power spectrum should be the power spectrum of the delta function : a constant .
we found that the pixels along the u and v axes of the power spectra were often contaminated by noise artifacts from the detector amplifiers , so we masked them out .
figures 3 and 4 show the calibrated azimuthally - averaged power spectra for our target stars . to compare different power spectra from the same target , we normalized each azimuthally - averaged power spectrum so that the geometric mean of the first 10 data points in each spectrum equals 1 . for altair and 61 cygni a and b we had more than three pairs of target and calibrator observations , i.e. calibrated power spectra ,
so we show the average of all the spectra and error bars representing the 68% confidence interval for each datum , estimated from the variation among the individual power spectra .
the error is primarily due to differences in the atmosphere - telescope transfer function between object and calibrator .
none of the calibrated power spectra deviate from a straight line by more than a typical error ; all the targets are unresolved to the accuracy of a our measurements .
to interpret our observations we compared them to models of the ir emission from the solar zodiacal cloud .
we constructed a model for exo - zodiacal emission based on the smooth component of the kelsall et al .
( 1998 ) model of the solar system zodiacal cloud as seen by cobe / dirbe , with emissivity @xmath3 and a temperature @xmath4 , where @xmath5 is the distance from the star in au , and @xmath6 is the luminosity of the star in terms of @xmath7 . for a dust cloud
consisting entirely of a single kind of dust particle of a given size and albedo , the @xmath6 exponent in the expression for the temperature is simply @xmath8 times the @xmath5 exponent ( backman & paresce 1993 ) .
the physics of the innermost part of the solar zodiacal dust is complicated ( see mann & macqueen 1993 ) , but our results are not sensitive to the details , because the hottest dust is too close to the star for us to resolve .
we assume that the dust sublimates at a temperature of 1500 k , and allow this assumption to define the inner radius of the disk .
we set the outer radius of the model to 3 au , the heliocentric distance of the inner edge of our own main asteroid belt .
our conclusions are not sensitive to this assumption ; decreasing the outer radius to 2 au or increasing it to infinity makes a negligible difference in the visibility of the model , even for a stars . the assumed surface density profile , however , does make a difference .
a collisionless cloud of dust in approximately circular orbits spiraling into a star due to poynting - robertson drag that is steadily replenished at its outer edge attains an equilibrium surface density that is independent of radius ( wyatt and whipple 1950 , briggs 1962 ) .
models that fit data from the helios space probes ( leinert et al 1981 ) , the fit by kelsall et al .
( 1998 ) to the cobe / dirbe measurements and good s ( 1997 ) revised fit to the iras data all have surface densities that go roughly as @xmath9 .
this distribution appears to continue all the way in to the solar corona ( macqueen & greely 1995 ) .
we find that in general , if we assume an @xmath10 surface density profile , our upper limit for the 1 au density of a given disk scales roughly as @xmath11 ; disks with more dust towards the outer edge of the 11.6 micron emitting region are easier to resolve .
likewise , the assumed temperature profile strongly affects our upper limits .
unfortunately , we know little about the temperature profile of the solar zodiacal cloud .
cobe / dirbe and iras only probed the dust thermal emission near 1 au , and helios measured the solar system cloud in scattered light , which does not indicate the dust temperature .
we found that a dust cloud model with the iras temperature profile ( @xmath12 ) was much easier to resolve than the model based on dirbe measurements that we present here , especially for g and k stars . to compare the models with the observations , we synthesized high resolution images of the model disks at an inclination of 30 degrees .
we calculated the ir flux of the stars from the blackbody function , and obtained the parallaxes of the stars from the hipparcos catalog ( esa 1997 ) .
we inferred stellar radii and effective temperatures for each star from the literature and checked them by comparing the blackbody fluxes to spectral energy distributions based on photometry from the simbad database ( egret et al .
for altair and vega , we use the interferometrically measured angular diameters ( 1974 ) ( they are 2.98 + /-0.14 mas and 3.24 mas ) .
stellar fluxes typically disagree with fitted blackbody curves by @xmath13 in the mid - infrared ( engelke 1990 ) , but our method does not require precise photometry , and the blackbody numbers suffice for determining conservative upper limits .
we computed the power spectra of the images , and normalized them just like the observed power spectra . in figures 3 and 4 , the azimuthally - averaged power spectra for our target stars are compared to the extrapolated cobe / dirbe model at a range of model surface densities .
disks with masses as high as @xmath14 times the mass of the solar disk will suffer collisional depletion in their inner regions , so they are unlikely to have the same structure as the solar disk . by neglecting this effect
we are being conservative in our mass limits .
the density of the densest model disk consistent with the data in each case is listed in table 1 .
our best upper limit is for altair ( spectral type a7 , distance 5.1 pc ) ; with 11 pairs of object and calibrator observations we were able to rule out a solar - type disk a few times @xmath14 as dense as our zodiacal cloud .
such a disk would have been marginally detectable by iras as a photometric excess .
iras detected no infrared excess in vega s spectral energy distribution at 12 microns , with an uncertainty of 0.8 jy .
this may be due to a central void in the disk interior to about 26 au ( backman & paresce 1993 ) .
aumann et al ( 1984 ) suggested that vega ( a0 , 7.8 pc ) could have a hot grain component ( 500 k ) with up to @xmath15 of the grain area of the observed component and not violate this limit .
the apparent upward trend in the visibility data may be a symptom of resolved flux in the calibrator stars .
we have only 3 object / calibrator pairs for vega , not enough to test this hypothesis .
our upper limit is a solar - type disk with approximately @xmath16 times the density of the solar disk .
this disk would have a @xmath17500 k emitting area of @xmath18 , about @xmath15 of the grain area of the observed component .
though 61 cygni is close to the galactic plane and surrounded by cool cirrus emission , backman , gillett and low ( 1986 ) identified an iras point source with this binary system and deduced a far - infrared excess not unlike vega s .
the color temperature of the excess suggests the presence of dust at distances @xmath19 au from either star .
however these stars are dim ( spectral types k5 and k7 ) and the region of the disk hot enough to emit strongly at 11.6 microns is close to the star and difficult to resolve ; we could not detect a solar - type dust disk around either of these objects at any density , assuming the cobe / dirbe model , or unless it had @xmath20 times the density of the solar disk , assuming the iras model .
70 oph is a binary ( types k0 and k4 ) with a separation of 24 pixels ( 2.6 arcsec ) .
we were able to assemble a power spectrum for b from 9 object / calibrator pairs , but the image of a fell on a part of the lws chip that suffered from many bad pixels and was unusable . the image of a may also have been distorted by off - axis effects .
70 oph b , like 61 cygni a and b , is dim , making any dust around it cool and hard to detect at 11.6 microns . iras could have barely detected a disk with @xmath22 times the emitting area of the solar disk around tau ceti ( g8 , 3.6 pc ) , the nearest g star .
we have only three object / calibrator pairs for this object , not enough data to improve on this limit .
we are grateful to dana backman , alycia weinberger , keith matthews and eric gaidos for helpful discussions , and to keith matthews and shri kulkarni for assistance with the observations .
this research has made use of the simbad database , operated at cds , strasbourg , france .
the observations reported here were obtained at the w. m. keck observatory , which is operated by the california association for research in astronomy , a scientific partnership among california institute of technology , the university of california , and the national aeronautics and space administration .
it was made possible by the generous financial support of the w. m. keck foundation .
aumann , h. h. 1984 , apj , 278 , l23 backman , d. e. , & paresce , f. 1993 , `` main sequence stars with circumstellar solid material : the vega phenomenon '' , _ protostars and planets iii _ ( eds levy e. h. & lunine , j. i. ) briggs , r. e. 1962 , aj , 67 , 710 cohen , m. , witteborn , f. c. , carbon , d. f. , davies , j. k. , wooden , d. h. , bergman , j. d. 1996 , aj , 112 , 2274 dermott , s.f . , gomes , r.s .
, dorda , d.d . ,
gustafson , b. .
, jayaraman , s. , xu , y.l . , & nicholson , p.d .
1992 , in ferraz - mello s. ed . ,
chaos , resonance and collective dynamical phenomena in the solar system .
kluwer , dordrect , p.333 egret , d. , wenger , m. , dubois , p. ( 1991 ) , in _ databases & on line data in astronomy _ , albrecht & egret ( eds . ) , kluwer acad .
publ . , pp . 7988 .
labeyrie , a. 1970 , a&a , 6 , 85 leinert , c. , richter , i. , pitz , e. , & planck , b. 1981 , a&a , 103 , 177 liou , j.c . , &
zook , h.a .
1996 , icarus , 123 , 491 macqueen , r. m. , & greeley , b. w. 1995 , apj , 440 , 361 mann , i. , & macqueen , r. m. 1993 , a&a , 275 , 293 moshir , m. , et al .
1992 , explanatory supplement to the iras faint source survey , version 2 , jpl d-10015 8/92 ( pasadena : jpl ) ccccccc august 3 & vega & r lyr & 800 & 90 & 2 & @xmath23 & & @xmath24 lyr&800 & 90 & 1 & & 61 cyg a & @xmath25 cyg & 800 & 90 & 4 & & 61 cyg b & @xmath25 cyg & 800 & 90 & 5 & & @xmath21 cet & @xmath26 cet & 800 & 90 & 2 & & & & 210 & 348 & 1 & august 4 & 70 oph b & @xmath27 oph & 84 & 864 & 5 & & & 74 oph & 84 & 864 & 4 & & altair & @xmath28 aql & 84 & 864 & 6 & @xmath29 & & @xmath27 aql & 84 & 864 & 5 & | we have begun an observational program to search nearby stars for dust disks that are analogous to the disk of zodiacal dust that fills the interior of our solar system .
we imaged six nearby main - sequence stars with the keck telescope at 11.6 microns , correcting for atmosphere - induced wavefront aberrations and deconvolving the point spread function via classical speckle analysis .
we compare our data to a simple model of the zodiacal dust in our own system based on cobe / dirbe observations ( kelsall et al .
1998 ) and place upper limits on the density of exo - zodiacal dust in these systems . |
the higgs sector in supersymmetric extensions of the standard model contains charged higgs - bosons as well as additional neutral higgs - bosons @xcite . among the latter is a pseudoscalar particle usually denoted @xmath0 . in this paper
, we calculate the production cross section for the @xmath0 in the process @xmath1 .
contributions to this process arise from triangle and box diagrams .
the triangle contributions consist of diagrams in which the @xmath0 and photon are on - shell external particles and the remaining particle is a virtual photon or @xmath4 in the @xmath5-channel . since @xmath6 is in the physical region , the photon pole contribution dominates the @xmath4 pole contribution in this set of diagrams @xcite .
moreover , because of the off - diagonal structure of the @xmath0 couplings to other bosons , the particles in the loop are either quarks , leptons or charginos . here
, we present the top quark , bottom quark , tau lepton and the two chargino contributions to the photon pole amplitude .
the box diagrams have a more complex particle structure , with leptons , charginos , neutralinos and scalar leptons in the loops . like the @xmath4 pole ,
these diagrams are non - singular at @xmath6 , and should not contribute a sizable correction to the photon pole terms .
they are not included in the present calculation .
the amplitude for the production of an @xmath0 of momentum @xmath7 and an @xmath8 of momentum @xmath9 in the collision of an @xmath8 of momentum @xmath10 and a @xmath11 of momentum @xmath12 and polarization @xmath13 by the exchange of a @xmath11 in the @xmath5-channel is @xmath14 where @xmath15 , and @xmath16 \nonumber & & \;\;-(-1)^2m_{\tau}^2\tan\beta\,c_0(t , m_a^2,m_{\tau}^2 ) + 2m_wm_1g_{11}c_0(t , m_a^2,m_1 ^ 2 ) \\ [ 6 pt]\nonumber & & \;\;+\;2m_wm_2g_{22}c_0(t , m_a^2,m_2 ^ 2)\biggr]\,,\end{aligned}\ ] ] here , @xmath17 and @xmath18 are the top and bottom quark masses , @xmath19 is the tau lepton mass , @xmath20 and @xmath21 are the chargino masses , @xmath22 is the @xmath23 mass and @xmath24 is a ratio of vacuum expectation values @xcite .
the chargino coupling constants @xmath25 and @xmath26 depend on the elements of two @xmath27 unitary matrices @xmath28 and @xmath29 which diagonalize the chargino mass matrix @xmath30 , where @xcite @xmath31 and are chosen to ensure that @xmath20 and @xmath21 are positive . for illustrative purposes , we assume that the symmetry breaking parameters @xmath32 and @xmath33 are real and consider two cases : @xmath34 and @xmath35 .
the couplings in these cases are @xmath36 for @xmath34 , and @xmath37 for @xmath35 . notice that , these couplings are symmetric in @xmath20 , @xmath21 and , unlike the @xmath0-top coupling
, there is no enhancement factor of @xmath38 @xcite .
we take @xmath39 . due to the reality of @xmath32 and @xmath33 , @xmath20 and @xmath21 in eqs.([g1],[g2
] ) are subject to certain constraints discussed below @xcite .
the scalar function @xmath40 is @xcite @xmath41 since one of the external particles is a photon , this function can be expressed in terms of inverse trignometric or hyperbolic functions @xcite as @xmath42 where @xmath43 & = & \;\left\ { \begin{array}{lll } 2\left(\sinh^{-1}(\sqrt{-\frac{\mbox{\rule[-3pt]{0pt}{10pt}}\displaystyle\beta } { \mbox{\rule{0pt}{11pt}}\displaystyle 4}}\ , ) \right)^{\!2 } & & \beta\leq 0 \\ [ 8 pt ] -2\left(\sin^{-1}(\sqrt{\frac{\mbox{\rule[-3pt]{0pt}{10pt}}\displaystyle\beta } { \mbox{\rule{0pt}{11pt}}\displaystyle 4}}\ , ) \right)^{\!2 } & & 0\leq\beta\leq 4 \\ [ 8 pt ] 2\left(\cosh^{-1}(\sqrt{\frac{\mbox{\rule[-3pt]{0pt}{10pt}}\displaystyle\beta } { \mbox{\rule{0pt}{11pt}}\displaystyle 4}}\ , ) \right)^{\!2 } - \frac{\mbox{\rule[-3pt]{0pt}{10pt}}\displaystyle\pi^2 } { \mbox{\rule{0pt}{11pt}}\displaystyle 2 } - 2i\pi\cosh ^{-1}(\sqrt{\frac{\mbox{\rule[-3pt]{0pt}{10pt}}\displaystyle\beta } { \mbox{\rule{0pt}{11pt}}\displaystyle 4 } } ) & & \beta\geq 4 \end{array } \right.\,.\end{aligned}\ ] ] the cross section is given by @xmath44 and we have @xmath45 where @xmath46 and @xmath47 .
the presence of the @xmath48 in eq.([msq ] ) means it is necessary to introduce a cutoff in the calculation of the total cross section .
one approach to obtaining a finite cross section is use the effective photon or weizscker - williams approximation for the exchanged photon @xcite .
here , we integrate the exact amplitude and impose an angular cutoff . the expression for the total cross section
is @xmath49 where @xmath50 is an angular cutoff .
we investigated the effect of varying @xmath51 by comparing the standard model cross section with and without the @xmath4 exchange .
for @xmath52 as large as @xmath53 , the @xmath4 contribution is only 3%-4% of the total .
the result scales approximately as the logarithm of @xmath50 , and we use @xmath54 in the figures . to complete the calculation of the cross section for the @xmath55 process , it is necessary to fold the cross section , eq.([sigma ] ) , with the distribution @xmath56 of backscattered photons having momentum fraction @xmath57 @xcite to obtain @xmath58 with @xmath59 . here
, we have taken the usual upper limit on the allowed @xmath57 value , @xmath60 .
this cross section is plotted in fig.(1 ) for @xmath34 and in fig.(2 ) for @xmath35 .
the dashed line in each panel is the contribution from the top and bottom quarks and the tau lepton . for large @xmath24 ,
the tau contribution is important .
this is illustrated in the @xmath61 panel of fig.(1 ) , where the dot - dashed line is the contribution from the top and bottom quarks . in fig.(1 ) , the solid lines are @xmath62 gev and @xmath21 the largest value consistent with the constraint @xmath63 , which is needed to ensure that @xmath32 and @xmath33 are real . similarly , in fig.(2 ) , the solid lines correspond to @xmath62 gev and @xmath21 the largest value consistent with @xmath64 . unlike the @xmath34 case , when @xmath35 it is possible for @xmath20 and @xmath21 to be equal for @xmath65 provided the @xmath66 .
these values of @xmath20 and @xmath21 are within the range of chargino masses found in studies of minimal supersymmetric models @xcite . in most cases
, the inclusion of the chargino contribution leads to a significant increase in the cross section , especially for the larger values of @xmath24 . to assess the observability of this process , we assume that the dominant @xmath67 decay is @xmath68 . for @xmath69 , this ignores some contribution from chargino pair decay , but this is relatively small since in all but one of the examples we consider the lowest chargino mass exceeds @xmath70 gev .
even above the top threshold , @xmath71 decay dominates when @xmath72 @xcite .
the dotted lines in figures ( 1 ) and ( 2 ) are the cross section for the direct production of a background @xmath71 of invariant mass @xmath73 in @xmath74 collisions subject to an angular cut on the @xmath75 and @xmath76 direction relative to the that of the incident photon in the @xmath74 center of mass .
we find that an angular cut of @xmath77 on both the @xmath75 and the @xmath76 reduces the background @xmath71 cross section by about a factor of 10 while leaving the @xmath71 signal from @xmath0 decay essentially unchanged .
more restrictive cuts on the @xmath75 and @xmath76 angles can further suppress the background , but at the expense of a significant decrease in the signal @xcite .
the cut shown appears to be optimal .
we would like to point out that the @xmath55 cross sections calculated here are very likely to be much larger than those of the related process @xmath78 at 500 gev .
we have checked this for the production of the standard model higgs - boson using the complete ( standard model ) calculation of @xmath79 @xcite and the photon pole contribution to @xmath80 . at an @xmath81 center of mass energy of 500 gev
, we find the cross section @xmath82 for the production of a 200 gev @xmath83 is 0.08 fb , whereas @xmath84 fb for the same higgs - boson mass .
this enhancement is implicit in a previous calculation of scalar higgs - boson production @xcite . in ref.@xcite ,
the weizscker - williams approximation is used for the @xmath5 channel photon together with the on - shell @xmath85 amplitude .
this is essentially equivalent to setting @xmath86 in the parentheses of eq.([sigma ] ) and using @xmath87 as the cutoff in the remaining integral @xcite .
our comparison of the approximate results of ref.@xcite with an exact calculation suggests that the weizscker - williams approach tends to overestimate the cross section .
apart from minor variations depending on how the calculation is performed , it is nevertheless true that the @xmath5 channel cross section is substantially larger than its @xmath88 channel counterpart .
the production of @xmath0 has also been investigated in @xmath89 collisions @xcite . in this case , the background arises from the process @xmath90 and it is effectively suppressed by imposing an angular cut . with our choice of chargino masses , a comparison of the @xmath61
cross sections in @xmath74 production and @xmath89 production @xcite reveals a larger signal in the @xmath91 mode .
both channels are likely to be important in searches for the @xmath0 . to the extent that the photon pole contribution can be isolated , this method of searching for the @xmath0 has the advantage that the contributions from supersymmetry are significant and limited to one type supersymmetric particle .
should one observe a cross section larger than any standard model prediction , the case for the presence of chargino contributions is rather strong .
we would like to acknowledge conversations with c .- p .
yuan and x. tata .
this research was supported in part by the national science foundation under grant phy-93 - 07980 and by the united states department of energy under contract de - fg013 - 93er40757 .
99 , j. f. gunion , h. e. haber , g. kane and s. dawson , addision wesley publishing company , 1990 .
the same thing can be said about the production of scalar higgs - bosons in the reaction @xmath92 .
j. f. gunion and h. w. haber , nuc . phys . *
b272 * , 1 ( 1986 ) .
h. baer , a. bartl , d. karatas , w. majerotto , and x. tata , int . j. mod
a4 * , 4111 ( 1989 ) appendix b. x. tata and d. a. dicus , phys .
d * 35 * , 2110 ( 1987 ) .
g. t hooft and m. veltman , nuc .
b * 153 * , 365 ( 1979 ) .
a. abbasabadi , d. bowser - chao , d. a. dicus and w. w. repko , phys .
rev d * 52 * , 3919 ( 1995 ) .
o. j. p. boli , m. c. gonzalez - garcia and s. f. novaes , phys .
d * 49 * , 91 ( 1994 ) .
i. f. ginzburg , g. l. kotkin , s. l. panfil , v. g. serbo , and v. i. telnov , nucl .
methods * 219 * , 5 ( 1984 ) .
g. l. kane , c. kolda , l. roszkowski , and j. d. wells , phys .
d * 49 * , 6173 ( 1994 ) .
v. barger , k. chung , r. j. n. phillips , and a. l. strange , phys .
d * 46 * , 4914 ( 1992 ) .
if the @xmath0 mass should turn out to be very close to the @xmath93 mass , there could be a substantial background from @xmath94 .
we have not included this in our background calculation . to see this in detail , compare the simplified eq.([sigma ] ) with eq.(4 ) of ref.@xcite .
j. f. gunion and h. e. haber , phys . rev .
d * 48 * , 5109 ( 1993 ) . | we investigate the production of a pseudoscalar higgs - boson @xmath0 using the reaction @xmath1 at an @xmath2 collider with center of mass energy of 500 gev .
supersymmetric contributions are included and provide a substantial enhancement to the cross section for most values of the symmetry breaking parameters .
we find that , despite the penalty incurred in converting one of the beams into a source of backscattered photons , the @xmath3 process is a promising channel for the detection of the @xmath0 .
= -5pt = -5pt |
there is a wealth of physical phenomena inherent to junctions between superconducting ( s ) and normal ( n ) matter , such as the josephson effect @xcite , quasi - particle tunneling @xcite and andreev reflection ( ar ) @xcite .
traditionally , sn junctions are either based on composites of superconducting and normal - conducting materials , or on constrictions / thickness modulations of superconducting films .
a common feature of these junctions is their fixed character , since they are static constructions that can not be modified any more once they are fabricated .
however , recent works have shown that the highly inhomogeneous fields of ferromagnetic domains can be used to locally suppress superconductivity in thin s - films , resulting in the states of domain - wall superconductivity @xcite ( dws ) or reverse - domain superconductivity @xcite ( rds ) . in these two states ,
the superconductor can be seen as a network of sn junctions , which can exhibit the same flexibility as the underlying magnetic domains . in this article
, we demonstrate that tunable sn junctions can be created in a controlled way by using the highly inhomogeneous fields of ferromagnetic domains .
in particular , we show that in such junctions , the interfaces between the superconducting and the normal parts are highly transparent for incident electrons , which is a consequence of creating superconducting and normal state regions inside the _ same material_. in order to experimentally investigate sn interfaces that are induced by stray magnetic fields , a specially designed superconductor / ferromagnet ( s / f ) hybrid @xcite system was needed , exhibiting the following two qualities : ( i ) the opportunity to specifically realize superconductivity either above the magnetic domain walls ( dws ) or above the reverse domains ( rds ) of the substrate .
( ii ) transport currents had to cross effectively the interfaces between the induced superconducting and normal - state regions .
the preparation of such system is challenging as several strict requirements need to be fulfilled .
first of all , formation of magnetic stripe domains in the template is desirable @xcite .
alignment of a transport bridge perpendicular to such domains guarantees a bias current to cross them successively .
second , the magnetic domain pattern of the template must not change significantly when subjected to external fields , required for setting up the different states ( e.g. rds ) in the s - layer .
finally , the out - of plane component of the stray field above magnetic domains has to reach the upper critical field of the superconductor @xcite .
thereby , realization of dws is possible down to temperatures well below @xmath0 .
( color online ) ( a ) photograph of the as - grown barium hexaferrite single crystal .
( b ) sketch of the crystal after slicing under a cut angle @xmath1 with respect to its c - axis .
( c ) magnetic force microscopical image of the magnetic domains at the cut surface of a slice of the bafe@xmath2o@xmath3 single crystal .
( d ) sketch of a transport bridge processed perpendicular to the c - axis on the cut surface of a slice of the crystal.,width=325 ]
accounting for the above requirements , s / f hybrid systems were prepared by slicing a single crystal of barium hexaferrite ( bafe@xmath2o@xmath3 ) under a small tilt to its c - axis ( figure [ fig : sample preparation](b ) ) , and then processing superconducting aluminium bridges of 50 nm thickness on the cut surfaces perpendicular to the c - axis ( figure [ fig : sample preparation](d ) ) .
the superconducting and ferromagnetic components were electrically isolated by 5 nm sio@xmath4 in order to prevent any proximity effect .
the ferromagnetic crystal ( figure [ fig : sample preparation](a ) ) was grown from a sodium carbonate flux , following a recipe after @xcite . when cut along the proper crystallographic axis , single crystals of bafe@xmath2o@xmath3 exhibit a one - dimensional stripe - type domain structure ( figure [ fig : sample preparation](c ) ) with dominant in - plane magnetization and relatively small
out - of - plane component @xmath5 @xcite . to demonstrate that these magnetic domains do not change significantly in perpendicular external magnetic fields @xmath6 mt , the magnetization @xmath7 of one slice of the single crystal was measured with a vibrating sample magnetrometer as a function of @xmath8 ( see figure [ fig : m(h ) ] ) .
apparently , the magnetization of the ferromagnet depends almost linearly on the perpendicular applied magnetic field and saturates at @xmath9 1.7 t. from the slope @xmath10 am@xmath11t@xmath11 , one can indeed expect only minor changes of the domain structure for @xmath6 mt , since the corresponding variation of the magnetic moment is less than 7@xmath12 of the saturated magnetization ( @xmath13 am@xmath11t@xmath11 ) .
( color online ) the magnetization @xmath7 of a slice of the bafe@xmath2o@xmath3 single crystal at 5 k as a function of the external magnetic field @xmath14 normal to the cut surface .
the insert shows a magnification of the curve for @xmath15 150 mt.,width=253 ] furthermore , the influence of the external magnetic field on the size and position of the magnetic domains was studied at low - temperatures ( 77 k ) with a scanning hall - probe microscope @xcite .
as it is shown in figure [ fig : domain width ] , the width @xmath16 of the parallel domains increases linearly for @xmath17 mt with a rate of approximately 43 nm / mt . in accordance with the very small coercivity of these ferromagnets ( see figure [ fig : m(h ) ] ) , the observed domain walls returned to their initial positions within the experimental resolution of 1 @xmath18 m , each time @xmath8 was reduced to zero .
( color online ) the increase of the width @xmath16 of a parallel domain ( the bright areas in the inserts ) in the ferromagnetic substrate as a function of @xmath8 .
the shown data is the average of the results obtained for an increasing and a decreasing external magnetic field .
the inserts show two examples of the position of the same domain - wall at 0 mt and 150 mt , respectively.,width=274 ] in order to show that the prepared s / f hybrid systems are suitable for the study of stray field induced sn interfaces , figure [ fig : rds](b ) shows the normalized resistance @xmath19 ( @xmath20 being the normal state resistance ) of two transport bridges as a function of @xmath8 .
both curves were measured at 340 mk , i.e. well below the critical temperature @xmath21 k of the used aluminum ( see the insert of figure [ fig : rds ] ) . from the corresponding atomic and magnetic force microscopical images ( afm and mfm , respectively ) of figure [ fig : rds](a ) , it can be seen that the measured parts of the bridges lay entirely above magnetic domains of opposite magnetization .
the difference in the mfm signal above the two kinds of domains indicates a non - zero out - of plane component @xmath22 of the stray magnetic field . note that @xmath23 above the wide domains results directly from a finite cutting angle @xmath1 to the c - axis of the crystal .
therefore , by choosing @xmath1 , the strength of @xmath22 can be adapted to match the critical fields of the superconductor . for the present case , it was found that aluminium as a superconductor and @xmath24 are a good match . for the case of the bridge above the bright domain ( left panel of figure [ fig : rds](a ) ) , @xmath25 drops to zero around -53 mt ( see the red curve with circles ) . in a symmetric manner , @xmath25 of the bridge above
the dark domain shows a similar behavior around + 53 mt ( see the blue curve with diamonds ) .
these observations prove the possibility to realize the state of rds by applying compensation fields of @xmath26 mt to the designed al / bafe@xmath2o@xmath3 hybrids .
moreover , it becomes clear that in the state of rds at 340 mk , the superconducting order parameter is completely suppressed above the corresponding parallel domains ( i.e. above magnetic domains with magnetization in the same direction as @xmath14 ) .
( color online ) ( a ) afm - images of two transport bridges with their underlying magnetic domains ( mfm - images ) .
the mfm - images are vertically extended to illustrate the domains .
arrows indicate the z - component @xmath27 of the magnetization of the template .
( b ) the normalized resistance of the two bridges , measured at 340 mk with a bias current of 10 @xmath18a as a function of @xmath8 .
the insert shows the superconducting transition of a reference al - film on a si - substrate as a function of temperature ( critical current density @xmath28 a / m@xmath29 at 340 mk).,width=325 ] next , a long transport bridge of @xmath30 @xmath31 was investigated , which due to its relatively large size had to cross several magnetic domain walls of the substrate .
inspection of the sample with a magnetic force microscope revealed indeed nine domain walls underneath the bridge ( figure [ fig : fig_1](a ) ) .
figure [ fig : fig_2 ] displays the normalized dc - resistance of the bridge well below @xmath0 , measured as a function of bias current @xmath32 and @xmath8 .
as can be expected from the above presented results , two pronounced minima in resistivity are seen around @xmath33 mt , indicating that stray fields above magnetic domains are compensated by @xmath8 .
application of these compensation fields thus induces the rds state in the s / f hybrid system .
furthermore , a second key feature can be seen in figure [ fig : fig_2 ] .
while beyond the compensation fields the resistance quickly rises towards its value in the normal state , parts of the bridge remain superconducting when subjected to external fields lower than the compensation fields . particularly , in the case of zero applied field , when superconductivity is likewise suppressed above domains of opposite magnetization , the reduced resistance is a clear fingerprint of dws .
moreover , the insert in figure [ fig : fig_2 ] shows the transitions of the bridge from the normal state to the states of dws and rds ( at 0 mt and 53 mt , respectively ) as a function of temperature . the significant difference @xmath34 of the onsets of the transitions reflects the confinement of the superconducting order parameter above wide magnetic domains ( rds ) and narrow domain - walls ( dws ) @xcite .
( color online ) ( a ) , the transport bridge ( optical image shown in three dimensions ) with its underlying magnetic domains ( mfm image ) .
the mfm image is vertically extended to illustrate the domains .
arrows indicate the z - component @xmath27 of the magnetization of the template .
( b ) , a detailed mfm - image of a typical domain wall in the substrate.,width=325 ] ( color online ) the normalized resistance of the transport bridge of figure [ fig : fig_1](a ) at 340 mk , as a function of bias current @xmath32 and external field @xmath35 ( @xmath36 ) .
regions where rds and dws occur are indicated .
the insert shows the resistive transitions ( @xmath32=10 @xmath18a ) from the normal state to the states of dws ( blue curve with diamonds ) and rds ( red curve with circles ) .
the points corresponding to @xmath37=340 mk are indicated in the main panel.,width=325 ] above , the occurrence of the minima in figure [ fig : fig_2](a ) has been discussed , along with the reduction of resistance at zero applied field .
however , surprising are the values of the resistance reached at these points : as can be seen from figure [ fig : fig_1](a ) , approximately half of the area of the bridge is covered by each kind of domains .
nucleation of superconductivity above one type of domains should thus cause the bridge to loose roughly half of its resistance in the normal state .
by contrast , for compensation fields of both polarities , only half of the expected resistance is seen .
a similar observation can be made at @xmath38 when superconductivity survives above domain walls only . in that case , the drop in resistance is , a priori , expected to be equal to the ratio between the width of magnetic domains and domain walls .
but from detailed mfm studies it becomes clear that all changes in stray fields are confined to approximately 1 @xmath39 around domain walls ( figure [ fig : fig_1](b ) ) , whereas the domains are typically 25 @xmath39 wide ( figure [ fig : fig_1](a ) ) .
therefore , in absence of external fields , the observed reduction of resistance by @xmath40% is surprising . in order to investigate these remarkable features in more detail ,
the differential resistance @xmath41 of the transport bridge was measured as a function of bias current and temperature in both states , rds and dws .
simultaneously , the voltage drop @xmath42 over the bridge was also detected .
measurements were carried out via standard lock - in techniques at a frequency of 33 hz and an ac - modulation current of 2 @xmath18a .
the normalized differential conductance @xmath43 is shown in figure [ fig : fig_3 ] as a function of voltage for @xmath44 mt ( rds ) .
in that diagram , results are shown twice for clarity : the left 2d - panel displays a few conductance curves that are vertically shifted , whereas all obtained curves are given in a 3d - representation at the right . here , at lowest temperatures
, @xmath45 is sharply peaked at zero voltage , declining symmetrically to its minima at @xmath46 before recovering to its normal value at higher voltages .
together with some smaller local minima , these features gradually collapse with increasing temperature .
in order to interpret the conductance spectra of figure [ fig : fig_3 ] , several aspects must be taken into account : \(i ) the whole transport bridge is in the normal state for @xmath472 mv even at the lowest temperature ( 340 mk ) .
the reason for this is that the critical current density @xmath48 is exceeded due to the low resistance of the bridge .
therefore , in a certain low - voltage region where @xmath49 , a higher value for @xmath45 is expected , since the parts of the bridge above reverse - domains ( rd ) are superconducting .
\(ii ) the normalized differential conductance reaches 2.8 at 340 mk and zero voltage . assuming that this increase of @xmath50 was solely caused by the n @xmath51 s transition mentioned under point ( i )
, approximately 64% of the transport bridge had to become superconducting .
however , the hybrid system behaves similar for both polarities of @xmath52 ( see figure [ fig : fig_2 ] ) , meaning that an unequal distribution of parallel and reverse domains can not be the reason for the high conductances observed at positive _ and _ negative compensation fields .
moreover , as discussed above , the external field increases the width of the parallel domains by 43 nm / mt ( see figure [ fig : domain width ] ) .
accordingly , at 53 mt , a normalized conductance of only 1.8 instead of 2 should be expected , provided that parallel and reverse domains are equally distributed at @xmath53 . finally , the characteristic length @xmath54 , over which the cooper pair amplitude decays exponentially with the distance from an sn interface , is in the present case of the order of 400 nm .
m@xmath29/s , with the fermi velocity @xmath55 m / s @xcite and the electron mean free path @xmath56 nm ( @xmath57 and @xmath58 are electronic mass and charge , @xmath59 m@xmath60 is the density of conduction electrons @xcite ) .
the residual resistance @xmath61 was estimated according to @xmath62 , with the resistivity @xmath63 @xmath64 m of al at 295 k @xcite . ] due to this proximity effect , the superconducting state extends into the normal regions and vice versa , but the corresponding increase of @xmath45 at @xmath65 is only minor .
( color online ) differential conductance spectra of the transport bridge of figure [ fig : fig_1](a ) at 53 mt in the state of rds .
arrows mark two minima whose positions can be traced . in the left panel ,
curves are shifted for clarity and the vertical scale corresponds to the top curve.,width=325 ] taking account of the above considerations , the observed conductance of 2.8 can not be explained by a corresponding expansion of the superconducting state along the transport bridge .
below the superconducting gap @xmath66 ( i.e. for @xmath67 ) , an excess of the conductance can generally result from andreev reflection processes at sn - interfaces , if the latter are highly transparent for incident electrons . in the present case ,
normal and superconducting states are created _ inside the same material _ and , therefore , the presence of higly transparent sn - interfaces is reasonable .
accordingly , the observed excess of conductance suggests that the mechanism of charge transfer across the sn - interfaces is affected by ar . the theory of blonder , tinkham and klapwijk ( btk )
@xcite describes the effects of ar on the conductance of a single sn junction for the particular case of ballistic transport in the normal - state region .
from that theory it follows that inside the gap , the conductance can be enhanced up to twice its above - gap value . in the present case , as discussed above , the above - gap conductance in the rds - state at 53 mt can be estimated to be 1.8 . therefore , the observed zero - voltage conductance of 2.8 is smaller than twice the above - gap conductance ( @xmath68 ) , meaning that these findings are not in contradiction with the btk - theory .
moreover , the btk - theory predicts for highly transparent interfaces a flat conductance below the gap , which has been verified experimentally with superconducting point contacts ( see for example @xcite ) . by contrast , the @xmath50-curves of figure [ fig : fig_3 ] are sharply peaked at zero voltage .
such anomalies in the conductance spectra in the form of zero - bias peaks have been reported before in systems that deviate from the model of btk , such as for example planar nb / au contacts @xcite , junctions between superconductors and semiconductors @xcite and series of sns - junctions @xcite . in the present case , the used sample differs also significantly from the model system of btk , since the bridge crosses nine domain walls ( see figure [ fig : fig_1](a ) ) , each of them inducing one sn interface . moreover , due to the large size of the domains , the electric transport in the normal - state regions is not ballistic .
a theoretical description of such series of diffusive sns junctions will go beyond the ballistic theories @xcite , and will have to include nonlocal coherent effects in the normal - state regions @xcite .
( iii ) the positions of characteristic minima in the conductance spectra of figure [ fig : fig_3 ] and [ fig : fig_5 ] ( markers ) are compared with @xmath69 ( solid lines ) . for clarity , the results obtained at 0 mt are shifted by -0.2 .,width=325 ] two of the local minima of the conductance spectra of figure [ fig : fig_3 ] , marked as @xmath70 and @xmath71 , can be traced from 340 mk to nearly @xmath0 .
their relative position on the @xmath42-axis was compared to the superconducting gap function @xmath72 of the bcs - theory @xcite , using a value of 423 k for the ratio between cut - off frequency @xmath73 and boltzmann constant @xmath74 @xcite .
a solution @xmath69 of the above equation can be found by iteration , integrating numerically over energies @xmath75 while treating @xmath0 as a fitting parameter .
as illustrated in figure [ fig : fig_4 ] ( upper curve ) , @xmath76 follow strictly the superconducting gap @xmath66 in temperature . for the ideal case of a single ballistic sns junction , it is known that multiple andreev reflection ( mar ) leads to minima in the conductance curves at voltages smaller than the gap ( @xmath67 ) .
their positions follow @xmath69 in the same way as @xmath76
. however , the present case is quite different in that the measured @xmath77-curves belong to a series of diffusive sns junctions . when dividing @xmath42 by the number of sn interfaces @xcite and considering that @xmath42 dropped mainly over the normal - state regions of the bridge , it could be concluded that @xmath76 lay inside the gap and originate from mar ( typical values for @xmath78 are 200 @xmath18v for al @xcite ) .
however , it is also possible that series of ar processes lead to multiplication effects and to different effective voltages across subsequent sn interfaces .
therefore , even if caused by the same process , features in @xmath77 could repeatedly appear at different voltages , and result in the observed set of local minima . moreover ,
a multiplication effect in series of junctions might also lead to an increase of the conduction by factors higher than two @xcite .
+ + ( color online ) differential conductance spectra of the transport bridge of figure [ fig : fig_1](a ) at 0 mt ( dws ) .
arrows mark two minima whose positions can be traced . in the left panel ,
curves are shifted for clarity and the vertical scale corresponds to the top curve.,width=325 ] intriguingly , all observations described above can be made not only in the case of rds but also in absence of external fields when dws is realized ( figure [ fig : fig_5 ] ) . in that case ,
local minima are less pronounced , but nonetheless , two of them can be traced up to higher temperatures ( lower curve in figure [ fig : fig_4 ] ) . as before ,
their positions in the conductance spectra follow the collapse of @xmath66 .
it is remarkable that values of @xmath0 obtained by fitting are significantly different in the cases of rds ( @xmath79 k ) and dws ( @xmath80 k ) .
these findings reflect directly that due to quantum size effects , @xmath0 values of superconducting micro - structures differ significantly from those of bulk superconductors @xcite an effect that leads to the reduction of @xmath0 when superconductivity is confined above the domain walls of a underlying ferromagnet @xcite .
in conclusion , there are two major findings of this work : on one hand , it has been demonstrated that tunable sn junctions inside superconducting thin films can be created _ in a controlled manner _ by using magnetic templates .
this first conclusion is a direct result from the successful fabrication of a s / f hybrid system , that allows for setting up dws and rds in the s - layer , without changing the actual configuration of magnetic domains . on the other hand ,
the occurrence of andreev reflection , observed in the conductance of the s - layer of the hybrid system , proves the high transparency of sn interfaces induced by magnetic stray fields .
this result is based on the innovative approach to create sn junctions _ in the same material _ via local suppression of superconductivity . from a technological point of view
, generation of sn junctions via ferromagnets is attractive due to both , the natural tunability of magnetic domain structures and the here demonstrated high quality of sn interfaces .
potentially , inclusion of magnetic templates with pure in - plane magnetization will make it possible to invert the scheme of dws and to _ suppress _ superconductivity in a very narrow region above domain walls , realizing the domain - wall normal state ( dwn ) .
such configuration may lead to controllable phase coupling effects between two superconducting reservoirs , separated by a thin dwn region , and pave the way for the development of new types of tunable quantum interference devices .
r. j. soulen jr . and
j. m. byers and m. s. osofsky and b. nadgorny and t. ambrose and s. f. cheng and p. r. broussard and c. t. tanaka and j. nowak and j. s. moodera and a. barry and j. m. d. coey , science * 282 * , 85 ( 1998 ) . | using the highly inhomogeneous fields of a magnetic substrate , tunable junctions between superconducting and normal state regions were created inside a thin film superconductor .
the investigation of these junctions , created _ in the same material _
, gave evidence for the occurrence of andreev reflection , indicating the high transparency of interfaces between superconducting and normal state regions .
for the realization of this study , a ferromagnet with magnetic stripe domains was used as a substrate , on top of which a superconducting transport bridge was prepared perpendicular to the underlying domains .
the particular choice of materials allowed to restrict the nucleation of superconductivity to regions above either reverse - domains or domain walls .
moreover , due to the specific design of the sample , transport currents in the superconductor passed through a sequence of normal and superconducting regions . |
in recent years , a dynamical model had been developed@xcite to investigate electromagnetic pion production reactions in the region near the @xmath0(1232 ) resonance .
the model was subsequently extended@xcite to also investigate neutrino - induced reactions @xmath15 .
fairly consistent descriptions of all of the available data in the @xmath0 region have been obtained . in this work
, we further extend this model ( called the sl model ) to address two questions of current interest : ( 1 ) will hadron - quark duality first observed by bloom and gilman@xcite in inclusive @xmath16 be also seen in the neutrino - induced @xmath17 and @xmath18 reactions ? ( 2 ) how can parity violating asymmetry of inclusive @xmath19 reaction be used to improve our knowledge about the axial @xmath1-@xmath0 form factor ?
experimental data for testing our predictions can be obtained at new neutrino facilities@xcite and jefferson laboratory ( jlab)@xcite .
our first step is to construct neutral currents within the sl model .
this then allows us to extend our previous calculations@xcite of @xmath20 reactions to also predict @xmath21 reactions for examining some questions concerning the quark - hadron duality .
simply speaking , an inclusive lepton scattering observable exhibits quark - hadron duality when an average of this quantity over an appropriately chosen scaling variable in the resonance region is close to that in the deeply inelastic scattering ( dis ) region .
more details on this subject can be found in a recent review by melnitchouk , ent and keppel@xcite .
the quark - hadron duality was first observed@xcite for the @xmath22 structure function @xmath23 .
recent experimental data@xcite for both @xmath5 and @xmath24 have further confirmed more quantitatively this interesting observation .
theoretical attempts in understanding the quark - hadron duality within qcd were already made@xcite in 1970 s .
more recent works are reported in , for example , refs.@xcite .
within the standard model@xcite , it is natural to ask whether the quark - hadron duality should also exist in the neutrino - induced @xmath25 and @xmath18 reactions ( from nowon , @xmath26 and @xmath27 denote the neutrinos and charged leptons , respectively , of any generation within the standard model ) . in the absence of necessary neutrino data ,
this can be explored theoretically in the @xmath0 region using the extended sl model presented in this paper .
obviously , we can only explore the local quark - hadron duality in the @xmath0 region .
we will also report on our investigations of the spin dependent structure functions @xmath9 and @xmath10 of @xmath3 processes . to see the accuracy of our model in predicting these quantities ,
we first show that some recent data@xcite of the polarization observables associated with the exclusive @xmath11 and @xmath28 reactions agree well with the predictions from the considered sl model .
we then find that @xmath9 calculated from our model agree reasonably well with the recent data@xcite , but do not show quark - hadron duality when compared with the dis data@xcite .
the calculated @xmath10 is also found to be rather different from the dis data which can be described reasonably well by the wandzura - wilczek formula@xcite .
our second task in this work is to apply the extended sl model to address some questions concerning the @xmath1-@xmath0 form factors .
it has been well recognized that these form factors are important information for testing current hadron models and also lattice qcd calculations in the near future .
the vector parts of the @xmath1-@xmath0 form factors in the @xmath29 ( gev / c)@xmath30 region have been rather well determined by analyzing very extensive high precision data of electromagnetic pion production reactions , as reviewed in ref.@xcite . on the other hand , the axial vector @xmath1-@xmath0 form factor is not well determined mainly because of the short of high precision data of neutrino - induced pion production reactions in the @xmath0 region .
for example , the axial @xmath1-@xmath0 from factor determined in ref.@xcite is rather different from the one determined previously@xcite .
this leads to uncertainties in interpreting the axial @xmath1-@xmath0 form factor in terms of the hadron structure calculations such as those reported in refs.@xcite .
the situation will be improved when the data from new neutrino facilities become available in the near future .
alternatively , progress can be made by following refs.@xcite to investigate the parity violating asymmetry of inclusive @xmath2 reactions .
this polarization observable is due to the interference between the electromagnetic currents and neutral currents and hence can be used to explore the axial @xmath1-@xmath0 form factor .
experimental effort in this direction is being made@xcite at jlab . to facilitate this study
, we have applied the extended sl model to explore the dependence of the parity violating asymmetry of @xmath2 on the axial @xmath1-@xmath0 form factor . in section
ii , we will specify the considered models of electromagnetic currents ( @xmath31 ) , weak charged currents ( @xmath32 ) , and weak neutral currents ( @xmath33 ) .
we first recall their forms in the standard model@xcite and then specify how such currents are defined in terms of hadronic degrees of freedom of the sl model . in section iii , we give expressions of the inclusive cross sections and indicate how the structure functions are calculated within our hadronic model and parton model .
section iv is devoted to present results for studying the quark - hadron duality in the @xmath0 region .
the predicted @xmath34 structure functions will also be compared with the recent data .
the results for the parity violating asymmetry of the inclusive @xmath2 will be given in section v. a summary is given in section vi .
we first recall the electroweak currents defined in the standard model@xcite . in the considered @xmath0 excitation region
, we can eliminate heavy @xmath35 and @xmath36 bosons and keep interactions involving only up ( @xmath37 ) and down ( @xmath38 ) quarks . the interaction lagrangian for our study can then be written as @xmath39 a_\mu \nonumber \\ & & - \frac{g_f v_{ud}}{\sqrt{2 } } [ \bar{\nu } \gamma_\mu(1-\gamma_5)e j^{\mu\dagger}_{cc } + \bar{e } \gamma_\mu(1-\gamma_5)\nu j^\mu_{cc } ] \nonumber \\ & & - \frac{g_f}{\sqrt{2 } } [ \bar{\nu}\gamma_\mu(1-\gamma_5)\nu + \bar{e}(2g_v^e\gamma_\mu - 2 g_a^e \gamma_\mu\gamma^5)e ] j^\mu_{nc } \,,\end{aligned}\ ] ] where @xmath40 , @xmath41 gev@xmath42 , @xmath43 , @xmath44 , @xmath45 is the photon field , @xmath27 and @xmath26 are the field operators of the charged leptons and neutrinos , respectively .
the weinberg angle @xmath46 is known empirically to be @xmath47 and @xmath48 is the the cabibbo - kobayashi - maskawa ( ckm ) coefficient .
the electromagnetic current ( @xmath49 ) , weak charge current ( @xmath50 ) and weak neutral current ( @xmath51 ) carried by @xmath37 and @xmath38 quarks can be written as @xmath52 - 2 \sin^2\theta_w j_{em}^\mu \,.\end{aligned}\ ] ] with the simplification that only @xmath37 and @xmath38 quarks are kept , it is well known@xcite that the above currents can be classified according strong isospin .
we thus can write @xmath53 where the isospin components of the vector ( @xmath54 ) and axial vector ( @xmath13 ) currents are defined as @xmath55 here we have defined an isospin doublet field operator @xmath56 and @xmath57 is the usual pauli operator .
the above expressions will allow us to calculate electroweak structure functions of deeply inelastic @xmath58 , @xmath59 and @xmath8 processes within the parton model , as explained in , for example , ref.@xcite .
however , they can not be used directly for investigating meson production reactions in the resonance region where the perturbative qcd is not applicable . at the present time ,
the most tractable ways for investigating these reactions are in terms of hadronic degrees of freedom .
the starting point is a hadronic effective lagrangian constrained by the symmetry properties of the standard model .
accordingly , the resulting electroweak currents have the forms of eqs.(5)-(8 ) , but are written in terms of hadronic field operators .
this can be achieved by using the standard effective chiral lagrangian methods@xcite . in the sl model developed in refs.@xcite ,
the constructed electroweak currents are expressed in terms of the field operators of the nucleon ( @xmath1 ) , delta ( @xmath0 ) , pion ( @xmath60 ) , omega meson ( @xmath61 ) , and rho meson ( @xmath62 ) .
they can be written as @xmath63\cdot \frac{\vec{\tau}}{2}n + \frac{g_a}{2f}\bar{n } \gamma^\mu\gamma_5 [ \vec{v}_\mu\cdot \vec{\tau}]n \times \vec{\pi } \nonumber \\ & & + [ \vec{\pi } \times \partial^\mu \vec{\pi}]\cdot\vec{v}_\mu - \frac{g_{\omega\pi v}}{m_\pi}\epsilon_{\alpha\mu\nu\delta } [ \partial^\alpha\vec{v}^\mu]\cdot \vec{\pi } [ \partial^\nu\omega^\delta ] % \nonumber \\ & & -i \bar{\delta}_\mu \vec{t}\cdot \vec{v}_\nu \gamma^{\mu\nu}_vn \ , , \\ \vec{a}^{\mu}\cdot \vec{v}_\mu & = & g_a \bar{n } \gamma^\mu \gamma_5\frac{\vec{\tau}}{2}\cdot \vec{v}_\mu
n - f \partial^\mu \vec{\pi } \cdot \vec{v}_\mu - f_{\rho\pi a}(\vec{\rho}^\mu \times \vec{\pi})\cdot\vec{v}_\mu + \bar{\delta}_\nu \vec{t}\cdot \vec{v}_\mu \gamma^{\mu\nu}_a n \,,\end{aligned}\ ] ] where @xmath64 is an arbitrary isovector function , @xmath65 mev is the pion decay constant , and @xmath66 is the axial coupling strength of the nucleon .
the other parameters as well as the vertex @xmath67 and isospin operator @xmath68 of the @xmath1-@xmath0 transition are given in refs.@xcite .
with the electromagnetic current @xmath49 given explicitly in ref.@xcite , the relation eq.(5 ) and eq.(12 ) can be used to get @xmath69 and hence @xmath51 of eq.(8 ) is also completely determined within the sl model .
the neutral current @xmath51 is needed to extend the calculations of refs.@xcite to also study @xmath70 reactions and the parity violating asymmetry of the @xmath71 reaction .
the electroweak currents defined by eqs.(5)-(8 ) and ( 12)-(13 ) must be supplemented by additional effective lagrangians describing the haronic interactions , such as @xmath72 , @xmath73 , @xmath74 , @xmath75 , and @xmath76 couplings , to calculate the meson production amplitudes within the sl model .
the details of such a dynamical approach have been given in refs.@xcite .
here it is sufficient to just illustrate schematically the basic meson production mechanisms of the sl model . in fig.1
, we show the constructed non - resonant pion production mechanisms . for the @xmath31 current contributions , the waved - line is photon and all diagrams contribute . for @xmath32 and @xmath33 current contributions ,
the wave - line is the vector current @xmath54 or axial vector current @xmath13 . like the @xmath31 case ,
all terms in fig.1 contribute to @xmath33 amplitude but with different isospin weighting factors defined by eqs.(5 ) and ( 8) . for the @xmath32 contributions ,
all terms contribute to the amplitudes induced by the vector current @xmath54 except that the @xmath77 exchange in ( c ) should be excluded because it is an isoscalar interaction .
the @xmath32 s amplitude due to axial vector current @xmath13 contains only mechanisms ( a ) , ( b ) , ( e ) with only @xmath77 exchange , and ( f ) . in both @xmath32 and @xmath33 cases ,
the pion pole term due to the second term @xmath78 of axial current eq.(13 ) has additional contributions , as explained in ref.@xcite .
the @xmath0 excitation mechanism is illustrated in fig.2 .
for the later discussions in this paper , here we recall that the matrix element of the axial @xmath1-@xmath0 coupling @xmath79 in eq.(13 ) can be written as @xmath80 where @xmath81 and @xmath82 are the momenta of the @xmath0 and @xmath1 , respectively , @xmath83 , and @xmath84 .
the strengths of the form factors at @xmath85 are identified with the quark model predictions@xcite and are found to be @xmath86 with @xmath66 , @xmath87 , @xmath88 , @xmath89 , and @xmath90 . in ref.@xcite , it was found that all of the available data of @xmath14 data in the @xmath0 region can be well described if the @xmath91-dependence of each form factor in eq.(14 ) is taken to be ( @xmath92 ) @xmath93 with @xmath94 where @xmath95 ( gev / c)@xmath42 , @xmath96 ( gev / c)@xmath30 , and @xmath97 with @xmath98 gev is the nucleon axial form factor@xcite .
the corresponding vector form factor @xmath99 , which is associated with @xmath100 of eq.(12 ) and has been well determined@xcite by analyzing the data of pion electroproduction , has the same form as eq.(15 ) except that @xmath101 is replaced by the usual proton form factor @xmath102 with @xmath103 ( gev / c)@xmath30 .
we further mention that the electroweak pion production amplitudes ( denoted as @xmath104 from nowon ) calculated within the sl model can be written schematically in the center of mass ( c.m . )
frame of the final @xmath105 system as @xmath106 where @xmath107 , w or @xmath36 bosons , @xmath108 and @xmath109 are the relevant initial and final momenta , @xmath110 is the @xmath0 self - energy .
the non - resonant amplitudes @xmath111 are calculated from @xmath112 where @xmath113 are the non - resonant amplitudes illustrated in fig.1 , @xmath114 is the non - resonant @xmath105 scattering amplitude , and @xmath115 is the @xmath105 propagator . note that the second term of eq.(17 ) is the consequence of the unitarity condition .
this is neglected in the previous investigations@xcite of the effects due to the non - resonant amplitudes on the parity violating asymmetry of @xmath2 reaction .
furthermore , the non - resonant mechanisms considered in those works also differ from what are illustrated in fig.1 and well tested in extensive investigations of exclusive @xmath116 , @xmath117 , and @xmath118 reactions .
for example , the vector meson exchanges are not included in refs.@xcite and pseudo - scalar @xmath72 coupling , instead of the pseudo - vector coupling used in the sl model , is used in ref.@xcite . the resonant term in eq.(16 )
is defined by the dressed vertices which contain the influence of the non - resonant interactions as given by the following equation @xmath119 where @xmath120 is the @xmath121 @xmath1-@xmath0 form factor . the second term in eq.(18 )
is commonly called the meson cloud contributions to the @xmath1-@xmath0 transition .
explicit calculations of these meson cloud effects also mark an important difference between this work and all of the previous investigations@xcite of parity violating asymmetry of @xmath2 reactions .
we emphasize that eqs.(16)-(18 ) satisfy the unitarity condition which is essential in interpreting the meson production data , as explained in refs.@xcite as well as in many works reviewed in ref.@xcite .
with the lagrangian eq.(1 ) , the formula for calculating the exclusive cross sections for @xmath122 and @xmath123 processes are given in details in refs.@xcite . the formula for calculating @xmath124
cross sections can be easily obtained from ref.@xcite with minor modifications . in this paper , we focus on inclusive processes .
their cross sections can be more simply written in terms of structure functions . following ref.@xcite
, the symmetry properties require that the lepton scattering structure functions @xmath125 , @xmath126 and @xmath127 are in general related to the hadron tensor @xmath128 by @xmath129 where the convention @xmath130 is chosen .
the structure functions are functions of two independent invariant variables .
one usually chooses @xmath131 and the invariant mass @xmath132 in the resonance region , but chooses bjorken scaling variable @xmath133 and @xmath131 in the deeply inelastic region . here
@xmath108 and @xmath134 are the momenta of the exchanged boson ( @xmath135 , @xmath35 or @xmath36 ) and the initial nucleon , respectively .
the unpolarized inclusive @xmath34 cross sections can then be written as @xmath136 \,,\end{aligned}\ ] ] with @xmath137 where @xmath138 and @xmath139 are the incident and outgoing lepton energies , respectively , @xmath140 is the lepton scattering angle with respect to the incident lepton . for inclusive @xmath141 and @xmath142 processes , we have @xmath143 \,,\end{aligned}\ ] ] where the sign in front of @xmath127 is @xmath144 ( - ) for @xmath145 ( @xmath146 ) . for @xmath8 , the cross section formula is the same as eq.(22 ) except that the factor @xmath147 is removed . within the hadronic models , such as the sl model considered in this paper
, the hadron tensor @xmath128 in the region near the @xmath0 excitation can be calculated from summing all contributions from @xmath148 with @xmath149 .
we can write in general the hadron tensor @xmath128 for electroweak pion reactions as @xmath150 where @xmath151 denotes the considered current , and we have introduced concise notations @xmath152 and @xmath153 here @xmath154 denote the z - components of the nucleon spin - isospin , @xmath155 the z - component of the pion isospin , and @xmath156 is the @xmath157 amplitude with @xmath158 , @xmath159 , @xmath36 for @xmath160 , @xmath32 , and @xmath33 , respectively . in the dynamical approach of refs.@xcite ,
the current matrix element eq.(25 ) has the form of eq.(16 ) , consisting of a non - resonant term and a resonant term . from eqs.(19 ) and
( 23 ) , one can derive the expressions for calculating the structure functions @xmath125 , @xmath126 , and @xmath127 from the current matrix elements defined by eq.(25 ) .
it is convenient to calculate these structure functions in the center of mass ( c.m . )
frame of the initial @xmath161 and final @xmath105 systems .
the direction of the momentum - transfer is chosen to be the quantization @xmath162-direction : i.e. @xmath163 for the exchanged boson and @xmath164 for the initial nucleon .
we then have for the contribution from current @xmath165 with @xmath151 , @xmath166 \ , , \\
w_3 & = & - \frac{2m_n}{\mid { \bf q}\mid } \sum im[<f\mid j_\alpha^x\mid i >
< f\mid j_\alpha ^y \mid i > ^ * ] \,,\end{aligned}\ ] ] where @xmath167 , and @xmath168 is three - momentum transfer in the laboratory frame ( i.e. @xmath169 ) . in practice , the calculation of any term of the structure functions defined by eqs.(26)-(28 ) can be obtained from appropriate combinations of the following integrations @xmath170 \nonumber \\ & = & \frac{1}{2}\sum_{s , s',t',t_{\pi } } \int d\omega_{\pi}\frac{|{\bf k}_{c}|m_n}{16\pi^3w } < k(t_\pi)p^\prime(s',t')|j_{\alpha}^\mu(q)|p(s , t ) > < k(t_\pi)p^\prime(s',t')|j_{\beta}^\nu(q)|p(s , t)>^ * \ , . \nonumber \\\end{aligned}\ ] ] were @xmath171 , and @xmath172 is final pion momentum in the final @xmath105 c.m . system .
we will also examine the spin dependent structure functions of @xmath3 .
they are defined@xcite by writing the hadron tensor for a polarized target with spin vector @xmath173 ( @xmath174 ) as @xmath175 where @xmath176 \ , .
\end{aligned}\ ] ] with some derivations , one can show that @xmath177 \ , , \\ g_2&=&\frac{1}{1+\frac{q^2}{\omega^2}}[x^{y0 } - x^{xy } ] \,,\end{aligned}\ ] ] with @xmath178 \
x^{xy}&= & - m_n im[i^{xy}_{em , em}\mid_{s = s(z ) } ] \,,\end{aligned}\ ] ] where @xmath179 is the same as that defined in eq.(29 ) except that the initial nucleon projection @xmath180 is not summed over but is fixed in the chosen direction defined by @xmath181 .
@xmath182 ( @xmath183 ) means that the initial nucleon spin is polarized in the direction perpendicular ( parallel ) to the incident electron direction . for investigating quark - hadron duality
, we would like to compare the structure functions calculated from using the above formula for hadronic models with those calculated from the quark currents
eqs.(2)-(4 ) in the deeply inelastic region . with the standard definitions @xmath184 where @xmath185 , the parton model gives@xcite ( keeping only the contributions from the @xmath37 and @xmath38 quarks in the considered @xmath0 region ) @xmath186 \ , , \\ % & & + \frac{1}{9}(s(x , q^2)+\bar{s}(x , q^2 ) ) ] f_{1}(x , q^2 ) & = & f_{2}(x , q^2)/2x \ , , \\ g_1(x , q^2 ) & = & \frac{4}{18}[u^\uparrow(x , q^2)-u^\downarrow(x , q^2 ) + \bar{u}^\uparrow(x , q^2)-\bar{u}^\downarrow(x , q^2 ) ] \nonumber \\ & & + \frac{1}{18 } [ d^\uparrow(x , q^2)-d^\downarrow(x , q^2 ) + \bar{d}^\uparrow(x , q^2)-\bar{d}^\downarrow(x , q^2 ) ] \,.\end{aligned}\ ] ] @xmath188 @xmath190 ( u(x , q^2)+\bar{u}(x , q^2 ) ) \nonumber \\ & & + x[\frac{1}{4}+(-\frac{1}{2 } + \frac{2}{3}\sin^2\theta_w)^2](d(x , q^2)+\bar{d}(x , q^2 ) ) \ , , \\
f_3(x , q^2)&= & [ \frac{1}{2}-\frac{4}{3}\sin^2\theta_w](u(x , q^2)-\bar{u}(x , q^2 ) ) \nonumber \\ & & + [ \frac{1}{2}-\frac{2}{3}\sin^2\theta_w](d(x , q^2)-\bar{d}(x , q^2 ) ) \,.\end{aligned}\ ] ] here @xmath191 , @xmath192 with @xmath193 are the parton distribution functions ( pdf ) determined from fitting the data in deeply inelastic region . in eq.(40 ) @xmath194 are the spin dependent parton distribution functions .
the structure functions for the neutron target can be obtained from eqs.(38)-(44 ) by interchanging the @xmath37 and @xmath38 parton distribution functions . in actual calculations , the strange quark and sea quark contributions
are included , but are found to be very small in the considered @xmath0 region .
thus our results presented below are from eqs.(38)-(44 ) .
this is consistent with the considered hadronic sl model which also neglects any possible reaction mechanisms involving intermediate strange hadrons .
we now turn to exploring the quark - hadron duality in the @xmath0 excitation region .
this is done by comparing the structure functions calculated from eqs.(38)-(44 ) using the cteq6 parton distribution functions @xcite with those from eqs.(26)-(28 ) and ( 33)-(36 ) using the hadronic sl model described in section ii .
it is well known@xcite that more quantitative tests of quark - hadron duality need to include target mass corrections .
furthermore , the role of the higher - twist effects must be better understood . for simplicity
, we will not take such a more involved procedure and will only compare all results from the sl model with the parton model predictions at @xmath195 ( gev / c)@xmath30 .
thus our goal here is more qualitative .
we will focus on exploring whether the neutrino - induced reactions show the similar quark - hadron duality observed in @xmath3 .
furthermore we will also consider the spin dependent structure functions and parity violating asymmetry of @xmath196 to which the procedures for including the target mass corrections have not been developed .
we follow the usual criterion@xcite that the local quark - hadron duality is seen if the the predictions from our hadronic model are `` oscillating '' around the predictions from parton model such that their averaged values could be very close after target mass corrections are included . following the previous works , as reviewed in ref.@xcite , we present the calculated structure functions as functions of the nachtmann scaling variable defined by @xmath197 where @xmath198 is the bjorken scaling variable
. one can show@xcite that @xmath199 is the fraction of the plus light - front momentum of the nucleon carried by the struck quark in the infinite momentum frame .
the use of this variable includes some of the target mass corrections , as discussed in ref.@xcite .
we note here that in the @xmath200 about 0.7 region the parton model results at @xmath195 ( gev / c)@xmath30 will correspond to @xmath201 about 2 gev which is much larger than the considered @xmath0 region ( 1.1 gev @xmath202 w @xmath202 1.4 gev ) .
for example , for @xmath203 ( gev)@xmath30 the @xmath0 peak ( @xmath204 gev ) occurs at @xmath205 which corresponds to @xmath206 gev at @xmath195 ( gev / c)@xmath30 .
first we consider the @xmath16 processes .
as reported in refs.@xcite , the recent data of structure functions @xmath4 and @xmath5 from jlab have further established the quark - hadron duality in the entire resonant region .
this is illustrated in fig.3 along with the results calculated from the sl model ( solid curves near @xmath207 ) in the @xmath0 region .
from now on , we will only consider the data in the @xmath0 region .
in fig.4 we compare our hadronic model calculations at @xmath208 ( gev / c)@xmath30 ( solid curves , from left to right ) with these data .
each solid curve covers the same @xmath0 region with @xmath209 gev @xmath210 w @xmath211 gev .
we see that the predictions from the employed hadronic model ( solid curves ) agree well with the data and oscillate around the parton model predictions ( dashed curve ) . to pursue further
, it is natural to ask , from the point of view of standard model , whether the quark - hadron duality observed in @xmath16 should also be seen in neutrino - induced processes .
experimental data for such investigations are still absent , but could be obtained at new neutrino facilities in the near future . to facilitate these developments , we here present predictions for @xmath187 and @xmath189 processes . since our model allows us to also predict the structure functions for the neutron target and hence we will also provide predictions of @xmath212 for the isospin @xmath213 deuteron - like target .
calculations for general nuclear targets , such as those considered in ref.@xcite , are beyond the scope of this work . our predictions for structure functions are shown in fig.5 for @xmath3 , fig.6 for @xmath145 , and fig.7 for @xmath8 .
clearly all cases exhibit the similar feature of quark - hadron duality .
experimental confirmations of our predictions ( solid curves ) shown in figs.5 - 7 will be useful for making further progress .
if they are confirmed , our next step is to examine the local quark - hadron duality @xmath214 by including the target mass corrections@xcite and considering the available information about the role of the higher twist effects . as reviewed in ref.@xcite , an another criterion of the quark - hadron duality
is that the structure functions in the resonance region should slide along the parton model predictions as @xmath131 increases .
this is clearly the case in figs.5 - 7 for @xmath215 ( gev / c)@xmath30 . in fig.8
we further show within the employed sl hadronic model that this should also be the case up to rather high
@xmath216(gev / c)@xmath30 where the high precision data for the @xmath0 region could be obtained from the experiments with 12-gev upgrade of jlab .
the existing slac data@xcite do not have enough high accuracy for investigating the @xmath0 at @xmath217 about 6 ( gev / c)@xmath30 .
we now turn to discussing the spin dependent structure functions @xmath9 and @xmath10 .
we first note from eqs.(33)-(36 ) that these functions depend on the interference terms @xmath218 and @xmath219 .
such terms also determine various polarization observables of the exclusive @xmath11 and @xmath12 precesses , as discussed in , for example , ref.@xcite .
it is therefore important to test the sl model predictions against the recent data from such polarization measurements@xcite .
it is clear from figs.9 - 12 that the sl model can describe the jlab data very well and certainly can be used here to investigate the spin dependent structure functions .
the details of these comparisons can be found in refs.@xcite and also in the captions of figs.9 - 12 .
our results for @xmath9 and @xmath10 of @xmath22 are shown in fig.13 . in the left side of
fig.13 we again see that our predictions for @xmath9 ( solid curves ) agree reasonably well with the data@xcite .
( note that the determination of these @xmath9 data from the polarization data of ref.@xcite involved some model dependent input@xcite . ) to get the parton model predictions using eq.(40 ) , we need to known the spin dependent parton distribution functions .
however the determination of such information in the considered large @xmath220 region is still in developing stage and therefore no attempt is made here to do calculation using eq.(40 ) .
instead , we assume that the parton model predictions can be identified with the dashed curve which is from fitting the data of @xmath9 measured in deeply inelastic scattering @xcite .
the left side of fig.13 then indicates that @xmath9 clearly does not exhibit the local quark - hadron duality .
such a disagreement was in fact expected@xcite by considering the constraints imposed by the ellis - jaffe integral@xcite and drell - hearn - gerasimov sum rule@xcite .
it is known@xcite that the naive parton model , considered here , can not predict @xmath10 .
there are however some @xmath10 data@xcite from deeply inelastic scattering which can be described well by using the wandzura - wilczek formula@xcite @xmath221-g_1(x , q^2 ) \,.\end{aligned}\ ] ] the dashed curve in the right side of fig.13 is obtained from the above formula using the dashed curve for @xmath9 in the left as the input .
clearly it disagrees with the predictions ( solid curves ) from the considered hadronic sl model . unlike the result for @xmath9
, there is no simple explanation of such a similar breakdown of local quark - hadron duality .
the quark - hadron duality for the spin dependent structure functions of @xmath222 and @xmath189 are not investigated here since there are no corresponding pqcd calculations or data from deeply inelastic scattering data to compare with .
the formula for calculating parity violating asymmetry @xmath13 of @xmath2 reactions have been given in literatures@xcite .
for our purposes , we write it in terms of structure functions defined in section iii . with some straightforward derivations ,
we have @xmath223 with @xmath224 here the electromagnetic structure functions @xmath225 are calculated from eqs.(26)-(27 ) with @xmath160 . the interference term @xmath226 between the electromagnetic and neutral currents
can also be calculated from eqs.(26)-(28 ) with the following replacements @xmath227 obviously these quantities can also be calculated from integration @xmath228 defined in eq.(29 ) .
we next focus on the first two terms of @xmath1 of eq.(48 ) which depend on @xmath229 . here
we note that @xmath229 is the symmetric part of the hadron tensor defined in eq.(19 ) .
hence they can only have the contributions from the vector parts of neutral currents because of the vector structure eq.(5 ) of the electromagnetic current . to see this more clearly
, we use eq.(8 ) to write the neutral current as@xmath230 with @xmath231 obviously the @xmath232 $ ] part of eq.(50 ) will not contribute to @xmath229 and hence the first two terms of @xmath1 of eq.(48 ) can then be written as @xmath233 where @xmath234 is defined in eq.(49 ) and @xmath235 is the same as @xmath226 with @xmath236 replaced by the isoscalar vector current [ @xmath237 defined in eq.(51 ) .
thus only the non - resonant isospin @xmath238 @xmath105 state contribute to @xmath235 . with the above relation ,
eq.(47 ) for the asymmetry can then be written as @xmath239 \nonumber \\ & = & - q^2 [ 8.99 \times 10^{-5 } ] ( 1.075 + \delta_v + \delta_a ) \,,\end{aligned}\ ] ] where @xmath131 is in unit of ( gev / c)@xmath30 and @xmath240 in eq.(52 ) we have evaluated ( @xmath241 which is a model - independent constant and is the main contribution to the asymmetry .
this term was identified by cahn and gilman@xcite and later investigators@xcite .
we now note that the term @xmath242 defined by ( 54 ) depends only on @xmath243 which is antisymmetric as defined in eq.(19 ) .
because of the vector structure of electromagnetic current eq.(5 ) , @xmath243 can only have the contributions from the axial vector parts of neutral currents .
clearly , the information about the axial vector @xmath1-@xmath0 form factor , which is the quantity we hope to learn about as discussed in section i , is isolated in @xmath242 through its dependence on @xmath243 .
therefore it is important to identify the region where @xmath242 is much larger than @xmath244 such that the extracted axial @xmath1-@xmath0 form factor has less model dependence . as mentioned above , only the non - resonant isospin @xmath245 @xmath105 final states contributes to @xmath235 .
thus we expect that @xmath244 is weaker than @xmath242 near the @xmath0 resonance .
this is illustrated in fig.14 for incident electron energy @xmath246 gev .
we see that @xmath242 ( solid curve ) is indeed much larger than @xmath244 at energies near @xmath204 gev of the @xmath0 position .
this is also the case for other typical electron kinematics which can be conducted at jlab , as illustrated in fig.15 .
the contribution from @xmath242 ( solid curves ) are clearly much larger than @xmath244 ( dashed curves )
. therefore the asymmetry data at resonance position @xmath204 gev can be used to extract the contributions from the axial vector currents .
we now focus on examining how @xmath242 depends on the axial @xmath1-@xmath0 form factor . recalling eqs.(16)-(18 ) for the current matrix elements , @xmath242 , like other observables , contains contributions from the resonant and non - resonant amplitudes .
as seen in fig.16 , the non - resonant contribution ( dotted curve near the bottom ) is very small compared with the full model prediction ( solid curve ) .
this agrees with previous investigations@xcite , although those earlier works neglected the unitary condition as discussed in section i. the large difference between the solid and the dotted curves in fig.16 indicates that @xmath242 is a useful quantity for extracting the dressed axial @xmath1-@xmath0 form factor defined by eq.(18 ) .
when the pion cloud effects ( second term of eq.(18 ) ) on the @xmath1-@xmath0 transition is turned off , we obtain the dash curves . this is consistent with our previous findings in refs.@xcite that pion cloud effects on @xmath1-@xmath0 transition are significant .
we present in fig.17 our predictions of asymmetry @xmath13 for several incident electron energies @xmath247 gev and at the @xmath0 peak . here
we note that our result for @xmath248 gev is about 10 @xmath249 larger than that shown in fig.6 of ref.@xcite . furthermore
, their @xmath13 decreases with @xmath131 in striking difference with our predictions .
this is perhaps mainly due to the use of different @xmath1-@xmath0 form factor , but the differences in treating the non - resonant amplitudes , which interfere with the dominant @xmath1-@xmath0 transition amplitude , also play some roles .
experimental tests of our predictions in fig.17 will be useful in examining the extent to which the axial @xmath1-@xmath0 form factor eq.(15 ) , which was determined in our study@xcite of @xmath250 reaction , is valid . as pointed out in ref.@xcite ,
the form factor eq.(15 ) is rather different from the form @xmath251 with @xmath252 and @xmath253 ( gev / c)@xmath30 which was determined in earlier works@xcite .
it is therefore interesting to see how these two form factors can be distinguished by the parity violation asymmetry of @xmath254 .
this is illustrated in fig.18 .
experiment test of our prediction shown in fig .
18 will help distinguish these two axial @xmath1-@xmath0 form factors . to end this section , we return briefly to the investigation of quark - hadron duality presented in section iv .
it is also interesting to explore@xcite whether the quark - hadron duality also exists in the parity violating asymmetry @xmath13 .
this can be done here by comparing the predictions of our hadronic model and that of parton model .
the formula for calculating @xmath13 within the parton model was derived in ref.@xcite . keeping only the @xmath37 and @xmath38 quarks in the considered large @xmath220 region ,
one finds for a proton target @xmath255-c_{1d}[d(x)+\bar{d}(x ) ] + y[2c_{2u}(u(x)-\bar{u}(x))-c_{2d}(d(x)+\bar{d}(x ) ) ] } { 4(u(x)+\bar{u}(x))+(d(x)+\bar{d}(x ) ) } \nonumber \\\end{aligned}\ ] ] where @xmath256 note that @xmath257 is the ratio between the longitudinal and transverse total cross sections .
we use the values of @xmath257 calculated from the sl model .
the variable @xmath258 depends on incident electron energy @xmath138 and the energy transfer @xmath61 .
the other coefficients in eq.(55 ) depend on the electroweak coupling constants for leptons and quarks @xmath259 when the radiative corrections within the standard model is included @xcite @xmath260 .
these values are only slightly different from those calculated from eq.(59 ) .
the asymmetry of @xmath254 predicted from our hadronic model ( solid curves ) and parton model ( dashed curves ) for several typical electron kinematics are compared in fig.19 .
if our hadronic model results ( solid curves ) are confirmed experimentally , the parity violating asymmetry in the @xmath0 region obviously does not show local quark - hadron duality .
the situation is similar to the results for the spin dependent structure functions shown in fig.14 .
the dynamical model developed in refs.@xcite ( the sl model ) has been extended to include the weak neutral current contributions for investigating all possible electroweak pion production reactions in the region near the @xmath0 excitation .
the main purpose is to examine the quark - hadron duality in the neutrino - induced reactions and to explore how the axial @xmath1-@xmath0 transition form factor can be determined by using the parity violating asymmetry of @xmath254 .
the experimental data for testing our predictions can be obtained at jlab and new neutrino facilities .
we have found that the @xmath3 structure functions @xmath4 and @xmath5 predicted by the sl model are in good agreement with the recent data@xcite which had verified more quantitatively the quark - hadron duality first observed by bloom and gilman@xcite .
the predicted structure functions for @xmath7 and @xmath8 processes also show quark - hadron duality to the extent similar to what has been observed in @xmath3 .
furthermore , we also predict that quark - hadron duality should also be seen in all electroweak reactions on the neutron or equivalently the isospin @xmath213 deuteron - like target .
these results suggest that the sl model can be a candidate hadronic model for developing a theoretical understanding of quark - hadron duality within the standard model .
for example , it will be interesting to explore which parts of the predictions from the sl model can be related to the predictions from the parton model .
this is clearly a difficult question to answer and is beyond the scope of this paper .
we have also investigated the spin dependent structure functions @xmath9 and @xmath10 of @xmath3 .
it is found that the quark - hadron duality is not seen in the calculated @xmath9 and @xmath10 , while our results for @xmath9 and some polarization observables associated with the exclusive @xmath11 and @xmath12 reactions are in good agreement with the recent data .
experimental data for @xmath10 in the @xmath0 region are clearly needed . if our predictions are confirmed , we perhaps have more complete information for exploring why the quark - hadron duality breaks down in the spin dependent structure functions .
we should also emphasize that our investigations of quark - hadron duality are rather qualitative since they are based on the naive parton model . to test local quark - hadron duality @xmath214
, we need to include target mass corrections@xcite and consider the roles of higher twist effects . in particular , the procedures for including these effects in the calculations of spin dependent structure functions @xmath9 and @xmath10 and parity violating asymmetry @xmath13 must also be developed . in the investigation of parity violating asymmetry @xmath13 of the inclusive @xmath254
, we have shown that the non - resonant contribution is small at the @xmath204 gev @xmath0 peak and hence a precise measurement of @xmath13 can be used to improve the determination of the axial @xmath1-@xmath0 transition form factor .
we have also predicted that the parity violating asymmetry @xmath13 , like the spin dependent structure functions @xmath9 and @xmath10 , does not exhibit quark - hadron duality .
predictions for the experiments which can be conducted at jlab have been given .
we would like to thank a. biselli , r. fatemi , y. liang and c. smith for useful discussions on the analyses of their experimental data .
this work was supported by the u.s .
department of energy , office of nuclear physics division , under contract no .
w-31 - 109-eng-38 , and by japan society for the promotion of science , grant - in - aid for scientific research ( c ) 15540275 .
m. anselmino , b.l .
ioffe , and e. leader , yad . fiz . *
49 * , 214 ( 1989)[sov .
. phys . * 49 * , 136 ( 1989 ) ] ; f. close , in _ excited baryons 1988 _ , proceedings of the topical workshop on excited baryons , edited by g. adams , n.c .
mukhopadhyay , and p. stoler ( world scientific , singapore , 1989 ) ; v. d. burkert and b. l. ioffe , j. exp .
. phys . * 78 * , 619 ( 1994 ) .
( left ) and @xmath5 ( right ) of @xmath3 for the proton target at @xmath203 ( gev / c)@xmath30 .
the dashed curves are calculated from using the cteq6 parton distribution functions at @xmath195 ( gev / c)@xmath30 .
the solid curves in the region near @xmath207 are calculated from the sl model .
the data are from liang et al.@xcite.,title="fig:",width=226 ] ( left ) and @xmath5 ( right ) of @xmath3 for the proton target at @xmath203 ( gev / c)@xmath30 .
the dashed curves are calculated from using the cteq6 parton distribution functions at @xmath195 ( gev / c)@xmath30 .
the solid curves in the region near @xmath207 are calculated from the sl model .
the data are from liang et al.@xcite.,title="fig:",width=226 ] ( left ) and @xmath5 ( right ) of @xmath3 for the proton target .
the dashed curves are calculated from using the cteq6 parton distribution functions at @xmath195 ( gev / c)@xmath30 .
the solid curves are the results at @xmath261 ( gev / c)@xmath30 ( from left to right ) calculated from the sl model .
the data are from liang et al.@xcite.,title="fig:",width=226 ] ( left ) and @xmath5 ( right ) of @xmath3 for the proton target .
the dashed curves are calculated from using the cteq6 parton distribution functions at @xmath195 ( gev / c)@xmath30 .
the solid curves are the results at @xmath261 ( gev / c)@xmath30 ( from left to right ) calculated from the sl model .
the data are from liang et al.@xcite.,title="fig:",width=226 ] ( left ) and @xmath5 ( right ) of @xmath3 for the proton ( top ) and an i=0 deuteron - like target ( bottom ) .
the dashed curves are calculated from using the cteq6 parton distribution functions at @xmath195 ( gev / c)@xmath30 .
the solid curves are the results at @xmath262 ( gev / c)@xmath30 ( from left to right ) calculated from the sl model.,title="fig:",width=226 ] ( left ) and @xmath5 ( right ) of @xmath3 for the proton ( top ) and an i=0 deuteron - like target ( bottom ) .
the dashed curves are calculated from using the cteq6 parton distribution functions at @xmath195 ( gev / c)@xmath30 .
the solid curves are the results at @xmath262 ( gev / c)@xmath30 ( from left to right ) calculated from the sl model.,title="fig:",width=226 ] ( left ) and @xmath5 ( right ) of @xmath3 for the proton ( top ) and an i=0 deuteron - like target ( bottom ) .
the dashed curves are calculated from using the cteq6 parton distribution functions at @xmath195 ( gev / c)@xmath30 .
the solid curves are the results at @xmath262 ( gev / c)@xmath30 ( from left to right ) calculated from the sl model.,title="fig:",width=226 ] ( left ) and @xmath5 ( right ) of @xmath3 for the proton ( top ) and an i=0 deuteron - like target ( bottom ) .
the dashed curves are calculated from using the cteq6 parton distribution functions at @xmath195 ( gev / c)@xmath30 .
the solid curves are the results at @xmath262 ( gev / c)@xmath30 ( from left to right ) calculated from the sl model.,title="fig:",width=226 ] ( left ) , @xmath5 ( center ) , and @xmath263(right ) of @xmath7 for the proton ( top ) and an i=0 deuteron - like target ( bottom ) .
the dashed curves are calculated from using the cteq6 parton distribution functions at @xmath195 ( gev / c)@xmath30 .
the solid curves are the results at @xmath262 ( gev / c)@xmath30 ( from left to right ) calculated from the sl model.,title="fig:",width=170 ] ( left ) , @xmath5 ( center ) , and @xmath263(right ) of @xmath7 for the proton ( top ) and an i=0 deuteron - like target ( bottom ) .
the dashed curves are calculated from using the cteq6 parton distribution functions at @xmath195 ( gev / c)@xmath30 .
the solid curves are the results at @xmath262 ( gev / c)@xmath30 ( from left to right ) calculated from the sl model.,title="fig:",width=170 ] ( left ) , @xmath5 ( center ) , and @xmath263(right ) of @xmath7 for the proton ( top ) and an i=0 deuteron - like target ( bottom ) .
the dashed curves are calculated from using the cteq6 parton distribution functions at @xmath195 ( gev / c)@xmath30 .
the solid curves are the results at @xmath262 ( gev / c)@xmath30 ( from left to right ) calculated from the sl model.,title="fig:",width=170 ] ( left ) , @xmath5 ( center ) , and @xmath263(right ) of @xmath7 for the proton ( top ) and an i=0 deuteron - like target ( bottom ) .
the dashed curves are calculated from using the cteq6 parton distribution functions at @xmath195 ( gev / c)@xmath30 .
the solid curves are the results at @xmath262 ( gev / c)@xmath30 ( from left to right ) calculated from the sl model.,title="fig:",width=170 ] ( left ) , @xmath5 ( center ) , and @xmath263(right ) of @xmath7 for the proton ( top ) and an i=0 deuteron - like target ( bottom ) .
the dashed curves are calculated from using the cteq6 parton distribution functions at @xmath195 ( gev / c)@xmath30 .
the solid curves are the results at @xmath262 ( gev / c)@xmath30 ( from left to right ) calculated from the sl model.,title="fig:",width=170 ] ( left ) , @xmath5 ( center ) , and @xmath263(right ) of @xmath7 for the proton ( top ) and an i=0 deuteron - like target ( bottom ) .
the dashed curves are calculated from using the cteq6 parton distribution functions at @xmath195 ( gev / c)@xmath30 .
the solid curves are the results at @xmath262 ( gev / c)@xmath30 ( from left to right ) calculated from the sl model.,title="fig:",width=170 ] ( left ) , @xmath5 ( center ) , and @xmath263(right ) of @xmath8 for the proton ( top ) and an i=0 deuteron - like target ( bottom ) .
the dashed curves are calculated from using the cteq6 parton distribution functions at @xmath195 ( gev / c)@xmath30 .
the solid curves are the results at @xmath262 ( gev / c)@xmath30 ( from left to right ) calculated from the extended sl model described in this work.,title="fig:",width=170 ] ( left ) , @xmath5 ( center ) , and @xmath263(right ) of @xmath8 for the proton ( top ) and an i=0 deuteron - like target ( bottom ) .
the dashed curves are calculated from using the cteq6 parton distribution functions at @xmath195 ( gev / c)@xmath30 .
the solid curves are the results at @xmath262 ( gev / c)@xmath30 ( from left to right ) calculated from the extended sl model described in this work.,title="fig:",width=170 ] ( left ) , @xmath5 ( center ) , and @xmath263(right ) of @xmath8 for the proton ( top ) and an i=0 deuteron - like target ( bottom ) .
the dashed curves are calculated from using the cteq6 parton distribution functions at @xmath195 ( gev / c)@xmath30 .
the solid curves are the results at @xmath262 ( gev / c)@xmath30 ( from left to right ) calculated from the extended sl model described in this work.,title="fig:",width=170 ] ( left ) , @xmath5 ( center ) , and @xmath263(right ) of @xmath8 for the proton ( top ) and an i=0 deuteron - like target ( bottom ) .
the dashed curves are calculated from using the cteq6 parton distribution functions at @xmath195 ( gev / c)@xmath30 .
the solid curves are the results at @xmath262 ( gev / c)@xmath30 ( from left to right ) calculated from the extended sl model described in this work.,title="fig:",width=170 ] ( left ) , @xmath5 ( center ) , and @xmath263(right ) of @xmath8 for the proton ( top ) and an i=0 deuteron - like target ( bottom ) .
the dashed curves are calculated from using the cteq6 parton distribution functions at @xmath195 ( gev / c)@xmath30 .
the solid curves are the results at @xmath262 ( gev / c)@xmath30 ( from left to right ) calculated from the extended sl model described in this work.,title="fig:",width=170 ] ( left ) , @xmath5 ( center ) , and @xmath263(right ) of @xmath8 for the proton ( top ) and an i=0 deuteron - like target ( bottom ) .
the dashed curves are calculated from using the cteq6 parton distribution functions at @xmath195 ( gev / c)@xmath30 .
the solid curves are the results at @xmath262 ( gev / c)@xmath30 ( from left to right ) calculated from the extended sl model described in this work.,title="fig:",width=170 ] of @xmath22 as a function of the scaling variable @xmath264 calculated at @xmath265 ( gev / c)@xmath30 ( from left to right ) from the sl model .
the dashed curve is calculated from using the cteq6 parton distributions at @xmath266 ( gev / c)@xmath30.,width=226 ] ( defined in ref.@xcite ) vs @xmath267 of @xmath268 reaction at @xmath269 ( gev / c)@xmath30 and @xmath270 gev predicted by the sl model are compared with the data of joo et al .
@xmath271 ) are the pion angles in the center of mass frame of the final @xmath105 system.,width=377 ] ( defined in ref.@xcite ) vs @xmath272 for @xmath273 ( top ) and @xmath268 ( bottom ) at @xmath274 predicted by the sl model are compared with the data of joo et al .
@xmath275 is the pion scattering angle in the center of mass frame of the final @xmath105 system.,width=377 ] ( upper half ) ) and @xmath276(lower half ) ( as defined in ref.@xcite ) of @xmath273 reaction as a function of @xmath277 integrated over @xmath278 at @xmath270 gev and + @xmath279 ( gev / c)@xmath30 @xmath280 q@xmath30 @xmath281 ( gev / c)@xmath30 ( left ) , @xmath282 ( gev / c)@xmath30 @xmath280 q@xmath30 @xmath283 ( gev / c)@xmath30 ( right ) .
@xmath284 ) are the pion angles in the center of mass frame of the final @xmath105 system .
the solid curves are the predictions of the sl model .
the data are from biselli et al .
@xcite , title="fig:",width=226 ] ( upper half ) ) and @xmath276(lower half ) ( as defined in ref.@xcite ) of @xmath273 reaction as a function of @xmath277 integrated over @xmath278 at @xmath270 gev and + @xmath279 ( gev / c)@xmath30 @xmath280 q@xmath30 @xmath281 ( gev / c)@xmath30 ( left ) , @xmath282 ( gev / c)@xmath30 @xmath280 q@xmath30 @xmath283 ( gev / c)@xmath30 ( right ) .
@xmath284 ) are the pion angles in the center of mass frame of the final @xmath105 system .
the solid curves are the predictions of the sl model .
the data are from biselli et al .
@xcite , title="fig:",width=226 ] ( upper half ) ) and @xmath276(lower half ) ( as defined in ref.@xcite ) of @xmath273 reaction as a function of @xmath277 integrated over @xmath278 at @xmath270 gev and + @xmath279 ( gev / c)@xmath30 @xmath280 q@xmath30 @xmath281 ( gev / c)@xmath30 ( left ) , @xmath282 ( gev / c)@xmath30 @xmath280 q@xmath30 @xmath283 ( gev / c)@xmath30 ( right ) .
@xmath284 ) are the pion angles in the center of mass frame of the final @xmath105 system .
the solid curves are the predictions of the sl model .
the data are from biselli et al .
@xcite , title="fig:",width=226 ] ( upper half ) ) and @xmath276(lower half ) ( as defined in ref.@xcite ) of @xmath273 reaction as a function of @xmath277 integrated over @xmath278 at @xmath270 gev and + @xmath279 ( gev / c)@xmath30 @xmath280 q@xmath30 @xmath281 ( gev / c)@xmath30 ( left ) , @xmath282 ( gev / c)@xmath30 @xmath280 q@xmath30 @xmath283 ( gev / c)@xmath30 ( right ) .
@xmath284 ) are the pion angles in the center of mass frame of the final @xmath105 system .
the solid curves are the predictions of the sl model .
the data are from biselli et al .
@xcite , title="fig:",width=226 ] ( upper half ) ) and @xmath276(lower half ) ( as defined in ref.@xcite ) of @xmath273 reaction as a function of @xmath278 integrated over @xmath285 at @xmath270 gev and + @xmath279 ( gev / c)@xmath30 @xmath280 q@xmath30 @xmath281 ( gev / c)@xmath30 ( left ) , @xmath282 ( gev / c)@xmath30 @xmath280 q@xmath30 @xmath283 ( gev / c)@xmath30 ( right ) .
@xmath284 ) are the pion angles in the center of mass frame of the final @xmath105 system .
the solid curves are the predictions of the sl model .
the data are from biselli et al .
@xcite.,title="fig:",width=226 ] ( upper half ) ) and @xmath276(lower half ) ( as defined in ref.@xcite ) of @xmath273 reaction as a function of @xmath278 integrated over @xmath285 at @xmath270 gev and + @xmath279 ( gev / c)@xmath30 @xmath280 q@xmath30 @xmath281 ( gev / c)@xmath30 ( left ) , @xmath282 ( gev / c)@xmath30 @xmath280 q@xmath30 @xmath283 ( gev / c)@xmath30 ( right ) .
@xmath284 ) are the pion angles in the center of mass frame of the final @xmath105 system .
the solid curves are the predictions of the sl model .
the data are from biselli et al .
@xcite.,title="fig:",width=226 ] ( upper half ) ) and @xmath276(lower half ) ( as defined in ref.@xcite ) of @xmath273 reaction as a function of @xmath278 integrated over @xmath285 at @xmath270 gev and + @xmath279 ( gev / c)@xmath30 @xmath280 q@xmath30 @xmath281 ( gev / c)@xmath30 ( left ) , @xmath282 ( gev / c)@xmath30 @xmath280 q@xmath30 @xmath283 ( gev / c)@xmath30 ( right ) .
@xmath284 ) are the pion angles in the center of mass frame of the final @xmath105 system .
the solid curves are the predictions of the sl model .
the data are from biselli et al .
@xcite.,title="fig:",width=226 ] ( upper half ) ) and @xmath276(lower half ) ( as defined in ref.@xcite ) of @xmath273 reaction as a function of @xmath278 integrated over @xmath285 at @xmath270 gev and + @xmath279 ( gev / c)@xmath30 @xmath280 q@xmath30 @xmath281 ( gev / c)@xmath30 ( left ) , @xmath282 ( gev / c)@xmath30 @xmath280 q@xmath30 @xmath283 ( gev / c)@xmath30 ( right ) .
@xmath284 ) are the pion angles in the center of mass frame of the final @xmath105 system .
the solid curves are the predictions of the sl model .
the data are from biselli et al .
@xcite.,title="fig:",width=226 ] ( left ) and @xmath10 ( right ) for the proton target .
the dashed curves are from the fits to the deeply inelastic scattering data as explained in the text .
the solid curves are the results at @xmath286 ( gev / c)@xmath30 ( from left to right ) calculated from the sl model .
the data for @xmath9 are from fatemi et al .
@xcite.,title="fig:",width=226 ] ( left ) and @xmath10 ( right ) for the proton target .
the dashed curves are from the fits to the deeply inelastic scattering data as explained in the text .
the solid curves are the results at @xmath286 ( gev / c)@xmath30 ( from left to right ) calculated from the sl model .
the data for @xmath9 are from fatemi et al .
@xcite.,title="fig:",width=226 ] ( solid curve ) and @xmath287(dotted curve ) ( defined in eqs.(53)-(54 ) ) of @xmath288 as a function of invariant mass @xmath35 calculated from the extended sl model described in this work .
the results are for incident electron energy @xmath246 gev and scattering angle @xmath289.,width=226 ] ( solid curves ) and @xmath244(dotted curves ) ( defined in eqs.(53)-(54 ) ) of @xmath288 as a function of the incident electron energy @xmath290 calculated from the extended sl model described in this work .
the results are for invariant mass @xmath204 gev and electron scattering angles @xmath291(left ) and @xmath289(right ) .
, title="fig:",width=226 ] ( solid curves ) and @xmath244(dotted curves ) ( defined in eqs.(53)-(54 ) ) of @xmath288 as a function of the incident electron energy @xmath290 calculated from the extended sl model described in this work .
the results are for invariant mass @xmath204 gev and electron scattering angles @xmath291(left ) and @xmath289(right ) .
, title="fig:",width=226 ] ( defined in eq .
( 54 ) ) of @xmath288 as a function of @xmath131 evaluated with an invariant mass @xmath292 and electron scattering angle @xmath289 .
the solid curve is from the extended sl model described in this work .
the dashed curve is from turning off the pion cloud effect on the axial @xmath1-@xmath0 transition .
the dotted curve near the bottom is from keeping only the non - resonant amplitude in the calculation.,width=302 ] -dependence of the scaled parity violating asymmetry @xmath293 of @xmath254 calculated from the extended sl model described in this work .
the results are for @xmath204 gev and incident electron energy e= 0.8 ( solid curve ) , 1.5 ( dotted ) , 2.0 ( dashed ) , 4.0 ( dash - dotted ) , width=302 ] of @xmath254 calculated from the extended sl model described in this work .
the solid ( dashed ) curve is from using the axial @xmath1-@xmath0 form factor @xmath294 of ref.@xcite ( refs.@xcite ) .
the results are for invariant @xmath35=1.232 gev and electron angle @xmath289.,width=302 ] of @xmath254 as a function of the scaling variable @xmath264 predicted by the extended sl model described in this work ( solid curves ) is compared with that calculated from eqs.(55)-(59 ) using the cteq6 parton distribution functions ( dashed curves ) .
the results are for @xmath295 gev , @xmath296 ( gev / c)@xmath30 ( top left ) , @xmath297 gev , @xmath298 ( gev / c)@xmath30 ( top right ) , @xmath299 gev , @xmath300 ( gev / c)@xmath30 ( bottom left ) , and @xmath301 gev , @xmath195 ( gev / c)@xmath30 ( bottom right).,title="fig:",width=226 ] of @xmath254 as a function of the scaling variable @xmath264 predicted by the extended sl model described in this work ( solid curves ) is compared with that calculated from eqs.(55)-(59 ) using the cteq6 parton distribution functions ( dashed curves ) .
the results are for @xmath295 gev , @xmath296 ( gev / c)@xmath30 ( top left ) , @xmath297 gev , @xmath298 ( gev / c)@xmath30 ( top right ) , @xmath299 gev , @xmath300 ( gev / c)@xmath30 ( bottom left ) , and @xmath301 gev , @xmath195 ( gev / c)@xmath30 ( bottom right).,title="fig:",width=226 ] of @xmath254 as a function of the scaling variable @xmath264 predicted by the extended sl model described in this work ( solid curves ) is compared with that calculated from eqs.(55)-(59 ) using the cteq6 parton distribution functions ( dashed curves ) .
the results are for @xmath295 gev , @xmath296 ( gev / c)@xmath30 ( top left ) , @xmath297 gev , @xmath298 ( gev / c)@xmath30 ( top right ) , @xmath299 gev , @xmath300 ( gev / c)@xmath30 ( bottom left ) , and @xmath301 gev , @xmath195 ( gev / c)@xmath30 ( bottom right).,title="fig:",width=226 ] of @xmath254 as a function of the scaling variable @xmath264 predicted by the extended sl model described in this work ( solid curves ) is compared with that calculated from eqs.(55)-(59 ) using the cteq6 parton distribution functions ( dashed curves ) .
the results are for @xmath295 gev , @xmath296 ( gev / c)@xmath30 ( top left ) , @xmath297 gev , @xmath298 ( gev / c)@xmath30 ( top right ) , @xmath299 gev , @xmath300 ( gev / c)@xmath30 ( bottom left ) , and @xmath301 gev , @xmath195 ( gev / c)@xmath30 ( bottom right).,title="fig:",width=226 ] | a dynamical model@xcite of electroweak pion production reactions in the @xmath0(1232 ) region has been extended to include the neutral current contributions for examining the local quark - hadron duality in neutrino - induced reactions and for investigating how the axial @xmath1-@xmath0 form factor can be determined by the parity violating asymmetry of @xmath2 reactions .
we first show that the recent data of @xmath3 structure functions @xmath4 and @xmath5 , which exhibit the quark - hadron duality , are in good agreement with our predictions . for possible future experimental tests ,
we then predict that the structure functions @xmath4 , @xmath5 , and @xmath6 for @xmath7 and @xmath8 processes also show the similar quark - hadron duality . the spin dependent structure functions @xmath9 and @xmath10 of @xmath3 have also been calculated from our model .
it is found that the local quark - hadron duality is not seen in the calculated @xmath9 and @xmath10 , while our results for @xmath9 and some polarization observables associated with the exclusive @xmath11 and @xmath12 reactions are in reasonably good agreement with the recent data . in the study of parity violating asymmetry @xmath13 of @xmath2 reactions ,
the relative importance between the nonresonant mechanisms and the @xmath0 excitation is investigated by taking into account the unitarity condition .
predictions are made for using the data of @xmath13 to test the axial @xmath1-@xmath0 form factors determined previously in the studies of @xmath14 reactions .
the predicted asymmetry @xmath13 are also compared with the parton model predictions for future experimental investigations of quark - hadron duality . |
accurate knowledge of the properties of the stationary states of glue in the presence of the simplest of color sources , that of a static quark and antiquark separated by some distance @xmath0 , is an important stepping stone on the way to understanding confinement .
it is generally believed that at large @xmath0 , the linearly - growing ground - state energy of the glue is the manifestation of the confining flux whose fluctuations can be described in terms of an effective string theory .
the lowest - lying excitations are then the goldstone modes associated with spontaneously - broken transverse translational symmetry .
expectations are less clear for small @xmath0 .
the determination of the energies of glue in the presence of a static quark - antiquark pair is also the first step in the born - oppenheimer treatment of conventional and hybrid heavy - quark mesons@xcite .
a better understanding of hybrid quarkonium should provide valuable insight into the nature of light hybrid mesons , which are currently of great experimental and theoretical interest .
even the simplest property , the energy spectrum , of the stationary states of glue interacting with a static quark - antiquark pair is not accurately known .
the main goal of this work is to remedy this . here
, we present , for the first time , a comprehensive determination of the low - lying spectrum of gluonic excitations in the presence of a static quark - antiquark pair . in this initial study , the effects of light quark - antiquark pair creation are ignored .
a few of the energy levels for @xmath0 less than 1 fm have been studied before@xcite .
our results for these quantities have significantly improved precision , and we have extended the range in @xmath0 to 4 fm . most of the energy levels presented here have never been studied before . some of our results were previously reported@xcite .
we adopt the standard notation from the physics of diatomic molecules and use @xmath2 to denote the magnitude of the eigenvalue of the projection @xmath3 of the total angular momentum @xmath4 of the gluons onto the molecular axis @xmath5 .
the capital greek letters @xmath6 are used to indicate states with @xmath7 , respectively .
the combined operations of charge conjugation and spatial inversion about the midpoint between the quark and the antiquark is also a symmetry and its eigenvalue is denoted by @xmath8 .
states with @xmath9 are denoted by the subscripts @xmath10 ( @xmath11 ) .
there is an additional label for the @xmath12 states ; @xmath12 states which are even ( odd ) under a reflection in a plane containing the molecular axis are denoted by a superscript @xmath13 @xmath14 .
hence , the low - lying levels are labelled @xmath15 , @xmath16 , @xmath17 , @xmath18 , @xmath19 , @xmath20 , @xmath21 , @xmath22 , and so on . for convenience
, we use @xmath23 to denote these labels in general .
the glue energies @xmath24 were extracted from monte carlo estimates of generalized wilson loops .
the expectation value of the path - ordered exponential of the gauge field along a closed loop is known as a wilson loop .
the familiar static - quark potential @xmath25 can be obtained from the large-@xmath26 behaviour @xmath27 $ ] of the wilson loop for a rectangle of spatial length @xmath0 and temporal extent @xmath26 .
the two spatial segments of such a wilson loop can be replaced by a complicated _ sum _ of spatial paths , all sharing the same starting and terminating sites , in order to extract the @xmath24 energies , as long as the sum of paths transforms as @xmath23 under all symmetry operations . using several different sums of paths ( we typically use 3 to 22 such sums ) then produces a matrix of wilson loop correlators @xmath28 .
monte carlo estimates of the @xmath28 matrices were obtained in eight simulations using an improved gauge - field action@xcite .
our use of anisotropic lattices in which the temporal lattice spacing @xmath29 was much smaller than the spatial spacing @xmath30 was crucial for resolving the glue spectrum , particularly for large @xmath0 . to hasten the onset of asymptotic behaviour
, iteratively - smeared spatial links@xcite were used in the generalized wilson loops .
the temporal segments in the wilson loops were constructed from thermally - averaged links , whenever possible , to reduce statistical noise . to facilitate the removal of finite - spacing errors by extrapolating to the continuum limit , results for several values of the lattice spacing were obtained ; simulations were done using @xmath31 , and @xmath32 fm .
two runs corresponding to the same @xmath30 but different @xmath29 were done to provide a measure of the @xmath33 errors in our results ( our action has @xmath34 errors ) .
such information is important for carrying out the @xmath35 extrapolations .
agreement of energies obtained using different quark - antiquark orientations on the lattice was used to check the smallness of finite - spacing errors and to help identify the continuum @xmath2 value corresponding to each level ( there are only discrete symmetries on the lattice ) .
for this reason , results were obtained not only for cases in which the molecular axis coincided with an axis of the lattice , but also for separation vectors such as @xmath36 and @xmath37 .
the matrices @xmath28 were reduced in the data fitting phase to single correlators and @xmath38 correlation matrices using the variational method .
the lowest - lying glue energies were then extracted from these reduced correlators by fitting a single exponential and a sum of two exponentials , the expected asymptotic forms , in various ranges @xmath39 to @xmath40 of the source - sink separation .
the two - exponential fits were used to check for consistency with the single - exponential fits , and in cases of favourable statistics , to extract the first - excited state energy in a given channel .
three additional runs on small lattices were done to verify that finite - volume errors in our results were negligible .
we confirmed the smallness of the @xmath41 renormalization for two values of the qcd coupling by extracting the ground state potential from wilson loops in different orientations .
the hadronic scale parameter @xmath42 fm was used to determine the lattice spacing@xcite .
the additive ultraviolet - divergent self - energies of the static sources were removed by expressing all of our results with respect to @xmath43 .
finite - lattice spacing errors were removed by extrapolating our simulation results for @xmath44 $ ] to the @xmath35 continuum limit .
these extrapolations were carried out by fitting all of our simulation results to an ansatz @xmath45 : a ratio of a polynomial of degree @xmath46 over a polynomial of degree @xmath47 , where @xmath48 or @xmath49 , was found to work well for the continuum limit form @xmath50 , and @xmath51 was chosen empirically to be a sum of three terms @xmath52 , @xmath53 , and @xmath54 .
all fits yielded @xmath55 near unity .
continuum @xmath2 values were easily identified in all cases but one : we were unable to distinguish between a @xmath56 and @xmath57 interpretation for the on - axis @xmath58 level .
our continuum - limit extrapolations are shown in fig . [ fig : res ] .
the ground - state @xmath15 is the familiar static - quark potential .
a linearly - rising behaviour dominates the @xmath15 potential once @xmath0 exceeds about 0.5 fm and we find no deviations from the linear form up to 4 fm .
the lowest - lying excitation is the @xmath20 .
there is definite evidence of a band structure at large @xmath0 : the @xmath59 , @xmath19 , and @xmath21 form the first band above the @xmath20 ; the @xmath17 , @xmath18 , @xmath60 , and @xmath22 form another band .
the @xmath16 is the highest level at large @xmath0 .
this band structure breaks down as @xmath0 decreases below 2 fm .
in particular , two levels , the @xmath16 and @xmath18 , drop far below their large-@xmath0 partners as @xmath0 becomes small .
note that for @xmath0 above 0.5 fm , all of the excitations shown are stable with respect to glueball decay . as @xmath0 decreases below 0.5 fm , the excited levels eventually become unstable as their gaps above the ground state @xmath15 exceed the mass of the lightest glueball .
a universal feature of any low - energy description of a thin fluctuating flux tube is the presence of goldstone excitations associated with the spontaneously - broken transverse translational symmetry .
these transverse modes have energy separations above the ground state given by multiples of @xmath1 .
the level orderings and approximate degeneracies of the gluon energies at large @xmath0 match , without exception , those expected of the goldstone modes . however , the precise @xmath61 gap behaviour is not observed ( see fig .
[ fig : res ] ) . for separations
less than 2 fm , the gluon energies lie well below the goldstone energies and the goldstone degeneracies are no longer observed .
the two @xmath62 states are in violent disagreement with expectations from a fluctuating string .
the gluon energies also can not be explained in terms of a nambu - goto string ( whose quantization in four dimensions is problematical , besides ) .
these results are rather surprising .
our results cast serious doubts on the validity of treating glue in terms of a fluctuating string for quark - antiquark separations less than 2 fm .
note that such a conclusion does not contradict the fact that the @xmath63 energy rises linearly for @xmath0 as small as 0.5 fm .
a linearly - rising term is not necessarily indicative of a string : for example , the adiabatic bag model predicts a linearly - rising ground state much before the onset of string - like behaviour , even in the spherical approximation@xcite . for @xmath0
greater than 2 fm , there are some tantalizing signatures of goldstone mode formation , yet significant disagreements still remain .
to what degree these discrepancies can be explained in terms of a distortion of the goldstone mode spectrum arising from the spatial fixation of the quark and antiquark sources is currently under investigation .
this work was supported by the u.s .
doe , grant no .
de - fg03 - 97er40546 .
1 cm p. hasenfratz , r. horgan , j. kuti , j. richard , phys .
b * 95 * , 299 ( 1980 ) .
s. perantonis and c. michael , nucl .
* b 347 * , 854 ( 1990 ) ; i.j .
ford , r.h .
dalitz , and j. hoek , phys .
b * 208 * , 286 ( 1988 ) ; n. campbell _ et al .
_ , phys .
b * 142 * , 291 ( 1984 ) .
juge , j. kuti , and c. morningstar , nucl .
b ( proc . suppl . ) * 63 * , 326 ( 1998 ) . c. morningstar and m. peardon , phys
. rev .
d * 56 * , 4043 ( 1997 ) . | the spectrum of gluon excitations in the presence of a static quark - antiquark pair is presented .
our results are obtained from computer simulations of gluons on anisotropic space - time lattices using an improved gauge - field action .
measurements for quark - antiquark separations @xmath0 ranging from 0.1 fm to 4 fm and for various orientations on the lattice are made .
discretization errors and finite volume effects are taken into account .
surprisingly , the spectrum does not exhibit the expected onset of the universal @xmath1 goldstone excitations of the effective qcd string , even for @xmath0 as large as 4 fm . |
two purely electric mechanisms have been proposed to generate spin accumulations in electronic systems with spin - orbit interaction ( soi ) .
they are current - induced spin polarization @xcite and the spin hall effect @xcite .
both phenomena appear when a spin - orbit coupled conductor is traversed by an electric current .
current - induced spin polarization manifests itself in a weak bulk polarization of the electronic spins in a direction determined by the electric current and the soi , while the spin hall effect manifests itself in a spin accumulation at the lateral edges of the sample .
though the resulting polarizations are usually rather weak , both effects have been demonstrated experimentally @xcite .
they offer an elegant alternative to spin injection and manipulation with ferromagnets , zeeman fields or by optical means , and pave the way toward spintronic devices with purely electric control of electronic spins .
a dual aspect of these magnetoelectric effects is that they are accompanied by spin currents , flowing either inside the sample or leaving it .
early analytical investigations of these spin currents have been devoted to calculations of bulk spin conductivities , but it was soon realized that the latter are not always related to physically relevant quantities .
local approaches based on classical kinetic equations have been used to investigate the conversion of magnetoelectric spin accumulations at the edge of a spin - orbit coupled sample into spin currents leaving the sample @xcite .
these quasiclassical theories have been quite successful , however they neglect coherent effects and , in their present form , are not appropriate to investigate local fluctuations and spatial correlations of spin polarization .
coherent mesoscopic effects on spin currents have recently attracted quite some theoretical attention , both analytically @xcite and numerically @xcite , in both diffusive and ballistic systems , while so far only mesoscopic fluctuations of sample - integrated spin polarizations in the diffusive regime have been investigated @xcite . at present ,
little is known about local coherent fluctuations of spin polarizations , the scale on which they occur , how they are correlated , and how they influence spin currents and their fluctuations , neither have current - induced spin polarizations in chaotic ballistic systems been investigated so far .
it is the purpose of this article to start filling these gaps , by investigating the magnetoelectric generation of spin polarizations in ballistic mesoscopic samples . to that end
we construct a _ spin - probe model _ , which connects local spin - dependent chemical potentials to those imposed on a nearby probe . in this way
, we are able to express local current - induced spin polarizations in terms of transmission coefficients deriving from the scattering matrix . focusing on chaotic ballistic mesoscopic systems ,
we then use random matrix theory @xcite to calculate the statistics of these polarizations .
we find that , while the ensemble average polarization vanishes , it fluctuates from position to position ( in a given sample ) and from sample to sample .
for sufficiently large systems , the amplitude of these fluctuations by far exceeds the magnitude of all current - induced polarizations investigated to date .
simultaneously , we show that this polarization is mostly uncorrelated with spin currents in external leads and thus can not easily be converted into a large spin current , at least not in linear response . the spin - probe set - up . an open quantum dot with spin - orbit interaction
is traversed by an electric current ( @xmath0 ) .
the magnetoelectrically generated spin accumulation is measured locally by an external probe ( @xmath1 ) where spin - orbit interaction is absent . the charge and spin - resolved electrochemical potentials of the probe
are set such that no spin nor charge current flow between the dot and the probe . when the probe is contacted to the system via a tunnel barrier of low transparency
, we argue that the spin - dependent potentials in the probe can be identified with the local spin - resolved potentials in the sample at the contact with the probe .
this allows to relate spin accumulations to spin - dependent transmission coefficients . ]
[ fig : probe_sketch ] gives a sketch of the system we focus on .
we consider a two - dimensional chaotic ballistic quantum dot connected via reflectionless transport terminals to two external electron reservoirs .
the reservoirs are biased and an electric current flows through the dot . assuming that soi is strong enough that the spin - orbit time is shorter than the mean electronic dwell time in the dot , @xmath2 , spin rotational symmetry is totally broken inside the dot .
together with chaotic boundary scattering , this leads to spin relaxation .
does the current flowing through this soi coupled , spin - relaxing system generate a spin polarization ?
on one hand , the relative positions of the contacts to the external electrodes determine an average electronic drift . in that sense
, the situation is very similar to that considered by edelstein @xcite , and one would accordingly argue that a current - induced spin polarization should emerge . on the other hand ,
however , electronic distribution functions are not well defined locally in such a system where momentum and energy relax on the scale of the system itself .
one might thus wonder how local spin - dependent potentials can be consistently defined . here
, we circumvent this difficulty by introducing an additional mobile terminal whose purpose it is to probe the local chemical potentials on the dot .
we assume that spin currents are well defined in the probe as well as in the transport terminals , which presupposes that spin relaxation processes are frozen there .
in particular , soi is present only in the dot .
the probe is assumed to carry a single - transport channel .
it is connected at one end to a macroscopic external reservoir , which defines its electrochemical and spin - dependent chemical potentials unambiguously , while its other end is contacted to the quantum dot via a tunnel barrier of transparency @xmath3 .
when @xmath4 , the probe does not perturb the dot and we can identify the spin - dependent chemical potentials on the dot with the corresponding probe potentials defined by the condition that neither charge nor spin currents flow between the probe and the dot @xcite .
the probe being mobile , it can scan the whole dot area , and being single - channeled , it has maximal resolution .
it can thus map the spin - polarization in the dot with maximal accuracy .
the _ spin - probe model _ becomes a powerful theoretical tool to calculate spin accumulations once electric and spin currents in the probe and the two transport terminals are related to chemical potentials via the scattering approach to transport @xcite .
when the probe carries a single transport channel , time - reversal symmetry and the unitarity of the scattering matrix ensure that the reflection coefficients from the probe back to itself are diagonal in spin indices , independently of @xmath5 .
this allows to directly express the electrochemical and spin - dependent chemical potentials of the probe in terms of spin - dependent transmission coefficients from the transport leads to the probe , and the chemical potentials at those leads .
this is done below in eq .
( [ eq : accu ] ) .
we additionally find that , for a single - channel probe , the dependence that transmission coefficients have on @xmath5 factorizes .
this has the important , and somewhat surprising consequence that the chemical potentials ensuring the vanishing of the currents through the probe , being given by the ratio of transmission coefficients , do not depend on @xmath5 .
this is an exact result , valid as long as one has a single - channel probe , and we stress that it is true not only on average nor to some leading order .
consequently , we are able to obtain the value of the spin - dependent potentials on the probe in the limit @xmath6 , where it is justified to identify the potentials on the dot with those on the probe , while performing the calculation for the analytically simpler case @xmath7 . our scattering theory of spin accumulations can be applied to both diffusive and ballistic mesoscopic systems in the linear response regime .
because current - induced spin polarization has not been considered in these systems so far , we specialize here to ballistic systems , and accordingly apply random matrix theory ( rmt ) @xcite to calculate average and fluctuations of the spin - resolved local chemical potentials .
we perform our calculation for @xmath7 to avoid unnecessary complications arising from the presence of a poisson kernel for @xmath8 @xcite .
we find that , within rmt , there is no current - induced local spin polarization on mesoscopic average .
however , sample - dependent or probe position - dependent fluctuations make local spin - dependent chemical potentials fluctuate with a variance @xmath9 given by the voltage bias @xmath10 between the two transport leads and the number @xmath11 of channels each of them carries .
the above formula gives the leading order contribution in @xmath12 .
the exact formula is given below in eq .
( [ eq : var_mup ] ) . within rmt ,
the variance is the same for @xmath13 and @xmath14 .
the transmission coefficients from the transport terminals to the probe are elements of a scattering matrix which depends on the position of the probe .
therefore , as the probe scans the sample , these coefficients fluctuate .
below , we conjecture and numerically confirm that , within rmt , moving a single - channel terminal by a single fermi wavelength is sufficient to decorrelate transmission coefficients to and from that terminal .
this has for consequence that the total polarization integrated over the entire dot , while still averaging to zero , acquires a sample - to - sample variance given by the single - channel local variance multiplied by the total number @xmath15 of uncorrelated probe positions .
thus , @xmath16 because @xmath17 gives the chemical potential difference between the two possible spin orientation along the spin axis @xmath18 , dividing it by twice the level spacing converts it into a number of polarized spins in that direction .
this gives the typical number of polarized spins in a given sample as @xmath19 with the dot level spacing @xmath20 , the width @xmath21 of the transport terminals and the linear dot size @xmath22 .
( [ eq : typaccu ] ) is one of our main results .
it predicts that linear electric transport across a typical , ballistic mesoscopic quantum dot generates a giant polarization of the electronic spins inside the dot in the semiclassical limit @xmath23 , and for a large enough dwell time @xmath24 .
this typical polarization is in particular much larger than the average magnetoelectric polarization predicted by edelstein @xcite @xmath25 in a diffusive dot with elastic mean free path @xmath26 and with rashba soi whose strength @xmath27 is usually only a fraction of @xmath28 .
it is also much larger than the typical duckheim - loss polarization @xcite , @xmath29 predicted for @xmath30 . despite its magnitude , @xmath31 can not easily be converted into a large spin current . below we demonstrate this by calculating the correlator of the spin polarization with the spin current in external leads , and show that this correlator vanishes unless the probe is within a fermi wavelength of the contact to the terminal .
in other words , rmt predicts that only accumulations right at the boundary of the sample can be extracted and converted into spin currents in the linear regime , which is consistant with the universal fluctuations of spin currents found in refs .
it is at present unclear if more spin current can be extracted in nonlinear regimes of transport .
having summarized our method and main findings , we now proceed to present our theory more formally . in the next section [ section : scatt ]
, we construct a scattering theory of current - induced spin polarization . in section [ section :
rmt ] , we focus on ballistic systems and apply rmt to calculate the mean and variance of the local as well as of the sample - integrated polarizations .
we further demonstrate the power of our approach by calculating correlators of spin polarizations with spin currents .
section [ section : numerics ] is devoted to a numerical proof - of - principle of our theory and a quantitative confirmation of our main result , eq .
( [ eq : typaccu ] ) .
we conclude with a brief discussion of our results and of future challenges and demonstrate the @xmath5-independence of the probe voltages for @xmath32 in an appendix .
in its standard form , the scattering approach to transport gives a linear relation between currents and voltages measured in external leads @xcite .
it applies equally well to diffusive and ballistic systems .
extended to take account of spin currents and chemical potentials in a multiterminal geometry , the relation reads @xmath33 where greek letters label spin indices and latin letters label electric terminals . here , @xmath34 gives the number of transport channels in terminal @xmath35 , @xmath36 the transparency of the contact between that terminal and the dot , @xmath37 gives the electrochemical potential in terms of the fermi energy @xmath38 and the applied voltage @xmath39 , and @xmath40 give the chemical potential difference between the two possible spin orientation along the spin axis @xmath18 @xcite .
we further defined @xmath41 and @xmath42 as the charge current and the three components of the spin current vector respectively , and introduced the spin - dependent transmission coefficients @xcite @xmath43,\ ] ] where @xmath44 , @xmath45 are pauli matrices and @xmath46 is the identity matrix . the trace in eq .
( [ tr_coeff_spin ] ) is taken over the spin degree of freedom and @xmath47 is a 2@xmath482 matrix of spin - dependent transmission amplitudes from channel @xmath49 in lead @xmath50 to channel @xmath51 in lead @xmath35 .
we are interested in the mesoscopic counterpart of the magnetoelectric geometry of ref .
@xcite , and consider a quantum dot connected to two external transport terminals @xmath22 and @xmath52 .
an electric bias voltage @xmath10 is applied between the terminals , @xmath53 , where we further assume that there is no spin accumulation , @xmath54 , @xmath55 . to measure local chemical potentials
, we introduce an external probe with controllable chemical potentials , which scans the dot surface .
the situation is sketched in fig . [
fig : probe_sketch ] .
the chemical potentials in the probe are determined by the condition that no charge nor spin current flows through it , i.e. they are solutions of @xmath56 and depend a priori on the transparency of the contact between the probe and the dot .
( [ eq : stm ] ) determines the local chemical potentials @xmath57 as a function of the electrochemical potentials @xmath58 imposed on the longitudinal leads .
we consider a single - channel probe @xmath59 , which not only delivers the sharpest resolution of the dot s chemical potentials landscape , but moreover ensures that the spin reflection matrix @xmath60 from the probe to itself is _ exactly _ ( i.e. not only on average ) diagonal in spin indices .
this follows from time - reversal symmetry , current conservation and gauge invariance @xcite , which additionally require @xmath61 , @xmath62 .
one then straightforwardly obtains the chemical potentials in the probe as @xmath63 e v \big/ 2 [ \mathcal{t}^{(00)}_{pl } + \mathcal{t}^{(00)}_{pr}].\ ] ] in particular , for @xmath64 , the electrochemical potential in the probe is given by @xmath65 e v \big/4 \gamma_p \ , , \ ] ] which reproduces a result first derived in ref .
@xcite . in diffusive samples , this gives a linear average decay of @xmath66 from @xmath67 to @xmath68 as the probe is moved across the sample from regions more easily accessible from the left lead ( where @xmath69 ) to regions which are more easily accessible from the right .
we next make the key observation that the @xmath5-dependence of the transmission coefficients @xmath70 between the probe and the transport terminals @xmath71 exactly factorizes when the probe is single - channeled .
this remarkable property is apparently mentioned here for the first time and is discussed in some detail in appendix a. consequently , the chemical potentials in eq .
( [ eq : accu ] ) remain the same , regardless of the transparency of the tunnel barrier between the dot and the probe , when the latter carries only a single channel .
our aim being to obtain the current - induced spin polarization and thus the spin - dependent chemical potentials on the dot , we are quite naturally interested in the limit @xmath72 , where the probe does not perturb the dot and accordingly the spin - dependent chemical potentials on the dot equal those on the probe .
the observation we just made allows us to access that limit via an analytically simpler calculation at @xmath7 which we present in the next section . to get the total spin polarization , we scan the probe over the whole sample .
the transmission amplitudes @xmath47 in eq .
( [ tr_coeff_spin ] ) are elements of a scattering matrix that depends parametrically on the probe position . for a wide probe , @xmath73 , we argue semiclassically that the correlator @xmath74 decays to zero when the distance between @xmath1 and @xmath75 is of order the width of the probe .
this is so , because beyond that , correlated escapes of pairs of trajectories no longer contribute @xcite .
we next make the leap of faith that this semiclassical argument can be extended to the case of present interest of a single - channel probe , for which we accordingly conjecture that two measurements performed a distance @xmath76 apart deliver uncorrelated outcomes . to the best of our knowledge ,
this problem has not been investigated yet , and we will numerically confirm the validity of this conjecture below . further assuming a sharp decay of the correlator at @xmath77 , we obtain that the sample - integrated chemical potentials are given by the sum @xmath78 over @xmath15 uncorrelated probe positions on the surface of our two - dimensional quantum dot .
having expressed local and sample - integrated polarizations in terms of transmission coefficients , eq .
( [ eq : accu ] ) , we next proceed to evaluate them quantitatively .
we focus on ballistic chaotic cavities , and accordingly use rmt to calculate the average and variance of @xmath57 and @xmath17 .
this calculation is easier for @xmath7 , since then the ensemble - average scattering matrix vanishes @xcite . because of the factorization property of the @xmath5-dependence of transmission coefficients mentioned above , the spin - dependent potentials on the dot are equal to those on the probe for @xmath6 , which in their turn remain unchanged as @xmath79 .
thus we can extract the dot s spin - dependent potentials @xmath80 at point @xmath81 underneath the probe from a calculation at @xmath7 , because @xmath82 .
because we are interested in systems with time reversal symmetry but totally broken spin rotational symmetry , we take the scattering matrix as an element of the circular symplectic ensemble of random matrices @xcite . rmt does not depend on the exact form of soi , but only requires that soi totally breaks spin rotational symmetry .
our results are thus expected to apply for rashba and dresselhaus soi equally well , as long as one has a ballistic chaotic quantum dot with totally broken spin rotational symmetry . for @xmath32
, we get from ref .
@xcite @xmath83}{n_{l } + n_{r}+1/2 } \ , , \ ] ] from which we conclude , together with eq .
( [ eq : accu ] ) , that @xmath84 rmt thus predicts the vanishing of the spin polarization on average .
the latter is understood in the standard mesoscopic sense , as an average over changes in the dot shape , the overall chemical potential or the position of the leads .
chemical potentials nevertheless exhibit spatial fluctuations in a single sample , whose magnitude can be estimated by calculating @xmath85 . of further interest
are also mesoscopic fluctuations of the sample - integrated chemical potentials whose magnitude we analyze with @xmath86 . to calculate the variance of @xmath57 in eq .
( [ eq : accu ] ) , we take the variance of the numerator and divide it by the squared average of the denominator , a procedure which neglects subdominant corrections @xmath87 .
the variance of the spin - dependent transmission coefficients have been calculated in ref .
@xcite . for @xmath88
, we have @xmath89}{n_{\rm t } ( 2 n_{\rm t}-1 ) ( 2 n_{\rm t}-3 ) } \ , , \\ \label{eq : covar } { \rm covar } ( \mathcal{t}^{(\beta 0)}_{pl } , \mathcal{t}^{(\beta 0)}_{pr } ) = \frac{-4 n_p n_l n_r}{n_{\rm t } ( 2 n_{\rm t}-1 ) ( 2 n_{\rm t}-3 ) } \ , , \end{aligned}\ ] ] with @xmath90 . for our case with @xmath32
, we obtain @xmath91 for @xmath92 , eq .
( [ eq : var_mup ] ) goes into eq.([eq : varmup_n ] ) . with the further assumptions , discussed in the previous section , that ( i ) spin relaxation processes
are frozen in the probe and ( ii ) the correlator @xmath93 sharply decays when the distance between the probe positions @xmath1 and @xmath75 reaches the fermi wavelength , we get the typical total spin - dependent chemical potentials @xmath94 as justified above , we identify these potentials with those on the dot . assuming that they are generated by a spin polarization , we convert them into the variance of the spin polarization by dividing them with @xmath95 . for @xmath96 , we obtain the variance of the total number of polarized spins on the dot as @xmath97 which , for @xmath98 , gives eq .
( [ eq : typaccu ] ) . the validity of eqs .
( [ eq : var_mup][eq : var_tot ] ) is numerically confirmed below in fig .
[ fig : accu_tot ] .
the power of our scattering theory is perhaps even more evident when one realizes that it allows to calculate correlation functions between local spin polarizations inside the sample and spin currents in the leads .
the spin current in lead @xmath22 is given by @xmath99-\frac{e}{h } \sum_\beta \mathcal{t}_{lp}^{(\alpha \beta ) } \mu_p^{(\beta ) } \ , , \end{aligned}\ ] ] and each of these three terms gives a separate contribution to the correlator @xmath100 .
for @xmath101 , the contribution of the third term vanishes because it contains the average of three spin - dependent transmission coefficients .
furthermore , rmt gives @xmath102 if @xmath103 .
therefore , for @xmath101 , the first two terms do not contribute either , unless the probe is placed in the immediate vicinity , i.e. within a single fermi wavelength of the contact to the @xmath22 lead . in this case , nonvanishing covariances appear in @xmath104 , see eq .
( [ eq : covar ] ) , and one obtains @xmath105 where @xmath106 requires that the probe be placed within @xmath77 of the boundary between the soi - coupled dot and the lead . summing over all possible probe positions , one obtains a universal correlator @xmath107 for @xmath108 , which is consistent with the universality of spin currents fluctuations @xcite .
rmt thus predicts that coherent polarizations correlate with the spin currents only if they are right at the sample exit .
we numerically illustrate the validity of our scattering theory of spin polarizations and confirm our main prediction , eqs .
( [ eq : var_mup]-[eq : var_tot ] ) .
we use tight - binding models defined by the hamiltonian @xmath109 where the operators @xmath110 ( @xmath111 ) create ( destroy ) a spin-@xmath112 fermion with spin @xmath113 and @xmath114 indicate elements of pauli matrices . to give a proof - of - principle for our approach
, we consider a diffusive model , while quantitative confirmations of our theory are solely based on the rmt model . for the diffusive model , @xmath115 labels a site on a two - dimensional lattice of size @xmath116 with spacing @xmath117 , giving the size @xmath118 of the hamiltonian matrix .
we further take @xmath119 + w_i \delta_{\alpha,0 } \delta_{i , j}$ ] , corresponding to an on - site disorder potential with randomly distributed @xmath120 $ ] , a nearest - neighbor hopping @xmath121 and a rashba spin - orbit interaction @xmath122 .
these parameters give a spin - orbit length @xmath123 and an elastic mean free path @xmath124 close to half filling . for the rmt model , we take @xmath125 to be randomly distributed quaternions @xcite , with time - reversal symmetry requiring @xmath126 , and @xmath127 for @xmath128 . generally , the indices @xmath129 have no particular physical meaning in rmt and the size @xmath130 of the hamiltonian matrix is a free parameter . here
we attach the system to external leads , and also consider @xmath35 and @xmath50 to label sites on a two - dimensional lattice whose hamiltonian has an infinite - range random hopping .
once the sharp decay of spin - spin correlations is confirmed , @xmath131 gives the total number of uncorrelated probe positions .
for both models , all our simulations fix the fermi energy slightly below half - filling . to construct the scattering matrix , we follow the method of ref .
we introduce a @xmath132 projection matrix @xmath21 which connects the system to external leads .
this matrix has only one nonzero element per row , at a spot determined by the coupling between the system and the leads .
for an ideal contact , all nonzero matrix elements of @xmath21 are unity .
the @xmath133 scattering matrix @xmath134 associated with @xmath135 is given by @xmath136 with the overall fermi energy @xmath38 .
the transmission coefficients of eq .
( [ tr_coeff_spin ] ) are straightforwardly obtained from @xmath134 . for the diffusive model , this method is certainly not the most efficient numerically .
we still rely on it , because for the rmt model , the infinite - range hopping renders the implementation of recursive green s function algorithms problematic .
note that it is necessary to first generate random hamiltonians from which random scattering matrices are constructed in order to consistently define probe positions and investigate whether spin - spin correlations decay sharply beyond a fermi wavelength . to investigate the decay of these correlations , we generate a random @xmath135 and take @xmath137 where for each choice of @xmath21 , only a single component of the column vector @xmath138 ( whose index indicates its length ) is nonzero and equal to one .
the position of that nonzero component determines the position of the probe on the two - dimensional lattice .
our hamiltonians are defined on rectangular two - dimensional lattices , and we contact the transport modes at two of its edges , which is represented by the two identity matrices in eq .
( [ eq : w ] ) .
we first show numerical proof - of - principle for the spin - probe mode . in fig .
[ fig : local_accumul ] , we present data for the local spin polarizations @xmath57 , @xmath18 .
the three panels clearly show that , for transport in the @xmath139-direction and a rashba soi , there is a finite average magnetoelectrically induced polarization along the @xmath140-axis only , as predicted in ref .
furthermore , there are local ( as well as sample - to - sample ) fluctuations of the spin polarization in all directions , as predicted in ref .
all accumulations vanish on average in the rmt model , moreover spatial fluctuations occur on the smallest possible scale of a single site , as expected , and we therefore do not show them here .
we next confirm our conjecture of a sharp decay of correlations , and to that end , we calculate the spin - spin correlation function @xmath141 . for both diffusive as well as rmt models we found no significant anisotropy ,
i.e. @xmath142 was the same for @xmath143 in the @xmath139 or @xmath140 direction .
this is of course expected for the rmt model , but would deserve some more investigations in the diffusive case .
the fermi wavelength being equal to the lattice spacing close to half filling , one expects @xmath144 for the rmt model .
this is confirmed in the inset of fig .
[ fig : accu_tot ] , where we additionally see that @xmath145 is also sharply peaked at @xmath146 in the diffusive model , with some correlation persisting over few , @xmath1472 - 3 sites .
this sharp decay of @xmath145 in the diffusive model suggests that our conjecture of a fast decay of the correlator @xmath148 applies beyond rmt .
this also deserves further investigations .
main panel : numerically obtained variance of the sample - integrated spin - dependent potential @xmath149 in the rmt model ( circles ) vs. the theoretical prediction of eq .
( [ eq : vars_tot ] ) ( dashed line ) for @xmath150 $ ] and @xmath151 $ ] .
deviations from the theory occur at smaller number of channels , for which fluctuations in the denominator of eq .
( [ eq : accu ] ) can no longer be neglected .
the behavior of @xmath152 and @xmath153 is the same as the one of @xmath149 .
inset : spin - spin correlation function @xmath154 for the rmt model ( black circles ) and the diffusive model ( red squares ) with @xmath155 , @xmath156 . in both cases , we find that the correlator does not depend on the spin orientation @xmath157 .
, title="fig : " ] + finally focusing on the rmt model , fig . [ fig : accu_tot ] confirms our main results , eq .
( [ eq : var_mup]-[eq : var_tot ] ) , by showing the scaling of the variance of the sample - integrated spin - dependent chemical potential in symmetric samples @xmath158 as a function of @xmath11 and the size @xmath130 of the random matrix hamiltonian
. small deviations from our analytical predictions occur at small @xmath11 for which our procedure of replacing the denominator in eq .
( [ eq : accu ] ) by its average squared when calculating @xmath85 is no longer justified .
all these results numerically confirm our scattering theory of spin polarizations .
we have demonstrated how introducing an external probe allows to relate local current - induced spin polarizations in a mesoscopic system to spin - resolved transmission coefficients . focusing on ballistic systems , we used random matrix theory to compute the statistics of these polarizations .
we predicted that a ballistic quantum dot with large contacts to external transport leads but still large dwell time typically carries a spin polarization that exceeds those previously investigated by a parametrically large factor [ see eqs .
( [ eq : typaccu][eq : dl ] ) ] .
perhaps most importantly , we used this spin - probe model to calculate for the first time quantum correlations between local spin polarizations and spin currents , beyond the classical fick law .
we showed that the reported large spin polarizations correlate only weakly with spin currents , thus they can not be easily converted into spin currents and vice - versa . as disappointing as this result may be , it is consistent with the universality of spin current fluctuations @xcite .
the reported accumulations are coherent in nature , and one might wonder how they disappear as the system is coupled to external sources of noise . in this respect
it is interesting to note that for a multichannel probe , eq .
( [ eq : vars_tot ] ) becomes , for @xmath159 , @xmath160 enlarging the spin - probe damps the variance of the polarizations in a similar way as a dephasing probe damps conductance fluctuations @xcite , with however a different parametric dependence on @xmath161 .
note that the probe generating this damping is spin - conserving on time - average , as there is no spin current flowing through it . beside decoherence effects , there are a number of interesting open problems , related for instance to spatial fluctuations of spin polarizations and our assumption of a sharp decay of correlations at @xmath77 .
finding out how the polarizations correlate to spin currents in diffusive systems , as well as in true ballistic systems treated beyond rmt is also of interest .
ways to extract the large typical polarizations we reported in the nonlinear transport regime or by optical means as in ref .
@xcite should also be explored .
this work has been supported by the nsf under grant dmr-0706319 .
we acknowledge the hospitality of the basel center for quantum computing and quantum coherence at the early stage of this project .
we thank m. duckheim and d. loss for an introduction to magnetoelectric effects and discussions of their letter , ref .
@xcite , and m. bttiker for discussions on probe physics .
we demonstrate that the potentials ensuring zero current through a single - channel probe do not depend on the transparency of the contact between the probe and the system . according to eq .
( [ eq : accu ] ) , the chemical potentials in an external probe connected to a voltage - biased two - terminal cavity are given by the ratio of the difference and the sum of transmission coefficients between the transport leads and the probe .
we show that the @xmath5-dependence of the coefficients @xmath162 factorizes when @xmath32 so that @xmath5 does not affect the chemical potential .
our starting point is eq .
( [ eq : heidel ] ) for the scattering matrix , which we rewrite here , @xmath163 the matrix @xmath164 projects onto the system s states coupled to the transport leads ( @xmath165 ) and the probe ( @xmath166 ) .
we can write a perturbative series in @xmath167 for the elements of the scattering matrix connecting the leads to the probe ( @xmath168 is a channel index on the @xmath71 leads ) the transparency of the contact between probe and system is contained exclusively in the matrix @xmath1 .
we are free to choose any orthogonal basis @xmath170 for the system and pick one where the @xmath171 coupling matrix between probe and system is @xmath172 .
one then has for the @xmath173 matrix @xmath174 @xcite .
one gets @xmath175 one sees that , because the probe is single - channeled , the @xmath5-dependent part of @xmath176 factorizes , and the same obviously holds for transmission coefficients @xmath177 .
this is not the case for a multichannel probe , in which case @xmath178 is no longer unique .
thus , for @xmath32 , the electrochemical potential in eq .
( [ eq : accu_charge ] ) do not depend on the strength of the tunnel barrier .
the argument is straightforwardly extended to the spin - dependent chemical potentials in eq .
( [ eq : accu ] ) , because for a single - channel probe , the reflection coefficient @xmath179 is diagonal in spin - indices .
vasko and n.a .
prima , sov .
solid state * 21 * , 994 ( 1979 ) ; l.s .
levitov , y.v .
nazarov , and g.m .
eliashberg , sov .
jetp * 61 * , 133 ( 1985 ) ; a.g .
aronov and yu.b .
lyanda - geller , jetp lett .
* 50 * , 431 ( 1989 ) .
she for a review on the spin hall effect , see : h .- a .
engel , e.i .
rashba , and b.i .
halperin , _ theory of spin hall effects in semiconductors _
, handbook of magnetism and advanced magnetic materials , wiley and sons ( 2007 ) ; arxiv : cond - mat/0603306 .
e. saitoh , m. ueda , h. miyajima , and g. tatara , appl .
. lett . * 88 * , 182509 ( 2006 ) ; t. kimura , y. otani , t. sato , s. takahashi , and s. maekawa , phys .
98 * , 156601 ( 2007 ) ; * 98 * , 249901(e ) ( 2007 ) .
caveat1 because of coulomb interactions , such a direct relation does not necessarily exist for the electrochemical potential on the dot , see e.g. : m. bttiker , j. phys . :
matter * 5 * , 9361 ( 1993 ) .
caveat2 for @xmath18 , @xmath40 gives the chemical potential difference between the two possible spin orientations along the corresponding spin axis , while @xmath181 gives the usual electrochemical potential .
we refer to electrochemical and spin - dependent potentials collectively as `` chemical potentials '' .
note the plural .
`` chemical potential '' still retains its standard meaning .
jac06 correlated escape of pairs of classical trajectories and its influence on semiclassical transport was first discussed in : ph .
jacquod and r.s .
whitney , phys .
b. * 73 * , 195115 ( 2006 ) .
the only nonvanishing semiclassical contributions to @xmath182 for @xmath183 have trajectories with correlated escape ; ph . jacquod , unpublished ( 2008 ) . | we construct a novel scattering theory to investigate magnetoelectrically induced spin polarizations .
local spin polarizations generated by electric currents passing through a spin - orbit coupled mesoscopic system are measured by an external probe .
the electrochemical and spin - dependent chemical potentials on the probe are controllable and tuned to values ensuring that neither charge nor spin current flow between the system and the probe , on time - average . for the relevant case of a single - channel probe , we find that the resulting potentials are exactly independent of the transparency of the contact between the probe and the system .
assuming that spin relaxation processes are absent in the probe , we therefore identify the local spin - dependent potentials in the sample at the probe position , and hence the local current - induced spin polarization , with the spin - dependent potentials in the probe itself .
the statistics of these local chemical potentials is calculated within random matrix theory . while they vanish on spatial and mesoscopic average , they exhibit large fluctuations , and we show that single systems typically have spin polarizations exceeding all known current - induced spin polarizations by a parametrically large factor .
our theory allows to calculate quantum correlations between spin polarizations inside the sample and spin currents flowing out of it .
we show that these large polarizations correlate only weakly with spin currents in external leads , and that only a fraction of them can be converted into a spin current in the linear regime of transport , which is consistent with the mesoscopic universality of spin conductance fluctuations .
we numerically confirm the theory . |
supergiants are the descendants of massive o and b - type main sequence stars . unlike their main sequence progenitors ,
cool supergiants are characterized by a helium - burning core and a deep convective envelope .
due to their extended radii , low - atmospheric densities , slow rotation and long convective turnover times , supergiants provide an opportunity to study stellar magnetism at the extremes of parameter space .
in fact , observations of late - type supergiants show characteristics consistent with magnetic activity , such as luminous x - ray emission and flaring , and emission in chromospheric uv lines - phenomena suggesting the presence of dynamo - driven magnetic fields . motivated by the activity - related puzzles of late - type supergiants , the near complete lack of direct constraints on their magnetic fields , and recent success of measuring fields of red and yellow giants ( e.g. aurire et al .
2008 ) , we have initiated a program to search for direct evidence of magnetic fields in these massive , evolved stars . here
we summarize the recent results of grunhut et al .
circular polarization ( stokes @xmath0 ) spectra were obtained with the high - resolution ( r@xmath168000 ) espadons and narval spectropolarimeters at the canada - france - hawaii telescope and bernard lyot telescope , as part of a large survey investigating the magnetic properties of late - type supergiants . to date , we have observed more than 30 stars : 4 a - type stars , 8 f - type stars , 11 g - type stars , 7 k - type stars , and 3 m - type stars .
we applied the least - squares deconvolution ( lsd ; donati et al .
1997 ) technique to all our data in order to increase the s / n and detect weak zeeman signatures .
our investigation shows that many late - type supergiants host detectable stokes @xmath0 zeeman signatures , which are frequently complex .
overall , we find that approximately 1/3 of our sample reveal detectable zeeman signatures in stokes @xmath0 .
however , we find no clear differences between classical activity indicators ( such as caii h&k or h@xmath2 emission ) of those stars with or without detections .
however , we do find a weak correlation between the caii core equivalent width and the magnetic field strength for those stars with multiple observations .
in addition , we also see clear temporal variability of the stokes @xmath0 profiles for those targets with multiple observations . | we present the results of a recent survey of cool , late - type supergiants - the descendants of massive o- and b - type stars - that has systematically detected magnetic fields in these stars using spectropolarimetric observations obtained with espadons at the canada - france - hawaii telescope .
our observations reveal detectable , often complex , stokes @xmath0 zeeman signatures in least - squares deconvolved mean line profiles in a significant fraction of the observed sample of @xmath130 stars . |
cosmic strings are a special case of the vorticity phenomenon , which appears in many forms - in the flow of rivers and streams , in smoke rings , in hurricanes and tornadoes , etc .
a special case of vorticity are the _ quantum vortices _
@xcite which include vortices in superfluids , flux tubes in superconductors and cosmic strings .
global cosmic strings are analogues to vortices in superfluids while local cosmic strings are analogues to flux tubes in superconductors @xcite .
all these vortices are described by a complex wave function or scalar field with a multivalued phase function @xmath0 , such that @xmath1 over a closed contour surrounding the string . as a result
different physical quantities get quantized : the circulation of a superfluid vortex , the magnetic flux in a superconductor , the magnetic flux in a local cosmic string .
vorticity was first investigated by helmholz and kelvin who set down the foundations for the study of vorticity in hydrodynamics . in this work
we try to implement concepts born in the field of classical hydrodynamics to cosmic strings . as we shall see
the interaction between these two fields can be very rewarding .
turbulent fluid is regarded as full of vortex filaments @xcite .
quantum turbulence in superfluids has likewise been defined as a superfluid state featuring a tangle of quantum vortex filaments @xcite .
this paper is concerned with helicity of a tangle of cosmic strings .
the concept of _ fluid helicity _ was introduced by moffatt @xcite as a useful measure of the degree of linkage of ordinary fluid vortice loops .
it is defined as the volume integral of the scalar product of a velocity field @xmath2 and its curl , the vorticity @xmath3 .
following this , moffatt @xcite and others @xcite considered helicity integrals in general , i.e. the space integral over the scalar product of a vector field and its curl @xmath4 and found them to have a topological interpretation as the linking of the field lines of the divergence free field @xmath5 .
thus the conservation of helicity can be explained for systems in which the field lines may not cross each other , such as vortex lines in an inviscid fluid and the magnetic field lines in a perfectly conducting fluid @xcite . for these flows
the field lines are said to be frozen into the flow , and the helicity is a topological invariant mathematically related to the hopf invariant @xcite classifying the nontrivial homotopy classes of maps from @xmath6 to @xmath7 .
, which are isomorphic to circles in @xmath6 .
the hopf invariant is equivalent to the helicity integral of a vector field who s field lines lie tangent to the preimage circles . in this sense
the hopf invariant is a helicity integral term , but the inverse is usually not the case . ] superfluids and cosmic strings are both described by field theories of a complex field @xmath8 which obeys a nonlinear field equation @xcite , e.g. the nonlinear ginzburg - pitaevskii equation @xcite or the higgs equation .
cosmic strings and superfluid vortices both occur in the presence of spontaneous symmetry breaking @xcite . a vortex filament or cosmic string is a configuration of @xmath9 which approaches asymptotically the broken symmetry vacuum , and is characterized by a phase @xmath0 which changes by @xmath10 ( @xmath11 ) when one goes once around the filament or string axis .
@xmath12 , is termed the _ winding number of the string_. single valuedness of the field @xmath9 on the filament axis requires that @xmath13 along it .
the curves along which @xmath13 define the position of the strings . a vortex filament or cosmic string
must either be infinite in both directions , close on itself ( cosmic string loop , vortex ring ) , or terminate on a boundary of the system .
otherwise the behavior of the phase just beyond the filament s free ends would be ambiguous .
we assume throughout that all strings are closed un - knotted loops confined to a finite region .
the analogy between vortices in superfluids and global cosmic strings has inspired the construction of a _
helicity _ for global cosmic strings .
bekenstein @xcite defined a helicity for global strings , and vachaspati and field @xcite , sato and yahikozawa @xcite and sato @xcite did this for local strings .
these works tried to relate helicity for cosmic strings with the topological and geometrical structure of these strings ( such as linking , knotting , writhing and twisting ) .
in all these works bizarre physical conclusions were drawn .
bekenstein concluded that an isolated unknotted loop of global string is restricted to a plane @xcite i.e. , only planar configurations may exist .
this is strange since dynamically there seems to be no constraint on a loop accumulating a distortion continuously in an arbitrary direction .
furthermore simulations of the formation and evolution of cosmic strings show the formation of loops which are not planar @xcite .
bekenstein also claimed that the contortion of a single knot is quantized in integers .
this is also strange because it says a single knot may not wiggle freely about , but is frozen in a configuration of integer contortion . for local cosmic strings there is a relation between the electroweak magnetic helicity of a tangle of strings and the violation of baryon number conservation via what is known as the chiral anomaly @xcite .
vachaspati and field , sato and yahikozawa and sato found the helicity to change continuously with the shape of the strings , which implies that baryon number changes continuously too .
this clashes with the expectation that baryon number is quantized in integers .
we show here that the topological interpretation of helicity as the linking of field lines themselves is not fully incorporated in these works , and hence an important term was left out , which accounts for all the mentioned unacceptable conclusions .
helicity arises from internal structure within a string , determined by the twist and writhe of the string , in addition to the external relations between strings , i.e. linking and knotting .
the missing term arises from the internal structure of field lines within the strings which we name _
internal helicity_. it was previously believed @xcite that since the strings length is much greater than its width , the internal structure of the string becomes unimportant and physical quantities of interest can be averaged over the string cross - section . however , as we will see this is not always the case . in order to clarify the contribution of internal helicity , the relation of helicity to the topological structure of field lines will be further developed here .
field lines of vector fields which are divergence - free do not have endpoints .
this property allows us to examine such field structures in terms of the topology of closed curves .
the link between helicity and topological invariants of curves was first conjectured by moffatt in 1981 @xcite and then developed by berger and field in 1984 @xcite . in 1992 moffatt and ricca @xcite managed to derive the extra term , arising from the internal structure of magnetic flux tubes , directly from the field equations of motion for a magnetohydrodynamic fluid .
this is not possible when calculating the helicity of cosmic strings since their equations are second order non - linear equations , and therefore the calculation is too complicated .
however , using formulas describing curves , developed by mathematical biologists investigating the structure of dna @xcite , the internal helicity may be calculated . by adding it to the external helicity we correct the mentioned results .
in section [ sec_2 ] we review the concept of _ helicity _ of a solenoidal ( divergence free ) vector field , and dwell on the relation of the helicity to the linking of the field lines of the solenoidal field . in section [ sec_3 ]
we review the construction and calculation of the helicity for a tangle of global cosmic string loops , as a function of different geometrical and topological properties of the strings @xcite .
next we correct the previously mentioned conclusions of bekenstein by means of the internal helicity . in section [ sec_4 ]
we deduce new constraints on the permitted linkages of string loops .
we find that the linking possibilities of string loops depends on their winding numbers . in section [ sec_5 ]
we review how the electroweak helicity of local cosmic strings in the weinberg - salam model is related to the baryon number , and clarify how baryon number conservation is violated .
the baryon number is found to be quantized , in contrast to previous results .
again , we show how internal helicity accounts for the new results .
the term _ helicity _ is used in particle physics for the scalar product of the momentum and spin of a particle .
moffatt @xcite adopted this term to describe the volume integral over the scalar product of a velocity field @xmath2 and its curl , the vorticity @xmath3 @xmath14 which he named the _ helicity of the flow_. this idea stemmed from the essentially kinematical interpretation of the quantity @xmath15 .
the streamlines of the flow passing near any representative point @xmath16 in a small volume element @xmath17 are ( locally ) helices about the streamline through @xmath16 , and the contribution @xmath18 to @xmath19 from @xmath17 is positive or negative according as the screw of these helices is right - handed or left - handed .
more generally , the concept of helicity describes the volume integral over the scalar product of any general vector and its curl .
in particular , the magnetic helicity of magnetohydrodynamic flow @xmath20 had already been shown by woltjer @xcite to be conserved for a perfectly conducting fluid .
later moffatt @xcite showed that the _ fluid helicity _ of an ideal barotropic fluid is conserved as well .
both these cases share the property that the divergence free field lines ( vortex lines or magnetic field lines ) are frozen into the fluid , and that in consequence knots and linkages of the field lines are inevitably conserved .
it was through struggling to understand the physical meaning of this result that moffatt was led to the topological interpretation of helicity in terms of links and knots in convected vector fields generally , and to the characterization of helicity as a _
topological invariant _ for these cases .
moreau @xcite observed that conservation of fluid helicity can be deduced from noether s theorem .
helicity thus appears to have status comparable to energy , momentum and angular momentum in this sense .
as we shall see , unlike these quantities helicity also has a strong topological character . to demonstrate this we review the calculation of helicity for the simple case of two linked flux tubes and
show that the helicity counts the gauss linking number of the tubes @xcite .
consider two flux tubes linked once ( fig .
[ flux ] ) , carrying fluxes @xmath21 and @xmath22 respectively , of the vector field @xmath23 ( the flux is zero everywhere outside the filaments ) .
4.5 cm within each filament , the @xmath24-lines are unlinked curves which close on themselves after just one passage round the filament .
the helicity @xmath19 of the two flux tubes has the same form as the pseudo - scalar quantity described in ( [ b_helicity ] ) .
does not depend on the gauge of @xmath25 ; for if @xmath26 is replaced by @xmath27 , then @xmath19 is unchanged . ]
adopting the coulomb gauge for @xmath26 ( i.e. @xmath28 ) , and imposing the further condition @xmath29 as @xmath30 , @xmath25 is given by the biot - savart law : @xmath31 so that from ( [ b_helicity ] ) , @xmath32 each tube may be built of many infinitely small flux tubes carrying fluxes @xmath33 and @xmath34 . replacing @xmath35 by @xmath36 : allowing for the fact that each flux filament is integrated over twice once as @xmath37 and once as @xmath38 , we find that @xmath39 with @xmath40 which is the gauss formula for the linking of two curves .
the sign of @xmath41 depends on the relative orientations of the field in the two filaments ( according to the right hand rule).-lines within either tube on its own are unlinked closed curves . ]
it is simple to obtain the total helicity of the two flux tubes .
each pair of filaments ( one from each tube ) make a contribution @xmath42 to the total helicity , so that summing over all such pairs , this is now given by @xmath43 helicity has been shown to be an invariant for flows in which the divergence free field lines are frozen into the flow ( woltjer @xcite for a perfectly conducting fluid and moffatt @xcite for ideal barotropic flow ) .
combining this knowledge with ( [ link_hell ] ) , the interpretation of the invariant in terms of conservation of linkage of the ( vortex / magnetic ) field lines which are frozen in the fluid is immediate .
states of a higgs field containing a tangle of cosmic strings resemble a tangle of quantum vortex filaments which arise in quantum turbulent flow in superfluids .
they are both described by a complex wave function @xmath8 and a winding number @xmath12 .
the phase @xmath0 changes by @xmath10 when one goes once around the filament or string axes .
the curve @xmath13 defines the position of the strings and the passage from @xmath13 to the vacuum value of @xmath44 takes place over a certain lengthscale denoted by @xmath45 .
this lengthscale is very small in comparison with the strings lengths and distances between the strings , i.e. the strings are very thin . motivated by this resemblance bekenstein @xcite designed a helicity for cosmic strings in analogy with fluid helicity of a superfluid .
we first review the original calculation @xcite of the helicity of a tangle of strings , and then correct it for the left - out contribution of internal helicity . in a superfluid
the velocity is proportional to the gradient of the phase of the wave function ; bekenstein defines the vector @xmath46 where @xmath47 is taken to be a function which rises rapidly from its value at a string axis @xmath48 to near its asymptotic value , @xmath49 , over an interval of @xmath50 which corresponds to a distance of order @xmath45 .
@xmath51 is a kind of regularised velocity in that the singularity of @xmath52 on the axis is defused by the @xmath53 function ( far from the axis @xmath51 will drop inversely with the distance from it as befits a vortex velocity field ) .
we have @xmath54 where @xmath55 is the derivative of @xmath47 with respect to @xmath56 .
@xmath55 has a rather sharp peak at a distance @xmath57 from a string axis and the integral of @xmath58 over all @xmath56 is unity .
the helicity of the field @xmath51 is defined as @xmath59 unlike ordinary fluid helicity , this one vanishes .
this is clear since the vectors @xmath51 and @xmath60 are perpendicular by construction .
nevertheless , string loops ( and vortex rings ) can link , and knot , and their linkage is correlated with other topological properties of the strings to be introduced later . throughout it
is assumed that linked and knotted loops can occur over some periods with no intersections .
situations where there are intersecting strings are viewed as representing singular moments , not generic ones .
( simulations of the development of superfluid vortices @xcite and global cosmic strings @xcite confirm this assumption ) .
bekenstein develops from the zero helicity a quantitative relation between the linkage and other properties of the loops . in order to evaluate the helicity explicitly for a tangle of cosmic string loops ,
it is convenient to rewrite ( [ h ] ) as a two - point functional of @xmath51 ( this is valid if the loops are confined to a finite region @xcite ) @xmath61 = \int d^3r \int d^3r'\ , \frac{(\nabla \times { \bf v } ) \times(\nabla \times { \bf v } ) ' \cdot ( { \bf r}-{\bf r}')}{4\pi|{\bf r}-{\bf r}'|^3}.\ ] ] we follow bekenstein in the following discussion of the reduction of @xmath62 $ ] , and the conclusions that may be drawn from it . for field configurations of size large compared to @xmath45 , and where all the string axis are always well apart on scale @xmath45 , @xmath19 receives a contribution only when each of @xmath63 and @xmath64 lies very close to a string axis .
this stems from the fact that the @xmath58 function in ( [ curl_v ] ) confines the integrals to very near the axis .
hence , bekenstein assumes @xmath19 is equivalent to a double line integral taken along the string axis .
in addition the cross section of each loop is very little distorted by the presence of the other loops and nearby portions of itself and therefore is assumed to possess a line symmetry . under these assumptions and
a few manipulations bekenstein concludes that @xmath65 for the contribution of a little piece of string , where @xmath66 here is the vectorial line element along the piece of string considered . inserting ( [ c_flux ] ) into ( [ helic ] )
we obtain a double contribution to helicity from each pair of loops , because @xmath66 may progress through one loop ( call it @xmath67 ) while @xmath68 does so through the other ( @xmath69 ) and vice - versa .
in addition we obtain a contribution from each string loop alone so that @xmath70 = \frac{1}{2}\sum_{i \neq j } 2{\cal l}_{ij}(2\pi|n_i|)(2\pi|n_j| ) + \sum_i { \cal w}_i(2\pi|n_i|)^2.\ ] ] where @xmath71 is the _ gauss linking number _
( [ gauss_link ] ) of the two loops @xmath67 and @xmath69 and @xmath72 is the _ writhing _ number of loop @xmath73 ( fuller @xcite ) . )
differs from the original one in @xcite by a factor of @xmath74 which was missed in the previous calculation . ] while @xmath75 is a topological invariant of the linking of the string loops and must be an integer , @xmath76 is not a topological invariant .
it changes with the shape of the loop and can obtain any value .
@xmath76 measures the deviation of a loop from its plane ; the writhe of a planar loop is zero .
when all loops are confined to a finite region , @xmath77 , and ( [ hel_eq ] ) gives @xmath78 we thus obtain a constraint connecting the linking of a collection of loops with their geometrical shapes .
particularly simple is the case of an isolated loop . then @xmath79
for such a loop we must have @xmath80 too . according to this result
_ an isolated un - knotted loop must be a planar loop at any instant _
this is hard to accept since dynamically there seems to be no constraint on a loop accumulating a distortion continuously in an arbitrary direction .
furthermore simulations of the formation and evolution of cosmic strings @xcite show the formation of loops which are not planar ! similarly , according to ( [ const ] ) two loops linked once may never be planar since the contribution of their writhe to the helicity must cancel out the contribution from their linkage .
this restriction is again hard to accept . to obtain
another constraint bekenstein introduces the relation @xcite @xmath81 @xmath82 measures the knotting of the curve and can jump by increments of @xmath83 , and @xmath84 is the _ contortion _ of the loop defined by @xmath85 which integrates the torsion over the whole loop.the axis of a string can be treated as a curve in differential geometry .
a curve @xmath86 ( @xmath87 is the arc length ) has @xmath88 as unit tangent vector , has a unit normal @xmath12 , and a unit binormal @xmath89 .
they are connected by the frenet - serret equations @xmath90 where @xmath91 , a positive scalar , is called the curvature of the curve , and @xmath92 , which may take on either sign , is called the torsion . ] inserting @xmath93 from ( [ knotting ] ) into ( [ const ] ) gives the following constraint connecting the contortion , knotting and linking of a collection of loops : @xmath94 this relation says that the sum of contortions of a collection of unit winding number loops is quantized in integers @xcite .
this implies that the contortion of any single knot is quantized in integers : the knot can not wiggle freely about , but is frozen in a configuration of integer contortion .
this is a very strong constraint on the dynamics of such a configuration , and together with our previous observations , makes us suspect that the calculation has missed a term .
the first clue of the missing term was presented by moffatt @xcite , who notes that close inspection of an expression of the form @xmath95 integrated over a single flux loop @xmath67 ( of an arbitrary divergence - free vector field ) , shows that the helicity in this limit has an additional contribution from pairs of points @xmath63 , @xmath64 separated by a distance comparable to the cross - sectional span of the loop . in addition to the writhe @xmath93 ( [ writh ] ) , which we introduced earlier as the sole contribution to helicity from a single loop , moffatt conjectures that the close points contribute an extra term namely @xmath96 where @xmath97 is the contortion of loop @xmath73 defined earlier , and @xmath98 is the flux of the loop .
actually it is easy to see that bekenstein s calculation in sec .
[ global_helicity ] ignores the thickness of the string , i.e. the contribution from close pairs of points @xmath63 and @xmath64 .
equation ( [ c_flux ] ) reduces a volume integral to a line integral along the core of the string and thus treats the string as though it where infinitely thin .
moffatt speculates that the additional contribution , from close pairs of points @xmath63 and @xmath64 , is related to the contortion so that the helicity will be a topological invariant of the loop , i.e. , @xmath99 by ( [ knotting ] ) .
but , thinking this over we realize that the contortion is also just a property of the flux tube s ( string s ) axis and not of its internal structure hence we should expect a different term to appear .
later moffatt @xcite explicitly calculated the helicity of a magnetohydrodynamic system of flux tubes to find that his earlier speculation was not exact , and that @xmath82 is not a topological invariant after all since its value jumps discontinuously at inflexion points ( points of zero curvature ) . in this later calculation moffatt
manipulates the magnetohydrodynamic equation of motion of a configuration of field lines moving with the flow .
this calculation is much too complicated to carry over to the case of cosmic strings since the field equations for these are non - linear .
hence we will follow in the footsteps of berger and field @xcite who also calculate the internal helicity due to the internal structure of a flux tube , but we will present a slightly different calculation , and a slightly different result .
berger and field s result was that @xmath100 where @xmath101 is any real number , while we obtain the same result as moffatt @xcite with @xmath102 and @xmath103 , an integer , is the linking number of the field lines in the flux tube .
we first calculate exactly the contribution to the helicity of a single vortex loop @xmath104 .
first we consider the inner structure of the string to consist of nested toroidal vorticity surfaces ( a peace of the loop is shown in fig . [ f2 ] ) .
then we may build up the string by increments of vorticity flux @xmath105 composed of many thin flux tubes @xmath106 that wind around the inner flux tube @xmath98 @xmath103 times .
the increment in helicity for each such flux tube @xmath106 is @xmath107 where @xmath103 is the gauss linking number of element @xmath108 with @xmath98 ( the inner flux tube ) .
the increment in helicity for the whole of @xmath105 is simply @xmath109 .
( the flux @xmath105 simply counts the number of field lines intersecting the increment in the cross section of the string , and hence the number of field lines wrapped around the flux tube in the center . )
we thus have according to ( [ helic ] ) @xmath110 where @xmath111 is along the axis of the string and @xmath112 is along one of the field lines in @xmath106 wrapping @xmath98 , and vice - versa . finally , integrating over all the nested flux tubes @xmath113 where @xmath12 is the winding number of the cosmic string ( the flux @xmath98 is easily calculated using the definition of @xmath51 , stoke s theorem and ( [ q_phase ] ) ) . throughout this calculation
we have assumed that @xmath106 and @xmath98 have the same linking number as a single field line in @xmath114 with the axis of the string .
we have also assumed that all field lines have the same linking number with the axis of the string ( and with each other ) .
this is plausible for a continuous field . as we see
, the internal helicity is due to the linking of the @xmath60 field lines inside the string .
this linking occurs when the field @xmath60 is twisted : if the field lines do not link the internal helicity is zero .
we would now like to express the linking number of the field lines @xmath103 as a sum of @xmath93 and some other term or terms , so that we may reconcile this calculation with bekenstein s result ( [ hel_eq ] ) .
the linking number @xmath103 of @xmath108 with @xmath98 is equal to the linking number of a single field line in @xmath108 with the axis of the string .
since these are two closed curves at a very short distance apart , the axis of the string can be treated as a curve in differential geometry , forming the edge of a ribbon whose other edge is traced by the field line .
this type of problem appears first to have been addressed by clugreanu @xcite , who considered two neighboring closed curves @xmath115 , @xmath116 forming the boundaries of a ( possibly knotted ) ribbon of small spanwise width @xmath117 , and showed that for sufficiently small @xmath117 the linking number @xmath103 of @xmath115 and @xmath116 can be expressed in the form @xmath118 where @xmath93 and @xmath84 are respectively the writhe and the contortion of @xmath115 , and @xmath119 is an integer representing the number of rotations of the unit spanwise vector @xmath120 on the ribbon relative to the frenet pair ( the unit normal and unit binormal . ] ) in one passage round @xmath115 .
( the contortion @xmath84 actually measures the rotation round the curve of the frenet pair itself . )
8.5 cm 2.5 cm the frenet pair has discontinuous behavior in going through a point of inflexion ( zero curvature ) . as a result @xmath84 and @xmath119
are not well defined if @xmath115 has inflexion points .
if @xmath115 is continuously deformed through an inflexional configuration , then @xmath84 is discontinuous by @xmath121 , but @xmath119 is simultaneously discontinuous by an equal and opposite amount @xmath122 , so that the sum @xmath123 does vary continuously through inflexion points @xcite .
this sum is known as the _ twist _ of the ribbon @xcite @xmath124 thus , as we remarked earlier , @xmath125 is not a topological invariant while @xmath103 ( [ calig ] ) is . with the help of ( [ calig ] ) and ( [ twist ] ) we are now ready to complete the calculation of the helicity for a single ( un- knotted ) cosmic string loop @xmath104 .
we take @xmath115 above to be the string s axis and @xmath116 to be along one of the field lines twisted around the axis .
then for the @xmath73-th loop @xmath126 our final expression for the helicity of a tangle of cosmic string loops is therefore @xmath127= \sum_{i \neq j}{\cal l}_{ij}(2\pi|n_i|)(2\pi|n_j| ) + \sum_i{\cal w}_i(2\pi n_i)^2
+ { \cal t}_i(2\pi n_i)^2.\ ] ] contrast this ( correct ) result with ( [ hel_eq ] ) .
the twist @xmath128 is the contribution from the close pairs of points @xmath63 and @xmath129 inside the string ; it is not a property of the axes of the string alone , but rather of the internal structure of the field lines inside the string core .
in the previous sections we learned that helicity measures the linkage of the field lines of @xmath130 , and that the total helicity of global cosmic strings is zero whatever their configuration may be . these properties must put some constraint on the possible configurations of the strings . in this section
we show that the knotting and external linkage of a cosmic string determines the internal structure of the field lines inside the string , and thus external and internal helicity cancel each other out so that the total helicity is zero . since the field lines in a string must close on themselves , and their twist depends on external knotting and linking , we will get constraints on the possible configurations of string loops .
these constraints are weaker than those proposed by bekenstein , and are free from the latter s paradoxical aspects .
let us start by examining @xmath52 ( the gradient of the phase ) of the most simple configuration , a single loop . as we know
the phase varies when one goes once around the string by @xmath10 , where @xmath12 is the winding number of the string .
2.5 cm what happens if the phase varies along the string too ? because of periodicity this is only possible if @xmath131 where @xmath66 is along the string axis .
but in this case there will be another topological requirement . because the phase varies by @xmath132 around the perimeter of the loop , for every surface spanning the loop there must be at least one place where
the amplitude @xmath13 so that the field @xmath9 will be single - valued .
this is actually the requirement that the loop be threaded by another @xmath133 unit winding - strings or a single string with winding number @xmath133 .
hence an isolated loop must have @xmath134 ; for a totally symmetric loop @xmath52 will have only one component , in the direction around the string . along the loop
must be zero for an isolated loop .
however , any configuration where the average of the component of @xmath52 along the loop is zero can be continuously deformed into a configuration where the component of @xmath52 along the loop equals zero at every point .
hence , as we shall see the linking of the @xmath130 field lines is not affected . ]
how does this affect the configuration of the field lines of @xmath130 ?
we recall that by definition ( [ vec_v ] ) , @xmath2 is parallel to @xmath52 and @xmath130 ( [ curl_v ] ) is perpendicular to @xmath2 .
hence if @xmath2 has a component only around the string , @xmath130 will have a component only along the string , and its field lines will not be twisted or linked .
since linkage of the field lines is a topological property , it will not change for continuous deformation of the loop ; therefore the field lines of any isolated loop with arbitrary shape will be unlinked .
the immediate result is that _ the total helicity of any isolated loop with arbitrary shape is zero_. this corrects bekenstein s conclusion ( [ const ] ) that an isolated loop must be a planar loop at any instant so that the helicity will be zero .
we next investigate a configuration of two linked loops @xmath135 and @xmath136 with winding numbers @xmath137 and @xmath138 respectively .
we now have a component of @xmath52 in the direction along the strings too .
the phase will vary around the perimeter of loop @xmath135 by @xmath139 and around the perimeter of @xmath136 by @xmath140 ( fig .
now that @xmath141 has components along the string and around it , we must introduce a new concept , the _ chirality _ ( handedness ) of a string loop . a loop is right - handed ( and has positive chirality ) when the component of @xmath52 along the string is in the direction of the thumb of the right hand , whose fingers point along the component of @xmath52 around the string .
if the component of @xmath52 along the string is in an opposite direction to the thumb , the loop is left - handed ( and the chirality is negative ) .
8.5 cm 3.5 cm along with the definition of chirality comes the definition of the sign of the directions along and around the string . we define the positive direction around the string to coincide with the direction of @xmath52 around it , and the positive direction along the string to coincide with the direction of the thumb of the right hand whose fingers follow the direction of @xmath52 around the string axis . in a
right - handed ( left - handed ) string loop whose phase changes around the string by @xmath10 and along the string by @xmath142 , @xmath12 and @xmath133 will have equal ( opposite ) signs .
it is easy to see that _ two strings may link only if they have the same chirality _ ; otherwise there would be a discontinuity in the phase . what will be the structure of the @xmath130 field lines in such a configuration ?
now that @xmath52 has components in two directions , @xmath130 will also have at least two components , one along the string and one around it .
the field lines will be twisted and linked ( fig .
[ f6 ] ) .
in such a case we expect to have two contributions to helicity , a contribution from the external linkage of the string loops , and a contribution from the internal linkage of the field lines inside the strings .
these two contributions must cancel each other so that this configuration may exist .
the external helicity for such a configuration with @xmath143 and winding numbers @xmath137 and @xmath138 is simply @xmath144 where the sign of @xmath41 is determined by the relative direction of @xmath130 along the perimeter of both loops
. we will show below that @xmath41 is positive for right - handed configurations and negative for left - handed ones . )
is always positive by our current definition , there is no need to take the absolute of the winding numbers as in ( [ twist_hel ] ) . ] but computing the internal helicity @xmath145 is not straightforward .
the internal helicity of magnetic flux tubes in a magnetohydrodynamic fluid or of vortices in an ordinary fluid is arbitrary since there is no restriction on the linking of the field lines inside the flux tube . in those systems the internal structure of the field lines
is not determined by the topology of the flux tubes or vortex filaments and , therefore , the total helicity can not be determined exactly from their configuration .
by contrast we shall now show that the internal helicity of cosmic strings is solely determined by the winding numbers of the strings and their topological configuration , so that the total helicity for a configuration of linked vortices can be expressed as a function of their linkage and winding numbers alone .
3.6 cm we now attempt to derive the internal linking number @xmath103 , that measures the linkage of the field lines inside a cosmic string loop .
the loop is characterized by winding numbers @xmath12 and @xmath133 around and along the perimeter of the loop respectively .
any ( un - knotted ) cosmic string loop may be deformed continuously into a planar circle loop , and the twist of the field lines inside it uniformly distributed , without changing the linking of the field lines .
this is possible since the linking is a topological property of the field lines , and does not change by continuous deformations .
it would further simplify our calculation if we could perform it in cylindrical symmetric coordinates , by virtually cutting the loop and standing it upright , such that it acquires cylindrical symmetry ( fig .
[ f7 ] ) . we may convince ourselves that this is possible since very close to the axis of the string the field lines have cylindrical symmetry around the string , and for reasons of continuity the linkage of the field lines can not change as we move away from the string .
the linking number @xmath103 is simply the number of times @xmath146 is wrapped around the @xmath147 axis ( which coincides with the axis of the string ) .
this must be an integer for the string to form a loop .
the vectors @xmath2 and @xmath130 are computed in cylindrical symmetric coordinates @xmath148 as follows : first @xmath149 where @xmath47 in ( [ vec_v ] ) is taken to be @xmath150 , a function of @xmath37 only .
the gradient of @xmath0 must obey @xmath151 where @xmath152 is the length of the string . hence for uniformly distributed twist
@xmath153 the calculation of @xmath130 gives @xmath154 we are interested in the number of times each field line encircles the @xmath147 axis .
often we associate flow lines with vector fields by imagining how a test particle would move if the vector field were its velocity field .
let us assume the field line traces out the trajectory of a point particle traveling with constant velocity during a period of time @xmath155 .
the velocity components of the particle are equal to the components of @xmath130 .
the following simple equations must apply so that the field lines may reconnect when the string forms a loop @xmath156 by eliminating @xmath155 between the two equations and using @xmath130 from ( [ sym_curl_v ] ) we obtain the result @xmath157 thus the internal linking number @xmath103 depends exclusively on the winding numbers of the loop . to evaluate the total helicity for two linked un - knotted cosmic string loops , we first determine the sign of the gauss linking number @xmath41 for left - handed and right - handed configurations , which depends on the direction of @xmath158 along the perimeter of both loops . by our definition above , the winding numbers around the axis of the string are always positive , since we have defined the positive direction around the string to coincide with the direction of @xmath52 around the axis of the string . hence from ( [ sym_curl_v ] ) the component of @xmath130 along the axis
is always in the positive direction i.e. along the thumb of the right hand whose fingers follow the positive direction around the axis of the string .
[ f8 ] makes it clear that a right - handed configuration will have @xmath159 and the left - handed configuration will have @xmath160 . the sign of the internal linking number @xmath103 also depends on whether the configuration is left - handed or right - handed . as we explained earlier , in a right - handed configuration @xmath12 and @xmath133 have equal signs while in a left - handed configuration , they have opposite signs .
therefore , by ( [ int_link ] ) @xmath103 is negative for a right - handed pair and positive for a left - handed one . _
the signs of the external and internal linkages are opposite_. we are now able to write down the total helicity for a pair of linked strings . for two right - handed linked strings with winding numbers @xmath137 and @xmath138 , we have @xmath161 and @xmath162 ( since the axis of each loop encircles the other loop ) and by ( [ h_ext ] ) and ( [ h_int ] ) the total helicity is @xmath163 for two left - handed linked loops the total helicity is @xmath164 the external and internal helicities thus cancel each other out without constraining the shapes of the loops .
we have thus removed the constraint on the geometrical shape of a single loop , a pair of linked loops or any other configuration .
the contributions to helicity are purely topological .
but ( [ int_link ] ) introduces a new topological constraint . the internal gauss linking number @xmath103 of the field lines in the string
must be an integer .
hence the winding number @xmath133 along the perimeter of the string is not free to take on just any value , i.e. , a string loop with winding number @xmath12 can not be linked arbitrarily with other loops . for the right handed pair of linked loops
@xmath135 and @xmath136 , for example , the internal linking number of @xmath135 is @xmath165 and the internal linking number of @xmath136 is @xmath166 .
therefore we must have @xmath167 so that both @xmath168 and @xmath169 be integers .
this means that _ only string loops with the same winding number may link once_. in the next section we generalize this result to a generic configuration of linked un - knotted loops .
examining a couple more examples will help us formulate a more general result .
3.5 cm fig .
[ f9 ] depicts a right - handed cosmic string loop @xmath135 with winding number @xmath137 linked twice with a cosmic string loop @xmath136 with winding number @xmath138 such that their gauss linking number is @xmath170 . as explained earlier the change in phase along the perimeter of a loop
is determined by the winding numbers of the strings threading it . since @xmath136 links @xmath135 twice the change in phase along @xmath135 is @xmath171 so that @xmath172 ; correspondingly the phase along @xmath136 changes by @xmath173 so that @xmath174 . hence the internal linking number of @xmath135 is @xmath175 and that of @xmath136 is @xmath176 .
thus the values of @xmath138 must be restricted to the integers that divide @xmath177 and are greater than or equal to @xmath178 .
for a few examples see table [ tab1 ] , only such loops may link twice .
the total helicity for this configuration is @xmath179 @xmath180 our second example in fig .
[ f10 ] depicts a left - handed string loop with winding number @xmath137 linked once with two other left - handed string loops , one with winding number @xmath138 and the other with winding number @xmath181 . 3.5 cm the external gauss
linking numbers are @xmath182 and @xmath183 and @xmath184 .
the change in phase around the perimeter of the strings , and the internal linking numbers are determined by the same arguments as above , such that @xmath185 , @xmath186 and @xmath187 .
the total helicity will thus be @xmath188 as we would expect . by generalizing from these examples
we see that the internal linking number @xmath189 for a loop @xmath67 linked with an arbitrary number of other loops is @xmath190 where @xmath191 is equal to + 1 for right handed loops and -1 for left handed ones .
this equation is also a constraint on the winding numbers and the gauss linking numbers of the loops linking @xmath67 , since @xmath189 must come out an integer .
the total helicity of loops linked in an arbitrary manner thus vanishes : @xmath192
local cosmic strings are solutions to lagrangians invariant under local gauge transformations .
the simplest model for a local cosmic string is the abelian higgs model ; it has @xmath193 local symmetry .
nielsen and olesen @xcite found straight string ( cylindrically symmetric ) solutions to this model characterized by a magnetic field confined to the string s core whose flux is quantized in proportion to the winding number @xmath12 of the string .
another important property of these strings is that they have finite energy per unit length ( in contrast to the global strings ) since the gauge field @xmath194 can cancel out the gradient of the phase at spatial infinity .
the nielsen - olesen string is actually a two dimensional vortex solution with energy density localized around a point in space , which can trivially be embedded into the extra dimension to produce an infinite string - like object whose energy density is localized along a line . by embedding the vortex along a closed curve , rather than a straight line
, the ends of the string can be joined to produce a closed string of finite length .
thus we may have a configuration that is a tangle of such local cosmic strings which are both knotted and linked with each other .
since these strings have quantized magnetic flux tubes in their cores , we actually have a tangle of knotted and linked magnetic flux tubes , and each configuration may be characterized by a magnetic helicity quantifying the linking of the cosmic strings , and the internal linkage of the magnetic field lines inside each string . in this section we refer to the result @xcite that the change in the electroweak magnetic helicity of a tangle of local cosmic strings , in the standard model , is related to baryon number violation .
previous works indicated a continuous change in baryon number . in contradiction to these works , we show here , by using our knowledge of the topological interpretation of helicity as the linking of field lines that , _ baryon number on local strings is quantized _ in this model .
sato and yahikozawa @xcite scrutinize the geometrical and topological properties of the abelian higgs model with vortex strings by calculating the topological term @xmath196 @xmath197 which originates from the chiral anomaly , and is the cp violating @xmath0 term in the action .
the definition of the _ dual _ is the following : @xmath198 .
@xmath196 is evaluated for cases where all vortex strings are closed , with winding number @xmath199 , and the number of the vortex strings is conserved i.e. vortex string reconnection does not occur . using a _
topological formulation _ which they developed , and via a complicated calculation , sato and yahikozawa show that @xmath200_{t = -\infty}^{t = + \infty}.\ ] ] the first term is the linking number between each pair of strings and comes from the _ mutual interactions _ between different vortex strings .
the second term is the writhing number ; it comes from the _ self - interactions _ of each vortex string , and can vary with the shape of the string .
it is plain that the value of @xmath196 changes continuously with the shape of the strings as time elapses .
equation ( [ sato_res ] ) strongly resembles bekensteins result ( [ hel_eq ] ) for the helicity of global cosmic strings .
we shall now show that @xmath196 is just the total change in time of the magnetic helicity of the abelian higgs model with local vortex strings , and hence equal to the change in the external and internal linking numbers of the magnetic field lines each weighed by the square of the quantized flux in the corresponding vortex . by construction @xmath201 , so we may write @xmath202 using maxwell s equations @xmath203 and @xmath204 and integration by parts , @xmath205 where we have used the relation @xmath206and the assumption that all fields die off at spatial infinity .
our final result is thus @xmath207 where @xmath19 stands for the magnetic helicity .
as we already know , the magnetic helicity may be written as the sum of the external linkages of the flux tubes and the internal linkage of the field lines in each flux tube , times the square of the flux ( [ twist_hel ] ) @xmath208 the two terms @xmath209 add to @xmath189 , the internal linking number ( [ n_hel ] ) , and the magnetic flux in a local string of unit winding number in this model is @xmath210 . as @xmath211 .
hence , @xmath212 and via stoke s theorem and ( [ q_phase ] ) the magnetic flux is @xmath213 . ] therefore , according to ( [ i_rel ] ) and ( [ mag_hel ] ) the term @xmath196 is simply equal to @xmath214_{t = -\infty}^{t = + \infty}.\ ] ] comparing this with ( [ sato_res ] ) we see that sato and yahikozawa have missed out the twist term . this resulted from neglecting the internal structure of the vortex strings in their calculation .
our result is crucially different from theirs : according to their result the term @xmath196 may have a continuous range of values , while according to our result @xmath196 is quantized because @xmath215 is quantized by the internal and external linking numbers of the magnetic flux tubes threading the strings cores .
this fact has an important physical consequence , as we shall see presently .
the origin of the excess of matter over antimatter in our universe remains one of the fundamental problems .
a widely discussed model @xcite is baryogenesis in the course of the broken symmetry electroweak transition , where the baryonic asymmetry is induced by the quantum chiral adler - bell - jackiw anomaly @xcite . in models of fermions
coupled to gauge fields , certain current - conservation laws are violated by this anomaly .
the term @xmath216 calculated on the background of the nielsen - olesen strings has been shown to cause violation of baryon number conservation via this anomaly .
the anomaly arises when a classical symmetry of the lagrangian does not survive the process of quantization and regularization : the symmetry of the lagrangian is not inherited by the effective action .
an axial current that is conserved at the level of the classical equations of motion can thus acquire a nonzero divergence through one - loop diagrams that couple this current to a pair of gauge boson fields . the feynman diagram that contains
this anomalous contribution is a triangle diagram with the axial current and the two gauge currents at its vertices .
left and right handed fermions contribute terms with opposite signs to the anomaly ; however , in chiral theories in which the gauge bosons do not couple equally to right- and left - handed species , the sum of these terms can give a non - zero contribution . in theories such as qed or qcd in which the gauge bosons couple equally to right- and left - handed fermions , the anomalies automatically cancel . within the standard model of weak interactions ( the glashow - weinberg - salam @xmath217 invariant theory ) ,
the requirement from experiment that weak interaction currents are left - handed forces us to choose a chiral gauge coupling .
the coupling of the @xmath218 boson to quarks and leptons can be derived by assigning the left - handed components of quarks and leptons to doublets of an @xmath219 gauge symmetry , and then identifying the @xmath218 bosons as gauge fields that couple to this @xmath219 group , while making the right - handed fermions singlets under this group .
the restriction of the symmetry to left - handed fields leads to the helicity structure of the weak interactions effective lagrangian , and thus causes the chiral anomaly of the baryon current to arise .
the baryon number current in this model is @xmath220 and the anomalous baryonic current conservation equation is @xmath221 where @xmath222 and @xmath223 are the field strengths for the three @xmath219 and one @xmath193 gauge fields @xmath224 ( @xmath225 ) and @xmath226 respectively , and @xmath227 and @xmath228 are their charges .
@xmath229 is the number of quark families .
the baryon number current can be integrated to yield the change in the baryon number between two different times , if we assume that the baryonic flux through the surface of the three - volume of interest vanishes : @xmath230 equations ( [ current_anomaly ] ) and ( [ delta_charge ] ) relate the change in the baryon number to terms of the form @xmath231 which we calculated in the previous section for an abelian model on the background of nielsen - olesen vortices .
this term was found to be related to the change in the sum of the linking of the strings from their initial to final configurations ( [ fin_i ] ) .
this leads us to an exciting idea @xcite : if chiral fermions are coupled to electroweak vortices , baryon number may be violated because of the anomaly as vortices link and de - link . for this to happen
there must be a stable string solution in this model , and the change in the baryon number must be related to the helicity of an electroweak field whose flux is confined to the string .
it is generally believed that the standard electroweak model is free from topological defects .
the reason is that the first homotopy group of @xmath232 can be shown to be trivial i.e. @xmath233 @xcite .
however , this does not mean that the model is free from non - topological defects .
it has been shown by manton @xcite that the configuration space of classical bosonic weinberg - salam theory has a non - contractible loop .
hence an unstable , static , finite - energy solution of the field equations can exist .
vachaspati @xcite showed that exact vortex solutions exist , which are stable to small perturbations for large values of the weinberg angle @xmath234 and small values of the higgs boson mass .
this expectation has been confirmed by the numerical results of james , perivolaropoulos and vachaspati @xcite . defining @xmath235 where @xmath236 , the static vortex solution that extremizes the energy functional was found to be : @xmath237 and @xmath238 the subscript @xmath239 on the functions @xmath240 and @xmath194 means that they are identical to the corresponding functions found by nielsen and olesen for the usual abelian - higgs string .
in other words , the weinberg - salam theory has a vortex solution which is simply the @xmath239 vortex of the abelian higgs model embedded in the non - abelian theory . substituting this solution into ( [ current_anomaly ] ) we have @xmath241 where @xmath242 and @xmath243 .
the change in the baryon number is now @xmath244 via equations ( [ delta_charge ] ) , ( [ z_current ] ) and ( [ i_rel ] ) and with @xmath245 playing the role of baryon number charge .
we see that the change in the baryon number is proportional to the change in the @xmath246 field s helicity . for a configuration of tangled strings with arbitrary winding numbers @xmath247 @xmath248 where the flux of each vortex is @xmath249 . ] and hence the factor of @xmath250 for the flux . ]
the final result for a configuration of strings with unit winding number is @xmath251 and baryon number is quantized .
earlier works failed to fully implement the topological interpretation of helicity as the measure of the linkage of field lines of the divergence free field .
since field lines may be twisted and linked inside a vortex core , the inner structure of the field inside the core may not be ignored when calculating helicity terms . neglecting internal helicity
had led to peculiar results : unacceptable constraints on string configurations and continuous violation of baryon number .
once we add the contribution from the internal structure of the strings , we find new constraints on the configurations of linked global cosmic strings , which are physically more pleasing , and we also find that , the baryon number is quantized as we would expect .
our work is another demonstration of the fact that helicity counts the linkage of the divergence free field lines .
we explain how there is no contradiction between the fact that helicity counts the linkage of field lines , and that the total helicity of any configuration of knotted linked loops always manages to vanish .
the vector @xmath252 ( [ vec_v ] ) constructed for the purpose of obtaining a helicity term for global cosmic strings is proportional to @xmath141 .
hence it is perpendicular to @xmath253 ( [ curl_v ] ) and the helicity must vanish . on the other hand
the helicity counts the linkage of field lines of @xmath253 , and these lines must have external and internal linkages for linked configurations of cosmic string loops .
what we have discovered is that the field lines arrange themselves in such a manner that the external and internal helicities exactly cancel each other .
the field lines inside a string may not link in an arbitrary manner , their linkage is constrained by the topological configuration of the string and by its linking relations to other strings .
further , we have discovered a constraint on the permitted configurations of un - knotted linked loops , in the form of ( [ gen_int_link ] ) which must produce integer values .
this constraint results from the combination of the topological character of helicity integrals with the topological nature of cosmic strings as topological defects . in this work we confined ourselves to the problem of un - knotted linked loops .
the problem of knotted configurations is far more complicated since it is hard to separate the external contribution to helicity ( depending on the topology of the knot ) from the internal contribution depending on the twist of the field lines inside the knot .
moffatt @xcite proposed a method for calculating the external helicity , but it turns out that for some knotted configurations this method is ambiguous ( it may be that it works only for _
chiral _ knots , i.e. knots that are not isotopic to their mirror images ) .
hence , we leave the treatment of knots to later work . for local cosmic strings in the standard model , the change in the helicity of the electroweak magnetic field confined to the strings core is related to the change in baryon number over a period of time ( [ z_helic ] ) .
thus , baryon number conservation is violated as cosmic strings link and de - link . the idea that violation of baryon number
may be related to the change in the linking of electroweak strings has been developed by cohen and manohar @xcite , vachaspati and field @xcite and garriga and vachapati @xcite .
however , they neglected the contribution of the writhe of the strings to helicity .
sato @xcite adds the writhe term , but misses out the twist term .
this led to his conclusion that baryon number conservation is violated as vortices change their shape , which implies that baryon number may change continuously and , therefore , is not quantized . only by including both the writhe and the twist terms , which together give the internal linkage of the field lines ,
do we recover a quantized baryon number !
charge quantization has previously been related to flux quantization .
it was suggested by dirac in 1931 @xcite to explain the quantization of the electric charge .
this idea was developed further by jehle @xcite . in general , quantization of charges
is not a direct consequence of standard ( non - gut ) field theories . particularly , the quantization of electric charge has no widely accepted explanation within the standard model . during the last decade
it has been suggested @xcite that the electric charges can be heavily constrained within the framework of the standard model .
this is achieved partly by constraints related to the classical structure of the theory ( such as the requirement that the lagrangian be gauge - invariant ) , and from the cancelation of anomalies , at the quantum level . the novel idea in this work
is that quantization of the baryon number may arise from the topological nature of helicity integrals calculated on the background of local cosmic strings and from the flux quantization in the strings .
i thank prof .
d. bekenstein for his guidance during the course of this work and for many discussions .
i am also grateful to dr .
l.sriramkumar for many stimulating discussions .
this research is supported by a grant from the israel science foundation , established by the israel academy of sciences and humanities .
99 l. onsager , nuovo cim . * 6 * , suppl . 2 , 246 ( 1949 ) .
feynman , in : progress in low temperature physics , vol .
1 , ed . c.j .
gorter ( north - holland , amsterdam , 1955 ) .
a. vilenkin , and e .. p.s .
shellard , in : cosmic strings and other topological defects , ( cambridge u.p . ,
cambridge , 1994 ) h.k .
moffatt , [ 1981 ] , j. fluid mech . *
106 * , 27 ( 1955 ) .
donnelly and c.e .
swanson , j. fluid mech . * 173 * , 387 ( 1986 ) ; k.w .
schwartz , phys . rev . *
b 38 * , 2398 ( 1988 ) .
moffat , j. fluid mech . * 35 * , 117 ( 1969 ) .
moffatt and r.l .
ricca , proc .
. lond . * a 439 * , 411 ( 1992 ) .
berger and g.b .
field , j. fluid .
147 * , 133 ( 1984 ) .
arnold , ( in russian ) , proc .
summer school in differential equations , ereran , armenian ssr acad .
( trans . in sel .
* 5 * , 327 ( 1986 ) . )
kuznetsov and a.v .
mikhailov , phys .
lett . * a 77 * , 37 ( 1980 ) .
h. hopf , math .
annalen * 104 * , 637 ( 1931 ) .
whitehead , proc .
n. a. s. * 33 * , 117 ( 1947 ) .
davis and e.p.s .
shellard , phys .
lett . * 63 * , 2021 ( 1989 ) .
ginzburg , and l.p .
pitaevskii , zh .
* 24 * , 1240 ( 1958 ) .
jetp * 7 * , 858 ( 1958 ) )
kibble , j. phys . *
a9 * , 1387 ( 1976 ) .
bekenstein , phys . lett .
* b 282 * , 44 ( 1992 ) . t. vachaspati and g.b .
field , phys .
lett . * 73 * , 373 ( 1994 ) .
m. sato , and s. yahikozawa , nucl .
b436 * , 100 ( 1995 ) .
m. sato , _ shapes of cosmic strings and baryon number violation _ ,
_ hep - ph_/9508375 .
t. vachaspati and a. vilenkin , phys . rev . * d 30 * , 2036 ( 1984 )
shellard and b. allen , in : the formation and evolution of cosmic strings , eds .
g. gibbons , s.w . hawking and t. vachaspati ( cambridge u.p . , cambridge , 1990 ) .
white , am .
j. maths * 91 * , 693 ( 1969 )
fuller , proc .
usa * 68 * , 815 ( 1971 ) .
l. woltjer , proc .
44 * , 489 ( 1958 ) .
l. woltjer , proc .
44 * , 833 ( 1958 ) .
moreau , sminaire danalyse convexe montpellier , expos no . 7
g. clugreanu , rev . math .
pures appl . * 4 * , 5 ( 1959 ) .
g. clugreanu , czechoslovak .
* 11 * , 588 ( 1961 ) .
fuller , proc .
usa * 75 * , 3557 ( 1978 ) .
nielsen and p. olesen , nucl . phys . *
b61 * , 45 ( 1973 ) .
cohen , and a.v .
manohar , phys .
b265 * , 406 ( 1991 ) . j. garriga , and t. vachaspati , nucl
. phys . * b438 * , 161 ( 1995 ) .
s. weinberg , the quantum theory of fields , ( cambridge u.p .
, cambridge , 1996 ) .
peskin and d.v .
schroeder , an introduction to quantum field theory , ( addison - wesley , new york , 1995 ) .
dolgov , rep .
. phys . * 222 * , 309 ( 1992 ) .
n. turok , in : formation and interaction of topological defects , p.283 , eds .
davis and r. brandenberger ( plenum press , new york , 1995 ) .
hindmarsh and t.w.b .
kibble , rep .
58 * , 477 ( 1995 ) .
adler , phys . rev . * 177 * , 2426 ( 1969 ) .
j.s bell and r. jackiw , nuovo cim . *
60a * , 47 ( 1969 ) .
s. weinberg , phys .
* 19 * , 1264 ( 1967 ) .
a. salam , elementary particle theory , ed .
n. svarthholm , ( almquist and forlag , stockholm , 1968 ) .
n.s . manton , phys .
* d 28 * , 2019 ( 1983 ) .
t. vachaspati , phys .
rev . lett . * 68 * , 1977 ( 1992 ) .
t. vachaspati , nucl .
b397 * , 648 ( 1993 ) .
y. nambu , nucl . phys . *
b130 * , 505 ( 1977 ) .
m. james , l. perivolaropoulos and t. vachaspati , nucl .
b395 * , 534 ( 1993 ) .
dirac , proc .
( london ) * a133 * , 60 ( 1931 ) . h. jehle , phys . rev . * d 3 * , 306 ( 1971 ) .
r. foot , h. lew and r.r .
volkas , j. phys .
* g 19 * , 361 ( 1993 ) . | we apply the concept of helicity from classical hydrodynamics to elucidate two problematical issues in cosmic string physics .
helicity , the space integral of the scalar product of a velocity - like field with its vorticity field ( curl ) , can be defined for a complex scalar field in analogy with fluids .
we dwell on the topological interpretation of helicity as related to the linking of field lines of the vorticity field .
earlier works failed to fully implement this interpretation for cosmic strings by missing a term connected with the linking of these lines inside the strings . as a result
paradoxical conclusions were drawn : global cosmic string loops may not take on certain simple shapes , and baryon number is not quantized in integers in the presence of local cosmic strings in gauge theory .
we show that both paradoxes are removed when internal contributions to helicity are properly taken into account .
in particular , quantization of baryon number can be understood within a special case of the glashow - weinberg - salam model if cosmic strings are the unique mechanism for baryosynthesis . in addition
, we find a new constraint on the permitted linkages of cosmic strings in a string tangle . |
nanoscale magnetic materials are the subject of a currently increasing interest which is related both to the novel fundamental aspects of magnetic structure and dynamics on the nanometer - scale , and promising future technological applications , such as magnetic data storage elements , sensors or spin electronic devices ( for a review see , e.g. , ref . ) .
one novel approach to engineer single , i.e. well separated and individually addressable nanoscale magnets , is to utilize self - organization of nanostructures , such as carbon fullerenes , carbon nanotubes , silicon - carbon - nanorods and others .
however , owing to a weak magnetic exchange between unpaired spins these materials often exhibit only a very small magnetic moment so that external magnetic fields barely interact with these nanostructures .
one possible way to overcome this problem is to prepare self - organized nanoscale materials based on transition - metal ( tm ) oxide low - dimensional ( low - d ) structural units where complex strong electronic correlations in the spin- and charge sectors are expected .
indeed , a variety of fascinating phenomena related to spin- and charge correlations in low - d tm oxides were found in the past years , ranging from high - temperature superconductivity in 2d - cuprates , a spin - peierls transition or a haldane spin - gap behavior in 1d - spin chain compounds , to superconductivity in 1d - spin ladder materials ( for recent reviews see e.g. ref . ) .
-nt wall : ( a ) double layer of octahedrally coordinated v(1,2 ) ions .
tetrahedrally coordinated v(3 ) ions are located between the two layers .
( b ) zig - zag v(1,2)-chains in a layer . ]
recently a new class of nanoscale spin magnets , mixed valence vanadium oxide multiwall nanotubes ( vo@xmath3-nts ) , has been introduced.@xcite the structure of the vo@xmath3-nt walls @xcite is closely related to that of barium vanadium oxide bronze bav@xmath4o@xmath5h@xmath6o ( fig .
[ structurevox]).@xcite one finds there a v - metal ion in distorted octahedral and tetrahedral coordination , respectively .
v - octahedra are coupled in edge - sharing zig - zag vo - chains .
arrays of vo - chains form a double layer with tetrahedral v - sites lying between the two layers .
mixed valence vanadium oxides , vo@xmath3 ( @xmath7 ) , are themselves very rich in electronic and magnetic properties owing to an intimate interplay between spin , orbital and charge degrees of freedom.@xcite when assembled in nanotubes ( nt ) using dodecylamine , c@xmath8h@xmath9nh@xmath6 , as a structure - directing surfactant molecule , vo@xmath3-nts show up diverse properties ranging from spin frustration and semiconductivity to ferromagnetism by doping with either electrons or holes.@xcite such a rich behavior makes this material very interesting from the viewpoint of fundamental research as well as regarding possible future applications mentioned above . in particular , its magnetic properties can be widely tuned by intercalation or electrochemical doping ( e.g. by lithium or iodine ) , thereby offering the possibility to realize novel nanoscale magneto - electronic devices .
a detailed analysis of the magnetic properties is , therefore , important in order to correlate them with structure and morphology of vo@xmath3-nts for tailoring the magnetic properties of this nanostructured material . in order to obtain a deeper insight into the magnetism of vo@xmath3-nts we have measured magnetization and nuclear magnetic resonance ( nmr ) of pristine , i.e. li - undoped , samples .
the static magnetic data give evidence for the occurrence of magnetically nonequivalent vanadium sites in the structure .
these sites can be presumably attributed to v@xmath10 ( @xmath11 , @xmath0 ) ions in the octahedral and tetrahedral oxygen coordination , respectively .
the spins in the octahedral sites ( chain sites ) , are strongly antiferromagnetically correlated .
one part of them is coupled in dimers and exhibits a spin - gap behavior . the other part forms trimers .
they , together with much weaker correlated tetrahedral magnetic sites , dominate the low - temperature static magnetic response .
nmr measurements on @xmath12v nuclei reveal relaxation channels with two distinctly different relaxation rates .
though the longitudinal relaxation rate @xmath13 exhibits a strong temperature dependence the manifestation of a spin gap is obscured due to the presence of an appreciable amount of magnetic sites at low temperatures .
we discuss possible mechanisms of vanadium nuclear spin relaxation and their interplay with the static magnetic properties .
in particular , both nmr and the static susceptibility data give strong indications that owing to the pronounced structural low - dimensionality and mixed valency of vo@xmath3-nts different spin arrangements , namely individual spins , spin dimers and trimers , coexist in comparable amounts in this complex material . in an external magnetic field of @xmath14 t (
left ordinate ) and of @xmath15 ( right ordinate ) .
the dashed line is a curie - weiss fit of the low temperature data to eq .
( [ chicw ] ) .
( for details see the text . ) ]
samples of vo@xmath3-nts were synthesized using a hydrothermal procedure described in ref . .
the high quality of the samples has been verified by a number of techniques , such as high - resolution and analytical transmission electron microscopy ( tem ) , electron diffraction , x - ray photoemission spectroscopy and electron energy - loss spectroscopy ( eels)(for details see ref . ) . in this preparation method
the nanotubes occur in two topological modifications , i.e. concentric shell tubes with closed edges of the vo@xmath3 sheets and scrolled tubes with open edges , with the latter topology being a predominant one . @xcite as one can see in fig . 1 of ref . , however , the tube walls in our samples consist of a large number ( @xmath16 ) of vo@xmath3 layers .
therefore one expects that such differences in the microscopic topology , yielding absence or presence of open vanadium edge sites plausibly different from those in the bulk of the wall , should not affect appreciably the bulk magnetic properties . here
, static magnetization measurements have been performed in the temperature range 2 - 500k with a squid magnetometer from quantum design in magnetic fields up to 5 t and with a home - made vibrating sample magnetometer ( vsm ) in fields up to 15 t. upon heating , however , we observed indication that , above @xmath17370k , the material degenerates so that we restrict the data presented here to temperatures below 370k .
presumably this sample degradation is due to the changes of the oxygen stoichiometry .
we note , however , that on the time scale of two months repeated magnetization measurements of the samples kept in air yielded identical results which assures the sample stability if overheating is avoided .
the @xmath12v - nmr experiments were carried out with a tecmag pulse solid - state nmr spectrometer in a field of 7.05 t in a temperature range 15 - 285k .
_ experiment .
_ the temperature dependence of the static susceptibility @xmath18 obtained in a magnetic field of @xmath19 t is shown in fig.[chit ] .
these data are similar to the findings in ref . .
one can distinguish between a low - temperature regime below @xmath20k characterized by a curie - like behavior , which is reflected by a linear temperature dependence of @xmath15 ( cf .
fig.[chit ] ) . at higher temperatures
, however , there are strong deviations from the linear behavior .
the latter can be attributed to an additional contribution to the static susceptibility @xmath21 arising at higher temperatures . in this case
@xmath21 can be obtained by subtracting from the raw data the curie - weiss term @xmath22 and a small temperature independent contribution @xmath23 owing to the van - vleck paramagnetism of the v ions : @xmath24 here , @xmath25 is the curie constant , assuming the spin @xmath0 and the @xmath26 factor @xmath27 , and @xmath28 is the curie - weiss temperature .
the best fit of the low - temperature data yields the concentration of v@xmath10 spins contributing to the curie - weiss susceptibility @xmath29% of all v ions , @xmath30k and @xmath31 emu / mole .
the additional contribution @xmath21 is shown in fig.[chis ] . to the total susceptibility , eq .
( [ chidiff ] ) . a fit to a dimer model , eq .
( [ chidimer ] ) , is shown by a solid line ( red color online ) which is practically undistinguishable from the data .
( for details see the text ) ] consistently with the results of ref . , @xmath21 can be explained with a model of noninteracting antiferromagnetic dimers : @xmath32\right).\label{chidimer}\ ] ] fitting of the data in fig.[chis ] to eq .
( [ chidimer ] ) implies that at @xmath33% of all v sites there are spins @xmath34 which form antiferromagnetic dimers .
the dimer gap amounts to @xmath35k . with these parameters the fit yields @xmath36 with a good accuracy .
hence , the susceptibility data suggest that in a large temperature range the static magnetic response of vo@xmath3-nts can be very well described by a curie - weiss contribution @xmath22 , the van - vleck paramagnetism @xmath23 and the presence of antiferromagnetic dimers . however
, below @xmath37k the experimental data deviate from this description , ( fig.[chis ] ) . here
, @xmath38 is practically zero owing to a large value of @xmath39 .
thus the strong upturn of @xmath21 below 15k implies the occurrence of another contribution to the static susceptibility not accounted for in eqs .
( [ chidiff ] ) and ( [ chidimer ] ) . in order to address the low temperature magnetism , the field dependence of the magnetization @xmath40
up to 15 t , at @xmath41k , has been measured .
the data presented in fig.[mh ] display a nearly linear field dependence of @xmath42 at high magnetic fields and a pronounced curvature at lower fields without measurable hysteresis .
this @xmath40 curve can be very well fitted ( fig.[mh ] ) as a sum of two contributions @xmath43 the first part of eq .
( [ mfit ] ) describes an appreciable linear contribution to @xmath42 determined by a field - independent susceptibility @xmath44 emu / mole .
the second part accounts for a nonlinearity of @xmath40 due to the alignment of the spins by a magnetic field at low temperatures . here
@xmath45 is the brillouin function with @xmath46 , and @xmath47 being the mean - field parameter .
assuming the @xmath26 factor @xmath27 and the spin @xmath0 one obtains from the fit a small mean - field parameter @xmath48 mole / emu and the concentration of such polarisable , i.e. quasi - free , spins @xmath49 of all v sites at @xmath41k .
k measured in ascending and descending fields .
note the absence of hysteresis .
a fit to eq .
( [ mfit ] ) is shown by a solid line ( red color online ) which is practically undistinguishable from the data .
contribution owing to quasi - free spins described by the brillouin function is shown by a dashed line .
( for details see the text ) ] _ discussion .
_ we will start the discussion of the magnetization data by considering the response at @xmath50k , where the susceptibility can be described well in terms of non - interacting dimers and the curie - weiss law .
different contributions to the magnetic susceptibility can be related to the occurrence of nonequivalent vanadium sites in the crystal structure of vo@xmath3-nts.@xcite the octahedrally coordinated v(1 ) and v(2 ) sites are found in the double layers ( fig .
[ structurevox ] ) .
tetrahedrally coordinated v(3 ) sites occur between the layers .
the bond valence sums calculated for a closely related material bav@xmath4o@xmath5h@xmath6o suggest that vanadium in the v(3 ) position has a valency close to 4 + ( @xmath11 , @xmath0 ) , whereas vanadium in the positions v(1 ) and v(2 ) may be 4 + as well as 5 + ( @xmath51 , @xmath52).@xcite the mixed - valency in vo@xmath3-nts is confirmed by recent eels measurements which yield 60% of v@xmath10 and 40% of v@xmath53 in our samples.@xcite thus the concentration of magnetic v sites with @xmath34 amounts to 60% .
this estimate is consistent with our nmr data ( see section [ nmr ] below ) but , however , at first glance numerically contradicts the static magnetization data .
indeed , the number of magnetically active sites contributing to the curie - weiss susceptibility and to the dimer susceptibility makes up @xmath54%+@xmath55%=45% only .
the number @xmath56 , however , does not necessarily count only single @xmath34 v sites , which we will discuss in the following .
it is reasonable to assume that owing to magnetic frustration the v(3 ) spins between the layers are interacting very weakly and that all magnetic v(3 ) sites are therefore contributing to the curie - like behavior of the static susceptibility.@xcite the amount of v(3 ) sites in the unit cell is @xmath57% .
a larger number of weakly interacting curie - like 1/2-spins @xmath29% estimated from the analysis of the susceptibility ( fig .
[ chit ] ) suggests that quasi - free spins are not confined to the interlayer v(3 ) sites only .
the mixed valency of vanadium in vo@xmath3-nts results in a heterogeneous charge distribution in the zig - zag chains comprising the walls ( fig .
[ structurevox ] ) . assuming that most of tetrahedral v(3 ) sites are magnetic ( @xmath58% ) and recalling that the eels data yield 40% of nonmagnetic v@xmath53 ions in the structure which
thus should mostly reside in the chains , one finds almost equal number of spin- ( @xmath0 ) and spinless ( @xmath52 ) sites in the chains .
thus , about a half of the sites in the chain are magnetic .
recalling our estimate of the concentration of spins in dimers @xmath33% which all should reside in the chains , we obtain that those spins constitute there almost 2/3 of magnetic chain sites .
hence , there is still a significant number of spins in the chains amounting to @xmath59 of all v sites in the unit cell , which are not coupled in dimers .
if they were individual quasi - free spins , they would be expected to contribute to the curie - weiss susceptibility , similar to the v(3 ) interlayer sites .
this would yield the total concentration of curie - like spins @xmath60 , which is almost twice as large as the experimentally determined value @xmath29% .
therefore , this apparent discrepancy suggests that the remaining spins in the chain which are not involved in antiferromagnetic dimers are not independent but might form longer spin - chain fragments .
indeed , it has been recently shown both theoretically and experimentally that the ground state configuration of a strongly hole - doped antiferromagnetic heisenberg chain is composed of spin chain fragments of different length , i.e. of spin dimeres , trimers and monomers ( individual spins ) etc.@xcite remarkably , for small external fields , trimers respond similarly to free spins at low temperatures ( see e.g. ref . ) .
thus @xmath29% of curie - like spins estimated from our susceptibility data may comprise magnetic interlayer v(3 ) sites , i.e. quasi - free spins , as well as trimers in the v chain . in case of antiferromagnetic trimers ,
the number of magnetically active spins involved in the curie - like response is three times larger than for single spins , i.e. for a number of spins @xmath61 arranged in trimers only @xmath62 will be counted in @xmath56 . assuming that the remaining spins in the chains which are not involved in dimers , all form trimers , i.e. @xmath63% , one obtains a correct estimate of the total concentration of magnetic @xmath0 sites in the sample @xmath64% , which is in a good agreement with the eels data . to summarize
the above discussion the approximate fractional weights of different spin species are depicted in the diagram in fig.[diagram ] .
the above data analysis hence strongly suggests the occurrence in the v chains of spin - chain fragments of different lengths , i.e. individual spins , antiferromagnetic dimers and trimers .
one might speculate if even longer fragments occur .
the presence of chain fragments with an even number of sites , e.g. of quadrumers , however , can be ruled out by the experimental data : for quadrumers , one would expect the gap @xmath65k , @xcite which is not observed in our data .
we also note , that the singlet - triplet excitation in the trimer occurs at @xmath66 which can not be investigated since the sample thermally degenerates at 370k . -nts .
( for details see the text ) ] finally , we discuss the magnetization data at a low temperature @xmath41k ( fig.[mh ] ) which reveals a substantially different magnetic response as compared to the results at @xmath67k . in particular , the concentration of quasi - free spins , i.e. monomers ( individual spins ) or trimers , is strongly reduced down to 3% .
this clearly indicates that most of the monomers / trimers are coupled antiferromagnetically at this temperature .
this scenario is corroborated by the observation of a significant linear contribution @xmath68 to the @xmath42 vs. @xmath69 curve .
the fact that @xmath70 is constant in the whole field range implies that the antiferromagnetic coupling of the monomers / trimers is larger than the energy @xmath71 of the applied field of 15 t .
this scenario of antiferromagnetic coupling , however , does not straightforwardly explain the _ additional _ contribution to @xmath72 below @xmath37k ( cf .
fig.[chis ] ) .
one may speculate that anisotropic coupling mechanisms , like e.g. the dzyaloshinskii - moriya exchange , have to be considered .
this requires , however , a detailed symmetry analysis of the exchange paths between vanadium spins in vo@xmath3-nts . _ experiment . _
the @xmath12v - nmr experiments were carried out in a field of 7.05 t in a temperature range 15 - 285k .
@xmath12v nucleus has a spin @xmath73 and possesses a nuclear quadrupole moment .
therefore it is sensitive to both , the magnetic degrees of freedom and to the charge environment .
the nmr spectra were acquired by a point - by - point frequency sweeping . at each frequency point
the signal has been obtained by a fourier transformation of the hahn spin echo @xmath74 with the pulse separation time @xmath75 .
the final frequency - swept spectrum has been calculated by sequential summing of the signals at three successive frequency points . in fig.[spectrum ] a spectrum at @xmath76k is shown as an example .
the spectrum is broadened due to a different orientation of nanotubes in a powder sample and contains several unresolved quadrupole satellites .
the relaxation times @xmath77 and @xmath78 were measured at the frequency of the maximum intensity of the spectrum which was assigned to the main @xmath79 nmr transition .
the transversal relaxation time @xmath77 has been determined from the hahn spin echo decay which can not be described by a single exponent . however it could be very well fitted assuming the two - exponential behavior with two distinctly different relaxation times . to measure the longitudinal relaxation time @xmath78
a method of stimulated echo employing a pulse sequence @xmath80 with @xmath81 has been used .
the time dependence of the stimulated spin echo intensity @xmath82 , which is a measure of the decay of the longitudinal magnetization characterized by the relaxation time @xmath78 , has been analyzed following the standard equation for the spin @xmath73 derived in ref . .
also in this case the decay of @xmath82 can not be described by a one - center fitting .
the fit requires the presence of two contributions of different weights with distinctly different relaxation rates .
thus for the following discussion it is useful to define the fast and the slow relaxation processes .
the temperature dependence of the fast and slow transverse and longitudinal relaxation rates @xmath83 and @xmath13 is shown in fig.[rates ] .
slow contribution to @xmath83 is temperature independent whereas the fast one as well as both contributions to @xmath13 increase with increasing temperature .
the ratio of the weights of these two contributions to the relaxation decay @xmath84 at @xmath76k is close to 2.4 .
however the weight of the slow component strongly increases with decreasing temperature yielding @xmath85 at @xmath86k ( fig.[ratio ] ) . and @xmath83 : ( a ) as a function of temperature ; ( b ) as a function of inverse temperature .
dashed line denotes the expected behavior of @xmath13 according to eq .
( [ spingap ] ) due to the opening of the spin gap @xmath87k . ] _ discussion .
_ as discussed in section [ statmag ] vanadium ions in vo@xmath3-nts occur in different valence states and can be magnetic and nonmagnetic .
the ions possessing an electronic magnetic moment produce a fluctuating magnetic field at the nuclei .
it leads to a fast relaxation of the nuclei of the magnetic ions due to the on - site hyperfine interaction .
the magnitude of the transferred hyperfine field at the nuclei of nonmagnetic ions is much smaller and the relaxation is more slow .
the temperature independent slow component of @xmath83 can obviously be assigned to the conventional magnetic dipolar interaction ( see , e.g. , ref . ) . however with
increasing temperature a more effective relaxation channel appears .
the fact that the temperature dependence of the fast @xmath83 contribution is similar to that of the both components of @xmath13 ( fig .
[ rates ] ) gives evidence that the origin of this contribution is the relaxation due to the fluctuating magnetic field produced by electron spins at the nuclei . in this case
according to the redfield s theory @xcite @xmath13 and @xmath83 are determined by the magnitude of the transferred field @xmath88 ( @xmath89 and @xmath90 ) and the correlation time @xmath91 of the electron spins : @xmath92\tau_0/(1+\omega_n^2\tau_0 ^ 2),\\ t_2^{-1}=\gamma_n^2[\overline{h_z^2}\tau_0 + \overline{h_y^2}\tau_0/(1+\omega_n^2\tau_0 ^ 2)].\end{aligned}\ ] ] here @xmath93 is the nuclear gyromagnetic ratio and @xmath94 is the nmr frequency , respectively . in the regime of fast fluctuations , @xmath95 , which is expected far away from magnetically ordered phases , the observed strong increase of the relaxation rates with temperature ( fig .
[ rates]a ) suggests the enhancement of @xmath88 at the positions of the nuclei owing to , e.g. , an increase of the concentration of the spin centers .
moreover , the change of the ratio of the weight coefficients of two contributions to the relaxation decay @xmath84 with temperature ( fig . [ ratio ] ) can be explained by taking into account the magnetic susceptibility data which indicate that part of magnetic ions form antiferromagnetic dimers and exhibit a spin gap behavior . usually owing to a spin gap
the huge slowing down of relaxation with decreasing temperature is observed , as , e.g. in the case of the low - dimensional vanadium oxides , it has been observed in the nmr relaxation measurements of a two - leg spin - ladder cav@xmath6o@xmath96,@xcite or a quarter - filled ladder compound nav@xmath6o@xmath96.@xcite the occurrence of a spin gap @xmath39 yields an exponential temperature dependence of the longitudinal relaxation rate @xmath13 for @xmath97 : @xcite @xmath98 . \label{spingap}\ ] ] with the dimer gap @xmath35 k deduced from our magnetization data one would expect a strong decrease of the nuclear relaxation below 100k , which is not observed experimentally ( cf .
fig.[rates ] ) .
obviously , in our case this picture is disguised by the presence of nondimerized magnetic ions providing fast relaxation even at low temperature .
these magnetic centers can be assigned to the interlayer v(3 ) sites as well as to trimers in the v chains ( see section [ statmag ] ) .
the nuclei in dimerized ions are slowly relaxing at low temperatures because of the non - magnetic singlet ground state of a dimer and fast relaxing at high temperatures owing to thermally activated singlet - triplet excitations across the dimer gap .
hence the existence of a spin gap should be seen as a change of the weight coefficients of the respective @xmath13 contributions .
moreover , thermal activation of dimers at high temperatures should effectively look like an increase of the concentration of magnetic centers .
this should enhance the magnitude of the fluctuating magnetic field and the relaxation rate of the slow - relaxing nuclei .
indeed , as can be seen in fig .
[ ratio ] with increasing temperature the relaxation rate of the slow nuclei grows stronger than that of the fast ones .
the common temperature dependence of the ratios of the weights and of the relaxation times of two contributions ( fig . [ ratio ] ) justifies the same origin of both dependences which can be due to the occurrence of the spin gap for a part of magnetic vanadium ions .
the weight coefficients at high temperatures provides the ratio of the fast and slow relaxing centers approximately as 70%:30% .
this ratio is in a fair agreement with the eels estimation of the ratio between magnetic v@xmath10 and nonmagnetic v@xmath53 amounting to 60%:40%.@xcite therefore in the high temperature regime it is reasonable to associate the fast- and the slow - relaxing centers with v@xmath10 and v@xmath53 sites , respectively . from the weight ratio @xmath84 at low temperatures
one obtains the number of slow - relaxing
nuclei amounting to @xmath99 which comprises the nonmagnetic v@xmath53 sites as well as the dimerized centers .
as the high - temperature nmr estimate of the v@xmath53 sites amounts to @xmath100 , the concentration of the magnetic @xmath0 v@xmath10 sites coupled in antiferromagnetic dimers is , therefore , about 30% of all vanadium , in a good agreement with the estimate @xmath33% from the static magnetization data ( section [ statmag ] ) .
the nmr estimate of the fast - relaxing , i.e. paramagnetic centers , at low temperatures yields @xmath101 which is significantly larger than the concentration of the curie - like spins @xmath102% obtained from the analysis of the curie - like susceptibility @xmath103 ( eq . [ chicw ] ) .
this apparent contradiction can be resolved if not only individual spins but also trimers would contribute to @xmath103 at low temperatures , as has been suggested in the analysis of the static magnetic data in section [ statmag ] . assuming that the nuclei in a trimer belong to the fast - relaxing
centers , owing to a magnetic ground state of a trimer , all @xmath61 sites involved in trimers will be counted in the weight of the respective contribution to the nmr relaxation rate @xmath13 , whereas effectively only @xmath62 of those sites is counted in @xmath104 .
thus the analysis of the nmr data strongly supports the static magnetic results , both justifying the occurrence of comparable amounts of individual spins , antiferromagnetic dimers and trimers as depicted in fig.[diagram ] .
in summary , magnetization and @xmath12v nmr measurements of the samples of multiwalled vanadium - oxide nanotubes reveal a complex evolution of the static and dynamic magnetic properties as a function of temperature .
the results give strong indications that different spin arrangements beyond a simplified picture suggested in ref .
are realized in vo@xmath3-nts . in particular
, the data analysis strongly suggests that in addition to individual spins and antiferromagnetic dimers showing a spin gap of the order of 700k an appreciable amount of spin trimers occur in the samples .
though the estimate of the fractional weights of different spin species summarized in fig.[diagram ] might change to some extent owing to the experimental uncertainties , and also because structural imperfections as well as the proportion of the majority scrolled tubes to the minority concentric tubes may vary to some extent from sample to sample , the data give evidence for the coexistence of different spin configurations in comparable amounts which is related to the complex low - dimensional crystallographic structure as well as mixed valency of this nanoscale magnet .
this work was supported by the deutsche forschungsgemeinschaft ( dfg ) through project kl 1824/2 .
the work of ev was supported by the dfg grant 436rus17/38/06 , joint dfg - russian academy of sciences project on physics of novel materials 436rus113/780/0 - 1 and also by the russian foundation for basic research through grant no .
04 - 02 - 17137 . | measurements of the static magnetic susceptibility and of the nuclear magnetic resonance of multiwalled vanadium - oxide nanotubes are reported . in this nanoscale magnet
the structural low - dimensionality and mixed valency of vanadium ions yield a complex temperature dependence of the static magnetization and the nuclear relaxation rates .
analysis of the different contributions to the magnetism allows to identify individual interlayer magnetic sites as well as strongly antiferromagnetically coupled vanadium spins ( @xmath0 ) in the double layers of the nanotube s wall .
in particular , the data give strong indications that in the structurally well - defined vanadium - spin chains in the walls , owing to an inhomogeneous charge distribution , antiferromagnetic dimers and trimers occur .
altogether , about 30% of the vanadium ions are coupled in dimers , exhibiting a spin gap of the order of 700k , the other @xmath1% comprise individual spins and trimers , whereas the remaining @xmath2% are nonmagnetic . |
there has been a lot of speculations about the appearance of new particles in the interior of neutron stars .
a traditional neutron star consists of neutrons , protons , and leptons only .
cameron suggested first in 1959 , that hyperons will also appear at high baryon density @xcite .
a possible phase transition to quark matter was conjectured shortly after the quark model was introduced @xcite .
pion condensation @xcite as well as kaon condensation were considered later @xcite .
finally , strange stars , built out of absolutely stable strange quark matter , have been studied within the mit bag model @xcite .
the appearance of one or the other exotic phase in the core of neutron stars is still a matter of active debates .
nevertheless , recent developments in the field indicate , that hyperons are probably the first exotic particle to appear in neutron star matter at twice normal nuclear density .
this result was consistently derived within various , different models : nonrelativistic potential models @xcite , quark - meson coupling model @xcite , relativistic mean - field models @xcite , relativistic hartree - fock @xcite , brueckner - hartree - fock models @xcite , and chiral effective lagrangians @xcite .
hence , the onset of hyperons at @xmath0 seems to be rather insensitive to the underlying details of the hyperon - nucleon interaction used .
if there are a lot of hyperons in the interior of neutron stars , then it might be also possible to form dibaryons with strangeness . besides the famous h - dibaryon proposed by jaffe @xcite built out of six quarks ,
the most recent version of the nijmegen potential predicts the existence of deeply bound states of two hyperons , involving the @xmath1 s and @xmath2 s with maximum isospin @xcite .
their results can be understood in terms of the underlying su(3 ) symmetry for the baryon - baryon interactions ( see @xcite ) .
denotes the optical potential of the h - dibaryon at normal nuclear density . for an attractive potential , the equation of state
is considerably softened ( taken from @xcite ) . ]
let us assume in the following , that there exist bound states with hyperons .
what would be the consequences for the properties of dense matter , as present in the interior of neutron stars ? bound systems in dense matter have been discussed before in connection with the liquid - gas phase transition of nuclear matter .
they dissolve at rather low density due to effects in analogy to the mott transition ; deuterons disappear at @xmath3 and @xmath4 particles around @xmath5 , as detailed e.g. in @xcite .
a similar transition will happen for bound states with hyperons .
hence , weakly bound systems will have no influence on the equation of state of neutron star matter unless , there is a deeply bound state which can be treated as a quasiparticle .
the h - dibaryon as well as heavier quark bags with strange quarks ( strangelets ) are such deeply bound states .
they can actually be formed as precursor of the deconfinement phase transition in the core of neutron stars .
the general condition for the appearance of a strangelet in s - wave with baryon number @xmath6 and charge @xmath7 is that the effective energy of the particle equals its chemical potential @xmath8 here , @xmath9 stands for the effective potential felt by the strangelet at finite density . in order to estimate the effects of these exotic states , we focus on the lightest possible one , the h - dibaryon with @xmath10 and @xmath11 .
we use an extended relativistic mean - field model which incorporates the h - particle in a consistent scheme .
the lagrangian part for the h - dibaryon field reads @xmath12 with a minimal coupling scheme for the scalar and vector meson fields @xmath13 for the coupling constant to the vector field , we choose the simple quark counting rule as guidance , i.e. we set @xmath14 .
we fix the scalar coupling to an optical potential at normal nuclear matter density in the range @xmath15 figure [ fig : eosh ] shows the resulting equation of state when using the parameter set tm1 and including effects from hyperons ( see @xcite for details ) .
the equation of state turns out to be quite sensitive to the choice of the optical potential of the h - dibaryon .
the phase transition to the h - dibaryon condensate is of second order in all cases studied . for an attractive optical potential , @xmath16 mev ,
the equation of state has a plateau - like behaviour , i.e. the pressure is very slowly rising with energy density .
the drastic softening of the equation of state has crucial consequences for the maximum mass of a neutron star , which will be lowered below the acceptable value of the hulse - taylor pulsar of @xmath17 .
hence , one can constrain the properties of the h - dibaryon in dense matter from pulsar data .
these constraints are summarized in fig .
[ fig : hconstrain ] as a function of the vacuum mass of the h - dibaryon and its optical potential at @xmath18 .
the calculations have been done by using two completely different parameterizations for the nucleon and the hyperon coupling constants ( denoted as set tm1 and gm91 in the figure ) and our findings seems to be rather insensitive to these choices . according to these calculations ,
the shaded area in the lower left side of the plot can be ruled out , as the corresponding neutron star equation of state gives too low a maximum mass .
therefore , a deeply attractive potential for the h - dibaryon with a mass close to the @xmath19 threshold is incompatible with the measured hulse - taylor pulsar mass . also , a h - dibaryon mass below the @xmath1n threshold requires even a repulsive potential in matter in our model calculation to get sufficiently heavy neutron stars .
finally , we remark that the existence of deeply bound states of two hyperons can have an indirect effect on the properties of neutron stars , though . if the hyperon - hyperon interaction is deeply attractive , it can give rise to a phase transition to hyperonic matter @xcite .
this phase transition can result in a drastically modified mass - radius relation for neutron stars , like neutron star twins with similar masses but smaller radii than ordinary neutron stars @xcite .
hence , measurements of the mass and the radius of neutron stars will provide important insights into the equation of state and may reveal that there is indeed something strange going on at high density .
there is the possibility to probe the hyperon - hyperon interaction in less remote places than neutron stars .
terrestrial relativistic heavy - ion collisions will produce dozens of hyperons in a single central collisions of two heavy nuclei , and for rhic even close to 200 hyperons are expected @xcite .
the hyperons , sitting close in phase - space , can coalesce to form a bound state or a resonance which is decaying afterwards .
the final products will be measured then in the detectors .
the production rates for dibaryons with hyperons at rhic has been estimated in the coalescence model to be in the range of @xmath20 to @xmath21 per single event .
effects from the details of the dibaryon wavefunction have been found to be rather small @xcite .
the rapidity distribution is rather flat , so that dibaryons are also produced at forward and backward rapidities where the decay length is considerably longer than at midrapidity .
the production rates can be enhanced dramatically in two ways which are not taken into account in the above coalescence estimates .
if a quark - gluon plasma is formed , strange quark - antiquark pairs are produced more abundantly so that the total initial number of produced strange hadrons increases .
if matter is created in a chirally restored phase , the effective hyperon masses are reduced drastically which will result in an enhanced hyperon - antihyperon yield .
@lllll & & & & & & @xmath22 & 3.25 & 3.25 & 22.1 & 22.1 @xmath23 & @xmath242.37 & @xmath242.30 & @xmath2416.0 & @xmath2415.6 @xmath25 & 0.13 & 0.0 & 42.2 & 40.0 @xmath26 & @xmath243.27 & @xmath243.33 & 26.6 & 28.3 @xmath27 & 4.27 & 4.71 & @xmath241.44 & 0.0 @xmath28 & 3.43 & 3.19 & @xmath2412.3 & @xmath2411.7 @xmath29 & @xmath244.51 & @xmath244.51 & 16.6 & 16.6 + to detect strange dibaryons , their weak decay patterns have to be investigated .
starting point is the weak nonleptonic decay of hyperons in vacuum .
both , s - wave and p - wave amplitudes have been measured and correspond to a parity violating and a parity conserving amplitudes , respectively .
the standard approach for describing these amplitudes is by means of chiral perturbation theory .
the p - wave amplitudes are beyond leading order and are usually derived within the pole model .
the basic version fails to describe the p - wave amplitudes .
this constitutes the so called s - wave / p - wave puzzle for the weak nonleptonic decay of hyperons ( for a discussion and more elaborate ways to remedy the situation see @xcite and references therein ) .
we have found a simple way to parameterize both amplitudes , s - wave and p - wave , in terms of pure su(3 ) symmetry for the weak interactions @xcite .
the effective lagrangian involves the baryon octet @xmath30 , the pseudoscalar octet @xmath31 , and the gell - mann matrix @xmath32 , which ensures the @xmath33 rule and a change of hypercharge by one unit : @xmath34 + f \cdot { \rm tr } \bar b \left[p,\lambda_6\right ] b \nonumber \\ & & { } + g \cdot { \rm tr } \bar b p \gamma_5 b \lambda_6 + h \cdot { \rm tr } \bar b \lambda_6 \gamma_5 b p + j \cdot { \rm tr } \bar b \ { p,\lambda_6\ } \gamma_5 b \label{eq : lagweak } \quad .\end{aligned}\ ] ]
the first two terms give the s - wave contributions , while the other three the p - wave contributions .
table [ tab : weakamp ] compares the resulting amplitudes with the measured ones when setting @xmath35 , @xmath36 , @xmath37 , @xmath38 , and @xmath39 ( in units of @xmath40 ) .
we conclude that the above model as defined in eq .
( [ eq : lagweak ] ) gives a reasonable description of all amplitudes .
the weak decays of possible dibaryons with hyperons are then calculated by adopting a hulthen - type wavefunction with varying binding energy .
the nonmesonic decay is described by pion- and kaon- exchange diagrams , where the weak coupling constants are given by eq .
( [ eq : lagweak ] ) and the strong coupling constants by su(6 ) symmetry .
we note in passing , that the state - of - the - art calculations for the weak nonmesonic decay of hypernuclei rely on the ( wrong ) pole model ( see @xcite and references therein ) . while this will not affect the pion exchange diagrams , accidentally
, it will certainly alter the coupling constants for the kaon exchange diagrams .
the use of the su(3 ) symmetric model might be interesting to pursue in this direction to investigate possible effects , e.g. for the ratio of neutron- to proton - induced decay which comes out too small in the standard approach .
figure [ fig : lala ] depicts the branching ratio for a bound @xmath19 system versus the binding energy , and fig .
[ fig : xi0p ] shows the case for a possible @xmath42p dibaryon .
the mesonic decay channel dominates for binding energies of a few mev .
then , for higher binding energies , the nonmesonic decay channels take over . for @xmath19 ,
the dominant nonmesonic decay mode is to a @xmath44 and a proton , the same as for the hypothetical h - particle .
the @xmath42p decays mainly to a @xmath45 and a proton for binding energies of just 1.5 mev and more .
this weak decay has the same decay topology as the weak decay of the @xmath46 and the @xmath47 , and both particles have readily been measured in relativistic heavy - ion collisions @xcite .
other interesting decay topologies occur for the doubly negatively charged dibaryons @xmath48 , @xmath49 , and @xmath50 , all of which have been predicted to be bound in the recent nijmegen model @xcite . here
, a negatively charged object decays into two negatively charged tracks a unique decay prong which should be easily seen in tracking devices .
the calculated lifetimes of all dibaryons are lying just below the one for the free @xmath45 with decay lengths in the range of @xmath515 cm .
there are in principle three ways to detect possible short - lived strange dibaryons ( for an experimental investigation we refer to the detailed feasibility study of coffin and kuhn for the case of the h - dibaryon at the star detector @xcite ) .
firstly , if the dibaryon is bound , one can look for exotic decays in a tpc , like a charged track splitting into two charged ones . secondly ,
if the mass of the strange dibaryon is close to threshold , it will show up in the invariant mass spectrum of its decay products by using background subtraction .
note , that this method is also sensitive to dibaryon resonances , if their decay widths are not too large @xcite .
thirdly , the hyperon - hyperon interaction and possible broad dibaryon resonances can be probed in correlation functions as the low momentum part is sensitive to final state interactions @xcite . | the formation of dibaryons with strangeness are discussed for the interior of neutron stars and for central relativistic heavy - ion collisions .
we derive limits for the properties of h - dibaryons from pulsar data .
signals for the formation of possible bound states with hyperons at bnl s relativistic heavy - ion collider ( rhic ) are investigated by studying their weak decay patterns and production rates . |
the technique of interferometry has been widely used in radio astronomy to image the sky using arrays of antennas . by correlating the complex voltage signals between pairs of antennas ,
the field - of - view of a single element can be sub - divided into `` synthesized beams '' of higher angular resolution . in the small - angle approximation ,
the interferometer forms the fourier transform of the sky convolved with the autocorrelation of the aperture voltage patterns .
in standard radio interferometric data analysis , as described for example in the text by @xcite and the proceedings of the nrao synthesis imaging school @xcite , the correlations or visibilities are inverse fourier transformed back to the image plane . however , there are applications such as estimation of the angular power spectrum of fluctuations in the cosmic microwave background ( cmb ) where it is the distribution of and correlation between visibilities in the aperture or @xmath0-plane that is of most interest . in standard cosmological models ,
the cmb is assumed to be a statistically homogeneous gaussian random field @xcite . in this case , the spherical harmonics of the field are independent and the statistical properties are determined by the power spectrum @xmath1 where @xmath2 labels the component of the legendre polynomial expansion ( and is roughly in inverse radians ) .
@xcite showed that in cold dark matter inspired cosmological models , there would be features in the cmb power spectrum that reflected critical properties of the cosmology .
recent detections of the first few of these `` acoustic peaks '' at @xmath3 in the power spectrum @xcite have supported the standard inflationary cosmological model with @xmath4 .
measurement of the higher-@xmath2 peaks and troughs , as well as the damping tail due to the finite thickness of the last scattering surface , is the next observational step .
interferometers are well - suited to the challenge of mapping out features in the cmb power spectrum , with a given antenna pair probing a characteristic @xmath2 proportional to the baseline length in units of the observing wavelength ( a @xmath5 projected baseline corresponds to @xmath6 , see [ sec : basic ] ) .
there are many papers in the literature on the analysis of cmb anisotropy measurements , estimation of power spectra , and the use of interferometry for cmb studies .
general issues for analysis of cmb datasets are discussed in @xcite .
@xcite present a bayesian method for the analysis of cmb interferometer data , using the 3-element cosmic anisotropy telescope data as a test case .
a description of analysis techniques for interferometric observations from the degree angular scale interferometer ( dasi ) were presented in @xcite , while @xcite report on the power spectrum results from first - season of dasi observations .
@xcite discusses cmb interferometry with application to the proposed amiba instrument .
@xcite have recently presented an approach similar to ours , and demonstrate their technique on simulated very small array ( vsa ) data ; a brief comparison of their algorithm with ours is given in appendix [ app : hobson ] . in this paper
, we describe a fast gridded method for the @xmath0-plane analysis of large interferometric data sets .
the basis of this approach is to grid the visibilities and perform maximum likelihood estimation of the power spectrum on this compressed data .
our use of gridded estimators is significantly different from what has been done previously .
in addition to power spectrum extraction , this procedure has the ability to form optimally filtered images from the gridded estimators , and may be of use in interferometric observations of radio sources in general .
we have applied our method to the analysis of data from the cosmic background imager ( cbi ) .
the cbi is a planar interferometer array of 13 individual 90-cm cassegrain antennas on a 6-m pointable platform @xcite .
it covers the frequency range 2636 ghz in 10 contiguous 1 ghz channels , with a thermal noise level of @xmath7k in 6 hours , and a maximum resolution of @xmath8 limited by the longest baselines .
the cbi baselines probe @xmath2 in the range 5003900 .
the 90-cm antenna diameters were chosen to maximize sensitivity , but their primary beamwidth of @xmath9 ( fwhm ) at 31 ghz limits the instantaneous field of view , which in turn limits the resolution in @xmath2 .
this loss of aperture plane resolution can be overcome by making mosaic observations , i.e. observations in which a number of adjacent pointings are combined @xcite . in the cbi observations , mosaicing a field several times larger than the primary beam has resulted in an increase in resolution in @xmath2 by more than a factor of 3 , sufficient to discern features in the power spectrum .
the first cbi results were presented in @xcite , hereafter paper i , using earlier versions of the software that did not make use of @xmath0-plane gridding , and were far too slow to be used on larger mosaiced data sets .
it was therefore essential to develop a more efficient analysis method that would be fast enough to carry out extensive tests on the cbi mosaic data .
the software package described below has been used to process the first year of cbi data . in the companion papers (
* hereafter paper ii ) and ( * ? ? ?
* hereafter paper iii ) , the results from passing cbi deep - field data and mosaic data respectively through the pipeline are presented .
this paper is paper iv in the series . the output from this pipeline is then used to derive constraints on cosmology ( * ? ?
* hereafter paper v ) .
finally , analysis of the excess of power at high-@xmath2 seen in results shown in paper ii in the context of the sunyaev - zeldovich effect is carried out , again using the method presented here , in ( * ? ? ?
* hereafter paper vi ) . an introduction to the properties of the cmb power spectrum ,
the response of an interferometer to the incoming radiation , and the computation of the primary beam are given in sections
[ sec : pscmb ] , [ sec : basic ] , and
[ sec : gauss ] respectively .
the gridding process is presented in [ sec : gridmethod ] , followed by a description of the likelihood function and construction of the various covariance matrices in [ sec : liklhd ] .
details on the maximum likelihood solution and the calculation of window functions and component bandpowers is given in
[ sec : relax ] , while [ sec : image ] presents our method for making optimally filtered images from the gridded estimators .
finally , a description of the cbi implementation of this method and the performance of the pipeline , including demonstrations using simulated cbi datasets , is given in
[ sec : estliklhd ] , followed by a summary and conclusions in [ sec : conclude ] .
at small angles , the curvature of the sky is negligible and we can approximate the spherical harmonic transform of the the temperature field @xmath10 in direction @xmath11 as its fourier transform @xmath12 @xcite , where @xmath13 is the conjugate variable to @xmath11 .
we adopt the fourier convention @xmath14 of @xcite , @xcite , and @xcite . in terms of the multipoles @xmath2 , @xmath15 which we simplify to @xmath16 for the @xmath17 of interest in this paper . for the low levels
anisotropy seen in the cmb on these scales , it is convenient to give @xmath18 in units of @xmath19k and thus @xmath1 will be in units of @xmath20 . because the cmb is assumed to be a statistically homogeneous gaussian random field , the components of its fourier transform are independent
gaussian deviates .
@xmath21 where @xmath22 . because @xmath10 is real
, its transform must be hermitian , with @xmath23 , and therefore @xmath24 note that it is common to write the cmb power spectrum @xmath1 in a form @xmath25 @xcite .
constant @xmath26 corresponds to equal power in equal intervals of @xmath27 .
although the power spectrum @xmath1 is defined in units of brightness temperature , the interferometer measurements carry the units of flux density @xmath28 ( janskys , 1 jy = @xmath29 w m@xmath30 hz@xmath31 ) . in particular , the intensity field on the sky @xmath32 has units of specific intensity ( w m@xmath30 hz@xmath31 sr@xmath31 or jy / sr ) , and thus to convert between @xmath33 and @xmath18 we use @xmath34 with the planck factor @xmath35 where @xmath36 corrects for the blackbody spectrum .
note that we have treated the temperature @xmath18 as small fluctuations about the mean cmb temperature @xmath37 k @xcite , and thus the @xmath36 appropriate to @xmath38 is used with @xmath39 at @xmath40 ghz .
we are not restricted to modeling the cmb .
for example , we might wish to determine the power spectrum of fluctuations in a diffuse galactic component such as synchrotron emission or thermal dust emission . in this case
, one might wish to express @xmath41 in jy / sr , but take out a power - law spectral shape @xmath42 where @xmath43 is the spectral index , and @xmath44 is the conversion factor that normalizes to the intensity @xmath45 at the fiducial frequency @xmath46 .
note that this normalization is particularly useful for fitting out centimeter - wave foreground emission , which tends to have a power - law spectral index in the range @xmath47 that is significantly different from that for the thermal cmb ( @xmath48 ) .
in addition , foregrounds will also tend to have a power spectrum shape different from that of cmb , which must be included in the analysis ( see [ sec : foreground ] below ) .
a visibility @xmath49 formed from the correlation of an interferometer baseline between two antennas with projected separation ( in the plane perpendicular to the source direction ) @xmath50 meters observed at wavelength @xmath51 meters measures ( in the absence of noise ) the fourier transform of the sky intensity modulated by the response of the antennas @xcite @xmath52 where @xmath53 is the primary beam , and @xmath54 is the conjugate variable to @xmath11 . for angular coordinates
@xmath11 in radians , @xmath13 has the dimensions of the baseline or aperture in units of the wavelength .
the fourier domain is also referred to as the _ uv_-plane or aperture plane in interferometry for this reason .
we define the direction cosines @xmath55 between the point at right ascension and declination @xmath56 and the center of the mosaic @xmath57 . for the cbi , data are taken keeping the phase center fixed on the pointing center @xmath58 by shifting the phase with the beam and rotating the platform to maintain constant parallactic angle during a scan , so that the response to a point source at the center of the field @xmath59 is constant , and thus @xmath60 in equation ( [ eq : point1 ] ) , where @xmath61 is the normalized primary beam response at the observing frequency of visibility @xmath62
. then , by application of the fourier shift theorem , it is easy to show that @xmath63 where @xmath64 is the fourier transform of the primary beam @xmath65 , and @xmath32 is the sky brightness field ( expressed in units such as jy / sr ) with transform @xmath66 . the instrumental noise on the complex visibility measurement is represented by @xmath67 .
the _ uv_-plane resolution of an interferometer in a single pointing is thus limited by the convolution with @xmath68 .
however , these sub - aperture spatial frequencies can be recovered by using the phase gradient in the exponential @xmath69 $ ] from a raster of mosaic pointings @xmath70 , provided that the spacing of the pointings is sufficiently small to avoid aliasing as discussed in appendix [ app : estim ] . to aid us later on ,
we introduce a convolution kernel @xmath71 and thus @xmath72 where @xmath73 is the planck conversion factor for the cmb given in equation ( [ eq : fplanck ] ) .
it is easiest to write these in operator notation , with @xmath74 where @xmath75 and @xmath76 are the visibility and noise vectors respectively , @xmath77 is our kernel , and @xmath78 is the transform of the temperature field . in this representation @xmath78 can be thought of as a vector of cells in _
uv _ space .
in order to determine the response of the array to the radiation field , we need to know the primary beam @xmath79 of the antenna elements and its fourier transform @xmath80 . in general , for each frequency channel , each baseline has a primary beam which is the fourier transform of the cross - correlation of the voltage illumination functions across the aperture of each antenna see @xcite for a detailed treatment of the interferometer response . for a real and symmetric primary beam that is identical between antennas , then the transforms are symmetric and real we can ignore the differences between cross - correlation and convolution and write @xmath81 for the voltage illumination function @xmath82 across the radius of the aperture @xmath83 at frequency @xmath84 , where @xmath85 is the fourier transform of @xmath86 .
the cbi beams have been measured and are nearly identical and symmetric , and thus we will use a single mean primary beam and its transform for the array . for a heterogeneous array ,
the individual beams can be used with some added complication . for most antennas , such as those used in the cbi
, the primary beam width scales linearly with observing wavelength , and thus @xmath87 is approximately constant with wavelength .
then , we can define @xmath88 as the normalized aperture autocorrelation function , and write @xmath89 for a channel centered at wavelength @xmath90 , with @xmath91 normalizes the response to give unity gain on - sky at the beam center ( @xmath92 ) . if @xmath93 , then @xmath94 .
the two - dimensional primary beam response , @xmath79 , is determined by means of measurements of a bright radio source over a two - dimensional grid of offset pointings centered on the source .
the central lobe of @xmath79 for the cbi is well approximated by a circular gaussian , which is characterized by its dispersion @xmath95 , which is related to the full width at half - maximum ( fwhm ) @xmath96 by @xmath97 .
the fourier transform of an infinite circular gaussian is given by @xmath98 where @xmath99 is the gaussian dispersion in fourier space .
the function @xmath88 is therefore @xmath100 for gaussian radius @xmath101 . for the cbi the measured primary beam
( see paper iii ) has @xmath102 , so @xmath103 at 31 ghz ( @xmath104 cm ) , which corresponds to @xmath105 cm . for the cbi software pipeline , instead of using a gaussian approximation to @xmath88 we have chosen to model the antenna illumination @xmath106 as a gaussian truncated at both the dish edge and the secondary blockage radius @xmath107 where for the cbi antennas @xmath108 cm .
note that if @xmath106 and @xmath88 were infinite circular gaussians , then @xmath109 .
a best - fit taper parameter @xmath110 is obtained using the measured primary beam @xmath111 , giving @xmath112 cm or an edge taper of 0.118 ( @xmath113 db of power ) at the dish edge .
we then numerically tabulate the autocorrelation @xmath88 assuming @xmath112 cm which is then interpolated in the code when @xmath68 is required .
this model is a better fit to the observed beam than a pure gaussian beam ( see figure 1 in paper iii for a plot of this model ) .
the resolution in @xmath0 or @xmath2 space is set by the width of @xmath114 . for a gaussian approximation to the beam ,
the dispersion in multipole @xmath2 is @xmath115 , and the fwhm is @xmath116 . for @xmath117 at 31 ghz we have fwhm @xmath118 ( @xmath119 ) . given that features are expected in the power spectrum of widths significantly less than this , it is highly desirable to reduce the effective resolution width of the cbi by at least a factor three using mosaicing .
the principal problem in using likelihood ( see [ sec : liklhd ] ) to determine confidence limits on the power spectrum for cbi data is the large number of visibilities compounded by the large number of mosaic pointings ( typically @xmath120 or larger ) .
even a modest reduction in the number of matrix elements passed to the likelihood calculation will greatly aid the computation .
this suggests that we grid the visibilities before computing the likelihood function . for an effective resolution in the aperture plane determined by the primary beam and mosaic
size , there is little use in sampling below this smearing scale , and we can define an optimum gridding scheme which minimizes the quantity of data and information loss ( the gridding is a form of compression ) .
we implement this by defining _ estimators _
@xmath121 for the true complex brightness transform which are linear combinations of the measured visibilities .
these estimators bin together data from the different frequency bands and mosaic pointings .
thus , a direct sum of visibilities taken at the same @xmath122 but over the whole mosaic @xmath123 will result in an estimator that has a higher effective resolution in the _ uv _ plane .
the result is that we can speed up the likelihood estimation at the cost of complicating the covariance matrix . in general , this matrix can be computed relatively quickly as it is a @xmath124 process , and thus this is a worthwhile trade - off versus the @xmath125 cost of calculating the likelihood .
the estimators derived in appendix [ app : estim ] are not orthogonal combinations of the original visibilities , and thus some information loss is expected .
however , the tests performed in [ sec : test ] show that these estimators are unbiased , and comparisons to results obtained using the visibilities directly show that there is no noticeable loss in sensitivity .
thus , our gridding can be considered to be an efficient form of ( lossy ) compression using the beam as a signal template . in appendix
[ app : estim ] , we argue that a @xmath126 formed by a linear combination of visibilities will give a estimate of the weighted average of @xmath127 or @xmath128 around _ uv _ locus @xmath129 .
we have from equation ( [ eq : moskern ] ) @xmath130 where the kernel @xmath131 is defined in equation ( [ eq : moskern ] ) and the kernel for the conjugate visibilities @xmath132 is defined in equation ( [ eq : moskernalt ] ) .
in particular , @xmath133 where @xmath134 the normalization factor given in equation ( [ eq : zk1 ] ) , and @xmath135 is the visibility weight given in equation ( [ eq : viswt ] ) . by operating with the gridding kernel on equation ( [ eq : visop ] ) , we get @xmath136 where we define @xmath137 as the convolution kernel that operates on the transform of the intensity ( the gridded version of @xmath77 ) , and @xmath138 is the gridded noise .
the conjugate to @xmath77 defined in equation ( [ eq : pk ] ) is given by @xmath139 which gathers the conjugate visibilities under the transformation @xmath140 .
although it is not necessary to do so , it is convenient to construct the @xmath126 on a regular lattice in @xmath129 with a spacing @xmath141
. thus the grid `` cells '' represented by the @xmath126 represent an interpolation using @xmath131 of the visibilities onto the _ uv_-plane .
this will be useful when using the estimators to form filtered images ( [ sec : image ] ) .
if it is desired that the visibilities be used directly , for example when the datasets are small , then the ungridded case can be recovered by setting @xmath142 and @xmath143 , giving @xmath144 and @xmath145 , with no loss of generality in the derivations .
to carry out the power spectrum estimation , we form the likelihood of the data given covariance matrices for the signal , noise , and foregrounds .
since the estimators are linear combinations of the visibilities , which we assume are made up of gaussian noise and gaussian signal components , we can use a multivariate gaussian probability distribution to describe the estimators also . because @xmath146 is complex
, it is easier to deal with the real and imaginary parts by packing them together in a double - length real vector @xmath147 written here as a column - vector of length @xmath148 .
the log - likelihood function for a real multivariate gaussian probability distribution is @xmath149 where @xmath150 is the transpose of @xmath151 , and @xmath152 is a block matrix of the real and imaginary covariances .
the vector @xmath153 represents the parameters of the model or theory against which the data is being measured ( see below ) .
these parameters are contained in @xmath154 . in terms of @xmath155 and @xmath156 , we can write @xmath157 and therefore @xmath158 \nonumber \\ \langle { \rm im}\,{\mbox{\boldmath $ \delta$}}\,{\rm im}\,{\mbox{\boldmath $ \delta$}}^{\rm t } \rangle & = & \phantom{-}{1\over2}\,{\rm re}\,\left [ \langle { \mbox{\boldmath $ \delta$}}\,{\mbox{\boldmath $ \delta$}}^\dag \rangle - \langle { \mbox{\boldmath
$ \delta$}}\,{\mbox{\boldmath $ \delta$}}^{\rm t } \rangle \right ] \nonumber \\ \langle { \rm re}\,{\mbox{\boldmath $ \delta$}}\,{\rm im}\,{\mbox{\boldmath $ \delta$}}^{\rm t } \rangle & = & -{1\over2}\,{\rm im}\,\left [ \langle { \bf \delta}\,{\bf \delta}^\dag \rangle - \langle { \mbox{\boldmath $ \delta$}}\,{\mbox{\boldmath $ \delta$}}^{\rm t } \rangle \right ] \nonumber \\ \langle { \rm im}\,{\mbox{\boldmath $ \delta$}}\,{\rm re}\,{\mbox{\boldmath $ \delta$}}^{\rm t } \rangle & = & \phantom{-}{1\over2}\,{\rm im}\,\left [ \langle { \mbox{\boldmath $ \delta$}}\,{\mbox{\boldmath $ \delta$}}^\dag \rangle + \langle { \mbox{\boldmath $ \delta$}}\,{\mbox{\boldmath $ \delta$}}^{\rm t } \rangle \right ] \label{eq : cimre}\end{aligned}\ ] ] where @xmath159 is the hermitian transpose of @xmath155 ( a row - vector containing the complex conjugate of a column - vector ) , and @xmath160 is the tensor or outer product of @xmath155 and @xmath161 , which is a matrix with elements @xmath162 .
it is important to include the covariances of @xmath163 as well as those for @xmath160 .
normally , only one of a given visibility @xmath49 or its conjugate @xmath164 will correlate with @xmath165 .
however , for short baselines @xmath166 ( less than 127.3 cm for the 90 cm cbi dishes ) , there may be overlap between the support for a given visibility and both another visibility and its conjugate , as shown in figure [ fig : support ] , and thus both may be nonzero . note that the correlation between distant conjugate pairs is small , since the overlap occurs far out in the antenna response @xmath68 , although it is non - negligible on the shortest cbi baselines where the overlap occurs at the @xmath167 point ( illustrated in figure [ fig : support ] ) for perpendicular 1-meter baselines with the beam response @xmath168 . outside the baseline range @xmath169 one of @xmath170 or @xmath171 will be zero
. the covariance matrix @xmath154 can be split into a sum of independent contributions from instrumental noise @xmath172 , the cmb signal @xmath173 , and foreground signals @xmath174 and @xmath175 .
we further split @xmath176 into a sum of terms @xmath177 from separate @xmath2 bands of the power spectrum , @xmath178 the factors @xmath179 are the `` bandpowers '' @xcite for bins with centers at @xmath180 , and are the model parameters to be determined by maximizing the likelihood .
the factor @xmath181 is amplitude of the covariance due to a residual gaussian foreground , and @xmath182 is the amplitude of the covariance contributed by known point sources ; there may more than one of each of these types of foreground covariance matrices .
the @xmath182 and @xmath181 can be treated as adjustable parameters to determined by maximum likelihood , or they can be held fixed at _ a priori _ values , in which case @xmath174 and @xmath183 are constraint matrices with their corresponding terms in equation ( [ eq : csnbandv ] ) behaving like additional noise terms . in the following sections we consider each of the terms @xmath172 , @xmath177 , @xmath174 , and @xmath183 in turn .
if we write @xmath184 then in each case we calculate the contributions to the covariance matrix for the real and imaginary parts of the estimators using equations ( [ eq : creim ] ) and ( [ eq : cimre ] ) @xmath185 & -{1\over2}\,{\rm im}\ , [ { \bf m } - \overline{\bf m } ] \\ \phantom{-}{1\over2}\,{\rm im}\ , [ { \bf m } + \overline{\bf m } ] & \phantom{-}{1\over2}\,{\rm re}\ , [ { \bf m } - \overline{\bf m } ] \end{array } \right)\ ] ] with the individual covariance matrices given by insertion of the appropriate contribution to @xmath186 and @xmath187 for that component , e.g. @xmath188 and @xmath189 to compute the block elements of @xmath177 .
the instrumental noise correlations are assumed to be gaussian and independent between different baselines , and frequency channels . for the cbi ,
tests have been carried out on the data which show this to be true to a high level of accuracy . in this case , the noise contribution to the real and imaginary parts of the visibilities are independent zero - mean normal deviates with @xmath190 and thus we can write @xmath191 for real noise matrix @xmath192 , where @xmath193 . it can be shown that the noise contributions @xmath138 to the estimators defined in equation ( [ eq : rop ] ) have the contributions to the covariance elements @xmath186 and @xmath187 defined in equation ( [ eq : mmbar ] ) given by @xmath194 using the covariances of @xmath195 given in equation ( [ eq : ndef ] ) .
this is assembled into the covariance matrix @xmath172 using equation ( [ eq : csbreimest ] ) . in general
, the gridding kernel @xmath131 will map a given visibility to more than one estimator , and thus @xmath172 will have non - zero off - diagonal elements . furthermore , if there are noise correlations between baselines or channels , then the structure of @xmath172 will be even more complicated .
the cmb contribution to the visibility covariance matrix is given by the covariance of the @xmath196 term in equation ( [ eq : rop ] ) @xmath197 where @xmath198 and @xmath199 are given in equations ( [ eq : cofu ] ) and ( [ eq : cofualt])respectively .
then , the elements of @xmath200 and @xmath201 for estimators @xmath202 and @xmath203 are @xmath204 with @xmath205 where to aid in breaking up the cmb covariance matrices into bands we write the integrations in terms of polar fourier coordinates @xmath206 ( @xmath207 ) . as an illustration ,
consider the case without gridding .
then , @xmath145 , and using equation ( [ eq : pk ] ) in ( [ eq : mkkint ] ) we get @xmath208 for the covariance matrix element between visibilities @xmath49 and @xmath165 .
we furthermore write the radial integral over @xmath209 as a sum with respect to @xmath210 of equation ( [ eq : cvflat ] ) @xmath211 where @xmath212 is the _ variance window function _ ( e.g. @xcite ) . we define the bandpowers @xmath179 by constructing @xmath210 piecewise with respect to a fiducial shape @xmath213 @xmath214 where @xmath215 breaks the power spectrum into non - overlapping bands .
the standard choice for the shape is @xmath216 for equal power per log-@xmath2 interval , with @xmath217 then giving the bandpowers in units of @xmath218 .
then , to calculate @xmath177 , we construct band versions of the covariance matrix elements in equation ( [ eq : mdef ] ) @xmath219 where @xmath220 and @xmath221 .
these are then combined following the prescription in equation ( [ eq : csbreimest ] ) to assemble the @xmath222 .
the variance window function @xmath223 is the convolution of the @xmath224 and @xmath225 , and thus its width is characteristic of the square of the fourier transforms primary beam , or fwhm @xmath226 .
thus we would expect in a single field to be able to achieve a limiting resolution of @xmath227 for @xmath118 at 31 ghz .
this will be increased by the mosaicing by a factor roughly equal to the extent of the half - power width of the mosaic relative to that of a single field . in practice ,
the limiting useful width for the @xmath2 bins for the bandpowers will be set by the band - band correlations introduced in the maximum likelihood estimation procedure ( see [ sec : relax ] and [ sec : test ] for further discussion and examples ) .
consider a set of @xmath228 point sources at positions @xmath229 with flux densities @xmath230 ( @xmath231 ) .
the intensity field at frequency @xmath84 is then given by @xmath232 which is assumed to be uncorrelated with other intensity components like the cmb .
the effect @xmath233 on the visibilities @xmath49 ( e.g. eq.[[eq : visi]])is then given by the sum over sources @xmath234 where @xmath235 is the contribution to visibility @xmath62 of source @xmath236 .
we assume that the positions of the sources can be determined with negligible uncertainty through radio surveys , and that the errors are due to uncertainties in the measurements of the flux densities .
then , the covariance between the source contributions to visibilities @xmath62 and @xmath237 is @xmath238 where @xmath239 is the flux density covariance matrix between sources @xmath236 and @xmath240 at frequencies @xmath241 and @xmath242 respectively .
there is a similar covariance matrix @xmath243 .
t these can be passed through the gridding procedure using equation ( [ eq : linop ] ) to make @xmath244 and used to construct the covariance elements @xmath245 using equation ( [ eq : mmbar ] ) .
this covariance matrix can be greatly simplified if we can subtract off the mean source flux densities , leaving a zero - mean residual error .
let the true source flux density @xmath230 be the sum of the measured flux density @xmath246 and an error @xmath247 .
if our measurements of these foreground sources are accurate , then the residuals @xmath247 should be uncorrelated between sources ( they are due to measurement errors ) and have zero mean . in this case , we can make corrected visibilities @xmath248 @xmath249 to be used in place of @xmath250 in subsequent analysis . then , we are left with the fluctuating component @xmath251 which we must deal with statistically .
the covariance between the source error contributions to the visibilities , assuming the flux density errors are independent between sources ( but not between frequency channels for the same source ) , is given by @xmath252 and similarly for @xmath253 .
finally , if the covariance is separable , e.g. @xmath254 then we can write @xmath255 the other covariance @xmath256 can be computed in the same way . because we have assumed that the covariance is separable , we can speed up the covariance calculation as only the vector @xmath257 for each source is needed .
we can grid this onto the estimators @xmath258 and then @xmath259 which are used to build @xmath174 .
there are two components to the source flux density uncertainties @xmath260 , one from the uncertainties on the source frequency spectrum , and the other from the uncertainties on the flux density measurements and any extrapolation of the measured flux densities to the observing frequencies @xmath241 ( using the estimated source spectrum ) . as an example , consider a source with a flux density @xmath261 measured with standard deviation @xmath262 at frequency @xmath46 , and a known power - law frequency spectrum with spectral index @xmath43 , @xmath263 then , it is easy to show that @xmath264 with the fractional uncertainty in the flux density @xmath265 remaining independent of the frequency . on the other hand , consider the case where there is now an uncertainty @xmath266 in the spectral index between @xmath241 and @xmath46 . then , our extrapolation factor @xmath267 , which we write as @xmath268 propagates to the extrapolated flux density as @xmath269 which can be negative for two channels flanking the fiducial frequency ( e.g. @xmath270 ) the errors will be anti - correlated .
note that we have approximated the resulting distribution as gaussian .
in general it is not , e.g. for a gaussian distribution in @xmath43 we find a log - normal distribution in @xmath271 .
although the dominant spectral error is due to the extrapolation from a frequency @xmath46 outside the range of the cmb instrument , there is an additional error due to an error in the spectral index over the frequency channels @xmath241 of the visibilities .
this is as if you extrapolated using one spectrum appropriate for the band center @xmath272 of the instrument , but when the flux densities @xmath273 are extrapolated from band center @xmath274 there is an error from using the wrong @xmath43 over the band .
this is handled using equation ( [ eq : sigmaa ] ) with another @xmath266 appropriate to the uncertainty in the spectral index over the @xmath241 .
for the cbi analysis , we have approximated both the flux density error and the spectral extrapolation error as a single equivalent flux density error . for the cbi ,
the frequency span ( 2636 ghz , or @xmath275 ) is small enough that we can approximate the spectral index uncertainty as an effective flux density uncertainty @xmath276 extrapolated to band center @xmath272 from @xmath46 using @xmath277 @xmath278 ^ 2\,\sigma^2_\alpha \big)\ ] ] where @xmath43 need not equal @xmath277 and should reflect the spectral index over the observing band , not the one used for extrapolation from @xmath46 . in principle , if the true mean flux densities for the sources are correctly subtracted from the visibilities , and the covariance matrix @xmath174 is built using the correct elements @xmath279 , then inclusion of this as a noise term in @xmath154 using @xmath280 would remove the effects of these sources from our power spectrum estimation in a statistical sense .
however , there are a number of factors that make this difficult .
if the source flux density measurements have a calibration error , then the errors will not be independent between sources .
also , the fainter sources ( which are still significant contributors to the signal ) have flux densities that are often extrapolated from much lower frequencies ( e.g. the `` nvss '' sources in paper ii and paper iii ) .
furthermore , since there are a relatively small number of discrete sources contributing to a given field , it is not clear we are in the statistical limit where the flux density covariance is an accurate description of what is happening to the data . for these reasons , for the cbi analysis
we treat the covariance matrix @xmath174 constructed using the approximation in equation ( [ eq : sigmeff ] ) as a constraint matrix for the nuisance parameters due to the sources , and set @xmath182 to a high enough amplitude to _ project out _ the contaminated modes in the data .
because the source modes are spread out by the effect of the synthesized beam ( the `` point - spread function '' in imaging terms ) , setting @xmath182 to too high a value will start to down - weight modes that are not significantly contaminated , while too low a value will eat into the noise and cmb signal power in those modes without down - weighting them sufficiently thus biasing the affected bandpowers low .
the exact values to be used thus depend upon the signal and noise levels in the data ; we refer the reader to papers ii and iii for descriptions of what was chosen for the cbi analysis .
see @xcite for a description of the constraint matrix formalism and the technique of projection . in [ sec :
pscmb ] it was mentioned that a single foreground component could be modeled with a modified covariance matrix , power spectrum shape and frequency dependence .
as long as these foregrounds can be treated as a gaussian random field , they can be processed in the same manner as the cmb .
therefore , once the amplitude and shape of the foreground fluctuation power spectrum @xmath281 is input , we compute the foreground covariance matrix elements @xmath282 where the variance window functions @xmath283 and @xmath284 are given by substituting for @xmath137 in equation ( [ eq : wijvarpi ] ) a new @xmath285 built from a kernel @xmath286 using a frequency factor @xmath287 appropriate to the foreground in question .
the matrix @xmath183 is then obtained by substitution of @xmath288 and @xmath289 as usual using equation ( [ eq : csbreimest ] ) .
although it is possible to break up the gaussian foreground component into bands as we did the cmb , it is preferable to compute the foreground covariance matrix in a single band using its shape @xmath290 , to reduce the degeneracy with the cmb you can not distinguish between the two in narrow @xmath2 bands where the shapes are unimportant .
an example of a foreground that strongly affects the cbi data is that from point sources below the limit for subtraction contaminating the cbi fields .
residual _ statistical background , in the limit where there are many sources per field , can be modeled as a white noise gaussian field with constant angular power spectrum and power - law frequency spectrum .
each individual source has a flux density drawn from a differential number count distribution @xmath291 which represents the number of sources per steradian with flux densities between @xmath28 and @xmath292 at observing frequency @xmath84 .
the angular clustering in these sources is very small , and can be neglected .
the contribution of a source @xmath236 to visibility @xmath49 was given by @xmath235 in equation ( [ eq : vsrc ] ) .
the sources are independently distributed in flux density and on the sky , so @xmath293 where the angular average can be written as an integral over @xmath294 with @xmath295 as the normalizing solid angle .
this integral is just a fourier transform , and so @xmath296 which is the same as the cmb visibility covariance matrix @xmath297 in equation ( [ eq : mkkintvis ] ) with @xmath298 and @xmath299 .
similarly , the other covariance @xmath300 reduces to @xmath301 .
thus , in the stochastic limit the residual source background behaves as a gaussian random field and can thus be treated as we do the cmb signal in [ sec : viscor ] but with a power spectrum shape @xmath302 which is constant over @xmath2 for a given pair of frequency channels .
the amplitude of the covariance matrix is the ensemble average of the source power per solid angle , which is obtained by integration over the flux density and spectral index distributions @xmath303 where we have again assumed that the spectrum is a power - law with spectral index @xmath43 over the range of interest for the @xmath241 , and integrate over the number counts over the flux density range from @xmath304 to @xmath305 .
we also assume that there is a large number of these faint sources over the solid angles of interest ( e.g. the cbi primary beam ) and thus the poisson contribution to the probability can be ignored and we can use the mean source density given by the number counts @xmath306 at the fiducial frequency @xmath46 for which @xmath307 is given . the spectral index distribution as a function of flux density @xmath308 must be estimated from radio surveys , though it can be uncertain at the high frequencies and faint levels at which the cmb experiments are carried out . if @xmath309 and thus is independent of flux density , then it can be shown ( e.g. appendix [ app : counts ] ) that the integrals in equation ( [ eq : resampint ] ) can be evaluated at a single frequency @xmath84 in the band and scaled using an effective spectral index @xmath310 @xmath311 where @xmath312 is the amplitude of the fluctuation power per solid angle ( in units of jy@xmath313/sr ) at @xmath84 . in terms of the logarithmic average @xmath26 for the cmb , @xmath314 which rises at high @xmath2 with respect to the cmb .
see appendix [ app : counts ] for an example analytic calculation using power - law source counts and a gaussian spectral index distribution .
the frequency range of the cbi is insufficient to distinguish nonthermal foreground emission from the thermal cmb , and thus this is treated as a constraint matrix ( i.e. @xmath181 is not solved for as a parameter ) .
therefore , in the cbi analysis we construct the covariance matrix @xmath315 using the matrix elements in equation ( [ eq : mkkfg ] ) built assuming unit power ( 1 jy@xmath313/sr ) and the frequency dependence @xmath316 from equation ( [ eq : creskk ] ) .
the value used for @xmath181 is equal to the source fluctuation power @xmath317 calculated as an _ a priori _ estimate based on knowledge of the residual foreground source populations ( see appendix [ app : counts ] ) .
we are not restricted to cmb , gaussian foreground , and discrete point sources as the components of our signal or noise in the covariance matrix @xmath154 in equation ( [ eq : csnbandv ] ) .
this approach can be generalized to deal with other signals of interest .
for example , extended sources with a known profile , such as the sunyaev - zeldovich effect from clusters of galaxies , could be modeled either analytically or numerically giving a power spectrum shape ( e.g. @xcite ) . in the case of a signal with a known distribution on the sky , a template can be used .
examples of this include dust emission in the millimeter - wave bands , or the anomalous centimeter - wave emission observed at ovro @xcite and by cobe @xcite .
in particular , the latter foreground , which is posited as due to spinning dust grains by @xcite , correlates with the 100@xmath19 m dust emission as measured by iras and dirbe , and thus a template of emission can be constructed .
unfortunately , with its low intrinsic fringe rates and extremely short ( @xmath318 ) spacings , the cbi is susceptible to ground pickup . to remove this , we observe for each field a trailing field displaced @xmath319 in right ascension @xmath319 later , and difference the corresponding visibilities
therefore , we must take this differencing into account in our correlation analysis .
this effectively modifies the window function , quenching low spatial frequencies further .
let us write @xmath320 for switching offset @xmath321 ( e.g. @xmath319 in ra @xmath322 for the cbi fields near the celestial equator ) .
then , from equation ( [ eq : visi ] ) we find @xmath323 + e^{\rm sw}_k\ ] ] where the switched noise @xmath324 . in terms of the kernel of equation ( [ eq : pk ] ) , @xmath325 \label{eq : pksw}\ ] ] and
thus for our switched visibilities we compute everything as before , but substituting @xmath326 for @xmath327 . note that if the trail field offsets @xmath328 were constant in arc length rather than in right ascension ( this is approximately true since the declination range of the mosaic is limited ) we could write the convolution kernel as @xmath329\ ] ] where the leading factor of 2 dominates ( you essentially get twice the cmb power ) .
note that a noticeable effect of the differencing is that the window function will have a ripple of `` wavelength '' @xmath330 superimposed on its envelope .
for example , the @xmath319 switching in ra that the cbi uses corresponds to @xmath331 at the celestial equator , and thus the ripple has @xmath332 in @xmath333 .
this corresponds to 180 in @xmath2 , but is azimuthally averaged in the _
uv_-plane and thus the peak - to - peak amplitude is reduced .
we have expressed the estimators as a real vector @xmath151 and obtained expressions for the components of its covariance matrix , and now turn to the problem of solving for the maximum likelihood estimators for the bandpowers using equation ( [ eq : likx ] ) .
as shown below , we will be carrying out a large number of matrix operations using @xmath154 and its component matrices ( e.g. @xmath172 , @xmath177 , etc . ) and thus these will need factorization . because @xmath154 is positive definite we use optimized cholesky decomposition routines ( dchdc from linpack , or dpotrf from lapack ) to carry out the required factorizations . the large number of visibilities _ times _ the number of mosaic pointings makes this computation extremely costly ( the matrix inversions and/or solution of systems of equations are order @xmath125 processes ! ) , especially for a large number of bands @xmath334 .
clever perturbative or gradient search methods can help to reduce the overhead in finding the maximum in parameter space .
one such method is the quadratic relaxation technique of ( * ? ? ?
* bjk ) . to summarize here ,
if one taylor expands the log - likelihood around the maximum likelihood bandpowers @xmath335 to second order @xmath336 then we can move toward the maximum using the quadratic approximation @xmath337^{-1 } \ , { \partial \,\ln l({\bf q } ) \over \partial q_{b ' } } .\ ] ] the first derivative ( gradient ) is given by @xmath338\ ] ] and the second derivative ( curvature matrix ) by @xmath339 \nonumber \\ & & { } + { 1\over2}\,{\rm tr}\bigg [ { \bf c}^{-1 } { \partial c \over \partial q_{b } } { \bf c}^{-1 } { \partial c \over \partial q_{b ' } } \bigg ] .
\label{eq : curvf}\end{aligned}\ ] ] note that the partial derivatives of the covariance matrix are just the band signal covariance matrices @xmath340 defined above .
the final approximation is to replace the curvature matrix with its expectation value , which is the fisher information matrix @xmath341 .
\label{eq : fisher}\ ] ] this yields @xmath342_{b b ' } \ , { \rm tr}\left [ ( { \mbox{\boldmath $ d$}}\,{\mbox{\boldmath $ d$}}^{\rm t } - { \bf c } ) ( { \bf c}^{-1}\,{\bf c}^{\rm s}_{b'}\,{\bf c}^{-1 } ) \right ] \label{eq : delqb}\ ] ] for the iterative correction to the bandpowers
. this amounts to making a quadratic approximation to the shape of the likelihood function around the maximum , and iteratively approaching it . at each step
, the total covariance matrix @xmath154 must be updated using the new bandpowers @xmath343 . a convergence criterion based on the magnitude of the corrections
@xmath344 will allow approach to the true @xmath345 to be controlled .
the inverse of the fisher matrix @xmath346_{b b'}$ ] evaluated at maximum likelihood is the covariance matrix of the parameters @xcite .
the diagonals @xmath346_{bb}$ ] give an estimated gaussian error bar for the derived bandpowers @xmath345 , though the full fisher matrix must be used to take the ( usually significant ) band - band correlations into account . as the width of the @xmath2 bins for the bands @xmath347 is reduced , anti - correlation between adjacent bands increases due to the intrinsic @xmath2-space resolution of the data .
the presence of known or residual point source foregrounds in equation ( [ eq : csnbandv ] ) is dealt with by either fixing the amplitudes @xmath182 or @xmath181 and treating @xmath348 or @xmath349 as additions to the noise matrix @xmath172 , or by solving for the @xmath182 or @xmath181 and treating them as extra bandpowers @xmath217 with associated entries in the fisher matrix . in practice , for the cbi , it is necessary to hold fixed the @xmath181 because the contribution from the source foreground with a white noise power spectrum and appropriate frequency spectrum is largely indistinguishable from an overall offset of the cmb power spectrum .
in addition , the uncertainties on the individual known source contributions to an aggregate @xmath174 will be substantial , and thus solving for a single amplitude @xmath182 will not be as useful as it might appear . in this case , the @xmath174 acts as a constraint matrix and the @xmath182 can be set to an arbitrarily high value , which will effectively _ project _ out the modes corresponding to the known sources by down - weighting the relevant combinations of the estimators in the likelihood @xcite .
consider observations taken of separate sets of single fields or mosaics @xmath350 where there is effectively no correlation between fields from separate @xmath350 and the fields within a given set @xmath350 are related by the mosaic covariance given in the previous sections . in this case
, we can assemble a giant data vector @xmath351 from the @xmath352 individual field vectors @xmath353 ( e.g. eq.[[eq : xvec ] ] ) , with the block diagonal covariance matrix @xmath354 which in turn can be written as sums of block - diagonal noise and signal covariance matrices @xmath355 and @xmath356 etc .
, with blocks given by @xmath355 and @xmath356 , etc . because they are block diagonal , we can write the log - likelihood in equation ( [ eq : likx ] ) as the sum over datasets @xmath357 we proceed as before , with the same bandpowers @xmath358 and with the block - diagonal band covariance matrix @xmath359 and thus all matrices are block - diagonal and composed of the individual single field or mosaic matrices . therefore , @xmath360 \nonumber \\
\delta q_b & = & { 1\over2}\,\sum_{b'}\left [ f^{-1 } \right]_{b b ' } \ , \sum_f { \rm tr}\left [ ( { \mbox{\boldmath $ d$}}_f\,{\mbox{\boldmath $ d$}}_f^{\rm t } - { \bf c}_f ) ( { \bf c}_f^{-1}\,{\bf c}^{\rm s}_{b'f}\,{\bf c}_f^{-1 } ) \right]\end{aligned}\ ] ] which is used to iteratively approach the @xmath345 using the bjk scheme as in the single dataset case . to compare the bandpowers obtained from the data to model power spectra we need to define a set of filter functions which project models @xmath361 into a set of bandpowers @xmath362 @xmath363 as in @xcite . in the ensemble limit
, the expectation value @xmath364 will approach the underlying signal covariance matrix @xmath176 .
we can then use the expression for the minimum variance estimate of the bandpower to derive the filter functions @xmath365 @xcite . since @xmath366_{b b ' } \ , { \rm tr}\left [ ( { \bf c}^{-1}\,{\bf
c}^{\rm s}_{b'}\,{\bf c}^{-1 } ) \ , { \bf c}^{\rm s } \right]\ ] ] and @xmath367 the normalized filter functions can be computed using the band averaged fisher matrix ( e.g. eq.[[eq : fisher ] ] ) @xmath368_{b b ' } \ , { \rm tr}\left [ ( { \bf c}^{-1}\,{\bf c}^{\rm s}_{b'}\,{\bf c}^{-1})\ , \frac{\partial { \bf c}^{\rm s}}{\partial \cal{c}_{\ell}}\right]\ ] ] with respect to the @xmath369 that is built into the @xmath176 . because of the @xmath370 used in the construction of the @xmath177 in equation ( [ eq : mkkbin ] ) , @xmath371 and thus @xmath372 is orthonormal with respect to the bands defined by @xmath370 .
calculating the filter functions at each @xmath2 becomes somewhat prohibitive in both processor time ( the problem scales as an extra @xmath373 operations ) and storage since the calculation of equation ( [ eq : windows ] ) can only proceed once we have relaxed to the maximum likelihood solution .
for this reason , in practice we sample the full filter functions in bands at intervals @xmath374 where the @xmath374 are narrower than the bands @xmath347 , with @xmath375_{b b ' } \ , { \rm tr}\left [ ( { \bf c}^{-1}\,{\bf
c}^{\rm s}_{b'}\,{\bf c}^{-1 } ) \ , { \bf c}^{\rm s}_{b_f}\right].\ ] ] in principle , this is equivalent to assuming a flat window over the ` fine ' band @xmath374 and as long as the curvature of the exact window function is small over the intervals @xmath374 this should provide an adequate sampling of the continuous limit . to obtain model bandpowers
we can then either interpolate the samples @xmath376 to obtain an approximate form for @xmath365 for use in equation ( [ eq : bpaverage ] ) or _ pre - average _ the model spectrum over the fine bands @xmath374 as @xmath377 where @xmath378 are bandpowers calculated using flat filters ( @xmath379 ) .
we find that a fine band width @xmath380 is sufficient to adequately sample the window functions and ensure normality and orthogonality to within @xmath381 with respect to integration over the bands ( e.g. eq.[[eq : winortho ] ] ) .
example window functions calculated in this manner for mock deep fields and mosaics are shown in the lower panels of figure [ fig : mockdeep ] and figure [ fig : mockmos ] respectively .
a further complication at the parameter end of the process is that the likelihood of the bandpowers can not be assumed to be a gaussian .
this is especially so in cases where the error in the bandpowers is sample or cosmic variance limited .
assuming the bandpowers to be gaussian distributed can lead to the well known problem of _ cosmic bias _ where the likelihood of low power models can be overestimated and conversely that of high power models can be underestimated . @xcite
have shown how one can avoid this problem while still retaining gaussianity in the @xmath382 analysis by treating certain functions of the bandpowers as gaussian distributed .
very good fits to the non - gaussian distribution of the bandpowers can be obtained by use of the _ offset log - normal _ and _ equal variance _
approximations to the likelihood .
both approximations use offsets @xmath383 in the bandpowers which describe the contributions to the error in the bandpowers due to components other than the cmb . for
the range of scales probed by instruments such as cbi these components will include the foregrounds such as point sources in addition to the usual noise ` on the sky ' offset @xmath384 , where @xmath385 is the offset due to the noise contribution to the error such that the quantity @xmath386 has a normal distribution @xcite .
for accurate parameter fits we therefore require estimates of bandpowers for all the components making up the total covariance @xmath154 .
an approximation for these can be obtained by modifying the minimum variance estimator for the bandpowers @xmath217 at the maximum likelihood @xmath387^{-1 } \ , { \rm tr}\left [ ( { \bf c}^{-1}\,{\bf c}^{\rm s}_{b'}\,{\bf c}^{-1 } ) \ , { \bf c}^{\rm x } \right]\ ] ] where we have substituted @xmath388 in equation ( [ eq : delqb ] ) for the observed measure for the signal covariance @xmath389 .
we then set @xmath388 to the noise @xmath172 , foreground source @xmath390 or gaussian residual foreground @xmath391 covariance components as desired ( or use the maximum likelihood values @xmath392 and @xmath393 if these are included as parameters in the solution rather than being fixed ) .
examples of these are shown in paper ii and paper iii for the deep field data and mosaic data respectively .
the offset to the log - normal is then obtained by summing the @xmath394 over the components , @xmath395 ( e.g. @xcite ) .
this formalism is used in paper v to approximate the shapes of the likelihood functions in order to derive limits on the cosmological parameters .
although not the primary goal of this method , an image can be constructed by fourier transforming back to the sky plane using equation ( [ eq : fourier ] ) .
if the estimators @xmath126 are constructed on a regular lattice in @xmath396 with spacing @xmath141 and _ uv _ extent @xmath397 , then the resulting image will have an extent on the sky given by the inverse of the spacing @xmath398 and a resolution given by @xmath399 . in the continuum limit ( see appendix [ app : estim ] ) , we can define an estimator @xmath400 for the temperature field @xmath401 @xmath402 where @xmath121 is the continuous functional form ( e.g. eq[[eq : estcont ] ] ) for the estimators , with @xmath403 . in practice
the lattice of estimators @xmath126 is embedded in a wider grid padded with zero in the unsampled cells , and an fft is carried out . for our standard gridding normalization @xmath404 given in equation ( [ eq : zk1 ] ) , the units of @xmath155 will be flux density units ( jy ) and thus its inverse fourier transform will produce a map in units of jy / beam , where the beam area is given by the point - spread - function ( psf ) of the image . for a single field ,
the psf is just the image generated using equation ( [ eq : image ] ) using estimators @xmath405 computed by introducing unit `` visibilities '' into equation ( [ eq : linop ] ) @xmath406 the situation for the mosaics is somewhat more complicated , as the mosaic offsets must be taken into account in constructing equivalent visibilities for point sources @xmath407 obtained by evaluating equation ( [ eq : visi ] ) with no noise and @xmath408 . in this case
one would evaluate the psf at various positions @xmath409 in the map .
because our estimators use the kernel @xmath131 as given in equation ( [ eq : qik ] ) which includes the beam transform @xmath68 , we are effectively multiplying the image on the sky by the primary beam squared once in the kernel , and once due to the instrument itself ( e.g. eq.[[eq : pk ] ] ) .
images made directly from the @xmath155 will therefore be strongly attenuated in the ( noisy ) outskirts .
as mentioned in appendix [ app : estim ] , the optimal weighting for the imaging of the cmb component is to use the planck factor in equation ( [ eq : fplanck ] ) to correct for the thermal frequency spectrum ( e.g. eq.[[eq : qtherm ] ] ) , while our standard intensity weighting given in equation ( [ eq : viswt ] ) is optimized for a flat non - thermal power - law spectrum with spectral index @xmath410 .
we can also filter the gridded estimators in such a way as to enhance or down - weight certain signals or noise .
we can do this with optimal or wiener filtering ( e.g. @xcite ) , @xmath411 where the choice of the filter @xmath412 depends on the application . for example , the covariances @xmath388 calculated from equation ( [ eq : csbreimest ] ) can be used to construct optimal filters for each component contributing to the observations . for the signal component described by the covariances
@xmath388 we can construct a wiener filter to be applied to the gridded @xmath0 estimators @xmath413 the amplitudes for the signal models such as the bandpowers @xmath217 or the source amplitudes @xmath182 can be set to their maximum likelihood values or to fiducial model amplitudes .
the wiener filtered image is then recovered by fourier transforming .
examples of images created using this method are described in the next section , and shown in figure [ fig : image ] .
wiener - filtered images are also used in paper vi to explore the possibility of detection of the sunyaev - zeldovich effect in the data at high @xmath2 .
the algorithm described above was coded as a scalar fortran ( f77 compatible ) program designated cbigridr , with a parallelized fortran 90 version using openmp directives also available for use on multiprocessor machines .
in addition , the bjk likelihood relaxation was coded in a second parallelized fortran 90 program called mlikely using parallel versions of lapack matrix algebra routines ( e.g. [ sec : relax ] ) .
together , these two programs make up the cbi analysis pipeline .
this pipeline has undergone numerous tests and development since its inception in april 2001 , and has been used to produce the power spectra and to provide the bandpowers as input to the cosmological parameter analysis given in the companion papers .
we now give a brief description of our implementation . in order to carry out the numerical integrations , a fine - grain rectangular lattice in _
uv_-space was used .
the fine - grain grid size @xmath414 ( in units of wavelength ) was chosen to adequately sample the phase turns in the _ uv _ plane due to the mosaic size and differencing ; for a standard @xmath120 cbi mosaic with @xmath415 spacing , the maximum field separation along the grid direction is @xmath416 , which gives oscillations in the _ uv _ plane with wavelength of @xmath417 , giving @xmath418 for two samples per cycle . to evaluate the projection operators @xmath419 ( e.g. eq.[[eq : rop ] ] ) , we store a small fine - grain lattice around each @xmath129 .
the maximum radius in _ uv _ space needed for the support of @xmath137 is @xmath420 . for cbi @xmath421 cm , and @xmath422 cm at @xmath423 ghz
, we get @xmath424 . a grid of size @xmath425 cells with @xmath426 will fit both the sampling and radius requirements .
the estimators are evaluated on a coarse - grain lattice of @xmath129 , with a spacing of @xmath427 .
the fine - grain lattices on which we accumulate the @xmath419 will be cross - correlated to form the covariance elements @xmath428 and @xmath429 ( e.g. eq.[[eq : mkkint ] ] ) , it is desirable to have the coarse grid size locked to integer multiples of the fine - grid cells .
this coarse grid does not have to sample the highest mosaic frequencies , but only the effective width of @xmath137 .
we find that for single cbi fields , @xmath430 is adequate . for cbi mosaics , a hybrid lattice with @xmath431 in the inner part ( @xmath432 ) and @xmath433 in the outer part
was found to work well . as stated in
[ sec : pscmb ] , the choice of sign of the exponential of the fourier transform in equation ( [ eq : fourier ] ) is a convention .
this choice varies throughout the literature on the subject , but in practice depends upon the way the baseline vectors are defined in the data and how the correlation products are made ( e.g. which antenna gets the quadrature phase shift ) .
we note that in coding our algorithm to conform to the imaging standards of the aips , and difmap @xcite packages using the cbi data , we had to use the opposite sign convention from the one presented in equation ( [ eq : fourier ] ) . to process a dataset , a spectral weighting @xmath434 and shape function @xmath435 are chosen .
the visibilities @xmath49 are looped over , and any source subtraction ( [ sec : src ] ) is applied . for each estimator
@xmath202 that @xmath49 contributes to either directly or as a conjugate , its contribution to the fine - grain lattice @xmath436 for @xmath437 is accumulated , e.g. @xmath438 where @xmath439 is a @xmath425 unit ( fine - grain ) lattice .
this means that for @xmath440 estimators , the storage required for @xmath137 is only @xmath441 double - precision complex numbers .
if the data were differenced ( as for cbi data ) , then @xmath442 from equation ( [ eq : pksw ] ) is used .
the contributions to the noise covariance elements @xmath443 and @xmath444 ( eq.[[eq : eeexmat ] ] ) , and the @xmath445 ( eq.[[eq : dsrcrms ] ] ) are also accumulated at this time .
finally , this visibility s contributions are added to estimator @xmath126 and normalization @xmath134 .
the storage for @xmath137 , @xmath443 , and @xmath444 dominate the memory requirements . for example , the cbi mosaics use around @xmath446 estimators , and thus storage for @xmath447 double - precision complex numbers is needed .
a single packed @xmath448 array is needed to hold @xmath443 and @xmath444 .
the @xmath388 matrices are calculated in place and written out row by row , and thus need not be stored .
there are no instances where matrices of dimension @xmath449 are stored ; the storage for a matrix of this size would be prohibitive as our largest cbi mosaics have @xmath450 .
when all the visibilities have been processed , the estimators are normalized by @xmath134 , split into real and imaginary parts , and written out to disk .
the covariance matrices @xmath451 , @xmath452 , and any @xmath453 are constructed by looping over pairs of rows corresponding to the real and imaginary parts of each estimator , e.g. rows @xmath202 and @xmath454 for estimator @xmath202 .
for each @xmath455 , the stored @xmath456 and @xmath457 are cross - correlated along with the shape function @xmath458 to form the bandpower covariance elements @xmath459 and @xmath460 of equation ( [ eq : mkkint ] ) , combined to make @xmath452 using equation ( [ eq : csbreimest ] ) , and stored .
the relevant columns of these rows of @xmath451 are formed from the stored @xmath461 and @xmath462 . at this point , for each @xmath174 desired ( there may be more than one , in the cbi analysis we use 3 ) , the relevant @xmath463 are combined using equation ( [ eq : mkksrc ] ) to form @xmath464 and @xmath462 , which in turn are used to make @xmath453 .
after all columns for these rows of the covariance matrices are computed , they are written to disk , and this process is repeated for the pair of rows corresponding to the next estimator @xmath202 . when all rows are complete , the output file is complete .
note that different binnings of the @xmath222 can be run without regridding using the original @xmath137 , saving significant time .
if a residual foreground covariance matrix @xmath183 is desired , the procedure outlined above is repeated in its entirety using the description in [ sec : foreground ] .
the spectrum and shape appropriate for the source population or foreground emission is used during the gridding and covariance matrix construction .
other than these factors , the same gridding as in the cmb and noise estimators must be used .
the output files from cbigridr are then used as input to mlikely .
these can be for single fields or mosaics , or for combinations of independent fields or mosaics ( [ sec : comblike ] ) . at this point ,
the pre - factors @xmath182 and @xmath181 for any @xmath465 and @xmath183 covariance matrices are chosen and fixed .
relaxation to the likelihood maximum is carried out as described in [ sec : relax ] , and the resulting bandpowers @xmath358 and inverse fisher matrix elements @xmath346_{b b'}$ ] are written out . if desired , the bandpower window functions @xmath376 ( [ sec : bwin ] ) can be computed if cbigridr was run to produce narrow - bin @xmath466 .
the component bandpowers @xmath467 , @xmath468 and @xmath469 ( [ sec : comp ] ) can also be computed at this time .
finally , filtered images using the formalism of [ sec : image ] can be computed from the estimators , the @xmath154 ( at maximum likelihood ) , and the component covariance matrices .
results from this are shown below and in paper vi .
the timing for cbigridr depends upon the degree of parallelization as well as the processor speed on a given machine , and the number of visibilities gridded , number of foreground sources , and number of bins @xmath347 for the bandpowers . as an example , the processing of the 14-h mosaic field of paper iii ( the largest of the datasets ) involved gridding 228819 visibilities from 65 separate nights of data in 41 fields to 2352 complex estimators .
a total of 916 sources were gridded into three source covariance matrices .
a total of 7 different binnings for @xmath222 were run at this time from the same gridding .
the execution time using the parallel version of cbigridr was @xmath470 running on 22 processors on a 32-processor alpha gs320 workstation at the canadian institute for theoretical astrophysics .
it then took @xmath471 on the same computer for mlikely to process 4704 double - precision real estimators in 16 @xmath222 bands , with 3 @xmath174 matrices , one @xmath183 and one @xmath172 .
this included the time needed to calculate the component bandpowers @xmath472 , but not the window functions .
the speed of this fast gridded method has allowed us to carry out numerous tests on both real and simulated dataset , which would not have been possible carrying out maximum likelihood ( e.g. using even the optimized mlikely ) on the 200000-plus visibilities .
the performance of the method was assessed by applying it to mock cbi datasets .
simulated cbi datasets were obtained by replacing the actual visibilities from the data files containing real cbi observations of the various fields used in paper ii and paper iii with the response expected for a realization of the cmb sky drawn from a representative power spectrum , plus uncorrelated gaussian instrumental noise with the same variance as given by the scatter in the actual cbi visibilities .
the differencing of the lead and trail fields used in cbi observations was included ( e.g. [ sec : difference ] ) .
this mock dataset had the same _ uv _ distribution as the real data , and gives an accurate demonstration of expected sensitivity levels and the effect of cosmic variance .
the power spectrum chosen for these simulations was for a model that fit the cobe and boomerang data @xcite .
figure [ fig : mockdeep ] shows the power spectrum estimation derived following the procedure detailed above .
the mock datasets were drawn as realizations for the 08h cbi deep field from paper ii .
the binning of the signal covariance matrix @xmath222 was chosen to be uniform in @xmath2 with bin width @xmath473 . because a single realization of the sky drawn from the model power spectrum will have individual mode powers that deviate from mean given by the power spectrum due to this intrinsic so - called `` cosmic variance '' plus the effect of the thermal instrumental noise , we analyze 387 realizations each taken from a different realization of the sky and a different set of instrumental noise deviates .
the mean @xmath217 for each band @xmath347 converge to @xmath474 , which is obtained by integrating the model @xmath475 over the window functions @xmath365 ( e.g. eq.[[eq : cbpred ] ] ) , within the sample uncertainty for the realizations .
furthermore , the standard deviation of the @xmath217 from the mean for each band agrees with the value obtained from the diagonals of the inverse of the fisher matrix .
the choice of the @xmath2 bin size is driven by the trade off between the desired narrow bands for localizing features in the power spectrum and the correlations between bins introduced by the transform of the primary beam .
there is an anti - correlation between adjacent bands seen in @xmath476_{bb'}$ ] at the level of @xmath477% to @xmath478% for @xmath473 with a single field .
we have found that correlations up to about @xmath479% give plots of the @xmath217 that are more visually appealing than those made with narrower band and higher correlation levels due to the increasing scatter in the bandpowers about the mean values .
bins of this size do not achieve the best possible @xmath2-resolution , and thus our cosmological parameter runs use finer - binned bands since the correlations are taken into account in the analyses . the band window functions @xmath480 are shown in the lower panel of figure [ fig : mockdeep ] , and were computed using narrow binnings @xmath376 ( e.g. eq.[[eq : wbbf ] ] ) with @xmath481 .
the small - scale structures seen in the window functions , particularly visible around the peaks , are due to the differencing which introduces oscillations ( see [ sec : difference ] ) . as shown in equation ( [ eq : winortho ] ) , a window function @xmath480 is normalized to sum to unity within the given band @xmath347 , and to sum to zero in the other bands , and thus there must be compensatory positive and negative `` sidelobes '' of the window function outside the band .
figure [ fig : mockmos ] shows the power spectrum derived for a simulated mosaic of @xmath120 fields separated by @xmath415 using the actual cbi 20h mosaics fields from paper iii as a template .
this mosaic field was chosen as it had incomplete mosaic coverage , and thus would be the most difficult test for the method .
the binning for @xmath222 shown used @xmath482 , which gave adjacent band anti - correlations of @xmath483 to @xmath484 in the @xmath485 .
again the mean of the 117 realizations converges to the value expected within the error bars , showing that there is no bias introduced by the method , even in the presence of substantial holes in the mosaic ( see paper iii for the mosaic weight map ) .
furthermore , the rms scatter in the realizations converges to the mean of the inverse fisher error bars , as in the single - field case . as in the previous figures ,
the bandpower window functions are shown in the lower panels . in figure
[ fig : mockmosruns ] are shown three randomly chosen realizations from the ensemble , plotted along with the input power spectrum .
this shows the level of field - to - field variations that we might expect to see in cbi data .
there are noticeable deviations from the expected bandpowers in individual realizations , particularly at low @xmath2 where cosmic variance and the highly - correlated bins conspire to increase the scatter .
these differences are within the expected scatter when bin - bin correlations and limited sample size is taken into account , but care must be exercised in interpreting single field power spectra . in particular ,
the acoustic peak structures are obscured by the sample variations .
however , the average bandpowers for the 3 runs ( shown in figure [ fig : mockmosruns ] as open black circles ) are better representations of the underlying power spectrum .
although this is not a proper `` joint '' maximum likelihood solution ( e.g. [ sec : comblike ] ) as is done for the real cbi mosaic fields , the improvement seen using the 3-field average leads us to expect that the combination of even 3 mosaic fields damps the single field variations sufficiently to begin to see the oscillatory features in the cmb power spectrum .
while we do not show the equivalent plots of the deep fields from figure [ fig : mockdeep ] , the same behavior is seen ( with even larger field - to - field fluctuations in the relatively unconstrained first bin , though still consistent with the error bars ) . the effect of adding point sources to the mock fields , and then attempting power spectrum extraction , is shown in figure [ fig : mockdeepsrc ] .
a set of 200 realizations were made in the same manner as in the runs in figure [ fig : mockdeep ] , but the list of point source positions , flux densities and uncertainties , and spectral indices from lower frequency used in the analysis in paper ii ( the `` nvss '' sources ) was used to add mock sources to the data .
the flux density of the sources actually added to the data were perturbed using the stated uncertainties as 1-@xmath486 standard deviations .
the errors used were 33% of the flux density except for a few of the brighter sources which were put in with 100% uncertainties .
we then used the methodology described in [ sec : src ] to compute the constraint matrices .
the first method of correction used was to subtract the ( unperturbed ) flux densities from the visibilities , and build the @xmath174 from @xmath487 built using the uncertainties ( shown as the red triangles ) .
in addition , we also did no subtraction , but built @xmath174 from @xmath487 using the full ( unperturbed ) flux densities ( shown as the blue squares ) .
this is equivalent to assuming a 100% error on the source flux densities , and thus canceling the average source power in those modes . in both cases
the factor @xmath280 was used .
the simulations show that both methods are effective , with no discernible bias in the reconstructed cmb bandpowers .
finally , the production of images using the gridded estimators described in [ sec : image ] is demonstrated in figure [ fig : image ] .
the series of plots show the effect of wiener filtering using the noise and various signal components on an image derived from one of the mock 08h cbi deep field realizations with sources from the ensemble shown in figure [ fig : mockdeepsrc ] .
the planck factor weighting of equation ( [ eq : qtherm ] ) was used during gridding to optimize for the thermal cmb spectrum , though in practice this makes little difference due to the restricted frequency range of the cbi .
the estimators for this realization were computed by subtracting the mean values of the source flux densities and putting the standard deviations into @xmath174 with @xmath280 ( the red triangles in figure [ fig : mockdeepsrc ] ) .
the filtering down - weights the high spatial frequency noise seen in the unfiltered image , and effectively separates the cmb and source components as shown by comparing panels ( c ) and ( d ) to the total signal in panel ( b ) .
the signal in this realization is dominated by the residuals from two bright point sources that had 100% uncertainties put in for their flux densities and thus escaped subtraction .
the effectiveness of @xmath174 in picking out the sources in the image plane underlines its utility as a constraint matrix in the power spectrum estimation .
we have outlined a maximum likelihood approach to determining the power spectrum of fluctuations from interferometric cmb data .
this fast gridded method is able to handle the large amounts of data produced in large mosaics such as those observed by the cbi .
software encoding this algorithm was written , and tested using mock cbi data drawn from a realistic power spectrum .
the results of the code were shown to converge as expected to the input power spectrum with no discernible bias . for small datasets , this code was also tested against independently written software that worked directly on the visibilities .
in addition , the pipeline was run with gridding turned off as described in [ sec : liklhd ] , again for small test data sets .
no bias or significant loss in sensitivity was seen in these comparisons .
this software pipeline was used to analyze the actual cbi data , producing the power spectra presented for the deep fields and mosaics in paper ii and paper iii respectively .
the output of the pipeline also was used as the input for the cosmological parameter analysis in paper v and the investigation of the sunyaev - zeldovich effect in paper vi .
this method is of interest for carrying out power spectrum estimation for interferometer experiments that produce a large number of visibilities but with a significantly smaller number of independent samples of the fourier plane ( such as close - packed arrays such as vsa or dasi ) .
the cbi pipeline analysis is carried out in two parts , the gridding and covariance matrix construction from input uv - fits files in cbigridr and the maximum likelihood estimation of bandpowers using quadratic relaxation in mlikely .
the software for the pipeline is available by contacting the authors .
we close by noting that our formalism can be extended to deal with polarization data . in the case of cmb polarization ,
there are as many as six different signal covariance matrices of interest in each band , with estimators ( or visibilities ) for parallel - hand and cross - hand polarization products , and thus development of a fast method such as this is critical . in september 2001 polarization capable versions of cbigridr and mlikely
were written and tested .
we describe the method , the polarization pipeline , and results in the upcoming paper ( myers et al .
2002 , in preparation ) .
stm was supported during the early years of the cbi by a alfred p.
sloan fellowship from 1996 to 1999 while at the university of pennsylvania .
genesis of this method by stm greatly benefited by a stay in july 2000 at the itp in santa barbara , supported in part by the national science foundation under grant phy99 - 07949 .
the national radio astronomy observatory is a facility of the national science foundation operated under cooperative agreement by associated universities , inc .
the cbi was funded under nsf grants ast-9413934 , ast-9802989 and ast-0098734 , with contributions by maxine and ronald linde , and cecil and sally drinkward , and the strong support of the california institute of technology , without which this project would not have been possible .
in addition , this project has benefited greatly from the computing facilities available at cita , and from discussions with other members of the group at cita not represented as authors on this paper .
suppose we were to construct a simple linear `` dirty '' mosaic on the sky obtained by a linear combination of the dirty ( not deconvolved ) images of the individual fields ( e.g. @xcite ) .
in the _ uv _ plane , this reduces to summing ( integrating ) up the visibilities from each mosaic `` tile '' with some weighting function , e.g. @xmath488 where for the time being we ignore the contribution from the complex conjugates of the visibilities ( see below ) . for illustrative purposes ,
let us consider only a single frequency channel and write the estimator as a function @xmath121 , where @xmath403 , which in the absence of instrumental noise is given by @xmath489 with kernel @xmath490 , sky and aperture plane sampling given by @xmath491 , and where @xmath492 is the visibility at pointing position @xmath11 and _ uv _ locus @xmath493 from equation ( [ eq : visi ] ) .
in practice , the sampling function is just a series of delta functions @xmath494 over the measured visibilities @xmath495 each with weight @xmath496 . as an ansatz , we let the mosaicing kernel have the form @xmath497 where @xmath498 is the interpolating kernel . furthermore , let us assume that the _ uv_-plane coverage is the same for all mosaic pointings , and thus @xmath499 is separable @xmath500 where @xmath501 and @xmath502 are the sampling and weighting in the two domains . combining these and rearranging terms , we get @xmath503 where in equation ( [ eq : mos2 ] ) we used the fact that the final right - hand side integral in equation ( [ eq : mos2a ] ) is the fourier transform @xmath504 of the mosaic function @xmath505 . for an infinite continuous mosaic , @xmath506 and thus @xmath507
if we wish to recover @xmath508 in this limit , then @xmath509 with normalization @xmath510 will fulfill our requirements .
we have chosen @xmath511 as the _ uv _ kernel as it reproduces the least - squares estimate of the sky brightness in the linear mosaic @xcite .
then , equation ( [ eq : mos2 ] ) becomes @xmath512 which has a width controlled by the narrower of the width of @xmath513 or the width of @xmath504 .
thus , by widening the mosaic @xmath514 to a larger area than the beam @xmath111 , we will fill in the desired information inside the @xmath68 smeared patches in the _ uv_-plane .
thus , a properly sampled @xmath515 mosaic will fill in a @xmath515 sub - grid within each _ uv _ cell you would have normally had for a single pointing , and thus an @xmath515 mosaic consisting of @xmath124 `` images '' each is equivalent to a _
uv _ super - grid of size @xmath516 ( e.g. @xcite ) .
note that for a non - continuous mosaic , there will be `` aliases '' in the _ uv _ plane separated by the inverse of the mosaic spacing in the sky @xcite . ideally , we would like the separation between aliased copies to be larger than the extent of the beam transform , which is satisfied for @xmath517 which for @xmath421 cm corresponds to @xmath518 at @xmath519 ghz and only @xmath520 at @xmath423 ghz , the centers of the extremal cbi bands .
the spacing used in the cbi mosaics is a compromise between the aliasing limits over the bands and the desire to have a fewer number of pointings on a convenient grid .
we chose to observe with pointing centers separated by @xmath415 , which is sub - optimal above @xmath521 ghz .
however , the effect of aliasing is small , with the overlap point @xmath522 occurring at the @xmath523 point of @xmath68 at 31 ghz , and the @xmath524 point for the highest frequency channel at @xmath423 ghz .
we obtain the gridding kernel @xmath525 of equation ( [ eq : linest ] ) corresponding to equation ( [ eq : qvu ] ) by using the discrete sampling in equation ( [ eq : fsamp ] ) @xmath526 with visibility weights @xmath496 , and normalization factor @xmath134 .
the discrete form of the normalization derived in equation ( [ eq : zu ] ) is @xmath527 then , @xmath528 is the weighted sum of visibilities used for the estimators . note that because @xmath529 , there are also visibilities @xmath165 for which @xmath530 lies within the support range around @xmath129 , i.e.@xmath531 .
thus , we should add in the complex conjugates @xmath532 @xmath533 to do this , we construct another kernel @xmath534 @xmath535 which will gather the appropriate @xmath532 , giving @xmath536 for the final form of our linear estimator . for estimated visibility variances
@xmath537 , the optimal weighting factor ( in the least - squares estimation sense ) is given by @xmath538 but may also include factors based on position in the mosaic or frequency channel .
for example , up until now we have neglected the frequency dependence of the observed visibilities .
if we are reconstructing an intensity field with a uniform flux density spectrum , then no changes need be made .
if there is a frequency dependence , such as that for a power - law foreground ( e.g. eq.[[eq : alpha ] ] ) or the thermal spectrum of the cmb ( e.g. eq.[[eq : fplanck ] ] ) , then the visibilities should be scaled and weighted by the appropriate factor @xmath539 when gridded in order to properly estimate @xmath540 or @xmath541 respectively .
for example , for the cmb using equation ( [ eq : fplanck ] ) for the spectrum , we find @xmath542 in practice for the cbi , the frequency range of the data is not great enough for the spectral weighting factor to matter , and we therefore use the default weighting given in equation ( [ eq : viswt ] ) .
this will therefore be slightly non - optimal in the signal - to - noise sense ( it will not be the minimum - variance estimator ) but it will not introduce a bias in the bandpowers .
the choice of the normalization @xmath134 is somewhat arbitrary , as it only determines the units of the @xmath126 and not the correlation properties .
however , this can be important if we wish to use the estimators to make images using the formalism of [ sec : image ] .
for instance , the normalization given in equation ( [ eq : zi ] ) has the drawback of diverging in cells where all the @xmath543 are vanishingly small ( such as the innermost and outermost supported parts of the _ uv_-plane ) , and will produce images with heightened noise on short and long spatial wavelengths .
it is therefore more convenient to use the alternate normalization @xmath544 which when inserted into equation ( [ eq : linest2 ] ) will properly normalize the weighted sums of visibilities .
this will then produce images with the desired units of janskys per beam ( see [ sec : image ] ) .
we therefore use equation ( [ eq : zk1 ] ) for the normalization in our cbi pipeline .
we wish to calculate @xmath312 ( cf .
eq.[[eq : creskk ] ] ) using equation ( [ eq : resampint ] ) with @xmath545 . if @xmath546 is independent of flux density , then @xmath547 where we have left in the possibility that the upper flux density cutoff will depend on spectral index ( see below ) and set the lower flux density cutoff to zero ( the results for realistic power - law counts with @xmath548 are insensitive to the lower cutoff , but one can easily be included ) . as an example for the calculation of the fluctuation power due to residual sources in the gaussian limit , consider power - law integral source counts @xmath549 where @xmath550 is the mean number density of sources with flux density _ greater _ than @xmath307 at frequency @xmath46 , and a gaussian spectral index distribution at frequency @xmath46
@xmath551 first consider the case where there is a fixed flux density upper cutoff @xmath305 at the frequency where the number counts are defined .
the two parts of equation ( [ eq : creskk ] ) separate easily , where the source count part of the integral is @xmath552 for the distribution in equation ( [ eq : gausalpha ] ) , the integral over @xmath43 becomes @xmath553 where @xmath554 , and @xmath555 where @xmath556 is the mean of the extrapolated spectral index distribution which remains a gaussian , and the effective spectral index @xmath310 for the spectral component is shifted from the mean spectral index of the input distribution @xmath277 by the combination of the scatter in the @xmath43 and the lever arm @xmath557 from the frequency extrapolation . putting these together ,
we get @xmath558 one can also deal with the case where there is an upper flux density cutoff @xmath559 imposed at a frequency @xmath560 other than @xmath46 where the @xmath561 distribution is defined . in this case , the flux density cutoff in equation ( [ eq : resampdef ] ) is @xmath562 where @xmath563 , and @xmath564 is the cutoff @xmath559 extrapolated to @xmath46 using @xmath277 .
then , @xmath565 and thus @xmath566 with @xmath567 where @xmath568 gives the modification of the effective spectral index due to the change in the frequency at which the cutoff is done .
one often has an upper flux density cutoff at two different frequencies .
for example , sources that are extrapolated to be bright at the cmb observing frequency will have been detected and subtracted . if there is a flux density cutoff of @xmath559 imposed at a frequency @xmath560 as before , but an additional upper cutoff of @xmath569 at another frequency @xmath570 , then there is a critical spectral index @xmath571 above which the effective cutoff @xmath564 of equation ( [ eq : snumax ] ) changes from that appropriate to @xmath560 to that at @xmath570 ( assuming @xmath572 ) .
thus , the integral over @xmath43 in equation ( [ eq : resampcut ] ) will be broken into two pieces , @xmath573 where @xmath574 . the quantities in @xmath575 are as defined in equations ( [ eq : snumax ] ) through ( [ eq : alphaeffhat ] ) , and the parameters in @xmath576 are defined in the same way but using the higher frequency @xmath570 .
the truncated gaussian integrals are just the integrated probabilities for the normal distribution @xmath577 with @xmath578 the error function .
then , @xmath579 \label{eq : j2}\end{aligned}\ ] ] where @xmath580 and @xmath581 . as an example , consider the source counts presented in paper ii ( 4.3.2 ) , with @xmath582 above @xmath583 mjy at @xmath584 ghz and @xmath585 , which gives @xmath586 as the raw source power . in the analysis described there , mason et al .
find that a gaussian 1.4 ghz to 31 ghz spectral index distribution with @xmath587 and @xmath588 fits the observed data .
the cbi and ovro direct measurements have a cutoff of @xmath589 mjy at 31 ghz ( sources brighter than this have been subtracted from the cbi data and have residual uncertainties placed in a source covariance matrix ) , and sources above @xmath590 mjy at @xmath591 ghz have already been accounted for in a second source matrix .
therefore , the critical spectral index is @xmath592 from equation ( [ eq : alpcrit ] ) . for @xmath593 , the 31 ghz cutoff holds .
since the cutoff and source distribution are at the same frequency as the observations @xmath594 , there is no extrapolation factor @xmath595 and the spectral index distribution is unchanged ( @xmath596 ) .
then , @xmath597 and @xmath598 , so @xmath599 = 0.003\,{\rm jy}^2\,{\rm sr}^{-1}\ ] ] for the flat - spectrum tail of the spectral index integral .
the rest of the integral uses the 1.4 ghz cutoff , which we extrapolate using the mean spectrum to 31 ghz using equation ( [ eq : snumax ] ) , @xmath600 because @xmath595 , we have to modify the quantities in equation ( [ eq : alphaeffhat ] ) by explicitly expanding the terms in @xmath568 , and canceling remaining terms in @xmath557 , giving @xmath601 which can then be inserted into equation ( [ eq : j1 ] ) , giving @xmath602 for @xmath603 , @xmath604 , and thus we expect @xmath605 for the amplitude of the residual sources in the cbi fields . in paper
ii , it is noted that there is a 25% uncertainty on @xmath606 , and more importantly the power - law slope of the source counts could conceivably be as steep as @xmath607 . taking the extreme of @xmath607
, we get @xmath608 using the above procedure . we thus conservatively
estimate a 50% uncertainty on the value of @xmath609 derived in this manner .
note that in paper ii we actually use the value of @xmath610 derived using a monte - carlo procedure emulating the integrals in equation ( [ eq : intj ] ) but using the actual observed distribution of source flux densities and spectral indices .
the agreement between these two estimates shows the efficacy of this procedure in practice .
recently , ( * ? ? ? * hm ) have independently proposed a binned _ uv_-plane method that is somewhat similar to ours , though it is more directly related to the `` optimal maps '' of @xcite .
hm use a gathering mapping @xmath611 ( @xmath186 in their notation ) @xmath612 rather than our scattering kernel @xmath131 of equation ( [ eq : linop ] ) . in the hm method
, the vector @xmath613 can be thought of as a set of ideal pixels in the _
they show that the likelihood depends upon binned visibilities @xmath493 and noise @xmath138 @xmath614 where @xmath615 the hm kernel @xmath616 is chosen to equal 1 if the @xmath617 of visibility @xmath618 lies in cell @xmath62 , though other more complicated kernels could be imagined .
the hm method will also give a calculational speedup through the reduction in number of independent gridded estimators , and the use of the method is demonstrated using simulated vsa data in their paper . .
note that both @xmath619 and its conjugate @xmath620 have overlapping support for visibility @xmath618 , and this must be taken into account in computing the covariance matrix element.[fig : support],width=624 ] | we describe an algorithm for the extraction of the angular power spectrum of an intensity field , such as the cosmic microwave background ( cmb ) , from interferometer data .
this new method , based on the gridding of interferometer visibilities in the aperture plane followed by a maximum likelihood solution for bandpowers , is much faster than direct likelihood analysis of the visibilities , and deals with foreground radio sources , multiple pointings , and differencing .
the gridded aperture - plane estimators are also used to construct wiener - filtered images using the signal and noise covariance matrices used in the likelihood analysis .
results are shown for simulated data .
the method has been used to determine the power spectrum of the cosmic microwave background from observations with the cosmic background imager , and the results are given in companion papers . |
in algebraic statistics @xcite , a _ model for discrete data _ is a map @xmath0 whose coordinates @xmath1 are polynomial functions in the parameters @xmath2 .
the parameter vector @xmath3 ranges over an open subset @xmath4 of @xmath5 such that @xmath6 lies in the positive orthant @xmath7 .
the image @xmath8 represents a family of probability distributions on an @xmath9-element state space , provided we make the extra assumption that @xmath10 is the zero polynomial . a given data set is a vector @xmath11 of positive integers .
the problem of _ maximum likelihood estimation _ is to find parameters @xmath3 which best explain the data @xmath12 .
this leads to the following optimization problem : @xmath13 under suitable assumptions we have an optimal solution @xmath14 to the problem ( [ mle ] ) , which is an algebraic function of the data @xmath12 . our goal is to compute the degree of that algebraic function .
we call this number the _ maximum likelihood degree _ of the model @xmath15 .
equivalently , the _ ml degree _ is the number of complex solutions to the critical equations of ( [ mle ] ) , for a general data vector @xmath12 . in this paper
we prove results of the following form : [ thm1 ] let @xmath1 be polynomials of degrees @xmath16 in @xmath17 unknowns . if the maximum likelihood degree of the model @xmath18 is finite then it is less than or equal to the coefficient of @xmath19 in the generating function @xmath20 equality holds if the coefficients of the polynomials @xmath21 are sufficiently generic . as an example , consider a model given by @xmath22 quadratic polynomials in @xmath23 parameters .
the solution to ( [ mle ] ) satisfies the two critical equations @xmath24 if the @xmath21 s are general quadrics then these equations have @xmath25 complex solutions .
the formula for the maximum likelihood degree in theorem [ thm1 ] gives @xmath26 for special quadrics @xmath21 , the ml degree can be much lower than @xmath25 .
a familiar example is the _ independence model for two binary random variables _ :
@xmath27 here the ml degree is only one because the maximum likelihood estimate @xmath14 is a rational function (= algebraic function of degree one ) of the data @xmath12 : @xmath28 this paper is organized as follows . in section 2
we present the algebraic geometry for studying critical points of a rational function @xmath29 on an irreducible projective variety @xmath30 .
the critical equations @xmath31 are interpreted as sections of the sheaf @xmath32 of @xmath33-forms with logarithmic singularities along the divisor @xmath34 defined by @xmath35 . in theorem
[ chernclassthm ] , we show that if @xmath34 is a global normal crossing divisor then the ml degree equals the degree of the top chern class of @xmath36 .
if @xmath30 is projective @xmath17-space then this leads to theorem [ mle ] . in section 3
we study the case when @xmath30 is a smooth toric variety , and we derive a formula for the ml degree when the @xmath21 s are laurent polynomials which are generic relative to their newton polytopes . for instance , example [ rectangles ] shows that the ml degree is @xmath37 if we replace ( [ 2by2table ] ) by @xmath38 section 4 is concerned with the relationship of the ml degree to the bounded regions of the complement of @xmath39 in @xmath5 .
the number of these regions is a lower bound to the number of real solutions of the critical equations , and therefore a lower bound to the ml degree .
we show that for plane quadrics all three numbers can be equal .
however , for other combinations of plane curves the ml degree and the number of bounded regions diverge , and we prove a tight upper bound on the latter in theorem [ thm : viro ] .
also , following work of terao @xcite and varchenko @xcite , we show in theorem [ thm - linear ] that the ml degree coincides with the number of bounded regions of the arrangement of hyperplanes @xmath40 when the @xmath21 s are ( not necessarily generic ) linear forms .
section 5 revisits the ml degree for toric varieties , replacing the smoothness assumption by a much milder condition .
theorem [ toricformula4 ] gives a purely combinatorial formula for the ml degree in terms of the newton polytopes of the polynomials @xmath21 .
this section also discusses how resolution of singularities can be used to compute the ml degree for nongeneric polynomials .
section 6 deals with topological methods for determining the ml degree .
theorem [ euler ] shows that , under certain restrictive hypotheses , it coincides with the euler characteristic of the complex manifold @xmath41 , and theorem [ semiconti ] offers a general version of the semi - continuity principle which underlies the inequality in theorem [ thm1 ] . in section 7
we relate the ml degree to the sheaf of logarithmic vector fields along @xmath34 , which is the sheaf dual to @xmath42 .
this paper was motivated by recent appearances of the concept of ml degree in statistics and computational biology .
chor , khetan and snir @xcite showed that the ml degree of a phylogenetic model equals @xmath43 , and geiger , meek and sturmfels @xcite proved that an undirected graphical model has ml degree one if and only if it is decomposable .
the notion of ml degree also makes sense for certain parametrized models for continuous data : drton and richardson @xcite showed that the ml degree of a gaussian graphical model equals @xmath44 , and bout and richards @xcite studied the ml degree of certain mixture models .
the ml degree always provides an upper bound on the number of local maxima of the likelihood function .
our ultimate hope is that a better understanding of the ml degree will lead to the development of custom - tailored algorithms for solving the critical equations @xmath45 .
there is a need for such new algorithms , given that methods currently used in statistics ( notably the _ em - algorithm _ ) often produce only local maxima in ( [ mle ] ) .
in this section we work in the following general set - up of algebraic geometry .
let @xmath30 be a complete factorial algebraic variety over the complex numbers @xmath46 .
we also assume that @xmath30 is irreducible of dimension @xmath47 . in applications to statistics , the variety @xmath30 will often be a smooth projective toric variety .
suppose that @xmath48 is a rational function on @xmath30 .
since @xmath30 is factorial , the local rings @xmath49 are unique factorization domains .
this means that the function @xmath35 has a global factorization which is unique up to constants : @xmath50 here @xmath51 is a prime section of an invertible sheaf @xmath52 where @xmath53 is the divisor on @xmath30 defined by @xmath51 . in our applications
we usually assume that @xmath54 where @xmath9 is the number considered in the introduction .
for instance , if @xmath1 are polynomials and @xmath55 then @xmath56 ; namely , @xmath57 are the homogenizations of @xmath58 using @xmath59 , and @xmath60 ( see the proof of theorem [ thm1 ] for details ) . by ( [ globfac ] ) , we can write the divisor of the rational function @xmath35 uniquely as @xmath61 where the @xmath62 s are ( possibly negative ) integers . let @xmath34 be the reduced union of the codimension one subvarieties @xmath63 , or , as a divisor , @xmath64 .
we are interested in computing the critical points of the rational function @xmath35 on the open set @xmath65 complementary to the divisor @xmath34 .
especially , we wish to know the number of critical points , counted with multiplicities .
a critical point is by definition a point @xmath66 where the differential @xmath33-form @xmath67 vanishes .
if @xmath68 is a smooth point on @xmath30 , and @xmath69 are local coordinates , then @xmath70 . hence @xmath68 is a critical point of @xmath35 if and only if @xmath71 we next rewrite the critical equations ( [ criticalequations ] ) using the factorization ( [ globfac ] ) . around each point @xmath66 , we may choose a local trivialization for the sheaf @xmath52 and express @xmath51 locally by a regular function .
by slight abuse of notation , we denote that regular function also by @xmath51 .
for instance , if @xmath55 then this means replacing the homogeneous polynomial @xmath51 by a dehomogenization . since @xmath35 has neither zeros nor poles on the open set @xmath72 , the vanishing of @xmath73 is equivalent to the vanishing of the logarithmic derivative @xmath74 we now recall some classical definitions and results concerning the sheaf of differential @xmath33-forms with logarithmic singularities along @xmath34 .
the standard references on this subject are dligne s book @xcite and saito s paper @xcite .
we define @xmath75 as a subsheaf of the sheaf @xmath76 of @xmath33-forms with poles at most on @xmath34 and of order one .
this sheaf is the image of the natural map @xmath77 which is given by the inclusion @xmath78 and the homomorphisms sending @xmath79 . for experts we note that our definition differs from the one in @xcite when @xmath34 is not normal crossing .
saito s sheaf is the double dual of our @xmath80 , which explains why his is always locally free when @xmath30 is a surface ( * ? ? ?
* corollary 1.7 ) .
ours need not be locally free even for surfaces
. however , our definition gives a natural exact sequence .
[ exactlem ] if @xmath30 is factorial and complete then we have an exact sequence @xmath81 the local sections of the sheaf @xmath82 are rational @xmath33-forms which can be written as @xmath83 , where @xmath84 is a regular @xmath33-form .
ssince the @xmath85 s are distinct prime divisors and @xmath30 is factorial , the local rings @xmath86 are discrete valuation rings with parameter @xmath51 .
thus @xmath87 is invertible in this local ring for @xmath88 , and @xmath89 is regular if and only if @xmath51 divides @xmath90 .
this implies that the homomorphism which sends @xmath89 to the vector @xmath91 is well defined , and it induces an isomorphism from the quotient @xmath92 onto @xmath93 .
assume now that @xmath30 is smooth .
then both sheaves @xmath94 and @xmath95 are locally free of rank @xmath96 .
hence the intermediate sheaf @xmath97 is torsion free of the same rank .
our next result shows that @xmath97 is locally free if and only if the divisors @xmath85 are smooth and intersect transversally .
[ prop1 ] let @xmath66 be a smooth point , @xmath98 local coordinates at @xmath68 and @xmath99 the divisors which contain @xmath68 .
then the sheaf @xmath97 is locally free at @xmath68 if and only if the @xmath100-matrix @xmath101 has rank @xmath102 at @xmath68 .
any local section of @xmath82 can be written in the form @xmath103 this observation gives rise to a local exact sequence @xmath104 the surjective map on the right takes @xmath105 to the sum on the right hand side of ( [ omegasum ] ) . the injective map on the left takes the @xmath102-tuple @xmath106 to @xmath107 the exactness of the sequence ( [ localexseq ] ) follows from the proof of lemma [ exactlem ] . if the section @xmath89 in ( [ omegasum ] ) is identically zero in @xmath108 then @xmath89 is in particular regular , and so @xmath51 divides each @xmath90 .
now , since @xmath30 is reduced , a coherent sheaf @xmath109 is locally free of rank @xmath17 if and only if @xmath110 for each point @xmath68 .
since tensor product is right exact , it follows that this condition is verified for @xmath97 if and only if the matrix of @xmath111 , evaluated at @xmath68 , has rank precisely @xmath102 . since the functions @xmath112 vanish at @xmath68 , this is exactly the asserted condition that the jacobian marix @xmath113 has rank @xmath102 at @xmath68 . in the above situation where @xmath30 is smooth and @xmath75 is locally free
we shall say that the divisor @xmath34 has _ global normal crossings _ ( or gnc ) .
[ chernclassthm ] let @xmath30 be smooth and assume that @xmath34 is a gnc divisor
. then 1 .
the section @xmath114 of @xmath75 does not vanish at any point of @xmath34 , 2 . if the divisor @xmath34 intersects every curve in @xmath30 ( in particular , if @xmath34 is ample ) then @xmath115 vanishes only on a finite subset of @xmath116 , 3 .
if the above conclusions hold , then the number of critical points of @xmath35 on @xmath72 , counted with multiplicities , equals the degree of the top chern class @xmath117 we abbreviate @xmath118 . by the proof of proposition [ prop1 ] it follows that if @xmath119 has rank @xmath102 at @xmath68 , then @xmath75 is locally free of rank @xmath17 with generators @xmath120 and some choice of @xmath121 of the @xmath122 .
if we write @xmath123 in this basis , the coefficients of @xmath124 are the constants @xmath62 while the coefficients of the @xmath122 are some regular functions .
the first assertion follows immediately since the exponents @xmath62 are all nonzero .
the second assertion follows from the first : let @xmath125 be the zero set of the section @xmath123 .
since @xmath125 does not intersect @xmath34 , it follows that @xmath126 .
thirdly , if @xmath109 is a locally free sheaf of rank @xmath17 on a smooth variety @xmath30 of dimension @xmath17 , and @xmath127 is a section of @xmath128 with a zero scheme @xmath125 of dimension @xmath129 , then the length of @xmath130 equals the degree of the top chern class @xmath131 .
the _ total chern class _ of a sheaf @xmath109 is the sum @xmath132 .
this is a polynomial in @xmath133 whose coefficients are elements in the chow ring @xmath134 . recall that every element in @xmath135 has a well - defined _ degree _ which is the image of its degree @xmath17 part under the degree homomorphism @xmath136 .
suppose that @xmath30 is smooth and @xmath34 is a gnc divisor on @xmath30 which intersects every curve .
then the number of critical points of @xmath35 , counted with multiplicities , is the degree of the coefficient of @xmath19 in the following polynomial : @xmath137.\ ] ] the total chern class @xmath138 is multiplicative with respect to exact sequences , i.e. , if @xmath139 is an exact sequence of sheaves , then @xmath140 .
hence the sequence ( [ lemmasequence ] ) implies the result . in the next section ,
we apply the formula ( [ totalchernclass ] ) in the case when @xmath30 is a smooth projective toric variety .
the chow group @xmath141 has rank one and is generated by the class of any point .
this canonically identifies @xmath141 with @xmath142 and so any top chern class can be considered to be a number .
[ toricformula ] suppose @xmath30 is a smooth toric variety with boundary divisors @xmath143 and @xmath34 is gnc and meets every curve .
the number of critical points of @xmath35 , counted with multiplicity , equals the coefficient of @xmath19 in @xmath144.\ ] ] by virtue of equation we need only compute the total chern class @xmath145 . for this
we use the exact sequence in @xcite , @xmath146 where @xmath147 , and the fact that @xmath148 is trivial .
we now apply the results of the previous section to models @xmath149 . to illustrate how this works ,
we first prove theorem [ thm1 ] for generic polynomials .
the proof of the statement that the ml degree of generic polynomials is an upper bound on the ml degree of special polynomials ( when this number is finite ) is deferred to theorem [ toricformula3 ] which is a generalization of theorem [ thm1 ] .
see also theorem [ semiconti ] where this semi - continuity principle is stated in general .
the polynomials @xmath150 are assumed to be generic among all ( nonhomogeneous ) polynomials of degrees @xmath16 in @xmath151 , and @xmath152 are positive integers .
we take @xmath30 to be projective space @xmath153 with coordinates @xmath154 .
our object of interest is the following rational function on @xmath55 : @xmath155 the global factorization ( [ globfac ] ) of this @xmath156 has @xmath56 prime factors , namely , @xmath157 and @xmath158 with @xmath159 .
the chow ring of @xmath55 is @xmath160/ \langle h^{d+1 } \rangle$ ] , where @xmath161 represents the hyperplane class . by our genericity hypothesis ,
the @xmath162 prime factors of @xmath156 are smooth and global normal crossing .
they correspond to the following divisor classes : @xmath163 projective space @xmath153 is a smooth toric variety with @xmath164 torus - invariant divisors @xmath165 , each having the same class @xmath161 .
hence the formula in ( [ toricformula2 ] ) specializes to @xmath166 since we work in the chow ring of projective space @xmath153 , the coefficient of @xmath167 is the same as the coefficient of @xmath19 in the generating function in ( [ thm1formula ] ) .
we now generalize our results from polynomials of fixed degrees to laurent polynomials with fixed newton polytopes . recall that the _ newton polytope _ of a laurent polynomial @xmath168 is the convex hull of the set of exponent vectors of the monomials appearing in @xmath35 with nonzero coefficient . given a convex polytope @xmath169 with vertices in @xmath170 , by a _ generic laurent polynomial with newton polytope @xmath171
_ we will mean a sufficiently general @xmath172-linear combination of monomials with exponent vectors in @xmath173 .
in the next theorem we consider @xmath9 laurent polynomials @xmath174 having respective newton polytopes @xmath175 . because the @xmath21 s are laurent polynomials , i.e.
, their monomials may have negative exponents , we only consider those critical points of @xmath176 which lie in the algebraic torus @xmath177 .
the number of such critical points ( counted with multiplicity ) will be called the _ toric ml degree _ of the rational function @xmath35 .
let @xmath178 denote the _ minkowski sum _ of the given newton polytopes , and let @xmath30 be the projective toric variety defined by @xmath171 .
let @xmath179 be the primitive inner normal vectors of the facets of @xmath171 .
they span the rays of the fan of @xmath30 .
let @xmath180 denote the corresponding torus - invariant divisors on @xmath30 .
each of the newton polytopes @xmath181 is the solution set of a system of linear inequalities of the specific form @xmath182 the divisor on @xmath30 defined by the laurent polynomial @xmath21 is linearly equivalent to @xmath183 .
the @xmath184 are integers which can be positive or negative .
the divisor on @xmath30 defined by @xmath185 is linearly equivalent to @xmath186 we abbreviate the _ support _ of this divisor by @xmath187 a toric variety @xmath30 is smooth if all the cones in its normal fan are unimodular .
[ toricformula3 ] if the toric variety @xmath30 is smooth and the toric ml degree of the rational function @xmath35 is finite then it is bounded above by the coefficient of @xmath19 in the following generating function with coefficients in the chow ring of @xmath30 : @xmath188 equality holds if each @xmath21 is generic with respect to its newton polytope @xmath181 . note
that theorem [ thm1 ] is the special case of theorem [ toricformula3 ] when @xmath181 is the standard @xmath17-dimensional simplex @xmath189 scaled by a factor of @xmath190 .
let us first assume that @xmath21 is a generic laurent polynomial with newton polytope @xmath181 .
let @xmath191 $ ] be the homogeneous coordinate ring @xcite of @xmath30 with one variable for each torus - invariant divisor @xmath165 .
given a laurent polynomial @xmath192 with newton polytope @xmath181 , the corresponding rational function on @xmath30 is @xmath193 where @xmath85 is as defined above and @xmath51 is homogeneous of degree @xmath85 .
therefore the rational function on @xmath30 we are interested in is @xmath194 we next show that the divisor of @xmath156 is gnc .
note that @xmath51 is a generic section of a line bundle on @xmath30 that is generated by its sections .
this implies , by the bertini - sard theorem and by induction on @xmath9 , that the divisors @xmath195 meet transversally in the dense torus of @xmath30 . for points in the boundary of @xmath30
, we simply restrict to the torus orbit determined by the corresponding facet where the restricted @xmath51 s remain generic sections of the restricted bundles .
the reduced divisor of poles and zeros of @xmath156 is @xmath196 + @xmath197 where @xmath198 is defined as in ( [ whatisi ] ) .
since @xmath199 is the divisor corresponding to @xmath171 it is ample on @xmath30 by construction .
so @xmath199 meets every curve on @xmath30 and therefore so does @xmath34 and we can apply corollary [ toricformula ] .
a variable @xmath200 appears as a factor in @xmath156 if and only if @xmath201 , in which case @xmath202 appears in both the numerator and denominator of ( [ toricformula2 ] ) , and we get the expression ( [ toricchern ] ) . consider now arbitrary laurent polynomials @xmath203 in @xmath151 such that @xmath204 has only finitely many critical points in @xmath205 .
let @xmath206 be the coefficient of @xmath19 in ( [ toricchern ] ) .
let @xmath207 be the space of all @xmath9-tuples of laurent polynomials with the given newton polytopes .
consider the critical equations of @xmath208 and clear denominators .
the resulting collection of @xmath17 laurent polynomials defines an algebraic subset @xmath209 in the product space @xmath210 . saturate @xmath209 to remove any components along the hypersurfaces @xmath40 and get a new algebraic subset @xmath211 .
the map from @xmath211 onto @xmath207 is dominant and generically finite , and the generic fiber of this map consists of @xmath206 points .
our given laurent polynomials @xmath203 represent a point @xmath212 in @xmath207 .
let @xmath213 be the isolated critical points of @xmath35 .
for each @xmath214 , consider any irreducible component @xmath215 of @xmath211 containing the point @xmath216 in @xmath217 .
by krull s principal ideal theorem , the component @xmath215 of @xmath211 has codimension @xmath218 and hence it has dimension @xmath219 . as the generic fiber is finite , the dimension of @xmath220 is exactly @xmath221 and the projection to @xmath207 is dominant .
since @xmath222 is an isolated solution of the critical equations , the projection map to @xmath207 is open ( * ? ? ?
* ( 3.10 ) ) , so the intersection of @xmath215 with an open neighborhood of @xmath216 maps onto an open neighborhood of @xmath212 .
hence every generic point @xmath223 near @xmath212 has a preimage @xmath224 near @xmath216 , and these preimages are distinct for @xmath225 .
we conclude that @xmath226 .
this semicontinuity argument is called the `` specialization principle '' stated in mumford s book ( * ? ? ?
* ( 3.26 ) ) and also works when the @xmath222 have multiplicities , as shown in theorem [ semiconti ] below .
@xmath227 we illustrate theorem [ toricformula3 ] with two examples which we revisit in section 5 .
[ rectangles ] consider @xmath9 generic polynomials @xmath228 where the support of @xmath21 consists of monomials @xmath229 with @xmath230 and @xmath231 , and suppose the @xmath62 s are generic .
the newton polytope of @xmath21 is the rectangle @xmath232 the minkowski sum of these rectangles is another rectangle , and @xmath233 . in the numerator of ( [ toricchern ] ) , the contribution of the two torus - invariant divisors @xmath34 and @xmath234 corresponding to the left and the bottom edge of this rectangle survives .
the denominator comes from the product of the divisors of @xmath1 : @xmath235 now , the coefficient of the term @xmath236 modulo the chow ring relations @xmath237 gives the toric ml degree @xmath238 [ haseightrays ] let @xmath239 be generic polynomials in @xmath240 and @xmath241 with supports @xmath242 the corresponding newton polytopes @xmath243 are shown in figure [ fig2 ] .
the normal fan of the minkowski sum has eight rays and is shown in figure [ fig3 ] .
theorem [ toricformula3 ] applies because the toric surface @xmath30 is smooth .
we label the eight rays by @xmath244 in counterclockwise order , starting with @xmath245 .
the chow ring @xmath246 is the polynomial ring @xmath247 $ ] modulo the ideal @xmath248 the three divisors corresponding to the polygons @xmath249 in figure [ fig2 ] are @xmath250 if all @xmath62 are positive , then the support of the divisor @xmath251 is @xmath252 .
it follows that the toric ml degree is the coefficient of @xmath236 in @xmath253 this coefficient is @xmath254 , which means that the toric ml degree is @xmath255 .
the _ toric ml degree of the model _
@xmath256 is the toric ml degree defined above for generic @xmath12 . in this case
, there is no cancellation among the coefficients in ( [ whatisi ] ) , and @xmath198 is the set of all indices @xmath257 such that for some @xmath181 the supporting hyperplane normal to @xmath258 does not pass through the origin .
the toric ml degree of @xmath15 is a numerical invariant of the polytopes @xmath259 .
a combinatorial formula for this invariant will be presented in theorem [ toricformula4 ] of section 5 .
as in the introduction , we consider @xmath9 polynomials @xmath260 in @xmath17 unknowns @xmath261 .
we now assume that all coefficients of the @xmath21 s are real numbers , and we also assume that @xmath152 are positive integers .
however , we do not assume that the union of the divisors of the @xmath21 s has global normal crossings .
this is the case of interest in statistics .
consider the arrangement of hypersurfaces defined by the @xmath21 s and let @xmath262 be the complement of this arrangement .
a connected component of @xmath263 is a _ bounded region _ if it is bounded as a subset of @xmath5
. then the following observation holds .
[ bounded ] for any polynomial map @xmath264 and any @xmath265 , @xmath266 the function @xmath267 is continuous , and on the boundary of the closure of each bounded region its value is zero . hence it has to have at least one ( real ) critical point in the interior of each region .
the second inequality holds trivially , since the ml degree was defined as the number of critical points of @xmath268 in @xmath269 , counted with multiplicities .
this observation raises the question whether the inequalities above could be realized as equalities .
we next show that this is the case when @xmath150 are quadrics in the plane . here
the ml degree is @xmath270 by theorem [ thm1 ] . for each @xmath9
, there are @xmath9 quadrics @xmath150 in @xmath271 such that @xmath272 hence all critical points are real
. we will take @xmath9 quadrics that define `` nested '' ellipses with center at the origin , as suggested by figure [ fig1 ] .
the proof follows by induction : assume we have @xmath273 bounded regions with @xmath274 ellipses .
observe that the @xmath275st ellipse contains @xmath276 bounded regions .
then we add a new long and skinny ellipse which replaces the @xmath276 regions with @xmath277 regions .
the total count comes out to be @xmath278 .
we will see such an equality holding for @xmath9 linear hyperplanes in @xmath5 below .
however , even in the plane @xmath271 , the number of critical points and the number of bounded regions of @xmath263 diverge for curves of degree @xmath279 .
theorem [ thm1 ] implies that for @xmath9 generic plane curves of degrees @xmath280 the ml degree is @xmath281 the optimal upper bound for the number of bounded regions of @xmath263 is smaller than the ml degree , by the following unpublished result due to oleg viro .
[ thm : viro ] ( viro ) let @xmath150 be real plane curves of degrees @xmath280 , and let @xmath282 be the number of odd degree curves among them .
then @xmath283 and this bound is optimal .
the proof is by induction . for @xmath284
the bound above is harnack s inequality @xcite , and it is optimal .
suppose the statement is proved for @xmath274 curves , and @xmath285 defines a curve of degree @xmath286 .
we will take this new curve so that it has the maximum number of bounded regions allowed by harnack s inequality , i.e. @xmath287 if @xmath286 is even , and one less than that if @xmath286 is odd .
one can achieve @xmath288 by taking a curve with @xmath289 unnested ovals and one more distinguished piece ( that gives an extra bounded region when @xmath286 is even ) such that some line intersects this distinguished piece in exactly @xmath286 points .
we can arrange this last curve in such a way that when we superimpose the distinguished piece on the arrangement given by @xmath274 curves , the last curve will intersect the curve given by @xmath21 in @xmath290 points .
now if we trace this last distinguished piece , every time we encounter an intersection point an extra bounded region is created , except the last point in case @xmath286 is odd . together with the remaining @xmath289 ovals we get the bound . in order to get any meaningful lower bound on the number of bounded regions of @xmath291 one needs to make some assumptions . without any assumptions
the lower bound is zero : for @xmath21 of even degree we take an empty ( real ) curve , and for @xmath21 of odd degree we take the union of an empty curve with a line .
if we let all the lines intersect in a single point there will not be any bounded region . if we insist on at least having a gnc configuration , then by the same construction the lower bound we get is the number of bounded regions in a generic arrangement of @xmath282 lines where @xmath282 is the number of odd degrees @xmath190 .
this idea leads us to studying the ml degree of a hyperplane arrangement .
[ thm - linear ] let @xmath15 be given by @xmath9 linear polynomials @xmath1 with real coefficients .
then the ml degree of @xmath15 is equal to the number of the bounded regions of @xmath291 , and all critical points of the optimization problem ( [ mle ] ) are real .
this theorem does not assume any hypothesis such as global normal crossing .
under the gnc hypothesis , the hyperplanes would be in general position and the number of bounded regions equals @xmath292 , as predicted by theorem [ thm1 ] .
theorem [ thm - linear ] is essentially due to varchenko @xcite .
we shall give a new proof . in light of proposition
[ bounded ] , we need to show that the number of bounded regions of @xmath291 equals the number of complex solutions of the ml equations .
let @xmath293 for @xmath294 .
the ml equations are @xmath295 consider the map @xmath296 given by @xmath297 . here @xmath298 is the homogenization of @xmath21 .
we let @xmath299 be the central hyperplane arrangement in @xmath300 given by the @xmath51 .
we assume that the intersection of all the hyperplanes in @xmath299 is just the origin ; otherwise , the linear forms @xmath51 depend on fewer than @xmath17 coordinates , and then we get infinitely many critical points . the zariski closure of @xmath301 in @xmath302 is a @xmath17-dimensional complex variety @xmath303 .
the solution set on @xmath303 of the @xmath17 linear equations @xmath304 consists of finitely many points provided @xmath152 are generic . obviously , the solutions to ( [ new - eq ] ) lift to such complex solutions . in other words
, the degree of the projective variety @xmath303 is an upper bound on the ml degree of @xmath15 .
now we will compute the degree of @xmath303 .
this variety is the projective spectrum of the @xmath305-graded algebra @xmath306\,$ ] where @xmath307 .
terao ( * ? ? ?
* theorem 1.4 ) showed that the hilbert series of @xmath308 is equal to @xmath309 here @xmath310 is the mbius function of the intersection lattice @xmath311 of the arrangement @xmath312 . from this series
we shall determine the leading coefficient of its hilbert polynomial .
this coefficient has the form @xmath313 , where @xmath314 is the degree of @xmath303 . for large enough @xmath315 ,
the coefficient of @xmath316 in the hilbert series ( [ terao ] ) equals @xmath317 this is the hilbert polynomial of the graded algebra @xmath308 .
its leading term is @xmath318 we conclude that the degree of the projective variety @xmath303 is @xmath319 . by zaslavsky @xcite ,
this number equals the number of bounded regions of @xmath291 .
a family of important statistical models where theorem [ thm - linear ] applies is the _ linear polynomial model _ of @xcite .
such a model is given by a polynomial in @xmath315 unknowns @xmath320 with indeterminate coefficients , @xmath321 together with @xmath9 data points @xmath322 .
the model is parametrized by @xmath323 the ml degree is the number of bounded regions of this arrangement .
we now return to the setting of section 3 , with the aim of relaxing the restrictive smoothness hypothesis in theorem [ toricformula3 ] .
our aim is to derive a combinatorial formula for the toric ml degree of any model @xmath324 defined by generic laurent polynomials satisfying a mild hypothesis .
the derivation of theorem [ toricformula4 ] involves resolution of singularities in the toric category . in the end of the section
we shall comment on using resolution of singularities for bounding the ml degree in general .
given a polytope @xmath171 in @xmath5 and a linear functional @xmath325 on @xmath5 , we write @xmath326 for the face of @xmath171 at which @xmath325 attains its minimum .
two linear functionals @xmath325 and @xmath327 are _ equivalent _ if @xmath328 .
the equivalence classes are the relative interiors of cones of the _ inner normal fan _ @xmath329 .
if @xmath123 is a cone in @xmath329 , or @xmath123 is a cone in any fan which refines @xmath329 , then we write @xmath330 for @xmath331 in the relative interior of @xmath123 . if @xmath35 is a polynomial with newton polytope @xmath171 then @xmath332 denotes the _ leading form _ consisting of all terms of @xmath35 which are supported on @xmath333 . as in section 3
, let @xmath1 be laurent polynomials with newton polytopes @xmath334 . consider any fan @xmath335 which is a common refinement of the inner normal fans
suppose @xmath337 is a cone in @xmath335 and let @xmath338 be the dimension of @xmath339 .
there exists a @xmath338-dimensional linear subspace @xmath340 of @xmath5 and vectors @xmath341 such that @xmath342 lies in @xmath340 for all @xmath343 .
the subspace @xmath340 is unique and satisfies @xmath344 .
let @xmath345 denote the _
normalized mixed volume _ on the subspace @xmath340 .
here `` normalized '' refers to the lattice @xmath346 , as is customary in toric geometry @xcite . for any @xmath338-element subset @xmath347 of @xmath348
we abbreviate @xmath349 and @xmath350 if @xmath351 . if @xmath352 and @xmath353 then we simply write @xmath354 for the mixed volume in ( [ mixedvolume ] ) .
if @xmath355 and @xmath356 is full - dimensional then ( [ mixedvolume ] ) equals @xmath33 ; this happens in the last sum of ( [ longmultline ] ) .
we are now ready to state our more general toric ml degree formula . as in section 3 , let @xmath30 be the toric variety corresponding to the minkowski sum @xmath357 and @xmath358 the normal fan with rays @xmath359 .
we consider the function @xmath360 .
each polytope @xmath181 corresponds to a divisor @xmath85 so the divisor of @xmath35 is @xmath361 .
let @xmath198 be the support of @xmath34 as in .
label the rays of @xmath358 so that @xmath362 are the indices not in @xmath198 .
for each subset @xmath363 of @xmath364 let @xmath365 denote the smallest cone of @xmath335 which contains the vectors @xmath366 for @xmath367 .
if no such cone exists then @xmath365 is just a formal symbol and the expression ( [ mixedvolume ] ) is declared to be zero for @xmath368 .
the mild smoothness hypothesis we need is that every singular cone of @xmath335 contains at least one ray from @xmath198 .
equivalently all cones @xmath365 are smooth .
[ toricformula4 ] suppose every singular cone of @xmath358 contains some ray in the support of the divisor @xmath34 .
then , the toric ml degree of the rational function @xmath35 is bounded above by the following alternating sum of mixed volumes : @xmath369 equality holds if each @xmath21 is generic relative to its newton polytope @xmath181 . in order to apply corollary [ toricformula ]
we must resolve the singularities of @xmath30 .
for toric varieties this is done in two steps .
first we get a simplicial toric variety without adding any new rays to the fan .
second we resolve the remaining singular ( but simplicial ) cones by adding new rays .
this procedure is described in detail in @xcite .
typically the first step involves taking the pulling subdivision at each ray in the fan .
however , under the given hypothesis it is enough to perform pulling subdivisions only at the rays in the support of @xmath34 to obtain a simplicial fan @xmath370 .
this fine detail will be important below .
our hypothesis holds for this intermediate fan as well , and subsequently we take a smooth refinement @xmath371 of @xmath370 by adding new rays in the relative interiors of each of the singular cones .
let @xmath372 be the induced map .
we will show that we get no new critical points under the resolution .
hence the number of critical points can be computed on @xmath373 .
we finally claim that the chern class formula expands into the given combinatorial formula .
we investigate critical points of the pullback of our rational function : @xmath374 for generic @xmath21 , the same argument as in the proof of theorem [ toricformula3 ] shows that the reduced divisor of poles and zeros @xmath375 of @xmath376 is gnc .
what we must show is that all critical points of @xmath376 on @xmath373 are off the exceptional locus hence actually critical points of @xmath156 on @xmath30 .
there are two types of new cones in @xmath371 .
the first come from the triangulation step . by our construction
, any such cone must contain a ray @xmath258 in the support of @xmath34 .
this ray corresponding to the strict transform under @xmath377 is in the support of @xmath378 , and its variable appears as a factor in @xmath376 . by part 1 of theorem [ chernclassthm ]
, @xmath376 has no critical points along the torus - invariant divisor @xmath379 , hence no critical points on any torus orbit contained in @xmath379 .
the second type of new cone comes from the desingularization step .
these cones all contain at least one new ray @xmath380 corresponding to an exceptional divisor @xmath381 in @xmath373 .
we will show there are no critical points on @xmath381 .
equivalently we show that there are no critical points on each torus orbit contained in @xmath381 . given a torus orbit
let @xmath382 be the corresponding cone of @xmath383 containing @xmath380 .
there is some minimal cone @xmath337 of @xmath335 containing @xmath382 .
let @xmath384 be any cone of @xmath383 containing @xmath382 that is maximal with respect to being contained in @xmath337 .
since @xmath337 is refined in @xmath383 it must be a singular cone , and so by the hypothesis it has some generating ray in the support of @xmath385 , or equivalently the linear function of this cartier divisor is not identically zero on @xmath337 .
the pullback keeps the same linear functional which can not be zero on the subset @xmath384 of @xmath337 . as a consequence
, @xmath384 contains a ray @xmath258 in the support of @xmath378 .
this means @xmath386 appears as a factor in @xmath376 for some nonzero integer @xmath387 . if @xmath258 is a generator of @xmath382 , then as above there are no critical points on @xmath379 and thus no critical points on the orbit corresponding to @xmath382 .
suppose on the other hand @xmath258 is not a generator of @xmath382 .
let @xmath388 be the variables corresponding to the generators of @xmath382 in an affine chart of @xmath373 that contains @xmath384 .
note that the variable @xmath200 corresponding to @xmath258 is not among these variables . because @xmath337 is the minimal cone of @xmath335 containing @xmath382 , the face @xmath389 is contained in the face @xmath390 , and
hence it is contained in the face @xmath391 .
so @xmath392 , obtained by setting @xmath388 to zero , does not contain any of the variables corresponding to @xmath384 ; in particular it does nt contain @xmath200 . on the other hand , @xmath393 where @xmath394 is in the ideal generated by @xmath388 . since @xmath200 is not among the @xmath395 we have @xmath396 where @xmath397 means @xmath398 .
hence we have @xmath399 we conclude that @xmath376 has no critical points on the torus orbit corresponding to @xmath382 as desired
. thus the toric ml degree of @xmath35 on @xmath373 is the same as that on @xmath35 .
since @xmath376 has no critical points on the exceptional locus and @xmath34 is ample on @xmath30 , @xmath400 meets any curve off the exceptional locus and therefore the ml degree must be finite .
it is computed in the chow ring of @xmath373 as the coefficient of @xmath19 in @xmath401 here @xmath402 are the strict transforms of the @xmath165 not in the support of @xmath385 .
the @xmath403 are those exceptional divisors which are not in the support of @xmath404 , and @xmath405 are the proper transforms of the divisor classes of the @xmath51 s .
we can now expand our chern class product replacing @xmath406 in the denominator by @xmath407 in the numerator .
the intersection product of any collection of prime torus - invariant divisors is the cycle of the cone they span or @xmath129 if there is no such cone .
hence the coefficient of @xmath19 is the sum of all terms of the form @xmath408 here @xmath409 and @xmath410 ranges over all dimension @xmath387 cones of @xmath383 spanned by rays not in the support of @xmath411 .
this product is exactly the mixed volume @xmath412 .
to finish we note that if @xmath410 contains an exceptional divisor @xmath403 , the minimal cone @xmath337 of @xmath335 containing @xmath410 must have dimension strictly larger than @xmath387 .
this is because @xmath410 does not have any rays in the support of @xmath411 hence is not maximal in @xmath337 . as a consequence all of the faces @xmath413
have a translate that lies in a subspace of dimension strictly less than @xmath414 and the corresponding mixed volume is @xmath129 . in conclusion , the exceptional divisors do not contribute to the top chern class product and the formula reduces to the stated one . in two variables
we recover a particularly simple formula : [ areaarea ] let @xmath415 be generic laurent polynomials in two variables @xmath416 with newton polygons @xmath417 . if the origin lies on none of the lines spanned by edges of their minkowski sum @xmath357 then the toric ml degree equals the area of @xmath171 plus the areas of each of the @xmath181 .
this is a special case of theorem [ toricformula4 ] when no facets pass through the origin .
therefore the only term is @xmath418 the euclidean area of each @xmath181 is @xmath419 .
the euclidean area of the minkowski sum @xmath420 equals @xmath421 .
the stated formula is the sum of these expressions .
now that we are equipped with the volume formulas in theorem [ toricformula4 ] and corollary [ areaarea ] , let us revisit the two - dimensional examples from section 3 .
the newton polygons @xmath422 in example [ rectangles ] are axis - parallel parallelograms .
the first term in the formula ( [ rectangleformula ] ) is the area of their minkowski sum @xmath423 , and the second term is the sum of the areas of the @xmath181 , as in corollary [ areaarea ] .
the third and fourth term are the two correction terms stemming from the fact that the origin is a vertex of each newton polytope .
these terms disappear if we replace one @xmath21 by @xmath424 .
the number @xmath255 in example [ haseightrays ] can also be derived using theorem [ toricformula4 ] .
the three polygons in figure [ fig2 ] have areas @xmath425 and @xmath426 respectively .
the area of their sum is @xmath427 .
the two divisors corresponding to @xmath428 and @xmath429 pass through the origin and yield correction terms of @xmath33 and @xmath430 respectively .
finally we add back @xmath33 for the vertex at the origin .
altogether @xmath431 .
.1 cm our discussion so far indicates that we get the sharpest results when @xmath30 is smooth and @xmath34 is gnc . in the toric case
the smoothness hypothesis could be largely removed as in theorem [ toricformula4 ] .
the gnc condition can also be relaxed for certain other cases as we saw in the previous section . in general ,
if the pair @xmath432 does not satisfy the smoothness and gnc hypotheses then we must appeal to hironaka s theorem on resolution of singularities ( see e.g. @xcite ) .
this furnishes a proper projective morphism @xmath433 such that @xmath373 is smooth and @xmath434 has gnc .
we need to compute the divisor @xmath435 of the pullback of the function @xmath35 .
if @xmath405 is the proper transform of the divisor @xmath85 and @xmath436 are the exceptional divisors of @xmath377 then @xmath437 where @xmath438 are certain ( possibly negative ) integers .
these integers are @xmath439 where @xmath440 is the multiplicity of the full transform of @xmath85 along @xmath441 .
the underlying reduced divisor is @xmath442 .
the number of critical points is now gotten by applying theorem [ chernclassthm ] to @xmath443 instead of @xmath432 .
this procedure can be very complicated in practice .
we illustrate it with a simple example .
[ 4 curves ] let @xmath23 , @xmath22 , @xmath444 , @xmath445 , @xmath446 , and @xmath447 .
the divisor @xmath34 is not a gnc divisor since at the origin all the four curves defined by @xmath448 meet . in order to resolve this singularity
we blow up @xmath449 at @xmath450 to obtain @xmath373 which is smooth .
we note that @xmath451 is @xmath452 where @xmath161 is the proper transform of the generic hyperplane section , @xmath234 is the exceptional divisor , and @xmath453 is the proper transform of a line through the origin .
we have four cases : we consider first the general case where @xmath454 and @xmath455 . in this case @xmath375
consists of the proper transforms of the four original curves , the exceptional divisor , and the pullback of the line at infinity .
after cancellations , we just need to compute the coefficient of @xmath236 in @xmath456 where @xmath457 and @xmath458 are the irreducible divisors corresponding respectively to the circle and the ellipse .
this coefficient is @xmath459 . in @xmath373 the two curves intersect in three points , and their self - intersection also yields three points .
hence the ml degree is nine . in the special case where @xmath460
, we need to compute the coefficient of @xmath236 in @xmath461 , which is @xmath462 . since @xmath463 the ml degree drops down to seven .
if @xmath464 then the coefficient of @xmath236 in @xmath465 is @xmath466 .
since @xmath467 , the ml degree is five .
finally , if both @xmath468 and @xmath469 are zero , then the coefficient of @xmath236 in @xmath470 is @xmath471 : since @xmath472 , the ml degree further drops down to three .
the number of bounded regions of the complement of the four curves in @xmath271 is seven . by proposition
[ bounded ] , this is a lower bound on the ml degree when all @xmath62 are positive .
this example shows that , for specific _ negative _ values of the @xmath62 s , the number of critical points may be smaller than this lower bound .
a well - known result in the theory of hyperplane arrangements @xcite states that the number of bounded regions of a real arrangement equals the euler characteristic of the complement of its complexification .
the euler characteristic is @xmath473 where @xmath474 is the mbius number of the intersection lattice , and , by theorem [ thm - linear ] , this is precisely the ml degree of the associated linear model .
here we extend this relationship between topology and the ml degree to statistical models which are given by nonlinear polynomials @xmath21 . working in the general
setting of section 2 , we shall prove the following : [ euler ] let @xmath30 be a smooth complete algebraic variety over @xmath172 of dimension @xmath17 , and let @xmath34 be the reduced divisor associated to @xmath267 .
assume that the hypotheses ( a ) , ( b ) and ( c ) below hold .
then the ml degree equals @xmath475 where @xmath476 is the topological euler characteristic . invoking hironaka s theorem on resolution of singularities
, we fix a blow up @xmath477 such that @xmath478 is smooth , and the rational function @xmath35 pulls back to a proper morphism @xmath479 onto @xmath480 .
the three * hypotheses * are as follows : 1 . the inverse image @xmath481 of the divisor @xmath34 can be written as @xmath482 where @xmath483 is the support of the divisor @xmath484 of the pullback @xmath479 of the rational function @xmath35 , while @xmath485 is the horizontal divisor consisting of the sum of all components of @xmath375 which map onto @xmath486 .
the restriction of @xmath479 to @xmath487 is a topological fiber bundle over @xmath488 .
the number of critical points of @xmath479 on @xmath489 is finite .
[ wowo ] hypothesis ( a ) is crucial and depends on the exponents @xmath62 .
for instance , consider the rational function @xmath490 on @xmath491 .
we get @xmath478 by blowing up the origin twice .
the exceptional curve of the first blow - up belongs to the fiber @xmath492 and hence is not supported on @xmath484 .
hypothesis ( a ) is not satisfied for this example .
if we take instead @xmath493 then hypothesis ( a ) is satisfied because the exceptional curve belongs to the fiber @xmath494 and is hence supported on @xmath484 , hypothesis ( b ) implies that the cohomology ranks of @xmath487 can be computed from that of @xmath495 and the fibers using knneth s formula .
the alternating sum of the ranks is zero for the fibers , and we get @xmath496 .
in fact , the hypothesis ( b ) could be replaced by the more general condition @xmath496 .
any proper map @xmath497 for @xmath30 smooth is a topological fiber bundle if it is a submersion , i.e. @xmath498 for all points in @xmath30 .
therefore to check hypothesis ( b ) , we need only find a controlled stratification ( in the sense of thom - mather theory @xcite ) of @xmath485 into locally closed smooth sets such that for all points on each strata @xmath499 , @xmath500 .
in particular , this last condition will imply that the critical points of @xmath479 on @xmath489 are the same as the critical points of @xmath35 on @xmath41 .
our method follows closely the proof of the lefschetz hyperplane section theorem ( cf .
moreover , since the complement is not necessarily compact we shall use borel - moore homology @xcite ( see also @xcite ) .
we note that for compact spaces @xmath30 the ordinary homology groups coincide with the borel - moore homology groups . in the borel - moore homology theory
we have the following useful exact sequence to be used below : if @xmath30 is locally compact , @xmath156 is closed in @xmath30 , and @xmath501 , we have @xmath502 thus in this situation the borel - moore euler characteristic is additive : @xmath503 finally , if @xmath504 is an even - dimensional orientable manifold then poincar duality holds between borel moore homology and ordinary cohomology , and @xmath505 coincides with the topological euler number @xmath506 . in our situation
we get @xmath507 the last equation follows from ( [ additive ] ) . hypothesis ( b )
implies @xmath508 ( see remark [ wowo ] ) , and hence it suffices to show that the ml degree equals @xmath509 in other words , we may now simply erase the tilde and consider the case when @xmath30 is smooth and @xmath35 defines a proper morphism @xmath510 .
let @xmath511 denote the set of critical points of @xmath35 on @xmath30 .
by hypothesis ( c ) , this set is finite and the ml degree equals its cardinality counting multiplicities : @xmath512 the multiplicity @xmath513 of a critical point @xmath514 of @xmath35 is known as the _ milnor number _ at @xmath514 of the hypersurface @xmath515 .
milnor @xcite showed that this algebraic invariant of a singularity has the following topological interpretation .
consider a coordinate chart around the point @xmath514 and intersect the fiber @xmath516 with a ball of radius @xmath517 around @xmath514 .
for @xmath518 this intersection is the _
milnor fiber_. milnor @xcite showed that the milnor fiber is homotopy equivalent to a bouquet of @xmath513 spheres of dimension @xmath519 .
each singular fiber is obtained ( cf .
@xcite ) from a smooth fiber by replacing the milnor fiber by a contractible set .
the borel - moore exact sequence ( [ bmsequence ] ) implies that the euler number of a singular fiber @xmath376 is obtained from the euler number of a smooth fiber @xmath156 by adding @xmath520 .
then the euler number of the union of the singular fibers equals @xmath521 times the euler number of a smooth fiber @xmath156 plus the correction @xmath522 . applying knneth s formula to the fiber bundle defined by @xmath35 on @xmath41 minus the union of the singular fibers , and then applying the additivity formula ( [ additive ] )
, we conclude that @xmath523 as desired .
@xmath524 for another illustration consider example [ 4 curves ] with @xmath491 .
the generic ml degree was @xmath43 but it decreased by @xmath430 when @xmath525 is a general solution of @xmath526 .
this is consistent with theorem [ euler ] , because for such @xmath12 the divisor @xmath34 loses the component at infinity .
the difference is a projective line minus @xmath527 points , which has euler number @xmath528 .
consider our hypotheses when @xmath529 is the blow - up of @xmath30 at the origin .
if @xmath530 then the exceptional curve is part of @xmath531 and theorem [ euler ] is valid . on other hand , if @xmath532 then it maps to @xmath533 under a rational map of degree @xmath534 , so hypothesis ( a ) does not hold .
the philosophy of this example is that , even if the divisor @xmath34 is locally biholomorphic to an arrangement of hyperplanes , genericity of the exponents @xmath62 may be necessary for the topological formula of theorem [ euler ] to hold .
theorems [ thm1 ] , [ toricformula3 ] and [ toricformula4 ] offer combinatorial formulae for the ml degree and hence ( using theorem [ euler ] ) for the euler number of the complement of an arrangement of generic hypersurfaces @xmath535 in @xmath536 or in a toric manifold . in each of these theorems ,
the combinatorial number becomes an upper bound for the ml degree when the coefficients of the @xmath21 s are special .
this semi - continuity principle will be explained by the following general topological result , in which also the underlying manifold is allowed to vary .
[ semiconti ] assume we are given a one - parameter smooth proper family @xmath537 of complex manifolds over the unit disk @xmath538 , and a family of rational functions @xmath539 on @xmath537 , such that 1 .
for @xmath540 the divisor @xmath541 defined by @xmath539 has gnc , and 2 . for @xmath542 the divisor @xmath543 defined by @xmath544 has the same homology class as @xmath545 for the natural differentiable trivialization of the family @xmath537 .
then the ml degree of @xmath544 is less than or equal to the ml degree of @xmath539 . in order to understand the second hypothesis ,
let us recall ehresmann s theorem @xcite : any proper submersion @xmath546 of differentiable manifolds is a differentiable fiber bundle , i.e. , if @xmath504 is a sufficiently small open set in @xmath547 , there is a local diffeomorphism between @xmath548 and @xmath549 , for a fiber @xmath156 , and this diffeomorphism is compatible with the two projections to @xmath504 .
_ sketch of proof .
_ let @xmath550 be the total space of the family @xmath551 .
we consider the function @xmath552 , given by @xmath553 .
consider the locus @xmath554 given by the vanishing of the vertical differential of @xmath212 : this is the local complete intersection defined by the @xmath17 partials @xmath555 , where @xmath98 are local coordinates on the fibers provided by the implicit function theorem . at each critical point @xmath514 of @xmath544 ,
the locus @xmath556 has dimension @xmath33 , and thus , in a neighborhood of @xmath514 , the morphism @xmath557 is finite , whence its degree is locally constant .
this establishes the desired semi - continuity . in general
, it might be difficult to show that a given rational function @xmath544 has a perturbation as above , or we might want to calculate the ml degree with more algebraic precision .
the results in the next section may help .
in this section we will show that the formula for the ml degree given by the logarithmic chern number ( theorem [ chernclassthm ] ) holds in greater generality .
returning to the setting of chapter 2 , we consider logarithmic vector fields along @xmath34 .
again , our definition differs slightly from the one given by saito @xcite .
recall that the _ tangent sheaf _
@xmath560 is @xmath561 , the dual of the @xmath33-forms on @xmath30 .
the inclusion of @xmath95 into @xmath75 , studied in lemma [ exactlem ] , dualizes to an inclusion of the logarithmic vector fields into the tangent sheaf .
[ moresheaves ] we have the following exact sequence of sheaves on @xmath30 : @xmath562 if @xmath30 is smooth , then the rightmost homomorphism is onto , and the total chern class of the sheaf of logarithmic vector fields equals @xmath563 dualizing the sequence @xmath564 , we get @xmath565 hence the @xmath566-sequence gotten by dualizing the sequence ( [ lemmasequence ] ) has the form @xmath567 this is the first statement of proposition [ moresheaves ] . if @xmath30 is smooth then the cotangent sheaf @xmath95 is free , and we have @xmath568 .
the formula ( [ chernoftheta ] ) follows from the multiplicativity of the total chern class .
[ logvecchara ] the two leftmost maps in the exact sequence of proposition [ moresheaves ] characterize the logarithmic vector fields on @xmath30 along @xmath34 as those vector fields @xmath569 which satisfy @xmath570 for all @xmath214 .
in other words , for each @xmath571 , the vector field @xmath572 has the property that there exist functions @xmath90 such that @xmath573 .
an interesting case will be the one where @xmath30 is smooth and the sheaf @xmath574 has zero - dimensional support and length @xmath514 . here the top chern class of @xmath558 and the top chern class of @xmath75 differ by @xmath575 , and the count in part 3 of theorem [ chernclassthm ] changes accordingly . as in the proof of theorem [ chernclassthm ] , set @xmath576 and let @xmath577 be the subscheme of @xmath30 defined by the vanishing of @xmath123 .
the restriction of @xmath578 to the open set @xmath579 is the critical locus of @xmath35 .
the reason why the sheaf @xmath558 is important is that it enters directly into the algebraic description of @xmath578 .
indeed , the section @xmath123 of @xmath580 corresponds to an exact sequence @xmath581 where @xmath582 is the kernel of @xmath583 .
hence the @xmath566-sequence gotten by dualizing the previous sequence has the form @xmath584 this leads to the following more refined formula for the ml degree .
[ thetatheorem ] assume that @xmath30 is smooth , @xmath34 meets every curve , the sheaf @xmath558 is locally free , and @xmath125 does not intersect the divisor @xmath34 .
then the number of critical points of @xmath35 on @xmath585 equals @xmath586 .
the sequence ( [ zsequence ] ) gives the following description of the equations defining the critical points .
the ideal sheaf of the subscheme @xmath578 is generated by the functions @xmath587 , where @xmath588 varies among the logarithmic vector fields and @xmath589 is derived from @xmath588 as in remark [ logvecchara ] .
this holds because the section @xmath590 factors through the homomorphism @xmath591 , thus the homomorphism dual to the section @xmath123 also factors through the dual homomorphism @xmath592 .
let @xmath595 with @xmath596 and @xmath597 .
a vector field @xmath598 is logarithmic if and only if @xmath599 and @xmath600 , for some function @xmath601 .
observe that @xmath602 .
we conclude that the sheaf @xmath558 is locally free of rank @xmath603 .
let @xmath608 with @xmath609 and @xmath610 .
logarithmic vector fields @xmath588 have the form @xmath611 with @xmath612 divisible by @xmath613 .
this implies @xmath614 , @xmath615 , and @xmath616 , for any functions @xmath617 .
thus @xmath558 is locally free of rank two .
the origin does not belong to the subscheme @xmath125 if and only if @xmath618 .
* acknowledgements : * this project grew out of discussions with lior pachter at the december 2003 workshop on _ computational algebraic statistics _ at the american institute for mathematics ( aim ) at palo alto .
many thanks to both lior and aim for getting us started , and many thanks to msri berkeley and the organizers of the spring 2004 msri program _
topology of real algebraic varieties _ , which gave the four of us the opportunity to collaborate .
we also thank oleg viro for letting us include theorem [ thm : viro ] .
amit khetan was supported by an nsf postdoctoral fellowship ( dms-0303292 ) .
bernd sturmfels was supported by the hewlett packard visiting research professorship 2003/2004 at msri berkeley and in part by the nsf ( dms-0200729 ) .
b. chor , a. khetan and s. snir .
maximum likelihood on four taxa phylogenetic trees : analytic solutions , _ the @xmath619th annual conference on research in computational molecular biology recomb 2003 _ , berlin , april 2003 , pp . | maximum likelihood estimation in statistics leads to the problem of maximizing a product of powers of polynomials .
we study the algebraic degree of the critical equations of this optimization problem .
this degree is related to the number of bounded regions in the corresponding arrangement of hypersurfaces , and to the euler characteristic of the complexified complement . under suitable hypotheses
, the maximum likelihood degree equals the top chern class of a sheaf of logarithmic differential forms .
exact formulae in terms of degrees and newton polytopes are given for polynomials with generic coefficients . |
the purpose of this paper is to assess the possibility of detecting certain tetraquarks in present and future facilities . among many possible tetraquarks , the double charm tetraquark @xmath1 with quantum numbers
@xmath2 is particularly interesting since it is very sensitive to the chosen effective interaction : * it is very delicate , it is either weakly bound or slightly unbound with respect to the two - body hadronic decay @xmath3 .
* its structure can be either predominantly `` molecular '' or predominantly `` atomic '' with consequences for the production and decay .
double charm tetraquarks were intensively studied by many authors .
various approaches were applied , from lattice qcd and chiral heavy quark effective theories to nonrelativistic potential models .
it was shown that , although the predictions of these theories agree in the barion and meson sector , they give dramatically different results for tetraquarks . for this reason , the double charm tetraquarks present an important laboratory for discriminating between different hadronic models .
moreover , our estimates for the production cross section of such states gives us some hope that they can be experimentally detected in near future .
the most important ingredient in the production of the @xmath4 tetraquark is double charm production .
experimental data for such events are very puzzling .
the production of prompt @xmath5 at @xmath6 factories as well as the production of @xmath7 at selex are much larger than expected .
therefore the comparison of @xmath4 production with @xmath8 at b factories and the comparison @xmath4 production with @xmath7 at selex could shed some new light on the mechanism responsible for such large double charm production . in section 2
we presents our results for the @xmath4 production at high energy colliders were we believe that the dominant mechanism for the initial double charm production is double gluon fusion . from experimental data we also estimate phenomenologically the production of the @xmath4 tetraquark at b factories and at selex .
the results of detailed four body calculations in nonrelativistic constituent quark model ( section 3 ) encourage us to further investigate this state .
since we found the @xmath4 tetraquark to be weakly bound , we propose ( section 4 ) the branching ratio between hadronic and radiative decays as the most promising mechanism for the detection of these states .
the most promising mechanism for the production of the @xmath4 tetraquark is the formation of the @xmath9 diquark followed by hadronisation into @xmath10 .
an alternative mechanism would exploit binding of @xmath11 and @xmath12 mesons if when they are produced with small relative momenta .
one might expect that the latter mechanism could drastically enlarge the production rate if the dominant configuration is molecular .
due to the very messy environment in hadron colliders , however , such a weakly bound system would too soon dissociate into free mesons by the interaction with surrounding partons of initial hadrons .
the first step is to create two @xmath13 pairs with the @xmath14 quarks close in the phase space and in colour antisymmetric state , so that in the second step they bind into the @xmath15 diquark .
the binding energy of such a system is @xmath16 200 mev @xcite . in the third step the diquark
gets dressed either with a light @xmath17 or @xmath18 quark into a @xmath19 or @xmath20 baryon or with a light @xmath21 antidiquark into the @xmath10 tetraquark .
the probabilities for this two types of dressing can be estimated using the analogy of a single heavy quark fragmentation .
the branching ratio of the @xmath22 and @xmath23 production at the fermilab and at lep experiment is 0.9 and 0.1 @xcite , respectively , therefore we expect the same ratio in the hadronisation @xmath24 and @xmath25 , respectively .
the double charmed baryons were probably detected at selex @xcite .
it was estimated that @xmath26 of the singly charmed baryons they see result from the decay of doubly charmed baryons .
the most probable mechanism for the double charm production at selex is production of the single @xmath13 pair in the processes @xmath27 or @xmath28 while the second @xmath13 pair is created in the fragmentation of the heavy quark @xmath29 .
however , theoretically it is still unclear why the selex has such a large cross section for double charm production .
since selex is a fixed target experiment the @xmath15 diquark is most likely to be produced with high lab momenta which might be helpful in the detection as discussed in @xcite .
but since selex found , with their cuts , only about fifty candidates for double charmed baryons , the statistics for detecting double charmed tetraquarks should be improved .
next , we look at the production and detection of the @xmath4 tetraquark in b - factories . since the total mass of four @xmath30 mesons is close to the c.m .
energy , the @xmath31 quarks created in this process have small relative momenta which is very important in @xmath4 production .
this feature also ensures a smaller number of additional pions created in the @xmath32 annihilation and thus a cleaner reconstruction of @xmath4 .
, @xcite has reported a measurement of prompt @xmath5 production in @xmath33 annihilation at @xmath34 = 10.6 gev and found that the most of the observed @xmath5 production is due to the double @xmath13 production @xmath35 which correspond to @xcite , @xcite @xmath36 or to about 2000 events from their 46.2 fb@xmath37 data sample . the theoretical nonrelativistic qcd prediction for this process
is an order of magnitude smaller @xcite , @xcite , @xcite , @xcite , so this process is still not well understood @xcite . but
it is very likely that the analogous mechanism would also enlarge the cross section for the prompt production of @xmath15 diquark and thus the @xmath7 baryon and @xmath4 tetraquark .
the @xmath15 diquark production cross section can then be estimated to be @xmath16 0.15 pb , which correspond to @xmath16 @xmath38 @xmath7 @xcite and about @xmath16 @xmath39 @xmath4 per year .
we now present our calculation for the @xmath4 production at high energy colliders .
the two colliding nucleons in tev machines can be considered as two packages of virtual gluons whose number is huge for low bjorken-@xmath40 .
therefore we expect @xcite , @xcite that in these facilities the dominant mechanism for double charm production would be a double gluon - gluon fusion : @xmath41 .
the usual hard production mechanism is heavy quark production followed by fragmentation , however this mechanism does not include all the possible feynman diagrams . in ref .
@xcite , @xcite it has been shown that at high energy collider a significant rate of events with double heavy quark pairs is expected . to compute all the fourth order @xmath42 feynman diagrams we have to consider the two different mechanisms leading to the same final state : the usual single parton scatterings and the double parton scatterings .
at high energy in single parton interactions the partonic sub - process is dominated by the gluon - gluon fusion @xmath43 while the double parton interactions are dominated by the process @xmath41 where the two distinct interactions occur in the same hadronic event .
we give an estimate of the production cross section at high energy in the region of small transverse momenta where the multiple parton interactions provide the leading contribution to the cross section @xcite , @xcite .
we compute the production cross section of two @xmath31-quarks , @xmath44 , very close in momentum space @xmath45,@xmath46 @xmath47 , as a function of @xmath48 .
we consider the heavy quark production in the kinematical range of the lhcb @xmath49 , and for completeness for the alice @xmath50 , tevatron @xmath51 and rhic @xmath52 experiments ; in the last case we calculate also the production cross sections in proton - nucleus interactions @xcite .
the results are shown in fig [ sl0 ] one can notice that the cross section @xmath53 at small @xmath48 is almost uniform and then it is approximately proportional to the momentum volume @xmath54
. in the second step , the two @xmath14 quarks join into a diquark .
we assume simultaneous production of two independent c quarks with momenta @xmath55 .
since they appear wherever within the nucleon volume , we modulate their wave functions with a gaussian profile with the `` oscillator parameter '' @xmath56 fm corresponding to the nucleon rms radius @xmath57 where the normalisation factor @xmath58 here @xmath59 is the average distance between two nucleons in the target nucleus for proton - nucleus experiment at rhic and we use the value @xmath60 fm or zero otherwise .
we make an impulse approximation that this two - quark state is instantaneously transformed in any of the eigenstates of the two - quark hamiltonian .
then the amplitude of the diquark formation @xmath61 is equal to the overlap between the two free quarks and the diquark with the same centre - of - mass motion . by approximating the diquark wave function with a gaussian with the oscillator parameter @xmath62 fm we get @xmath63 for the production cross section we take into account that @xmath64 is practically constant and can be taken out of the integral @xmath65 where factors in front of the integral are due to the projection on the colour and spin triplet states .
if we insert the values of @xmath66 obtained from the fig [ sl0 ] , we get @xmath67 nb and @xmath68 nb for lhcb and alice at lhc , @xmath69 nb at tevatron and @xmath70 nb and @xmath71 nb at rhic for proton - proton and proton - nucleus interaction , respectively .
the last step of the @xmath4 production is dressing of the heavy diquark .
it either acquires one light quark to become the doubly - heavy baryon ccu , ccd or ccs , or two light antiquarks to become a tetraquark . with this
we neglect the possible dissociation of the heavy diquark into a @xmath72 pair so the results are the upper estimate for the real @xmath4 production . assuming that the probability for dressing the @xmath15 diquark into the @xmath73 tetraquark is 0.1 @xcite , as pointed out at the beginning of the section , yields the production rate of the dimeson 20900 , 9700 , 600 and 1 events / hour for lhc at luminosity @xmath74 @xmath75s@xmath37 , tevatron at luminosity @xmath76 @xmath75s@xmath37 and rhic at @xmath77 luminosity @xmath78 @xmath75s@xmath37 , respectively .
the structure of the @xmath79 tetraquark has been studied in ref . @xcite .
we summarize here those features which are particularly relevant for the detection .
there are two extreme spatial configurations of quarks in a tetraquark .
the first configuration which we call _ atomic _ is similar to @xmath80 , with a compact @xmath9 diquark instead of @xmath81 , around which the two light antiquarks are moving in a similar manner as in the @xmath82 baryon . the second configuration which we call _ molecular _ resembles deuteron ,
the two heavy quarks are well separated and the two light antiquarks are bound to them as if we had two almost free mesons .
the atomic configuration is more likely to appear in strongly bound tetraquarks while the molecular configuration can be expected in weakly bound systems .
we present results using two different one - gluon exchange potentials .
the bhaduri potential @xcite quite successfully describes the spectroscopy of the meson , as well as baryon ground states .
this is an important condition since in the tetraquarks we have both quark - quark and quark - antiquark interactions .
the al1 potential @xcite slightly improves the meson spectra by introducing a mass - dependent smearing of the colour - magnetic term .
we expand tetraquark wave function with gaussinas of three sets of jaccobi coordinates . in this basis
we were able to reconstruct the wave functions of deeply bound tetraquarks as well as of two free mesons - the threshold state .
this is important if one is searching for weakly bound tetraquarks with molecular structure .
we found that the @xmath4 is weakly bound for both the bhaduri and al1 potential in contrast to the results of calculations in harmonic oscillator basis @xcite where asymptotic channel can not be accommodated as shown in table [ tabela1 ] .
.column 1 : type of potential , column 2 : lowest meson - meson threshold for a given potential in mev , column 3 : our results in mev , column 4 : results in mev of ref.@xcite , column 5 : mean distance in fm between two heavy quarks @xmath83 . [ cols="^,^,^,^,^",options="header " , ] in fig .
[ t1 ] we present the probability densities @xmath84 for finding ( anti)quark @xmath85 and ( anti)quark @xmath86 at the interquark distance @xmath87 and the ratio of the projections on colour sextet state @xmath88 and colour triplet state @xmath89 where e.g. @xmath90 here particles 1 and 2 are the two heavy quarks @xmath14 and particles 3 and 4 the light antiquarks @xmath91 and @xmath92 .
the wave function between heavy quarks is broad and has an exponential tail @xmath93 at large distances where @xmath94 , @xmath95 is the binding energy of the system and @xmath96 the reduced mass of the @xmath11 and @xmath12 mesons . at small distances
the dominant colour configuration is @xmath97 .
here we have a diquark - antidiquark structure and this region present about a third of the total probability while for @xmath98 fm sextet colour configuration become larger .
the ratio of these two configurations stabilise at 2 , since here we have a molecular structure of the two colour singlet mesons which has in diquark antidiquark basis @xmath99 colour decomposition , while octet configuration @xmath100 is negligible . , of the two light antiquarks @xmath101 and of a light antiquark and a heavy quark @xmath102 in @xmath4 as a function of the interquark distance .
the ratio of the projection on colour sextet and colour triplet configurations is also shown . ]
now we show that additional weak three - body interaction can transform the molecular structure of the @xmath4 tetraquark into atomic . for the radial part
we take the simplest possible radial dependence the smeared delta function of the coordinates of the three interacting particles @xcite .
the colour factor in the two - body bhaduri or al1 potential is proportional to the first ( quadratic ) casimir operator c@xmath103 ; c@xmath104 .
it is then natural that we introduce in the three - body potential the second c@xmath105 ( cubic ) casimir operator c@xmath106 .
a deeper discussion of the properties that the colour dependent three - body interaction must fulfil can be found in @xcite . in the baryon sector
the three - body interaction was used to better reproduce the baryon ground state spectroscopy @xcite .
a colour structure is there irrelevant since there is only one colour singlet state and thus the colour factor is just a constant which can be included into the strength of the potential . in tetraquarks
the situation is different since there are two colour singlet states : @xmath107 and @xmath108 ( or @xmath109 and @xmath110 after recoupling ) .
the three - body force operates differently on these two states and one can anticipate that in the case of the weak binding it can produce large changes in the structure of the tetraquark .
this can not be otherwise produced simply by reparameterization of the two - body potential , so the weakly bound tetraquarks are a very important laboratory for studying the effect of such an interaction .
a drastic change in the width of the probability density can already be seen for strength @xmath111 = -20 mev ( fig .
[ sl3body ] ) , where the binding energy of the tetraquark becomes - 14 mev for bhaduri potential . here
@xmath4 loses the molecular structure , the triplet - triplet colour configurations become dominant and the @xmath4 tetraquark becomes similar to @xmath112 . in the baryon sector
such an interaction would merely lower the states by about @xmath111 so it would have no dramatic effect nor would it spoil the fit to experimental data . since the predicted energies of ground state baryons for the bhaduri and al1 potential are above the experimental values , this is actually a desirable feature . in the @xmath4 tetraquark as a function of interquark distance for three different values of the strength of the three - body potential . ]
in order to identify a weakly bound @xmath4 tetraquark we have to distinguish the pion or photon emitted by the @xmath113 meson bound inside the tetraquark from the one resulting from free @xmath12 meson decay .
we can exploit the fact that the phase space for @xmath114 decay is very small .
this has a strong impact on the branching ratio between radiative and hadronic decay .
since the @xmath113 meson inside the tetraquark with molecular structure is not significantly influenced by the other d meson in the tetraquark , we expect that the partial width for the magnetic dipole m1 transition would be very close to the width of the free meson while the width for hadronic @xmath114 decay will decrease with stronger binding and will become energetically forbidden below the @xmath115 threshold .
the hadronic decay of the @xmath4 tetraquark is a three - body decay which is commonly represented by the dalitz plot . . ]
tetraquark obtained with integration of the dalitz plot @xmath116 where @xmath117 is the mass of the tetraquark .
the width at @xmath118 is normalized to be equal to the width of the free @xmath12 meson decay . ]
if the @xmath4 tetraquark is below the @xmath119 threshold but above the @xmath120 and @xmath121 , as was the case in our nonrelativistic potential models , the partial decay rate for the @xmath122d+d+@xmath123 is given by @xmath124 where particles 1 and 2 are two final @xmath11 mesons and particle 3 is a @xmath123 emerging from the decaying tetraquark . here
@xmath125 and @xmath126 and m is the mass of the tetraquark . since the total masses of the @xmath127 and @xmath128 are so close there is a strong isospin violation in the decay which can not be reproduced with the bhaduri or al1 potential where the @xmath113 and the @xmath11 isospin doublets are degenerate .
we shall not try to modify the interaction to accommodate the dependence of the decay on the isospin of the particles , but we shall rather work with the experimental masses taken from the pdg @xcite where we see that @xmath129 the allowed region of integration over @xmath130 and @xmath131 for three different binding energies is plotted in fig .
if we assume @xmath132 is constant , which is very plausible in our case , the allowed region will be uniformly populated with experimental events so that the measured partial decay rate @xmath133 will be proportional to the kinematically allowed area from fig .
this is shown in fig .
[ slb ] , where we assumed that for the molecular state , the width of the @xmath4 tetraquark with zero binding energy would be the same as the width of the free @xmath12 meson .
let us now consider also the posibility that @xmath4 is not a bound @xmath134 state but a resonant state above the @xmath3 threshold .
then if the resonance is situated near the threshold , there will be a significant fraction of hadronic @xmath135 decays beside the @xmath136 decay .
this region of positive binding energy is also presented in fig .
[ sla ] and fig .
[ slb ] .
in order to estimate the decay width we make a comparison with charmonium .
the charmonium state @xmath137 has the width of @xmath138 mev and is 36 mev above @xmath139 threshold , which is also the dominant decay mode .
let us assume , that the @xmath4 tetraquark resonant state , which would be 36 mev above the @xmath119 threshold would have the same partial width for the decay into @xmath11 and @xmath12 meson the area of the integrated dalitz plot for the binding energy @xmath140 mev is then 37 times larger then at the threshold @xmath141 . since the experimental width for @xmath142 is @xmath143 kev @xcite we can estimate that the decay width for the @xmath144 three - body decay would be @xmath145 mev .
so we expect about 15@xmath146 direct @xmath147 decays . the dalitz plot would not be uniformly populated but there will be a strong band where @xmath148 reflecting the appearance of the @xmath149 decay chain . in this estimation
we have neglected the interference between these two decay mechanism , since the width of the @xmath12 is three orders of magnitude smaller then the width of the tetraquark .
we have shown that the @xmath79 tetraquark production is comparable to double charm baryon production ( possibly 10% ) . therefore they may be seen in selex if statistics is improved .
similarly , it is comparable to prompt @xmath150 production which is reasonably abundant in @xmath6-factories . in high energy colliders
we may expect an optimistic number of events due to double @xmath151 production via double two - gluon fusion ( see sect.2 ) .
therefore time has come to start the hunt ! regarding the detection of the @xmath152 tetraquark we propose a nice opportunity the very small phase space of the @xmath142 decay which is very sensitive to the binding energy of @xmath12 to @xmath30 .
one possibility would be to measure the branching ratio between the pionic and gamma decay of @xmath153 .
janc d. , rosina m. : few - body systems * 31 * , 1 ( 2001 ) gelman b. a. , nussinov s. : phys .
b551 * , 296 ( 2003 ) cho p. , leibovich k. : phys . rev . *
d54 * , 6990 ( 1996 ) .
yuan f. , qiao c.f .
, chao k.t .
: phys . rev .
* d56 * , 321 ( 1997 ) ; ibid , 1663 ( 1997 ) .
baek s. , ko p. , lee j. , song h.s . :
j. korean phys . soc . *
33 * , 97 ( 1998 ) .
kiselev v.v .
b332 * , 411 ( 1994 ) .
ioffe g.l . ,
kharzeev d.e .
: phys . rev .
* d69 * , 0140016 ( 2004 ) .
berezhnoy a.v .
, likhoded a.k .
: phys . atom .
nucl . * 67 * , 757 ( 2004 ) .
del fabbro a.,treleani d. : phys . rev . * d61 * , 077502 ( 2000 ) .
del fabbro a.,treleani d. : phys . rev . * d63 * , 057901 ( 2001 ) .
mattson m. et al .
( selex collaboration ) : phys .
lett . * 89 * , 112001 ( 2002 ) affolder t. et al .
( cdf collaboration ) : phys .
lett . * 84 * , 1663 ( 2000 ) abe k. et al .
( belle collaboration ) : phys .
* 88 * , 052001 ( 2002 ) abe k. et al .
( belle collaboration ) : phys .
lett . * 89 * , 142001 ( 2002 ) trnqvist
n. a. : phys .
lett . * 67 * , 556 ( 1991 ) .
trnqvist n. a. : nuovo cim .
* a107 * , 2471 ( 1994 ) .
manohar a. v. , wise m. b. : nucl.phys .
* b399 * , 17 ( 1993 ) .
bhaduri r. k. , cohler l. e. , nogami y. : nuovo cim . * a65 * , 376 ( 1981 ) silvestre - brac b. : few - body systems * 20 * , 1 ( 1996 ) janc d. , rosina m. : hep - ph/0405208 silvestre - brac b. , semay c. : z. phys . *
c57 * , 273 ( 1993 ) dmitrasinovic v. : phys . lett . *
b499 * , 135 ( 2001 ) dmitrasinovic v. : phys . rev . * d67 * , 114007 ( 2003 ) pepin s. , stancu fl .
: phys . rew . *
d65 * , 054032 ( 2002 ) hagiwara k. et al .
( particle data group ) : phys .
* d66 * 010001 ( 2002 ) janc d. , rosina m. : hep - ph/0405208 | the feasibility of tetraquark detection is studied . for the @xmath0 tetraquark
we show that in present ( selex , tevatron , rhic ) and future facilities ( lhcb , alice ) the production rate is promising and we propose some detectable decay channels . |
the filamentous fungi are the most diverse of any eukaryotic organisms , thriving as mutualists , decomposers and pathogens the world over .
one suspected contributor to their tremendous ecological success is their unusual mode of life .
unlike plant and animal cells , the cells of filamentous fungi are generally multinucleate , and can even harbor genetically different nuclei , bathed by a common cytoplasm . as the tube - like hyphae grow , each extending incrementally at its tips , nuclei flow from the colony interior to fill the free space created at the tips .
we show that the flow is driven by pressure gradients across the colony , and that nuclei follow complex multi - directional trajectories , reminiscent of cars traveling through a city .
we hypothesize that the complexity of nuclear paths is a deliberate effort by the fungus to keep genetically different nuclei well mixed . disrupting the exquisite hydraulic engineering of the cell
( e.g. by knocking out the ability of hyphae to fuse , to make the multi - connected network ) causes genetically different nuclei to become un - mixed during growth .
our fluid dynamics video includes 11 short segments : 1 .
a time lapse sequence , accelerated 7500 fold showing a colony invading a small block of agar .
2 . confocal imaging of hh1-gfp transformed nuclei flowing toward a growing tip .
as we pan deeper into the colony , we see how these tip hyphae are fed as branches of trunk hyphae , each supplying nuclei to many tips .
the flow speed in a tip hyphae is @xmath0 0.1 @xmath1 m/s , flow speed in trunk hyphae can be 30 or 80 times greater .
3 . in the colony interior , nuclei
flowing through the complex network of hyphae follow torturous and even multidirectional paths , looking in confocal microscopy , like the headlights of cars navigating a microscopic city .
flow speeds in the colony interior can reach 10 - 20 @xmath1 m/s .
4 . just like cars in a city , speeds vary between hyphal roads .
some roads remain grid - locked while in others nuclei flow at speeds of up to 10 - 20 @xmath1 m/s .
5 . under some conditions , in small colonies ,
the nuclei form spontaneous traffic jams .
6 . what drives the fluid flow ?
we measured the variation in nuclear speeds across hyphae .
when collapsed by hyphal diameter and axial flow speed , we see a common poiseuille - flow profile in each hypha , indicating that the flows are hydrodynamic : driven by pressure gradients across the colony . 7 .
we can manipulate the pressure gradients by applying hyper - osmotic solutions to the colony .
these treatments apply a uniform pressure gradient counter to the normal direction of flow .
flow in each fungal hyphae is transiently reversed , indicating that the hyphal network architecture far from being random
is finely tuned to create mixing flows from spatially coarse pressure gradients .
what is the function of these flows ?
since real fungi can harbor many genetically different nucleotypes , we hypothesize that physical mixing , associated with the flows , helps to preserve nuclear diversity as the fungus grows .
we show a sequence of nuclear mixing in a chimeric colony formed by fusing two _ n. crassa _ colonies ; one with hh1-dsred and one with hh1-gfp transformed nuclei .
we can manipulate the hyphal architecture genetically .
we show micrographs of conidiophores ( spore - bearing hyphae ) of chimeric wild type and _ so _ mutant colonies .
both colonies have dsred ( red ) and gfp ( green ) expressing nucleotypes . in the wild type colony ,
the red and green nuclei remain well mixed in the colony . in the _ so _ colony
they segregate out , so that conidiophores end up almost exclusively red or exclusively green ( shown ) .
_ so _ mutations stop the hyphae from fusing to create an interconnected network . 10 .
conidiophores are branched structures ( without interconnections ) so pose particular challenges for mixing .
taylor dispersion ( enhanced diffusion due to fluid shear across a hypha ) can keep nuclei well - mixed but only if flow speeds in the conidiophore exceed 200 @xmath1 m/s . to achieve such fast flows each conidiophore must feed into at least 2000 hyphal tips .
11 . a time - lapse video showing the growth and collapse of conidiophores , accelerated 7500 fold . 2000 separately growing hyphae places impossible weight upon the conidiophore , causing it to collapse .
since spores are thought to be dispersed by air - flows across the colony and collapsed conidiophores are likely to experience reduced wind speeds , rates of spore liberation are reduced .
loss of spore dispersal effectiveness represents the high price paid by the colony for maintaining mixing flows . | the syncytial cells of a filamentous fungus consist of a mass of growing , tube - like hyphae .
each extending tip is fed by a continuous flow of nuclei from the colony interior , pushed by a gradient in turgor pressure .
the myco - fluidic flows of nuclei are complex and multidirectional , like traffic in a city .
we map out the flows in a strain of the model filamentous fungus _
n. crassa _ that has been transformed so that nuclei express either hh1-dsred ( a red fluorescent nuclear protein ) or hh1-gfp ( a green - fluorescent protein ) and report our results in a fluid dynamics video . |
for over half a century , one of the major puzzles in particle physics has been the question of why quark and lepton families replicate .
although we have accumulated a wealth of data concerning the masses and mixings of quarks and leptons , we still appear to be far from a true understanding of family structure . in the absence of a concrete model to consider ,
it is natural to postulate the existence of some family symmetry @xcite that plays a role in determining the observed particle spectrum .
once we consider such a family symmetry , we face a plethora of options . the symmetry may be ( 1 ) discrete , ( 2 ) continuous and local , or ( 3 ) continuous and global . within each of these categories
, one may choose any of a number of symmetry groups , and the overall family symmetry may even be a combination of the three possibilities . of course , any exact family symmetry of the underlying theory must be spontaneously broken at some energy scale since we know that the quark and lepton masses are very different from one family to the next . for option ( 1 ) , spontaneously broken discrete symmetries , domain walls are the only model - independent predictions , and these can not be studied in particle physics laboratories . in case ( 2 ) , the masses of the family gauge bosons of spontaneously broken local continuous symmetries can be constrained , _
e.g. _ , from @xmath10-@xmath11 mixing @xcite . from a phenomenological point of view
, however , possibility ( 3 ) is particularly enticing , as it implies the existence of massless nambu
goldstone bosons , called `` familons , '' from the spontaneously broken family symmetry
. this family symmetry may be either abelian or non - abelian ; nambu
goldstone bosons associated with the spontaneous symmetry breaking of an abelian lepton number symmetry are often called `` majorons.''-parity breaking @xcite . in this paper
, we study a number of probes , many of which are applicable to both abelian and non - abelian symmetries .
we use the generic name `` familon '' to denote the associated nambu goldstone bosons in either case . ]
the existence of new massless particles has many implications in particle physics , astrophysics , and cosmology , and , as we will see , may be probed in a wide variety of experiments .
moreover , the couplings of familons at low energies are determined by the non - linear realization of the family symmetry .
these couplings are , _
e.g. _ , of the form @xmath12 where @xmath13 is the family symmetry breaking scale , _
i.e. _ , the familon decay constant , @xmath14 are the familons , @xmath15 are the generators of the broken symmetry , and the @xmath16 are fermion fields in terms of which the flavor symmetry is defined .
the strength of the familon coupling is therefore inversely proportional to @xmath13 and can be constrained for a given family symmetry group in a model - independent manner .
familon couplings between the first and second generations have been studied extensively and will be reviewed below .
in contrast , however , couplings involving the third generation are largely unexplored , although they may have rather rich phenomenological and cosmological implications @xcite .
current constraints in the lepton sector are relatively weak , with the best bounds coming from @xmath17 bounds @xcite , and there are at present no corresponding bounds reported in the hadronic sector ( see , however , ref .
@xcite ) . at the same time , it is a logical possibility that the familon couples preferentially to the third generation , and models have been proposed in which this is the case @xcite .
it is therefore interesting to explore the possibilities for improving ( or setting ) bounds on familon scales for the third generation , especially in light of the upcoming @xmath8 physics experiments . in this paper
, we will study what we believe to be the most sensitive probes of couplings of familons to the third generation , primarily to @xmath18 leptons and @xmath19 quarks .
we show that dedicated analyses of existent data from cleo , argus , and lep could probe family symmetry breaking scales up to @xmath20 gev and may be significantly improved at future @xmath8 factories .
simply because this is largely unexplored physics , there is a high discovery potential for familons at these facilities .
familon couplings to the third generation are also of interest from a cosmological point of view .
the mass of the @xmath18 neutrino is still allowed to be as large as 18.2 mev experimentally @xcite .
a heavy @xmath18 neutrino has interesting consequences for both big - bang nucleosynthesis ( bbn ) and large scale structure formation @xcite , as will be discussed in sec .
[ sec : cosmology ] . since a heavy neutrino ( @xmath21 ev
, where @xmath22 is the expansion rate of the universe in units of 100 km / sec / mpc ) must decay in order not to overclose the universe , an invisible decay into a lighter neutrino and a massless boson , such as a familon ( or majoron ) , is typically required .
( the three neutrino mode is strongly disfavored and therefore the familon mode is most preferred . )
there is therefore an interesting interplay between experimental searches for familons and scenarios requiring heavy neutrinos , and , as we will see , future collider experiments and analyses may severely constrain a number of such cosmologically motivated scenarios .
this paper is organized as follows .
we begin in sec .
[ sec : familons ] with a discussion of familon interactions .
in particular , we emphasize that the familon interactions of particles in the same gauge multiplet are expected to be comparable . in sec . [
sec : collider ] we consider constraints on familon interactions that may be inferred from current experimental data , concentrating on familon couplings to the third generation .
current bounds on third generation couplings from astrophysical considerations are presented in sec .
[ sec : astrophysics ] .
we then describe some promising prospects for detecting familons in @xmath8 physics at future experiments in sec .
[ sec : bfactories ] . finally , we note some of the interesting cosmological implications in sec . [
sec : cosmology ] and give our conclusions in sec .
[ sec : conclusions ] .
the standard model contains 15 particle states in each of the 3 generations .
these states are distinguished by the su(3)@xmath23su(2)@xmath24u(1)@xmath25 gauge interactions , which divide each generation into 5 multiplets : @xmath26 , @xmath27 , @xmath28 , @xmath29 , and @xmath30 .
the gauge interactions therefore break the flavor symmetry group from u(45 ) to u(3)@xmath31 . in the standard model , the flavor group u(3)@xmath31
is broken explicitly to @xmath32 by yukawa couplings . however , in extensions of the standard model in which one hopes to gain some understanding of the pattern of fermion masses and mixings , some subgroup of the flavor group may be an exact symmetry of the lagrangian that is broken spontaneously by the vacuum , and it is this possibility we consider here
. the massless nambu - goldstone bosons of the spontaneously broken flavor symmetry , familons @xcite , have interactions given by the couplings @xmath33 where @xmath14 are the familon fields , and @xmath34 are flavor currents .
the interactions are suppressed by @xmath13 , the scale at which the flavor symmetry is spontaneously broken .
note that familons are derivatively coupled , and so do not mediate long - range ( @xmath35 ) forces .
the most general current @xmath36 composed of two fermion fields takes either the form @xmath37 or @xmath38 where @xmath39 are the projection operators @xmath40 , @xmath41 and @xmath42 are generational indices , and @xmath43 are the spontaneously broken generators of the family symmetry . the fields @xmath44 and @xmath45 are fermion mass eigenstates , which we assume here to be also flavor eigenstates .
( the more general case is described below . ) using the form of the current given in eq .
( [ av ] ) , the familon interaction may be written as @xmath46 t_{ij}^a \psi_j \ , \label{famint}\end{aligned}\ ] ] where in the last step we have integrated by parts and then substituted the equations of motion .
the second line of eq .
( [ famint ] ) is of course only valid for on - shell fermions such as external leptons , whereas in hadronic matrix elements and processes including off - shell fermions , the derivative coupling of the first line must be used .
we see that familons may mediate or be produced in family - changing processes .
they may also couple to identical fermions @xmath47 , but only through axial couplings .
what processes are mediated by familons depends on the particular family symmetry group that is broken . for example , for o(@xmath48 ) groups , the generators @xmath49 are anti - symmetric , and so do not generate flavor - diagonal interactions .
however , they do generate interactions like @xmath50 , where we have considered axial vector current interactions as an example .
familons from o(@xmath48 ) groups may therefore mediate neutral meson mixing , which we will consider in sec .
[ sec : mesons ] .
the situation is reversed for su(@xmath48 ) groups . here ,
flavor - diagonal couplings exist .
however , if we consider any su(2 ) subgroup and form the complex familon @xmath51 , the off - diagonal interactions are given by @xmath52 , and we see that @xmath53 exchange can not induce neutral meson mixing . up to this point , we have ignored possible mass mixing effects . in general ,
if the flavor eigenstates @xmath54 are related to the mass eigenstates @xmath55 by @xmath56 where @xmath57 is a @xmath58 unitary mixing matrix , the familon interactions are given by @xmath59 where @xmath60 .
mass mixings may therefore generate flavor - diagonal interactions from flavor off - diagonal interactions , and vice versa .
for example , in the case of an abelian u(1 ) symmetry , mass mixing effects may generate flavor - changing interactions .
they may also extend non - maximal family symmetries to couplings involving all three generations ; for example , a u(2 ) symmetry between the first and second families , may , after rotation to mass eigenstates , result in familon interactions involving the third generation . while the phenomenology of familons varies from group to group , it is important to note that gauge symmetry relates the familon interactions of particles in the same gauge multiplet . as an example , let us consider a spontaneously broken lepton flavor symmetry .
the familon interaction is then given by @xmath61 where the su(2 ) lepton doublets @xmath62 are in the flavor eigenstate basis .
this interaction therefore generates familon interactions for both the charged leptons and neutrinos . in the presence of neutrino masses
, the flavor eigenstates may not correspond to mass eigenstates .
the familon interactions in the mass basis are then @xmath63 where @xmath64 , and we have defined @xmath65 and @xmath66 .
@xmath67 and @xmath68 are therefore related by a similarity transformation , and in the presence of mass mixing , the couplings of the interactions of @xmath69 and @xmath70 are not necessarily identical . however , in the absence of fine - tuning , we expect these couplings to be of the same magnitude .
bounds on one familon interaction may thus be considered to imply comparable bounds on the other interactions linked by gauge symmetry .
because the familon interactions of particles in the same gauge multiplet are comparable in the absence of fine - tuning , there are many more relations in theories with enlarged gauge groups . for example , for su(5 ) grand unified theories ( guts ) , the particles @xmath71 are expected to have comparable familon interactions , as are the particles @xmath72 .
a particularly relevant example for our study below is that , in the gut framework , bounds on familon decays of @xmath8 mesons imply bounds on familon decays of tau neutrinos in the absence of fine - tuning .
flavor mixing effects also induce familon couplings of fields with different generational indices . in the quark sector , for example , substituting the quark doublet @xmath73 for @xmath74 in the discussion above , eq .
( [ massbasis ] ) becomes @xmath75 where @xmath76 and @xmath77 are related by the ckm matrix through @xmath78 we see that in general , couplings to all generations are induced by flavor mixings .
for example , a familon with flavor - diagonal coupling to @xmath79 in the up sector couples not only to @xmath80 , but also to , for example , @xmath81 and @xmath82 .
the induced couplings to first and second generation quarks in this case are ckm - suppressed , but may still lead to significant bounds when , as is often the case , these induced couplings are much more strongly constrained .
we will consider the constraints on mixing - induced couplings from @xmath83 decays in sec .
[ sec : decays ] and from supernova cooling in sec .
[ sec : mixing ] . finally , note in eq .
( [ av ] ) that the strength of the interaction depends not only on @xmath13 , but also on @xmath43 and the couplings @xmath84 . in the following sections
, we will present a variety of bounds on combinations of these couplings , and it is important that we define our conventions and normalizations .
we will always define our interaction as @xmath85 and similarly for @xmath86 and @xmath87 ; the superscripts of the couplings will often be omitted when they are obvious from the context . in presenting our bounds
, it will be convenient to define @xmath88 where @xmath89 .
in addition , as many of our bounds are to a good approximation independent of the chirality of the interaction and so only dependent on the combination @xmath90 , we define @xmath91
as described in the previous section , familons may take part in flavor - changing processes , and bounds on such processes lead to lower bounds on the familon energy scale . for familons mediating transitions between the first and second generation ,
such bounds are rather stringent . in contrast , similar bounds involving the third generation are much weaker , with the previously reported constraints limited only to bounds from rare @xmath18 decays .
we are thus motivated to focus on the third generation . in sec .
[ sec : decays ] , we begin by reviewing and contrasting such bounds , and then discuss the implications of flavor eigenstate mixings .
we then go on to derive new bounds from a variety of processes . in sec .
[ sec : mesons ] we consider familon - mediated processes such as neutral meson mixing and rare leptonic decays of mesons . finally , in sec .
[ sec : lep ] we consider possible analyses at lep and extrapolate a preliminary aleph bound on @xmath92 to a bound on @xmath93 .
we begin by considering bounds from decays of mesons and leptons to familons . normalizing the relevant familon scale according to eq .
( [ conven ] ) , we find @xmath94 where @xmath95 . in the limit of exact flavor su(3 ) symmetry , the form factor @xmath96 at zero momentum transfer
has a fixed normalization , @xmath97
. for leptonic decays @xmath98 , the exact tree - level partial decay width in the limit of massless @xmath99 is given by @xmath100 where here @xmath101 .
the strongest bound on any flavor scale is derived from the constraint on exotic @xmath83 decay . using the above expressions
, the experimental result @xmath102 @xcite leads to the bound @xmath103 note that the limit on @xmath104 bounds only the vectorial familon coupling ; the axial coupling is unconstrained . for the leptonic sector ,
jodidio _ et al .
_ report the constraint @xmath105
@xcite , which they obtain under the assumption of a vector - like familon coupling .
this can be converted into the bound @xmath106 for familon interactions of arbitrary chirality , the slightly weaker constraint @xmath107 may be obtained from the bound @xmath108 ( 90% cl ) @xcite .
we now compare these bounds to those available in the third generation .
the argus collaboration @xcite has bounded the branching fractions of @xmath18 decays into light bosons and found the limits @xmath109 and @xmath110 .
these imply the following constraints on the flavor scale : @xmath111 we see that the bounds on flavor scales in the leptonic sector are significantly less stringent for third generation couplings than for those involving only the first two .
the discrepancy is even more pronounced in the hadronic sector , where there are as yet no reported bounds on flavor scales from @xmath8 decays .
it is also worth noting , however , that strong bounds on a particular flavor scale , such as the one on @xmath112 , may imply significant bounds on other flavor scales as well .
these bounds are induced by the flavor - mixing effects discussed in sec .
[ sec : familons ] and are thus model - dependent . as an example
let us now assume that flavor and mass eigenstates coincide for up - type quarks .
a given familon coupling in the up sector requires , by gauge invariance , a corresponding coupling in the down sector .
for example , from eqs .
( [ quarkmassbasis ] ) and ( [ ckmrelation ] ) we see that the coupling @xmath113 induces the coupling @xmath114 , which mediates the rare decay @xmath115 . assuming complex familons
, the hermitian conjugate coupling gives a similar contribution @xmath116 to the decay into the complex conjugate familon .
summing both decay widths and comparing to the bound on @xmath112 in eq .
( [ kpifam ] ) , one can derive the mixing induced bound @xmath117 under similar assumptions , we find @xmath118 .
note , however , that such bounds do not apply if the mass and flavor bases are aligned in the down sector @xcite or if the couplings are purely axial . in this section
we derive new constraints on the scale of spontaneous flavor symmetry breaking by considering non - standard familon contributions to neutral meson mixing and existing bounds on rare leptonic decays such as @xmath119 . a familon contribution to neutral meson mixing requires a real flavor group to be spontaneously broken in the corresponding sector , such that the same real familon scalar field couples to the quark current and its hermitian conjugate . for concreteness ,
let us consider the @xmath120 system ; similar formulae hold ( at least approximately ) for other neutral meson systems .
assuming the general coupling structure @xmath121 \ , \label{mesonmixing}\end{aligned}\ ] ] we find a familon contribution to the mass splitting of @xmath122 eq .
( [ deltamsq ] ) may be derived by taking the matrix element of the non - local operator @xmath123 between @xmath1 and @xmath2 states and using the definition of the pseudoscalar decay constant , @xmath124 .
the subscripts @xmath125 , @xmath126 , @xmath127 , and @xmath128 in eq .
( [ oper ] ) are color indices . between two color singlet states , there are two contributions .
the first one arises from @xmath129 and @xmath130 with a familon in the @xmath131-channel . in this case , the momentum transfer through the familon propagator is @xmath132 , and after a vacuum insertion , it is easy to verify that this contribution is as in eq .
( [ deltamsq ] ) , but without the factor of @xmath133 .
however , there is also a @xmath134-channel contribution from @xmath135 and @xmath136 , which may be evaluated by a fierz transformation and then a vacuum insertion as before . for a heavy light system like the @xmath1 meson
, one may assume the free - quark picture , in which the momentum transfer is governed by the energy of the `` static '' @xmath19 quark @xmath137 , and , in the numerator , the derivative acting on the quark current gives again a factor of @xmath138 .
using @xmath139 and including the relative color factor of @xmath140 , one can estimate the @xmath134-channel contribution to be @xmath141 times the @xmath131-channel contribution , which leads to eq .
( [ deltamsq ] ) .
our result should be fairly reliable for the @xmath1 meson . for @xmath142 and @xmath10 mesons ,
the evaluation of the @xmath134-channel momentum transfer is more ambiguous .
however , because this contribution is suppressed relative to the @xmath131-channel part , we expect the result of eq .
( [ deltamsq ] ) to be reasonably accurate in these cases as well .
we also note that a vector - like familon interaction does not contribute to the mass splitting , at least in the heavy quark approximation @xmath143 .
although one might expect a vector contribution to appear in the @xmath134-channel contribution after the fierz rearrangement , one finds that the term proportional to @xmath144 contains axial vector and pseudoscalar contributions of equal magnitude but opposite sign . the constraint on the flavor scale @xmath13 results in principle from the requirement that the combined standard model and familon contributions do not exceed the measured value .
however , when considering nonstandard contributions , it is also uncertain what one should take as the standard model contribution .
for example , the reported value @xcite for @xmath145 is derived from @xmath120 mixing under the assumption that the standard model gives the only contribution .
as a conservative bound , we simply compare the familon contributions directly to the corresponding measured values .
the results are summarized in table [ tab : mixing ] .
we take the decay constants to be @xmath146 mev and @xmath147 mev from recent lattice results @xcite , and @xmath148 mev @xcite .
since we use the measured mass splitting ( not its error ) , the bounds from @xmath1 and @xmath10 will only improve when the size of the standard model contribution can be quantified independently .
for the @xmath142 , where only the upper bound on the mass splitting is known , future experiments will improve the bound .
.bounds on the flavor scale from contributions to neutral meson mixing from familon exchange as given in eq .
( [ deltamsq ] ) .
note that these limits do not apply to vector - like couplings , and that this process requires a real flavor group so that a real familon scalar field couples to a current operator and its hermitian conjugate , as in eq .
( [ mesonmixing ] ) . [ cols=">,>,>",options="header " , ] [ tab : decay ] currently , there are no reported experimental bounds on decays @xmath149 .
one can , however , infer a constraint from aleph s preliminary bound on @xmath150 @xcite . by searching for events with large missing energy
, they placed the constraint @xmath151 ( 90% cl ) .
one can rescale this constraint to obtain an upper bound on @xmath152 .
the analysis for @xmath153 relies on the @xmath154 distribution @xcite , where @xmath154 is defined by @xmath155 in each hemisphere of @xmath19-tagged events .
here , @xmath156 is half of the center - of - momentum energy , @xmath157 , where @xmath158 and @xmath159 are the visible invariant masses in the same and opposite hemispheres , respectively , and @xmath160 is the total visible energy in the hemisphere .
@xmath161 improves the estimate of the actual missing energy in the hemisphere by correcting for the fact that the hemisphere with larger invariant mass typically has higher energy .
the backgrounds from @xmath162 and @xmath163 are suppressed by rejecting events with identified @xmath164 or @xmath165 in the relevant hemisphere . up to this point
, we do not expect significant differences in efficiencies between the @xmath150 mode and the @xmath93 mode .
they then required @xmath166 .
the efficiencies for this requirement obviously differ between the two decay modes , since the mode @xmath150 has two missing neutrinos , resulting in a harder @xmath154 spectrum than that of the @xmath93 mode .
the @xmath154 spectrum of both modes may be calculated by convoluting the theoretical missing energy distribution in three - body ( @xmath167 ) and two - body ( @xmath168 ) decays with the measured @xmath19 fragmentation function @xcite .
we find that the ratio of efficiencies is 0.43 with little dependence on the details of the fragmentation function . by scaling the reported @xmath169 upper bound by this factor
, we find @xmath170 using the expression of eq .
( [ hier ] ) with the substitution of @xmath171 for @xmath172 , this corresponds to a limit on the flavor scale of @xmath173 note that this analysis does not require an energetic strange particle , and so the constraint of eq .
( [ bsbound ] ) is actually on the sum @xmath174 .
thus , for @xmath175 , the bound on the flavor scale given in eq .
( [ bsfbound ] ) improves by a factor @xmath176 .
the bound of eq .
( [ bsfbound ] ) is enhanced by the fact that the sm decay width is greatly suppressed by @xmath177 , which increases the sensitivity of @xmath19 decays to small exotic decay widths .
in this section , we discuss constraints on third generation familon couplings from astrophysics .
we begin in sec .
[ sec : tree ] with constraints on direct ( tree - level ) couplings .
second and third generation particles are absent in almost all astrophysical objects .
the exception is supernovae , where all three neutrino species are thermalized in the core .
we therefore consider what bounds on familon couplings to @xmath18 neutrinos may be obtained by supernova observations .
couplings of familons to the third generation may also radiatively induce couplings to first generation particles .
although such induced couplings are suppressed by loop factors , they are so stringently bounded by constraints from supernovae , white dwarfs , and red giants that interesting bounds also result .
these are studied in sec .
[ sec : loop ] .
finally , mixings of flavor eigenstates may also induce couplings of familons to the first generation ; such effects are discussed in sec .
[ sec : mixing ] .
it is important to note that , while the bounds derived in this section are rather strong in certain cases , they are also typically more model - dependent than , for example , the accelerator bounds of the previous section . we therefore specify the necessary conditions for each bound in detail in each case . in 1987 ,
the kamiokande group and the imb group independently detected neutrinos emitted from supernova sn 1987a .
they observed that the neutrino pulse lasted for a few seconds .
furthermore , their results indicate that neutrinos carried off about @xmath178 erg from the supernova . the observed duration time and neutrino flux
can be well explained by the generally accepted theory of core collapse , and the observations confirmed the idea that most of the released energy in the cooling process is carried off by neutrinos .
exotic light particles , such as familons , may affect the agreement of theory and observation , since they can also carry off a significant energy fraction .
the core of the supernova is hot ( @xmath179 ) and dense , and so neutrinos are thermalized in the core and can be a source of familon emission .
if the energy fraction carried away by familons is substantial , the duration time of the neutrino pulse becomes much shorter than the observed value . in order not to affect the standard cooling process
, the familon luminosity @xmath180 must be smaller than the neutrino luminosity , _
i.e. _ , less than @xmath181 erg / sec .
this constraint can be satisfied in two different regimes of the familon coupling strength .
for sufficiently high flavor scales @xmath13 , the familon interaction is weak enough that familons are rarely produced and the familon luminosity @xmath180 is suppressed . on the other hand , for sufficiently low flavor scales ,
although familons are readily produced , they interact so strongly that they become thermalized and trapped in the core as well , thus decreasing the familon luminosity .
therefore , there are two parameter regions consistent with observations , high and low @xmath13 , with an excluded region in the middle .
these types of constraints have been discussed by choi and santamaria @xcite in the context of a majoron model .
we modify their discussions slightly for the familon case . to simplify the analysis
, we will look at two extreme scenarios .
first we consider a diagonal familon coupling to @xmath182 , as in the case of an abelian family symmetry , and second we analyze a purely off - diagonal coupling , as in the case of an o(2 ) family symmetry . , most of the interesting mass range from a cosmological point of view is therefore excluded . ] for a general family symmetry , one expects familons with both diagonal and off - diagonal couplings ; a generalization to such cases is straightforward . in this subsection
, we also neglect possible mismatches between the flavor and mass eigenstates , and assume that the relative angles relating the two are small .
such mismatches will be discussed in sec .
[ sec : mixing ] . finally , we assume that @xmath183 , @xmath184 are negligible compared to @xmath185 , as suggested from laboratory constraints as well as the corresponding masses of the charged leptons . here
we consider a purely diagonal familon coupling to @xmath182 , such as in models with a u(1 ) family symmetry acting on the third - generation lepton doublet @xmath186 .
the relevant interaction is given by @xmath187 let us first consider the case where the familon can freely escape the core of the supernova .
based on the interaction given in eq .
( [ l_on ] ) , potentially significant processes of familon production are the neutrino scatterings @xmath188 and @xmath189 , the latter process being allowed due to background matter effects .
the familon luminosities due to these processes are given in ref .
@xcite : @xmath190 if _ either one _ of the above luminosities is larger than @xmath191 erg / sec , the cooling process of the supernova may be dominated by familon emission , and the duration time of the neutrino pulse becomes shorter than @xmath192 . imposing the constraint that the familon luminosities given in eqs .
( [ q_f1 ] ) and ( [ q_f3 ] ) are smaller than @xmath193 erg / sec , we obtain the following constraints : @xmath194 \displaystyle \left(\frac{m_{\nu_\tau}}{\text{mev}}\right ) \left(\frac{\text{gev}}{f^l_{\nu_\tau\nu_\tau}}\right ) & \leq & 4.0\times 10^{-4 } \ ; , & m_{\nu_\tau}\leq 95 \text { kev } \ .
\end{array } \right .
\label{constraint_3}\end{aligned}\ ] ] familon volume emission is sufficiently small when both eqs .
( [ constraint_1 ] ) and ( [ constraint_3 ] ) are satisfied
. on the other hand , if familon interactions are strong enough , familons effectively scatter off the neutrinos in the background and get thermalized and trapped in the supernova .
once this happens , a thermal sphere of familons is formed , just like the thermal neutrino sphere , and familons can only be emitted from the surface .
the familon luminosity essentially obeys the formula of blackbody emission with the surface temperature of the familon sphere .
the important point is that , once the familon is trapped , the familon luminosity _ decreases _ as the familon interaction becomes stronger .
this can be understood in the following way : as the familon interaction becomes stronger , familons can be thermalized with a lower temperature .
( notice that the scattering rate increases for higher temperature . )
the surface temperature of the familon sphere then decreases , and hence the luminosity is suppressed .
therefore , the familon luminosity can be small enough when the scale @xmath13 is sufficiently low .
following ref .
@xcite we find that the cooling through familon emission is sufficiently suppressed ( _ i.e. _ , is less than @xmath195 ) when _ either one _ of the following constraints is satisfied : @xmath196 of course , if the familon has strong interactions with other light particles ( the photon , electron , or neutron ) , familons may be trapped by other processes as well .
this gives additional regimes where the earlier constraints of eqs .
( [ constraint_1 ] ) and ( [ constraint_3 ] ) can be evaded .
= 0.6 in fig .
[ fig : sn ] , we show the upper and lower bounds on the diagonal coupling @xmath197 as a function of the neutrino mass @xmath198 .
the region above the upper line is allowed because familon emission is sufficiently suppressed by the flavor scale @xmath13 .
this line is basically determined by eq .
( [ constraint_1 ] ) ; the slight bump is due to eq . ( [ constraint_3 ] ) .
as we can see , the lower bound on @xmath13 is at most 1 tev for the maximum allowed value of the tau neutrino mass ( 18.2 mev ) , and it becomes less stringent as the mass becomes smaller .
the lower boundary is determined by eq .
( [ lowerbound_1 ] ) , which supercedes eq .
( [ lowerbound_3 ] ) .
the region below this line is also allowed because the familon is trapped in the core and the contribution to the cooling is again sufficiently small .
here we discuss sn 1987a constraints on a familon which has only an off - diagonal coupling , such as in the case of an o(2 ) family symmetry .
the low - energy lagrangian of the model can be written as @xmath199 the inclusion of @xmath200 in the discussion is straightforward .
the familons are produced by @xmath188 via @xmath134-channel @xmath201 exchange , @xmath202 via @xmath134-channel @xmath182 exchange , or the decays @xmath203 .
following ref .
@xcite again , @xmath204 we require that _ all _ of these familon luminosities are smaller than @xmath193 erg / sec , and obtain the following constraints : @xmath205 where the lifetime of @xmath182 is given by @xmath206 increasing the interaction strength further into the excluded region , familons eventually become trapped and rendered harmless again .
this occurs when _ any one _ of the following constraints are satisfied : @xmath207 as mentioned before , if the familon has strong interactions with other light particles , these interactions may lead to thermalization of familons as well , resulting in additional allowed regions for low flavor scales .
the resulting excluded region is fairly similar to that of the diagonal coupling case .
the constraint from the decay process eq .
( [ constraint_6 ] ) is important only for smaller @xmath208 or larger masses @xmath185 , values that are outside our range of interest .
in addition , the small bump in fig .
[ fig : sn ] now disappears due to the absence of the @xmath209 process .
the dominant constraints are therefore from eqs .
( [ cool1 ] ) and ( [ lowerbound_4 ] ) in the off - diagonal case , which differ from the dominant constraints of eqs .
( [ constraint_1 ] ) and ( [ lowerbound_1 ] ) in the diagonal coupling case only by a small constant factor . the boundary of the excluded region is therefore given by the lines of fig .
[ fig : sn ] shifted downwards by a factor of 1.2 in @xmath13 . in the previous subsection , we considered astrophysical constraints on tree - level familon couplings to @xmath210 .
in addition , however , astrophysical bounds may also be used to constrain familon couplings to all other particles , as these couplings may induce couplings of familons to electrons and nucleons at the loop level . while these induced couplings are suppressed by the usual loop factors , the bounds on familon couplings to first generation particles
are so stringent that these constraints may be strong in certain cases .
in fact , we will see below that the contributions to induced couplings are proportional to fermion masses , and so these constraints are particularly relevant for couplings of familons to third generation fermions . in this subsection
, we will estimate the induced couplings for various choices of the family symmetry group and determine what lower bounds on flavor scales @xmath13 result from current astrophysical constraints .
for simplicity , we will limit our discussion here to familons with flavor - diagonal couplings to the third generation , and ignore possible rotations relating the flavor and mass eigenstates .
extensions of this analysis to more general cases are straightforward . to evaluate the strength of the induced coupling
, we will begin by considering the low energy effective theory below the flavor scale @xmath13 . in this approach ,
the theory is specified by the flavor symmetry , that is , the low energy derivative couplings of the familon , and no further knowledge of the mechanisms of flavor symmetry breaking is required . with the assumptions given above
, the dominant contribution to the induced couplings is from the @xmath211-@xmath212 mixing graph shown in fig .
[ fig : feynman ] . here
@xmath213 is any one of the third generation particles directly coupled to the familon , and @xmath214 , @xmath215 , or @xmath216 .
( there are also additional contributions from penguin - like @xmath217 diagrams , but these are suppressed by mixing angles , _
e.g. _ , @xmath218 in the case of @xmath219 . )
let us define the @xmath212-@xmath213 coupling as @xmath220 and the @xmath211-@xmath213 coupling as @xmath221 where @xmath222 , @xmath223 , and @xmath224 .
the induced @xmath211-@xmath212 mixing from the fermion loop is divergent , and the logarithmically - enhanced contribution is @xmath225 z_{\mu } \partial_{\nu } f \nonumber\\ & = & \frac{2n_c}{(4\pi)^2 f } m_\chi^2 \ln \frac{\lambda^2}{m_\chi^2 } ( g_l - g_r ) ( g_l^z - g_r^z ) z^{\mu } \partial_{\mu } f \ , \label{zphimix}\end{aligned}\ ] ] where @xmath226 is the number of colors of the fermion @xmath213 , and @xmath227 is the effective ultraviolet cutoff of the order of the flavor scale @xmath13 . note that @xmath228 and the fermion charge @xmath229 drops out : the ward
takahashi identity guarantees that the @xmath230 piece in the @xmath211 vertex gives a completely transverse vacuum polarization amplitude proportional to @xmath231 , which vanishes when contracted with @xmath232 .
the leading contributions to the induced couplings are determined by the amount of current non - conservation , _
i.e. _ , the masses of the particles in the loop , and the third generation couplings therefore give the most important contributions .. this operator may be present at tree - level if the familon couples to `` higgs number , '' but we will not consider this case . ] = 0.3 the mixing of eq .
( [ zphimix ] ) is logarithmically - enhanced and so typically gives the leading contribution if present . however , there are cases in which this term is not present or is highly suppressed .
first , it may be that the amount of current non - conservation is itself suppressed by inverse powers of the flavor scale .
for instance , in the singlet majoron model @xcite , lepton number conservation in the low - energy theory is violated only by neutrino masses which are of order @xmath233 due to the seesaw mechanism .
the neutrino loop contribution to the @xmath211-majoron mixing is then @xmath234 and highly suppressed .
second , if the familon coupling is vector - like so that @xmath235 , the logarithmically - enhanced term is absent . for example , a familon coupled to the ( possibly generation - dependent ) baryon number current has this property . finally , this mixing is also absent if the familon has only flavor off - diagonal couplings . in all of these cases , the contribution of eq .
( [ zphimix ] ) is absent or suppressed , and the leading contributions to @xmath211-@xmath212 mixing come from non - logarithmically - enhanced threshold corrections which may be of order @xmath233 .
such corrections are sensitive to physics at the flavor symmetry breaking scale , and are therefore model - dependent . with these caveats in mind
, we now assume that the leading contribution given in eq .
( [ zphimix ] ) is present , and determine bounds on @xmath13 for various flavor symmetries .
the logarithmically - enhanced mixing induces an effective coupling @xmath236 where in the last step we have integrated by parts and substituted the equations of motion for @xmath55 .
for example , for a familon coupled to @xmath237 , @xmath238 , @xmath239 , and @xmath240 . for a familon coupled to @xmath241 , the contributions from both @xmath242 and @xmath243 must be summed . the effective coupling of eq . ( [ bracketed ] )
is constrained from various sources . for the case
@xmath244 , a stringent constraint is provided by red giants .
familon - electron couplings lead to additional sources of red giant cooling , which , if too large , would destroy the agreement between the observed population of red giants in globular clusters and stellar evolution theory .
such constraints have been studied extensively in the literature @xcite .
the current best upper limit on the coupling is @xcite @xmath245 for @xmath246 .
the strongest bound on the family symmetry breaking scale is for familon couplings that are dominantly proportional to @xmath247 , as , for example , when a familon is coupled only to @xmath237 .
such a case results in the bound @xmath248 where we have taken @xmath249 .
weaker , but still significant constraints are obtained if the familon coupling is dominated by @xmath138 , as when the familon couples only to @xmath250 . the bound in this case is @xmath251 notice that , in the case where the familon couples to @xmath237 and @xmath241 with the same charge , the bound of eq .
( [ frbb ] ) also holds for the corresponding flavor scales
. however , possibly stronger bounds may also be possible if model - dependent non - logarithmically enhanced terms proprtional to @xmath247 are present . if the familon contribution to the induced coupling is dominantly proportional to @xmath252 , we find the constraint @xmath253 similar bounds may be obtained from induced familon couplings to nucleons using constraints from supernova sn 1987a by rescaling the bounds on axion couplings .
these constraints are somewhat more ambiguous because of the loss of coherence in axion emission due to nucleon spin fluctuations caused by scattering effects in the supernova core @xcite .
more realistic estimates were addressed in refs .
the constraints yield results comparable to the red giant bound on electron couplings , but with larger error bars . finally , we stress again that these bounds are for specific flavor symmetries . for certain examples mentioned above , the logarithmically - enhanced contribution to the induced coupling is absent . for such flavor symmetries , the threshold corrections at the flavor scale
must be studied separately for each model , and the flavor scale can be constrained only after the model - dependent coefficients are known . however , from the numerical estimates above , it is clear that the induced loop - level bounds can provide interesting constraints on flavor - diagonal familon couplings .
such bounds are particularly interesting for couplings to the third generation , as they are enhanced for large fermion masses .
note also that couplings to the top quark are stringently bounded and are extremely difficult to bound by other means . in this section ,
we have so far parametrized and constrained possible familon couplings individually by introducing effective flavor scales , neglecting possible mismatches between flavor and mass eigenstates . however , as noted in sec .
[ sec : familons ] , when bounds on a particular familon coupling are very stringent , such as in the case of bounds on familon couplings to the first generation from supernovae , one can also obtain interesting bounds on other familon couplings from flavor - mixing effects . in this subsection
, we will consider such bounds in the quark sector . in extensions of the standard model with massive neutrinos ,
similar arguments hold in the leptonic sector .
additional constraints may also be obtained if the gauge symmetry is enlarged .
let us assume that the flavor and mass eigenstates coincide in the down sector .
a generic familon coupling term for left - handed down - type quarks is then @xmath254 where @xmath255 and @xmath256 are generational indices .
notice that , in this section , the familon @xmath212 is a real scalar for the diagonal couplings ( @xmath257 ) , and a complex scalar for the off - diagonal ones ( @xmath258 ) .
( thus , in the off - diagonal case , there is also a hermitian conjugate term in the lagrangian . )
the up - type coupling required by gauge invariance in terms of down quark mass eigenstates is @xmath259 where @xmath260 .
therefore , a constraint on @xmath261 obtained from supernova cooling through familon
nucleon coupling implies similar constraints on the expressions @xmath262 , @xmath263 , and so on . of course
, different contributions to the same effective coupling @xmath264 may also have opposite signs , which must be checked in the specific model under investigation . to derive constraints on the flavor symmetry breaking scales
, we must convert the quark level couplings to the effective nucleon - familon couplings of the form @xmath265 .
this can be done through a generalized goldberger - treiman relation . with the interaction given in eq .
( [ phi - q - q ] ) , we obtain @xmath266 where @xmath267 gev is the nucleon mass , and the coefficients @xmath268 are given by @xcite @xmath269 here , we have assumed that the flavor symmetry is anomaly - free for su(3)@xmath270 .
if the flavor symmetry is anomalous under su(3)@xmath270 , there are anomaly - induced contributions to eq .
( [ phi - n - n ] ) ; see refs .
@xcite for discussions of constraints on axions .
the effective couplings of eq .
( [ phi - n - n ] ) are constrained by supernova sn 1987a . in ref .
@xcite , the upper bound on @xmath271 is given as a function of @xmath272 .
for simplicity , we adopt the most conservative constraint on @xmath271 , @xmath273 and use this to estimate bounds on the flavor symmetry breaking scales . under the assumption that only one down - type familon coupling exists at a time , that is , ignoring possible cancellations between two different contributions , we find bounds on third - generation couplings @xmath274
an even stronger constraint is obtained for @xmath275 : @xmath276 this constraint is , however , weaker than the laboratory bound . note that the above bounds are obtained under the assumption that the mass and flavor eigenstates are identical in the down sector .
if , on the other hand , these eigenstates were assumed to be identical in the up sector , familon coupling to the @xmath131- and @xmath216-quark arises due to the mixing effect , and the supernova constraints on @xmath277 and @xmath278 could be used instead . in particular , an interesting bound is derived for @xmath279 : @xmath280 this bound is about one order of magnitude stronger than the bound on @xmath281 since , in this case , the effective familon coupling to the nucleon is dominated by the @xmath131-quark contribution , and the interaction is therefore not as highly cabbibo - suppressed as in the @xmath281 case .
of course , there is no reason why the familon coupling is diagonalized in one sector .
however , as the up and down sectors can not be diagonalized in the same basis , one generally expects similar mixing - induced constraints in all cases .
in this section , we estimate what constraints on familon couplings may be obtained from the current cleo data set and the upcoming cleo iii , babar and belle experiments .
we make no attempt to conduct detailed experimental studies appropriate to each of these experimental settings .
rather , our intent here is to describe a number of analyses that are likely to significantly improve the present limits on familon energy scales , and , we believe , merit further study .
we will begin with investigations of hadronic couplings of @xmath19 quarks to familons , and then consider decays of the @xmath18 lepton to familons . in all of the experiments mentioned above
, one may search for the exclusive decays @xmath283 and @xmath284 .
these exclusive modes have smaller branching fractions than the inclusive modes @xmath7 , but have clear experimental signatures due to their simple two - body kinematics .
the form factor of @xmath285 which is necessary to calculate branching fractions , has been estimated by colangelo _
et al . _
@xcite to be @xmath286 , based on sum rules .
estimates of @xmath287 based on the quark model are 0.34 , 0.36 , 0.30 , or 0.35 , depending on which quark model parameters are assumed @xcite .
we could not find other estimates of this particular form factor in the literature , but various estimates for @xmath288 transitions give comparable but slightly larger values .
this is reassuring , since they must agree in the flavor symmetric limit . the decay rate @xmath289 is given by @xmath290 where @xmath291 .
( if the coupling is purely axial , there is , of course , no contribution to @xmath289 ; searches for decays to @xmath292 are required to bound such couplings . )
neglecting the mass difference between the @xmath19 quark and @xmath8 meson , and using the naive spectator model for the @xmath8 meson decay , one finds @xmath293 the concept of the search for such exclusive decay modes is relatively simple . after applying the standard cuts to suppress continuum @xmath294 and lepton pair events , one looks for events at the @xmath295 resonance that have either an isolated @xmath296 , or an isolated charged meson @xmath297 , @xmath298 together with large missing energy . in the center - of - momentum frame
, the energy of the meson @xmath299 , @xmath298 , or @xmath296 must be in the narrow range @xmath300 < e_p < \frac{\sqrt{s}}{4 } \left [ \left ( 1 + \frac{m_p^2}{m_b^2}\right ) + \beta \left ( 1 - \frac{m_p^2}{m_b^2}\right ) \right ] , \ ] ] where @xmath301 , and @xmath302 is the mass of the meson .
one can also require that , after excluding the isolated energetic meson whose energy is in the above range , all the tracks and energy deposits in the calorimeters reconstruct @xmath303 and have total energy @xmath304 in the center - of - momentum frame . of the existing analyses ,
the one most similar to that described above is a search for @xmath305 by the cleo collaboration @xcite .
the reported upper bounds on the branching fractions are @xmath306 @xmath307 , @xmath308 @xmath309 , and @xmath310 @xmath311 .
the reach for the @xmath312 mode is expected to be worse than for @xmath313 or @xmath314 , as the continuum backgrounds are larger and the detection efficiencies are worse for mesons .
however , we expect the sensitivity to such meson decays to be greater than to the @xmath315 mode , because the mesons have more - or - less fixed energy , unlike in the @xmath315 case .
@xmath316 separation is probably difficult with the current cleo data set @xcite , but this analysis may still give us an upper bound on @xmath317 somewhere at the @xmath318 to @xmath319 level @xcite .
such a constraint would be competitive with the upper bound inferred from the aleph @xmath4 study discussed in sec .
[ sec : lep ] .
particle identification will be much better at cleo iii , which will allow @xmath316 separation , and will be even better at babar and belle .
the higher luminosity at these machines will also help , and an upper bound of @xmath320 may be possible @xcite .
such a bound would imply a bound on the flavor symmetry breaking scale of @xmath321 gev !
we now turn our attention to the lepton sector .
the decay rate for @xmath323 is given in eq .
( [ hier ] ) . a search for @xmath324 suffers from the standard model background @xmath325 .
a conventional method for bounding the branching fraction to familons is to fit the momentum spectrum of the electron ( muon ) from tau decay to a linear combination of the standard model spectrum , which drops approximately linearly for large momenta , and a possible contribution from the familon mode , which is flat .
the argus bound quoted in sec .
[ sec : decays ] was obtained by this method .
cleo has not reported a similar analysis . however , in a recent cleo analysis of the michel parameter in @xmath18 decays @xcite , the electron momentum distribution of 33531 @xmath326 events was presented in uniform 0.25 gev bins .
this distribution may be fit beautifully by the standard model alone , and contains about 90 events in the highest momentum bin . for reference ,
the contribution of a familon decay mode with @xmath327 , a branching fraction near the current argus limit , would contribute 28 events in each bin , leading to a significant excess at high momentum . to see the possible sensitivity given the current cleo data set , we generated 33531 standard model @xmath328 events .
the momentum spectrum from our monte carlo simulation ( points with error bars ) , along with the predicted standard model spectrum ( solid histogram ) , is shown in fig .
[ fig : emom]a . for comparison
, we also plot the spectrum given a hypothetical familon branching fraction of @xmath327 normalized to the same number of events . in fig .
[ fig : emom]b , we plot the ratio to the standard model prediction to make the familon contribution more visible .
note that the spectrum with the familon contribution differs considerably in the high momentum bins . by fitting the monte carlo data to the linear combination of the standard model and familon modes , we find that cleo can obtain an upper bound on the familon branching fraction of @xmath329 ( 95% cl ) .
this would already improve the argus bound @xcite on @xmath330 slightly , leading to a lower bound of @xmath331 gev .
note that in this analysis , only events with the @xmath332 decay of the other @xmath18 were used .
this requirement was motivated in the original study by the desire to study spin correlations between decaying @xmath18 pairs , but is not necessary for our purpose .
the statistical power of our analysis may therefore be boosted by including events with additional decay modes of the other @xmath18 .
pairs by selecting events with @xmath333 ( see ref . @xcite for the definition of @xmath334 ) .
such a selection enhances the right - handed @xmath335 ( left - handed @xmath336 ) decaying to leptons , and thereby suppresses the electron momentum spectrum at the end point .
the sensitivity to familon contributions at this endpoint is then improved . ] in this analysis , the systematic effects appear to be under control .
the momentum dependence of the electron identification efficiency can be calibrated by using the actual data , for instance , by using radiative bhabha events , and this calibration improves with statistics .
in addition , the background is small in the above cleo data sample .
indeed , all measurements of @xmath18 decay parameters are statistically limited and we expect a similar situation for the familon analysis .
note also that in the above analysis we simply fit to the standard model contribution , allowing its normalization to vary . in principle
, one can determine this normalization by measuring the efficiencies of @xmath18 identification in each decay mode through methods analogous to the multi - tag methods employed in the measurement of @xmath337 in @xmath211 decays @xcite .
we therefore conclude that a dedicated analysis could well lead to an upper bound on @xmath338 below the @xmath319 level .
the @xmath339 mode is more difficult because the muon identification efficiency is less well - calibrated and the statistics is slightly poorer , with 21680 events in the cleo analysis .
again , however , the uncertainties are dominated by statistical errors .
we therefore expect an upper bound on @xmath340 only slightly worse than that on the @xmath341 mode .
although we are concerned primarily with @xmath8 factories in this section , we should note that the above analysis may also be applied at lep .
for example , in a recent opal analysis of @xmath18 polarization @xcite , a large sample of 25000 @xmath342 events was studied . by fitting the @xmath343 momentum distribution as described above , an upper bound on @xmath344 at the @xmath345 level could be derived .
( here , we have simply scaled the cleo results given above to the opal statistics ; we have checked that the momentum distributions are sufficiently similar that such an approximation is valid . ) combining the four lep experiments , we expect an upper bound of @xmath346 . at the asymmetric @xmath8 factories ,
the boosted center - of - momentum system would somewhat complicate the analysis of the electron ( muon ) momentum spectrum .
we are not aware of any studies at these colliders .
however , given that the event samples available at these machines will be much larger than the current cleo data set , we expect babar and belle to place constraints significantly better than the @xmath319 level , and possibly at the @xmath318 level .
the study of familon decays at these machines is extremely promising , and worthy of further study .
it must be mentioned , furthermore , that the future @xmath8 factories will be able to improve the upper bound on the mass of @xmath210 significantly to the level of 3 mev , or possibly even 1 mev @xcite .
as we will see in the next section , the interplay between bounds on the @xmath210 mass and bounds on branching ratios to familons is very interesting from a cosmological point of view .
finally , we note that the bounds on @xmath347 branching ratios are expected to be even better at a tau - charm factory .
@xcite has shown that one can reach the level of @xmath348 using the standard optics or even @xmath349 using a monochromator .
this would raise the lower bound on the flavor scale to @xmath350 gev .
the @xmath351 mode is more difficult , and is limited by the @xmath352 separation capability @xcite .
however , a bound better than the @xmath319 level using a rich detector for particle identification is expected @xcite .
non - standard properties of neutrinos are always interesting in cosmology , and in fact , heavy unstable neutrinos are advocated in certain scenarios to obtain reasonable agreement between theory and observation . the heavy neutrino is typically taken to be the tau neutrino , and we will assume this to be the case in this section . once a decaying neutrino is required , its decay into a lighter neutrino and a massless boson is the most harmless .
visible neutrino decays are usually severely constrained from sn 1987a . as mentioned in sec .
[ sec : astrophysics ] , the energy released from sn 1987a was mostly carried away in neutrinos , and the visible luminosity of sn 1987a was much smaller . however , if neutrinos decay into visible particles , such as photons or electrons , neutrinos emitted from sn 1987a that decay before reaching the earth may increase the apparent visible luminosity of sn 1987a to levels much larger than observed @xcite . in addition , scenarios with @xmath18 neutrinos decaying into three neutrinos are also dangerous , since , in the absence of fine - tuning , such models also predict large flavor violating @xmath18 decays ( like @xmath353 ) by su(2)@xmath354 gauge symmetry . in particular , in the cosmological models to be described below , the resulting flavor - violating @xmath18 decay rates are already excluded by current bounds .
this is because , for these three - body decays , the flavor - violating @xmath18 branching fraction is of order @xmath355 ( to be compared with @xmath356 for the two - body familon decays ) , and hence is extremely enhanced for scenarios with neutrino masses in the currently allowed range .
therefore , if we adopt a massive unstable neutrino as a solution to cosmological problems , massless bosons are good candidates for its decay products .
the familon is an example of such a massless boson .
( in the literature , a particular example of a familon , the majoron , is often considered . ) if neutrinos decay to familons , cosmological scenarios , which require specific ranges for neutrino masses and lifetimes , then predict rates for familon signals in future experiments , or may even be excluded from current familon bounds , assuming an absence of fine - tuning . in this section ,
we first review some of the potentially interesting cosmological scenarios that require massive neutrinos .
then , assuming that the massive tau neutrino decays through @xmath357 , where @xmath358 , we will discuss how well these scenarios may be constrained by current and future collider searches for familons . among the several cosmological motivations for massive neutrinos
is the `` crisis '' in the standard bbn scenario .
the standard bbn scenario contains only one free parameter , the baryon to photon ratio @xmath359 ; the abundances of the light elements are predicted once we fix @xmath359 . until a few years ago , the theoretical prediction with @xmath360 was in good agreement with observations .
recently , however , it has been claimed that the predictions of the standard bbn are disfavored by observations of the light element abundances @xcite : normalizing @xmath359 with the d and @xmath361he abundances , the observed @xmath362he abundance is claimed to be smaller than the standard bbn prediction .
there are several arguments against this viewpoint on the observational side .
for example , the apparent discrepancy vanishes if one adopts a larger systematic error in the observed @xmath362he abundance @xcite , or if the recently measured d abundance in high red - shift qso absorber systems is regarded as a primordial one @xcite . on the other hand ,
if we regard this `` crisis '' as a genuine problem with the standard bbn theory , it can be taken as an indication of new physics beyond the standard model .
there are several attempts to solve this crisis by a modification of the standard scenario @xcite . here
, we concentrate on a solution that uses massive unstable neutrinos to reduce the predicted @xmath362he abundance .
since the @xmath362he abundance decreases as the energy density at the neutron freeze out time decreases , the @xmath362he abundance becomes smaller if @xmath363 , the `` effective number of neutrino species '' at the neutron freeze out time , is reduced . in the standard bbn ,
@xmath363 is 3 , but it can be smaller if heavy neutrinos decay and effectively convert their energy density into lighter particles .
for example , in the presence of the decay mode @xmath357 , if all the tau neutrinos are converted into thermal familons and light neutrinos , @xmath364 .
bbn scenarios with massive neutrinos decaying to familons have been discussed in refs . .
the most recent calculation shows that a massive neutrino with @xmath365 mev and @xmath366 sec can resolve the conflict between theory and observation @xcite .
decaying massive neutrinos are also interesting for large scale structure formation .
the standard cold dark matter ( cdm ) scenario , which assumes a flat universe , a scale - invariant initial spectrum , and that the universe is mostly filled with slowly moving ( `` cold '' ) particles @xcite , is very attractive in explaining the origin of large scale structure .
however , if the normalization of the power spectrum is fixed by the anisotropy in the temperature of the cosmic background radiation observed by cobe , the standard cdm scenario predicts too large density fluctuations at small scales ( @xmath367 mpc ) .
attempts to explain the scale dependence of the density perturbations include proposals of cdm with a small component of hot dark matter or with a cosmological constant @xcite , or scenarios with a tilted initial density fluctuation @xcite . as pointed out in refs .
@xcite , cdm with late decaying neutrinos also provides a solution to this problem .
if the neutrino lifetime is long enough , neutrinos dominate the energy density of the universe at the temperature @xmath368 . after this stage ,
the mass density of the neutrino , @xmath369 , scales as @xmath370 .
neutrinos then decay at time @xmath371 .
once they decay , the energy density of the neutrinos is converted to radiation energy density , resulting in an increase of the radiation energy density without affecting the background photons .
this then delays the time of matter - radiation equality , the matter - dominated era starts later , and , with the cobe normalization , the density perturbations at small scales are reduced . due to the neutrino decay , the energy density of the radiation is increased by the factor @xmath372 , where @xmath373 is the temperature just before the neutrino decay . as we can see
, physics ( approximately ) depends on the combination @xmath374 . to obtain the correct density fluctuation at small scales , this combination must lie in the range @xmath375 @xcite . in these scenarios ,
the neutrino mass must also lie in a specific interval .
the neutrino must be heavier than about 50 ev ; otherwise , its mass density is always smaller than ( or at most comparable to ) the mass density of the cdm , and the scenario does not work well . on the other hand ,
if the neutrino mass is above @xmath376 1 mev , the neutrinos decouple from the thermal bath after becoming non - relativistic , and their number density is reduced . in this case , the constraint given above is not applicable . furthermore
, if the neutrino mass is in the range @xmath376 1 10 mev while the lifetime is longer than @xmath376 1 sec , the neutrino mass density may be so large at the neutron freeze out time that @xmath362he can be overproduced .
such lifetimes and large neutrino masses are therefore also disfavored from bbn considerations @xcite . in this section ,
we consider the mass range @xmath377 , with the above caveats in mind .
to summarize , we consider the following two cosmological scenarios : * bbn : @xmath365 mev , and @xmath366 sec .
* structure formation : @xmath378 , with @xmath379 .
these scenarios require decays to familons @xmath380 .
as we discussed previously , this process is related to the decay modes @xmath5 and @xmath93 though su(2)@xmath354 and gut gauge symmetries , respectively .
thus , searches for these decay modes are interesting tests of these scenarios .
let us start with the @xmath18 familon decay mode .
if the decay mode @xmath380 exists , by su(2)@xmath354 symmetry , the charged @xmath18 lepton must also have flavor - changing couplings to familons : @xmath381 if there is no fine - tuning , @xmath382 .
notice that , if the right - handed leptons also transform under the flavor group , they also couple to familons , and such interactions may increase the rare @xmath18 decay rate .
the following argument is therefore conservative . from the above lagrangian
, we obtain the decay rate @xmath383 and , using eq .
( [ tau_nu ] ) , we find @xmath384 in fig .
[ fig : mvst_tau ] contours of constant @xmath385 are shown in the @xmath386 plane , assuming @xmath387 .
= 0.6 from eq .
( [ gamma_tau ] ) , we see that , in the absence of fine - tuning , the bbn scenario with decaying neutrinos predicts @xmath388 to @xmath389 .
the current bounds @xmath390 and @xmath110 @xcite therefore do not constrain this scenario . however , if the sensitivity of future experiments is improved by one order of magnitude or more , the predictions of this scenario may be tested , and , if the scenario is correct , exotic @xmath18 decays may be seen .
( see fig .
[ fig : mvst_tau ] . )
the scenarios motivated by structure formation are also interesting . in this case , eq . ( [ gamma_tau ] ) implies @xmath391 to @xmath392 . comparing this result with the current bound , part of the parameter region of this scenario
is already excluded . as discussed in sec .
[ sec : taubounds ] , future experiments may reach a sensitivity for @xmath344 of @xmath319 or possibly @xmath318 .
thus , if the cdm scenario were realized , the familon decay mode is likely to be found if the neutrino is lighter than @xmath393 kev .
the region disfavored by bbn @xcite is also shown by a light shading .
up to now , we have only used su(2)@xmath354 gauge symmetry to relate the neutrino - familon interaction to existing and future constraints on @xmath18 decays . however ,
if we assume that the same familon also couples to down - type quarks , the above cosmological scenarios may also be probed by rare @xmath19 decays .
such is the case in su(5 ) guts , where the lepton doublet and right - handed down - type quarks are in the same multiplet , and so we also have a coupling of the form @xmath394 and similarly for @xmath216 . with this lagrangian , we obtain @xmath395 where we have used @xmath396 gev , and @xmath397 sec . comparing eq .
( [ gamma_b ] ) with eq .
( [ gamma_tau ] ) , we can see that the branching ratio @xmath152 is enhanced by about two orders of magnitude relative to @xmath385 .
this results from an enhancement by a factor @xmath398 , and also the fact that the total decay rate of the @xmath19 quark is @xmath177 suppressed , and so is even smaller than that of the @xmath18 lepton , despite its larger mass .
contours of constant @xmath399 are shown in fig .
[ fig : mvst_bot ] . for the cosmologically - motivated scenarios , the ranges of @xmath399 are @xmath400 to @xmath320 ( bbn ) , and 1 to @xmath349 ( structure formation ) . because the @xmath19 decay rates are so enhanced , in each case , the high value in the predicted range is above bounds which can be expected from the current data .
future experiments may reach a sensitivity of @xmath401 ( corresponding to @xmath402 in sec .
[ sec : bbounds ] ) and will thus cover most of the preferred parameter regions .
= 0.6 finally we comment on @xmath185 measurements at future @xmath8 factories .
as we mentioned earlier , the upper bound on @xmath185 will be significantly improved at future @xmath8 factories , to the level of 1 3 mev .
the parameter region that will be covered has a significant overlap with the neutrino mass required in the above mentioned cosmological scenarios , and hence such mass measurements provide another probe of these scenarios .
the bbn scenario with massive unstable neutrinos will be fully tested by the tau neutrino mass measurement in future @xmath8 factories . on the other hand , for the structure formation scenario ,
most of the interesting parameter region ( @xmath403 mev ) may be covered by the search for @xmath404 , and even if the neutrino mass is above @xmath376 1 mev , this scenario can be checked by the direct measurement of the tau neutrino mass ( though this region is disfavored by bbn ) .
therefore , in this case , measurements of the mass and the branching ratios will have complementary roles .
if global family symmetries play a role in determining the patterns of masses and mixings of the quarks and leptons , they must be spontaneously broken .
familons ( or majorons ) , the massless goldstone bosons associated with these broken symmetries , allow rare opportunities to probe the physics at very high mass scales in a multitude of low energy settings , and their discovery will signal a breakthrough in attempts to understand the flavor structure of the standard model .
the experimental investigation of familons has in the past focussed on familons coupled to the first two generations .
as we reviewed , such investigations have led to stringent lower bounds on the flavor breaking scale @xmath13 of @xmath405 gev and @xmath406 gev in the leptonic and hadronic sectors , respectively .
in contrast , bounds for familons coupling to the third generation are much less thoroughly studied . in the lepton sector ,
constraints from rare @xmath18 decays lead to constraints @xmath407 gev ; in the hadronic sector , no bounds have been previously reported .
the lack of study of third generation familons is conspicuous , especially in light of their cosmological relevance and the upcoming @xmath8 factory experiments , which hold promise for studying @xmath19 and @xmath18 decays with great precision . motivated by these considerations , we have presented a large and eclectic group of bounds which we believe place the most stringent constraints on third generation familon couplings . as emphasized in sec .
[ sec : collider ] , the experimental and astrophysical implications of familons vary strongly with the underlying flavor symmetry and depend on whether the flavor symmetry is real or complex , the familon couplings are axial or vector - like , and whether they are flavor - diagonal or non - diagonal .
for instance , bounds on @xmath83 decay in the presence of mixing effects may in general impose bounds of order @xmath408 gev on third - generation flavor breaking scales , but there are classes of models in which such bounds do not apply .
it is therefore important to consider a wide variety of experimental signatures .
different signatures are also related through some well - motivated theoretical considerations : in the absence of fine - tuning , the familon couplings of particles related by gauge symmetry are expected to be similar in strength .
probes of @xmath18 decays to familons are therefore indirect probes of @xmath182 familon couplings , and in su(5 ) grand unified theories , these are both related to @xmath19 decays .
we began by considering bounds from currently available accelerator data .
present values for neutral meson mass splittings imply bounds on the flavor scale of @xmath409 to @xmath410 gev for real familons .
bounds from neutral meson decays to leptons are significantly weaker , at the level of @xmath411 gev , and require both hadronic and leptonic couplings .
more promising are bounds from exotic @xmath19 decays at lep . by extrapolating from current bounds on @xmath92 , we estimate that an analysis of currently available lep data may provide a sensitivity to @xmath399 at the level of @xmath412 , leading to probes of flavor scales of the order of @xmath413 gev .
familons also have astrophysical implications , as they may lead to anomalously fast cooling of supernovae , red giants , and white dwarfs .
bounds from direct couplings to @xmath210 are generally weak .
however , couplings of familons to particles of the third generation may also induce couplings to electrons and nucleons radiatively or through flavor mixing effects .
bounds on such couplings are model - dependent , but may be stringent ; in the simple case where a familon couples diagonally to @xmath134 quarks , a bound of @xmath414 gev from radiatively induced couplings may be set . finally , having evaluated a host of new bounds , we considered the prospects for analyses at future @xmath8 factories .
such colliders are ideal experimental environments for searches for rare @xmath18 and @xmath19 decay modes and are expected to have greatly improved statistics .
we find that probes of branching fractions of @xmath319 ( @xmath320 ) for @xmath18 ( @xmath19 ) decays may be possible . as discussed in sec .
[ sec : cosmology ] , under the assumption that the flavor scales for these decays are naturally related to the scales for @xmath210 couplings , such precise probes are sensitive to parameter regions favored by various bbn and structure formation scenarios , where a massive unstable neutrino is motivated by possible discrepancies in cosmological data .
in fact , parts of the parameter regions in such scenarios are already excluded , and future searches will be able to explore large portions of the cosmologically - favored parameter space . given the present lack of analyses studying third generation familon couplings , studies at all of these experiments , and particularly the @xmath8 factories , are strongly encouraged . after the completion of this work , ref .
@xcite appeared , in which the impact of decays @xmath415 on bbn was examined .
tau neutrino masses of 0.1 to 1 mev and lifetimes of 50 to @xmath416 s were found to be allowed , implying rates for @xmath417 and @xmath418 within reach of future experiments .
( see figs .
[ fig : mvst_tau ] and [ fig : mvst_bot ] . )
we thank l. hall and m. suzuki for discussions and r. patterson , a. seiden , and s. stone for valuable correspondence .
this work was supported in part by the director , office of energy research , office of high energy and nuclear physics , division of high energy physics of the u.s .
department of energy under contract de ac0376sf00098 , and in part by the national science foundation under grant phy9514797 .
j.l.f . and t.m .
thank the aspen center for physics for hospitality while this work was being completed .
j.l.f . is supported in part by a miller institute research fellowship .
is also supported by the alfred p. sloan foundation .
is supported by a basf research fellowship and the studienstiftung des deutschen volkes .
m. girone , aleph collaboration , talk presented at the international europhysics conference on high energy physics , 1926 august 1997 , jerusalem , israel , session 10 , talk # 1003 , http://www.cern.ch/hep97/talks/t1003.ps.gz .
m. kawasaki , p. kernan , h .- s .
kang , r. j. scherrer , g. steigman , and t. p. walker , nucl . phys .
* b419 * , 105 ( 1994 ) .
d. a. dicus , e. w. kolb , and v. teplitz , phys .
* 39 * , 168 ( 1977 ) ; m. fukugita and t. yanagida , phys .
b * 144 * , 386 ( 1984 ) ; g. gelmini , phys .
b * 146 * , 311 ( 1984 ) ; m. s. turner , g. steigman , and l. krauss , phys .
lett . * 52 * , 2090 ( 1984 ) .
g. g. raffelt , phys .
d * 33 * , 897 ( 1986 ) ; d. s. p. dearborn , d. n. schramm , and g. steigman , phys .
56 * , 26 ( 1986 ) ; t. altherr , e. petitgirard , and t. del rio gaztelurrutia , astropart .
* 2 * , 175 ( 1994 ) , hep - ph/9310304 .
m. c. gonzalez - garcia , invited talk at the 3rd workshop on the tau - charm factory , marbella , spain , 1 - 6 jun 1993 ; r. alemany , j. j. gomez - cadenas , m. c. gonzalez - garcia , and j. w. f. valle , cern
ppe9349 , hep - ph/9307252 .
r. davis et al . , phys .
rev . lett . * 69 * , 1856 ( 1992 ) ; a. liddle and d. lyth , phys .
b * 291 * , 391 ( 1992 ) ; f. lucchin _ et al . _ ,
j. * 401 * , l49 ( 1992 ) ; j. gelb _ et al .
_ , _ ibid .
_ * 403 * , l5 ( 1993 ) ; r. cen _ et al .
_ , _ ibid .
_ * 399 * , l11 ( 1992 ) . | we study the physics of spontaneously broken family symmetries acting on the third generation .
massless familons ( or majorons ) @xmath0 associated with such broken symmetries are motivated especially by cosmological scenarios with decaying tau neutrinos .
we first note that , in marked contrast with the case for the first two generations , constraints on third generation familon couplings are poor , and are , in fact , non - existent at present in the hadronic sector .
we derive new bounds from @xmath1@xmath2 mixing , @xmath3 , @xmath4 , and astrophysics .
the resulting constraints on familon decay constants are still much weaker than those for the first and second generation .
we then discuss the promising prospects for significant improvements from searches for @xmath5 , @xmath6 , and @xmath7 with the current cleo , argus , and lep data .
finally , we note that future constraints from cleo iii and the @xmath8 factories will probe decay constants beyond @xmath9 gev , well within regions of parameter space favored by proposed scenarios in neutrino cosmology . |
let @xmath8 denote a subsequence of @xmath0 ternary symbols , @xmath9 . then the conditional probability , @xmath10 , that the @xmath11th symbol @xmath12 in the sequence will be @xmath13 or @xmath14 may not only depend on the number of individual symbols but , in general , may also depend on their specific order in @xmath15 . ignoring the configuration - dependence of @xmath10 leads to a solvable mean - field - like theory of these ternary sequences .
we define the conditional probability , @xmath16 as follows : @xmath17 where @xmath18 , @xmath19 and @xmath20 denote the number of @xmath21 s , @xmath22 s and @xmath14 s respectively in the sequence such that @xmath23 , and @xmath24 , @xmath25 and @xmath26 denote the _ a priori _ probabilities of choosing the respective symbols of which only two are _ independent _ because @xmath27 .
we can parametrize the deviations of @xmath28 from their unbiassed values @xmath29 by writing @xmath30 where @xmath31 $ ] and @xmath32 .
it is clear that the @xmath33 s are a measure of the memory built into the system which in turn leads to correlations between the symbols . since one of the three symbols is definitely going to be found at any given place in the subsequence , _
i.e.,_@xmath34 , we need only to define @xmath35 or @xmath36 .
now , the probability that the @xmath37 symbol is _ not _
@xmath21 is given by @xmath38 where @xmath39 ; \quad q_2 \equiv \frac { 1}{n}\left [ n_0 g_2 + n_1 g_0 + n_2 g_1 \right]\ ] ] so that we may identify @xmath35 either with @xmath40 or with @xmath41 .
that is to say , we may either have @xmath42 or have @xmath43 , where we define the @xmath44-complementary of @xmath15 as @xmath45 , with @xmath46 mod @xmath47 .
we choose the first definition : @xmath48 clearly , these definitions , eqs.(1 , 4 & 5 ) , ensure that @xmath49 . the other choice for @xmath35
can be shown to lead to the same result with a simple coordinate transformation .
taking @xmath21 and @xmath22 as the independent symbols , we can write eqs.(1 & 4 ) in the form , @xmath50 @xmath51}{n}\mu _ { 1}\right)\ ] ] where @xmath52 $ ] and @xmath53 $ ] . a ternary sequence of @xmath0 symbols is completely described by the probability , @xmath54 , that there are @xmath18 number of zeroes and @xmath19 number of ones in the sequence : @xmath55 since the average number of any symbol in the sequence will be @xmath56 asymptotically ( _ i.e. , _ no global bias in the system ) , we rewrite the above equation in terms of the variables , @xmath57 and @xmath58 . in doing so , we make use of the correspondence @xmath59 , @xmath60 and @xmath61 as can be seen from the definitions of @xmath62 and @xmath63 .
@xmath64)q(x , y;n)\end{aligned}\ ] ] where the probabilities @xmath65 and @xmath66 are given by , @xmath67}{3n}x + \frac{[\mu _ {
0}+2\mu _ { 1}]}{3n}y \right ) \quad \\ & & \nonumber \\ p_1 ( x , y;n ) & = & \frac{1}{3 } \left ( 1+\frac{[\mu _ { 1}-\mu _ { 0}]}{3n}x - \frac{[2\mu _ { 0}+\mu _ { 1}]}{3n}y \right)\end{aligned}\ ] ] the continuum version of eq.(9 ) is then given by @xmath68}{\partial x } + \frac{\partial [ ( \lambda x -\lambda _ { 0}y ) q]}{\partial y}\right)\end{aligned}\ ] ] where @xmath69 and @xmath70 represent the diffusion and drift constants respectively .
we have introduced the parameter @xmath71 to allow for the possible existence of transient time in the problem .
the @xmath72 s are defined by , @xmath73 a standard method of solving eq.(12 ) , with the initial condition @xmath74 , is to first fourier transform it with respect to the variables @xmath62 and
@xmath63 : @xmath75 \frac{\partial { \tilde q}}{\partial q_x } + [ \lambda _ 1 q_x - \lambda _ 0 q_y ] \frac{\partial { \tilde q}}{\partial q_y } \right)\ ] ] where @xmath76 and @xmath77 are the fourier conjugates of @xmath62 and @xmath63 respectively . this first order equation can then be solved by the method of characteristics .
in particular , we have to solve the equations , @xmath78 considering the second and the third terms leads to the equation , @xmath79 that can be solved for @xmath80 in terms of @xmath81 or _ vice versa _ : @xmath82^{1/2}\ ] ] using these relations , we can immediately obtain their @xmath0-dependence .
@xmath83^{1/2}\ ] ] which in turn helps us define the variables , @xmath84 in terms of which eq.(14 ) can be written as @xmath85 where @xmath86 . with the initial condition , @xmath87
, we immediately have the solution , @xmath88 from which we identify the variance , @xmath89 \frac{\tau}{(1 - 2\gamma ) } ; \quad 2\gamma \neq 1\ ] ] it is clear from the above expression that @xmath90 , where @xmath91 whenever @xmath92 and @xmath93 whenever @xmath94 . even in the case ,
@xmath95 , the exponent @xmath91 but there will be logarithmic corrections .
the existence of a critical value , @xmath96 , for @xmath97 implies that the parameter space , @xmath98 , is divided into two regions , diffusive and superdiffusive .
as shown in fig.1 , the diffusive region defined by the condition @xmath99 is the elliptical region , @xmath100 , inscribed within the triangular phase - space .
interestingly , the area of this diffusive region is @xmath101 which is roughly @xmath102 of the available phase - space area .
this may be contrasted with the binary case where it is exactly @xmath103 .
now the question arises whether a diffusive subregion of the ternary is likely to be mapped into a superdiffusive region of the binary due to a coarse - graining process .
since a mapping of a set of three symbols to a set of two symbols always introduces a global bias in the resulting binary system , it is necessary to reformulate the binary case so that we can identify long - range correlation in a biassed sequence . and
@xmath104 refer to the lines @xmath105 and @xmath106 respectively .
the ellptical region of the triangular phase space corresponds to the _ diffusive _ behavior whereas the region exterior to it corresponds to the _ superdiffusive _ behavior of the ternary sequences .
the entire region between the lines @xmath107 and @xmath104 , inclusive of the shaded elliptical region , gets mapped into the superdiffusive regime of the binary.,width=312,height=264 ]
the conditional probability , @xmath108 , of finding zero as the @xmath109 symbol in a binary sequence that already consists of @xmath18 zeros is given by the definition , @xmath110g_1 \right)\ ] ] where @xmath24 and @xmath111 are the _ a priori _ probabilities of choosing symbols zero and one respectively . if a biassed coin ( _ i.e. , _
@xmath112 ) is used for building up a sequence , then the average number of zeros in the sequence , @xmath113 , is expected to be @xmath114 asymptotically , where the intrinsic bias @xmath115 is equal to @xmath24 if and only if the symbols are added blindly without reference to the existing sequence ( _ i.e_. , @xmath116 does not depend on @xmath18 ) . in order that the equality @xmath117 holds good in general , it is necessary that the dependence of @xmath116 on @xmath18 is through the variable , @xmath118 . in analogy with the unbiassed case [ 4 ] ,
we write @xmath119 where @xmath120 $ ] pametrizes the deviation from the intrinsic bias of the coin .
we then have @xmath121 in order to check whether the equality @xmath117 holds good for @xmath33 in the range @xmath122 $ ] , we have generated [ 8 ] a large number of sequences each consisting of upto hundred thousand symbols for @xmath123 .
we find that @xmath124 for @xmath125 and decreases fast to the value @xmath126 for @xmath127 . on the other hand ,
the variance @xmath128 turns out to be proportional to @xmath129 for @xmath130 , and to @xmath0 for @xmath131 .
this implies that @xmath44 , defined as above , does not lead to the expected constant value for @xmath132 , whereas @xmath133 correctly defines the correlation exponent .
since the above definition of @xmath44 is derived from eq.(23 ) for @xmath116 , we have to check whether eq.(23 ) is a valid definition also for the biassed case . treating @xmath44 as a parameter to be fixed later
, we can easily show that the probability , @xmath134 , that there are @xmath18 zeros in the sequence satisfies the following equation in the continuum limit : @xmath135}{\partial x}\ ] ] solving this equation , we can show that the distribution , @xmath136 , peaks at @xmath137 asymptotically where @xmath138 $ ] .
thus , the peak will remain fixed at @xmath139 for all values of @xmath33 only if @xmath140 .
this leads to the following definition for @xmath116 : @xmath141 \right ) = 2b(1 - g_0 ) + ( 2g_0 - 1)\frac{n_0}{n}\ ] ] it is a general definition applicable even to an unbiassed sequence ( _ i.e. , _ reduces to the definition , eq.(23 ) , for @xmath142 ) .
the variance of the distribution is given by , @xmath143 since @xmath119 , long - range correlation in the sequence will be characterized by the values of @xmath33 in the range @xmath144 , -1 + [ 1/b])$ ] .
again , we have numerically checked [ 8 ] that the condition @xmath140 ( @xmath145 in our case ) does ensure the constancy of @xmath146 and , more importantly , we have confirmed that the variance has the above expected behaviour . in other words ,
the lrc behavior of a binary sequence is characterized by the exponent , @xmath147 , for @xmath130 .
assume that the symbols zero and one of the ternary are identified as @xmath21 while the symbol two is identified as @xmath22. then the _ a priori _ probability for @xmath21 and @xmath22 will be @xmath148 the condition for non - trivial correlations then turns out to be @xmath149 , above the lower line shown in fig.1 .
the entire region of the phase space between this line and the upper line , @xmath105 , corresponds to long - range correlation when mapped into the biassed binary .
this is so because the correlation parameter , @xmath33 , for the binary is related to @xmath150 and @xmath151 by the identity , @xmath152 which in turn implies a correspondence between the range @xmath153 of the ternary and the range @xmath154 of the biassed binary ; every line in between and parallel to the upper and lower lines in fig.1 is mapped into a point in the range @xmath154 and _ vice versa_. in particular , the shaded part of the ellipse in fig.1 is the diffusive area that is mapped into the superdiffusive regime of the biassed binary , which in general is characterized by a parameter , @xmath33 , in the range @xmath144 , -1 + [ 1/b])$ ] .
conversely , the fact that the correlation parameter of the biassed binary sequence under study has a value in the range @xmath144 , -1 + [ 1/b])$ ] does not necessarily mean that the parent ternary sequence also has long - range correlations . even if we know
_ a priori _ that there are long - range correlations between symbols of the parent sequence , the parameter @xmath33 is not a true measure of its strength .
we have worked out the exact phase diagram for a ternary sequence with long - range correlation .
motivated by the fact that the diffusive portion of the phase space is larger for the ternary than for the binary , we have studied the mapping between these two .
we have shown that long - range correlation for the binary does not necessarily imply long - range correlation for the ternary .
this exact result has deeper implications for the coarse - graining of many - alphabets sequences [ 8 ] .
for example , if we do not know that the original sequence has long - range correlations or ( and ) if we do not know the coarse - graining procedure that has led to the representative sequence under study , then we may not be able to make accurate inferences about the correlation properties of the original sequence .
a systematic numerical study of this problem will be reported elsewhere .
it could be of interest to do a similar study for a non - markov sequence .
kanter i. and kessler d. a. , phys .
* 74 * ( 1995 ) 4559 ; kokol p. , brest j. and zumer j. , cybernetics and systems , * 28*(1 ) ( 1997 ) 43 ; ebeling w. and poschel t. , europhys
* 26 * ( 1994 ) 241 , and references therein .
kak m. and logan j. , in _ fluctuations phenomena ( studies in statistical physics ) _ ed . by montroll e. w. and lebowitz j. l. , north - holland publishing company ( 1979 ) , page 10 , point out that a coarse - grained description may not preserve the markovian character of the process under study . | we present an exactly solvable mean - field - like theory of correlated ternary sequences which are actually systems with two independent parameters . depending on the values of these parameters , the variance on
the average number of any given symbol shows a linear or a superlinear dependence on the length of the sequence .
we have shown that the available phase space of the system is made up a diffusive region surrounded by a superdiffusive region .
motivated by the fact that the diffusive portion of the phase space is larger than that for the binary , we have studied the mapping between these two .
we have identified the region of the ternary phase space , particularly the diffusive part , that gets mapped into the superdiffusive regime of the binary .
this exact mapping implies that long - range correlation found in a lower dimensional representative sequence may not , in general , correspond to the correlation properties of the original system .
the dynamical behavior of complex systems consisting of a large number of hierarchically organized subsystems is known to carry the signature of long - range spatio - temporal correlations ( lrc ) that might be present in them .
a variety of physical [ 1 ] , biological [ 2 ] , linguistic [ 3 ] and even financial systems [ 2 ] provide examples of lrc systems . a standard method for studying correlations in such
a system is to first divide its state space into a finite number of distinctly labelled regions , map the sequence of states assumed by the system into a sequence of these symbols and then study the statistical properties of this representative sequence .
since such a _ coarse - graining _ procedure is not expected to lead to a loss of long - range correlations in the system , we may choose to divide the state - space into two distinct regions and study the statistical properties of the resulting binary sequence .
in fact , we generate a correlated binary sequence , parametrizable by a known procedure , and try to find the parameter values that correspond to the representative sequence .
recently , it has been shown [ 4 ] that the strength of long - range correlations in a binary sequence of length , @xmath0 , can be characterized by a parameter @xmath1 $ ] so that non - trivial correlations between the symbols correspond to the parameter region , @xmath2 .
more specifically , the variance , @xmath3 , of the the number of a symbol has been shown to have a form , @xmath4 , where @xmath5 for @xmath6 ( _ diffusive _ ) and @xmath7 for @xmath2 ( _ superdiffusive _ ) [ 5 ] .
this exact mean - field - like theory of correlated binary sequences seems to provide a paradigm for studying the correlational properties of generic symbolic sequences such as even natural language texts .
the implicit assumption in this approach is that an lrc system can be represented by a correlated binary sequence .
there may be no _ a priori _ reason why this assumption should hold good .
for example , it has been argued [ 6 ] that we need a minimum of ten letters ( not less than five letters , in any case ) to be able to design a foldable model of amino acid sequences .
so , the minimum number of symbols required for a sequential representation of the system ( or equivalently , the extent to which the state - space of a system can be coarse - grained ) may depend on the specific behaviour of the system under study . even if a binary sequential representation is acceptable , the problem of how to arrive at this representation remains if the number of symbols originally associated with the system is odd .
thus , it is necessary to examine whether a coarse - graining procedure could introduce long - range correlation , besides the one that might be present in the sequence already [ 7 ] . to this end
, we present an exact mean - field - like theory of correlated ternary sequences and identify the diffusive subregion of its phase space that gets spuriously mapped into the superdiffusive region of the phase space associated with the binary sequence .
this exact mapping implies that long - range correlation found in a lower dimensional representative sequence may not , in general , correspond to the correlation properties of the original system . |
the origin of supersonic jets from young stellar objects ( ysos ) remains unclear .
the current consensus holds that magnetic fields tied to either an accretion disk or the star - disk boundary can launch _ and _ collimate material into a jet .
these magneto - centrifugal " scenarios have been explored analytically in a variety of configurations including the popular disk - wind ( @xcite ; @xcite ) and x - wind ( @xcite ) models .
numerical simulations of magneto - centrifugal mechanisms have shown mixed but promising results .
@xcite demonstrated that disk wind models can produce well collimated jets when the feed - back on the disk is ignored . @xcite
have shown that magneto - centrifugal mechanisms can launch winds ; however , in their simulations the winds do not collimate into jets .
the potential difficulties involved in turning magneto - centrifugal winds into jets have been noted before .
@xcite discussed the slow ( logarithmic ) rate at which mhd models produce collimation .
@xcite demonstrated that x - wind models will tend to produce wide - angle winds rather than discrete cylindrical jets .
along with these issues , recent numerical studies have shown that pure hydrodynamic collimation can be surprisingly effective at producing well collimated supersonic jets .
frank & mellema ( 1996 , 1997 ) and mellema & frank ( 1997 ) have demonstrated that isotropic or wide angle yso winds interacting with toroidal density environments readily produce oblique inward facing wind - shocks .
these shocks can be effective at redirecting the wind material into a jet .
if the wind from the central source is varying , this `` shock focusing '' mechanism can , in principle , produce jets on the observed physical scales ( mellema & frank 1997 ) .
similar mechanisms have been shown to work in other jet - producing contexts as well ( peter & eichler 1996 ; frank , balick & livio 1996 ; borkowski , blondin & harrington 1997 ) .
the variety of available collimation models begs the question of which process actually produces protostellar jets .
observations that can distinguish signatures of different theoretical models are obviously needed . unfortunately seeing into the collimation region
is difficult .
a number of studies indicate that collimation occurs on scales of order @xmath4 ( burrows 1996 ; ray 1996 ) , at or below current observational limits .
in addition the many magnitudes of extinction common for star forming systems often obscure the innermost region where jets form
. thus critical observations concerning the formation of the jets will have to come from downstream of the collimation regions ; i.e. from the jets themselves . if the collimation process is mhd dominated , then magnetic fields will remain embedded in the jets as they propagate . in particular both disk - wind and x - wind models of jet collimation will produce jets with strong toroidal fields .
this is evidenced by the fact that while the sonic mach numbers of the jets may be high ( @xmath5 ) the fast mode mach numbers will be low ( @xmath6 , @xcite ) .
direct observation of magnetic fields in protostellar jets would help clarify issues surrounding jet origins .
unfortunately such measurements have generally proven to be difficult to obtain ( ray 1997 ) .
a promising alternative is look for less direct tracers of strong fields in yso jets .
if jets are produced via mhd processes , dynamically significant magnetic stresses should affect the beam and jet head as they interact with the environment .
thus the propagation characteristics of protostellar jets may hold important clues to their origins . in this letter we present the first results of a campaign of radiative mhd simulations of yso jets .
the goal of our ongoing study is to search for observable characteristics that distinguish hydrodynamic from magnetohydrodynamic jets .
here we present models that articulate significant radiative mhd effects while also making contact with the extensive bibliography of previous numerical jet studies .
radiative mhd jets are the next logical step in the explication of astrophysical jet dynamics .
we have deliberately used simplified initial conditions in our simulations .
our initial set - up , similar to those used past studies , demonstrates new features introduced by the interaction of radiative losses and magnetic stresses in the context of those aspects of jet physics that are well understood .
beginning with the work of norman and collaborators the behavior of axisymmetric jets ( @xmath0-d ) without radiative losses has been successfully cataloged and explained ( norman 1993 and references therein ) .
the dynamics of the _ bow shock _
( which accelerates ambient material ) , _ jet shock _ ( which decelerates material in the jet beam ) , and _ cocoon _ ( decelerated jet gas surrounding the beam ) have been well - studied in these simulations .
one should note that these investigations have tended to focus on so called `` light '' extra - galactic jets where the density of the material in the jet beam ( @xmath7 ) is lower than that in the ambient medium ( @xmath8 ) .
the jets emanating from ysos are , however , thought to be `` heavy '' in the sense that the ratio @xmath9 .
another fundamental difference between extra - galactic and yso jets is the presence in the latter of strong post - shock emission from collisionally excited atomic and molecular lines .
thus , while extra - galactic jets can be considered `` adiabatic '' , yso jets must be considered `` radiative '' . beginning with the work of blondin , knigl & fryxell ( 1990 )
the dynamics of radiative jets has been explored in considerable detail ( stone & norman 1994 ; raga 1994 ; suttner 1997 ) .
these simulations all show that as pressure support is lost , the bow shock / jet shock pair collapse into a thin shell . they also revealed dynamical and thermal instabilities associated with this shell .
mhd simulations of non - radiative jets were first carried out by clarke , norman & burns 1986 using a strong , purely toroidal magnetic field @xmath10 .
their results showed that `` hoop '' stresses associated with the radially directed tension force inhibit sideways motion of shocked jet gas .
material that would have spilled into the cocoon is forced into the region between the jet and bow shock , forming a `` nose - cone '' of magnetically dominated low @xmath11 gas ( @xmath12 ) .
hoop stresses also collapse the beam near the nozzle , producing strong internal shocks .
lind ( 1989 ) performed similar calculations , initializing their simulations with a jet in hydromagnetic equilibrium and presenting a more complete exploration of parameter space .
they confirmed that nose - cones form in jets with low initial @xmath11 .
kssl , mller & hillebrandt ( 1990ab ) explored mhd jets with a variety of initial field configurations ( poloidal , toroidal , and helical ) .
cases with significant toroidal fields always developed nose - cones .
cases with poloidal fields developed loops with field reversals in the cocoon susceptible to tearing mode instabilities and reconnection .
we note studies of mhd instabilities in _ just _ the jet beam have also been carried out ( hardee , clarke & rosen 1997 ) mhd simulations of heavy yso jets have been carried out by todo 1992 .
these models did _ not _ include radiative losses .
their jets showed similar forms to those simulated by kssl mller & hillebrandt 1990a , but todo 1992 also were able to identify the presence of both slow and fast mode shocks in the jet and to present a more quantitative analysis of stability issues .
very recently , cerqueira 1997 , in a study parallel to ours , have reported the first 3-d simulations of radiative mhd jets using an sph code .
those simulations show results that are similar to what we present below .
our work and that of cerqueira are complimentary in that very different methods are used .
their study was carried out in 3-d , while ours is @xmath13-d at this point .
however , our grid - based axisymmetric simulations have a factor 10 higher resolution . especially in radiatively unstable structures ,
high numerical resolution is important to capture the dynamics properly .
our simulations evolve the equations of ideal mhd in cylindrically symmetric coordinates ( @xmath14 ) .
since all three components of vector fields are accounted for the model is @xmath0-d .
the mhd code used is based on the mhd version of the total variation diminishing method .
it is an extension of the second - order finite - difference , upwinded , conservative scheme , originally developed by harten ( 1983 ) .
the mhd code is described in @xcite ( 1-d version ) , @xcite ( multidimensional cartesian version ) , and @xcite ( multidimensional cylindrical version ) .
the code contains routines that maintain the @xmath15 condition at each time step .
however , since the simulations described actually contain only the toroidal component of magnetic field in @xmath0-d , @xmath15 is trivially satisfied although @xmath15 is not enforced in the code .
cooling is calculated from look - up tables for a coronal cooling curve @xmath16 taken from @xcite .
full ionization is assumed and the cooling is applied in between hydro time - steps via an integration of the thermal energy @xmath17 ( @xcite ) .
tests show the method can recover steady state radiative shocks to within @xmath18 accuracy when the cooling region is resolved .
a floor " on the temperature is set at @xmath19 .
the code has been extensively tested in 1.5 dimensions ( franklin , noriega - crespo & frank 1997 ) with and without cooling and against uchida et al . simulations ( 1992 ) with satisfactory results .
we have performed simulations that compare the evolution of 4 cases : a non - radiative weak field jet ; a non - radiative strong field jet ; a radiative weak field jet ; a radiative strong field jet . since the first three cases have been studied before , our goal was to confirm that the code recovers features seen in previous investigations available in the literature and to extend the sequence into the radiative , strong - field regime . in each simulation
the jet was driven into the computational domain ( @xmath20 ) as a fully collimated supersonic / super - fast mode beam @xmath21 .
the properties of the jet common to all the simulations were as follows ( @xmath22=jet , @xmath23=ambient medium ) : @xmath24 ; @xmath25 ; @xmath26 ; @xmath27 .
the initial gas pressure in the jet @xmath28 , and hence @xmath29 are varied radially to obtain hydromagnetic equilibrium .
thus the sonic mach number @xmath30 must be defined as a radial average of @xmath31 . in all our simulations
@xmath32 the computational domain spanned @xmath33 with a jet radius @xmath34 or 30 grid cells . in all the simulations a magnetic field
was imposed in the _
jet only_. the field was purely toroidal . this simplification can be justified on theoretical grounds ( camenzind 1997 ; @xcite ) , since most disk - wind models rely on a dominant toroidal field to produce tightly collimated jets .
hydromagnetic equilibrium between gas and magnetic pressure in the jet was imposed as an initial condition ( priest 1983 , begelman 1997 ) after choosing a form for @xmath35 the equilibrium condition is solved for the initial radial gas pressure distribution @xmath36 . in our simulations
we used the same @xmath35 and @xmath36 as lind ( 1989 ) .
as with the sonic mach number the alfveic mach number @xmath37 for the jet is a radial average of @xmath38 .
the fast mode , sonic and alfvenic mach numbers are related by @xmath39 . for our strong field simulations the average initial value of the field is @xmath40 @xmath41 and an average plasma beta @xmath42 . it should be noted that this configuration , known as a z - pinch in the plasma physics community , is almost always unstable to both pinch and kink modes . since our simulations
are axisymmetric we are unable to track kink modes ( but , see cerqueira 1997 ; todo 1993 ) .
the pinch modes are quite important , however , and , as begelman 1997 has shown , the beam will be unstable to these instabilities when @xmath43 where @xmath44 is the ratio of specific heats .
this condition is satisfied for the lind 1989 initial configurations .
we note that since these simulations include radiative cooling @xmath45 quickly flattens out as the jet propagates .
we first present an equation to estimate the speed of the jet head , @xmath46 ( the bow shock ) that accounts for magnetic ( and gas ) pressure effects .
we define @xmath47 and @xmath48 where @xmath49 is effective jet head radius over which the ram pressure , @xmath50 is applied .
> from 1-d momentum balance , ignoring the ambient pressure , we find @xmath51 where @xmath52 is the familiar expression for the speed of a `` cold '' , pressure - less jet ( dal pino & benz 1993 ) , and the remaining terms describe an enhancement factor " for the head speed due to finite jet pressure .
this factor in equation [ bige ] is _ always greater than or equal to unity _ , as we should expect . for the limiting case @xmath53 equation [ bige ]
gives @xmath54 . in fig 1 and 2
we present the results of our simulations with @xmath55 .
we will use equation 4.1 to interpret the results presented below .
* a ) non - radiative weak field jet * : here the magnetic field is set quite low so that alfvnic mach number is @xmath56 .
fig 1 shows clearly the well - known bow shock / jet shock configuration at the head of the jet ( norman 1993 ) .
the scale of the bow shock is relatively large because of the post - shock thermal pressure .
note however that the speed of the jet ( @xmath57 ) is relatively low compared with extragalactic jets .
thus the post _ jet _ shock temperature in this simulation is low ( @xmath58 ) .
the pressure in the cocoon is therefore also small compared with extragalatic jet simulations and only weak internal waves are forced into the the jet beam .
blondin , fryxell & knigl 1990 found a similar result in their comparisons of non - radiative and radiative yso jet simulations . *
b ) non - radiative strong field jet * : here a stronger field is used so that the alfvnic mach number is @xmath59 ( @xmath60 ) . in this simulation
the structure of the jet head has changed dramatically due to dynamical influences of the field .
note the large separation of the jet and bow shocks . as has been described in other mhd jet studies ( clarke 1986 ; lind 1989 ; kssl 1990ab ) such `` nose - cone '' structures occur because of the pinching effect of the magnetic hoop stresses .
post jet - shock material , which would otherwise back flow into the cocoon , is forced to remain between the two shocks .
the bow shock is accelerated forward producing higher propagation speeds ( note the times at which the images where taken ) . in previous studies
the acceleration was attributed to the magnetic pressure in the nose cone ( kssl 1990b ) .
we demonstrate below that the increased speed can be better attributed in this case to a cross section effect produced by magnetic pinch forces .
note the strong convergence occurring just beyond the jet nozzle and the reflected shocks in the beam downstream .
this occurs due to the pinch instability described in section 3 and has been seen in all mhd jet simulations with strong toroidal fields .
we note also that these results look quite similar to the those presented by kssl 1990b ( see their figure 10a ) .
* c ) radiative weak field jet * : the magnetic field in this case is the same as case a. again the standard bow shock / jet shock configuration is apparent .
maps of temperature show that the post - bow / jet shock region is nearly isothermal .
the loss of post - shock thermal energy reduces the transverse width of the bow shock . densities behind the shocks become high with compression ratios of @xmath61 ( compared to the value @xmath62 obtained in the non - radiative case ) .
animations of this model show that the structure at the jet head is highly time - dependent with the region between the two shocks becoming quite thin at times . at late times in the simulation
the jet head undergoes the non - linear thin shell instability ( vishniac 1994 ) , which also has been observed in other radiative jet simulations ( blondin , fryxell & knigl 1990 ; stone & norman 1994 ) .
note again the lack of structure in the jet beam . in both cases
a ) and c ) the jet heads propagate at a velocity @xmath63 km which compares well with @xmath64 predicted by equation [ bige ] for a cold jet ( @xmath65 ) with @xmath66 . *
d ) radiative strong field jet * : in this simulation the field is the same as for the non - radiative , strong field jet ( case b ) and the structure of the jet head is similar to what obtains in that simulation .
a narrow bow shock appears some distance ahead of a high density region associated with the jet shock .
note that the loss of thermal energy behind the shocks has reduced the scale of the bow shock @xmath49 compared with its non - radiative twin .
fig 2 shows the evolution of this simulation at five different times . for this case the initial collapse of the beam just beyond the nozzle produces a series of strong shocks and reflections in the jet as it propagates down the length of the grid .
fig 2 shows the periodic density enhancements in the beam formed from these reflections .
if such features are not a consequence of the imposed axi - symmetry they may have important consequences for the emission characteristics of real yso jets .
figs 1 and 2 show that a jet shock forms close to the bow shock .
the reduced width of the jet shock / bow shock pair is similar to what is seen in the evolution of the weak field radiative jet ( blondin , fryxell & knigl 1990 ) .
the most important conclusion to be reached from these simulations comes from comparison of the third and fourth panels of fig 1 .
for radiative jets , the weak and strong field cases look dramatically different from each other in terms of the morphology of the jet head and beam . in the weak field case
there are no shocks in the beam .
the strong field case shows multiple shock reflections ( fig 2 ) .
the head of the weak field jet is quite `` blunt '' compared with the strongly tapered strong field jet .
the average propagation speed for the head of the strong field jet is @xmath67 , a 30% increase over the weak field case , even though both simulations have the same jet / ambient density ratio , @xmath68 .
the combination of higher shock speeds and strong pinch forces in the radiative strong field jet produces a compression ratio in the head almost twice as large as in the radiative weak field case .
if these results are born out in more detailed studies , particularly those in 3-d , then indirect diagnostics of the presence of dynamically strong fields in jets should exist .
as noted , the increased speed of mhd jet heads has sometimes been attributed to magnetic pressure in the nose - cone .
equation [ bige ] , however , demonstrates that the increased momentum provided by finite pressure alone can not account for the enhanced head speed in the mhd cases we have simulated . using @xmath55 , @xmath69 , @xmath70 appropriate to our simulations
, this relation gives @xmath71 , assuming the same head - scale factor @xmath72 that was used successfully to estimate the jet head speeds for our non - mhd simulations .
in fact , the finite pressures nominally add only about 2% to the jet head speeds according to equation [ bige ] . on the other hand
, it is apparent that the tapered shape of the nose cone formed by magnetic hoop stresses has streamlined the flow around the jet head .
that effectively increases the geometry factor , @xmath73 , which can significantly enhance the jet head speed , @xmath46 . from the simulation data we estimate in case d that the radius of the jet head is @xmath74 , leading to @xmath75 .
inserting this into equation [ bige ] gives a jet head speed of @xmath76 , closer to the @xmath77 estimated from the simulation itself .
the results of these simulations demonstrate that propagation - based diagnostics for radiative mhd jets may exist .
the strong effect of magnetic pinches in these axisymmetric calculations change both the structure of the jet head and the beam .
the increased velocity of the bow shock is also a distinctive feature of mhd jets .
these morphological and kinematic characteristics would alter the observed emission properties in a real yso jet .
the strong shocks in the beam would produce increased excitation of both atomic and molecular lines .
the increased speed of the jet head will alter both the degree of ionization and excitation .
this study is too preliminary , however , to provide observers with a definitive accounting of the differences between real radiative hydrodynamic and mhd jets .
we leave this task to future studies ( frank 1998 ) .
the most obvious deficit in our models is the imposed axisymmetry .
cerqueira 1997 have recently reported 3-d sph calculations which also show that strong toroidal field components dramatically alter the morphology of the jets , consistent with our results .
their numerical resolution was too low , however , to see the detailed structure of shocks in both the beam and the jet head , which are both clearly captured in our simulations .
since these kinds of structures are not seen in observations , cerqueira 1997 ( who did observe the narrowing of the jet head due to pinch forces ) concluded that real yso jets can not have significant toroidal fields .
if future high resolution grid - based and sph based codes continue to find such effects , that may pose a serious challenge to mhd jet models that rely on collimation via hoop stresses .
it is also worth noting the role of reconnection .
the initial conditions used in our models do not allow for field reversals to occur . if a poloidal ( @xmath78 ) component exists , then field lines embedded in the beam will be decelerated upon passage through the jet shock .
if the fields and cooling are not strong enough to inhibit the formation of a cocoon , these lines will be carried backwards forming field reversals ( frank 1997 ; kssl , mller & hillebrandt 1990b ) .
such a topology is unstable to resistive tearing mode instabilities and magnetic reconnection .
the presence of reconnection in jets could have important consequences for the interpretation of shock emission diagnostics ( hartigan , morse & raymond 1993 ) .
reconnection would provide an alternate means for converting kinetic energy into a thermal energy ( the field acts as a catalyst , jones 1997 ) which is then channeled into collisionally excited emission .
we wish to thank ralph pudritz , jack thomas and guy delemarter for the very useful and enlightening discussions on this topic .
support for this work was provided at the university of rochester by nsf grant ast-9702484 and the laboratory for laser energetics , and at the university of minnesota by nsf grants ast93 - 18959 , int95 - 11654 and ast96 - 16964 , by nasa grant nag5 - 5055 and the university of minnesota supercomputing institute .
the work by dr was supported in part by kosef through the 1997 korea - us cooperative science program .
burrows , c. , stapelfeldt , k. , watson , a. , krist , j.e . ,
ballester , k. , clarke , j. , crisp , d. , gallagher , j. , griffiths , r. , hester , j. , hoessel , j. , holtzman , j. , mould , j. , scowen , p. , trauger , j. , & westphal , j. , 1996 , apj , 473 , 437 fig . 1 : : comparison of 4 yso jet simulations . grey - scale log density . a : weak field non - radiative ( t = 1,971 y ) .
b : strong field non - radiative ( t = 1,460 y ) .
c : weak field radiative ( t = 1,971 y ) .
d : strong field radiative ( t = 1,460 y ) .
note that compression ratios behind bow / jet shocks in the radiative models are higher by an order of magnitude compared with the non - radiative cases .
each frame has been auto - scaled to its min / max resulting in different greyscale ranges .
fig . 2 : : propagation of the radiative strong field model at 4 times ( from top to bottom t = 698 , 1398 , 2095 , 2793 y ) .
each frame has been auto - scaled to its min / max resulting in different greyscale ranges .
the min / max values in the frames are approximately @xmath79 and @xmath80 . | we present multi - dimensional simulations of magnetized radiative jets appropriate to young stellar objects .
magnetized jets subject to collisionally excited radiative losses have not , as yet , received extensive scrutiny .
the purpose of this letter is to articulate the propagation dynamics of radiative mhd jets in the context of the extensive jet literature .
most importantly , we look for morphological and kinematic diagnostics that may distinguish hydrodynamic protostellar jets from their magnetically dominated cousins . our simulations are axisymmetric ( @xmath0-d ) . a toroidal
( @xmath1 ) field geometry is used .
our models have high sonic mach numbers ( @xmath2 ) , but lower fast mode mach number ( @xmath3 ) .
this is approximately the case for jets formed via disk - wind or x - wind models - currently the consensus choice for launching and collimating yso jets .
time - dependent radiative losses are included via a coronal cooling curve .
our results demonstrate that the morphology and propagation characteristics of strongly magnetized radiative jets can differ significantly from jets with weak fields .
in particular the formation of _ nose - cones _ via post - shock hoop stresses leads to narrow bow shocks and enhanced bow shock speeds .
in addition , the hoop stresses produce strong shocks in the jet beam which constrasts with the relatively unperturbed beam in radiative hydrodynamic jets .
our simulations show that pinch modes produced by magnetic tension can strongly effect magnetized protostellar jets .
these differences may be useful in observational studies designed to distinguish between competing jet collimation scenarios .
= cmmi10 scaled1 = cmmi10 = cmmi7 = cmmib10 scaled1 = cmmib10 = cmmib10 scaled 833 = cmbsy10 scaled1 = cmbsy10 = cmbsy10 scaled 833 1 = 1 = 1 = 9 = 9 = 9 = 10 = 10 = 10 = |
dark matter is at present one of the most exciting fields of particle physics and cosmology .
recent investigations of the cmb ( maxima , boomerang , dasi ) together with large scale structure results fix the non - baryonic dark matter contribution to the mass of the universe to @xmath0 25@xmath1 , of which still up to 38@xmath1 could be hot ( neutrinos ) . with the early nucleosynthesis constraint of @xmath2= 0.04 , the overwhelming contribution would consist of dark energy .
terrestrial direct dark matter search experiments just now are _ starting _ to enter into the range of sensitivity required , it we consider the neutralinos to be the candidate for cold dark matter .
the situation is similar in the hot dark matter search . here
double beta decay is the most sensitive means to look for an absolute neutrino mass scale .
direct search for wimps can be done \(a ) by looking for the recoil nuclei in wimp- nucleus elastic scattering .
the signal could be ionisation , phonons or light produced by the recoiling nucleus .
the typical recoil energy is a few 100ev / gev wimp mass .
\(b ) by looking for the modulation of the wimp signal resulting from the seasonal variation of the earth s velocity against the wimp wind. the expectation for neutralino elastic scattering cross sections and masses have been extensively analysed in many variants of susy models .
[ fig : bedn - wp2000 ] represents the present situation .
the susy predictions are from the msugra model @xcite and the mssm with relaxed unification conditions @xcite .
[ fig : ell - new-01 - 11 ] shows the result of a study at post - lep benchmark points based again on the msugra @xcite .
present experiments only just touch the border of the area predicted by the mssm .
the experimental dama evidence for dark matter lies in an area , in which msugra models do not expect dark matter .
they would require beyond - gut physics in this frame @xcite
. summarizing the present experimental status , present experiments and also future projects ( see @xcite ) can be categorized in two classes : \1 .
sensitivity ( or sensitivity goal ) just for confirmation of dama .
\2 . sensitivity to enter deeply into the range of susy predictions .
only very few experiments may become candidates for category 2 in a foreseeable future ( see figs .
[ fig : bedn - wp2000],[fig : ell - new-01 - 11 ] , and as far as at present visible , of those only genius will have the chance to search for modulation , i.e. to check , like dama , positive evidence for a dark matter signal .
the cryogenic experiments cdms , cresst and edelweiss operate at present with 600 g of detectors and less , after a decade of development .
expansion to several tens of kg or better 100 kg as required for modulation search , is still a _ very _ large step .
cdms has collected only 10.6kgd of data over this time ( in 1999 , since then not working ) , edelweiss only 4.53kgd .
the superheated droplet detectors picasso / simple are working at present on a scale of 15 and 50 g detectors , respectively .
zeplin and the very ambiguous and nice project drift are in the stage of prototype construction .
the situation is different for genius , which is based on conventional techniques .
if classifying the sugra models into more g@xmath3 - 2-friendly ( i , l , b , g , c , j ) and less g@xmath3 - 2-friendly models , according to @xcite , the former ones have good prospects to be detectable by lhc and/or a 1tev collider .
genius could check not only the larger part of these ones , but in addition two of the less g@xmath3 - 2-friendly models ( e and f ) , which will be difficult to be probed by future colliders ( see fig .
[ fig : ell - new-01 - 11 ] ) .
this demonstrates nicely the complementarity of collider and underground research .
genius would already in a first step , with 100 kg of _ natural _ ge detectors , cover a significant part of the susy parameter space for prediction of neutralinos as cold dark matter ( figs .
[ fig : bedn - wp2000],[fig : ell - new-01 - 11 ] ) . for this purpose
the background in the energy range @xmath4kev has to be reduced to @xmath5 ( events / kgykev ) . at this level solar neutrinos as source of background
are still negligible .
of particular importance is to shield the detectors during production ( and transport ) to keep the background from spallation by cosmic rays sufficiently low ( for details see @xcite .
the sensitivity of genius for dark matter corresponds to that obtainable with a 1km@xmath6 amanda detector for _ indirect _ detection ( neutrinos from annihilation of neutralinos captured at the sun ) ( see @xcite ) .
interestingly both experiments would probe different neutralino compositions : genius mainly gaugino - dominated neutralinos , amanda mainly neutralinos with comparable gaugino and higgsino components ( see fig.38 in @xcite ) .
neutrinos could still play an important role as hot dark matter in the universe .
recent results from cmb and lss measurements still allow for a sum of neutrino masses of @xmath7 4.4.ev . under the assumption that neutrinos are degenerate in mass , the common mass eigenvalue is , from decay , @xmath81.4ev(90@xmath1 c.l . ) if the lma msw solution is realized in nature @xcite .
new approaches and considerably enlarged experiments , will be required in future to fix the neutrino mass with higher accuracy .
genius is probably the most straightforward and promising of them .
as discussed @xcite it could probe with 1ton of enriched @xmath9 a scale of the effective neutrino mass down to 0.02ev .
this means that with 100 kg of enriched @xmath9 , a scale down to 0.04ev could be investigated .
as a first step of genius , a small test facility , genius - tf , is at present under installation in the gran sasso underground laboratory @xcite . with about 40 kg of natural ge detectors operated in liquid nitrogen , genius - tf could test the dama seasonal modulation signature for dark matter . up to summer 2001 ,
already six 2.5 kg germanium detectors with an extreme low - level threshold of @xmath0500 ev have been produced .
terrestrial dark matter experiments searching for cold dark matter only now are starting to enter into the range of susy predictions .
genius and genius - tf could play a decisive role here , as also in fixing the contribution of neutrinos to hot dark matter .
klapdor - kleingrothaus et al .
mpi - report * mpi - h - v26 - 1999 * and hep - ph/9910205 and in proc . of beyond99 ,
castle ringberg , germany , 6 - 12.06.1999 , ed .
klapdor - kleingrothaus and i.v .
krivosheina , iop , bristol ( 2000 ) 915 .
klapdor - kleingrothaus , in proc . of neutrino98 , ( eds )
y. suzuki et al .
* 77 * ( 1999 ) 357 .
klapdor - kleingrothaus , in proc . of wein98 , p. herczeg , c.m . hoffman and h.v .
klapdor - kleingrothaus ( eds . ) , world scientific , singapore ( 1999 ) 275 . | the new project genius will cover a wide range of the parameter space of predictions of susy for neutralinos as cold dark matter . together with dama it will be the only experiment which can probe the seasonal modulation signal
. concerning hot dark matter genius will be able to fix the ( effective ) neutrino mass with high accuracy .
a genius test facility has just been funded and will come into operation by end of 2002 . |
entanglement and nonlocality are some of the most emblematic concepts embodied in quantum mechanics@xcite .
recent developments in quantum computation@xcite and quantum information ( e.g. @xcite ) evoke interest in the properties of quantum entanglement .
now we have known that a state of a composite quantum system is called entangled ( or inseparable ) if it can not be represented as a convex combination of tensor products of its subsystem states@xcite . in the case of pure states ,
one has obtained a complete understanding , that is , a pure state is entangled if and only if it violates bell s inequality9,10,11 .
but the case for mixed states is much more complicated ( e.g.12,13 ) . besides the above definition of entangled state ,
another qualitative criterion proposed by peres@xcite is more practicable to determine whether a given state is inseparable , although it is sufficient and necessary only for sone simple quantum system , such as @xmath0 and @xmath1 systems . on the other hand ,
one is more interested in the quantitative measure of entangled states
. various entanglement measure has been proposed .
for example , for two - party system , there are following kinds of measure : the entanglement of formation@xcite , the relative entropy of entanglement@xcite , and the robustness of entanglement@xcite , _ etc .
_ consider the relative entropy associated with entangled state @xmath2 of two - qubit system . in order to obtain the relative entropy of entanglement of @xmath2 ,
the crucial point is to find a separable state @xmath3 such that minimizes the relative entropy @xmath4 defined as @xmath5 in general case , to find @xmath3 is a laborious work .
then recall the robustness of entanglement .
it is said that given an entangled state @xmath2 and a separable state @xmath3 there always exists a minimal real @xmath6 for which @xmath7is separable .
such minimal @xmath8 , denoted by @xmath9 , is called robustness of @xmath2 relative to @xmath3 .
then the entanglement embodied in @xmath2 can be measured by the absolute robustness of @xmath2 , which is obtained by means of finding a separable state @xmath10 such that @xmath11 is minimal .
we can see that both in relative entropy and in robustness , finding a specific separable state is of essence .
additionally , let s consider another useful tool to study entanglement , _
i.e. _ , lewenstein - sanpera ( l - s ) decomposition@xcite .
lewenstein _ et.al .
_ have shown that any state of two - qubit system can be decomposed as @xmath12 , \]]where @xmath13is a separable state , and @xmath14 denotes a single pure entangled projector @xmath15 , that is , @xmath14 represents a pure entangled state .
it is known that there are many different l - s decomposition with varying values of @xmath16 and various @xmath14 s . among them
is the unique optimal decomposition , the one with the largest @xmath16 value in eq .
( 3 ) , @xmath17where @xmath18is the degree of separability possessed by @xmath19 .
in other words , the entanglement possessed by @xmath19 can be represented as @xmath20 , where @xmath21 is the entanglement of its pure state expressed in terms of the von neumann entropy of the reduced density matrix of either of its subsystems:@xmath22here again , in the sense of maximizing or minimizing some quality , a specific separable state @xmath23 appears . from the above consideration
, we show that there are some specific separable states playing an important role in determining the entanglement measure .
then it is meaningful to determine the properties of these separable states .
a fact provokes our attention . that is , at least for two - qubit system the set of separable states , denoted by @xmath24 , is compact and convex19
. then boundary of @xmath24 does exist .
we argue in this paper that these special separable states mentioned above are on the boundary of @xmath25 . in the following , we give the criterion to determine whether a separable state is on the boundary . taking this into account ,
we present more clear interpretations on some concepts concerning quantum entanglement .
in this paper , we focus on the two - qubit quantum system .
the states can be represented in the hilbert space @xmath26 .
so we also call it 2@xmath272 quantum system .
we know that all @xmath28 hermitian matrices span a 16-dimensional real hilbert space , denoted by @xmath29 .
all density matrices of states forms a subset of @xmath30 .
we choose hilber - schimdt metric , that is , for @xmath31 , the norm is defined as @xmath32 ^{\frac{% 1}{2}}=\left [ tr\left ( a^{2}\right ) \right ] ^{\frac{1}{2}}$ ] , and for @xmath33 , the distance of @xmath34 and @xmath35 is@xmath36 ^{\frac{1}{2}}.\ ] ] for two - qubit system , a given state @xmath19 is separable , i.e. @xmath37 , if and only if the partial transposition of @xmath19 , denoted by @xmath38 or @xmath39 @xmath40 , is also positive@xcite , or equivalently , the partial time - reversal of @xmath41 , denoted by @xmath42 , is positive .
consider an arbitrary separable state @xmath43 .
mixing it with the maximal separable state @xmath44 , we have @xmath45 .\]]obviously , @xmath46 for each @xmath47 . in the geometrical view
, @xmath48 is on the line segment connecting @xmath3 and @xmath49 , then consider another form of mixture of @xmath3 and @xmath49 , @xmath50 at this time , @xmath51 is on the extension line from @xmath49 to @xmath3 . from the theorems in @xcite , we know that @xmath52 can be in the interior of @xmath24 , on the boundary of @xmath24 , or outside of @xmath24 . that depends on the values of @xmath53 .
so for sufficiently large @xmath54 , @xmath51 may be outside of @xmath24 .
let us now obtain our main result , which refer to the boundary of @xmath24 .
* theorem * for a given separable state @xmath55 , if and only if either of the following conditions is satisfied , @xmath56 is on the boundary of @xmath24 , @xmath57 there exists at least one vanishing eigenvalue for @xmath58 ; @xmath59 there exists at least one vanishing eigenvalues for @xmath60 . or in the other words , the rank of @xmath56 or @xmath61 is no larger than 3 . _
proof _ note that @xmath62 and @xmath24 is compact and convex and that @xmath49 is in the interior of @xmath24 .
if @xmath63 is on the boundary of @xmath24 , let us consider the mixture with form of eq.(8 ) , that is , @xmath64 from @xcite , we know that such @xmath65 does not belong to @xmath66 .
therefore , for a given @xmath67 , there are two possibilities for @xmath65 .
that is , either @xmath65 is positive and thus represents an entangled state , or @xmath65 is non - positive and does not represent any physical state . in the former case ,
any point on the line segment from @xmath49 to @xmath65 represents a real physical state , and particularly , all the points ( except @xmath68 ) on the line segment from @xmath63 to @xmath65 are entangled states .
let @xmath69 , then @xmath70 .
correspondingly @xmath71 . under this condition
, the distance between @xmath63 and @xmath65 can be arbitrarily small .
note that both @xmath63 and @xmath72 are clearly positive and that @xmath65 is also positive . but
@xmath73 is non - positive .
so by connectivity of the space and continuity of @xmath74 we conclude that @xmath72 must have vanishing eigenvalue . for the other case in which
@xmath65 is non - positive and does not represent real state , similar analysis shows that @xmath63 has at least one vanishing eigenvalue . to prove the reverse .
we first consider a separable state @xmath3 the eigenvalues of which can be denoted by @xmath75 , that is , @xmath3 has at least one vanishing eigenvalue .
we assume that @xmath3 is in the interior of @xmath66 . due to the compactness and convexity of @xmath24 and with the help of @xcite , we know that there always exists a @xmath54 such that the mixture @xmath51 in the form @xmath76 is also in the interior of @xmath24
. the eigenvalues of @xmath77 , denoted by @xmath78 , can be easily calculated , that is , @xmath79 obviously , for @xmath80 , we have @xmath81 . then @xmath82 is non - positive and can not belong to @xmath24 .
this result contradicts with the assumption made above .
hence @xmath3 can not be in the interior and can only be on the boundary of @xmath24 , i.e. @xmath83 . on the other hand , let s consider a separable state @xmath3 such that @xmath84 has one vanishing eigenvalue , _
i.e. _ @xmath85 . also assume that @xmath3 is in the interior of @xmath24 .
then @xmath51 is the interior point of @xmath24 for some @xmath86 .
so @xmath87 is necessarily positive .
denote the eigenvalues of @xmath88 by @xmath89 .
similar analysis will demonstrate that one of the eigenvalues of @xmath88 , which is actually @xmath90 , is negative .
it is contrary to the assumption .
so @xmath83 .
having obtained the description of the boundary of @xmath24 , we now return to think about some concepts of quantum entanglement . in @xcite ,
vedrel _ et.al .
_ propose the concept of relative entropy of entanglement and give the numberical results for some examples of two - qubit entangled states .
for instance , given entangled state @xmath2 as@xmath91where @xmath92 . the separable state which minimize the relative entropy @xmath93 ( cf .
eq.(1 ) ) is said to be@xmath94straightforward computation shows that @xmath3 has a vanishing eigenvalue .
then @xmath3 is on the boundary of @xmath24 .
this result also holds in the other examples mentioned in @xcite .
so we can say , at least for these specific cases , the relative entropy of entanglement can be achieved on the boundary of the set of separable states .
it is possible to generalize this result .
recall that the robustness of entanglement proposed by vidal _
et al._@xcite .
the geometrical interpretation has been obtained in @xcite , in which the boundary of @xmath24 is not yet explicitly defined .
the discussion in this letter can be supplementary to @xcite .
we can say that for a given entangled state @xmath2 of two - qubit system , if @xmath8 in eq.(2 ) is minimal among all separable states @xmath3 , _
i.e. _ @xmath8 is the absolute robustness of @xmath2 , then @xmath3 and @xmath95 in eq.(2 ) are necessarily on the boundary of @xmath24 .
now let us give our interpretation for l - s decomposition .
recall the form of l - s decomposition represented by eq.(3 ) . using hilbert - schmidt metric
defined by eq.(6 ) , we can express the @xmath16 in eq .
( 3 ) as @xmath96as shown in fig.1 , @xmath16 can be regarded as the ratio of the length of line segment @xmath97 to that of @xmath98 . with @xmath14 ( point @xmath34 in fig . 1 ) and @xmath41 ( point @xmath35 ) fixed , @xmath16 reaches the maximal when @xmath3 is on the boundary of @xmath24 , _
i.e. _ , the length of @xmath98 is minimal . in the geometrical view , this result is straightforward .
so we can always rewrite eq .
( 5 ) as @xmath99 , \]]where @xmath63 is on the boundary and @xmath100 is the maximal @xmath16 associated with @xmath14 . obviously , for eq.(4 ) @xmath23 is on the boundary of @xmath24 and can be written as @xmath101 while @xmath102 can be expressed as @xmath103 . in @xcite ,
englert _ et al .
_ have erred to what is involved in l - s decomposition , that is , if @xmath104 is the optimal decomposition , then ( a ) the state @xmath105 is non - separable for @xmath106 ; and ( b ) the state @xmath107 is either non - positive or non - separable for each @xmath108 .
it is said in @xcite that their verification is rather complicated even in seemingly simple cases .
but in our view , it is easy to understand .
now we give our explanation as follows . for conclusion ( a ) ,
since the optimal decomposition of @xmath19 has been obtained , @xmath101 is necessarily on the boundary of @xmath24 .
thus the convexity of @xmath24 guarantees that all points on the line segment _ ad _ ( cf fig .
1 ) can not belong to @xmath24 and without choice represent entangled states .
these states are expressed as the convex sum of @xmath109 and @xmath101 , i.e. , @xmath110 .
in addition , this conclusion also holds without the `` optimal '' condition .
that is , for any decomposition determined by eq .
( 14 ) , conclusion ( a ) is valid . for conclusion ( b ) , let @xmath111or equivalently , @xmath112the right - handed of eq . ( 16 ) is just the optimal decomposition of @xmath19 .
we already know that @xmath102 is the largest @xmath100 of all possible decomposition ( cf eq.(14 ) ) .
if the mixture on the left - hand of eq.(16 ) does represent @xmath19 , the weight of @xmath113 must be less than @xmath102 .
in other words , the left - hand of eq.(16 ) can not be a real l - s decomposition about @xmath41 . hence @xmath114 .
theore , either @xmath113 can not be positive , or can represent a non - separable state . the conclusion ( b )
can be demonstrated in fig .
2 . of course
the `` optimal '' condition is necessary in this case .
so far we have discussed the boundary of the set of all separable states . using this concept , we give straightforward explanations for some entanglement measure of the mixed states , mainly for l - s decomposition .
we feel that introducing boundary may be helpful for better understanding of entanglement .
1 @xmath115 ( point @xmath34 ) denotes an entangled pure state .
@xmath19 ( point @xmath35 ) represents a given non - separable mixed state .
@xmath24 is the set of all separable states .
@xmath116 ( point @xmath117 ) is in the interior of @xmath24 and @xmath56 ( point @xmath118 ) is on the boundary of @xmath24 .
2 @xmath119 and @xmath120 are two different entangled pure states . @xmath119 and @xmath101 construct the optimal l - s decomposition of @xmath19 .
@xmath113 is outside of set @xmath24 . @xmath121 and @xmath113 does not construct the l - s decomposition of @xmath19 . | the set of all separable quantum states is compact and convex .
we focus on the two - qubit quanum system and study the boundary of the set .
then we give the criterion to determine whether a separable state is on the boundary .
some straightforward geometrical interpretations for entanglement are based on the concept and presented subsequently . *
pacs : 03.65.bz . * [ theorem]acknowledgement [ theorem]algorithm [ theorem]axiom [ theorem]claim [ theorem]conclusion [ theorem]condition [ theorem]conjecture [ theorem]corollary [ theorem]criterion [ theorem]definition [ theorem]example [ theorem]exercise [ theorem]lemma [ theorem]notation [ theorem]problem [ theorem]proposition [ theorem]remark [ theorem]solution [ theorem]summary |
the unreduced colored jones polynomial is link invariant that assigns to each link @xmath2 a sequence of laurent polynomials @xmath3 indexed by a positive integer @xmath4 , the color . in @xcite dasbach and lin
conjectured that , up to a common sign change , the highest @xmath5 ( the lowest resp . ) coefficients of the polynomial @xmath6 of an alternating link @xmath2 agree with the first @xmath5 coefficient of the polynomial @xmath7 for all @xmath4 .
this gives rise to two power series with integer coefficients associated with the alternating link @xmath2 called the head and the tail of the colored jones polynomial .
the existence of the head and tail of the colored jones polynomial of adequate links , a class of links that contains alternating links , was proven by armond in @xcite using skein theory . independently ,
this was shown by garoufalidis and le for alternating links using @xmath8-matrices @xcite and generalized to higher order tails .
+ write @xmath9 to denote the kauffman bracket skein module of @xmath10 .
let @xmath2 be an alternating link and let @xmath11 be a reduced link diagram of @xmath2 . write @xmath12 to denote the @xmath13-smoothing state of @xmath11 , the state obtained by replacing each crossing by an @xmath13-smoothing .
the state @xmath14 of @xmath11 is defined similarly . using the kauffman skein relation kauffman
@xcite showed that the @xmath13 state ( respectively the @xmath15 state ) realizes the highest ( respectively the lowest ) coefficient of the jones polynomial of an alternating link .
we extend this result to the colored jones polynomial by using _ the colored kauffman skein relation _ [ section3 ] .
we show that the @xmath4-colored @xmath13 state and the @xmath4-colored @xmath15 state realize the highest and the lowest @xmath0 coefficients of the @xmath1 unreduced colored jones polynomial of an alternating link .
furthermore we show that this gives a natural layout to prove the stability of the highest and lowest coefficients of the colored jones polynomial of alternating links .
in other words we prove that the head and the tail of the colored jones polynomial exist .
in this section we recall the definition of adequate links and its connection to the kauffman states .
we also review the basic of the kauffman bracket skein module and the jones - wenzl idempotent .
let @xmath2 be a link in @xmath16 and @xmath11 be an alternating knot diagram of @xmath2 . for any crossing in @xmath11
there are two ways to smooth this crossing , the @xmath13-smoothing and the @xmath15-smoothing .
see figure [ smoothings1 ] .
we replace a crossing with a smoothing together with a dashed line joining the two arcs . after applying a smoothing to each crossing in @xmath11
we obtain a planar diagram consisting of collection of disjoint circles in the plane .
we call this diagram a _ state _ for the diagram @xmath11
. the @xmath13-smoothing state , obtained from @xmath11 by replacing every crossing by an @xmath13 smoothing , and the @xmath15-smoothing state for @xmath11 are of particular importance for us .
write @xmath12 and @xmath14 to denote the @xmath13 smoothing and @xmath15 smoothing states of @xmath11 respectively . for each state @xmath17 of a link diagram
@xmath11 one can associate a graph @xmath18 obtained by replacing each circle of @xmath17 by a vertex and each dashed link by an edges .
a link diagram @xmath11 is called @xmath13-adequate ( @xmath15-adequate , respectively ) if there are no loops in the graph @xmath19 ( the graph @xmath19 , respectively ) . a link diagram
@xmath11 is called adequate if it is both @xmath13-adequate and @xmath15-adequate .
it is known that a reduced alternating link diagram is adequate .
see for example @xcite .
it is important to see that the if a link diagram is @xmath13-adequate then any state that has only one @xmath15-smoothing will have one fewer circle than the @xmath13-smoothing state .
let @xmath20 denotes a @xmath21-manifold that is homeomorphic to @xmath22 $ ] , where @xmath23 is a connected oriented surface .
we will also use @xmath24 to denote the field generated by the indeterminate @xmath13 over the rational numbers .
( j. przytycki @xcite and v. turaev @xcite ) the _ kauffman bracket skein module _ of @xmath25 , denoted by @xmath26 , is the @xmath24-free module generated by the set of isotopy classes of framed links in @xmath20(including the empty knot ) modulo the submodule generated by the kauffman relations @xcite : @xmath27{0.06\linewidth } \vspace{0pt } \scalebox{0.04}{\includegraphics{simple - crossing } } \end{minipage } - a \begin{minipage}[h]{0.06\linewidth } \vspace{0pt } \scalebox{0.04}{\includegraphics{a - smoothing } } \end{minipage } - a^{-1 } \begin{minipage}[h]{0.06\linewidth } \vspace{0pt } \scalebox{0.04}{\includegraphics{b - smoothing } } \end{minipage } , \hspace{20 mm } ( 2)\hspace{3 mm } l\sqcup \begin{minipage}[h]{0.05\linewidth } \vspace{0pt } \scalebox{0.02}{\includegraphics{simple - circle } } \end{minipage } + ( a^{2}+a^{-2})l .
\end{aligned}\ ] ] where @xmath28 consists of a framed link @xmath2 in @xmath20 and the trivial framed knot .
relation ( 1 ) is usually called _ the kauffman skein relation_. since the manifold @xmath20 is homeomorphic to @xmath22 $ ] one may consider an appropriate version of link diagrams in @xmath23 to represent the classes of @xmath26 instead of framed links in @xmath20 . if the manifold @xmath20 has a boundary then one could define a relative version of the kauffman bracket skein module .
our first example is the kauffman bracket skein module of the sphere @xmath29 .
this module can be easily seen to be isomorphic to @xmath24 .
this isomorphism is provided by the kauffman bracket and it is induced by sending @xmath11 to @xmath30 . in particular
it send the empty link to @xmath31 . along with the module @xmath29
, we will also need the relative skein module @xmath32 , where the rectangular disk @xmath33 has @xmath4 designated points on the top edge and @xmath4 designated points on the bottom edge .
in fact , the module @xmath32 admits a multiplication given by vertical concatenation of two diagrams in @xmath32 . with this multiplication @xmath32 becomes an associative algebra over @xmath34 known as the _
@xmath1 temperley - lieb algebra _ @xmath35 .
+ for each positive integer @xmath4 , the algebra @xmath35 contains an element of special importance to us .
the@xmath1 _ jones - wenzl idempotent
_ ( projector ) , denoted @xmath36 , was first discovered by jones @xcite .
we shall adapt a diagrammatic presentation for jones - wenzel projector that is due lickorish @xcite . in this notation
one thinks of @xmath36 as an empty box with @xmath4 strands entering and @xmath4 strands leaving the opposite side .
see figure [ jw ] .
the jones - wenzl idempotent has two defining properties .
the defining properties of @xmath36 are : @xmath37{0.21\linewidth } \vspace{0pt } \scalebox{0.115}{\includegraphics{idempotent_mult } } \put(0,+80){\footnotesize{$n$ } } \end{minipage } = \hspace{5pt } \begin{minipage}[h]{0.1\linewidth } \vspace{0pt } \hspace{50pt } \scalebox{0.115}{\includegraphics{idempotent2 } } \put(-60,80){\footnotesize{$n$ } } \end{minipage } , ( 2 ) \hspace{22 mm } \begin{minipage}[h]{0.09\linewidth } \vspace{0pt } \scalebox{0.115}{\includegraphics{idempotent - with - hook } } \put(-70,+82){\footnotesize{$n - i-2 $ } } \put(-20,+64){\footnotesize{$1 $ } } \put(-2,+82){\footnotesize{$i$ } } \put(-28,20){\footnotesize{$n$ } } \end{minipage } = 0 .
\end{aligned}\ ] ] the second equation of [ properties8 ] holds for @xmath38 and it is usually called _ the annihilation axiom_. the @xmath1 jones - wenzl idempotent has an important recursive formula due to wenzl @xcite and this formula is stated graphically as follows : @xmath39{0.05\linewidth } \vspace{0pt } \scalebox{0.12}{\includegraphics{nth - jones - wenzl - projector } } \put(-20,+70){\footnotesize{$n$ } } \end{minipage } = \begin{minipage}[h]{0.08\linewidth } \hspace{8pt } \scalebox{0.12}{\includegraphics{nth - jones - wenzl - projector - first } } \put(-42,+70){\footnotesize{$n-1 $ } } \put(-8,+70){\footnotesize{$1 $ } } \end{minipage } \hspace{9pt } - \big ( \frac{\delta_{n-2}}{\delta_{n-1}}\big ) \hspace{9pt } \begin{minipage}[h]{0.10\linewidth } \vspace{0pt } \scalebox{0.12}{\includegraphics{nth - jones - wenzl - projector - second } } \put(2,+85){\footnotesize{$1 $ } } \put(-52,+87){\footnotesize{$n-1 $ } } \put(-25,+47){\footnotesize{$n-2 $ } } \put(2,+10){\footnotesize{$1 $ } } \put(-52,+5){\footnotesize{$n-1 $ } } \end{minipage } , ( 2 ) \hspace{8 mm } \hspace{2 mm } \begin{minipage}[h]{0.05\linewidth } \vspace{0pt } \scalebox{0.12}{\includegraphics{1th - jones - wenzl - projector } } \put(-20,+70){\footnotesize{$1 $ } } \end{minipage } = \begin{minipage}[h]{0.05\linewidth } \vspace{0pt } \scalebox{0.12}{\includegraphics{lonelystrand } } \end{minipage } \end{aligned}\ ] ] where @xmath40 we assume that @xmath41 is the empty tangle .
a proof of wenzl s formula can be found in @xcite and @xcite .
the defining properties of the jones - wenzl idempotent imply the following identities @xmath42{0.1\linewidth } \vspace{0pt } \scalebox{0.12}{\includegraphics{deltan } } \put(-29,+34){\footnotesize{$n$ } } \end{minipage } , ( 2 ) \hspace{8 mm } \begin{minipage}[h]{0.08\linewidth } \vspace{0pt } \scalebox{0.115}{\includegraphics{idempotent } } \put(-34,+82){\footnotesize{$n$ } } \put(-19,+82){\footnotesize{$m$ } } \put(-46,20){\footnotesize{$m+n$ } } \end{minipage } = \begin{minipage}[h]{0.09\linewidth } \vspace{0pt } \scalebox{0.115}{\includegraphics{idempotent2 } } \put(-46,20){\footnotesize{$m+n$ } } \end{minipage } \end{aligned}\ ] ] and @xmath43{0.09\linewidth } \vspace{0pt } \scalebox{0.115}{\includegraphics{simple - loop } } \put(-13,+82){\footnotesize{$n$ } } \put(-29,+70){\footnotesize{$m$ } } \end{minipage } = \frac{\delta_{m+n}}{\delta_{n}}\hspace{1 mm } \begin{minipage}[h]{0.06\linewidth } \vspace{0pt } \scalebox{0.115}{\includegraphics{nth - jones - wenzel-2 } } \put(-19,+82){\footnotesize{$n$ } } \end{minipage } , ( 2 ) \hspace{8 mm } \begin{minipage}[h]{0.08\linewidth } \vspace{0 pt } \scalebox{.15}{\includegraphics{little_1.eps } } \scriptsize { \put(-25,+72){$i$ } \put(-5,+72){$j$ } \put(-10,+10){$i+j$ } } \end{minipage } = a^{-ij } \begin{minipage}[h]{0.1\linewidth } \vspace{0pt } \scalebox{.15}{\includegraphics{little_2 } } \scriptsize { \put(-10,+10){$i+j$ } } \end{minipage } \end{aligned}\ ] ] let @xmath2 be a zero - framed link in @xmath16 . recall that the unreduced colored jones polynomial of @xmath2 , denoted by @xmath6 , can be obtained by decorate every component of @xmath2 , according to its framing , by the @xmath1 jones - wenzl idempotent and
consider this decorated framed link as an element of @xmath44 .
we are only concerted with the coefficients of the colored jones polynomial
so for our purpose the framing of the link @xmath2 could be chosen arbitrary .
+ following @xcite , write @xmath45 , where @xmath46 , @xmath47 are two laurent series if and only if the first @xmath4 coefficients of @xmath46 and @xmath47 coincide up to a common sign change .
the tail of the unreduced colored jones polynomial of a link @xmath2 , if it exist , is defined to be a series @xmath48 , that satisfies @xmath49 for all @xmath4 .
we start this section by proving the colored kauffman skein relation in [ lemma1 ] .
this relation is implicit in the work of yamada in @xcite .
the colored kauffman skein relation will be used in the next section to understand the highest and the lowest coefficients of the colored jones polynomial .
+ the following two lemmas are basically due to yamada @xcite .
we include the proof here with modification for completeness .
[ lemma1](the colored kauffman skein relation ) let @xmath50 .
then we have @xmath51{0.15\linewidth } \vspace{0 pt } \scalebox{.5}{\includegraphics{colored_crossing.eps } } \scriptsize { \put(-75,+82){$n+1 $ } \put(-8,+82){$n+1 $ } } \end{minipage } = a^{2n+1 } \begin{minipage}[h]{0.20\linewidth } \vspace{10pt } \scalebox{.5}{\includegraphics{coloredsmoothingb } } \scriptsize { \put(-85,+82){$n+1 $ } \put(-20,+82){$n+1 $ } \put(-60,48){$n$ } \put(-30,48){$n$ } \put(-45,+56){$1 $ } \put(-45,+18){$1 $ } } \end{minipage } + a^{-(2n+1 ) } \begin{minipage}[h]{0.20\linewidth } \vspace{10pt } \scalebox{.5}{\includegraphics{coloredsmoothinga } } \scriptsize { \put(-85,+82){$n+1 $ } \put(-20,+82){$n+1 $ } \put(-60,60){$n$ } \put(-32,60){$n$ } \put(-25,+42){$1 $ } \put(-65,+42){$1 $ } } \end{minipage}\end{aligned}\ ] ] applying the kauffman relation we obtain @xmath51{0.15\linewidth } \vspace{0 pt } \scalebox{.5}{\includegraphics{colored_crossing.eps } } \scriptsize { \put(-75,+78){$n+1 $ } \put(-8,+78){$n+1 $ } } \end{minipage } & = & a \begin{minipage}[h]{0.18\linewidth } \vspace{5pt } \scalebox{.5}{\includegraphics{smoothing_b1 } } \scriptsize { \put(-85,+78){$n+1 $ } \put(-20,+78){$n+1 $ } \put(-24,+64){$1 $ } \put(-50,+64){$1 $ } \put(-24,+10){$n$ } \put(-50,+10){$n$ } \put(-1,+10){$1 $ } \put(-70,+10){$1 $ } \put(-9,+57){$n$ } \put(-65,+57){$n$ } } \end{minipage } + a^{-1 } \begin{minipage}[h]{0.18\linewidth } \vspace{0pt } \scalebox{.5}{\includegraphics{smoothing_a } } \scriptsize { \put(-85,+78){$n+1 $ } \put(-20,+78){$n+1 $ } \put(-24,+64){$1 $ } \put(-50,+64){$1 $ } \put(-24,+10){$n$ } \put(-50,+10){$n$ } \put(-1,+10){$1 $ } \put(-70,+10){$1 $ } \put(-9,+57){$n$ } \put(-65,+57){$n$ } } \end{minipage}\\&= & a \begin{minipage}[h]{0.18\linewidth } \vspace{5pt } \scalebox{.5}{\includegraphics{smoothing_b } } \scriptsize { \put(-85,+75){$n+1 $ } \put(-20,+75){$n+1 $ } \put(-24,+64){$1 $ } \put(-50,+64){$1 $ } \put(-24,+10){$n$ } \put(-50,+10){$n$ } \put(-1,+10){$1 $ } \put(-70,+10){$1 $ } \put(-9,+57){$n$ } \put(-65,+57){$n$ } } \end{minipage } + a^{-1 } \begin{minipage}[h]{0.18\linewidth } \vspace{5pt } \scalebox{.5}{\includegraphics{smoothing_a } } \scriptsize { \put(-85,+75){$n+1 $ } \put(-20,+75){$n+1 $ } \put(-24,+64){$1 $ } \put(-50,+64){$1 $ } \put(-24,+10){$n$ } \put(-50,+10){$n$ } \put(-1,+10){$1 $ } \put(-70,+10){$1 $ } \put(-9,+57){$n$ } \put(-65,+57){$n$ } } \end{minipage } \end{aligned}\ ] ] using property ( 2 ) in [ properties2 ] we obtain the result .
note that if we special @xmath4 to @xmath31 in the previous lemma we obtain the kauffman skein relation .
[ lemma2 ] let @xmath50 .
then we have @xmath51{0.15\linewidth } \vspace{0 pt } \scalebox{.4}{\includegraphics{colored_crossing.eps } } \scriptsize { \put(-55,+52){$n$ } \put(-1,+52){$n$ } } \end{minipage } = \displaystyle\sum\limits_{k=0}^{n}c_{n , k } \begin{minipage}[h]{0.15\linewidth } \vspace{0pt } \scalebox{.4}{\includegraphics{goodbasis } } \scriptsize { \put(-60,+64){$n$ } \put(-5,+64){$n$ } \put(-58,30){$k$ } \put(-7,30){$k$ } \put(-43,+56){$n - k$ } \put(-43,+16){$n - k$ } } \end{minipage } \end{aligned}\ ] ] where @xmath52 lemma [ lemma1 ] implies that @xmath53{0.15\linewidth } \vspace{0 pt } \scalebox{.4}{\includegraphics{colored_crossing.eps } } \scriptsize { \put(-55,+52){$n$ } \put(-1,+52){$n$ } } \end{minipage } = \displaystyle\sum\limits_{k=0}^{n}c^{\prime}_{n , k } \begin{minipage}[h]{0.15\linewidth } \vspace{0pt } \scalebox{.4}{\includegraphics{goodbasis } } \scriptsize { \put(-60,+64){$n$ } \put(-5,+64){$n$ } \put(-58,30){$k$ } \put(-7,30){$k$ } \put(-43,+56){$n - k$ } \put(-43,+16){$n - k$ } } \end{minipage } \end{aligned}\ ] ] where @xmath54 is a polynomial with integral coefficients in a. let us prove by induction on @xmath4 that we have @xmath55 for @xmath56 relation ( [ induction ] ) holds since this is just the kauffman skein relation . applying the identity ( [ goo ] ) on the colored each term of the colored kauffman skein relation ,
we obtain : @xmath57{0.15\linewidth } \vspace{0pt } \scalebox{.4}{\includegraphics{goodbasis } } \scriptsize { \put(-60,+64){$n$ } \put(-5,+64){$n$ } \put(-48,30){$k$ } \put(-7,30){$k$ } \put(-43,+56){$n - k$ } \put(-43,+16){$n - k$ } } \end{minipage } & = & a^{2n-1}\displaystyle\sum\limits_{k=0}^{n-1}c^{\prime}_{n-1,k } \begin{minipage}[h]{0.17\linewidth } \vspace{0pt } \scalebox{.4}{\includegraphics{goodbasis } } \scriptsize { \put(-60,+64){$n$ } \put(-5,+64){$n$ } \put(-48,30){$k$ } \put(-7,30){$k$ } \put(-43,+56){$n - k$ } \put(-43,+16){$n - k$ } } \end{minipage}+ a^{-2n+1}\displaystyle\sum\limits_{k=0}^{n-1}c^{\prime}_{n-1,k } \begin{minipage}[h]{0.17\linewidth } \vspace{0pt } \scalebox{.4}{\includegraphics{goodbasis } } \scriptsize { \put(-60,+64){$n$ } \put(-5,+64){$n$ } \put(-48,30){$k+1 $ } \put(-7,30){$k+1 $ } } \end{minipage}\\&=&a^{2n-1}\displaystyle\sum\limits_{k=0}^{n-1}c^{\prime}_{n-1,k } \begin{minipage}[h]{0.16\linewidth } \vspace{0pt } \scalebox{.4}{\includegraphics{goodbasis } } \scriptsize { \put(-60,+64){$n$ } \put(-5,+64){$n$ } \put(-48,30){$k$ } \put(-7,30){$k$ } \put(-43,+56){$n - k$ } \put(-43,+16){$n - k$ } } \end{minipage}+ a^{-2n+1}\displaystyle\sum\limits_{k=1}^{n-1}c^{\prime}_{n-1,k-1 } \begin{minipage}[h]{0.15\linewidth } \vspace{0pt } \scalebox{.4}{\includegraphics{goodbasis } } \scriptsize { \put(-60,+64){$n$ } \put(-5,+64){$n$ } \put(-48,30){$k$ } \put(-7,30){$k$ } } \end{minipage } \end{aligned}\ ] ] the skein elements @xmath58{0.13\linewidth } \vspace{0pt } \scalebox{.3}{\includegraphics{goodbasis } } \scriptsize { \put(-50,+48){$n$ } \put(-5,+48){$n$ } \put(-45,20){$k$ } \put(-7,20){$k$ } \put(-34,+43){$n - k$ } \put(-34,+14){$n - k$ } } \end{minipage}$ ] , where @xmath59 , are linearly independent and hence @xmath60 however @xmath61 hence @xmath62 relations ( [ rec1 ] ) and ( [ rec2 ] ) and the induction hypothesis yield the result . motivated by lemma [ lemma1 ] we define _ the @xmath4-colored states _ for a link diagram @xmath11 for every positive integer @xmath4 .
suppose that the link diagram has @xmath63 crossings .
label the crossings of the link diagram @xmath11 by @xmath31, .. ,@xmath63 .
an @xmath4-colored state @xmath64 for a link diagram @xmath11 is a function @xmath65 .
if the color @xmath4 is clear from the context , we will drop @xmath4 from the notation of a colored state .
given a link diagram @xmath11 and a colored state @xmath66 for the diagram @xmath11 .
we construct a skein element @xmath67 obtained from @xmath11 by replacing each crossing labeled @xmath68 by an @xmath4-colored @xmath13-smoothing and each crossing labeled @xmath69 by an @xmath4-colored @xmath15-smoothing , see figure [ smoothings ] .
two particular skein elements obtained in this way are important to us .
the skein element obtained by replacing each crossing by the @xmath4-colored @xmath13-smoothing will be called the
_ @xmath4-colored @xmath13-state _ and denoted by @xmath70 , where @xmath71 denotes the colored state for which @xmath72 for all @xmath73 in @xmath74 .
the @xmath4-colored @xmath15-state is defined similarly .
see figure [ colored state ] for an example .
following armond @xcite , we now define what we mean by an adequate skein element in @xmath75 .
consider a skein element @xmath17 in @xmath75 consists of arcs connecting jones - wenzl idempotents of various colors .
construct the diagram @xmath76 from @xmath17 by replacing each @xmath77 jones - wenzl idempotent in @xmath17 by the identity in @xmath78 .
if @xmath17 contains no crossings then we say that @xmath17 is adequate if @xmath76 consists of circles each one of them passes at most once from the regions where we had the boxes of the idempotents in @xmath17 .
in other words , a crossingless skein element @xmath17 is adequate if each circle in @xmath17 passes at most once from the same idempotent .
see figure [ cody_1 ] for a local picture of an adequate skein element .
if @xmath79 is in @xmath75 then @xmath80 will denote the minimum degree of @xmath79 expressed as a laurent series in @xmath13 .
furthermore , denote @xmath81 .
the following lemma is due to armond @xcite .
[ codys ] if @xmath82 is expressed as a single diagram containing the jones - wenzl idempotent , then @xmath83 . if the diagram for @xmath17 is an adequate skein diagram , then @xmath84 .
the colored kauffman skein relation provides a natural framework to understand the highest and the lowest coefficients of the colored jones polynomial . in this section
we will use this relation to prove that the the highest ( the lowest respectively ) @xmath0 coefficients of the @xmath1 unreduced colored jones polynomial agree up to a sign with the @xmath4-colored @xmath13-state ( the @xmath4-colored @xmath15-state respectively ) .
we use this result to prove existence of the the tail of the colored jones polynomial .
[ thm1 ] let @xmath2 be an alternating link diagram .
then @xmath85 assume that the link diagram @xmath2 has @xmath63 crossings and label the crossing of the link diagram by @xmath86 .
the colored kauffman skein relation implies that @xmath87 where @xmath88 and the summation runs over all functions @xmath89 .
now for any colored state @xmath66 of the link diagram @xmath2 there is a sequence of states @xmath90 such that @xmath91 , @xmath92 and @xmath93 for all @xmath94 except for one integer @xmath95 for which @xmath96 and @xmath97 . it is enough to show that the lowest @xmath0 terms of @xmath98 are never canceled by any term from @xmath99 for any @xmath66 . for any colored state @xmath66 of the link diagram @xmath2 one could write @xmath100 where @xmath101 is the skein element that we obtain by applying [ lemma2 ] to every crossing in @xmath102 .
the theorem follows from the following three lemmas .
[ m7 ] @xmath103 @xmath104 @xmath105 @xmath106 it is clear that @xmath107 and @xmath108 furthermore , @xmath109 hence @xmath110 finally , it is clear that @xmath111 hence , the result follows .
@xmath112 @xmath113
when we replace the idempotent by the identity in the skein element @xmath114 we obtain the diagram @xmath115 where @xmath116 is a link diagram composed of a disjoint union of unit circles each one of them bounds a disk and @xmath117 is the @xmath118-parallel of @xmath2 .
note that state @xmath119 is exactly the all @xmath15-state of the link diagram @xmath120 and it is also the all b - state of @xmath115 .
since @xmath2 is alternating then the the number of circles in @xmath121 is one less than the number of circles in @xmath119 . in other words ,
the number of circles in the all @xmath15-state of @xmath122 is one less than the number of circles in the all @xmath15-state of the diagram @xmath115 .
thus , @xmath123 moreover the skein element @xmath124 is adequate since @xmath2 is alternating . hence by lemma [ codys ] we have @xmath112 for the second part , note that the number of circles in the diagrams @xmath125 and @xmath126 differs by @xmath31 . hence by [ codys ] we obtain @xmath127 .
@xmath128 the previous two lemmas imply directly that the lowest term in @xmath114 is coming from the skein element @xmath129 and this term is never canceled by any other term in the summation .
[ thm2 ] let @xmath2 be an alternating link diagram and let @xmath130 be its corresponding @xmath131-colored @xmath15-state skein element .
then @xmath132 since @xmath2 is an alternating link diagram then the skein element @xmath130 must look locally as in figure [ local ] .
it follows from theorem @xmath133 and lemma @xmath134 in @xcite that @xmath135{0.21\linewidth } \vspace{0pt } \scalebox{0.24}{\includegraphics{proof_1 } } \put(-80,55){$1 $ } \put(-32,2){$n$ } \end{minipage } \doteq_{4n } \hspace{5pt } \begin{minipage}[h]{0.1\linewidth } \vspace{0pt } \hspace{50pt } \scalebox{0.24}{\includegraphics{proof_2 } } \put(-80,55){$1 $ } \put(-32,2){$n$ } \end{minipage } \end{aligned}\ ] ] the details of the previous equation can be found in @xcite and we will not repeat them here . the previous step can be applied around the circle until we reach the final idempotent : @xmath136{0.18\linewidth } \vspace{0pt } \scalebox{0.24}{\includegraphics{proof_1 } } \put(-80,55){$1 $ } \put(-32,2){$n$ } \end{minipage } \doteq_{4n } \hspace{5pt } \begin{minipage}[h]{0.1\linewidth } \vspace{0pt } \hspace{50pt } \scalebox{0.24}{\includegraphics{proof_3 } } \put(-70,55){$1 $ } \put(-32,2){$n$ } \end{minipage}\hspace{40pt}\doteq_{4n } \begin{minipage}[h]{0.1\linewidth } \vspace{0pt } \hspace{200pt } \scalebox{0.24}{\includegraphics{proof_4 } } \put(-60,38){$1 $ } \put(-32,2){$n$ } \end{minipage } \end{aligned}\ ] ] equation [ properties2 ] implies @xmath137{0.1\linewidth } \vspace{0pt } \hspace{50pt } \scalebox{0.24}{\includegraphics{proof_4 } } \put(-60,38){$1 $ } \put(-32,2){$n$ } \end{minipage}\hspace{40pt}\doteq_{4n}\frac{\delta_{n+1}}{\delta_{n } } \hspace{5pt } \begin{minipage}[h]{0.13\linewidth } \vspace{0pt } \hspace{200pt } \scalebox{0.24}{\includegraphics{proof_5 } } \put(-32,2){$n$ } \end{minipage}\hspace{40pt}\doteq_{4n } \begin{minipage}[h]{0.11\linewidth } \vspace{0pt } \hspace{205pt } \scalebox{0.24}{\includegraphics{proof_5 } } \put(-32,2){$n$ } \end{minipage } \end{aligned}\ ] ] applying this procedure on every circle in @xmath130 , we eventually obtain @xmath138 theorems [ thm1 ] and [ thm2 ] imply immediately the following result .
let @xmath2 be an alternating link diagram
. then @xmath139
i am grateful for oliver dasbach for his guidance and advice .
i am very thankful to kyle istvan for carefully reading the paper and offering many useful remarks .
i also would like to thank cody armond for various conversation about this paper .
przytycki , _ skein modules of 3-manifolds _ , bull . pol .
39(1 - 2 ) ( 1991 ) 91100 .
turaev , _ the conway and kauffman modules of the solid torus _ , zapiski nauchnykh seminarov pomi , 167 , 79 - 89 . | using the colored kauffman skein relation , we study the highest and lowest @xmath0 coefficients of the @xmath1 unreduced colored jones polynomial of alternating links .
this gives a natural extension of a result by kauffman in regard with the jones polynomial of alternating links and its highest and lowest coefficients .
we also use our techniques to give a new and natural proof for the existence of the tail of the colored jones polynomial for alternating links . |
since the first observations of quantum collective phenomena , the quantum systems with strongly interacting constituents have become of paramount interest in condense matter community @xcite .
a new wave of activity in the area of quantum many - body systems has been burgeoning with the advent of laser cooling techniques @xcite .
many of those quantum models , which were thought to be theoretical abstractions , have since been implemented with cold atoms @xcite .
one of such models , a one - dimensional system of hard - core bosons a tonks - girardeau ( tg ) gas @xcite has been thoroughly probed in recent experiments @xcite . while the density profile and the energy spectrum of hard - core bosons resemble that of non - interacting fermions , its momentum distribution ( md ) exhibits distinct features .
the ground state of a homogeneous tg gas is known to possess an infrared divergence in the thermodynamic limit , @xmath1 , @xmath2 @xcite , which , however , vanishes upon addition of a harmonic confinement @xcite . so far , there are no analytic results on the infrared behavior of the finite - temperature md for a finite number of bosons in a harmonic trap . in a sole harmonic confinement ,
the ground - state md decays as @xmath3 at the high momentum regime @xcite . yet in the presence of an optical lattice , which sets an upper momentum scale given by the recoil momentum , @xmath4 , where @xmath5 is the wave vector of the laser beam , this region can not be resolved in present state - of - the - art experiments .
the finite - temperature mds of @xmath6 strongly repulsive bosons , confined on a one - dimensional optical lattice and an additional harmonic trap ( a so called `` 1d tube '' ) , have been measured experimentally @xcite .
these measurements revealed that in the intermediate region , @xmath7 , a momentum profile can be approximated by a power - law , @xmath8 .
the exponent @xmath9 depends on temperature , density , and strength of the atom - atom interactions .
results of numerical monte - carlo simulations have corroborated the experimental finding @xcite .
what kind of momentum distribution emerges in a system of strongly repulsive bosons ? in the present work , we attest that _ levy distribution _ describe the md of @xmath0 thermalized hard - core bosons in various one - dimensional confinements , in particular , within a single 1d tube as well as within array of 1d tubes probed experimentally @xcite .
levy statistics @xcite are known to describe classical chaotic transport @xcite , processes of subrecoil laser cooling @xcite , fluctuations of stock market indices @xcite , time series of single molecule blinking events @xcite , or bursting activity of small neuronal networks @xcite .
the appearance of levy distribution in a system output is a strong indicator of a long - range correlation `` skeleton '' which conducts system intrinsic dynamics @xcite .
however the levy distribution has at no time emerged in the context of _ many - particle quantum _
systems before .
the great advantage of the levy - based analysis is its capability of calibration of the tg in different quantum regimes by a unique scaling exponent @xmath10 .
the paper is organized as follows . in sec .
ii , we introduce the lattice tg model and employ two complimentary approaches to study its finite - temperature properties . in sec .
iii , we fit exact results of the calculation within the grand - canonical formalism @xcite by the levy distribution . to provide an insight into experimental situation @xcite , we devoted sec .
iv to the analysis of momentum profiles averaged over the array of 1d tubes
. we proceed with the results obtained within the canonical formalism @xcite and conclude the section with the analysis of the experimental data from the ref .
@xcite . in sec .
v , we demonstrate the universality of the levy spline approximation by addressing the md of a tg gas in one - dimensional confinements of various geometry .
we elaborate on the case of a tg gas in a sole harmonic confinement , in a box , and in a sole optical lattice with impenetrable ( hard wall ) boundaries . finally , in sec .
vi , we summarize our results .
some of important technical details are deferred to the appendixes a and b.
a bose gas confined in a deep optical lattice is well described by the bose - hubbard hamiltonian @xcite @xmath11 where @xmath12 is the particle number operator on lattice site @xmath13 , @xmath14 is the hopping strength , the parabolicity @xmath15 is the amplitude of the external harmonic potential with the trapping frequency @xmath16 , @xmath17 is the mass of the atom , and @xmath18 is the lattice constant . the last term in ( [ eq : hamiltonian ] ) describes the on - site atom - atom interactions .
here we are interested in the tg regime , where the strength of the repulsive atom - atom interaction considerably exceeds the kinetic ( hopping ) energy , i.e. , @xmath19 , @xcite .
therefore , the interaction term can be substituted by the condition that two bosons can not occupy the same lattice site @xcite .
hard - core bosons in a 1d tube , thin ( black ) lines , at four different sets of parameters are compared to levy distributions , thick ( red ) lines .
dashed - doted line ( c ) corresponds to a gaussian approximation .
the parameters @xmath20 are indicated by symbols @xmath21 , @xmath22 , @xmath23 , @xmath24 in fig .
2 . the thin solid ( blue ) icon indicates the confinement geometry.,scaledwidth=45.0% ] the momentum distribution can be obtained from the reduced single - particle density matrix @xmath25 , reading @xmath26 where the momentum @xmath27 is written in units of recoil momentum @xmath28 , and the envelope @xmath29 is the fourier transform of the wannier function . note that the latter , as well as the hopping @xmath14 , is solely defined by the lattice depth , @xmath30 , measured in units of recoil energy , @xmath31 . to find the reduced single - particle density matrix ,
we employ here the grand - canonical @xcite and the canonical formalism @xcite ( see appendixes a , b ) .
the first is relevant for a system being in contact with a thermal cloud at constant temperature @xmath32 , while the second describes an isolated many - particle system .
the difference between the momentum distributions obtained within the grand - canonical and the canonical descriptions is mediated by the number of particles and becomes negligible for @xmath33 @xcite .
in addition , at finite temperatures @xmath34 , systems with different number of particles , but the same densities @xmath35 , possess the same momentum profiles @xcite . here , @xmath36 is the number of single - particle eigenstates with non - zero population at the trap center @xcite .
the latter also yields the critical number of bosons in a 1d tube required to form the mott - insulator in the trap center at zero temperature .
the typical examples of the md obtained here within the grand - canonical formalism are presented in fig . 1 for @xmath37 , where the results of both the grand - canonical and canonical descriptions practically identical .
in view of the strong non - gaussian behavior of the function @xmath38 ( see fig .
1c ) and an apparent power - law intermediate asymptotics @xcite , it is tempting to compare the md of a tg gas with symmetric levy distribution @xcite .
the latter , @xmath39 , is a natural generalization of the gaussian distribution , and it is defined by the fourier transform of its characteristic function ; i.e. , @xmath40 where the exponent @xmath41 , and @xmath42 is some constant @xcite .
for @xmath43 the distribution is the gaussian .
the value of @xmath10 is invariant under the scaling transformation @xmath44 , which can be used for fitting .
the levy distribution exhibits a power - law asymptotics , @xmath45 , as @xmath46 .
these `` heavy '' tails cause the variance of levy distributions to diverge for all @xmath47 .
however , this asymptotic limit is not relevant for our objective : the md for the system ( 1 ) is bounded by the width of @xmath48 , @xmath49 @xcite , so that the md variance remains finite .
( top ) , and the corresponding root mean square deviation from the exact momentum density profile , @xmath50 ( bottom ) , of tg gas as a function of temperature and particle density for two different amplitudes of the optical lattice , @xmath51 ( left column ) , and @xmath52 ( right column ) , see text for more details .
the amplitude of harmonic confinement is chosen as @xmath53.,scaledwidth=45.0% ] the fitting of the calculated md @xmath38 by the levy distribution @xmath54 was automatized and performed numerically by the global minimum search of the root mean square deviation , @xmath55 ^ 2\right\}^{1/2}/n(0)$ ] , in a 2d parametric space @xmath56 ( [ eq : levy ] ) . for the results shown on figs .
1,2 , we took @xmath57 spline points equally spaced on the interval @xmath58 .
the initial area , @xmath59 , covered by @xmath60 grid was iteratively converged to the global minimum by decreasing the grid step in @xmath10 and @xmath61 until the desired relative accuracy ( fixed to @xmath62 in all figures ) was reached . in fig .
2 , we show the dependence of scaling exponent @xmath10 ( top ) on the scaled temperature @xmath63 and the scaled particle density , @xmath35 ( obtained for @xmath64 , and @xmath65 ) , together with the associated mean square deviation @xmath50 ( bottom ) , at relatively low and high amplitude of the optical lattice . these diagrams ( see also fig
. 1 , for concrete examples of the md and their levy - spline approximations ) constitute the first main result of this work , namely , the convergence of the md of a tg gas towards the levy distribution with increasing the temperature . despite of infinitely strong on - site repulsive interactions , the systematic increase of the levy exponent @xmath10 with temperature , cf .
2 ( top ) , is consistent with the high - temperature limit where the ideal bose gas obeys the classical boltzmann - maxwell statistics with a gaussian md , i.e. , @xmath43 .
the single 1d tube realization , using a fixed temperature @xmath32 and a fixed particle number @xmath0 , is not directly accessible with the present state - of - art experiments .
the averaging over an array of tubes done in ref .
@xcite can be understood as an averaging over many realizations , with different parameters , @xmath32 and @xmath0 .
namely , the experimental setup @xcite produces an array of independent 1d tubes , with different numbers of particles , @xmath66 .
the probability of having a tube with @xmath0 particles is given by @xcite @xmath67 where the number of particles in a central tube @xmath68 is a unique parameter .
assuming the same initial temperature in all the tubes for a shallow 1d lattice potential , @xmath69 , the temperatures at the experimentally adjusted lattice depth , @xmath30 , can be obtained by using the conservation of entropy in each tube during the subsequent adiabatic increase of the lattice depth from @xmath69 to @xmath30 .
therefore , tubes with the different number of particles , @xmath66 , acquire different final temperatures , @xmath70 , at @xmath71 @xcite .
hard - core bosons , thin ( black ) lines , obtained within the canonical approach ( [ eq : cdmatrix ] ) and the md averaged over the array of 1d tubes according to the experimental procedure used in ref .
@xcite , thick ( red ) line , with its best levy - spline , thick ( blue ) dashed line.,scaledwidth=45.0% ] we implemented this averaging procedure with @xmath72 , for a set of individual mds pre - calculated within the canonical formalism .
the result for @xmath73 , @xmath74 and @xmath75 is depicted in fig .
3 . surprisingly enough , the averaged md can be perfectly approximated by the levy distribution with an `` average '' scaling exponent , @xmath76 ; this is so despite the sizable dispersion of @xmath10-values appearing in the different momentum profiles .
therefore , the averaged momentum profile appears as a superposition of several different profiles .
it is known that using a proper weight function , @xmath77 , one can construct a levy distribution from a parameterized set of gaussian distributions of different dispersion , @xmath78 @xcite . yet , to the best of our knowledge , there are no results concerning the superposition of many different levy distribution functions with _ different exponents _ @xmath10 . a next objective is the comparison of our scheme with the experimental data from ref .
@xcite , see fig . 4 . as one can deduce the levy distribution yields an excellent approximation for the md of the experimental system although the latter does not map precisely to a tg gas , but rather corresponds to a set of @xmath0 soft - core bosons with @xmath79 ( [ eq : hamiltonian ] ) @xcite . in the experiment ,
one used a power - law fit , @xmath80 , on an intermediate range ( @xmath7 ) @xcite , with @xmath81 for the data shown on figs .
4(a - d ) , respectively .
we emphasize , however , that in the experimentally accessible region , @xmath82 , the power law behavior with the exponent @xmath83 of the levy distribution is not yet valid , but is assumed only for much larger momentum values .
therefore , the theoretical estimates @xmath76 in fig . 4 exceed those intermediate range power - law fit - values , i.e. one consistently finds that @xmath84 . , solid ( red ) lines .
the thin solid ( blue ) icon indicates the confinement geometry.,scaledwidth=45.0% ]
so far we have been dealing with a tg gas on a lattice with an additional harmonic potential ( [ eq : hamiltonian ] ) , which is the only confinement where the md of a tg gas was experimentally studied in detail @xcite . in the present section ,
we demonstrate applicability of the levy - spline approximation to the numerically obtained md of hard - core bosons at finite temperatures in various 1d confinements : a sole harmonic trap , fig .
[ fig4 ] , a box , fig .
[ fig5](a , b ) , and a sole optical lattice with impenetrable boundaries , fig .
[ fig5](c , d ) .
a tg gas confined in a general potential , @xmath85 , is described by the sum of the single particle hamiltonians , @xmath86 with the hard - core constrain on the bosonic many - particle wave function : @xmath87 if @xmath88 , where @xmath89 is the position of @xmath90th particle , and @xmath91 is the 1d hard - core diameter . at low density ,
hard - bosons can be approximated by impenetrable point size particles , so that @xmath92 @xcite .
the md of @xmath0 hard - bosons in a box , @xmath93 if @xmath94 and @xmath95 otherwise , and in a harmonic confinement , @xmath96 , was obtained numerically within the canonical formalism with the help of efficient method @xcite , which is detailed in appendix [ perez ] . in the case of a sole lattice potential with @xmath97 lattice sites ,
the md was calculated within the grand - canonical formalism by using the same algorithm that was employed in sec .
iv ( see appendix a ) .
hard - core bosons in a sole 1d harmonic trap , thin ( black ) lines , at the two different temperatures , @xmath98 , @xmath99 are compared to levy distributions , thick ( red ) lines .
dashed line ( c ) underlines well known universal power - law asymptotic behavior of the tonks - girardeau gas momentum distribution at large momenta , @xmath100 , while dashed - dotted line illustrates failure of a gaussian approximation .
the thin solid ( blue ) icon indicates the confinement geometry.,scaledwidth=45.0% ] it is known , that the md of a tg gas in a homogeneous toroidal trap and in a harmonic confinement ( fig .
[ fig4 ] ) exhibits a power - law behavior , @xmath100 , for @xmath101 @xcite .
we also found the same power - law behavior for a tg gas in a box , see fig .
[ fig5](a , b ) .
this in fact turns out to be a general feature of a tg , which also persists at finite temperatures .
however , the power - law tail , @xmath102 , of the levy distribution with exponent @xmath10 can not decay faster than @xmath103 @xcite .
therefore , our levy - spline approximation is exclusively aimed to a finite momentum region @xmath104 , where the asymptotic @xmath100 behavior has not yet developed . with figs .
[ fig4]-[fig5 ] , we demonstrate that the md of a tg gas at finite temperatures in various 1d confinement potentials ( thin lines ) can be perfectly approximated by levy - splines ( thick lines ) over significant momentum range .
while at zero temperature we detect a notable deviation of the levy - spline from the actual md , with increasing the temperature the deviation becomes practically negligible .
the exponent of levy - splines for a tg gas in the confinement potentials presented here depends , in general , on the confinement characteristics , on the gas temperature and on the particle number or the filling factor , @xmath105 ( in the case of a lattice ) .
a more detailed analysis of this dependence will be addressed elsewhere .
hard - core bosons in a 1d box ( a , b ) of size @xmath106 and on a lattice ( c , d ) with @xmath107 lattice sites and the potential amplitudes @xmath108 , thin ( black ) lines , at the temperatures @xmath109 ( a ) ; @xmath110 ( b ) , @xmath111 ( c , d ) are compared to levy distributions , thick ( red ) lines . here
the frequency @xmath112 sets the energy scale , @xmath113 , for a particle in a box .
the momentum , @xmath27 , is measured in units of @xmath114(a , b ) and @xmath115 ( c , d ) , in the case of a lattice potential and a box , respectively .
the thin solid ( blue ) icons indicate the two corresponding types of confinement geometry.,scaledwidth=45.0% ]
we have presented a study of the finite - temperature momentum distribution ( md ) of tonks - girardeau ( tg ) gases confined in various 1d confinement potentials .
the md of a tg gas on an optical lattice with a superimposed harmonic confinement depends on the particle density , on the lattice depth and on the gas temperature .
we have shown that the tunable levy distribution fits momentum profiles up to one recoil momentum with high accuracy .
this allows for calibration of tg states with a unique scaling exponent .
thus , our approach completes the attempts to quantify the finite - temperature mds by using a power - law fitting for the intermediate region @xmath7 @xcite .
we demonstrate that the md of a tg gas confined in a 1d box , in a sole 1d harmonic potential and on a 1d optical lattice with hard wall boundaries can be approximated by the levy distribution function ( [ eq : levy ] ) on a finite momentum region @xmath104 , where for a tg gas the universal power - law , @xmath100 , at large momenta has not yet developed .
we thus conjecture that levy scaling of md is generic feature of a thermalized tg gas .
we want to emphasize one aspect which we consider to be crucial for the understanding of our approach .
it is known that the levy distribution has appeared as the stable distribution , i.e. the `` attractor '' for normalized sums of independent and identically - distributed random variables , with no finite mean values .
the latter has opened the door to anomalous _
statistics _ @xcite . in our approach
, we employ the levy distribution ( [ eq : levy ] ) merely as a _ mathematical function _ , which is a natural generalization of the standard gaussian function , thus leaving aside all the possible statistical interpretations and speculations concerning physical processes responsible for the md formation . a good example for
such `` applied '' approaches is the widespread use of the gaussian distribution , which describes the wave function of the ground state of the quantum harmonic oscillator @xcite , the green s function solution of the deterministic heat or diffusion equation @xcite , and frequently results as a form function in many different physical contexts .
while it is intuitively clear that it is the long - range correlations in the system that cause the emergence of levy distributions , the task to unravel the inherent physical mechanism(s ) yielding this anomalous distribution remains a challenge .
we think that the analysis of a reduced single - particle density matrix in a spirit of the theory of random levy matrices @xcite may shed additional light on this intriguing issue .
within the grand - canonical description , the number of bosons , @xmath0 , is a fluctuating quantity and the reduced single - particle density matrix is defined as the trace over the fock space , @xmath116 where the chemical potential @xmath117 is fixed to give the required number of particles @xmath118 in the system , @xmath119 with @xmath120 being the boltzmann factor , and @xmath121 is the grand - canonical partition function .
the latter coincides with that of non - interacting fermions , i.e. , @xmath122 $ ] , where @xmath123 stands for the single - particle energy spectrum . the trace over the the fock space ( [ eq : gcdmatrix ] ) can be evaluated exactly @xcite by mapping the problem of hard - core bosons on that of spinless fermions via the jordan - wigner transformation @xcite .
we next use the expressions for the elements of the reduced single - particle density matrix elaborated in @xcite , i.e. , @xmath124\right . \nonumber \\ & \left .
-\det\left[\mathbf{i } + \mathbf{d}(n , l)\right]\right\}\,,\mbox{for}\ , n\neq l,\end{aligned}\ ] ] @xmath125 and @xmath126_{n , n},\ ] ] for the main diagonal elements .
all operators entering above are the square matrices @xmath127 defined as follows : @xmath128 denotes the identity matrix , @xmath129 , @xmath130 , @xmath131 is the orthogonal matrix of eigenvectors satisfying the eigenproblem , @xmath132 , with single - particle version of hamiltonian ( [ eq : hamiltonian ] ) @xmath133 and @xmath134 is diagonal matrix of its eigenvalues , i.e. , @xmath135 .
thus , to obtain the entire matrix @xmath25 ( @xmath136 ) one has to compute @xmath137 determinants of @xmath127 matrices .
in the canonical formalism , the particle reservoir is absent , meaning that one has to find the number , @xmath138 , self - consistently from those many particle states , @xmath139 , together with their eigenenergies , @xmath140 , that contribute to the thermal superposition at a given temperature .
the reduced single - particle density matrix is then obtained as a sum of thermally weighted density matrices , @xmath141 , evaluated for each from @xmath138 eigenstates separately ; i.e. , @xmath142 the canonical partition function @xmath143 is expressed through the true many particle energy spectrum : @xmath144 .
the energy spectrum of hard - core bosons is the same as that for spinless fermions : @xmath145 , where @xmath146 denotes the numbers of single - particle eigenlevels occupied in the @xmath147-th many - particle state . in the case of a lattice confinement , @xmath148 .
the corresponding many - particle eigenstates can be represented as @xmath149 where @xmath131 is complete orthogonal set of single - particle eigenvectors ( see appendix a ) .
each contribution , @xmath150 , is related to the green function , @xmath151 : @xmath152 using the jordan - wigner transformation @xcite , the bosonic green function can be rewritten as scaler product of two fermionic wave - functions , and subsequently as a determinant of matrix product @xcite : @xmath153,\ ] ] where the matrix @xmath154 has @xmath155 columns @xmath156 in comparison to the grand - canonical approach , the number of operations needed to obtain the entire matrix @xmath157 is a factor of @xmath138 larger , and is growing with increasing temperature .
above we detailed the canonical formalism for hard - core bosons on a lattice . here , instead , we elaborate on a general case with arbitrary confinement , @xmath85 . in this case , the momentum distribution , @xmath38 , can be obtained from the reduced single - particle density matrix in the continuum : @xmath158 , where @xmath159 is given by the thermal superposition ( [ eq : cdmatrix ] ) with @xmath160 defined below . to calculate the reduced single - particle density matrix , @xmath161 , in the continuum we make use of efficient method @xcite , which represents it in terms of the single particle states : @xmath162 where @xmath127 matrix @xmath163 associated to the @xmath147-th many - particle state is
@xmath164 @xmath165 is a minor of matrix @xmath166 obtained by crossing @xmath90-th row and @xmath167-th column , and matrix @xmath166 itself is given as @xmath168 where @xmath169 are the eigenfunctions of single particle eigenproblem in a given trapping potential @xmath85 . in ( [ eq : perez_sup_2 ] ) , it is assumed that @xmath170 , while for @xmath171 : @xmath172 .
the set of @xmath173 , as before in appendix [ lattice ] , denotes the numbers of single - particle eigenlevels occupied in the @xmath147-th many - particle state .
additionally , whenever @xmath174 , ( [ eq : perez_sup_1 ] ) can be represented as @xmath175 @xcite , which is more efficient when implemented numerically . for a box and for a harmonic confinement , @xmath176 and @xmath177 , respectively , where @xmath178 are the hermite polynomials . for these confinements , the integral in ( [ eq : perez_sup_2 ] ) , and the density matrices @xmath161 can be obtained analytically for a small number of particles with the help of symbolic computational routines . however , expanding out @xmath147 times @xmath179 determinants of @xmath180 matrices for @xmath181 quickly becomes unwieldy .
therefore , in sec .
v , the reduced density matrices @xmath161 were calculated on a numerical grid .
the grid step was sufficiently small to ensure a smooth representation of all single particle wave - functions , @xmath182 , participating in excited many - particle states at a given temperature . | levy distribution , previously used to describe complex behavior of classical systems , is shown to characterize that of quantum many - body systems . using two complimentary approaches , the canonical and grand - canonical formalisms
, we discovered that the momentum profile of a tonks - girardeau gas , a one - dimensional gas of @xmath0 impenetrable ( hard - core ) bosons , harmonically confined on a lattice at finite temperatures , obeys levy distribution .
finally , we extend our analysis to different confinement setups and demonstrate that the tunable levy distribution properly reproduces momentum profiles in experimentally accessible regions . our finding allows for calibration of complex many - body quantum states by using a unique scaling exponent . |
the wilkinson microwave anisotropy probe ( wmap ) has provided strong evidence for an optical depth for electron scattering of 0.17 @xmath1 0.04 based on the large scale polarization pattern related to rescattering of cosmic microwave background ( cmb ) photons ( kogut et al .
if the reionization process is described as instantaneous and homogeneous , the measured optical depth implies a reionization redshift of @xmath2 in a spatially flat universe .
interestingly , the derived redshift for reionization is at the high end of expectations in models in which stellar populations are dominant reionization sources ( e.g. , cen 2003 ; fukugita & kawasaki 2003 ; venkatesan , tumlinson & shull 2003 ; wyithe & loeb 2003 ; sokasian et al .
thus , in some of these models , primordial stars , the so - called population iii stars , are assumed to be rather massive and hence have a large uv photon emission rate , as suggested by recent theoretical studies ( e.g. abel , bryan & norman 2002 ; bromm , coppi & larson 2002 ; bromm , kudritzki & loeb 2001 ; schaerer 2002 , 2003 ) .
the formation of such massive stars in the early universe has many important cosmological implications ( e.g. , carr , bond , & arnett 1984 ; oh , cooray kamionkowski 2003 ; yoshida , bromm & hernquist 2004 ) . because radiation below the lyman - limit is absorbed by neutral hydrogen , luminous sources at redshifts 10 and higher
are generally seen only in the near - ir band . in particular ,
if pop iii stars are formed primarily at redshifts between 10 and 30 , they are expected to contribute to the infrared background ( irb ) light at wavelengths between 1 and 5 @xmath0 m at present epoch ( bond , carr & hogan 1986 ) .
recent estimates based on theoretical models suggest that a large fraction of the irb total intensity may indeed be due to these stars ( santos et al .
2002 ; salvaterra & ferrara 2003 ; cooray et al .
a substantial irb could arise from the pop iii stars not only due to the direct emission associated with these stars , but also due to indirect processes that lead to free - free and lyman-@xmath3 emission from the ionized nebulae , or hii regions , surrounding these stars .
observationally , about 20 - 40% of the near - ir flux ( between @xmath4 1 @xmath0 m and 3 @xmath0 m ) has been resolved by point sources ( cambrsy et al .
2001 ; totani et al . 2001 ) ; while j- and k - band source counts go down to ab - magnitudes of @xmath4 28 and 25 , respectively , the cumulative surface brightness converges by a magnitude around 23 , say , in the k - band ( see for a review of existing data by pozzetti & madau 2000 ) . the missing ir flux could be either due to a high - z population of galaxies , such as the proto - galaxy population related to first stars ( santos et al .
2002 ; salvaterra & ferrara 2003 ) , or a population of low surface brightness galaxies at nearby redshifts .
while there still remains uncertainties , one can potentially understand the presence of first luminous sources and some details related to their population , such as the number density , by characterizing the background light from optical to infrared at wavelengths of a few micron . in this wavelength range ,
final products of these massive stars are also expected to contribute to irb , in addition to the direct emission from stars ; radiation from supernovae and primordial black holes , in the form of miniquasars , may be significant sources of emission .
the role of final products in contributing to the irb , relative to the progenitor stars , depends on the initial mass function of the first stars . as the end state of a massive star above @xmath5 is a complete collapse to a black hole (
heger & woosley 2002 ) , if the first stellar population is extremely top - heavy with all mass above this limit , then one expects roughly a similar number density of black holes as that of stars .
these black holes grow via accretion and/or mergers and may partly be responsible for early reionization if they radiate as miniquasars ( haiman & loeb 1998 ; madau et al .
2004 ; ricotti & ostriker 2003 ) . the uv output from such quasars , if not absorbed heavily by the surrounding torii , can contribute both to reionization and to the ir flux as viewed today .
the unresolved soft x - ray background at energies between 0.5 and 2 kev may rule out a large density of miniquasars at high redshifts as required to reionize the universe from the quasar emission alone ( dijkstra , haiman & loeb 2004 ) unless reionization by miniquasars , within uncertainties of models used in dijkstra et al .
, is completed by a redshift of 20 .
here , we suggest that if all massive stars are eventually converted to black holes and accrete efficiently at the eddington - rate ( as would be the case if miniquasars are responsible for eventual reionization of the universe ) , then the ir background is overproduced .
a more realistic situation could be that the mass function is broad , and only a small fraction of the total initial stellar mass is converted to black holes ; current understanding of the irb may limit the mass fraction of stars above 260 m@xmath6 to a level less than 10% .
this constraint can be relaxed if miniquasars radiate less efficiently below the eddington - limit ; such a scenario may be viable if stellar sources dominate the uv photon production and are , thus , responsible for reionization , though some fraction of stellar mass is converted to miniquasars . the initial mass function of the first stars could potentially be dominated by stars with masses in the range between 140 and 260 @xmath7 . in this case , the final product will be a complete disruption of the star via the pair - instability process ( heger & woosley 2002 ) . in the case of such massive stars , in general ,
one expects a typical supernovae to be much brighter than the star , suggesting that a modest contribution to the irb could come from supernovae , when compared to the fractional contribution from stellar emission alone .
however , though supernovae are in fact brighter than an average star , the duration over which most optical flux is emitted is relatively smaller .
we show that this results in a small but non - negligible contribution to the irb from supernovae associated with massive pop iii stars . while the background contribution is smaller , these supernovae are likely to be detectable directly in deep ir images above 1 @xmath0 m down to ab - magnitude limits at the level of 26 and fainter . in the next section ,
we discuss the contribution to irb from the first generation of stars , supernovae , and miniquasars under the assumption that the early stellar population is dominated by very massive stars . in the case of supernovae , we will suggest that , while the relative contribution to the irb may be lower , individual detections may be possible with deep - ir imaging .
we also suggest that the ir background itself may be used to constrain the presence of early mini - quasars . throughout the paper ,
we make use of a @xmath8cdm cosmological model consistent with current data ( e.g. , tegmark et al .
2004 ) with @xmath9 , @xmath10 , @xmath11 , the spectral index of the primordial power @xmath12 , and a normalization for the matter power spectrum with @xmath13 .
our calculations of the first - source contribution to the irb follow previous calculations in the literature related to the stellar contribution to the irb ( santos et al . 2002 ; salvaterra & ferrara 2003 ; cooray et al .
2004 ; kashlinsky et al .
these studies involve two basic ingredients : ( 1 ) the rate at which the volume density of first - sources evolves and ( 2 ) the flux spectrum .
we extend these calculations to supernovae and miniquasars , which have been so far ignored in previous estimates , under the assumption that ( 1 ) the supernova rate in the early universe traces that of the star formation and ( 2 ) the formation rate of miniquasars is simply related to the formation rate of seed black holes , taken to be proportional to the star - formation , and to a model description for accretion .
we use a simple model to describe the spectrum and the light curve of supernovae following results from heger et al .
( 2002 ) . for miniquasars , we make use of the average spectrum of quasars derived by sazonov et al .
( 2004 ) using fits to observational data over the range of wavelengths corresponding to x - rays and optical and , thereafter , using models to describe the emission , such as in the ir band . while the spectral shape is well defined in the average sense for observed luminous quasars at redshifts less than @xmath4 6 , it is unclear to which extent this spectrum can be applied to miniquasars at redshifts greater than 10 and with masses in the range of 10@xmath14 m@xmath6 to 10@xmath15 m@xmath6 . at energies below 13.6 ev , the sazonov et al .
( 2004 ) spectrum , @xmath16 , scales as @xmath17 though expectations for miniquasars are that the spectrum is harder such that @xmath16 is a constant ( madau et al .
2004 ; dijkstra , haiman & loeb 2004 ) .
regardless of the exact shape of the spectrum , there is also an uncertainty with respect to the overall normalization , or the total luminosity of a given miniquasar . here
, we make the assumption that miniquasars radiate efficiently at the maximum allowed by the eddington - limit . in making this choice ,
we are guided by observations of quasars out to redshifts of 6 which indicate maximal emission at the eddington - limit .
these quasars are more massive and more luminous than the miniquasar counterparts at high redshifts .
it is likely that the miniquasar emission is submaximal .
the limits we derive here on the fraction on miniquasars present during the reionization era should , thus , only be taken as a conservative limit on the low end given the current state of knowledge on the ir background . in order to calculate the formation rate of first stars and the subsequent supernovae or seed black holes , we use an analytical description for halo formation following the press - schechter formalism ( press & schechter 1974 ) .
we write the star - formation rate as @xmath18 where @xmath19 is the minimum mass of halos in which gas can cool , and we take the star - formation efficiency @xmath20 .
we consider two cases for @xmath19 with a minimum virial temperature of 400 k , involving molecular hydrogen ( e.g. , tegmark et al .
1997 ) , and 10@xmath15k involving atomic hydrogen line cooling ( barkana & loeb 2001 ) .
the respective star - formation rates are shown in fig . 1 ( top panel ) .
we mention that , although more detailed modeling of early star formation is available ( e.g. hernquist & springel 2003 ; yoshida et al .
2003 ) , the overall feedback effects from mini - quasars on star - formation are still uncertain ( see , e.g. machacek , bryan , abel 2003 ) .
we thus prefer using the model described above for the sake of simplicity .
note that the scenario with a minimum temperature @xmath21k likely provides a possible maximal star - formation rate .
given the rate of formation of stars , we can write the emissivity per comoving unit volume at a certain wavelength @xmath22 as a function of redshift @xmath23 as @xmath24 where @xmath25 is the luminosity per source mass as a function of frequency and @xmath26 is the lifetime over which this flux is emitted ; for a single source at a redshift of @xmath23 with a source mass @xmath27 , such as a first - star , the total luminosity , as a function of frequency , is @xmath28 and the observed flux today is @xmath29 $ ] , where @xmath30 is the cosmological luminosity distance out to a redshift of @xmath23 . the cumulative background is obtained by integrating the emissivity over redshift , yielding the specific intensity @xmath31 @xmath32 where @xmath33 is the observed frequency , @xmath34 is the redshift scaling of the frequency , and @xmath35 is the proper distance such that @xmath36 $ ] when the expansion rate for adiabatic cold dark matter cosmological models with a cosmological constant and a flat space - time geometry is @xmath37 $ ] . in terms of the individual source fluxes , and their comoving number density @xmath38 , the specific intensity of the background can also be written as @xmath39 , where @xmath40 and @xmath41 is the cosmological volume element given by @xmath42 , in terms of the luminosity distance .
this expression and equation ( [ eqn : back ] ) are equivalent .
we use the same formalism to compute the background radiation intensities from stars , supernovae and miniquasars .
we assume that the supernovae rate follows that of the star - formation . for miniquasars
, we follow ricotti & ostriker ( 2003 ) and write the growth rate of black holes as @xmath43 where the two terms represent the growth of black holes via accretion , @xmath44 , and the formation rate of seed black holes , @xmath45 .
the over - dot represents the derivative with respect to proper time .
we ignore black hole removal process , such as through ejection from proto - galaxies , and take @xmath46 where @xmath47 , the star - formation rate , is given by equation ( 1 ) , and @xmath48 represents the overall fraction of stars converted to seed black holes .
we model growth related to accretion as @xmath49 , where eddington time is set to be a constant at 10@xmath50 years and use the same parametrization in ricotti & ostriker ( 2003 ) for @xmath51 such that @xmath52^\phi$ ] with the condition that @xmath53 based on the definition that @xmath54 where @xmath55 and @xmath56 are time intervals over which the blackhole accretes and does not accrete , respectively . at the low end , following ricotti & ostriker ( 2003 ) , we set @xmath57 to be consistent with both the agn fraction at @xmath58 and today ( steidel et al . 2002 ) . in fig . 1 ( bottom panel )
, we show @xmath59 normalized to the baryon density and assuming @xmath60 for a variety of models related to @xmath51 with parameters as labeled on the figure .
the spectrum of stellar emission is computed as in santos et al .
( 2002 ) , including the nebular and free - free emission .
for the miniquasar spectrum , we make use of the model of sazonov et al . ( 2004 ) .
we assume that the sazanov et al.s template for the average quasars also applies to miniquasars . in fig . 2
, we compare the spectrum of a typical star with @xmath61 and that of a miniquasar at a redshift of 15 , as viewed today . here
we plot @xmath62 ( in units of w m@xmath63 sr@xmath64 ) .
the miniquasars are assumed to radiate at the eddington - limit such that the total integrated luminosity over the range from x - rays to far - infrared is @xmath65 ergs s@xmath64 ( m@xmath66 ) , though , as discussed in section 2 , it is likely that the miniquasar emission is submaximal . also , at observed wavelengths below @xmath4 2 @xmath0 m , corresponding to redshifted lyman-@xmath3 emission , the miniquasars flux spectrum @xmath16 scales as @xmath17 . for miniquasars
, it could be that the spectrum is harder ( in terms of energy ) , such that the expectation is that @xmath16 is constant at energies above 13.6 ev ( madau et al .
2004 ; dijkstra , haiman & loeb 2004 ) .
this will result in a flatter spectrum , as observed today , from optical / uv to ir wavelengths . as shown in fig . 2
, we assume that stellar emission is responsible for reionization of the universe ; the lyman-@xmath3 emission is related to ionizing photons that are first absorbed by the neutral medium and are re - emitted during recombinations . in the case of miniquasars , we do not consider uv absorption as we have implicitly assumed that the surrounding medium is already ionized by stars preceding the formation of a miniquasar in or near that location .
we will comment on this later in the discussion . as shown in fig . 2
, the level of the incident flux from a typical star and a miniquasar is similar .
the spectrum and light curve for population iii supernovae were calculated by heger et al .
in these calculations , the relevant ir emission , as observed today , characterizing the peak of the light curve , is delayed a month or so from the initial explosion and the resulting shock break out .
the latter includes emission at short wavelengths , below the lyman limit , in the supernova frame , which is likely be absorbed by the intergalactic medium at epochs prior to complete reionization .
the peak of the light curve extends over a month at most . in order to make a reasonable estimate of the supernovae contribution to the irb
, we use the peak emission and its duration only and ignore detailed aspects of the light curve when fluxes drop below more than a factor of 10 from the peak emission . including the full light curve only leads to minor corrections at a a few percent level . while the peak emission is flat such that @xmath67 is a constant in the estimate by heger et al .
( 2002 ) , to study any departures , we also allow the spectrum to vary as @xmath68 with @xmath3 between -0.5 and 0.5 , with @xmath69 as the fiducial case . when varying the spectral shape , we renormalize such that the total flux , or the luminosity , remains constant in the rest wavelengths between 10@xmath14 @xmath70 and 10@xmath15 @xmath70 corresponding to the uv and optical regimes . here , again
, we assume that the surrounding medium is reionized by stellar emission before the supernova explosion and that the rest uv emission is not absorbed .
the reionization by supernovae may not be important as the time scale over which the rest uv emission is expected is significantly smaller than that of a star ( a month vs. a few million years , respectively ) .
note that the supernova spectrum of a pop iii star is highly uncertain in terms of spectral shape at the uv end and more numerical models are needed to further address the extent to which supernovae may be an important source of ionizing photons .
a more likely scenario involving pop iii supernovae is the one described in oh , cooray & kamionkowski ( 2003 ) , where the blast wave propagates to the dense ionized igm and heats the electrons to a substantial temperature which subsequently cools via inverse compton - scattering off of the cosmic microwave background , similar to the same effect related to hot electrons in galaxy clusters ( the sunyaev - zeldovich effect ; sunyaev & zeldovich 1980 ) .
we summarize our results in fig . 3 , where we show the contributions to total irb light from stars , supernovae and miniquasars .
in the case of stars , we divide the contribution to the three main components ; direct emission from stars , lyman-@xmath3 recombination radiation from the ionized patch , and the free - free component related to electron - ion scatterings within each ionized patch surrounding these stars .
our calculations related to the irb light due to stars are consistent with those in santos et al .
( 2002 ) and salvaterra & ferrara ( 2003 ) , except that here we have assumed a slightly different redshift limit to the first generation of stars at a value of 10 instead of values such as 7 and 8 used in those calculations that are aimed at explaining the total missing content .
our results suggest that , while the emission associated with stars can explain most of the irb intensities , the fractional contribution from supernovae is an order of magnitude smaller than the stellar emission contribution .
the relative difference between stars and supernovae is easily understood based on the typical flux from a star vs. a supernovae and the ages over which these fluxes are emitted .
for example , typical @xmath4 200 m@xmath71 star at a redshift of 20 has a flux , ignoring lyman-@xmath3 emission , of order @xmath4 10@xmath72 ergs cm@xmath63 s@xmath64 hz@xmath64 ( @xmath74 njy or @xmath75 ) , while the peak flux of the supernovae at the same redshift is at the level of @xmath4 few times 10@xmath76 ergs cm@xmath63 s@xmath64 hz@xmath64 ( @xmath4 1 @xmath0jy or @xmath4 @xmath77 ) . even if all the stars at redshifts greater than 10 die as supernovae , though the instant flux is greater , the total output of @xmath78 from supernovae is smaller than stars as the stellar emission lasts over a million years while the peak flux from supernovae lasts just over a month or so .
thus , the apparent high ratio of the _ peak _ luminosity is largely compensated by the age difference so that the typical ratio of a supernova to stellar flux remains at most at the level of @xmath4 0.1 .
note that the stellar contribution to the irb also includes a substantial fraction from the lyman-@xmath3 emission of absorbed ionizing uv photons during subsequent recombinations .
we have not included such a contribution from individual pop iii supernovae because of uncertainty in the supernovae spectrum at the uv end .
note also that regions surrounding supernovae is likely to have been ionized by the progenitor stars .
even if the igm between proto - halos are partly ionized , the resulting lyman-@xmath3 emission is expected to be suppressed relative to the case of stellar sources .
this can be understood from the fact that the lyman-@xmath3 spectrum , as a function of frequency , can be written as @xmath79 , where @xmath80 is the fit to the lyman-@xmath3 profile of loeb & rybicki ( 1999 ) by santos et al .
( 2002 ) and the production rate of lyman-@xmath3 photons from recombinations is @xmath81 , where @xmath82 is the electron density of the surrounding medium , @xmath83 is the case - b recombination coefficient for hi as a function of the electron temperature @xmath84 , and @xmath85 is the ionizing photon production rate ( see santos et al .
2002 for details ) . to estimate @xmath85 , we assume that the supernovae spectrum is flat ( in rest @xmath16 below the lyman - limit ) and given the uncertainty in spectral shape
, we take the production rate to be that associated with ionizing photons at the wavelength corresponding to the lyman - limit . the model related to stars
is shown in fig . 2 , where the unabsorbed spectrum peaks at a wavelength of @xmath4 300 @xmath70 ( rest frame ) .
the difference in flux , as suggested above , translates to a stellar - to - supernovae ionizing photon production rate difference of @xmath4 10@xmath86 to 10@xmath87 .
this should then be compared to the ratio of ages , which is again , @xmath88 between stars and supernovae .
furthermore , in the case of supernovae , the lyman-@xmath3 radiation is more likely to be associated with tenuous igm with density @xmath89 @xmath90 compared to dense nebulae surrounding stars ( with density @xmath91 @xmath90 ) .
this lowers the lyman-@xmath3 flux by another factor of 10@xmath92 relative to that of stars , resulting in an overall reduction of 10@xmath86 to 10@xmath93 in the lyman-@xmath3 contribution from stars to supernovae .
thus , it is unlikely that the difference between predicted supernovae background and the missing irb flux can be reconciled with the expected increase associated with lyman-@xmath3 emission if the uv emission from supernovae are absorbed during the reionization process .
intriguingly , redshifted light from early miniquasars can explain the ir excess at wavelengths around 1 @xmath0 m , as shown in fig 3(c ) .
in fact , if _ all _ stars collapse to become seed black holes , we find that the flux from miniquasars that grow down to @xmath94 will _ exceed _ the measured ir fluxes by an order of magnitude or more .
a fractional contribution of @xmath95 to 10% of initial stellar mass to seed black holes appears consistent with observations .
this is based on our assumption that the miniquasars are radiating efficiently at the maximum given by the eddington - limit . as discussed at the beginning of section 2 ,
the extent to which such an assumption , based on observations of luminous quasars at redshifts less than 6 , applies to high - redshift miniquasars is unclear .
if the radiation from miniquasars , with masses between 10@xmath14 m@xmath6 and 10@xmath15 m@xmath6 , is submaximal , the derived limit on the fractional contribution from @xmath95 to 10% can be increased .
thus our result should be considered as a _ weak _ constraint . since the typical stellar age is @xmath96 years ( bromm , kudritzki & loeb 2001 ) , at a redshift of around 10 , using the highest star - formation rate in fig . 1 ( top panel )
, one finds a typical density of order a few times @xmath97 @xmath7 mpc@xmath98 in stars . from the bottom panel of fig . 1
, one finds that the density of miniquasars is typically
@xmath99 @xmath7 mpc@xmath98 .
this apparent increase in miniquasar density , however , is compensated by the fact that the miniquasars flux , at relevant wavelengths , is roughly an order of magnitude smaller than that of a typical very massive star ( see fig . 2 ) .
the exact numerical calculation reveals that the miniquasars do dominate the background if the fraction of stars converted to seed black holes is large . in other words ,
the irb offers a useful way to constrain the presence and abundance of miniquasars at @xmath100 ; under our assumption , an initial mass fraction of black holes , relative to stars , of order 10% and below seems consistent with current observations .
this conclusion may indeed be strengthened if ionizing photons from miniquasars are absorbed by the neutral igm ; if absorbed , we expect some fraction of the energy to be reradiated in the form of the lyman-@xmath3 line and to increase the irb at the corresponding frequencies such that the upper bound on the black hole mass fration is further reduced . on the other hand ,
as discussed in the case for supernovae above , we do not expect the lyman-@xmath3 radiation to be significant as long as the dense nebula surrounding quasar is fully ionized already by the stellar emission prior to the formation of the miniquasar .
the miniquasar emission can ionize the igm beyond the nebula , but due to the lower density , recombinations are unlikely to be significant such that the lyman-@xmath3 emission is suppressed relative to the case where stars form and ionize dense nebulae surrounding such stars initially .
our suggestion that the seed mass fraction must be below 10% , assuming efficient radiation at the eddington - limit , also suggests that the universe can not be reionized by early quasars alone . for reference , in fig . 3
, we show the model in which the uv photons from miniquasars are absorbed by neutral igm during the reionization process ( thick dot - dashed line in the figure ) . integrating the uv light below the hydrogen lyman - limit
, we find a ratio of number density of photons relative to mean density of baryons at the level of 0.3 at @xmath101 ; to fully reionize , one would expect this ratio to be at least unity or as high as 25 as the recombinations are important at high redshifts . in oh , cooray & kamionkowski ( 2003 ) , it was estimated that @xmath102 photons per baryon are needed to explain an optical depth to reionization of 0.12 when the clumping factor of ionized gas @xmath103 is 4 and the escape fraction of ionized photons is 0.3 , consistent with other estimates ( e.g. , dijkstra , haiman & loeb 2004 ) .
note also that ionization by a combination of hard x - rays and secondary electrons does not lead to complete ionization , as argued by madau et al .
( 2004 ) .
our constraint on the total mass fraction of miniquasar - seed blackholes is independent of similar suggestions in the literature , for example , based on the unresolved x - ray background ( dijkstra , haiman & loeb 2004 ) .
these authors argue that the x - ray background would be violated by a population of miniquasars if they are responsible for complete reionization . if we consider a scenario where miniquasars exist with stars , with a seed miniquasar mass fraction of 10% , we find that the x - ray background constraint at 1 kev is not violated by this miniquasar population even in the extreme case where the spectrum is normalized to the eddington - limit .
interestingly , the level of the irb background from the same population is consistent with current observations .
given the uncertainties in the available observations and in our analytical model , we are not yet able to exactly establish the allowed seed mass fraction .
although current observations of the irb are still uncertain , one can improve the limit related to the mass fraction of seed black holes further by future background measurements . in fig .
3(c ) , we show the 1@xmath104 noise , or detection , level related to a planned low - resolution spectroscopic background measurement on a rocket experiment ( t. matsumoto , private communication ) for irb fluctuation studies ( e.g. , magliocchetti et al .
2003 ; cooray et al .
the detection limit comes from the addition of 400 spectra over a 30 second integration time at an altitude above 300 km ( which removes the air - glow ) and after a foreground source subtraction down to 14 , 15.9 and 16.3 magnitudes in i- , j- and k - bands .
the background can be measured to a level of 1 nw m@xmath63 sr@xmath64 , whereas the limiting factor in these measurements will be related to the removal of the zodiacal light ( e.g. , cambrsy et al . 2001 ) . in this case of the rocket experiment , this removal is achieved with data taken as a function of the latitude such that , when combined with multifrequency aspect , removal down to the instrumental noise level is expected .
the proposed measurement would help establish the excess above galactic stars and foreground sources better than what is currently known and , in return , limit the extent to which high - z sources , either in the form of stars and/or miniquasars , may be present
. one may consider the possibility whether the constraint from the irb can be relaxed if the rest uv and optical emission from miniquasars is heavily absorbed by dust .
first , we note that the extinction by inter - galactic dust is unimportant at @xmath100 even if nearly all the first stars die as pair - instability supernovae ( see yoshida et al .
if the inter - stellar absorption due to dust within ( proto-)galaxies that are hosting miniquasars is significant , the reradiation by thermal dust at far - ir wavelengths at @xmath105 m and above can still be used to put limits on their abundance .
assuming the same spectrum as shown in fig . 2 for quasars , which allows for significant dust absorption , we predict a 850 @xmath0 m background of @xmath4 50 nw m@xmath63 sr@xmath64 , for @xmath106% while the observed background is below a few nw m@xmath63 sr@xmath64 ( hauser & dwek 2001 ) .
this suggests that even for the 5% seed mass case , which may explain the missing irb , absorption by dust can not be as significant as for low redshift luminous quasars . keeping the same seed mass fraction
, one can refine the extent to which dust absorption is significant by studying the unresolved background at these far - ir wavelengths . while @xmath4 40% of the 850 @xmath0 m and 350 @xmath0 m backgrounds have now been resolved , the counts are such that a small extrapolation to a slightly lower flux density resolves the whole background ( borys et al .
2003 ) . unlike the near - ir background where surface brightness converges down to ab magnitudes of @xmath107 , as observed
counts go deeper , such a convergence is not found at far - ir wavelengths .
thus , it ll be useful to return to this aspect in the future when source counts are well defined . as an early conclusion ,
we are forced to consider the possibility that the far - ir background can be fully explained with resolved source counts alone .
this , when combined with irb measurements , suggests that a substantial population of `` dusty '' miniquasars is unlikely to be present at @xmath100 .
the cosmic infrared background ( irb ) at wavelengths between 1 @xmath0 m and 3 @xmath0 m provides a useful probe of the global star - formation rate prior to and during the reionization .
we have studied the relative contributions from first stars , supernovae and early miniquasars to the infrared background ( irb ) .
in addition to massive stars , the irb flux may be dominated by a population of miniquasars , especially if the first generation of stars are very massive and the end product of the stars lead to a substantial population of black holes that will effectively radiate as miniquasars by matter accretion .
we use the press - schechter formalism that describes dark halo formation , and combine it with a simple star - formation model to calculate the cosmic star - formation history , the supernova rate , and the blackhole formation rate at high redshift .
we follow previous calculations of the source spectrum for stars , while for population iii supernovae , we use a simple description of the peak emission and its duration .
we describe the flux spectrum of miniquasars using an average spectrum as observed for luminous quasars at redshifts less than 5 based on observed data and models by sazonov et al .
we fix the freedom related to the overall normalization of this spectrum by assuming that the miniquasars radiate efficiently at the maximum allowed by the eddington - limit . under this assumption , we find that a mass fraction of seed black holes , relative to pop iii stars , at the level of @xmath108 and below is consistent with observations
. this upper limit on the seed mass fraction can naturally be extended to a higher value if miniquasars radiate below the eddington - limit , so the limit we derive should be considered as on the lower side rather than a strict upper limit .
future spectroscopic background measurements exploiting rockets will improve the observational uncertainties and hence the constraints on the high - redshift miniquasar population . while the integrated light from population iii supernovae to the irb is subdominant ,
one can potentially consider the possibility of directly detecting individual supernovae in high resolution deep ir imaging data . given the star formation rate density and the expected flux , we estimate the surface density of supernovae to be order few tens per sqr .
degree over a year , if the star - formation rate is close to 0.5 m@xmath71 yr@xmath64 mpc@xmath98 at @xmath94 , and assuming that all stars are massive ( @xmath109 m@xmath71 ) . for the high redshift supernovae , the typical ab magnitudes are at the level of 26 in 1.5 @xmath0 m ( heger et al .
these are within reach of deep ir imaging with existing large telescopes and from space and , in the long term , clearly within reach of missions such as the james webb space telescope ( jwst ) , which can detect point sources down to a magnitude limit of 31 around the same wavelengths in a 10@xmath15 sec exposure in its 16 arcmin@xmath14 field of view .
_ acknowledgments : _ this work is supported by the sherman fairchild foundation and doe de - fg 03 - 92-er40701 ( ac ) .
ny thanks support from japan society of promotion of science special research fellowship ( 02674 ) .
we thank an anonymous referee for a careful reading of the manuscript and for detailed comments that improved the discussion in this paper .
heger , a. , woosley , s. e. , baraffe , i. & abel , t. 2002 , in .
gilfanov , m. et al , eds . "
lighthouses of the universe : the most luminous celestial objects and their use for cosmology , springer - verlag , berlin [ preprint astro - ph/0112059 ] | the cosmic infrared background ( irb ) at wavelengths between 1 @xmath0 m and 3 @xmath0 m provides a useful probe of early star - formation prior to and during reionization . to explain the high optical depth to electron scattering , as measured by the wilkinson microwave anisotropy probe ( wmap ) , one requires significant star - formation activity at redshifts 10 and higher .
in addition to massive stars , the irb flux may be contributed by a population of early miniquasars .
we study the relative contributions from first stars , supernovae and quasars to the irb for reasonable star formation rates at high redshift .
if miniquasars radiate efficiently at the eddington - limit , current background measurements limit the fraction of mass in first stars that is converted to seed black holes to be roughly less than 10% . in the case of supernovae , though an individual supernova is much brighter than the progenitor star , due to the shorter lifetime of order few months , the fractional contribution to the irb remains at a level of 10% and below when compared to the same contribution from stars .
the bright supernovae may , however , be directly detectable by future large ground - based and space telescopes .
infrared : general stars : formation cosmology : observations diffuse radiation |
in the vicinity of the equilibrium state nuclear matter is stable with respect to particle density and fermi surface distortions and the excitation of both the isoscalar and the isovector sound modes is possible .
propagation of an isoscalar sound wave in nuclear matter depends crucially on the landau parameter @xmath0 in the quasiparticle interaction amplitude and on the relaxation processes . for zero temperature
, there is an underdamped zero - sound mode at @xmath1 because the phase velocity of the sound wave exceeds the speed of particles inside the fermi sphere .
a strong landau damping appears at @xmath2 where a nonzero transfer of energy from the wave to particles is possible @xcite and the wave transforms to an overdamped mode .
the picture of propagation of the zero - sound wave is essentially more complicated in the case of a hot nuclear matter . due to the existence of the temperature tail of the equilibrium distribution function
, the phase conditions for the landau damping are fulfilled here for positive values of the @xmath0 and a possibility for a propagation of the sound wave appears in the region @xmath2 @xcite .
the zero sound is transformed to the first - sound mode in the limit of strong interaction , @xmath3 @xcite or in the frequent collision regime at high temperatures @xcite .
it is necessary to stress that the sound velocity @xmath4 is directly related to the nuclear matter incompressibility coefficient @xmath5 for the first - sound mode only . in this case
, one has @xmath6 , where @xmath7 is the velocity of the first sound . in general , the sound velocity @xmath4 is a complicated function of both @xmath5 and the dimensionless collisional parameter @xmath8 , where @xmath9 is the relaxation time and @xmath10 is the eigenfrequency of the sound mode . in the present work
we obtain a simple analytical expression for the sound velocity @xmath4 which provides a description for both the frequent- and rare - collision limit as well as for the intermediate cases .
special attention is paid to the propagation of the isovector sound and for the zero- to first- sound transition for this mode .
it is well known @xcite that hydrodynamic approaches like the goldhaber - teller or steinwedel - jensen models give a reasonable description of the nuclear isovector giant resonances ( ivgr ) .
however it is not the case for the isoscalar giant resonances ( isgr ) , where the fermi surface distortion effects plays an important role @xcite and the traditional hydrodynamic model can not be applied . in our approach
, this situation is directly related to the peculiarities of the propagation of both isovector and isoscalar sounds . the interparticle collisions on the distorted fermi surface lead to the collisional damping of the sound wave .
two limiting regimes @xmath11 and @xmath12 provide the existence of the non - damped first and zero sounds , respectively , @xcite . in what follows we will use the collisional kinetic theory , taking into account the memory effects on the collision integral @xcite .
we will discuss a special feature of the temperature dependence of the attenuation of the sound mode in hot nuclear matter .
in particular , we will show that memory effects lead to a bell - shaped form of the attenuation coefficient @xmath13 as a function of the temperature , providing a correct behavior of @xmath13 in both limiting regimes @xmath14 and @xmath12 . with decreasing bulk density or increasing temperature
the nuclear matter reaches the regions of mechanical or thermodynamical instabilities with respect to small particle density distortions @xcite and to separation into liquid and gas phases @xcite . the general instability condition of the fermi liquid reads @xmath15 @xcite , where @xmath16 is the landau s parameter in the expansion of the quasiparticle interaction amplitude in legendre polynomial @xcite .
however the development of instability depends not only on the equation of state , but also on the dynamical effects such as the dynamical fermi - surface distortion ( fsd ) effect @xcite .
the fsd effects strongly reduce , by the factor @xmath17 the instability growth rate @xmath18 in the unstable region @xmath19 @xcite .
in the present work , we will consider the influence of the fermi - surface distortion , relaxation and memory effects on the instability growth rate @xmath18 .
the plan of the paper is as follows . in sec .
ii we derive the general equation of motion for the particle density vibrations in the presence of the memory effects . we introduce the renormalized incompressibility coefficient and the viscosity coefficient which are frequency dependent due to the memory effect . in sec .
iii we obtain a simple expression for the refraction and attenuation coefficients for both the isoscalar and the isovector sound waves which propagate in a hot nuclear matter .
we consider also the mixing of the isoscalar and isovector modes in an asymmetric nuclear matter taking into account the relaxation and the memory effects .
the development of the bulk instability in a low density nuclear matter in the presence of the relaxation and the memory effects is considered in sec .
the conclusion is given in sec .
we start from the collisional kinetic equation @xcite @xmath20 here @xmath21 is the collisional integral , @xmath22 is the self - consistent mean field and @xmath23 @xmath24 is the wigner distribution function in which we will take into account only the distortion of the fermi sphere with multipolarities @xmath25 @xmath26 the distribution @xmath27 corresponds to the spherical fermi surface and @xmath28 represents both the quadrupole deformation and the displacement of the fermi surface . for small deviations from a fermi sphere the right - hand side ( rhs ) @xmath21 of ( [ 2.1 ] ) is a collision integral linearized in @xmath28 and it can be represented in the form @xmath29 which takes into account the memory effects due to the memory kernel @xmath30 . in this paper
we will not use an explicit form of @xmath31 .
below we will need @xmath21 only for periodic oscillation of @xmath32 with the eigenfrequency @xmath33 assuming the restriction @xmath34 for the fermi surface distortion , the collision integral ( [ 2.3 ] ) can be written in the form of the extended @xmath35-approximation , see refs .
@xcite , @xmath36 where the relaxation time @xmath37 is @xmath10-dependent due to the memory effects .
the assumption in eq .
( [ 2.2 ] ) allows us to reduce the collisional kinetic equation ( [ 2.1 ] ) to the local equations of motion for the particle density @xmath38 the displacement field @xmath39 and the pressure tensor @xmath40 derived as the lowest @xmath41-moments of the distribution function @xmath42 : @xmath43 here @xmath44 is the velocity field and @xmath45 is the spin - isospin degeneracy factor . taking the first two @xmath41-moments of eq .
( [ 2.1 ] ) one obtains , see refs .
@xcite , @xmath46 where @xmath47 is the pressure due to motion of nucleons without distortion of the fermi sphere and @xmath48 is associated with quadrupole distortion of the fermi surface : @xmath49 the pressure tensor @xmath48 is responsible for the dissipative processes . taking the second @xmath50-moment of eq .
( [ 2.1 ] ) one obtains the following equation for the pressure tensor @xmath51 the equations of motion ( [ 2.5 ] ) and ( [ 2.6 ] ) are closed .
they can be applied to both the isoscalar and isovector sound excitations .
let us consider the isoscalar compression mode . to simplify the problem
, we can rewrite the expression in the parenthesis in eq .
( [ 2.5 ] ) near the equilibrium value of the density @xmath52 as @xmath53 , \label{2.7}\ ] ] where index @xmath54 refers the equilibrium state , @xmath55 is the energy density of particles @xmath56 @xmath57 is the kinetic energy density @xmath58 and @xmath59 is the potential energy density which is related to the mean field @xmath60 by @xmath61 note that @xmath62 is the chemical potential , which does not depend on the space coordinate @xmath63 for the equilibrium state of the nucleus .
we have used this fact when deducing eq .
( [ 2.7 ] ) . solving eq .
( [ 2.6 ] ) with respect to @xmath64 using eq .
( [ 2.7 ] ) and the continuity equation @xmath65 , we obtain an equation for the density vibration in the form @xmath66 here @xmath67 where @xmath68 is the static incompressibility and @xmath69 is the viscosity coefficient @xmath70 we point out that there is a significant difference between the static nuclear incompressibility coefficient , @xmath5 , i.e. , derived as a stiffness coefficient with respect to a change in the bulk density , and the dynamic one , @xmath71 of eq .
( [ 2.13])@xmath72 associated with the sound propagation .
this difference is due to the second term on the rhs of eq .
( [ 2.13 ] ) caused by the fermi - surface distortion effects .
the quantity @xmath69 in eq .
( [ 2.14 ] ) determines the time irreversible contribution to the pressure tensor @xmath73 and can be considered as the viscosity coefficient due to the relaxation occurring on the distorted fermi surface . expression ( [ 2.14 ] ) is valid independently of the nucleon s collision rate .
the viscosity goes to zero in both the rare , @xmath74 and frequent , @xmath75 collision limits .
assuming a plane wave solution @xmath76 one obtains from eq .
( [ 2.11 ] ) the following dispersion relation @xmath77 the solution of this equation defines the complex wave number @xmath78 ( @xmath10 is real ) . a simple solution to eq . ( [ 3.1 ] )
can be obtained in two limiting cases of the frequent collision ( first sound ) regime , @xmath79 and the rare collision ( zero sound ) regime , @xmath80 the sound velocity @xmath81 is given by @xmath82 where @xmath83 ( we adopted the kinetic fermi energy @xmath84 ) .
we point out that the value of @xmath85 is comparable with the static incompressibility @xmath86 mev @xmath87and we have @xmath88 .
the factor @xmath89 in this relation is due to the restriction @xmath25 for the multipolarity @xmath90 of the fermi surface distortion . in a general case of arbitrary @xmath91 this factor
is increased to @xmath92 @xcite .
the result ( [ 3.cc ] ) means that in contrast to the first sound ( frequent collision ) regime , the sound velocity of the compression mode can not , in general , be used directly to extract the static incompressibility of @xmath5 because of the additional contribution from the fermi surface distortion effects which result in the renormalization of the incompressibility @xmath5 into @xmath71 . using both asymptotic sound velocity @xmath7 and @xmath93 the solution to the dispersion relation ( [ 3.1 ] )
can be written as @xmath94 where the refraction coefficient @xmath95 and the attenuation coefficient @xmath96 ( both real ) are obtained from the following equation @xmath97 in the frequent collision ( first sound ) regime we obtain from eq .
( [ 3.nkapa ] ) @xmath98\qquad { \rm if\ \ \ \ \ } \omega \tau _ { r,\omega }
\label{3.nkapa1}\ ] ] in the opposite case of the rare collision ( zero sound ) regime we obtain @xmath99/(2\omega \tau _ { r,\omega } ) \qquad { \rm if\ \ \ \ \ } \omega \tau _
{ r,\omega } \gg 1 .
\label{3.nkapa0}\ ] ] the attenuation coefficient @xmath13 in both limiting regimes is a complicated function of the frequency @xmath10 because of the memory effect in the relaxation time @xmath37 . in the case of sound propagation in hot nuclear matter ,
the competition between the temperature smoothing effects in the equilibrium distribution function and dynamical distortions of the particle momentum distribution leads to the following expression for the relaxation time @xcite @xmath100 where @xmath101 is the temperature of nuclear matter and the @xmath10-dependence of @xmath37 is due to the memory effects in the collision integral .
below we will use @xmath102 @xcite and@xmath103 ( for the isoscalar mode ) @xcite .
equations ( [ 3.1 ] ) and ( [ 3.nkapa ] ) are valid for arbitrary collision times @xmath37 and thus describe both the zero and the first sound limit as well as the intermediate cases . from it
one can obtain the leading order terms in the different limits mentioned . in fig .
1 we have plotted both coefficients @xmath95 and @xmath13 as obtained from eq .
( [ 3.nkapa ] ) . in the high temperature limit
, the system goes to the frequent collision ( first sound ) regime with the saturated refraction coefficient @xmath95 @xmath104 ( we use the factor @xmath92 instead of @xmath89 assuming the contribution of the higher multipolarities @xmath105 in the fermi surface distortion as was mentioned above ) and the attenuation coefficient @xmath106 . in the opposite low temperature limit ,
the system is close to the zero sound regime with @xmath107 we point out a shift of both @xmath95 and @xmath13 by nonzero values at @xmath108 this is due to the memory effect in the relaxation time @xmath37 of eq .
( [ 3.tau ] ) : in the very high frequency limit , the system can exist close to the first sound regime at @xmath109 even at zero temperature .
the position of the maximum of @xmath110 in fig .
1 can be interpreted as the transition temperature @xmath111 of zero- to first- sound regimes in a hot fermi system .
the value of @xmath112 depends slightly on the sound frequency @xmath10 and it is shifted to smaller values with the increase of @xmath33 let us consider now the isovector sound mode in a symmetric nuclear matter with @xmath113 where @xmath114 and @xmath115 are the equilibrium neutron and proton density respectively .
the general equations of motion ( [ 2.5 ] ) and ( [ 2.6 ] ) are still correct .
however the energy density @xmath55 in eq .
( [ 2.7 ] ) is related now to the symmetry energy @xmath116 .
the corresponding first sound velocity @xmath7 for the isovector mode is given by , @xcite ch .
6 , @xmath117 where @xmath118 mev , and @xmath119 is the isovector landau parameter in the quasiparticle interaction amplitude .
the zero sound velocity @xmath120 for the isovector mode can be found from eqs .
( [ 2.5 ] ) and ( [ 2.6 ] ) in the rare collision limit @xmath121 taking into account eq .
( [ 2.10 ] ) , one obtains , see also eq .
( [ 3.cc ] ) , @xmath122 where @xmath123 we point out that , in contrast to the isoscalar mode , the fermi surface distortion effect leads to a relatively small increase of the isovector zero sound velocity @xmath93 eq .
( [ 3.11 ] ) , with respect to the first sound one @xmath124 eq . ( [ 3.10 ] ) .
the dispersion relation ( [ 3.1 ] ) takes the form @xmath125 where @xmath126 all relations ( [ 3.qq])-([3.tau ] ) are still correct in the isovector case if both asymptotic velocity @xmath120 and @xmath7 are taken from eqs .
( [ 3.10 ] ) and ( [ 3.11 ] ) . in fig .
2 we have plotted the coefficients @xmath95 and @xmath13 as obtained from eq .
( [ 3.nkapa ] ) for the isovector mode with the collision parameter @xmath127 mev @xcite .
we point out that the transition temperature @xmath112 of zero- to first- sound regimes for the isovector mode is significantly smaller than @xmath128 for the isoscalar one . in an asymmetric nuclear matter , both the isovector and the isoscalar modes are dependent on each other .
the particle density fluctuation @xmath129 takes a bispinor form @xmath130 @xmath131 , where @xmath132 and @xmath133 are the isoscalar and isovector components respectively .
a solution of the corresponding equations of motion ( [ 2.5 ] ) and ( [ 2.6 ] ) leads to the following dispersion relation , see also eqs .
( [ 3.1 ] ) and ( [ 3.12 ] ) , @xmath134 here , @xmath135 is the asymmetry parameter , @xmath136 and @xmath137 are the neutron and proton densities respectively , @xmath138 and the coupling constants @xmath139 and @xmath140 are given by @xmath141 as is seen from eq .
( [ 3.15 ] ) , the eigenfrequency @xmath10 and the corresponding sound velocity for both the isoscalar and the isovector modes are independent of each other in the linear order of the asymmetry parameter @xmath142 the structure of bispinor @xmath130 @xmath143 is different for both the isoscalar - like and the isovector - like modes . for the isoscalar - like mode with the eigenfrequency @xmath144 given by a solution to eq . ( [ 3.1 ] ) , the main contribution to the bispinor @xmath129 is due to the isoscalar component @xmath145 while the isovector component @xmath133 is proportional to the asymmetry parameter@xmath146 namely , @xmath147 the opposite situation takes place for the isovector - like mode with the eigenfrequency @xmath148 given by a solution to eq .
( [ 3.12 ] ) . in this case
, one has @xmath149 and @xmath150
let us consider now the bulk instability regime @xmath151 and introduce an instability growth rate @xmath152 ( @xmath18 is real , @xmath153 ) , see refs .
the amplitude of the density fluctuations , @xmath154 grows exponentially if @xmath153 . to prevent an unphysical infinite growth of the short wave length fluctuations ( see dotted line in fig .
3 ) , we will take into account the velocity dependent contribution to the effective interparticle interaction .
due to the corresponding change in the selfconsistent mean field @xmath60 in eq .
( [ 2.5 ] ) , an additional anomalous term @xmath155 appears in the dispersion relation ( [ 3.1 ] ) and the equation for the instability growth rate @xmath18 takes the following form @xcite @xmath156 here @xmath157 and @xmath158 where @xmath159 is the relaxation time @xmath160 @xcite .
the constant @xmath161 in the anomalous dispersion term in eq .
( [ 4.2 ] ) depends on the model . in the case of the effective skyrme forces
, one has @xcite
@xmath162 where @xmath163 and @xmath164 are the parameters of the velocity dependent part of the skyrme forces @xcite .
we point out that the instability regime with @xmath151 can be reached at a low bulk density @xmath165 the incompressibility @xmath5 is given by @xmath166 here , the landau parameters @xmath16 are related to the parameters @xmath167 of the effective skyrme forces .
namely , @xmath168 \,{\frac{m^{\ast } } { m}}+3\,\left ( 1-{\frac{m^{\ast } } { m}}\right ) , \,\,\,f_{1}=3\,\left ( { \frac{m^{\ast } } { m}}-1\right ) , \label{4.4}\ ] ] where @xmath169 for the commonly used set of parameters @xmath167 , the instability regime with @xmath170 is reached at @xmath171 in fig .
3 we have plotted the instability growth rate @xmath18 as obtained from eq .
( [ 4.1 ] ) .
the calculation was performed for the skyrme force siii . for the relaxation time
@xmath159 we used @xmath172 @xmath173mev and the bulk density @xmath174 was @xmath175 , where @xmath52 is the saturated density @xmath176 @xmath177 .
we also show in fig .
3 the result for the nonviscous nuclear matter neglecting the anomalous dispersion term @xmath155 and the fermi surface distortion effect ( dotted line ) . the non - monotonous behavior of the instability growth rate as a function of the wave number
@xmath78 is due to the anomalous dispersion term in eq .
( [ 4.1 ] ) , induced by the velocity dependent terms in the interparticle interaction .
the instability growth rate @xmath18 reaches a maximum , @xmath178 at a certain @xmath179 and @xmath18 goes to zero at @xmath180 the existence of the critical wave number @xmath181 for an unstable mode is a feature of the system with the anomalous dispersion term @xcite .
the distortion of the fermi surface leads to a decrease of the critical value @xmath182 , i.e. , the nuclear matter becomes more stable due to the fsd effect .
we can also see that the presence of viscosity and the fsd effect lead to a shift of the position @xmath183 of the maximum of @xmath184 to the left .
thus , the instability of the nuclear matter with respect to short - wave - length density fluctuations decreases due to the viscosity and the fsd effect and the most unstable mode is shifted to the region of the creation of larger clusters in the disintegration of nuclear matter . for a saturated nuclear liquid one
has for the force parameters @xmath185 and @xmath186 .
however both values @xmath187 and @xmath182 have a non - monotonous behavior as a function of the bulk density @xmath174 because of the additional @xmath188dependence of the fermi energy @xmath189 in eqs . (
[ 4.3 ] ) and ( [ 4.4 ] ) .
the particle density dependence of the values of @xmath183 and @xmath190 is shown in fig .
the instability growth rate @xmath184as well as the values of @xmath183 and @xmath182 are only slightly sensitive to the change of temperature at @xmath191 mev , where the temperature dependence of the bulk density @xmath174 can be neglected .
a more sophisticated consideration is necessary near the critical temperature @xmath192 mev where the nuclear matter is unstable with respect to the liquid - gas phase transition .
starting from the collisional kinetic equation ( [ 2.1 ] ) , we have derived the dispersion relations ( [ 3.14 ] ) and ( [ 4.1 ] ) for both the stable and the unstable regime of the density fluctuations in a heated nuclear matter .
the dispersion relations are influenced strongly by the fsd effect and the anomalous dispersion term .
the presence of the fermi surface distortion enhances the stiffness coefficient for a stable mode and reduces the instability growth rate for an unstable one .
there is a significant difference between the static nuclear incompressibility coefficient , @xmath5 , i.e. , derived as a stiffness coefficient with respect to a change in the bulk density , and the dynamic one , @xmath71 associated with the zero sound velocity , see eq .
( [ 2.13 ] ) .
the fsd effect is responsible for the collisional relaxation of the collective modes in the fermi liquid and for the non - markovian character of the nuclear matter viscosity ( memory effect in the viscosity @xmath69 , eq .
( [ 2.14 ] ) ) .
the memory effects in the viscosity play an essential role in the description of the temperature dependence of the refraction coefficient @xmath193 see eqs .
( [ 3.nkapa ] ) and ( [ 3.tau ] ) .
we have noted also the bell - shaped form of the attenuation coefficient @xmath13 as a function of the temperature @xmath194 see figs . 1 and 2 .
this peculiarity of @xmath195 provides a new criterion for the determination of the transition temperature @xmath112 between the zero - sound and first - sound regimes in hot nuclear matter .
our consideration provides a good basis for understanding the difference in the development of the spinodal instability in nuclear matter taking into account both the fsd and the viscosity effect .
we have shown that the instability growth rate @xmath184 in an unstable nuclear matter with velocity dependent effective interparticle interaction is a non - monotonous function of the wave number @xmath78 because of the anomalous dispersion term .
the anomalous dispersion term removes an unphysical infinite growth of the short wave length fluctuations .
the non - monotonous behavior of the instability growth rate @xmath184 is characterized by two wave numbers @xmath196 and @xmath182 .
we point out that both the fsd effect and the relaxation processes lead to a shift of @xmath197 to the longer wave length region , providing an increase of the relative yield of heavier clusters .
the main results were obtained assuming a quadrupole distortion of the fermi surface .
we point out , however , that the key expression ( [ 2.14 ] ) for the viscosity coefficient can be rewritten identically in the following form @xmath198 where @xmath7 and @xmath120 are the first and zero sound velocity respectively .
this expression can also be established from a general consideration , see ref .
@xcite ch.8 , and can be used for arbitrary multipolarities of the fermi surface distortion .
this work was supported in part by the us department of energy under grant # fg03 - 93er40773 .
we are grateful for this financial support .
one of us ( v.m.k . )
thank the cyclotron institute at texas a&m university for the kind hospitality . | the propagation of the isoscalar and isovector sound modes in a hot nuclear matter is considered .
the approach is based on the collisional kinetic theory and takes into account the temperature and memory effects .
it is shown that the sound velocity and the attenuation coefficient are significantly influenced by the fermi surface distortion ( fsd ) .
the corresponding influence is much stronger for the isoscalar mode than for the isovector one .
the memory effects cause a non - monotonous behavior of the attenuation coefficient as a function of the relaxation time leading to a zero - to - first sound transition with increasing temperature .
the mixing of both the isoscalar and the isovector sound modes in an asymmetric nuclear matter is evaluated .
the condition for the bulk instability and the instability growth rate in the presence of the memory effects is studied .
it is shown that both the fsd and the relaxation processes lead to a shift of the maximum of the instability growth rate to the longer wave length region . |
the theory of circle patterns is a rich fascinating area having its origin in classical theory of circle packings .
its fast development in recent years is caused by mutual influence and interplay of ideas and concepts from discrete geometry , complex analysis and the theory of integrable systems .
the progress in this area was initiated by thurston s idea @xcite,@xcite about approximating the riemann mapping by circle packings .
classical circle packings comprised of disjoint open disks were later generalized to circle patterns where the disks may overlap ( see for example @xcite ) .
different underlying combinatorics were considered .
schramm introduced a class of circle patterns with the combinatorics of the square grid @xcite ; hexagonal circle patterns were studied in @xcite and @xcite . the striking analogy between circle patterns and the classical analytic function theory is underlined by such facts as the uniformization theorem concerning circle packing realizations of cell complexes of a prescribed combinatorics @xcite , discrete maximum principle , schwarz s lemma @xcite and rigidity properties @xcite,@xcite , discrete dirichlet principle @xcite .
the convergence of discrete conformal maps represented by circle packings was proven by rodin and sullivan @xcite . for a prescribed regular combinatorics
this result was refined .
he and schramm @xcite showed that for hexagonal packings the convergence is @xmath2 the uniform convergence for circle patterns with the combinatorics of the square grid and orthogonal neighboring circles was established by schramm @xcite .
approximation issue naturally leads to the question about analogs to standard holomorphic functions .
computer experiments give evidence for their existence @xcite,@xcite however not very much is known . for circle packings with the hexagonal combinatorics the only
explicitly described examples are doyle spirals @xcite,@xcite which are discrete analogs of exponential maps and conformally symmetric packings , which are analogs of a quotient of airy functions @xcite . for patterns with overlapping circles
more explicit examples are known : discrete versions of @xmath3 , @xmath4 @xcite , @xmath5 , @xmath6 @xcite are constructed for patterns with underlying combinatorics of the square grid ; @xmath5 , @xmath1 are also described for hexagonal patterns @xcite , @xcite .
it turned out that an effective approach to the description of circle patterns with overlapping circles is given by the theory of integrable systems ( see @xcite,@xcite,@xcite ) .
for example , schramm s circle patterns are governed by a difference equation which is the stationary hirota equation @xcite ( see @xcite for a survey ) .
this approach proved to be especially useful for the construction of discrete @xmath5 and @xmath7 in @xcite,@xcite,@xcite with the aid of some isomonodromy problem .
another connection with the theory of discrete integrable equations was revealed in @xcite,@xcite : embedded circle patterns are described by special solutions of discrete painlev ii equations , thus giving geometrical interpretation thereof .
this research was motivated by the attempt to carry the results of @xcite,@xcite over square grid circle patterns with prescribed intersection angles giving schramm patterns as a special case .
namely , we prove that such circle patterns mimicking @xmath5 and @xmath1 are embedded .
this turned out to be not straightforward and lead to asymptotical analysis of solutions to _ discrete riccati equation_. as the riccati differential equation is known as the only equation of the first order possessing panlev property we are tempted to conclude that circle patterns are described by discrete equations with painlev property though there is no satisfactory generalization thereof on discrete equations .
we use the following definition for square grid circle patterns , which is slightly modified version of one from @xcite .
let @xmath8 be a subgraph of the 1-skeleton of the cell complex with vertices @xmath9 and @xmath10 .
square grid circle pattern for @xmath8 with intersection angles @xmath11 is an indexed collection of circles on the complex plane @xmath12 that satisfy : + 1 ) if @xmath13 then the intersection angle of @xmath14 and @xmath15 is @xmath11 , + 2 ) if @xmath16 then the intersection angle of @xmath14 and @xmath17 is @xmath18 , + 3 ) if @xmath19 ( or @xmath20 ) then the disks , defined by @xmath14 and @xmath21 ( @xmath14 and @xmath22 respectively ) are tangent and disjoint , + 4 ) if @xmath23 , @xmath24 , @xmath25 ( i.e. @xmath26,@xmath27 are tangent and @xmath14 intersects @xmath26 and @xmath27 ) and @xmath28 ( i.e. @xmath29 is one step counterclockwise from @xmath30 ) , then the circular order of the triplet of points @xmath31,@xmath32,@xmath33 agrees with the orientation of @xmath14 .
the intersection angle is the angle at the corner of disc intersection domain ( fig .
[ angle ] ) . to visualize the analogy between schramm s circle patterns and conformal maps , consider regular patterns composed of unit circles and suppose that the radii are being deformed to preserve the intersection angles of neighboring circles and the tangency of half - neighboring ones .
discrete maps taking the intersection points and the centers of the unit circles of the standard regular patterns to the respective points of the deformed patterns mimic classical holomorphic functions , the deformed radii being analog of @xmath34 ( see fig . [ cpmap ] ) .
( 20,50 ) ( 2,22)(1,0)15 ( 7,27 ) _ f _ in section [ cp ] we give definition of discrete @xmath35 as a solution to some integrable equation subjected to a non - autonomous constraint .
its geometrical properties of immersion and embeddedness are expressed in terms of solution for radii of corresponding circle patterns .
for this solution to be positive it is necessary that some discrete riccati equation has positive solution .
this equation is studied in section [ re ] where its general solution is expressed via the hypergeometric function .
section [ ipe ] completes the proof of embeddednes , discrete equations of painlev type being the main tool .
possible generalizations for non - regular combinatorics and for non - circular patterns are discussed in section [ cr ] .
[ skew - schramm - def ] discrete map @xmath35 , @xmath36 is the solution @xmath37 of @xmath38 @xmath39 with @xmath40 and the initial conditions @xmath41 where @xmath42 stands for cross - ratio of elementary quadrilaterals : @xmath43 constraint ( [ sg - c ] ) was obtained from some isomonodromy problem of lax representation of ( [ q ] ) , found for @xmath44 in @xcite ( see also @xcite,@xcite ) .
the definition of @xmath35 can be justified by the following properties : * if one thinks of @xmath45 as defined on the vertices of the cell complex with diamond - shaped faces ( see fig.[cpmap ] ) then ( [ q ] ) means that @xmath45 respects the cross - ratios of the faces and therefore is `` locally conformal '' , * asymptotics of ( [ sg - c ] ) as @xmath46 suggests that @xmath45 approximates @xmath5
. * remark .
* equation ( [ q ] ) with @xmath47 was used in @xcite to define _ discrete conformal maps_. the motivation was that @xmath45 maps vertices of squares into vertices of the `` conformal squares '' . consider the surface glued of these conformal squares along the corresponding edges .
this surface is locally flat but can have cone - like singularities in vertices .
if the map is an immersion then the corresponding surface does not have such singularities .
therefore it is more consistent to define as discrete conformal an _ immersion _
map on the vertices of cell decomposition of @xmath48 which preserves cross - ratios of its faces .
[ compatible]@xcite constraint ( [ sg - c ] ) is compatible with ( [ q ] ) .
compatibility is understood as a solvability of some cauchy problem .
in particular a solution to ( [ q]),([sg - c ] ) in the subset @xmath49 is uniquely determined by its values @xmath50 . indeed ,
constraint ( [ sg - c ] ) gives @xmath51 and defines @xmath45 along the coordinate axis @xmath52 as a second - order difference equation .
then all other @xmath53 with @xmath54 are calculated via cross - ratios ( [ q ] ) . in this paper
the following initial conditions for ( [ q]),([sg - c ] ) are considered : @xmath55 with real @xmath56 .
the solution @xmath45 defines a circle pattern with square grid combinatorics . for @xmath57
such circle patterns were introduced by schramm in @xcite . for any @xmath58
we obtain a natural generalization of schramm circle patterns . in what follows
we say that the triangle @xmath59 has _ positive ( negative ) orientation _
if @xmath60 [ geom - kite ] let @xmath61 , @xmath58 . * if @xmath62 and the triangle @xmath63 has positive orientation then @xmath64 and the angle between @xmath65 $ ] and @xmath66 $ ] is @xmath67 . * if @xmath62 and the triangle @xmath63 has negative orientation then @xmath64 and the angle between @xmath65 $ ] and @xmath66 $ ] is @xmath11 .
* if the angle between @xmath65 $ ] and @xmath68 $ ] is @xmath11 and the triangle @xmath63 has positive orientation then @xmath69 and @xmath70 .
* if the angle between @xmath65 $ ] and @xmath68 $ ] is @xmath71 and the triangle @xmath63 has negative orientation then @xmath69 and @xmath70 . _
proof : _ straightforward .
[ kite ] all the elementary quadrilaterals @xmath72 for the solution of ( [ q]),([sg - c ] ) with initial ( [ skew - shramm - initial ] ) are of kite form : all edges at the vertex @xmath73 with @xmath74 are of the same length .
moreover , each elementary quadrilateral has one of the forms enumerated in lemma [ geom - kite ] . _
proof : _ given initial @xmath75 and @xmath76 constraint ( [ sg - c ] ) gives @xmath77 and @xmath78 for all @xmath79 it is easy to check that @xmath45 has the following equidistant property : @xmath80 for any @xmath81 , @xmath82 .
lemma [ geom - kite ] and @xmath83 allows one to apply induction in @xmath84 , starting with @xmath85 .
proposition [ kite ] implies that for @xmath86 the points @xmath87,@xmath88 lie on the circle with the center at @xmath73 . for
the most @xmath89 ( namely for @xmath90 ) the behavior of thus obtained circle pattern is quite irregular : inner parts of different elementary quadrilaterals intersect .
a discrete map @xmath73 is called an immersion if inner parts of adjacent elementary quadrilaterals @xmath91 are disjoint .
consider the sublattice @xmath92 and denote by @xmath93 its quadrant @xmath94 where @xmath95 we use complex labels @xmath96 for this sublattice .
denote by @xmath97 the circle of the radius @xmath98 with the center at @xmath99 let @xmath100 be a square grid circle pattern on the complex plane .
define @xmath101 as follows : + a ) if @xmath86 then @xmath73 is the center of @xmath102 + b ) if @xmath103 then @xmath104 since all elementary quadrilaterals @xmath72 are of kite form equation ( [ q ] ) is satisfied automatically . in
what follows the function @xmath105 defined as above by a ) and b ) is called _ a discrete map corresponding to the circle pattern @xmath106 .
_ [ eqforr ] let the solution @xmath45 of ( [ q]),([sg - c ] ) with initial ( [ skew - shramm - initial ] ) is an immersion , then @xmath107 defined by ( [ rmap ] ) satisfies the following equations : @xmath108 for @xmath109 and @xmath110 @xmath111 for @xmath112 + conversely let @xmath113 satisfy ( [ square ] ) for @xmath114 and ( [ ri ] ) for @xmath115 then @xmath107 define a square grid circle patterns with intersection angles @xmath11 , the corresponding discrete map @xmath73 is an immersion and satisfies ( [ q]),([sg - c ] ) .
_ proof : _ circle pattern is immersed if and only if all triangles @xmath116 of elementary quadrilaterals of the map @xmath73 have the same orientation ( for brevity we call it orientation of quadrilaterals ) .
suppose that the quadrilateral @xmath117 has positive orientation .
let the circle pattern @xmath73 be an immersion
. for @xmath118 points @xmath119 lie on circle with the center at @xmath120 and radius @xmath121 , where @xmath122 ( see the left part of fig .
[ reqsq ] ) . using ( [ q ] )
one can compute @xmath123 and @xmath124 .
lemma [ geom - kite ] and proposition [ kite ] imply that @xmath125 is in line with @xmath126 and that the points @xmath127 are also collinear .
denote by @xmath128 the radii of the circle at @xmath124 and @xmath123 respectively .
define @xmath129 if ( [ sg - c ] ) is satisfied at @xmath130 then ( [ sg - c ] ) at @xmath131 is equivalent to ( [ square ] ) , @xmath132 being positive iff the quadrilaterals @xmath91 and @xmath133 have positive orientation . similarly starting with ( [ sg - c ] ) at @xmath134 , where @xmath135 ( see the right part of fig.[reqsq ] ) one can determine evolution of the cross - like figure formed by @xmath124 , @xmath136 , @xmath73 , @xmath137 , @xmath138 into @xmath125 , @xmath139 , @xmath140 , @xmath73 , @xmath136 .
equation ( [ sg - c ] ) at @xmath141 is equivalent to ( [ square ] ) and ( [ ri ] ) at @xmath142 .
@xmath143 is positive only for immersed circle pattern .
now let @xmath121 be some positive solution to ( [ square]),([ri ] ) .
we can rescale it so that @xmath144 .
this solution is completely defined by @xmath145 .
consider solution @xmath73 of ( [ q]),([sg - c ] ) with initial data ( [ skew - shramm - initial ] ) where @xmath56 is chosen so that the quadrilateral @xmath117 has positive orientation and satisfies the conditions @xmath146 and @xmath147 .
the map @xmath73 defines circle pattern due to proposition [ kite ] .
it can be uniquely computed from ( [ square]),([ri ] ) . to this end one
have to resolve ( [ square ] ) with respect to @xmath148 and use it to find @xmath125 from @xmath149 and to resolve ( [ ri ] ) for @xmath150 to find @xmath140 from @xmath151 .
one can reverse the argument used in derivation of ( [ square]),([ri ] ) to show that @xmath45 satisfies ( [ q]),([sg - c ] ) .
moreover , since @xmath152 is positive , at each step we get positively orientated quadrilaterals .
note that initial data ( [ skew - shramm - initial ] ) for @xmath73 imply initial data for @xmath121 : @xmath153 a discrete map @xmath73 is called embedded if inner parts of different elementary quadrilaterals @xmath91 do not intersect . [ convex ] if for a solution @xmath121 of ( [ square]),([ri ] ) with @xmath154 and initial conditions ( [ rinitial ] ) holds true : @xmath155 in @xmath156 , then the corresponding discrete map is embedded .
_ proof : _ since @xmath157 the corresponding discrete map is an immersion .
consider piecewise linear curve @xmath158 formed by segments @xmath159 $ ] , where @xmath160 , @xmath161 and the vector @xmath162 along this curve .
due to proposition [ kite ] this vector rotates only in vertices with @xmath86 as @xmath163 increases along the curve .
the sign of the rotation angle @xmath164 , where @xmath165 , @xmath166 is defined by the sign of expression @xmath167 ( note that there is no rotation if this expression vanishes ) , where @xmath142 is a label for the circle with the center in @xmath73 .
if @xmath168 define @xmath169 . now the theorem hypothesis and equation ( [ ri ] ) imply that the vector @xmath170 rotates with increasing @xmath163 in the same direction for all @xmath171 , and namely , clockwise for @xmath172 and counterclockwise for @xmath173 . due to the following lemma it is sufficient to prove ( [ sign ] ) only for @xmath174 .
[ dual ] if @xmath121 is a solution of ( [ square]),([ri ] ) for @xmath175 then @xmath176 is a solution of ( [ square],[ri ] ) for @xmath177 .
[ dualr ] consider the sector @xmath178 .
the terminal points of the curves @xmath158 lie on the sector border .
[ rot ] the curve @xmath158 has no self - intersection and lies in the sector @xmath179 .
lemma is proved by induction .
for @xmath180 it is obviously true since the curve @xmath181 is a line segment .
let it be true for @xmath182 .
note that for immersed @xmath45 the first and the last segments of @xmath183 lie in @xmath179 therefore the vector @xmath184 rotates counterclockwise around convex domain of @xmath179 confined by @xmath158 and can not make a full circle .
lemma is proved .
each curve @xmath158 cuts the sector @xmath179 into a finite part and an infinite part .
since the curve @xmath158 is convex and the borders of all elementary quadrilaterals @xmath91 for imbedded @xmath185 have the positive orientation the segments of the curve @xmath183 lie in the infinite part .
now the induction in @xmath171 completes the proof of proposition [ convex ] for @xmath186 .
the proof for @xmath187 is similar . q.e.d .
one can compute @xmath188 from ( [ sg - c ] ) and obtain the asymptotics @xmath189 from stirling formula for large @xmath190 : @xmath191 numerical experiments support the following conjecture .
* conjecture .
* @xmath192 .
let @xmath193 and @xmath194 be radii of the circles with the centers at @xmath195 , @xmath196 respectively ( see fig.[ricc ] ) . constraint ( [ sg - c ] ) and property ( [ equidistant ] ) gives @xmath197 from elementary geometric considerations one gets @xmath198 define @xmath199 and denote @xmath200 for brevity .
now the equation for radii @xmath201 takes the form : @xmath202 * remark .
* equation ( [ riccati ] ) is a discrete version of _ riccati _ equation .
this title is motivated by the following properties : * cross - ratio of each four - tuple of its solutions is constant as @xmath203 is mbius transform of @xmath204 , * general solution is expressed in terms of solution of some linear equation ( see below this linearisation ) . below we call ( [ riccati ] )
d - riccati equation .
[ r - positive ] solution of discrete riccati equation ( [ riccati ] ) is positive for @xmath205 iff @xmath206 proof is based on the closed form of the general solution of d - riccati linearisation .
it is linearised by the standard ansatz @xmath207 which transforms it into @xmath208 one can guess that there is only one initial value @xmath209 giving positive d - riccati solution from the following consideration : @xmath210 as @xmath211 , and the general solution of ( [ linear ] ) with limit values of coefficients is @xmath212 .
so @xmath213 for @xmath214 .
however @xmath215 defines only asymptotics of a solution . to relate it to initial values one needs some kind of _ connection formulas_.
fortunately it is possible to find the general solution to ( [ linear ] ) .
[ solution ] the general solution to ( [ linear ] ) is @xmath216 @xmath217 where @xmath218 and @xmath219 stands for the hypergeometric function . _
proof : _ solution was found by slightly modified _ symbolic method _ ( see @xcite for method description ) .
substitution @xmath220 transforms ( [ linear ] ) into @xmath221 we are looking for solution in the form @xmath222 where @xmath223 satisfies @xmath224 * remark . *
note that the label @xmath163 in ( [ sum ] ) is running by step 1 but is not necessary integer therefore @xmath223 is a straightforward generalization of @xmath225 playing the role of @xmath226 in the calculus of finite differences .
general solution to ( [ eq - for - v ] ) is expressed in terms of @xmath227-function : @xmath228 stirling formula ( [ stirling ] ) for large @xmath190 gives the asymptotics for @xmath229 @xmath230 substituting ( [ sum ] ) into ( [ x - linear ] ) , making use of ( [ eq - for - v ] ) and collecting similar terms one gets the following equation for coefficients : @xmath231 choice @xmath232 or @xmath233 kills the term with @xmath234 . to make series ( [ sum ] ) convergent we can use the freedom in @xmath163 to truncate ( [ sum ] ) on one side .
the choice @xmath235 or @xmath236 leads to divergent series . for @xmath237 equation ( [ recurrent ] ) gives @xmath238 for all non - negative integer @xmath239 and @xmath240 where @xmath241 .
substitution of solution of this recurrent relation in terms of @xmath242-functions and ( [ s - for - v ] ) yields @xmath243 [ convergent ] for both @xmath244 series ( [ sum-1 ] ) converges for all @xmath190 .
_ proof of lemma [ convergent ] _ : since @xmath245 for @xmath246 respectively and @xmath247 the convergence of ( [ sum-1 ] ) depends on the behavior of @xmath248 .
stirling formula ( [ stirling ] ) ensures that this expression is bounded by @xmath249 for some @xmath250 an @xmath251 which gives convergence .
series ( [ sum-1 ] ) is expressed in terms of hypergeometric functions : @xmath252 where @xmath253 @xmath254\left[(\frac{\gamma-1}{2})(1+\frac{\gamma-1}{2 } ) ...
( k-1\frac{\gamma-1}{2})\right]}{(1-(x\frac{\gamma-1}{2})) ... (k-(x+\frac{\gamma-1}{2}))}+ ... \ ] ] here we use the standard designation @xmath255 for hypergeometric function as a holomorphic at @xmath256 solution for equation @xmath257f_z - a b f=0.\ ] ] now we can complete the proof of proposition [ solution ] . due to linearity general solution of ( [ linear ] )
is given by superposition of any two linear independent solutions . as was shown each summond in ( [ series ] ) satisfies the equation ( [ linear ] ) .
to finish the proof of proposition [ solution ] one has to show that the particular solutions with @xmath258 and @xmath259 are linearly independent , which follows from lemma [ as - lin ] .
[ as - lin ] as @xmath260 function ( [ series ] ) has the asymtotics @xmath261 _ proof : _ for @xmath211 series representation ( [ hyper ] ) gives @xmath262 .
stirling formula ( [ stirling ] ) defines asymptotics of the factor @xmath263 ( see ( [ asymptotics ] ) ) .q.e.d .
_ proof of proposition [ r - positive ] : _ for positive @xmath204 it is necessary that @xmath264 : it follows from asymptotics ( [ eq - as - lin ] ) substituted into ( [ ansatz ] ) .
let us define @xmath265 it is the hypergeometric function @xmath266 with @xmath267 . a straightforward manipulation with series shows that @xmath268 where @xmath269 .
note that @xmath209 as a function of @xmath270 satisfies some ordinary differential equation of first order since @xmath271 satisfies riccati equation obtained by reduction of ( [ gauss - eq ] ) .
computation shows that @xmath272 satisfies the same ode .
since both expression ( [ p-0 ] ) and ( [ r - initial ] ) are equal to 1 for @xmath256 they coincide everywhere .
[ unique ] if there exists immersed @xmath73 satisfying ( [ q]),([sg - c]),([skew - shramm - initial ] ) it is defined by initial data ( [ right - sg - initial ] ) .
let @xmath121 be a solution of ( [ square ] ) and ( [ ri ] ) with initial condition ( [ rinitial ] ) . for @xmath273 define @xmath274 , @xmath275 .
then ( [ ri ] ) and ( [ square ] ) are rewritten as follows @xmath276},\ ] ] @xmath277 where @xmath278 property ( [ sign ] ) for ( [ qpainleve]),([ppainleve ] ) reads as @xmath279 equations ( [ qpainleve]),([ppainleve ] ) can be considered as a dynamical system for variable @xmath280 , where due to ( [ diag - r ] ) @xmath281 [ q - exists ] for each @xmath282 there exists @xmath283 such that the solution of system ( [ qpainleve]),([ppainleve ] ) subjected to ( [ initial - p ] ) and @xmath284 has property ( [ pqsign ] ) for all @xmath285 .
_ proof : _ due to lemma [ dual ] it is sufficient to prove ( [ pqsign ] ) only for @xmath286 .
define real function @xmath287 on @xmath288 implicitly by @xmath289 for @xmath290 and by @xmath291 for @xmath292 designate @xmath293 @xmath294 as in fig.[domain ] . system ( [ qpainleve]),([ppainleve ] ) defines the map @xmath295 . this map has the following properties : *
it is a continuous map on @xmath296 .
values of @xmath297 on the border of @xmath296 are defined by continuity in @xmath298 . * for @xmath299 holds true @xmath300 , i.e. @xmath301 can not jump in one step from @xmath302 into @xmath303 .
consider the solution to ( [ qpainleve]),([ppainleve ] ) with initial conditions on the segment @xmath304 determined by ( [ initial - p ] ) and @xmath305 , where @xmath306 then @xmath307 .
define @xmath308 .
then @xmath309 is a closed set as @xmath310 is continuous on @xmath302 and @xmath301 can not jump in one step from @xmath302 on the half - line @xmath311 . as a closed subset of a segment @xmath304
the set @xmath309 is a collection of disjoint segments @xmath312 .
[ segments ] there exists sequence @xmath313 such that : * @xmath314 is mapped by @xmath315 onto some curve @xmath316 with one terminal point on the curve @xmath317 and the other on the line @xmath318 . *
@xmath319 .
lemma is proved by induction .
for @xmath320 it is trivial .
let it holds true for @xmath171 .
@xmath310 maps @xmath321 with @xmath317 into @xmath322 and @xmath323 with @xmath324 into @xmath325 .
thus @xmath310 maps @xmath326 into some curve @xmath327 lying in @xmath328 with one terminal point in @xmath322 and with the other in @xmath325 .
therefore at least one of the connected components of @xmath329 has its terminal points on the border of @xmath302 as stated by lemma .
as the segments of @xmath330 constructed in lemma [ segments ] are nonempty there exits @xmath331 for all @xmath332 . for this @xmath333
holds true : @xmath334 .
if @xmath335 then @xmath336 would jump out of @xmath302 .
[ sg - imbed ] the discrete map @xmath35 , @xmath36 is embedded . _
proof : _ proposition [ q - exists ] ensures that for any @xmath337 there exist @xmath338 such that the solution to ( [ square]),([ri ] ) with these initial values is positive for @xmath96 , @xmath339
. then equation ( [ square ] ) defines recursively @xmath340 and implies that @xmath341 far all @xmath342
. therefore asymptotics ( [ eq - as - lin ] ) implies that @xmath343 is exactly as defined by initial condition ( [ rinitial ] ) for @xmath35 .
proposition [ convex ] completes the proof .
* remark 1 . * for @xmath344 system ( [ qpainleve]),([ppainleve ] ) for @xmath345 reduces to the special case of discrete painlev equation (
the case @xmath346 was studied in @xcite ) : @xmath347 where @xmath348 .
this equation allows to represent @xmath349 as a function of @xmath350 and @xmath351 : @xmath352 .
@xmath353 maps the torus @xmath354 into @xmath355 and has the following properties : * @xmath356 it is a continuous map on @xmath357 where @xmath358\}$ ] .
* for @xmath359 holds true @xmath360 , where @xmath361\}$ ] and @xmath362 , i.e. @xmath190 can not jump in one step from @xmath363 into @xmath364 .
these properties guarantees that there exists the unitary solution @xmath365 of this equation with @xmath366 in the sector @xmath367 .
this solution corresponds to embedded @xmath35 .
equation ( [ dpii ] ) is a special case of more general reduction of cross - ratio equation ( see @xcite,@xcite for the detail ) .
for @xmath370 formula ( [ diag - r ] ) gives infinite @xmath371 .
the way around this difficulty is re - normalization @xmath372 and limit procedure @xmath373 , which leads to the re - normalization of initial data ( see @xcite ) .
as follows from ( [ r - initial ] ) this re - normalization gives : @xmath374 circle pattern @xmath375 has radius function specified by ( [ square]),([ri ] ) with initial data ( [ 2-initial ] ) .
the symmetry @xmath376 of ( [ square]),([ri ] ) stated in lemma [ dual ] is the _ duality transformation _ ( see @xcite ) .
smooth analog @xmath377 for holomorphic functions @xmath378 is : @xmath379 note that @xmath380 .
@xcite circle pattern @xmath381 is a circle pattern dual to @xmath368 .
discrete maps corresponding to circle patterns @xmath368 and @xmath382 are embedded . _
proof : _ the circle radii for @xmath368 and @xmath383 are subject to ( [ square]),([ri ] ) with @xmath384 and @xmath385 respectively . for these values of @xmath175 proposition [ convex ] is true : the proof is the same since @xmath368 and @xmath383 are immersed . due to lemma [ dual ] it suffices to prove the property ( [ sign ] ) only for @xmath368 .
consider the discrete conformal map @xmath386 with @xmath387 the corresponding solution @xmath121 of ( [ square]),([ri ] ) is a continuous function of @xmath388 .
so there is a limit as @xmath373 , of this solution with the property ( [ sign ] ) , which is violated only for @xmath256 since @xmath389 . + q.e.d .
further generalizations of discrete @xmath185 and @xmath390 are possible .
* i. * one can relax the unitary condition for cross - ratios and consider solutions to @xmath391 subjected to the same constraint ( [ sg - c ] ) with the initial data @xmath392 this solution is a discrete analog of @xmath35 defined on the vertices of regular parallelogram lattice ( see fig .
[ recz ] ) .
however , thus obtained mappings are deprived of geometrical flavor as they do not define circle patterns .
another possibility is to de - regularize prescribed combinatorics by projection of @xmath393 on a plane as follows ( see @xcite ) .
consider @xmath394 .
for each coordinate vector @xmath395 where @xmath396 define unit vector @xmath397 in @xmath398 so that for any pair of indexes @xmath399 vectors @xmath400 form a basis in @xmath401 .
let @xmath402 be some 2-dimensional connected simply connected cell complex with vertices in @xmath403 .
suppose @xmath404 .
( we denote by the same symbol @xmath405 the set of the complex vertices . )
define the map @xmath406 by the following conditions : \1 ) @xmath407 , \2 ) if @xmath408 are vertices of @xmath405 and @xmath409 then @xmath410 .
it is easy to see that @xmath411 is correctly defined and unique .
+ we call @xmath405 a _ projectable _ cell complex iff its image @xmath412 is embedded , i.e. intersections of images of different cells of @xmath405 do not have inner parts .
using projectable cell complexes one can obtain not only regular square grid and hexagonal combinatorics but more complex ones , i.e. combinatorics of _ penrose tilings_. it is natural to define discrete conformal map on @xmath413 as a discrete complex immersion function @xmath45 on vertices of @xmath413 preserving cross - ratios of @xmath413-cells .
the argument of @xmath45 can be labeled by the vertices @xmath190 of @xmath405 .
hence for any cell of @xmath405 , constructed on @xmath414 the function @xmath45 satisfies the following equation for cross - ratios : @xmath415 where @xmath416 is the angle between @xmath417 and @xmath418 , taken positive if @xmath419 has positive orientation and taken negative otherwise .
now suppose that @xmath45 is a solution to ( [ g - q ] ) defined on the whole @xmath403 . * conjecture 1 .
* _ equation ( [ g - q ] ) is compatible with the constraint _
@xmath420 for @xmath421
this conjecture was proven in @xcite .
now we can define discrete @xmath422 for projectable @xmath405 as solution to ( [ g - q]),([g - c ] ) restricted on @xmath405 .
initial conditions for this solution are of the form ( [ right - sg - initial ] ) so that the restrictions of @xmath45 on each two - dimensional coordinate lattices is an immersion defining circle pattern with prescribed intersection angles .
this definition naturally generalize definition of discrete @xmath35 given in @xcite for @xmath423 .
* conjecture 2 . * _ discrete @xmath424 is an immersion .
_ schramm @xcite showed that there is square grid circle patterns mimicking @xmath425 but an analog of @xmath426 does not exist .
the obstacle is purely combinatorial .
there is a hope that combinatorics of projectable cells can give more examples of discrete analogs of classical functions .
the author thanks a.bobenko , r.halburd and a.its for useful discussions .
this research was partially supported by the epsrc grant no gr / n30941 .
a.i.bobenko , t.hoffman , yu.b.suris , hexagonal circle patterns and integrable systems : patterns with the multi - ratio property and lax equations on the regular triangular lattice , _ internat .
notices , _ * 3 * ( 2002 ) , 111 - 164 .
nijhoff , discrete painlev equations and symmetry reduction on the lattice . in : _ discrete integrable geometry and physics ;
a.i.bobenko and r.seiler , pp .
209 - 234 , oxford university press , 1999 . | square grid circle patterns with prescribed intersection angles , mimicking holomorphic maps @xmath0 and @xmath1 are studied .
it is shown that the corresponding circle patterns are embedded and described by special separatrix solutions of discrete painlev and riccati equations .
general solution of this riccati equation is expressed in terms of the hypergeometric function .
global properties of these solutions , as well as of the discrete @xmath0 and @xmath1 , are established . |
frequency division multiplexing ( ofdm ) has been widely adopted in wireless communications due to its high - speed data transmission and inherent robustness against the inter - symbol interference ( isi ) . however , traditional rectangularly pulsed ofdm exhibits large spectral sidelobes , resulting in severe out - of - band power leakage . in this case , traditional ofdm will introduce high interference to the adjacent channels . for alleviating this problem ,
various methods have been proposed for sidelobe suppression in ofdm @xcite-@xcite .
more specifically , the windowing technique @xcite extends the guard interval at the cost of spectral efficiency reduction .
cancellation carriers @xcite will consume extra power and incur a signal - to - noise ratio ( snr ) loss .
precoding methods @xcite-@xcite require complicate decoding processing to eliminate the interference . against this background ,
nc - ofdm @xcite is a class of efficient sidelobe suppression techniques by making the ofdm signal and its first _
n _ derivatives continuous .
however , a disadvantage of nc - ofdm is the high implementation complexity in the transmitter . for reducing the complexity ,
some simplified schemes have been proposed in @xcite and @xcite . in @xcite ,
time - domain nc - ofdm ( td - nc - ofdm ) was proposed for reducing the complexity of nc - ofdm , by transforming traditional frequency - domain processing into the time domain . on the other hand
, traditional nc - ofdm will introduce severe interference so as to degrade the bit error rate ( ber ) performance . for alleviating this problem , _
n_-continuous symbol padding ofdm ( ncsp - ofdm ) was recently developed in @xcite at the cost of increased complexity in the transmitter .
meanwhile , to enable low - complexity signal recovery in nc - ofdm , several techniques @xcite-@xcite have been proposed .
however , those techniques will also result in complexity load or efficiency reduction . for reducing the interference of nc - ofdm and avoiding complex signal recovery at the receiver ,
this letter proposes a low - interference nc - ofdm scheme by adding an improved smooth signal in the time domain .
firstly , in the proposed scheme , the smooth signal is generated in a novel way as the linear combination of designed basis signals related to rectangular pulse .
secondly , the basis signals are smoothly truncated by a preset window function to obtain the short - duration smooth signal with shorter length than a cyclic - prefixed ofdm symbol and low interference .
thirdly , the short - duration smooth signal is overlapped onto only part of the ofdm symbol to reduce the interference .
furthermore , we give an asymptotic expression of the spectrum feature of the low - interference nc - ofdm signal , to show that the proposed scheme can maintain the similar sidelobe suppression to traditional nc - ofdm .
lastly , the complexity analysis denotes that the proposed scheme can significantly reduce the complexity of nc - ofdm .
in conventional nc - ofdm @xcite , the _ _ i__th transmit symbol @xmath0 follows @xmath1 and @xmath2 where @xmath3 is the precoded symbol on the _ _ r__th subcarrier with the subcarrier index set @xmath4 , @xmath5 is the _
_ n__th - order derivative of @xmath6 with @xmath7 , the frequency spacing is @xmath8 , @xmath9 is the symbol duration , and @xmath10 is the cyclic prefix ( cp ) duration . for satisfying eq .
, the _ n_-continuous processing can be summarized as @xcite @xmath11 where @xmath12^t$ ] , @xmath13^t$ ] , @xmath14 is the identity matrix , @xmath15 , @xmath16 , @xmath17 with @xmath18 , and @xmath19
td - nc - ofdm is recently proposed in @xcite for transforming the processing of nc - ofdm into the time domain . in td - nc - ofdm
, the smooth signal is added with the time sampling interval @xmath20 where _ m _ is the length of the symbol . following the basic idea of td - nc - ofdm
, the proposed low - interference scheme also adds a smooth signal @xmath21 onto the ofdm signal @xmath22 in the time domain , given as @xmath23 where @xmath24 and @xmath25 is the length of a cp .
thus , to make the ofdm signal _
n_-continuous , @xmath21 should satisfy @xmath26 to construct @xmath21 with low interference , @xmath21 is just added in the front part of each cp - prefixed ofdm symbol .
thus , the proposed linear - combination design of @xmath21 is described as @xmath27 where @xmath28 indicates the location of @xmath21 with length _
l _ , and the basis signals @xmath29 belong to the basis set @xmath30 , defined as @xmath31^t , \tilde{n}\in \mathcal{u}_{2n}\bigg\ } , \label{eqn7 } \end{aligned}\ ] ] where @xmath32 . for achieving the low - interference smooth signal @xmath21 in eq . , the design of the basis signals @xmath33 and the linear combination coefficients @xmath34^t$ ] will be specified as follows . on the one hand ,
the time duration of @xmath33 is truncated by a preset window function @xmath35 , which is considered as a smooth and zero - edged window function , such as triangular , hanning , or blackman window function .
then , the truncated basis signals can be given by @xmath36 where @xmath37 denotes the unit - step function , and @xmath38 is calculated by the rectangularly ofdm pulse @xcite , given as @xmath39 on the other hand , by substituting eqs .
- into eq . , the coefficients @xmath40 can be calculated as @xmath41 where @xmath42 is a @xmath43 symmetric matrix , given as @xmath44 @xmath45 and @xmath46 . finally , @xmath21 is added onto only part of the cp - inserted ofdm symbol to achieve the _
n_-continuous symbol @xmath47 , as @xmath48 & 0\leq i\leq m_{\rm s}-1 , \\
\mathbf{q}_{\tilde{f}}{\mathbf{b}}_i & i = m_{\rm s } , \end{matrix}\right .
\label{eqn11}\ ] ] where @xmath49 , @xmath50 is the number of the ofdm symbols , and @xmath51 is initialized as @xmath52 since the back edge of @xmath53 equals to zero .
in general , the proposed low - interference nc - ofdm scheme is illustrated as follows .
initialization : @xmath52 , @xmath54 , @xmath55 , @xmath56 , @xmath57 , @xmath58 ; + according to the above processing , the main advantage of the improved smooth signal @xmath21 is that its length has been effectively truncated to _
l _ , by a careful design in eq . .
furthermore , for @xmath59 , only parts of the ofdm signal are overlapped with the interference @xmath21 .
more especially , since the front part of the ofdm symbol is cp , the interference to the real data is further mitigated . in section
v , we will show that the influence of @xmath21 has been effectively reduced .
assume that the first _
n_-1 derivatives of the smoothed ofdm signal are continuous , and the _ _ n__th - order derivative has finite amplitude discontinuity @xcite . thus , according to the definition of the power spectrum density ( psd ) @xcite and the relationship between spectral roll - off and continuity in @xcite , the psd of the low - interference nc - ofdm signal is expressed by @xmath60 where @xmath61 .
indicates that the spectrum of the low - interference nc - ofdm signal is related to the expectation of @xmath62 multiplied by @xmath63 . in this letter ,
the conventional blackman window function is used as an example , given as @xmath64 where @xmath65 . by substituting eqs . and into eq .
, the psd of the smoothed ofdm signal is expressed by @xmath66 where @xmath67 , @xmath68 , @xmath69 is the binomial coefficient , @xmath70 @xmath71 with @xmath72 , and @xmath73 .
shows that the power spectral roll - off of the smoothed signal , whose first _ n_-1 derivatives are continuous , decays with @xmath74 .
moreover , the expression of @xmath75 reveals that the sidelobe is affected by the length of @xmath76 , so that the selection of _ l _ is important to balance ber and sidelobe suppression performance .
[ fig2 ] compares the theoretical and simulation results of the low - interference scheme with _ _
l__=@xmath25=144 , _ _
m__=2048 , where the length of @xmath76 is equal to that of cp . considering
the ratio of cp is only 7% of the whole symbol , the assumption is reasonable for practical wireless systems such as lte @xcite .
it is shown in fig .
[ fig2 ] that the simulation results match well with the theoretical analyses with varying highest derivative order _ n_. in section v , with the above parameters , we will show the proposed scheme can maintain the similar spectral roll - off to traditional nc - ofdm . .
the complexity comparison , quantified by the numbers of real additions and multiplications , among nc - ofdm , ncsp - ofdm , and the low - interference scheme is shown in table i. compared to other methods , the proposed low - interference scheme has notable complexity reduction .
l _ is equal to or smaller than the length of cp , the complexity is significantly reduced in the transmitter .
for example , when _ _
l__=@xmath25=144 , _ _
k__=256 , _ _
n__=2 , and _ _ m__=2048 , at the transmitter side , the complexity of the proposed low - interference scheme is , respectively , 47.4% and 20.3% of those of nc - ofdm and ncsp - ofdm .
8.85cm|m1.4cm|m2.69cm|m3.45cm| * scheme & * number of real multiplications & * number of real additions + * nc - ofdm & @xmath77 & @xmath78 + * ncsp - ofdm & @xmath79 & @xmath80 + * low - interference scheme & @xmath81 & @xmath82 + * * * * * *
simulations are performed in a baseband - equivalent ofdm system with _ _
k__=256 , @xmath83 , and 16-qam digital modulation .
the carrier frequency is 2ghz , the subcarrier spacing is @xmath84khz , and the time - domain oversampling factor is 8 .
the psd is evaluated by welch s averaged periodogram method @xcite with a 2048-sample hanning window and 512-sample overlap . in order to show the system performance in the multi - path fading environment , lte extended vehicular a ( eva ) channel with 9 paths of rayleigh fading channels @xcite is considered . fig .
[ fig3 ] compares the psds between nc - ofdm and the proposed low - interference scheme with different _
n_. the low - interference scheme can obtain the sidelobe suppression performance similar to traditional nc - ofdm .
moreover , as _ l _ increases , the sidelobe suppression performance of the proposed scheme is improved .
for example , _
l _ is increased from 36 to 72 and 144 . in general , we show that _ _
l__=@xmath25=144 is a good choice for maintaining the sidelobe suppression performance as traditional nc - ofdm . .
[ fig4 ] shows the ber performance of nc - ofdm , ncsp - ofdm , and the low - interference scheme when _ _ n__=3 and _ _ l__=@xmath25=144 in 3gpp lte eva fading channel .
it is shown that traditional nc - ofdm will introduce severe interference to the transmit signal and hence to influence the ber performance , while ncsp - ofdm and our proposed low - interference scheme can both effectively mitigate the interference so as to achieve similar ber performance to ofdm , and to save the complexity for signal recovery at the receiver .
however , the proposed low - interference scheme is with much lower complexity compared to other schemes according to table i. in general , we show that the proposed scheme can make a promising tradeoff among ber performance , sidelobe suppression performance and computational complexity . .
in this letter , a low - interference nc - ofdm was proposed to reduce the interference and complexity as opposed to the original nc - ofdm .
the main idea is to generate the time - domain low - interference smooth signal in a novel way as the linear combination of carefully designed rectangularly pulsed basis signals .
both analyses and simulation results showed that the low - interference scheme was capable of reducing the interference to a negligible extent , while maintaining similar sidelobe suppression to traditional nc - ofdm but with much lower complexity .
t. weiss , j. hillenbrand , a. krohn , and f. jondral , mutual interference in ofdm - based spectrum pooling systems , " in _ proc .
( vtc ) _ , milan , italy , may 2004 , pp .
1873 - 1877 .
s. brandes , i. cosovic , and m. schnell , reduction of out - of - band radiation in ofdm systems by insertion of cancellation carriers , " _ ieee commun .
10 , no . 6 , pp .
420 - 422 , jun . 2006 .
m. ma , x. huang , b. jiao , and y. j. guo , optimal orthogonal precoding for power leakage suppression in dft - based systems , " _ ieee trans .
844 - 853 , mar .
2011 . c. d. chung , spectrally precoded ofdm , " _ ieee trans .
2173 - 2185 , dec . 2006 .
j. van de beek and f. berggren , _ n_-continuous ofdm , " _ ieee commun .
_ , vol . 13 , no
. 1 , pp . 1 - 3 ,
jan . 2009 .
j. van de beek and f. berggren , evm - constrained ofdm precoding for reduction of out - of - band emission , " in _ proc .
( vtc ) _ , anchorage , usa , sep . 2009 .
j. van de beek , sculpting the multicarrier spectrum : a novel projection precoder , " _ ieee commun .
881 - 883 , dec . 2009 .
m. ohta , a. iwase , and k. yamashita , improvement of the error characteristics of an _ n_-continuous ofdm system with low data channels by slm , " in _ proc .
( icc ) _ , kyoto , japan , jun .
2011 , pp . 1 - 5
. m. ohta , m. okuno , and k. yamashita ,
receiver iteration reduction of an _
n_-continuous ofdm system with cancellation tones , " in _ proc .
ieee global telecommun .
( globecom ) _ , kathmandu , nepal , dec .
2011 , pp . 1 - 5 . h. kawasaki , m. ohta , and k. yamashita , _ n_-continuous symbol padding ofdm for sidelobe suppression , " in _ proc .
( icc ) _ , sydney , australia , jun .
2014 , pp .
5890 - 5895 .
p. wei , l. dan , y. xiao , and s. li , a low - complexity time - domain signal processing algorithm for _ n_-continuous ofdm , " in _ proc .
conf . commun .
( icc ) _ , budapest , hungary , jun .
2013 , pp .
5754 - 5758 .
r. bracewell , _ the fourier transform and its applications _ , 2nd ed .
new york : mcgraw - hill , 1978 , ch .
143 - 146 .
_ user equipment ( ue ) radio transmission and reception ( release 12 ) _ , 3gpp ts 36.101 , v12.3.0 , mar . 2014 . [ online ] .
available : http : //www.3gpp.org/. n. c. beaulieu and m. o. damen , parametric construction of nyquist - i pulses , " _ ieee trans .
2134 - 2142 , dec . 2004 .
p. d. welch , the use of fast fourier transform for the estimation of power spectra : a method based on time averaging over short , modified periodograms , " _ ieee trans .
audio electroacoustics _ , vol .
70 - 73 , jun . | _ n_-continuous orthogonal frequency division multiplexing ( nc - ofdm ) was demonstrated to provide significant sidelobe suppression for baseband ofdm signals .
however , it will introduce severe interference to the transmit signals .
hence in this letter , we specifically design a class of low - interference nc - ofdm schemes for alleviating the introduced interference .
meanwhile , we also obtain an asymptotic spectrum analysis by a closed - form expression .
it is shown that the proposed scheme is capable of reducing the interference to a negligible level , and hence to save the high complexity of signal recovery at the receiver , while maintaining similar sidelobe suppression performance compared to traditional nc - ofdm . _
n_-continuous orthogonal frequency division multiplexing ( nc - ofdm ) ; sidelobe suppression ; time - domain _ n_-continuous ofdm ( td - nc - ofdm ) . |
in the past few years we have seen the fruition of a great astronomical endevour : the measurement of the rate of cosmological evolution via direct censuses of star formation tracers , metal content , and metal production rate ( lilly et al .
1995 ; pei & fall 1995 ; lilly et al.1996 ; madau et al . 1996 ; connolly et al . 1997 ; madau , pozzetti , & dickinson 1998 ) .
this achievement has come about thanks to several key developments including ( 1 ) deep sub - arcsecond imaging capabilities , especially in the optical with the _ hubble space telescope _ ( williams et al .
1996 ) , ( 2 ) multi - object spectroscopy , especially down to @xmath13 abmag with the 10 m keck telescopes ( e.g. steidel et al.1996a , b ; lowenthal et al .
1997 ) , and ( 3 ) the refinement of photometric redshift techniques ( e.g. connolly et al . , 1997 ) especially the lyman - limit technique ( e.g. steidel & hamilton 1992 ; steidel , pettini , & hamilton 1995 ) which allow redshift estimates to be made to greater depth , and out to @xmath14 . in a pioneering paper madau et al .
( 1996 ; hereafter m96 ) derived the @xmath15 evolution of the metal production rate ( in comoving coordinates ) , which is directly proportional to the volume density of the star formation rate ( sfr ) .
the `` madau plot '' provides a direct constraint on the evolution of galaxies , hence it s importance .
it shows a strong ( factor @xmath16 10 ) decline in from its peak at @xmath17 to the present epoch , as pointed out earlier by lilly et al .
this decline mirrors that seen in the luminosity density of quasars ( boyle & terlevich 1997 ) . on the other side of the peak , at @xmath18 ,
is apparently only equal to that of the present epoch .
subsequent work by connolly et al .
( 1997 ) fills in the @xmath19 gap and is consistent with m96 .
the overall shape of the curve is consistent with models of the observed evolution in the properties of damped lyman-@xmath20 systems ( pei & fall , 1995 ) .
this form for is consistent with hierarchical models of galaxy formation ( e.g. searle & zinn , 1978 ; zepf & ashman , 1993 ; baugh et al .
1998 ) .
one concern about the madau plot is that it is heavily biased to observations in the rest frame uv .
the problem is that dust efficiently scatters and absorbs uv radiation , greatly complicating the interpretation of the detected uv emission .
not only is the amount of dust important , but so is the geometry ( where it is relative to the stars ) , and the composition of the dust .
for example , in the local group the uv `` extinction law '' varies from galaxy to galaxy and within galaxies , ( e.g. fitzpatrick 1986 ) .
the term `` extinction law '' was originally defined for stars .
it quantifies the wavelength dependent combined effect of dust absorption and scattering out of the line of sight toward point sources .
the geometry is quite different in distant galaxies where individual stars can not be isolated , and instead galaxies in their entirety , or large portions of them are observed . in such cases
much of the scattered light may be recovered ( by scattering into the line of sight ) , and it matters a great deal how porous the dust distribution is and whether the stars are mixed with the dust or not ( e.g. witt , thronson , & capuano 1992 ; calzetti , kinney & storchi - bergmann 1994 ) .
there are a variety of terms for the wavelength dependent net dimming of the light including `` effective extinction law '' ( calzetti et al .
1994 ) and `` obscuration law '' ( calzetti 1997 ) . here
we adopt the term `` absorption law '' to emphasize that it is dust absorbing photons that causes the dimming .
absorption laws can be derived empirically ( e.g. calzetti et al .
1994 ; calzetti 1997 ) and hence average together all the above geometric factors into a simple wavelength dependent correction .
observations of high redshift galaxies in the sub - mm and mm bands prove that molecular gas and dust formation have occurred as far back as @xmath21 ( omont et al .
1996 ; ivison et al . 1998 ; guilloteau et al .
1997 ; frayer et al .
especially interesting are the recent spate of blank field sub - mm detection with scuba , which we discuss in more detail in [ ss : rc ] and [ ss : scuba ] . these can account for a significant fraction of the sub - mm background , and are mostly attributed to dusty star forming galaxies having @xmath22 . following from uv observations of local starbursts ( meurer et al .
1995 , hereafter paper i ) we have argued that the correction factor to is considerable , a factor @xmath23 at @xmath1 ( meurer et al .
, 1997 ; hereafter paper ii ) , while others ( e.g. madau 1997 ; pettini et al.1997 ) prefer fairly modest , @xmath16 factor of three , corrections .
the amount of high-@xmath15 absorption has a direct bearing on the interpretation of the madau plot ; large corrections at high-@xmath15 can push back the peak @xmath24 , as might be expected by monolithic collapse dominated galaxy formation ( e.g. eggen , lynden - bell , & sandage 1962 ; bower , lucey & ellis 1992 ; ortolani et al .
1995 ) . here
we reexamine the uv luminosity density at high-@xmath15 , with particular emphasis on the @xmath0-dropouts in the hdf .
our technique has two strengths .
firstly , it is based on the similarity between local starburst galaxies and lyman limit systems as summarized in [ s : comp ] .
the main assumption we make is that the high redshift galaxies have the same multi - wavelength spectral properties as local starbursts .
we use local starburst galaxies to calibrate the relationship between uv reddening and uv absorption ( in [ s : cal ] ) thus largely avoiding the uncertainties in population model fitting and the choice of effective uv absorption laws .
secondly , our calibration of the uv absorption law recovers the intrinsic ( i.e. dust - free ) uv emission that is absorbed by dust and reradiated in the far infrared .
hence , up to the modest amounts of absorption considered here ( a few magnitudes ) , uv observations are capable of recovering the total bolometric luminosity of starbursts .
we apply our calibrations to the @xmath0-dropouts in the hdf in [ s : appl ] . in [ s : test ] we demonstrate the validity of our starburst assumption by showing that new constraints on the seds of these high-@xmath15 systems are consistent with the seds of local uv selected starbursts .
implications of our results are discussed in [ s : disc ] , while [
s : sum ] summarizes the paper . throughout this paper
we adopt @xmath25 , @xmath9 .
progress in understanding the lyman - dropout galaxies would be accelerated if we could find analogs to these galaxies in the local universe .
fortunately , there is mounting evidence that the lyman dropout galaxies indeed resemble starburst galaxies in many respects .
while the magnitude of the extinction corrections to the rest - frame uv light ( the subject of the present paper ) is still a matter of debate , it is nevertheless clear that local starbursts have surface - brightnesses that are similar to the distant galaxies ( paper ii ) .
that is , the high - z galaxies appear to be ` scaled - up ' versions of the local starbursts with the same sfrs per unit area and comparable stellar surface - mass - densities . in this regard , starbursts are a far better match to the lyman - dropout galaxies than are ordinary late - type galaxies , which have sfrs per unit area that are several orders - of - magnitude smaller ( cf .
paper ii ; kennicutt 1998 ) .
the overall rest - frame uv - through - optical spectral properties of local starbursts and the lyman dropout galaxies are similar in terms of the spectral energy distribution ( cf .
sawicki & yee 1998 ) and the dominance of the uv and optical spectra by absorption and emission lines respectively ( cf . leitherer 1997 ; pettini et al .
the lyman limit systems in the hdf are clearly spatially resolved ( giavalisco , steidel , & macchetto 1996 ) , demonstrating that their rest frame uv light is not dominated by an agn .
in addition the spectra of the lyman - dropout galaxies is more similar to starburst galaxies than agn . indeed
the typical strengths of the uv stellar and interstellar absorption - line features are comparable in starbursts and lyman - dropout galaxies , indicative of broadly similar stellar and interstellar content ( e.g. conti , leitherer , & vacca 1996 ; lowenthal et al .
1997 ) . the local starbursts and
lyman dropout galaxies also have similar distributions of observed uv colors , ( paper ii , and here in
[ s : appl ] ) , - which we argue chiefly measures the amount of dust - reddening .
even the dynamical state of the interstellar medium in the lyman dropout galaxies resembles that in local starbursts . both are driving extensive outflows of gas at velocities of several hundred km s@xmath26 .
the signature of this is the systematic blueshift of the broad uv interstellar absorption - lines with respect to the rest - frame optical ( e.g. h@xmath2 ) and/or rest - frame uv ( ly@xmath20 ) nebular emission - lines ( pettini et al .
1998 ; franx et al . 1997
; kunth et al . 1998 ; gonzalez - delgado et al .
the velocity dispersion of the emission - line gas is also similar in the two galaxy classes ( pettini et al .
1998 ) . to reiterate , in all their properties measured to - date , the lyman dropout galaxies appear to be simply scaled - up versions of local starbursts : larger and more luminous , but otherwise remarkably similar .
the challenge is to determine the intrinsic luminosities from the uv fluxes and colors of dusty lyman limit systems .
the above similarities justifies our approach of calibrating the effects of dust on local starbursts and applying it to the high-@xmath15 galaxies .
figure [ f : uvir ] illustrates the heart of the technique .
it shows the relationship between the ratio of far infrared ( fir ) and ultraviolet ( uv ) fluxes ( here at 1600 ) and the uv spectral slope @xmath2 ( equivalent to a color ) for a sample of starburst galaxies .
since the fir flux is due to dust radiatively heated by the absorbed uv radiation , the @xmath27 axis is a measurement of dust absorption .
this figure then shows that dust absorption is correlated with uv reddening .
such a relationship is expected for dust predominantly located in a foreground screen ( witt et al .
1992 ; paper i ) , although this screen need not be homogeneous ( calzetti et al . 1994
; calzetti 1997 ) . because this correlation links @xmath28 to uv quantities , then regardless of the exact dust geometry , this figure provides a powerful _ empirical _
tool for recovering the radiation reprocessed by dust and thus determining the total absorption - corrected uv flux of starbursts , using uv quantities alone . in the next sub - section
we define our local calibrating samples and the quatities we use . in order to apply this tool and interpret the results we perform three calibrations
. 1 . is calibrated in terms of absorption at 1600 and this is fitted as a function of @xmath2 .
2.the spectroscopic index @xmath2 is calibrated in terms of broad - band colors ( with a @xmath15 correction ) , since photometric colors are easier to measure than spectroscopic ones .
the relationship between luminosity measured at 1600 is related to sfr .
these calibrations are detailed in the final three sub - sections .
the local sample used to derive the various calibrations in this paper is listed in table [ t : loc ] .
it includes the sources shown in fig .
[ f : uvir ] , and is drawn from the _ international ultraviolet explorer _ ( iue ) atlas of kinney et al .
( 1993 ) , and contains galaxies having an `` activity class '' consistent with being a starburst , i.e. sb nuc .
( starburst nucleus ) , sb ring ( starburst ring ) , bcdg ( blue compact dwarf galaxy ) , or bcg ( blue compact galaxy ) .
although the uv sources in these galaxies tend to be centrally concentrated we find that galaxies with optical diameters @xmath29 ( @xmath30 is measured at @xmath31 mag arcsec@xmath32 ) tend to fall above the relationship shown in fig .
[ f : uvir ] .
this is probably because significant uv emission extends beyond the @xmath33 iue aperture .
hence these large galaxies were excluded from the sample . further limiting the sample to galaxies having @xmath34
removes a few more outliers , but not just galaxies above the relationship .
furthermore , it also severely depletes the points with @xmath35 .
applying fits to data limited in this way changes our final @xmath36 results by @xmath37% , hence we retain @xmath38 as the diameter limit for the local sample .
the lack of systematic residuals for galaxies up to 12 times larger than the iue aperture indicates that the uv emission of these galaxies is very compact , e.g.in a circum - nuclear starburst .
l r r r r r c c l ngc 4861 & 0.00 & @xmath39 & @xmath40 & @xmath41 & & & & 3 i zw 18 & 0.02 & @xmath42 & @xmath43 & @xmath44 & & & & 1,3 ngc 1705 & 0.18 & @xmath45 & @xmath46 & @xmath47 & @xmath48 & 2.85 & 4.61 & 1,7mrk 153 & 0.00 & @xmath49 & @xmath50 & @xmath51 & & & & 1 tol1924416 & 0.30 & @xmath52 & @xmath53 & @xmath54 & @xmath55 & 2.91 & 4.87 & 1,3,7 ugc 9560 & 0.00 & @xmath56 & @xmath57 & @xmath58 & @xmath59 & 3.36 & 4.13 & 1,3,7 mrk 66 & 0.01 & @xmath60 & @xmath61 & @xmath62 & @xmath63 & 2.85 & 2.90 & 1,7 ngc 3991 & 0.02 & @xmath64 & @xmath65 & @xmath66 & & & & 1,3 ngc 3738 & 0.00 & @xmath67 & & & & & & 3 ugca 410 & 0.02 & @xmath68 & @xmath69 & @xmath70 & & & & 1 mrk 357 & 0.17 & @xmath71 & @xmath72 & @xmath73 & @xmath74 & 3.25 & & 1,7 ngc 3353 & 0.00 & @xmath75 & @xmath76 & @xmath77 & & & & 1,3 mrk 54 & 0.02 & @xmath78 & @xmath79 & @xmath80 & & & & 1 ngc 1140 & 0.11 & @xmath78 & @xmath81 & @xmath82 & @xmath83 & 3.18 & 2.66 & 1,7 mrk 36 & 0.00 & @xmath84 & @xmath85 & @xmath86 & & & & 1 mcg6 - 28 - 44 & 0.02 & @xmath87 & @xmath88 & @xmath80 & & & & 1 ngc 1510 & 0.00 & @xmath89 & @xmath90 & @xmath91 & @xmath92 & 3.11 & 4.95 & 1,3,7 mrk 19 & 0.14 & @xmath93 & @xmath90 & @xmath94 & & & & 1 ngc 4214 & 0.00 & @xmath93 & & & & & & 3 ngc 4670 & 0.04 & @xmath95 & @xmath96 & @xmath97 & & & & 1,3 ngc 1800 & 0.00 & @xmath95 & & & @xmath98 & 3.08 & 3.99 & 7 ugc 5720 & 0.00 & @xmath99 & @xmath100 & @xmath41 & & & & 1,3 ugc 5408 & 0.00 & @xmath101 & @xmath102 & @xmath103 & & & & 1 ngc 7673 & 0.14 & @xmath104 & & & @xmath105 & 4.46 & 2.13 & 3,7 ngc 3125 & 0.25 & @xmath106 & @xmath107 & @xmath108 & @xmath109 & 3.28 & 4.72 & 1,7 haro 15 & 0.08 & @xmath110 & @xmath88 & @xmath111 & @xmath112 & 2.85 & 2.32 & 1,7 ngc2537 & 0.15 & @xmath113 & @xmath114 & @xmath115 & & & & 1 ugc 3838 & 0.07 & @xmath116 & @xmath117 & @xmath118 & & & & 1 ngc 7793 & 0.00 & @xmath119 & & & @xmath120 & 4.60 & 0.55 & 3,7 ngc 5253 & 0.18 & @xmath121 & @xmath122 & @xmath123 & @xmath124 & 3.18 & 4.68 & 3,7 ngc 7250 & 0.62 & @xmath121 & @xmath125 & @xmath82 & @xmath74 & 3.14 & 3.48 & 1,7 mrk 542 & 0.07 & @xmath126 & & & @xmath127 & 4.00 & 0.79 & 7 ngc 7714 & 0.15 & @xmath128 & @xmath129 & @xmath130 & @xmath131 & 4.36 & 1.60 & 1,3,7 mrk 487 & 0.02 & @xmath132 & & & @xmath133 & 3.25 & 6.61 & 7 ngc 3049 & 0.04 & @xmath134 & @xmath135 & @xmath136 & @xmath137 & 4.00 & 0.23 & 1,7 ugc 6456 & 0.09 & @xmath138 & @xmath139 & @xmath140 & & & & 1 ngc 3310 & 0.00 & @xmath141 & @xmath142 & @xmath143 & & & & 1 ngc 5996 & 0.06 & @xmath144 & @xmath145 & @xmath108 & @xmath146 & 4.76 & 0.38 & 1,7 ngc 4385 & 0.03 & @xmath147 & @xmath148 & @xmath149 & @xmath150 & 5.48 & 0.79 & 1,7 mrk 499 & 0.00 & @xmath147 & & & @xmath151 & 4.65 & 3.57 & 7 ngc 5860 & 0.03 & @xmath152 & @xmath153 & @xmath154 & @xmath155 & 6.04 & 0.64 & 1,7 ic 1586 & 0.07 & @xmath152 & @xmath156 & @xmath157 & @xmath158 & 5.30 & 1.35 & 7 ngc 2782 & 0.00 & @xmath159 & @xmath160 & @xmath161 & & & & 1 eso383 - 44 & 0.13 & @xmath162 & @xmath156 & @xmath163 & & & & 1 ngc 5236 & 0.14 & @xmath162 & & & @xmath164 & 3.91 & 0.22 & 7 ngc 2415 & 0.16 & @xmath165 & @xmath148 & @xmath166 & & & & 1 mcg013033 & 0.05 & @xmath167 & @xmath168 & @xmath136 & & & & 1 ngc 1614 & 0.22 & @xmath169 & @xmath170 & @xmath171 & @xmath172 & 7.85 & 0.96 & 7 ngc 6217 & 0.14 & @xmath173 & @xmath174 & @xmath175 & @xmath59 & 5.08 & 0.39 & 1,7 ngc 6052 & 0.10 & @xmath176 & @xmath177 & @xmath178 & @xmath179 & 3.54 & 2.01 & 1,7 ngc 4500 & 0.00 & @xmath180 & @xmath181 & @xmath182 & & & & 1 ic 214 & 0.11 & @xmath183 & @xmath184 & @xmath185 & @xmath186 & 5.08 & 1.82 & 7 ngc 3504 & 0.02 & @xmath187 & @xmath188 & @xmath189 & & & & 1 ngc 4194 & 0.00 & @xmath190 & @xmath191 & @xmath192 & @xmath172 & 6.81 & 1.13 & 1,7 ngc 2798 & 0.01 & @xmath193 & @xmath194 & @xmath195 & & & & 1 ngc 3256 & 0.55 & @xmath196 & @xmath197 & @xmath198 & @xmath199 & 5.48 & 0.47 & 1,7 ngc 7552 & 0.00 & @xmath200 & @xmath201 & @xmath202 & @xmath203 & 6.11 & 0.1 & 1,3,7 the iue spectra in the kinney et al
. atlas were measured to determine the uv quantities required for fig .
[ f : uvir ] .
the ultraviolet flux at 1600 , @xmath204 is a generalized flux of the form @xmath205 , and @xmath206 is the flux density per wavelength interval .
it was measured with the _ iraf / stsdas _ package _ synphot _ employing a square passband with a central wavelength of 1600 and width of 350 .
this filter is meant to approximate the rest frame parameters of the standard wfpc2 filters f606w and f814w for objects with redshifts @xmath207 , and 4 , respectively ( i.e. @xmath0 and @xmath208 band dropouts ) .
the ultraviolet spectral slope @xmath2 is determined from a power law fit of the form @xmath209 to the uv continuum spectrum as defined by the ten ( rest wavelength ) continuum bands listed by calzetti et al .
these spectral fits were performed after first removing galactic extinction using the law of seaton ( 1979 ) and taking galactic extinction values @xmath210 from burstein & heiles ( 1982 , 1984 ) as listed by ned .
since the continuum spectrum is never exactly a pure power law , @xmath2 is subject to systematic uncertainties due to the placement of the continuum windows .
eighteen of the data points in fig [ f : uvir ] represent galaxies observed with one of iue s short wavelength ( sw ) cameras ( @xmath211 to 1975 ) only and not with either of the lw cameras ( @xmath212 to 3000 ) , hence they do not have data in calzetti s et al .
( 1994 ) tenth window ( @xmath213 ) .
their @xmath2 values were determined from the sw only measurements using the following relationship : @xmath214 this was determined from measuring @xmath2 of 16 high signal - to - noise iue spectra with no noticeable sw / lw break , both with and without the tenth window .
the uncertainty is the standard error on the mean . the only non - uv quantity in fig .
[ f : uvir ] is the far - infrared flux @xmath28 which is derived from _ infrared astronomical satellite _ ( iras ) 60 and 100 flux densities listed by ned and is defined by helou et al .
for convenience in notation we define the ratio @xmath215 typical error bars are shown in the bottom right of fig [ f : uvir ] .
we estimate an uncertainty in @xmath2 of 0.14 from the @xmath216 fits to the spectra , and the scatter about the mean relationship given in eq .
[ e : swoff ] ) .
we adopt an average uncertainty in @xmath204 of 5% , which we showed in paper i is the typical level of agreement between broad band hst uv fluxes and iue spectra .
this is higher than expected from the error spectra presented with the kinney et al .
( 1993 ) iue atlas , which are commonly at the 10% 20% level _ per pixel _ , and hence @xmath217% when integrated over the uv passband .
the level of agreement between hst and iue is worse probably because of the centering errors with iue .
the median uncertainty in @xmath28 for our sample is 7% , derived from the uncertainties in @xmath218 and @xmath219 as listed by ned .
combining the @xmath204 and @xmath28 errors we estimate that the error in @xmath220 is 0.04 dex .
these uncertainties represent random errors .
systematic errors will effectively move all data points with respect to the zeropoints of the axes .
c. robert ( 1999 , private communication ) estimates that there is a sytematic uncertainty of 0.15 in @xmath2 due to the placement of the continuum windows in the spectrum fit .
figure [ f : uvir ] illustrates that the random errors in the quantities are smaller than the scatter about the relationship .
hence , the scatter in fig .
[ f : uvir ] is largely intrinsic .
possible sources of intrinsic scatter are dust geometry , dust composition , and intrinsic spectral slope ( which depends on the star formation history ) , all of which may vary from galaxy to galaxy .
we assume that @xmath28 is due to the thermal emission of dust heated by the radiation it absorbs .
this consists primarily of non - ionizing photons with the addition of photons which are resonantly scattered by hydrogen until they are absorbed by dust grains .
we assume that ionizing photons do not significantly contribute to dust heating% of the bolometric luminosity of starbursts is emitted below 912 ( leitherer & heckman , 1995 ) .
the fraction of this that is directly absorbed by dust grains is estimated to be @xmath221% ( smith , biermann , & mezger 1978 ) . ] .
for these assumptions can be written as @xmath222 @xmath223 is the unattenuated flux density of the emitted spectrum , @xmath224 is the net absorption in magnitudes by dust as a function of wavelength , and @xmath225 is the flux . in our models @xmath225 is derived from the spectrum shortward of 912 using the standard assumption that each ionizing photon results in one recombination and consequent decay down to case - b population levels . customarily one adopts a form of the absorption ( or extinction ) law scaled by the optical reddening , e.g. @xmath226 . in eq .
[ e : irx1 ] the first term gives the bolometric flux of the absorbed radiation , normalized by @xmath204 , and the second term gives the fraction of the bolometric flux emitted by dust that is intercepted by the fir passband ( e.g. helou et al .
1988 ) .
strictly speaking in our model depends on the intrinsic spectrum and the absorption curve , even at wavelengths well outside the 1600 passband . however , for dust heating dominated by young stellar populations eq .
[ e : irx1 ] is well approximated by the formula @xmath227 here the first term gives the fraction of @xmath228 that is absorbed by dust ; the second term is similar to a bolometric correction , giving the maximum amount of heating available to the dust divided by the intrinsic 1600 flux ; and the third term is the same as the second term in eq . [
e : irx1 ] .
the ratio of eqs .
[ e : irx1 ] and [ e : irx2 ] asymptotes to unity as @xmath229 for all @xmath230 .
we have calculated this ratio under a variety of assumptions applicable to starbursts .
the constant star formation rate ( csfr ) solar metallicity models of leitherer and heckman ( 1995 ) having a salpeter ( 1955 ) imf slope and durations of @xmath231 to 100 myr were used as the initial spectra .
leitherer and heckman ( 1995 ) varied other parameters in their population models including metallicity , imf slope and cutoff , and other star formation histories ( i.e. instantaneous bursts ) .
they find that as long as the population is ionizing ( i.e.produces a strong ionizing flux , as expected for starbursts by their strong emission line spectra ) , these parameters have very little effect on the overall uv spectrum which has @xmath232 to @xmath233 in all cases .
this range in intrinsic @xmath2 is well covered by the models we examine here .
the absorption was parameterized by @xmath234 to 1 and a variety of forms of @xmath235 including the following extinction laws ( assuming extinction = absorption ) : galactic ( seaton , 1979 ) , lmc ( howarth , 1983 ) , and smc ( using a polynomial fit to the bouchet et al.1985 curve ) ; and the starburst absorption laws of kinney et al .
( 1994 ) , calzetti et al .
( 1994 ) , and calzetti et al .
( 1997 ) . in all cases
the ratio of equations [ e : irx1 ] and [ e : irx2 ] is within 30% of unity for @xmath236 mag .
we conclude that eq .
[ e : irx2 ] is a good approximation of when the dust is heated by intrinsically young stellar populations and as long as the form of @xmath235 is not extremely different from standard absorption / extinction laws .
equation [ e : irx2 ] can be rewritten as @xmath237 where b is the ratio of the two bolometric like corrections : @xmath238 @xmath239 @xmath240 here we assume that @xmath208 is constant , and estimate the best value for it .
we do this by cross comparing observations with models of the photometric properties of stellar populations ( @xmath241 ) and dust emission ( @xmath242 ) .
values of @xmath241 were calculated for the csfr population models of leitherer & heckman ( 1995 ) with standard salpeter ( 1955 ) imf slope , and an upper mass limit of 100 . for burst durations in the range of 0 to 100 myr , we find @xmath243 .
we also calculated @xmath241 directly from the observed sed of ngc 1705 , one of the least extincted sources in our local sample .
we started with the observed iue + optical sed published by storchi - bergmann et al .
( 1995 ) after correction for galactic extinction .
this was extended down to @xmath244 by extrapolation using the power law fit ( yielding @xmath245 ) .
the sed was extended redwards to @xmath246 m using the _ jhk _ photometry of jones as reported in meurer et al .
the total flux given by meurer et al .
( 1992 ) was converted to @xmath225 assuming case b recombination .
the integral in eq .
[ e : bcstar ] was evaluated directly from the sed , and @xmath228 was measured with synphot .
this yields @xmath247 .
note that this may be a bit overestimated ( by up to 20% ) since there appears to be a break between the sw and lw portions of the iue spectrum , with the sw portion being relatively underestimated .
we conclude that within the uncertainties the ngc 1705 measurement is consistent with the model estimates .
the bolometric dust correction @xmath248 was estimated empirically by directly calculating @xmath248 for the ultra - luminous infrared galaxies observed in the fir - submm by rigopoulou , lawrence , & rowan - robinson ( 1996 ) .
two galaxies ( mrk 231 and iras05189 - 2524 ) were rejected from their sample because they have strong agn emission .
a smooth sed was drawn through the sed observations of the remaining sources and integrated over @xmath249 mm to determine @xmath250 .
the individual @xmath251 measurements are shown in the top panel of fig .
[ f : bcfir ] , as a function of fir color @xmath252 . for comparison the fir colors of local uv selected starbursts are shown in the bottom panel of fig .
[ f : bcfir ] .
we find that six of the seven remaining galaxies in the rigopoulou et al . sample have @xmath253 with a dispersion of 0.13 .
the seventh , 08572 + 3915 has @xmath254 , much warmer than the majority of the uv selected starbursts , almost all of which have @xmath255 .
the top panel of fig .
[ f : bcfir ] shows that the empirical @xmath251 measurements are in good agreement with helou s et al.(1988 ) dust emission models .
these models are very simple .
they treat the dust distribution as emitting thermal radiation at a single temperature , with emissivity @xmath256 .
we caution that more realistic modeling of interstellar dust emission includes multiple temperature emission with @xmath257 to 2 ( e.g. dsert , boulanger , & puget 1990 ) . figure [ f : bcfir ] illustrates that single temperature models with @xmath257 provide a good approximation to the dust emission of starbursts for the limited purpose of calculating @xmath248 . adopting this model
we have calculated @xmath248 for the local uv selected starbursts , which tend to have somewhat cooler colors in the fir than ultraluminous infrared galaxies .
these have a median @xmath258 , with 80% of the sample having @xmath259 .
we adopt @xmath260 , and @xmath261 , yielding @xmath262 . here
the `` errors '' actually indicate the amount of leeway allowed by our model calculations and limited measurements of starbursts , as described above .
equation [ e : irxlin ] can now be written logarithmically as @xmath263 @xmath264 here we see the beauty of the approximation eq .
[ e : irx2 ] ; it allows the direct transformation of to 1600 absorption as shown on the right hand side of fig . [
f : uvir ] .
our adopted model for the - @xmath2 relationship is a simple linear fit of @xmath3 versus @xmath2 . to first order
this is what one would expect for absorption by a foreground screen of dust .
the adopted fit , which is shown in fig .
[ f : uvir ] , is @xmath265 it was determined from separate unweighted least squares fits of @xmath3 versus @xmath2 , and @xmath2 versus @xmath3 , which were then averaged .
the dispersion about this fit is 0.55 mag in @xmath3 , 0.28 in @xmath2 and 0.30 dex in @xmath266 ( the latter including only galaxies with @xmath267 ) .
consequently the standard error in the fit zeropoint is 0.08 mag in terms of @xmath3 and 0.04 in terms of @xmath2 .
since the scatter about the fit is significantly larger than the errors , it is largely intrinsic .
hence , a more sophisticated fit ( e.g. a full @xmath216 fit with weighting by errors ) will not improve the quality of the fit . instead , improvement can only occur with an increase in sample size .
this fit implies the spectral slope in the absence of dust absorption is @xmath268 .
this result is consistent with what is expected of naked ionizing populations , which leitherer & heckman ( 1995 ) show to have @xmath232 to 2.6 ( cf .
paper i ) . in table
[ t : alaws ] the slope of the absorption law , @xmath269 derived here is compared with other extinction and absorption laws .
the slope we derive is in the middle of the range for the starburst absorption laws derived in other studies . in particular
, it is between the values derived by calzetti et al.(1994 ) and calzetti ( 1997 ) .
note that the galactic extinction law has a very large slope because the 2175 absorption feature counteracts the reddening of @xmath2 to a large degree .
llcl & * absorption * & * 1.99 * & * this work * starburst & absorption & 1.44 & kinney et al .
( 1994 ) starburst & absorption & 1.97 & calzetti et al .
( 1994 ) starburst & absorption & 2.30 & calzetti ( 1997 ) galactic & extinction & 4.42 & seaton ( 1979 ) lmc/30 dor & extinction & 1.45 & howarth ( 1983 ) smc & extinction & 1.05 & bouchet et al .
( 1985 ) in paper ii we derived the following relationship between @xmath2 and @xmath270 color : @xmath271 where the color is in the `` ab '' system ( @xmath272 , where @xmath273 is the spectral flux density per frequency interval in units of @xmath274 ) .
this was derived from first principles using the central wavelengths of the f606w and f814w wfpc2 filters and assuming a featureless power law spectrum and negligible filter widths . however , the uv spectra of starbursts typically contain numerous absorption features of both stellar and interstellar origin ( kinney et al .
emission lines , even , are typically weak in starburst galaxies ( kinney et al . , 1993 ) . while the spectroscopically defined @xmath2 avoids the lines , broad band filters can not .
since the concentration of lines is particularly strong for @xmath275 one may expect an artificial reddening of the photometric @xmath2 at @xmath5 when these features are redshifted into the middle of the f606w passband .
in addition at @xmath276 , enters the f606w passband .
the lyman forest strongly suppresses the flux bluewards of in such high redshift objects , again reddening @xmath277 .
we have used iue starburst spectra to quantify these effects and model the calibration between photometric and spectroscopic @xmath2 .
the spectra we use were selected ( by eye ) to have high s / n and no discernible break between lw and sw portion of spectrum .
this resulted in a final sample of 15 galaxies , as indicated in table [ t : loc ] ; most of these were also used to calibrate the @xmath3 versus @xmath2 fit .
their spectroscopic @xmath2 were measured as described above .
these spectra were redshifted over the range @xmath278 . in order to adequately account for the f606w filter width
, the spectrum blueward of was taken to be a pure power law ( using the measured @xmath2 ) normalized to the mean flux density at @xmath279 , and attenuated by the lyman series forest which was modeled following the algorithm of madau ( 1995 ) .
the @xmath280 colors were then measured using synphot , yielding @xmath277 .
figure [ f : betaphot ] plots the difference between @xmath277 and the standard spectroscopic @xmath2 as a function of @xmath15 .
our calibration of the @xmath281 relationship is a quadratic polynomial fitted to the median points shown in fig .
[ f : betaphot ] : @xmath282 the fit is a @xmath216 fit where the errors were taken to be the mean difference between the median and the 16th and 84th percentiles for a given @xmath15 .
the fit has an rms dispersion of 0.06 .
the reduced @xmath216 is 0.19 , hence the model is adequate for the given amount of scatter . combining this with eq .
[ e : bphot ] yields : @xmath283 which is now on the same scale as the @xmath284 calibration of eq .
[ e : afit ] .
the transformation of uv luminosity to sfr is highly model dependent .
hence , for most of this paper the emphasis will be on luminosity rather than the sfr . however , for comparison to other studies and to get a feel for the physics behind the uv luminosity , it is often convenient to transform it to a sfr .
our transformation is based on the population models of leitherer & heckman ( 1995 ) .
we adopt a solar metallicity , continuous star formation history population of duration 100 myr , and a salpeter ( 1955 ) initial mass function ( imf ) with mass limits of 0.1 to 100 . for these parameters a sfr of 1 yr@xmath26 produces a 1600 flux density of @xmath285 ( @xmath286 ) , and hence an absolute magnitude of 18.15 abmag ( 20.81 stmag ) ; the equivalent bolometric flux corresponds to @xmath287 .
for the same star formation history and imf parameters , bruzual & charlot ( 1993 ) stellar population models produce 8% more flux at 1600 ( see also comparison in paper i ) . not only is the choice of the particular population model important in this calibration , so to are the parameters of the imf and the adopted star formation history .
for example , if the lower mass limit of the imf is raised to 1 then the luminosity increases by a factor of 2.55 for a given sfr ; if the duration of star formation is decreased to 10 myr ( 1 myr ) , then the luminosity decreases by a factor of 1.53 ( 12.9 ) .
the above calibrations were applied to a sample of f300w ( @xmath0 ) dropouts selected from the hdf v2.0 catalog ( williams et al .
we will compare the results from this sample to two other @xmath0-dropout samples extracted from the hdf by madau and collaborators - in m96 and madau et al .
( 1998 ; hereafter m98 ) .
m96 lays out all selection criteria in detail and lists all galaxies selected , and in addition it is the pioneering work in this field .
m98 improves on m96 in several ways ( see below ) .
however , the results of m98 are model dependent , their @xmath0-dropout selection criteria are not explicitly given , and their final sample is not listed . hence we adopt m96 as our primary comparison , and we model our sample selection on the criteria given therein .
our sample selection differs from that of m96 , primarily because we use @xmath270 rather than @xmath288 as a color selection criterion .
this is done because for @xmath0-dropouts , the f450w filter is much more strongly affected by forest absorption than f606w . in our model , selecting by @xmath289 places a fairly strong limit on @xmath2 and hence on @xmath3 .
figure [ f : two2cd ] shows the @xmath290 , @xmath291 two color diagram , and our selection box .
our selection criterion @xmath292 was chosen to encompass the strong @xmath0-dropout plume but avoid relatively low-@xmath15 interlopers .
our other selection criteria are virtually identical to m96 ; @xmath293 is the crucial @xmath0-dropout selection criterion , while @xmath294 abmag is the magnitude limit abmag in our selection .
however , for our other selection criteria no objects approach this limit . ] . only sources in the wf chips
are selected .
another important difference with m96 is that their color selection box is not rectangular .
they have an additional constraint @xmath295 , clipping the bottom right corner out of the box .
this was introduced to avoid regions that model calculation indicate may be populated by low redshift galaxies .
however , this corner , while relatively sparsely populated in the @xmath291 versus @xmath296 two color diagram , contains about the galaxies in the hdf with measured @xmath297 ( dickinson 1998 ) . since the clipped corner is red in @xmath296 the clipping biases against large @xmath3 .
the color selection criteria of m98 are illustrated in madau ( 1997 ) , where they were revised in order to incorporate more galaxies with @xmath297 .
less of the bottom right corner is clipped than in m96 .
we have opted for a rectangular selection box in order to minimize this bias and keep the color selection as simple as possible , and yet be very close to the m96 selection box .
a total of 100 sources meet our selection criteria .
these are listed in table [ t : udrop ] .
photometry for these galaxies can be found in williams et al .
( 1996 ) . in comparison
m96 selected 69 @xmath0-dropouts , and madau ( 1997 ) note that there are @xmath298 galaxies in their revised selection box .
dickinson ( 1998 ) using more robust photometry , a somewhat different color selection and a fainter limiting magnitude , isolates 187 @xmath0-dropouts in the hdf .
l l l l 2 - 35.0 & 2 - 689.2 & 3 - 389.0 & 4 - 317.0 2 - 76.0 & 2 - 726.0 & 3 - 504.0 & 4 - 363.0 2 - 80.0 & 2 - 741.0 & 3 - 510.0 & 4 - 382.0 2 - 82.0 & 2 - 751.0 & 3 - 515.11 & 4 - 389.0 2 - 122.0 & 2 - 793.0 & 3 - 550.0 & 4 - 445.0 2 - 127.0 & 2 - 802.11112 & 3 - 593.0 & 4 - 488.0 2 - 239.0 & 2 - 822.0 & 3 - 633.1 & 4 - 491.0 2 - 321.2 & 2 - 824.0 & 3 - 674.0 & 4 - 497.0 2 - 373.0 & 2 - 889.0 & 3 - 677.0 & 4 - 555.1 2 - 381.12 & 2 - 890.0 & 3 - 736.0 & 4 - 557.0 2 - 392.0 & 2 - 901.0 & 3 - 748.0 & 4 - 576.0 2 - 449.0 & 2 - 903.0 & 3 - 813.0 & 4 - 588.0 2 - 454.0 & 2 - 916.0 & 3 - 875.0 & 4 - 590.0 2 - 456.12 & 2 - 949.0 & 3 - 902.0 & 4 - 599.0 2 - 456.22 & 2 - 973.11 & 3 - 916.0 & 4 - 602.0 2 - 525.0 & 3 - 41.0 & 4 - 52.0 & 4 - 639.0 2 - 547.0 & 3 - 83.0 & 4 - 83.0 & 4 - 660.0 2 - 565.0 & 3 - 118.0 & 4 - 85.2 & 4 - 676.0 2 - 585.1 & 3 - 126.0 & 4 - 96.0 & 4 - 681.0 2 - 591.0 & 3 - 221.2 & 4 - 109.0 & 4 - 724.11 2 - 594.0 & 3 - 243.0 & 4 - 194.0 & 4 - 825.0 2 - 604.1 & 3 - 286.0 & 4 - 252.0 & 4 - 858.0 2 - 637.0 & 3 - 296.1 & 4 - 274.0 & 4 - 878.1 2 - 643.0 & 3 - 321.2 & 4 - 289.0 & 4 - 926.2 2 - 689.11 & 3 - 335.0 & 4 - 316.0 & 4 - 936.0 a small difference with m96 concerns how sub - structure is included .
the hdf catalog lists objects found with the focas software ( jarvis & tyson 1981 ) , which often splits a `` parent '' object into `` daughters '' . both parents and daughters
are listed in hdf v2.0 .
m96 carefully evaluate the image of each candidate @xmath0-dropout to determine whether it is likely to apply to the whole parent object or the daughter(s ) . here , in cases where both daughter(s ) and corresponding parent meet our selection criteria ,
only the parent object was included in our list of @xmath0-dropouts .
this could cause the integrated 1600 luminosity density @xmath299 to be overestimated .
this is because not all daughters need satisfy the @xmath0-dropout selection criteria for the parent to satisfy these same criteria . to estimate the maximum magnitude of this effect we considered the opposite extreme - when both parent and daughter(s ) meet the selection criteria only the ( lowest level )
daughters that satisfy the selection criteria were included .
this algorithm , results in @xmath299 being lower than our adopted sample by 16% ( 24% after absorption correction ) .
this algorithm will underestimate @xmath299 because some daughters , having the right @xmath15 to be selected as @xmath0-dropouts , will fail the magnitude limit , or because photometric errors ( larger for the fainter daughters ) may shift some daughters out of the color selection box . the ideal parent - daughter separation will , therefore , lie somewhere between these two extremes , and hence the effect on @xmath299 is likely to be a @xmath300% effect .
the top panel of figure [ f : udrop ] shows the color ( @xmath2 ) absolute magnitude @xmath4 of our hdf @xmath0-dropout sample . here
, @xmath2 was calculated from the @xmath270 color listed in the hdf v2.0 catalog ( williams et al .
1996 ) using eq .
[ e : bphot2 ] ; where @xmath15 was taken from the compilation of dickinson ( 1998 ) where possible , otherwise we adopt an average value @xmath301 . this is the same @xmath0-dropout @xmath302 adopted by m96 , and very close to @xmath303 of the galaxies in our sample with measured @xmath15 .
the apparent ab magnitude at @xmath304 was interpolated between the @xmath305 and the @xmath306 magnitudes .
this is effectively a @xmath307 correction .
the @xmath27 axis shows the absolute ab magnitude at @xmath308corrected for uv absorption following eq .
[ e : afit ] , while the broken diagonal lines indicate constant @xmath309 if there were no @xmath3 correction .
the hatched region shows the region that is incomplete due to our selection criteria .
the lower selection limit is slanted because of the @xmath3 correction and because of color terms in the @xmath307 correction and in the translation of the @xmath310 limiting magnitude into the @xmath305 and @xmath306 bands . the middle panel of fig .
[ f : udrop ] shows the @xmath2 distribution of the @xmath0-dropouts , while the bottom panel shows that of the calibrating local sample for comparison .
the color distribution of the local sample spans a larger range than the @xmath0-dropouts . the two - sided kolmogorov - smirnov test gives a 0.3% chance that the two distributions were randomly drawn from the same parent distribution .
these differences are partially due to our @xmath311 selection limit which corresponds to @xmath312 for @xmath207 , or equivalently @xmath313 mag .
the local sample has no specific @xmath2 selection .
if we apply a @xmath314 limit to the local sample then the two - sided kolmogorov - smirnov test gives a 3% chance that the local and @xmath0-dropout sample were randomly drawn from the same parent distribution ; i.e. the two samples are more similar when similar @xmath2 selection criteria are applied .
the @xmath0-dropout sample is _ bluer _ than the local sample having @xmath315 .
the 20th percentile , median and 80th percentile of the @xmath0-dropout distribution are 1.9 , 1.6 , and 1.2 , while for the ( @xmath2 limited ) local sample they are 1.8 , 1.4 , and 0.8 , respectively .
one should nt read too much into this difference , in that the two samples were selected quite differently .
the @xmath0-dropouts have well defined magnitude and color limits in the rest - frame uv while the local sample is composed of galaxies observed with iue for a wide range of programs with heterogeneous initial selection criteria .
luminosity functions at 1600 , @xmath316 are shown for our @xmath0 dropout sample in fig .
[ f : lf ] .
two cases are shown : with ( @xmath317 ) and without ( @xmath316 ) the @xmath3 correction .
the shape of @xmath316 agrees well with that given by dickinson ( 1998 ) for @xmath318 abmag .
the drop - off in the last bin of our @xmath316 is due to incompleteness .
dickinson s luminosity function extends further because of his superior photometric technique and consequent fainter limiting magnitude .
our luminosity function is 0.3 dex higher than that displayed by dickinson because he renormalizes to match the luminosity function of @xmath0-dropouts selected from a larger ground based survey ( steidel et al .
this factor of two difference probably reflects the field to field variation in @xmath299 .
the 1600 luminosity functions extend well beyond the range given for a sample of local uv selected galaxies by buat & burgarella ( 1998 ) .
their luminosity function shows a distinct break at @xmath319 , corresponding to @xmath320 abmag ; a sfr of 4 year@xmath26 .
the brightest galaxies in our @xmath0-dropout sample have @xmath321 abmag , @xmath322 , corresponding to a sfr of 19 , 220 year@xmath26 with and without absorption correction , respectively .
this is further confirmation of the assertion that @xmath0-dropouts are much more luminous than local uv selected starbursts .
however , there are local starbursts with similarly high sfrs .
these are the ultra - luminous infrared galaxies ( sanders & mirabel , 1997 ) which are selected in the far infrared . from direct summation
we derive the total absorption corrected uv luminosity of our selected @xmath0-dropouts to be @xmath323 . assuming that the redshift range of @xmath324 to 3.5 is uniformly sampled , then the volume containing the @xmath0-dropouts is @xmath325 ( m96 ) , and the corresponding volume emissivity is @xmath326 ( @xmath327 ) .
the error quoted here is the quadratic sum of the zeropoint errors from equations [ e : irxcal ] , [ e : afit ] and [ e : bphot2 ] , and the sampling error determined from the bootstrap technique ( e.g. babu & feigelson 1996 ) .
this @xmath328 is actually a lower limit for the volume sampled by the hdf because we have made no completeness correction .
this raises a difficult issue .
the standard practice is to estimate the completeness from the luminosity function , without dust absorption . doing so , madau ( 1997 ) estimates a completeness correction factor of 1.45 .
figures [ f : udrop ] and [ f : lf ] show that in the presence of dust , the limiting @xmath4 is not uniform .
rather it is much shallower for highly reddened galaxies than for lightly reddened sources .
hence the completeness correction could be higher .
in addition there may be field to field differences in @xmath328 , due to large scale clustering .
such variations could be as high as a factor of 2 ( dickinson , 1998 ) .
how accurate is our @xmath0-dropouts selection criteria for selecting high-@xmath15 galaxies ?
our sample includes all but three of the spectroscopically confirmed hdf galaxies with @xmath329 as compiled by dickinson ( 1998 ) .
the galaxy 1 - 54.0 was missed because it appeared on the pc1 chip rather than one of the wf chips ; 3 - 512.0 has @xmath330 and is @xmath208-dropout and hence too red to be a @xmath0-dropout ; and 3 - 577.0 was not included because it is too faint for inclusion in our catalog .
the remaining 23 spectroscopically confirmed high-@xmath15 galaxies account for 47% of @xmath328 .
we cross - checked our @xmath0-dropout selection against a list of all redshift measurements in the hdf ( kindly made available by mark dickinson ) .
none of our selection corresponds to galaxies with known @xmath331 .
dickinson ( 1998 ) notes a 90% success rate in finding high-@xmath15 galaxies in the appropriate redshift range using the @xmath0-dropout technique on their ground based sample .
while the photometric bands and selection criteria are somewhat different from ours , this suggests that the contribution of low-@xmath15 interlopers to @xmath328 should not be much higher than 10% .
in this paper we are primarily interested in the effects of dust on the high-@xmath15 uv emissivity .
if we calculate @xmath332 and @xmath299 without correcting for dust absorption , the quantities are a factor of @xmath333 lower than the above estimates , e.g. @xmath334 .
this is the net effect of dust absorption on _ our sample _ of @xmath0-dropouts . for comparison ,
m96 give @xmath335 for their sample of @xmath0-dropouts .
the samples are identical in their limiting magnitudes , so the difference in the uncorrected emissivities is due to the sample selection . in particular , fig .
[ f : two2cd ] shows that the portion of @xmath332 missing from the m96 sample is contained within the corner clipped from their otherwise rectangular color selection box .
this corner contains preferentially reddened galaxies , hence this clipping biases against large @xmath3 .
relative to @xmath299 estimated by
m96 our emissivity is a factor of 9.0 larger .
we reiterate that the actual absorption - corrected @xmath328 will be even higher than our estimates because we have not corrected for completeness , and we are not sensitive to galaxies having @xmath336 mag .
more recently , madau ( 1997 ) and m98 now estimate an uncorrected @xmath337 for the volume sampled by the hdf @xmath0-dropouts , only 4%less than our estimate without dust correction .
it is higher than in m96 because , as noted above , they adopt significant completeness corrections ( a factor of 1.45 ) and use revised @xmath0-dropout selection criteria that clip less of the bottom - right corner in the @xmath338 vs. @xmath296 plane .
the 1600 dust absorption factor , @xmath333 ( equivalent to @xmath339 mag ) , derived here for the @xmath0-dropouts is lower than our original estimate of a factor of 15 ( paper ii ) , albeit within the factor of three uncertainty we originally estimated .
sawicki & yee ( 1997 ) using the calzetti ( 1997 ) absorption law , derived similarly high dust absorption factors of 15 to 20 .
our new estimates are somewhat lower for a combination of reasons .
firstly , we now fit the - @xmath2 relation instead of adopting an absorption law ( calzetti et al .
1994 ) that is unconstrained by fir data .
secondly , in paper ii we adopted the median absorption of the spectroscopically confirmed @xmath0-dropouts . because of the @xmath340 correlation ( fig .
[ f : udrop ] ) this biases the net absorption . finally , when deriving @xmath2 from @xmath270 we now add a @xmath15 dependent correction ( eq .
[ e : bphot2 ] ) to remove lyman - forest and lyman edge reddening , thus resulting in bluer colors and a smaller derived @xmath3 .
the uncertainty in our new estimate of the mean dust absorption factor of @xmath0-dropouts amounts to 16% , and includes zeropoint and sampling errors .
it is likely to underestimate the true uncertainty since it does not include field to field variations .
m98 and pettini et al .
( 1998 ) estimate 1500 dust absorption factors of about three .
thus the difference between these groups and ours has narrowed considerabley , and is probably within realistic errors on the dust absorption factor .
however we reiterate our caution : @xmath333 represents a lower limit to the true mean 1600 dust absorption factor of galaxies at @xmath1 . as discussed in
[ ss : scuba ] , true 1600 dust absorption factors on the order of 10 are still not ruled out .
buat & burgarella ( 1998 ) using a method very similar to ours , derive a somewhat lower mean absorption of 1.2 mag at @xmath341 for a _ local _ sample of starbursts selected to be visible in both the uv and fir .
the difference probably is largely due to the lower luminosity of the galaxies in their sample , which has a median @xmath342 ( sfr @xmath343 ) .
the @xmath0-dropout sample has a median @xmath344 ( sfr @xmath345 ) , nearly a factor of 100 higher . because of the @xmath340 correlation ( fig .
[ f : udrop ] ) , we expect the local sample to have a lower average amount of dust absorption ( cf .
heckman et al . 1998 ) . in summary ,
our value for the dust absorption corrected uv emissivity @xmath346 at @xmath347 is a factor of 9.0 higher than the uncorrected value given by m96 and a factor of 5.4 times higher than the uncorrected value given by m98 .
the increased uv emissivity relative to m96 is due to a combination of an improved @xmath0-dropout selection ( factor of 1.7 ) , and correction for dust absorption ( factor of 5.4 ) .
our determination of the uv opacity in high-@xmath15 galaxies relies on the similarity of the @xmath0-dropouts to local starburst galaxies ; the intrinsic seds of the two groups must be similar for our method to work . in
[ s : comp ] we reviewed how the overall spectral properties of the two samples are similar .
here we concentrate on recent observational studies of high-@xmath15 galaxies that are especially relevant to estimating uv opacities . if the high-@xmath15 galaxies are intrinsically redder in the uv than local starbursts ( i.e. @xmath348 ) , then we would overestimate @xmath3 and hence @xmath328 .
this could occur if the high-@xmath15 galaxies are dominated by _ older _ stellar populations ( which would have a larger @xmath15 of formation ) or , if for some reason , their stellar imf were deficient in the highest mass stars .
rest - frame optical emission lines present an ideal test for this hypothesis .
starbursts are by nature dominated by highly ionizing populations . indeed the calibrating iue sample used in [ s : cal ] were classified as starbursts primarily by the presence of intense narrow optical emission lines , arising in the ionized medium surrounding the starbursts .
hence , the ratio of line flux to flux density @xmath349 is a measure of the color from below the lyman - limit to 1600 , modulated by dust absorption . for @xmath350 rest - frame optical emission lines
are redshifted into the observed - frame near infrared . with careful , and very long exposures ,
the brightest optical lines are detectable in the brightest @xmath0-dropout galaxies with four meter class telescopes .
detections and fluxes have been reported by several groups including pettini et al .
( 1998 ) , wright & pettini ( 1998 ) , bechtold et al .
( 1997 ) , and bechtold et al.(1998 ) .
the detections correspond to @xmath0-dropouts discovered from the ground or in the hdf , as well as other sources that meet the photometric criteria for being @xmath0-dropouts .
figure [ f : uvem ] shows the line fluxes of these sources normalized by the rest frame 1600 flux density @xmath349 of the sources compared to @xmath2 .
both @xmath349 and @xmath2 were estimated from published photometry , where possible ( otherwise from published spectra ) .
the pseudo - equivalent widths @xmath351 have been corrected for galactic extinction and cosmological stretching only .
no attempt has been made to correct for internal absorption .
results for three different lines are shown - , , and . in cases where the line was detected ,
it was multiplied by 3.0 to convert it to an flux . a sample of local starburst galaxies observed by storchi - bergmann et al .
( 1995 ) and mcquade , calzetti , & kinney ( 1995 ) is shown for comparison .
the relevant data is listed in table [ t : loc ] .
this uv selected sample has a high degree of overlap with the sample used in [ s : cal ] .
figure [ f : uvem ] shows that the detected @xmath0-dropouts have line flux to uv flux density ratios consistent with local starburst galaxies .
the one exception is the observation of the lensed galaxy ms1512-cb58 by bechtold et al .
wright & pettini ( 1998 ) have also detected this galaxy in emission , and the corresponding pseudo - equivalent width is well within the scatter of points for local starbursts as shown in the bottom panel of fig [ f : uvem ] .
the discrepancy between the and measurements may be due to the lensed nature of the source , or perhaps may reflect the difficulty of the observations which were obtained at @xmath352 , the edge of the @xmath353 band window . at this stage ,
none of the groups report non - detections in emission lines of known @xmath0-dropouts .
we conclude that @xmath0-dropouts are ionizing sources to the same degree as local starbursts , and hence do not have an intrinsically redder continuum than the local sample .
if there is significant uv extinction of galaxies at @xmath1 , then perhaps the dust can be detected directly in the rest frame fir or indirectly in the radio via the radio - fir correlation ( helou 1991 ) .
two groups have recently claimed ( marginal ) detections of sources in and near the hdf in these wavelength regimes some of which which they attribute to @xmath1 star forming galaxies . here
we show that the number and fluxes of these marginal detections are consistent with the expectations of our dust absorption model .
the sub - mm regime , corresponding to the rest frame fir , is a promising window to look for high-@xmath15 galaxies since the `` negative @xmath307-corrections '' for dust emission can cancel out the effects of fade - out with distance , making dusty galaxies easier to detect with increasing @xmath15 ( guiderdoni et al .
several groups have now obtained deep sub - mm imaging with scuba on the 15 m jcmt ( e.g. smail , ivison & blain , 1997 ; barger et al .
1998 ; eales et al.1998 ; hughes et al . 1998 ; lilly et al . 1998 ) .
all report 850detections at the mjy level that they largely attribute to galaxies having @xmath22 .
indeed one of the scuba detections , smm02399 - 0136 , has a clear optical counter - part and spectroscopic determination of @xmath354 ( ivison et al .
however , lilly et al .
( 1998 ) note that a sizable fraction of these sources may have @xmath355 .
furthermore , the emission line spectrum of smm02399 - 0136 is indicative of a seyfert 2 galaxy , cautioning us that not all of the dust seen by scuba need be warmed by star - formation alone . the study of hughes et al .
( 1998 ) is particularly relevant because it is of a field centered on the hdf .
they report five sources above their detection limit of 2mjy .
since this limit is set by source confusion ( beam fwhm@xmath356 ) , none of their optical identifications is totally secure .
they attempt to improve the accuracy of their identifications using photometry , including photometric redshifts .
they claim that 4/5 of their detections are best identified with sources having @xmath357 .
the brightest of these has @xmath358 mjy . in table
[ t : rad ] we present @xmath359 estimates for hdf @xmath0-dropouts based on our dust absorption model , and the optical photometry of williams et al .
( 1996 ) . to do this
we used eqns . [
e : irxcal ] and [ e : afit ] to estimate the rest frame @xmath28 from the rest frame @xmath204 and @xmath2 estimates .
@xmath28 was then converted to @xmath359 using the fir sed of the prototypical ir luminous starburst arp 220 as modeled by a spline fit through the the measurements of rigopoulou et al .
( 1996 ) and klaas et al .
( 1997 ) . only the galaxies with the brightest predicted
@xmath359 values are listed .
the table is arranged in order of decreasing predicted @xmath359 . the corresponding rest
frame 1600 flux densities ( uncorrected for dust absorption ) do not come out to be similarly sorted .
this is because it is the dust reprocessed flux that is important .
since our adopted @xmath0-dropout sample is color limited ( i.e. @xmath313 mag ) , the brightest predicted infrared fluxes correspond to the brightest and reddest @xmath0-dropouts .
the uv absorption @xmath3 listed in this table is calculated from the @xmath270 as outlined in
[ ss : bacal ] .
l r c c c c c c c 4 - 52.0 & 2.931 & 24.06 & 0.66 & 3.05 & 0.63 & 1.8 & 3.7 & @xmath3604 - 878.1 & ( 2.75 ) & 23.28 & 1.06 & 2.32 & 1.35 & 1.6 & 3.7 & 4 - 639.0 & 2.591 & 24.16 & 0.73 & 2.92 & 0.52 & 1.1 & 2.8 & 2 - 585.1 & 2.002 & 23.50 & 0.73 & 2.93 & 0.76 & 1.1 & 4.7 & 4 - 555.1 & 2.803 & 23.33 & 1.27 & 1.93 & 1.39 & 1.1 & 2.3 & 4 - 445.0 & 2.268 & 23.75 & 0.80 & 2.79 & 0.69 & 1.0 & 3.5 & 4 - 382.0 & ( 2.75 ) & 24.10 & 0.95 & 2.52 & 0.61 & 0.9 & 2.2 & 2 - 127.0 & ( 2.75 ) & 24.72 & 0.79 & 2.81 & 0.33 & 0.7 & 1.6 & 2 - 449.0 & 2.008 & 23.42 & 1.14 & 2.17 & 1.01 & 0.6 & 2.7 & @xmath3614 - 858.0 & 3.216 & 24.06 & 1.40 & 1.70 & 0.78 & 0.6 & 1.0 & none of these galaxies have predicted @xmath362 mjy , consistent with the fact that none were detected by hughes et al .. likewise , none of the claimed optical counterparts of @xmath363 detections meet our criteria of being @xmath0-dropouts .
one is clearly a @xmath364 galaxy .
three are fainter than our adopted inclusion limit .
the remaining galaxy hdf850.3 is in the pc1 chip and identified as 1 - 34.2 .
its @xmath365 is within our selection limits , but its @xmath366 color is marginally too blue to be considered a @xmath0-dropout , but consistent with a photometric @xmath367 estimated by mobasher et al .
( 1996 ) .
the non - intersection of our @xmath0-dropout catalogue and the 850@xmath368 sample of hughes et al .
suggests that the sub - mm detections correspond to infrared luminous starbursts or agn that are so dust enshrouded that they are not selected as @xmath0-dropouts either because they are too extincted and/or because they are too reddened .
the hughes et al.identification of two 850 sources ( hdf850.1 , hdf850.2 ) with fairly blue galaxies ( @xmath311 ) also suggests that the @xmath2- calibration breaks down for heavily extincted infrared - luminous starbursts .
further work is needed to confirm these identifications and secure reliable redshifts . the remarkably tight radio - fir correlation ( helou 1991 )
provides an indirect means to probe the fir emission , and hence dust emission of starbursts .
it is best defined at @xmath369 ghz ( 21 cm ; e.g.sanders & mirabel , 1997 ) .
the radio emission of a wide range of star forming galaxies , including starbursts , is dominated by synchrotron emission at this frequency and up to @xmath370 ghz .
hence the correlation can be used to infer fir fluxes from radio observations down to rest wavelengths of about 1 cm .
do we expect this correlation to hold at high-@xmath15 ?
inverse compton scattering from cosmic microwave background photons provides an alternative mechanism to synchrotron emission for cooling the relativistic electrons .
the energy density of these photons will scale as @xmath371 .
however it is the total energy density of photons that is important for inverse compton cooling .
even at @xmath1 the energy density of cosmic microwave background photons will be an order of magnitude lower than internally produced energy density of starbursts having typical bolometric surface brightnesses of @xmath372 ( e.g. paper ii ) at @xmath373 .
since synchrotron emission clearly occurs in the face of intense internally produced emission in low-@xmath15 galaxies , it does not seem likely that inverse compton cooling will quench it for high-@xmath15 galaxies where the total energy density is only marginally higher .
richards et al .
( 1998 ) present a deep imaging survey of the hdf and surrounding areas obtained with the vla at 3.5 cm ( @xmath374 ghz ) .
their @xmath375 detection limit is @xmath376jy , and their @xmath377 marginal detection limit corresponds to @xmath378jy .
they marginally detect two hdf @xmath0-dropouts , 4 - 52.0 and 2 - 449.0 as listed in the hdf v2.0 catalog . while the detections are provisional , richards et al .
are confident that they are not spurious because they also are detected in a 1.4 ghz image of the hdf .
the logarithmic ratio of @xmath273 flux densities in the rest frame radio ( 1 cm @xmath379 ghz ) and uv ( 1600 @xmath380 phz ) and can be written as : @xmath381 where @xmath382 is the radio - fir logarithmic ratio : @xmath383}{3.75 \times 10^{12}\,[{\rm hz}]}\right/ f_\nu [ { \rm jy } ] \left\ } \frac { } { } \right .. \ ] ] at @xmath384 cm , @xmath385 ( sanders & mirabel , 1997 ) .
since starbursts typically have radio spectral slopes @xmath386 ( where @xmath387 ) , we expect @xmath388 . for the prototypical starburst galaxy m82 , @xmath389 is derived from the data listed by carlstrom & kronberg ( 1991 ) .
this is somewhat lower than expectations because free - free emission is starting to make a contribution to the 1 cm flux of m82 ( carlstrom & kronberg 1991 ) .
we adopt a rest frame @xmath390 and calculate the expected @xmath273 at 3.5 cm ( @xmath391 ) of the @xmath0-dropouts in our hdf sample using the above formulation , and modest ( factor of a few at most ) `` @xmath307 '' corrections to transfer from the rest frame wavelengths of 1600 and 1 cm , and the observed frame @xmath306 and @xmath392 cm observations .
table [ t : rad ] lists the results of these calculations for the expected 10 sub - mm brightest @xmath0-dropouts . note that the @xmath273 ranking of the radio predictions is different from that of the fir predictions
this is because the @xmath307 corrections in the two wavelength regimes work in opposite direction with redshift at @xmath1 .
the sources in table [ t : rad ] include the two detections of richards et al .
note that all of the galaxies in this table have predicted @xmath393 below the formal detection limits of richards et al .. however the top five @xmath393 estimates are not far below their marginal detection limit , and include these two sources . in such cases
we expect the fortuitous coincidences of uncertainties to allow some positive detections .
these uncertainties include the measurement bias mentioned above , the flux uncertainties listed by richards et al.(@xmath160.1 dex uncertainty ) , the intrinsic scatter about the radio - fir relationship ( @xmath394 dex uncertainty ; devereaux & eales 1989 ) , and the random scatter about the @xmath3 versus @xmath2 fit ( @xmath160.1 dex uncertainty ) .
the total random uncertainty is @xmath394 dex or a factor of @xmath395 .
hence random errors alone can account for the difference in the predicted and observed @xmath393 in the two galaxies detected by richards et al .. why are nt these galaxies also detected in the rest frame fir ? scaling our predictions by the @xmath396 observed by richards et al .
, we then expect @xmath397 mjy for 4 - 52.0 and 2 - 449.0 respectively .
the latter is below the hughes et al .
detection limit , while the former is above it but by less than the flux uncertainty , and so could easily be missed .
we conclude that the richards et al .
radio detections are consistent with the hughes et al . scuba non - detections .
our calculations show that @xmath0-dropouts in the hdf are just at the detection limit of state of the art deep rest frame fir and radio observations .
somewhat deeper observations in these bands of either the hdf or other fields containing @xmath0-dropouts ( preferably the brightest and reddest ones ) have the potential for directly testing the @xmath2 relationship and verifying the existence of the radio - fir correlation at @xmath1 . hence they could be used to directly test the form of the uv absorption law . for example , the smc law absorption model adopted by pettini et al .
( 1997 ) would predict @xmath398 mjy , and @xmath399jy for all the @xmath0-dropouts in our sample .
hence mjy level imaging at 850 and @xmath400jy level radio continuum imaging should detect the brightest @xmath0-dropouts if our absorption model is correct , but would not detect any if lightly absorbed models were correct .
the color - magnitude diagram of hdf @xmath0-dropouts , fig .
[ f : udrop ] , shows that the highest luminosity galaxies , with or without @xmath3 correction , tend to be red .
similarly the bluest galaxies tend to have the lowest luminosity .
that is , there is a color - luminosity correlation ( also noted by dickinson , 1998 , private communications ) .
the @xmath2 - luminosity correlation is similar to that seen in local starburst galaxies ( heckman et al .
1998 ) . as is the case for local starbursts ,
this correlation is not solely due to selection effects . while fig .
[ f : udrop ] clearly shows that we select against low - luminosity red @xmath0-dropouts , there is no selection against high luminosity very blue objects .
the more luminous starbursts have a higher dust content .
presumably , dust content is related to metallicity .
this is certainly born out for local starbursts where [ o / h ] correlates with @xmath2 ( heckman et al .
1998 ) . from paper ii
, we know that the @xmath0-dropouts have similar surface brightnesses , hence the most luminous ones are the largest , and presumably the most massive galaxies . putting these together implies a mass - metallicity relationship for @xmath0-dropouts like that seen in starburst ( heckman et al .
1998 ) and normal galaxies in the local universe .
the presence of dust in the most luminous @xmath0-dropouts implies that they were previously enriched with metals .
hence we are not seeing the most luminous galaxies at the instant they `` turn on '' .
their first generation of star formation must have occurred at @xmath401 .
figure [ f : uvem ] illustrates the ( un ) usefulness of line to uv continuum ratios as indicators of dust absorption ; the @xmath27 axis is proportional to the ratio of the sfrs one would derive from emission lines and the uv continuum respectively , in the absence of dust absorption .
pettini et al .
( 1998 ) argue that the similarity of the sfr derived from fluxes and the uv continuum indicates that dust absorption is not extreme .
however , fig .
[ f : uvem ] shows little correlation between @xmath402 with @xmath2 . in local starburst systems , we know that @xmath2 is a good dust absorption indicator ( fig . [
f : uvir ] ) ; apparently @xmath402 is not .
the only ratio that noticeably correlates ( positively ) with @xmath2 is @xmath403 .
however , the slope of the correlation is about half that expected by application of the lmc extinction law .
the @xmath404}/f_{1600}$ ] ratio even shows an anti - correlation with @xmath2 .
this is probably due to the correlation of [ o / h ] with @xmath2 found for starbursts ( heckman et al .
, 1998 ) highly reddened starbursts have higher metallicities , lower electron temperatures in the ionized medium , and hence lower [ oiii]/h@xmath2 ratios than low reddening starbursts .
the @xmath403 and @xmath402 results illustrate a result that is often overlooked : the dust absorption estimated from continuum observations is usually less than that estimated from emission .
this was first noted by fanelli , oconnell , & thuan ( 1988 ) and subsequently confirmed by others including mas - hesse , arnault , & kunth ( 1989 ) , and calzetti et al .
the lack of correlation between @xmath402 and @xmath2 indicates that the effective extinction law of starbursts is such that @xmath405 .
while uncorrected uv continuum fluxes and line fluxes may imply the same sfrs , both are affected by similar amounts of dust absorption .
hence , emission line flux to uv continuum ratios do not provide strong constraints on dust absorption .
the sfr of the various scuba detections is typically derived to be @xmath406 ( e.g. lilly et al .
1998 ) . as shown in fig .
[ f : lf ] , this is consistent with the sfr derived for the brightest and reddest @xmath0-dropouts , after dust absorption correction .
such high values for the sfr are unprecedented for local uv selected starbursts , and typically only seen in fir selected starbursts such as arp220 .
this suggests that the most luminous starbursts at @xmath407 are less dust enshrouded than those seen locally - the ultra - luminous infrared galaxies . however , this conjecture is difficult to test , since there are very few uv observations of ultra - luminous infrared galaxies ( cf .
trentham et al .
1999 ) , hence the amount of uv absorption they suffer has not been properly characterized .
also shown in fig .
[ f : lf ] is lilly s et al .
( 1998 ) estimate of the number density of the scuba sources in the redshift range @xmath408 .
this is also in fair agreement with the number density of the @xmath0-dropouts having the same luminosity .
this comparison suggests that the scuba sources may be largely @xmath0-dropouts with modest amounts of dust absorption , rather than being completely opaque in the uv .
furthermore , it suggests that much of the sfr density at @xmath1 can be recovered from rest - frame uv measurements alone . working against this suggestion
is the fact that none of the hdf scuba sources detected by hughes et al .
( 1998 ) correspond to known @xmath0-dropouts . however , the numbers are small , and the association with optical galaxies is still not secure .
hughes et al .
estimate @xmath409 at @xmath1 from their scuba detections which is about half of the @xmath410 we derive .
this is only @xmath16 five times larger than the m96 estimate ( @xmath411 times higher than the m98 estimate ) . like us
they do not extrapolate the luminosity function of their detections , hence the @xmath410 is a lower limit consistent with our estimates .
alternatively , if the scuba sources are at @xmath1 and a non - intersecting set with respect to @xmath0-dropouts ( as suggested by the hughes et al .
identifications ) , then the sum of the hughes et al.@xmath410 and that derived from our @xmath0-dropout sample provides an improved lower limit : @xmath412 at @xmath1 which is a factor @xmath413 larger than that estimated by m96 , and @xmath414 times the uncorrected value estimated by m98 . m98 argue against such a high value of @xmath410 since it would over - produce low mass stars in the present universe if it went on for more than about one gyr .
however , this argument assumes that the lyman - dropouts have an imf with constant salpeter ( 1955 ) slope over the mass range of 0.1 to 125 .
recent observations of local starbursts suggest that their imf is deficient in stars less massive than 3with respect to the salpeter imf ( nota et al .
1998 ; siriani et al .
1999 ) since @xmath410 is very model dependent , a more robust prediction of interest is the lower limit to the fir emissivity . summing the predicted fir fluxes of all the hdf @xmath0-dropouts ( using eq .
[ e : irxcal ] ) and normalizing by the volume yields @xmath415 at @xmath416 evaluated at a rest wavelength of 80@xmath400 m .
if the scuba detections are too reddened or dimmed to be selected as @xmath0-dropouts then we can add them to our sample to get an improved lower limit : @xmath417 .
these are still lower limits because completeness corrections have not been made .
these predictions can be tested directly with improved fir , sub - mm , and radio observations ( e.g. with the upcoming sub - mm array , and the sirtf , first , and planck space missions ) .
our calibration of uv extinction is readily adapted to observations at redshifts other than @xmath1 and hence other spectral domains .
the main requirement is that the data include fluxes in two broad bands or coarse spectra covering the rest frame uv .
redshift information is also helpful , and that can be obtained photometrically to adequate accuracy from complementary rest frame optical - near ir data ( e.g. from the sloan digital sky survey , or deep pointed observations ) .
for example , serendiptous detections of galaxies with @xmath418 ( e.g. franx et al . 1997 ; dey et al .
1998 ) indicate that galaxy formation had commenced by this epoch .
one of these galaxies is a lensed arc at @xmath419 , whose broad band ( observed frame ) optical - near ir photometry yield @xmath420 ( soifer et al . 1998 ) , implying @xmath421 mag . upcoming all sky surveys in the uv by galex ( martin et al .
1997 ) and tauvex ( brosch , et al . 1994 ) will probe the cosmological uv emissivity to @xmath422 .
extending deep imaging with the hst into the near ir with nicmos allows lyman edge systems to be detected out to @xmath423 ( @xmath306-dropouts ) .
there are a few caveats to using our methods , especially at low-@xmath15 .
firstly , it is based on the assumption that the uv emissivity is dominated by starbursts .
possible contaminants are normal disk galaxies and old stellar populations .
ideally we want to include the uv light from normal disk galaxies when estimating star sfrs .
the problem is that it is not known whether normal disks follow the starburst reddening law derived here .
their cool and warm ism is dominated by a disk morphology , while starbursts typically have significant quantities of extra - planar ism ejected by galactic winds ( heckman , armus , & miley 1990 ) . in paper
i we suggested that winds shape the ism around starbursts into a foreground screen , thus producing the - @xmath2 correlation .
further work has to be done to determine if normal disks have a reddening - uv absorption relation and to calibrate it .
fortunately normal disk and starburst galaxies can be distinguished in the uv by effective surface brightness ( paper ii ) .
cosmological surface brightness dimming renders normal spiral galaxies difficult to detect by @xmath1 ( giavalisco et al .
1996 ; hibbard & vacca , 1997 ) , hence they do not contaminate the @xmath0-dropout sample . old stellar populations produce significant uv `` upturn '' emission .
if they significantly contaminate a uv survey then the assumption that uv light is due to recent star formation is seriously compromised .
old stellar populations can be distinguished and removed from uv surveys by extending the broad - band sed into the rest frame optical .
secondly , our technique has been calibrated to a fairly modest @xmath424 mag ( fig .
[ f : uvir ] ) .
while the local calibrating sample contains several well known dusty starbursts ( e.g.ngc 1614 , ngc 3504 , ngc 3256 , ngc 7552 ) , it does nt contain any of the `` ultra - luminous infrared galaxies '' ( ulirgs ) which are the most luminous starbursts known ( sanders & mirabel , 1997 ) .
it is not known if they fall on or continue the - @xmath2 relationship since the number of these galaxies observed in the uv is rather meager ( trentham et al .
the amount of totally obscured star formation in the local universe can be estimated by applying our absorption algorithm to a blind uv survey such as will be produced by galex , after correction for contaminants as suggested above .
this will yield an estimate of @xmath328 and also the corresponding density of uv flux processed to the fir .
comparison with @xmath425 derived from iras , iso , or sirtf observations will yield the fraction of completely buried star formation .
the strong similarity between local starburst galaxies and lyman limit systems at @xmath1 provides a strong impetus to use starbursts as templates for interpreting the properties of high-@xmath15 galaxies .
the fact that starburst galaxies redden as more of their uv light is absorbed by dust is demonstrated by a strong correlation between their fir to uv flux ratios and their ultraviolet spectral slopes @xmath2 ( color ) .
our @xmath3 versus @xmath2 calibration is a simple empirical fit to this correlation .
hence , by design it recovers the uv light that has been reprocessed to the fir .
we applied this calibration to a @xmath0-dropout sample selected from the hdf v2.0 catalog .
this sample has an identical limiting mag to that of madau et al .
( 1996 : m96 ) . however , the color selection is somewhat different , resulting in a fairly uniform limiting @xmath426 mag .
in contrast the m96 @xmath0-dropout sample has a non - uniform bias against large @xmath3 . for our adopted cosmological parameters of @xmath427 , @xmath9 ,
the rest - frame comoving uv emissivity of our sample is @xmath428 without applying @xmath3 corrections .
this is 1.7 times higher than the value derived by m96 , which illustrates the effects of the differing color selections . after correcting for dust absorption
we find @xmath429 , which is factors of nine and five times higher than the uncorrected values of m96 and m98 , respectively .
however , this @xmath328 is still a lower limit because we have not corrected for incompleteness , and we are not sensitive to galaxies having @xmath336 mag .
this study and other recent observations further demonstrate the close correspondence between local starbursts and high-@xmath15 lyman limit systems .
firstly , we find a correlation between @xmath2 ( color ) and @xmath309 ( uv luminosity ) for the @xmath0-dropouts , similar to the correlation seen in local starburst galaxies ( heckman et al .
1998 ) . to first order
the uv luminosity gives the mass of recently formed stars , while @xmath2 correlates well with metal content in local galaxies ( heckman et al .
this implies that there is a mass metallicity relationship out to @xmath1 .
since the highest luminosity galaxies are clearly reddened they must have produced enough metals to create the dust prior to the epoch of observation .
hence the first epoch of star formation occurred at @xmath401 in the most luminous @xmath0-dropouts .
secondly , observation of emission lines in the rest frame optical ( observed in the near - ir ) show that @xmath0-dropouts contain ionizing stellar populations to the same degree as local starbursts .
this can be seen by comparing the line flux to uv flux density ratios with @xmath2 for both samples . however , neither the local nor the @xmath0-dropout samples show a strong correlations in these diagrams as would be expected from naive application of uv extinction / absorption laws .
this suggests that the column density of dust towards emission is higher than towards the uv continuum both in local starbursts and out to @xmath1 .
consequently , the line flux to uv flux density ratio is not a good indicator of dust absorption .
thirdly , the marginal detection of two hdf @xmath0-dropouts at radio rest @xmath430 cm by richards et al .
( 1998 ) and non - detections of the @xmath0-dropouts in the hdf by hughes et al .
( 1998 ) is consistent with estimates of the amount of uv flux reprocessed into the fir .
deeper observations in both domains ( by a factor of @xmath431 ) are required to test the dust reprocessing model proposed here , and to confirm whether the fir - radio correlation holds at @xmath1 .
the methodology presented here is readily adapted to other redshifts . for those sources affected by modest amounts of absorption ( @xmath432 mag )
, the intrinsic uv flux can be recovered from uv quantities alone : flux measurements in two rest frame vacuum - uv bands are sufficient .
these observations can also be used to estimate the total uv flux reprocessed into the fir .
hence , lower limits to the intrinsic @xmath433 and @xmath434 can be determined .
integration of @xmath434 and comparison to the fir background spectrum ( fixsen et al .
1998 ; hauser et al . 1998 ) will constrain the amount of star formation totally obscured by dust from the ultraviolet . we thank two anonymous referees for suggestions that improved this paper .
mark dickinson , harry ferguson , piero madau , chris martin and max pettini stimulated our thinking in numerous discussions .
piero madau kindly calculated cosmic transmission curves for us .
we thank mark dickinson for providing the @xmath15 measurements of the hawaii and berkeley groups collated to the hdf v2.0 catalog .
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 .
this research was supported by nasa grant nag5 - 6400 .
babu , g.j . , & feigelson , e.d .
1996 , `` astrostatistics '' , chapman & hall : london .
barger , a.j .
, cowie , l.l . ,
sanders , d.b .
, fulton , e. , taniguichi , y. , sato , y. , kawara , k. , & okuda , h. 1998 , , 394 , 248 baugh , c.m . ,
cole , s. , frenk , c.s .
, & lacey , c.g .
1998 , , 498 , 504 bechtold , j. , yee , h.k.c .
, elston , r. , & ellingson , e. 1997 , , 477 , l29 bechtold , j. , elston , r. , yee , h.k.c .
, ellingson , e. , & cutri , r.m .
1998 , in `` the young universe : galaxy formation and evolution at intermediate and high redshift '' , eds .
s. dodorico , a. fontana , and e. giallongo .
asp conference series ; vol .
146 , p.241 ( astro - ph/9802230 ) bower , r.g . ,
lucey , j.r . , & ellis , r.s .
1992 , , 254 , 601 boyle , b.j . &
terlevich , r.j .
1998 , , 293 , l49 bouchet , p. , lequeux , j. , maurice , e. , prevot , l. , & prevot - burnichon , m.l . 1985 ,
, 149 , 330 brosch , n. , shemi , a. , netzer , h. , blasberger , a. , & topaz , j.m .
1994 , in `` advances in multilayer and grazing incidence x - ray / euv / fuv optics '' , spie , 2279 , 469 bruzual , a.g . , & charlot , s. 1993 , , 405 , 538 buat , v. , & burgarella , d. 1998 , , 334 , 772 burstein , d. , & heiles , c. 1982 , , 87 , 1165 burstein , d. , & heiles , c. 1984 , , 54 , 33 calzetti , d. 1997 , , 113 , 162 calzetti , d. , kinney , a.l . , & storchi - bergmann , t. 1994 , , 429 , 582 carlstrom , j.e . &
kronberg , p.p .
1991 , , 366 , 422 connolly , a.j . , szalay , a.s . ,
dickinson , m. , subbarao , m.u . , & brunner , r.j .
1997 , apjl , 486 , l11 conti , p.s . , leitherer , c. , & vacca , w.d .
1996 , , 461 , l87 dey , a. , spinrad , h. , stern , d. graham , j.r . , &
chaffee , f.h .
1998 , , 498 , l93 dsert , f .- x . , boulanger , f. , & puget , j.l .
1990 , , 237 , 215 devereaux , n.a . , & eales , s.a .
1989 , , 340 , 708 dickinson , m. 1998 , in _ the hubble deep field _ , eds . m. livio , s.m . fall and p.
madau , in press ( astro - ph/9802064 ) eales , s. , lilly , s. , gear , w. , dunne , l. , bond , j.r . , hammer , f. , le fevre , o. , & crampton , d. 1998 , , submitted ( astro - ph/9808040 ) .
eggen , o.j . ,
lynden - bell , d. , & sandage , a.r .
1962 , , 136 , 748 fanelli , m.n .
, oconnell , r.w . , & thuan , t.x .
1988 , , 334 , 665 fitzpatrick , e.l .
1986 , , 92 , 1068 fixsen , d.j . , dwek , e. , mather , j.c . ,
bennett , c.l . , & shafer , r.a .
1998 , , 508 , 123 franx , m. , illingworth , g. , kelson , d.d .
, van dokkum , p.g . , & tran , k .- v .
1997 , , 486 , l75 frayer , d.t . ,
ivison , r.j .
, scoville , n.z . ,
evans , a.s .
, yun , m. , smail , i. , barger , a.j . ,
blain , a.w .
, & kneib , j .- p .
1999 , , submitted , ( astro - ph/9901311 ) giavalisco , m. , livio , m. , bohlin , r.c . ,
macchetto , f.d . , & stecher , t.p .
1996 , , 112 , 369 giavalisco , m. , steidel , c.c . , &
macchetto , f.d .
1996 , , 470 , 189 gonzalez - delgado , r. , leitherer , c. , heckman , t.m . , lowenthal , j. , ferguson , h. , & robert , c. 1998 , 495 , 698 guilloteau , s. , omont , a. , mcmahon , r.g . ,
cox , p. , & petitjean , p. 1997
, , 328 , l1 guiderdoni , b. , bouchet , f.r . ,
puget , j.l .
, lagache , g. , & hivon , h. 1998 , , 390 , 257 hauser , m.g .
et al . , 1998 , , 508 , 25 heckman , t.m . ,
armus , l. , & miley , g.k .
1990 , , 74 , 833 heckman , t.m . ,
robert , c. , leitherer , c. , garnett , d.r . , & van der rydt , f. , 1998 , , 503 , 646 ivison , r.j . ,
smail , i. , le borgne , j .- f . ,
blain , a.w .
, kneib , j .-
p . , bzecourt , j. , kerr , t.h . , & davies , j.k .
1998 , , 298 , 583 helou , g. , khan , i.r . ,
malek , l. , & boehmer , l. 1988 , , 68 , 151 helou , g. 1991 , in the interpretation of modern synthesis observations of spiral galaxies , edited by n.c .
duric , & p. crane ( asp , san francisco ) , p. 125
hibbard , j.e .
, & vacca , w.d .
1997 , , 114 , 1741 howarth , i.d . 1983 , , 203 , 301 hughes , d. , et al .
1998 , , 394 , 241 ivison , r.j .
et al .
1998 , , 494 , 211 jarvis , j.f . , &
tyson , j.a .
1981 , , 86 , 476 kennicutt , r.c . , jr .
1998 , , 498 , 541 kinney , a.l . ,
bohlin , r.c .
, calzetti , d. , panagia , n. , & wyse , r.f.g .
1993 , , 86 , 5 kinney , a.l . ,
calzetti , d. , bica , e. , & storchi - bergmann , t. 1994 , , 429 , 172 klaas , u. , haas , m. , heinrichson , i. , & schulz , b. , 325 , l21 kunth , d. , mas - hesse , j.m . , terlevich , e. , terlevich , r. , lequeux , j. , & fall , s.m .
1998 , , 334 , 11 leitherer , c. , & heckman , t.m .
1995 , , 96 , 9 lilly , s.j . , tresse , l. , hammer , f. , crampton , d. & le fvre , o. 1995 , , 455 , 108 lilly , s. , le fvre , o. , hammer , f. , crampton , d. 1996 , , 460 , l1 lilly , s. , eales , s. , gear , w. 1998 , in `` the next generation space telescope : science drivers and technological challenges '' , 34th lige astrophysics colloquium , p. 39
lowenthal , j.d . ,
koo , d.c .
, guzmn , r. , gallego , j. , phillips , a.c . ,
faber , s.m . ,
vogt , n.p . , illingworth , g.d .
, & gronwall , c. 1997 , , 481 , 673 madau , p. 1997 , in star formation near and far , edited by s.s .
holt & l.g .
mundy , ( aip , woodbury ny ) , p. 481 madau , p. , ferguson , h.c . ,
dickinson , m.e .
, giavalisco , m. , steidel , c.c . , & fruchter , a. 1996 , , 283 , 1388 ( m96 ) madau , p. , pozzetti , l. , & dickinson , m. 1998 , , 498 , 106 ( m98 ) mas - hesse , j.m . , arnault , p. , & kunth , d. 1989 , , 157 , 131 martin , c. , friedman , p. , shiminovich , d. , madore , b. , bianci , l. , szalay , a. , heckman , t. , milliard , b. , malina , r. , siegmund , o. , welsh , b. , & rich , m. , , 191 , 63.04 mcquade , k , calzetti , d. , & kinney , a.l .
1995 , apjs , 97 , 331 meurer , g.r . ,
freeman , k.c .
, dopita , m.a .
, & cacciari , c. 1992 , , 103 , 60 meurer , g.r . , heckman , t.m . , leitherer , c. , kinney , a. , robert , c. , & garnett d.r .
1995 , , 110 , 2665 ( paper i ) meurer , g.r . , heckman , t.m .
, lehnert , m.d . ,
leitherer , c. , & lowenthal , j. 1997 , , 114 , 54 ( paper ii ) mobasher , b. , rowan - robinson , m. , georgakis , a. , & eaton , n. 1996 , , 282 , l7 nota , a , siriani , m. , leitherer , c. , de marchi , g. , & clampin , m. 1998 , aas , 192 , 8401 .
omont , a. , petitjean , p. , guilloteau , s. , mcmahon , r.g . ,
solomon , p.m. , & pcontal , e. 1996 , , 382 , 428 ortolani , s. , renzini , a. , gilmozzi , r. , marconi , g. , barbuy , b. , bica , e. , & rich , m.r .
1995 , nature , 377 , 701 pei , y.c . , & fall , s.m .
1995 , , 454 , 69 pettini , m. , kellogg , m. , steidel , c.c . ,
dickinson , m. , adelberger , k.l .
, & giavalisco , m. 1998 , , submitted rigopoulou , d. , lawrence , a. , & rowan - robinson , m. 1996 , , 278 , 1049 richards , e.a . ,
kellermann , k.i .
, fomalont , e.b . ,
windhorst , r.a . , & partridge , r.b .
1998 , , submitted ( astro - ph/9803343 ) sanders , d.b . , & mirabel , i.f .
1985 , , 298 , l31 sawicki , m. , & yee , h.k.c .
1997 , 115 , 1329 searle , l. , & zinn , r. 1978 , , 225 , 357 seaton , m.j .
1979 , , 187 , 73p sanders , d.b . ,
soifer , b.t . ,
elias , g. , madore , b.f . , matthews , k. , neugebauer , g. , & scoville , n.z .
1988 , , 325 , 74 .
salpeter , e.e .
1955 , , 121 , 161 siriani , m. , nota , a. , leitherer , c. , de marchi , g. , & clampin , m. 1999 , , submitted smail , i. , ivison , r.j . , &
blain , a.w .
1997 , , 490 , 5 smith , l.f . , biermann , p. , & mezger , p.g . 1978 , , 66 , 65 soifer , b.t , neugebauer , g. , franx , m. , matthews , k. , & illingworth , g.d .
1998 , , 501 , l171 steidel , c.c . , & hamilton , d. 1992 , , 104 , 941 steidel , c.c .
, pettini , m. , & hamilton , d. 1995 , , 110 , 2519 steidel , c.c . , giavalisco , m. , pettini , m. , dickinson , & m. , adelberger , k.l .
1996a , , 462 , l17 steidel , c.c . , giavalisco , m. , dickinson , m. , & adelberger , k.l .
1996b , , 112 , 352 steidel , c.c . , adelberger , k.l . , giavalisco , m. , dickinson , m. , & kellogg , m. 1998 , phil.trans.r.s . ,
in press ( astro - ph/9805267 ) storchi - bergmann , t. , kinney , a.l . , & challis , p. 1995
, , 98 , 103 trentham , n. , sanders , d.b . , &
kormendy j. 1999 , , in press ( astro - ph/9901382 ) williams , r.e .
, blacker , b. , dickinson , m. , dixon , w.v.d . ,
ferguson , h.c .
, fruchter , a.s . , giavalisco , m. , gilliland , r.l . , heyer , i. , katsanis , r. , levay , z. , lucas , r.a . , mcelroy , d.b . , petro , l. , postman , m. , adorf , h .-
, & hook , r.n .
1996 , , 112 , 1335 witt , a.n . ,
thronson , h.a .
, capuano j.m .
1992 , , 393 , 611 wright & pettini , m. 1998 , private communication .
zepf , s.e . , & ashman , k.m .
1993 , , 264 , 611 | we refine a technique to measure the absorption corrected ultraviolet ( uv ) luminosity of starburst galaxies using rest frame uv quantities alone , and apply it to lyman - limit @xmath0-dropouts at @xmath1 found in the hubble deep field ( hdf ) .
the method is based on an observed correlation between the ratio of far infrared ( fir ) to uv fluxes with spectral slope @xmath2 ( a uv color ) .
a simple fit to this relation allows the uv flux absorbed by dust and reprocessed to the fir to be calculated , and hence the dust - free uv luminosity to be determined . _ international ultraviolet explorer _
spectra and _ infrared astronomical satellite _ fluxes of local starbursts are used to calibrate the versus @xmath2 relation in terms of @xmath3 ( the dust absorption at 1600 ) , and the transformation from broad band photometric color to @xmath2 .
both calibrations are almost completely independent of theoretical stellar population models .
we show that the recent marginal and non - detections of hdf @xmath0-dropouts at radio and sub - mm wavelengths are consistent with their assumed starburst nature , and our calculated @xmath3 .
this is also true of recent observations of the ratio of optical emission line flux to uv flux density in the brightest @xmath0-dropouts .
this latter ratio turns out _ not _ to be a good indicator of dust extinction . in @xmath0-dropouts ,
absolute magnitude @xmath4 correlates with @xmath2 : brighter galaxies are redder , as is observed to be the case for local starburst galaxies .
this suggests that a mass - metallicity relationship is already in place at @xmath5 .
the absorption - corrected uv luminosity function of @xmath0-dropouts extends up to @xmath6 abmag corresponding to a star formation rate @xmath7 ( @xmath8 , @xmath9 assumed throughout ) .
the absorption - corrected uv luminosity density at @xmath1 is @xmath10 .
it is still a lower limit since completeness corrections have not been done and because only galaxies with @xmath11 mag are blue enough in the uv to be selected as @xmath0-dropouts .
the luminosity weighted mean dust absorption factor of our sample is @xmath12 at 1600 . |
statistical - thermal models ( cf .
@xcite for recent surveys and references therein ) have enjoyed remarkable success in describing hadron multiplicities observed in both heavy - ion and elementary collisions over a wide range of energies @xcite .
the final state multiplicities are reproduced in these models with very few parameters .
the prominent ones are the chemical freeze - out temperature @xmath2 and baryon chemical potential @xmath3 .
there are further chemical potentials to be considered below .
a compilation of the freeze - out parameters over the accessible range of energies reveals a continuous curve in the @xmath4 plane which may be characterized by an energy per hadron of the order of @xmath5 @xcite . while the occurrence of such a universal freeze - out curve is interesting in itself and is useful for extrapolations and interpolations as well , its particular meaning could be the approximate coincidence with the confinement border line , at least at high energies , as conjectured in @xcite .
we do not touch here on the disputed question as to why also the multiplicities of the rare multi - strange hadrons seem to obey chemical equilibrium ( see @xcite ) . instead
, we focus on one possible indicator of incomplete equilibrium , the strangeness saturation factor @xmath6 . this factor has been introduced @xcite to account for an apparent under - saturation of strange hadrons .
various previous analyses @xcite considered such a factor as necessary to accomplish a satisfactory description of data . for sps energies , @xmath7 ,
it has been shown @xcite that @xmath6 grows with increasing beam energy and centrality .
first attempts @xcite , based on very restricted data sets , find also at rhic energies a preliminary indication of rising @xmath6 with increasing centrality .
it is the subject of this paper to investigate in detail the dependence of the thermal parameters including @xmath6 on the centrality in collisions of au+au at @xmath0 . by now there is a sufficiently large basis of published data at this energy to allow a thorough analysis .
our paper is organized as follows . in section
ii we recall the formulation of our employed statistical - thermal model .
the analysed data are listed in section iii .
our fit procedures are described in section iv and results are presented in section v. the summary and conclusions can be found in section vi .
for suitably large systems at rhic collider energies , hadron multiplicities are analysed by employing the grand - canonical partition function @xmath8 $ ] , where @xmath9 is the statistical operator of the system , @xmath2 denotes the temperature , and @xmath10 and @xmath11 are respectively the chemical potentials and corresponding conserved charges in the system . to be specific ,
we use here the baryon charge ( corresponding to @xmath3 ) and the strangeness charge ( corresponding to @xmath12 ) as unconstrained quantities .
( there are other possible chemical potentials , e.g. , @xmath13 as electric charge potential .
however , we put @xmath14 with the reasoning explained below . ) the primordial particle numbers are accordingly , @xmath15^{-1 } \mbox{bw } ( m_i ) , \label{eq.1}\ ] ] where we include phenomenologically the strangeness saturation factor @xmath6 ( with @xmath16 the total number of strange and anti - strange quarks in hadron species @xmath17 ) to account for the possibility of incomplete equilibration in this sector ; @xmath18 and @xmath19 is the breit - wigner distribution ( to be replaced by a @xmath20-function for stable hadrons ) to be integrated from thresholds with appropriate widths @xmath21 ( in practice the interval @xmath22 $ ] is sufficient ) .
the fiducial volume @xmath23 drops in particle ratios .
@xmath24 is the degeneracy factor of particle species @xmath17 with vacuum mass @xmath25 .
( the study of the effect of in - medium masses ( cf .
@xcite ) is interesting because it might shed light on the chiral properties of the medium created in heavy - ion collisions at rhic . )
( [ eq.1 ] ) simplifies in the boltzmann approximation ( i.e. , discarding the spin - statistics factor @xmath26 ) and when neglecting the energy distribution of resonances .
we use here the full expression eq .
( [ eq.1 ] ) and include all hadron states with @xmath27 quarks and corresponding anti - quarks up to 1.7 or @xmath28 with masses and total widths according to the particle data group listing @xcite .
the final particle numbers to be compared with experiment are @xmath29 due to decays of unstable particles with branching ratios @xmath30 .
originally , such a description was thought to be justified for multiplicities measured over the entire phase - space , since many dynamical effects cancel out in ratios of hadron yields @xcite . at sufficiently high energy , such as at rhic , however , ratios of mid - rapidity yields are also found to be well described by the statistical - thermal model @xcite .
in contrast to sps energies , where both mid - rapidity and fully phase space integrated data are at our disposal ( see @xcite for a comparison ) , at rhic energy of @xmath0 sufficiently many data of identified hadrons are available only at mid - rapidity .
the data analysed here were compiled from phenix , phobos , brahms and star experimental results .
these experiments differ in their cuts on the data .
it turns out that the results for the deduced thermal freeze - out parameters depend in some cases sensitively on the employed data sets .
therefore , for definiteness table [ table : data ] lists the ratios used as a function of collision centrality after recalculation to a common centrality binning as in @xcite .
( 1 ) denotes yields corrected for weak feed - down , while yields labelled by ( 2 ) include a contribution from weak decays .
three centrality bins were selected with participant numbers @xmath31 ( peripheral ) , @xmath32 ( mid - central ) and @xmath33 ( central ) .
.ratios of hadrons in @xmath34 collisions at @xmath0 from various rhic experiments . [ cols="^,^,^,^,^",options="header " , ] + as seen in table i , in some cases the same ratios are at our disposal from various experiments .
we include these multiply given ratios in our analysis as separate data points to be fit .
the global results of the fits are displayed in fig .
[ results : graph ] . as is evident from fig .
[ results : graph]a , the chemical freeze - out temperature is roughly constant at around @xmath35 in fit i. while the strangeness chemical potential is fairly flat at @xmath36 ( with 10% variation , fig .
[ results : graph]c ) , the baryon chemical potential ( fig .
[ results : graph]b ) increases slightly with participant number in the range @xmath37 to @xmath38 ( a 14% increase ) in fit i. most striking , however , and the main result of our analysis , is the increase in strangeness saturation ( fig . [ results : graph]d ) with increasing centrality for all fit types .
this confirms previous arguing @xcite . on a quantitative level
, an analysis based on too few hadron species delivers non - reliable results , in particular also for @xmath6 ( see fig . [ results : graph]d , fit iv ) . fits i and ii , in contrast , deliver consistent results .
since we consider the phenomenologically introduced parameter @xmath6 as one possible indicator of deviations from equilibrium , the conclusion is that in central collisions equilibrium conditions for describing the chemical freeze - out are appropriate , while more peripheral collisions point to some off - equilibrium effects .
, b : baryon chemical potential @xmath3 , c : strangeness chemical potential @xmath12 , d : strangeness saturation factor @xmath6 ) as extracted from the experimental data for the various fit types ( fit i : solid , fit ii : dashed , fit iii : dotted , fit iv : dash - dotted curves ) . , title="fig:",width=264 ] , b : baryon chemical potential @xmath3 , c : strangeness chemical potential @xmath12 , d : strangeness saturation factor @xmath6 ) as extracted from the experimental data for the various fit types ( fit i : solid , fit ii : dashed , fit iii : dotted , fit iv : dash - dotted curves ) . , title="fig:",width=264 ] + , b : baryon chemical potential @xmath3 , c : strangeness chemical potential @xmath12 , d : strangeness saturation factor @xmath6 ) as extracted from the experimental data for the various fit types ( fit i : solid , fit ii : dashed , fit iii : dotted , fit iv : dash - dotted curves ) .
, title="fig:",width=264 ] , b : baryon chemical potential @xmath3 , c : strangeness chemical potential @xmath12 , d : strangeness saturation factor @xmath6 ) as extracted from the experimental data for the various fit types ( fit i : solid , fit ii : dashed , fit iii : dotted , fit iv : dash - dotted curves ) .
, title="fig:",width=264 ] to highlight the role of the parameter @xmath6 on the quality of the data fits , we exhibit in fig .
[ contour ] the @xmath39 contours for fixed values of @xmath6 in the @xmath40 plane .
for definiteness , only results for fit i and peripheral collisions are shown .
we have selected @xmath41 ( representing the assumption of full equilibrium ) , 0.85 ( the optimum choice ) and 0.5 ( as rather extreme choice ) , respectively .
the corresponding @xmath42 are 24.1/12 , 17.2/11 and 62.0/12 .
the figure evidences that lowering @xmath6 noticeably shifts @xmath2 up , while the up - shift of @xmath3 is less pronounced .
the change in @xmath42 quantifies the relevance of the parameter @xmath6 for the fits . after discussing the global trends ,
it is worthwhile to look into details .
instead of displaying the usual comparison of model results with data in one plot we consider each ratio individually to exhibit the centrality systematics . , b : @xmath44 , c : @xmath45 , d : @xmath46 , e : @xmath47 , f : @xmath48 , g : @xmath49 , h : @xmath50 )
. line codes as in fig .
[ results : graph ] , experimental results : star : stars , phenix : circles , phobos : squares , brahms : triangles , cf .
table i. , title="fig:",width=264 ] , b : @xmath44 , c : @xmath45 , d : @xmath46 , e : @xmath47 , f : @xmath48 , g : @xmath49 , h : @xmath50 ) .
line codes as in fig .
[ results : graph ] , experimental results : star : stars , phenix : circles , phobos : squares , brahms : triangles , cf .
table i. , title="fig:",width=264 ] , b : @xmath44 , c : @xmath45 , d : @xmath46 , e : @xmath47 , f : @xmath48 , g : @xmath49 , h : @xmath50 ) .
line codes as in fig .
[ results : graph ] , experimental results : star : stars , phenix : circles , phobos : squares , brahms : triangles , cf .
table i. , title="fig:",width=264 ] , b : @xmath44 , c : @xmath45 , d : @xmath46 , e : @xmath47 , f : @xmath48 , g : @xmath49 , h : @xmath50 ) .
line codes as in fig .
[ results : graph ] , experimental results : star : stars , phenix : circles , phobos : squares , brahms : triangles , cf .
table i. , title="fig:",width=264 ] , b : @xmath44 , c : @xmath45 , d : @xmath46 , e : @xmath47 , f : @xmath48 , g : @xmath49 , h : @xmath50 )
. line codes as in fig . [ results : graph ] , experimental results : star : stars , phenix : circles , phobos : squares , brahms : triangles , cf . table i. , title="fig:",width=264 ] , b : @xmath44 , c : @xmath45 , d : @xmath46 , e : @xmath47 , f : @xmath48 , g : @xmath49 , h : @xmath50 )
. line codes as in fig .
[ results : graph ] , experimental results : star : stars , phenix : circles , phobos : squares , brahms : triangles , cf .
table i. , title="fig:",width=264 ] , b : @xmath44 , c : @xmath45 , d : @xmath46 , e : @xmath47 , f : @xmath48 , g : @xmath49 , h : @xmath50 ) .
line codes as in fig .
[ results : graph ] , experimental results : star : stars , phenix : circles , phobos : squares , brahms : triangles , cf .
table i. , title="fig:",width=264 ] , b : @xmath44 , c : @xmath45 , d : @xmath46 , e : @xmath47 , f : @xmath48 , g : @xmath49 , h : @xmath50 )
. line codes as in fig .
[ results : graph ] , experimental results : star : stars , phenix : circles , phobos : squares , brahms : triangles , cf .
table i. , title="fig:",width=264 ] , b : @xmath51 , c : @xmath52 , d : @xmath53 , e : @xmath54 , f : @xmath55 , g : @xmath56 , h : @xmath57 , i : @xmath58 , j : @xmath59),title="fig:",width=264 ] , b : @xmath51 , c : @xmath52 , d : @xmath53 , e : @xmath54 , f : @xmath55 , g : @xmath56 , h : @xmath57 , i : @xmath58 , j : @xmath59),title="fig:",width=264 ] , b : @xmath51 , c : @xmath52 , d : @xmath53 , e : @xmath54 , f : @xmath55 , g : @xmath56 , h : @xmath57 , i : @xmath58 , j : @xmath59),title="fig:",width=264 ] , b : @xmath51 , c : @xmath52 , d : @xmath53 , e : @xmath54 , f : @xmath55 , g : @xmath56 , h : @xmath57 , i : @xmath58 , j : @xmath59),title="fig:",width=264 ] , b : @xmath51 , c : @xmath52 , d : @xmath53 , e : @xmath54 , f : @xmath55 , g : @xmath56 , h : @xmath57 , i : @xmath58 , j : @xmath59),title="fig:",width=264 ] , b : @xmath51 , c : @xmath52 , d : @xmath53 , e : @xmath54 , f : @xmath55 , g : @xmath56 , h : @xmath57 , i : @xmath58 , j : @xmath59),title="fig:",width=264 ] , b : @xmath51 , c : @xmath52 , d : @xmath53 , e : @xmath54 , f : @xmath55 , g : @xmath56 , h : @xmath57 , i : @xmath58 , j : @xmath59),title="fig:",width=264 ] , b : @xmath51 , c : @xmath52 , d : @xmath53 , e : @xmath54 , f : @xmath55 , g : @xmath56 , h : @xmath57 , i : @xmath58 , j : @xmath59),title="fig:",width=264 ] , b : @xmath51 , c : @xmath52 , d : @xmath53 , e : @xmath54 , f : @xmath55 , g : @xmath56 , h : @xmath57 , i : @xmath58 , j : @xmath59),title="fig:",width=264 ] , b : @xmath51 , c : @xmath52 , d : @xmath53 , e : @xmath54 , f : @xmath55 , g : @xmath56 , h : @xmath57 , i : @xmath58 , j : @xmath59),title="fig:",width=264 ] for a qualitative understanding one can use the boltzmann approximation ignoring widths .
( at rhic energies the use of boltzmann statistics introduces errors in primordial yields at the level of 10% for pions , 1.3 - 1.5% for kaons and less than one percent for all other hadrons .
thus , given the intrinsic systematic uncertainties inherent in any thermal model analysis owing to the poorly constrained particle properties of the heavy resonances , quantum statistics is important only for pions . ) the primordial density of hadron species @xmath17 is then given by @xmath60 thus , the ratio of particle to anti - particle primordial yields is simply , @xmath61 anti particle - to - particle ratios therefore fix @xmath62 .
we note that these ratios are only slightly affected by feed - down ( using the central fit i best - fit parameters the percentage influence of feed - down is @xmath63 : 1.0% , @xmath44 : 2.9% , @xmath45 : 0.5% , @xmath46 : 3.7% , @xmath47 : 0.7% , @xmath48 : 4.2% , @xmath49 : 0.08% , and @xmath64 : 0% ) .
thus , the trends in these ratios translate directly into trends in @xmath65 and @xmath66 , while they are roughly insensitive to @xmath1 and @xmath2 .
let us now consider the individual ratios and begin with @xmath67 .
the experimental ratio is consistent with being flat and slightly below or equal to unity , while our model yields a @xmath68 ratio slightly above 1 ( see fig .
[ results : modelpap]a ) . model and experiment
, however , agree within errors .
there is no significant difference between the results of the four fits . setting
@xmath69 drives the model ratio to very close to 1 ; deviations are due to feeding ( since @xmath3 and @xmath12 are both greater than zero ) . with only strong decays
included , the ratio is less than 1 .
therefore it is the influence of the weak decays that pushes the ratio above 1 .
this ratio is insensitive to @xmath12 and @xmath3 : with a @xmath70 change in @xmath3 the ratio varies by 0.2% , while a @xmath70 change in @xmath12 leads to a 0.3% change in the ratio .
a 0.2 change in @xmath1 furthermore affects the ratio by only 0.2% , while a @xmath70 change in @xmath2 leads to only a 0.06% change in the ratio .
therefore , this ratio fixes nothing , and serves just as the motivation for setting @xmath69 .
the ratio @xmath71 is above 1 ( see fig . [ results : modelpap]b ) and drives @xmath12 to positive values . due to their much smaller errors , the star ratios are heavily weighted in the fit .
this contributes to @xmath12 having a slight kink as seen in fig .
[ results : graph]c .
the effect of the above - quoted feeding accounts for the slightly different shape of the @xmath72 ratio compared with the @xmath65 trend which can be deduced from fig .
[ results : graph ] . with this ratio removed from the fit
, @xmath65 would increase steeply with centrality , driven by the corrected @xmath73 to @xmath74 double ratio .
this gives @xmath74 an increasing trend . in order to keep the @xmath73 ratio flat ,
@xmath66 then similarly increases which causes @xmath75 to drop .
therefore , the kaon ratio is what keeps @xmath12 flattish and gives it the ` kink ' .
the @xmath75 experimental ratios differ noticeably ( see fig . [
results : modelpap]c and fig .
[ results : modelpap]d ) , with star being the highest .
these star data points are favored in fits since they have much reduced errors ( consistently a factor of 3 less than those of phenix ) .
these star ratios , however , are too high if the star @xmath73 ratio is to be simultaneously reproduced by the model ( star @xmath75 supports a lower @xmath3 as is seen when the @xmath76 s are removed from the fit , as done in fit iv ) . with @xmath76 s excluded ,
the much smaller star errors lead to these @xmath75 data points being strongly favored in the mid - central and peripheral fits .
this leads to a strong increase of @xmath3 with centrality .
since the experimental @xmath73 ratio ( fig . [ results : modelpap]e and f ) is flattish , they support a similar behavior for @xmath66 and @xmath65 .
excluding all but the star @xmath75 points in the most central bin removes the strong increase in @xmath3 from the mid - central to central bins seen in fit iv . our suspicion that it is @xmath75 which causes @xmath3 to increase
is confirmed by repeating the fits with only anti
particle - to - particle ratios included ( i.e. , all particle - to - pion ratios excluded ) . since these ratios are insensitive to @xmath1 , and @xmath2 enters only together with the chemical potentials , in these reduced fits we fixed @xmath77 and @xmath78 ( both reasonable in light of the full results ) .
the approximate trends in @xmath65 and @xmath66 observed in the full fits were reproduced using just these ratios . with
the @xmath75 ratio excluded , the steady increase in @xmath3 disappears ; instead @xmath66 decreases similarly to @xmath65 as driven by the flatness in the @xmath73 ratio . the ratio @xmath79 ( see fig . [
results : modelpap]g ) depends strongly on @xmath65 and @xmath66 and is little affected by feed - down .
similar fractional errors to the uncorrected @xmath73 ratio means both are treated equally in fits .
although consistent within errors , the model trend decreases slightly with centrality , while the data are consistent with a weak increase . comparing this experimental ratio with @xmath73 , the @xmath74 to @xmath73 double ratio increases
thus , @xmath65 is driven up particularly from the mid - central to central bin .
as mentioned earlier , the @xmath72 ratio leads to a kink in the trend for @xmath12 .
this effect wins over the influence of the @xmath80 to @xmath76 double ratio which suggests an increase in @xmath65 .
unfortunately there is no centrality information on the ratio @xmath81 .
such data would allow even better determination of @xmath12 .
as seen in fig .
[ results : modelpap]h , our model results agree fairly well with the data for the central bin .
the particle - to - pion ratios read in the boltzmann approximation neglecting width @xmath82 i.e. , qualitatively , mixed particle ratios allow the determination of @xmath1 and @xmath2 , and hence the @xmath83 s independently . whereas the particle - to - anti particle ratios feature just the combination @xmath84
, the primordial particle - to - pion ratios feature also @xmath2 in the combination @xmath85 .
this allows the temperature to be fixed and then , given the values of @xmath84 fixed by the particle - to - anti - particle ratios , the chemical potentials are determined .
however , feed - down plays an important role in these ratios , so that the final observed ratios differ from the ratios of primordial densities quite substantially .
this makes a detailed analysis of the influence of these ratios on the parameters much more difficult than in the case of the particle - to - anti particle ratios .
in fact , for the best - fit parameters corresponding to fit i of the central bin , more than 70% of @xmath86 s come from decays .
a simple analysis shows how the ratio @xmath85 varies with temperature for a number of hadron species @xmath17 : lowering the temperature has the effect of bringing the particle - to - pion ratios down .
furthermore , the fractional effect of changing the temperature is greater for the heavier particles . after these general remarks let us consider in some detail the comparison of our model with data .
the ratio @xmath87 ( fig . [ results : modelppi]a ) is reasonably well reproduced in the fits . since the errors on the phenix and star experimental ratios are similar , neither is heavily biased in the fit
. the phenix data rise more rapidly than the star data ( especially from the mid - central to central bin ) .
since the particle - to - anti - particle ratios support a centrality - independent @xmath65 , the experimental @xmath88 ratios drive an increase in strangeness saturation , with the phenix data supporting a steeper increase in @xmath1 than the star data ( see fig . [ results : graph]d ) . in fit
i and ii @xmath88 is over - predicted .
in contrast , the uncorrected @xmath89 and @xmath90 are under - estimated ( small fractional errors on @xmath90 mean that this ratio strongly influences the fit ) . with the @xmath80 s removed in fit iii
the star @xmath88 data are well reproduced owing to the drop in @xmath2 and @xmath1 .
the big drop in this ratio when @xmath76 s are excluded from the fit is due to the further drop in @xmath1 and @xmath2 in fit iv seen in fig .
[ results : graph ] .
the effect of this drop is somewhat lessened by the accompanying drop in @xmath65 in the central and mid - central bins .
furthermore , the kaon to pion ratios are least affected by the temperature change in fit iv . the increase in the ratio @xmath91 with centrality ( fig .
[ results : modelppi]b ) also drives @xmath1 to increase . in the full fits ( fits i - iii ) it is somewhat over - estimated ( a similar result is found in @xcite ) , while the exclusion of the @xmath76 s ( fit iv ) results in much better agreement .
this is owing to the reduced @xmath1 in fit iv . comparing the results for the ratios @xmath92 and @xmath93 in figs .
[ results : modelppi]a and b one gets the impression that the phenix @xmath92 data is compatible with star s @xmath93 data , while the good fit in fit iii of star s @xmath92 data result in a ratio @xmath93 being at the upper limits of the error bars from star . the central @xmath94 ratio reproduction by the model ( fig .
[ results : modelppi]g ) follows the same trend as the @xmath95 .
let us now focus on the @xmath96 and @xmath97 ratios ( figs .
[ results : modelppi]c , d , e and f ) . while the @xmath98 ratio is over - estimated by the model in fits i - iii , @xmath89 is under - estimated except for the corrected @xmath89 in the central bin in fits i and ii . with @xmath76
excluded , @xmath98 is well reproduced , while @xmath89 is greatly underpredicted .
therefore , @xmath98 drives @xmath2 down , while @xmath89 tends to raise @xmath1 and @xmath2 . the flatness of the ratio @xmath99 ( fig .
[ results : modelppi]h ) supports a centrality - independent @xmath1 . in fits
i and ii there is fair agreement with the model , while this ratio is greatly under - estimated in fit iv owing to the greatly reduced @xmath1 and @xmath2 .
the increasing trend of the model results is due to the increase in @xmath1 with centrality and the fact that @xmath100 .
the increase of the ratio @xmath101 ( fig .
[ results : modelppi]i ) drives @xmath66 to increase as well as @xmath1 . on the other hand , it causes @xmath65 to drop .
again there is fair agreement with the model in fits i and ii , while this ratio is greatly under - estimated in fit iv .
our final conclusion is that fits i - ii give a quite satisfactory description of most ratios , with the exception of @xmath93 , @xmath94 and corrected @xmath96 .
for the @xmath94 and @xmath103 there are no data for mid - central and peripheral collisions .
thus our results may serve as predictions . in summary
, we present an analysis of the centrality dependence of thermal parameters deduced from hadron multiplicities within a statistical - thermal model .
the grand - canonical formalism is employed with quantum statistics and resonance widths included for all of the fits .
we rely entirely on published data for the reaction au + au at @xmath0 .
our aim is to employ the strangeness saturation factor @xmath6 as one possible indicator for deviations from equilibrium conditions in describing the chemical freeze - out . indeed , for non - central collisions
the systematic deviation of @xmath6 from unity can be considered as a hint to off - equilibrium effects , while for central collisions and using a sufficiently complete set of available multiplicity ratios at mid - rapidity there is no indication for deviations from equilibrium conditions .
we present a detailed comparison of the model results with data .
the trends in @xmath104 are fixed by the anti particle - to - particle ratios .
the @xmath72 ratio is responsible for keeping @xmath105 essentially flat .
this wins over the influence of the @xmath80 to @xmath76 double ratio which tends to increase @xmath65 .
the @xmath75 ratio drives @xmath66 up , which wins over @xmath73 s attempts to make @xmath65 and @xmath66 behave similarly .
the temperature and strangeness saturation are fixed by the particle - to - pion ratios . the observed increase in @xmath1
is driven predominantly by the increase in the @xmath88 ratio and the flatness of @xmath65 as fixed by the @xmath72 ratio .
a possible extension of the present work is the analysis of saturation in the non - strange sector , parameterized by @xmath106 being a factor similar to @xmath6 @xcite . since our fits to the data
deliver reasonably small values of @xmath39 we believe , however , that there is not too much room left for sizeable deviations of @xmath106 from unity .
this issue deserves a separate investigation .
more tempting is the analysis of data for @xmath107 .
the explorative study in @xcite points to a similar behavior of @xmath6 as we find for @xmath0 .
even more interesting is to test heavy flavor production ; due to their larger intrinsic masses , heavy flavors should have a higher sensitivity to the medium .
in particular , hadrons with open and hidden charm may require a corresponding factor @xmath108 .
99 p. braun - munzinger , k. redlich , j. stachel , nucl - th/0304013 in _ quark - gluon plasma 3 _ , ( eds . ) r. hwa , x.n .
wang ; + k. redlich , j. cleymans , h. oeschler and a. tounsi , acta phys .
polonica b 33 ( 2002 ) 1609 ; + j. letessier , j. rafelski , int . j. mod
e 9 ( 2000 ) 107 .
f. becattini , j. cleymans , a. kernen , e. suhonen , k. redlich , phys .
c 64 ( 2001 ) 024901 .
p. braun - munzinger et al . , phys .
b 344 ( 1995 ) 43 , b 365 ( 1996 ) 1 , b 465 ( 1999 ) 15 , b 518 ( 2001 ) 41 .
j. cleymans , k. redlich , phys .
81 ( 1998 ) 5284 , + j. sollfrank , j. phys .
g : nucl . part .
( 1997 ) 1903 .
j. cleymans , k. redlich , phys .
c 60 ( 1999 ) 054908 .
w. florkowski , w. broniowski , phys .
c 65 ( 2002 ) 064905 , phys .
87 ( 2001 ) 272302 ; + f. becattini , g. passaleva , eur .
j. c 23 ( 2002 ) 551 .
p. braun - munzinger , j. stachel , c. wetterich , phys .
b 596 ( 2004 ) 61 .
t. biro , b. mller , phys .
b 578 ( 2004 ) 78 ; + a. majumder , v. koch , phys . rev .
c 68 ( 2003 ) 044903 .
p.koch , b. mller , j. rafelski , phys .
142 ( 1986 ) 167 ; + j. letessier , j. rafelski , a. tounsi , phys .
c 50 ( 1994 ) 405 ; + c. slotta , j. sollfrank , u. heinz , aip conf . proc .
( woodbury ) 340 ( 1955 ) 462 .
f. becattini , m. gazdzicki , a. kernen , j. mannien , r. stock , phys .
c 69 ( 2004 ) 024905 ; + i.g .
bearden et al .
( na44 ) , nucl - ex/0202019 ; + f. becattini , m. gazdzicki , j. sollfrank , eur .
j. c 5 ( 1998 ) 143 .
j. cleymans , b. kmpfer , s. wheaton , phys .
c 65 ( 2002 ) 027901 .
j. cleymans , b. kmpfer , s. wheaton , nucl .
phys . a 715 ( 2003 ) 553c ; + j. cleymans , b. kmpfer , p. steinberg , s. wheaton , hep - ph/0212335 , j. phys .
g 30 ( 2004 ) s595 ; + b. kmpfer , j. cleymans , k. gallmeister , s. wheaton , hep - ph/0202134 , acta .
phys . hung .
. heavy ion phys . 18 ( 2003 ) 1 .
d. zschiesche , s. schramm , j. schaffner - bielich , h. stcker , w. greiner , phys .
b 552 ( 2002 ) 7 .
k. hagiwara et al .
d 66 ( 2002 ) 010001 .
m. kaneta , n. xu , nucl - th/0405068 . f. becanttini , j. phys .
g 28 ( 2002 ) 1553 ; + d. magestro , j. phys .
g 28 ( 2002 ) 1745 ; + a. baran , w. broniowski , w. florkowski , acta phys .
b 35 ( 2004 ) 779 .
j. adams et al .
( star ) , nucl - ex/0311017 ; + m. calderon de la barca sanchez , nucl - ex/0111004 . c. adler et al .
( star ) , phys .
( 2002 ) 092301 .
bearden et al .
( brahms ) , phys .
87 ( 2001 ) 112305 .
k. adcox et al .
( phenix ) , phys .
88 ( 2002 ) 242301 .
k. adcox et al .
( phenix ) , phys .
( 2002 ) 092302 .
back et al . (
phobos ) , phys .
87 ( 2001 ) 102301 . c. adler et al .
( star ) , phys .
b 595 ( 2004 ) 143 . c. adler et al . ( star ) , phys . rev .
c 66 ( 2002 ) 061901 . c. adler et al .
( star ) , phys .
87 ( 2001 ) 262302 . c. adler et al .
( star ) , phys .
86 ( 2001 ) 4778 . c. adler et al .
( star ) , phys .
c 65 ( 2002 ) 041901(r ) . j. castillo et al .
( star ) , nucl .
a 715 ( 2003 ) 518c .
j. adams et al .
( star ) , phys .
92 ( 2004 ) 182301 . c. suire
et al . ( star ) , nucl .
phys . a 715 ( 2003 ) 470c .
j. lettesier , j. rafelski , phys .
c 59 ( 1999 ) 947 . | we analyse the centrality dependence of thermal parameters deduced from hadron multiplicities in au + au collisions at @xmath0 .
while the chemical freeze - out temperature and chemical potentials are found to be roughly centrality - independent , the strangeness saturation factor @xmath1 increases with participant number towards unity , supporting the assumption of equilibrium freeze - out conditions in central collisions . |
the equilibrium problem ( ep ) @xcite which was considered as the ky fan inequality @xcite is very general in the sense that it includes , as special cases , many mathematical models such as : variational inequalities , fixed point problems , optimization problems , nash equilirium point problems , complementarity problems , see @xcite and the references therein .
many methods have been proposed for solving eps @xcite .
the most solution approximations to eps are often based on the resolvent of equilibrium bifunction ( see , for instance @xcite ) in which a strongly monotone regularization equilibrium problem ( rep ) is solved at each iterative step .
it is also called the proximal point method ( ppm ) .
this method was first introduced by martinet @xcite for variational inequalities , and then it was extended by rockafellar @xcite for finding a zero point of a monotone operator . in 2000 , konnov @xcite further extended ppm to ky fan inequalities for monotone or weakly monotone bifunctions . a special case of ep is the variational inequality problem ( vip ) .
the projection plays an important role in constrained optimization problems .
the simplest method for vips is the gradient projection method in which only a projection on the feasible set is computed .
however , in order to obtain the convergence , the method requires the restrictive assumption that operators are strongly ( or inverse strongly ) monotone .
to overcome this , korpelevich @xcite introduced the extragradient method ( double projection method ) where two metric projections onto the feasible set be implemented at each iteration .
the convergence of the extragradient method was proved under the weaker assumption that operators are only monotone ( even , pseudomonotone ) and @xmath1 - lipschitz continuous . some extragradient - like algorithms proposed for solving vips
can be found in @xcite and the references therein .
however , the projection is only found easily if the constrained set has a simple structure , for instance , as balls , hyperplanes or halfspaces .
so , several modifications of the extragradient method have been proposed in various ways @xcite .
for instance , the authors in @xcite replaced the second projection onto the feasible set in the extragradient method by one onto a half - space and proposed the subgradient extragradient method for vips in hilbert spaces .
in recent years , korpelevich s extragradient method has been naturally extended to eps for monotone ( more general , pseudomonotone ) and lipschitz - type continuous bifunctions and widely studied both theoretically and algorithmically @xcite . in the extended extragradient methods to eps
, we need to solve two strongly convex optimization programs on a closed convex constrained set ( see , algorithms [ vsh2013a ] , [ vsh2013b ] and [ dhm2014 ] in section [ algor ] ) .
they are generalizations of two projections in korpelevich s extragradient method .
the advantage of the extragradient method is that two optimization programs are solved at each iteration which seems to be numerically easier than the non - linear inequality ( or rep ) in ppm .
however , this might still be costly and affects the efficiency of the used method if the structure of feasible set and equilibrium bifunction are complex
. moreover , we are not aware of any modification of the extragradient method for eps . in this paper , motivated by the _
hybrid method without the extrapolation step _
@xcite for variational inequalities , the _ extragradient method _ @xcite and the _ hybrid method _
, we have proposed a new hybrid algorithm for solving eps . in this algorithm , by constructing a specially cutting - halfspace in the hybrid method , we only need to solve a strongly convex optimization program onto the feasible set at each iteration .
the absence of an optimization program in our algorithm ( compare with the extragradient method ) can be considered an improvement of the results in @xcite .
the remainder of the paper is organized as follows : section @xmath2 introduce our algorithm and some related works . in section @xmath3 , we collect some definitions and preliminary results used in the paper . section @xmath4 deals with proving the convergence of the algorithm .
some applications of our algorithm to gato differentiable eps and multivalued variational inequalities are presented in section [ appl ] . finally , in section @xmath5 we provide some numerical examples to illustrate the convergence of the proposed algorithm and compare it with others .
let @xmath6 be a real hilbert space , @xmath7 be a nonempty closed convex subset of @xmath6 and @xmath8 be a bifunction with @xmath9 for all @xmath10 .
the equilibrium problem ( ep ) for the bifunction @xmath11 on @xmath7 is to find @xmath12 such that @xmath13 the solution set of ep @xmath14 is denoted by @xmath15 . in this paper , we introduce the following hybrid algorithm for solving ep @xmath14 . [ h2015 ] @xmath16 where @xmath17 are two specially constructed half - spaces ( see algorithm @xmath18 in section @xmath4 below ) . in the special case ,
@xmath19 where @xmath20 is a nonlinear operator then ep becomes the following variational inequality problem ( vip ) : find @xmath12 such that @xmath21 then , our algorithm ( algorithm @xmath22 ) becomes the following _ hybrid algorithm without extrapolation step _ which was introduced in @xcite for vips .
[ ms2015 ] @xmath23 in 2008 , quoc et al .
@xcite extended korpelevich s extragradient method to eps in euclidean spaces in which two optimization programs are solved at each iteration .
recently , nguyen et al .
@xcite also have done in that direction and proposed the general extragradient method which consists of solving three optimization programs on the feasible set . in euclidean spaces , the convergence of the sequences generated by the extragradient methods @xcite was proved under the assumptions of pseudomonotonicity and lipschitz - type continuity of equilibrium bifunctions .
the problem which arises in infinite dimensional hilbert spaces is how to design an algorithm which provides the strong convergence . in 2012 , vuong et al .
@xcite used the extragradient method in @xcite and the hybrid ( outer approximation ) method to obtain the following strong convergence hybrid algorithm [ vsh2013a ] @xmath24 in 2013 , another hybrid algorithm ( * ? ? ?
* algorithm 1 ) was also proposed in this direction as [ vsh2013b ] @xmath25 the authors in @xcite proved that the sequences @xmath26 generated by algorithms @xmath27 and @xmath28 converges strongly to @xmath29 . note that , the set @xmath30 in algorithm @xmath28 , in general , is not easy to construct . in 2014 , in order to avoid the condition of the lipschitz - type continuity the bifunction @xmath11 , dinh et al .
@xcite replaced the second optimization problem in the extragradient method by the armijo linesearch technique and obtained the following hybrid algorithm [ dhm2014 ] @xmath31 where @xmath32 and @xmath33 .
arcording to algorithm @xmath34 , we still have to solve an optimization program on @xmath7 for @xmath35 , find an optimization direction for @xmath36 and compute a projection onto @xmath7 for @xmath37 at each step .
we emphasize that the projection @xmath38 in algorithms @xmath27 , @xmath28 and @xmath34 deals with the constrained set @xmath7 , while the sets @xmath39 and @xmath40 in algorithm @xmath22 are two half - spaces , and so @xmath41 can be expressed by an explicit formula ( see , for instance @xcite ) .
in this section , we recall some definitions and results for further use .
let @xmath7 be a nonempty closed convex subset of a real hilbert space @xmath6 .
we begin with some concepts of the monotonicity of a bifunction ( see @xcite for more details ) .
a bifunction @xmath8 is said to be * strongly monotone on @xmath7 if there exists a constant @xmath42 such that @xmath43 * monotone on @xmath7 if @xmath44 * pseudomonotone on @xmath7 if @xmath45 * lipschitz - type continuous on @xmath7 if there exist two positive constants @xmath46 such that @xmath47 from the definitions above , it is clear that a strongly monotone bifunction is monotone and a monotone bifunction is pseudomonotone , i.e. , @xmath48 for solving ep @xmath14 , we assume that the bifunction @xmath11 satisfies the following conditions : * @xmath11 is pseudomonotone on @xmath7 and @xmath9 for all @xmath10 ; * @xmath11 is lipschitz - type continuous on @xmath7 with the constants @xmath46 ; * @xmath11 is weakly continuous on @xmath49 ; * @xmath50 is convex and subdifferentiable on @xmath7 for every fixed @xmath51 it is easy to show that under the assumptions @xmath52 , the solution set @xmath15 of ep @xmath14 is closed and convex ( see , for instance @xcite ) . in this paper , we assume that the solution set @xmath15 is nonempty
. the metric projection @xmath53 is defined by @xmath54 since @xmath7 is nonempty , closed and convex , @xmath55 exists and is unique .
it is also known that @xmath56 has the following characteristic properties , see @xcite for more details .
[ lem.propertypc ] let @xmath53 be the metric projection from @xmath6 onto @xmath7 .
then * @xmath56 is firmly nonexpansive , i.e. , @xmath57 * for all @xmath58 , @xmath59 * @xmath60 if and only if @xmath61 let @xmath62 be a function . the subdifferential of @xmath63 at @xmath64 is defined by @xmath65 we recall that the normal cone of @xmath7 at @xmath10 is defined by @xmath66 a function @xmath67 is called weakly lower semicontinuous at @xmath68 if for any sequence @xmath26 in @xmath6 converges weakly to @xmath64 then @xmath69 it is well - known that the functional @xmath70 is convex and weakly lower semicontinuous .
any hilbert space has the kadec - klee property ( see , for instance @xcite ) , i.e. , if @xmath26 is a sequence in @xmath6 such that @xmath71 and @xmath72 then @xmath73 as @xmath74 .
finally , we have the following technical lemma . @xcite[lem.technique ] let @xmath75 , @xmath76 , @xmath77 be nonnegative real sequences , @xmath78 and for all @xmath79 the following inequality holds @xmath80 if @xmath81 and @xmath82 then @xmath83 .
in this section , we present our algorithm for more details and prove its convergence .
[ algor1 ] * initialization . * chose @xmath84 and set @xmath85 .
the parameters @xmath86 and @xmath87 satisfy the following conditions @xmath88 * step 1 . *
solve a strongly convex program @xmath89 if @xmath90 then stop .
+ * step 2 . * compute @xmath91 where @xmath92 and @xmath93 .
set @xmath94 and go back * step 1 .
* we have the following result . [ lem2 ]
if algorithm @xmath18 finishes at the iteration step @xmath95 , then @xmath96 .
assume that @xmath90 .
from the definition of @xmath97 , @xmath98 thus , from ( * ? ? ? * proposition 2.1 ) , one has @xmath96 .
the proof of lemma @xmath99 is complete .
we need the lemma below which is an infinite version of theorem 27.4 in @xcite and is similarly proved by using moreau - rockafellar theorem to find the subdifferential of a sum of a convex function @xmath63 and the indicator function @xmath100 to @xmath7 in a real hilbert space .
[ lem.equivalent_minpro ] let @xmath7 be a convex subset of a real hilbert space h and @xmath62 be a convex and subdifferentiable function on @xmath7 .
then , @xmath101 is a solution to the following convex optimization problem @xmath102 if and only if @xmath103 , where @xmath104 denotes the subdifferential of @xmath63 and @xmath105 is the normal cone of @xmath7 at @xmath101 .
based on lemma @xmath106 , we obtain the following central lemma which is used to prove the convergence of algorithm @xmath18 .
[ lem1 ] assume that @xmath107 .
let @xmath108 be the sequences generated by algorithm @xmath18 .
then , there holds the relation @xmath109 where @xmath110 is defined by step 2 of algorithm @xmath18 . from the definition of @xmath97 and lemma @xmath106 , @xmath111 thus , there exist @xmath112 and @xmath113 such that @xmath114 hence , @xmath115 this together with the definition of @xmath116 implies that @xmath117 by @xmath118 , @xmath119 from the last two inequalities , we obtain @xmath120 similarly , by replacing @xmath121 by @xmath122 , we also have @xmath123 substituting @xmath124 onto @xmath125 and a straightforward computation yield @xmath126 substituting @xmath127 onto @xmath128 we also obtain @xmath129 since @xmath107 and @xmath130 , @xmath131 .
thus , from the pseudomonotonicity of @xmath11 one has @xmath132 .
this together with @xmath133 implies that @xmath134 by the lipschitz - type continuity of @xmath11 , @xmath135 thus , @xmath136 the relations @xmath137 and @xmath138 lead to @xmath139 this together with the relation @xmath140 implies that @xmath141 thus , @xmath142 we have the following fact @xmath143 by the triangle , cauchy - schwarz and cauchy inequalities , @xmath144 this together with @xmath145 implies that @xmath146 thus , @xmath147 combining @xmath148 and @xmath149 we obtain @xmath150 thus , from the definition of @xmath110 we obtain @xmath151 lemma @xmath152 is proved .
[ lem4 ] let @xmath108 be the sequences generated by algorithm @xmath18 .
then , there hold the following relations * @xmath153 for all @xmath79 . * @xmath154 ( i ) . from the definitions of @xmath39 and @xmath40
, we see that they are the half - spaces .
thus , @xmath39 and @xmath40 are closed and convex for all @xmath79 .
lemma @xmath152 and the definition of @xmath39 ensure that @xmath155 for all @xmath79 .
it is clear that @xmath156 .
assume that @xmath153 for some @xmath79 .
from @xmath157 and lemma @xmath158(iii ) we see that @xmath159 for all @xmath160 .
this is also true for all @xmath161 because @xmath153 . from the definition of @xmath162 , @xmath163 or @xmath164 . by the induction , @xmath153 for all @xmath79 .
since @xmath15 is nonempty , so @xmath165 is .
thus , @xmath38 is well - defined .
+ ( ii ) . from the definition of @xmath40 and lemma @xmath158(iii . ) , @xmath166 . thus , from lemma @xmath158(ii ) we have @xmath167 onto @xmath168 , one has @xmath169 , therefore @xmath26 , are bounded .
substituting @xmath170 onto @xmath168 , one also has @xmath171 this implies that @xmath172 is non - decreasing .
hence , there exists the limit of @xmath172 .
by @xmath173 , @xmath174 passing the limit in the last inequality as @xmath175 , we obtain @xmath176 thus , @xmath177 from the definition of @xmath39 and @xmath178 , @xmath179 set @xmath180 , @xmath181 , @xmath182 , @xmath183 , and @xmath184 . from the definition of @xmath110 , @xmath185 .
thus , from @xmath186 , @xmath187 from the hypothesises of @xmath188 and @xmath189 , we see that @xmath82 and @xmath190 .
lemma @xmath191 and @xmath192 imply that @xmath193 , or @xmath194 this together with the relation @xmath195 and the inequality @xmath196 implies that @xmath197 in addition , the sequence @xmath198 is also bounded because of the boundedness of @xmath26 .
lemma @xmath199 is proved .
+ thanks to lemma @xmath99 , we see that if algorithm @xmath18 terminates at the iterate @xmath122 then a solution of ep can be found .
otherwise , if algorithm @xmath18 does not terminate then we have the following main result .
[ theo.1 ] let @xmath7 be a nonempty closed convex subset of a real hilbert space @xmath6 .
assume that the bifunction @xmath11 satisfies all conditions @xmath200 .
in addition the solution set @xmath15 is nonempty .
then , the sequences @xmath26 , @xmath201 generated by algorithm @xmath18 converge strongly to @xmath29 . from lemma @xmath199 , the sequence @xmath26 is bounded .
assume that @xmath202 is any weak cluster point of @xmath26 .
without loss of generality , we can write @xmath203 as @xmath204 . thus , @xmath205 because @xmath206 .
now , we show that @xmath207 . from @xmath128 , we get @xmath208 passing the limit in @xmath209 as @xmath74 and using lemma @xmath199(ii ) , the bounedness of @xmath201 and @xmath210 we obtain @xmath211 for all @xmath212 .
thus , @xmath207 . from the inequality @xmath213 , we get @xmath214 where @xmath215 . by the weak lower semicontinuity of the norm @xmath216 and @xmath203
, we have @xmath217 by the definition of @xmath218 , @xmath219 and @xmath220 .
thus , @xmath221 . by the kadec - klee property of the hilbert space @xmath6 , we have @xmath222 as @xmath74 . from lemma @xmath199
, we also see that @xmath201 converges strongly to @xmath223 .
theorem @xmath224 is proved .
in this section , we introduce several applications of algorithm [ algor1 ] to gato differentiable eps and multivalued variational inequalities .
we consider eps for gato differentiable bifunctions .
we denote @xmath225 by the gato derivative of the function @xmath50 at @xmath226 . for solving ep @xmath14
, we assume that the bifunction @xmath11 satisfies the following conditions : * @xmath11 is monotone on @xmath7 and @xmath9 for all @xmath10 ; * @xmath50 is convex and gato differentiable on @xmath7 ; * there exists a constant @xmath227 such that @xmath228 * @xmath229 for all @xmath230 .
if ep @xmath14 is reduced to vip @xmath231 for the operator @xmath20 then the condition @xmath232 is equivalent to the lipschitzianity of @xmath233 with the constant @xmath227 .
we need the following results .
* lemma 2)[lem3 ] suppose that the conditions @xmath234 hold .
then , * the operator @xmath235 is monotone on @xmath7 . * @xmath236 .
[ lem4 ] assume that @xmath20 is a l - lipschitz continuous operator and the bifunction @xmath8 is defined by @xmath19 for all @xmath230 .
then * @xmath11 is lipschitz - type continuous on @xmath7 with @xmath237 . *
@xmath238 if and only if @xmath239 , where @xmath210 .
\i . from the @xmath1 - lipschitz continuity of @xmath233 , the cauchy - schwarz and cauchy inequalities
, we have @xmath240 this implies that @xmath11 is lipschitz - type continuous on @xmath7 with @xmath237 .
+ ii . from the definition of @xmath11
, we have @xmath241 in which the third equality is followed from the fact that @xmath242 and the last equality is true because of the definition of the metric projection .
lemma [ lem4 ] is proved .
thanks to lemma [ lem3 ] , instead of ep ( [ eq : ep ] ) we solve vip ( [ vip ] ) for the operator @xmath235 onto @xmath7 . it is emphasized that @xmath243 and @xmath232 are slightly strong conditions .
however , in this case , we can use the existing methods for vips to solve eps .
for instance , using the subgradient extragradient method ( * ? ? ?
* algorithm 3.6 ) we obtain the following hybrid algorithm for solving ep ( [ eq : ep ] ) @xmath244 where @xmath245 .
if the conditions @xmath246 hold for all @xmath247 then @xmath26 generated by ( [ cgr2011 ] ) converges strongly to @xmath29 . in this subsection , we introduce the following strong convergence result .
[ theo3 ] let @xmath7 be a nonempty closed convex subset of a real hilbert space @xmath6 .
assume that the bifunction @xmath11 satisfies all conditions @xmath246 such that @xmath15 is nonempty .
let @xmath26 be the sequence generated by the following manner : @xmath248 and @xmath249 where @xmath250 are defined as in algorithm [ algor1 ] with @xmath237 .
then , the sequence @xmath26 converges strongly to @xmath29 .
set @xmath251 for all @xmath230 , where @xmath235 .
lemma @xmath252.ii .
ensures that @xmath253 .
the bifunction @xmath254 satisfies the conditions @xmath255 and @xmath256 automatically . from lemma [ lem3].i .
, we see that @xmath254 is monotone , and so it is also pseudomonotone or @xmath254 satisfies the condition @xmath257 .
lemma [ lem4].i . and @xmath232 ensure that the condition @xmath258 holds for the bifunction @xmath254 . from step 1 of algorithm @xmath18 and lemma [ lem4].ii .
, @xmath259 .
thus , theorem [ theo3 ] is directly followed from theorem [ theo.1 ] for @xmath260 . in this subsection
, we consider the following multivalued variational inequality problem ( mvip ) @xmath261 where @xmath262 is a multivalued compact operator . for a pair @xmath230 , we put @xmath263 it is easy to show that @xmath101 is a solution of mvip ( [ mvip ] ) if and only if @xmath101 is a solution of ep for the bifunction @xmath11 on @xmath7 .
we recall the following definitions . a multivalued operator @xmath262 is said to be : * monotone on @xmath7 if @xmath264 * pseudomonotone on @xmath7 if @xmath265 * @xmath1 - lipschitz continuous if there exists a positive constant @xmath1 such that @xmath266 if we denote @xmath267 by the hausdorff distance between two sets @xmath268 and @xmath269 then the definition iii .
means that @xmath270 we can easily check that if @xmath233 is pseudomonotone and @xmath1 - lipschitz continuous then @xmath11 is also pseudomonotone and lipschitz - type continuous with two constants @xmath237 .
note that , when @xmath233 is singlevalued then algorithm @xmath18 becomes the _ hybrid algorithm without the extrapolation step _ for variational inequalities @xcite .
when @xmath233 is multivalued then algorithm @xmath18 can be applied for the bifunction @xmath11 defined by ( [ eq : f ] ) .
a disadvantage of performing algorithm @xmath18 in this case is that it is not easy to chose an approximation of the bifunction @xmath271 .
in fact , we can prove the strong convergence of the following algorithm @xmath272 where @xmath250 are defined as in algorithm [ algor1 ] with @xmath237 .
in this section , we consider two previously known academic numberical examples in euclidean spaces .
the purpose of these experiments is to illustrate the convergence of algorithm @xmath18 and compare its efficiency with algorithms @xmath27 , @xmath28 and @xmath34 . of course , there are many mathematical models for eps in infinite dimensional hilbert spaces , see , for instance @xcite and the norm convergence of algorithms is more necessary than the weak convergence . the ability of the implementation of these algorithms has been discussed in sections @xmath273 and [ algor ] .
note that algorithm @xmath28 , in general , is difficult to compute numerical experiments because of the complexity of the sets @xmath39 .
however , in the examples below , the feasible set @xmath7 is a polyhedron expressed by @xmath274 , where @xmath233 is a matrix , @xmath275 is a vector .
thus , from the definition of @xmath39 in algorithm @xmath28 , we see that it is also a polyhedron and @xmath276 with @xmath277 .
after that @xmath278 can be sequentially constructed by adding a linear inequality constraint to the set of constraints of @xmath39 .
this is performed in matlab version 7.0 and the number of constraints increases when @xmath122 increases . the sets @xmath279 in algorithms @xmath27 and @xmath34 are simply constructed more at each step . while the sets @xmath280 in algorithm @xmath18 are two half - spaces , so we use the explicit formula in @xcite to compute @xmath41 . all convex quadratic optimization programs and the projections on polyhedrons
can solved easily by the malab optimization toolbox where the projections are equivalently rewriten to the distance optimization programs .
the algorithms are performed on a pc desktop intel(r ) core(tm )
m cpu @ 2.50ghz 2.50 ghz , ram 2.00 gb . for a given tolerance @xmath281 , we compare numbers of iterates ( iter . ) and execution time ( cpu in sec . ) of mentioned algorithms above with chosing different starting points .
_ example 1 .
_ we consider the bifunction @xmath8 in @xmath282 proposed in ( * ? ? ?
* example 3 ) as @xmath283 and the feasible set @xmath284\times[0,1]$ ] .
it is easy to show that @xmath11 is monotone ( so pseudomonotone ) and lipschitz - type continuous with @xmath285 .
the solution set of ep for @xmath11 on @xmath7 is @xmath286 . in this example , for a starting point @xmath287 then the sequence @xmath26 generated by algorithms @xmath18 , @xmath27 , @xmath28 and @xmath34 converges strongly to @xmath288 which is easily known because @xmath15 is explicit .
the termination criterion in all algorithms is @xmath289 .
the parameters are chosen as follows @xmath290 . in algorithm @xmath18 , we chose @xmath291 .
the results are shown in table @xmath292 .
.results for given starting points in _
example 1_. [ cols="^,^,^,^,^,^,^,^,^ " , ] although , the study of the numerical examples here is preliminary and it is clear that ep depends on the structure of the feasible set @xmath7 and the bifunction @xmath11 .
however , the results in tables @xmath292 and @xmath293 show the convergence of our proposed algorithm and compare its efficiency with the others .
the paper proposes a novel algorithm for solving eps for a class of pseudomonotone and lipschitz - type continuous bifunctions . by constructing the specially cutting halfspaces ,
we have designed the algorithm without the extra - steps .
this is the reason which explains why our algorithm can be considered as an improvement of some previously known algorithms . the strong convergence of the algorithm is proved and its efficiency is illustrated by some numerical experiments .
it is also emphasized that we still have to solve exactly an optimization problem in each step .
this , in general , is a disadvantage of the algorithm ( also , of the extragradient methods and the armijo linesearch methods ) when equilibrium bifunctions and feasible sets have complex structures .
however , contrary to several previous algorithms , our algorithm does not only avoid using the extra - steps which , in general , are inherently costly but also is numerically easer at its last step because the projection is only performed onto the intersection of two half - spaces .
the paper also help us in the design and analysis of more practical algorithms to be seen .
finally , it seems to be that the algorithm also has competitive advantage .
99 anh , p.k . ,
hieu , d.v . :
parallel hybrid methods for variational inequalities , equilibrium problems and common fixed point problems .
vietnam j. math .
( 2015 ) , doi:10.1007/s10013 - 015 - 0129-z blum , e. , oettli , w. : from optimization and variational inequalities to equilibrium problems , math
. program . * 63 * , 123 - 145 ( 1994 ) ceng , l. c. , hadjisavvas , n. , wong , n. c. : strong convergence theorem by a hybrid extragradient - like approximation method for variational inequalities and fixed point problems .
. optim . * 46 * , 635646 ( 2010 ) ceng , l.c .
, yao , j.c . : an extragradient - like approximation method for variational inequality problems and fixed point problems .
* 190 * , 205 - 215 ( 2007 ) combettes , p. l. , hirstoaga , s. a. : equilibrium programming in hilbert spaces
. j. nonlinear convex anal . *
6*(1 ) , 117 - 136 ( 2005 ) censor , y . , gibali , a. , reich , s. : the subgradient extragradient method for solving variational inequalities in hilbert space .
theory appl .
* 148*(2 ) , 318 - 335 ( 2011 ) censor , y. , gibali , a. , reich , s. : strong convergence of subgradient extragradient methods for the variational inequality problem in hilbert space , optim .
methods softw .
* 26*(4 - 5 ) , 827 - 845 ( 2011 ) daniele , p. , giannessi , f. , maugeri , a. : equilibrium problems and variational models , kluwer , ( 2003 ) dinh , b.v .
, hung , p.g .
, muu , l.d .
: bilevel optimization as a regularization approach to pseudomonotone equilibrium problems .
* 35 * ( 5 ) , 539563 ( 2014 ) goebel , k. , reich , s. : uniform convexity , hyperbolic geometry , and nonexpansive map - pings .
marcel dekker , new york ( 1984 ) goebel , k. , kirk , w.a . :
topics in metric fixed point theory .
cambridge studies in advanced math .
, vol . * 28*. cambridge university press , cambridge , ( 1990 ) hieu , d. v. : a parallel hybrid method for equilibrium problems , variational inequalities and nonexpansive mappings in hilbert space .
j. korean math .
* 52 * , 373 - 388 ( 2015 ) hieu , d. v. : the common solutions to pseudomonotone equilibrium problems .
iranian math .
( 2015 ) ( accepted for publication ) he , b .-s . , yang , z .- h . , yu an , x .-
m . : an approximate proximal - extragradient type method for monotone variational inequalities .
* 300 * , 362374 ( 2004 ) fan , k. : a minimax inequality and applications . in : shisha ,
o. ( ed . ) inequality iii , pp .
103 - 113 . academic press , new york ( 1972 ) konnov , i.v . :
combined relaxation methods for variational inequalities .
springer , berlin ( 2000 ) korpelevich , g. m. : the extragradient method for finding saddle points and other problems , ekonomikai matematicheskie metody . * 12 * , 747 - 756 ( 1976 ) lyashko , s. i. , semenov , v. v. and voitova t. a. : low - cost modification of korpelevich s methods for monotone equilibrium problems , cybernetics and systems analysis , * 47*(4 ) , 146 - 154 ( 2011 ) mastroeni , g. : on auxiliary principle for equilibrium problems . publ .
* 3 * , 12441258 ( 2000 ) muu , l.d . , oettli , w. : convergence of an adative penalty scheme for finding constrained equilibria .
nonlinear anal .
18*(12 ) , 1159 - 1166 ( 1992 ) yu . v. malitsky , v. v. semenov : a hybrid method without extrapolation step for solving variational inequality problems . j. glob . optim . * 61 * , 193 - 202 ( 2015 ) martinet , b. : r@xmath294gularisation d@xmath295 in@xmath294quations variationelles par approximations successives .
rech . op@xmath294r .
_ * 4 * , 154159 ( 1970 ) mordukhovich , b. , panicucci , b. , pappalardo , m. , passacantando , m. : hybrid proximal methods for equilibrium problems .
* 6 * , 1535 - 1550 ( 2012 ) nadezhkina , n. , takahashi , w. : strong convergence theorem by a hybrid method for nonexpansive mappings and lipschitz - continuous monotone mappings .
siam j. optim .
* 16 * , 1230 - 1241 ( 2006 ) nadezhkina , n. , takahashi , w. : weak convergence theorem by an extragradient method for nonexpansive mappings and monotone mappings , j. optim .
theory appl .
* 128 * , 191 - 201 ( 2006 ) nguyen , t. t. v. , strodiot , j. j. , nguyen , v. h. : hybrid methods for solving simultaneously an equilibrium problem and countably many fixed point problems in a hilbert space .
theory appl .
( 2013 ) doi 10.1007/s10957 - 013 - 0400-y nguyen , t. p. d , strodiot , j. j. , nguyen , v. h. , nguyen , t. t. v. : a family of extragradient methods for solving equilibrium problems . journal of industrial and management optimization .
11(2 ) , 619 - 630 ( 2015 ) .
doi:10.3934/jimo.2015.11.619 quoc , t.d . ,
muu , l.d . ,
hien , n.v .
: extragradient algorithms extended to equilibrium problems .
optimization * 57 * , 749 - 776 ( 2008 ) rockafellar , r.t . : convex analysis .
princeton , nj : princeton university press , 1970 rockafellar , r.t .
: monotone operators and the proximal point algorithm .
siam j. control optim .
* 14 * , 877898 ( 1976 ) solodov , m. v. , svaiter , b. f.:forcing strong convergence of proximal point iterations in hilbert space .
math . program .
* 87 * , 189 - 202 ( 2000 ) solodov , m. v , svaiter , b. f. : a new projection method for variational inequality problems .
siam j. control .
optim . * 37 * , 765776 ( 1999 ) strodiot , j. j. , nguyen , t. t. v , nguyen , v. h. : a new class of hybrid extragradient algorithms for solving quasi - equilibrium problems , j. glob . optim . * 56 * ( 2 ) , 373 - 397 ( 2013 ) takahashi , s. , takahashi , w. : viscosity approximation methods for equilibrium problems and fixed point in hilbert space . j. math .
appl . * 331*(1 ) , 506 - 515 ( 2007 ) takahashi , w. , toyoda , m. : weak convergence theorems for nonexpansive mappings and monotone mappings .
theory appl .
* 118 * , 417428 ( 2003 ) vuong , p. t. , strodiot , j. j. , nguyen , v. h. : extragradient methods and linesearch algorithms for solving ky fan inequalities and fixed point problems .
theory appl .
* 155 * , 605627 ( 2012 ) jaiboon , c. , kumam , p. : strong convergence theorems for solving equilibrium problems and fixed point problems of @xmath297 -strict pseudo - contraction mappings by two hybrid projection methods . j. comput .
math . * 234 * , 722 - 732 ( 2010 ) | in this paper , we introduce a new hybrid algorithm for solving equilibrium problems .
the algorithm combines the extragradient method and the hybrid ( outer approximation ) method . in this algorithm , only an optimization program is solved at each iteration without the extra - steps like as in the extragradient method and the armijo linesearch method .
a specially constructed half - space in the hybrid method is the reason for the absence of an optimization program in our algorithm .
the strong convergence theorem is established and several numerical experiments are implemented to illustrate the convergence of the algorithm and compare it with others .
van hieu dang@xmath0 ( communicated by the associate editor name ) |
kernel methods have proven to be a highly successful technique for solving many problems in machine learning , ranging from classification and regression to sequence annotation and feature extraction @xcite . at their heart
lies the idea that inner products in high - dimensional feature spaces can be computed in implicit form via kernel function @xmath5 : @xmath6 here @xmath7 is a feature map transporting elements of the observation space @xmath8 into a possibly infinite - dimensional feature space @xmath9 .
this idea was first used by @xcite to show nonlinear separation .
there exists a rich body of literature on reproducing kernel hilbert spaces ( rkhs ) @xcite and one may show that estimators using norms in feature space as penalty are equivalent to estimators using smoothness in an rkhs @xcite .
furthermore , one may provide a bayesian interpretation via gaussian processes .
see e.g. @xcite for details . more concretely , to evaluate the decision function @xmath10 on an example @xmath11 , one typically employs the kernel trick as follows @xmath12 this has been viewed as a strength of kernel methods , especially in the days that datasets consisted of ten thousands of examples .
this is because the representer theorem of @xcite states that such a function expansion in terms of finitely many coefficients must exist under fairly benign conditions even whenever the space is infinite dimensional .
hence we can effectively perform optimization in infinite dimensional spaces .
this trick that was also exploited by @xcite for evaluating pca . frequently the coefficient space is referred to as _
dual space_. this arises from the fact that the coefficients are obtained by solving a dual optimization problem .
unfortunately , on large amounts of data , this expansion becomes a significant liability for computational efficiency .
for instance , @xcite show that the number of nonzero @xmath13 ( i.e. , @xmath14 , also known as the number of `` support vectors '' ) in many estimation problems can grow linearly in the size of the training set . as a consequence ,
as the dataset grows , the expense of evaluating @xmath15 also grows .
this property makes kernel methods expensive in many large scale problems : there the sample size @xmath16 may well exceed billions of instances .
the large scale solvers of @xcite and @xcite work in primal space to sidestep these problems , albeit at the cost of limiting themselves to linear kernels , a significantly less powerful function class .
numerous methods have been proposed to mitigate this issue . to compare computational cost of these methods
we make the following assumptions : * we have @xmath16 observations and access to an @xmath17 with @xmath18 algorithm for solving the optimization problem at hand . in other words ,
the algorithm is linear or worse .
this is a reasonable assumption almost all data analysis algorithm need to inspect the data at least once to draw inference .
* data has @xmath3 dimensions . for simplicity
we assume that it is dense with density rate @xmath19 , i.e. on average @xmath20 coordinates are nonzero . *
the number of nontrivial basis functions is @xmath21 .
this is well motivated by @xcite and it also follows from the fact that e.g. in regularized risk minimization the subgradient of the loss function determines the value of the associated dual variable . * we denote the number of ( nonlinear ) basis functions by @xmath2 .
[ [ reduced - set - expansions ] ] reduced set expansions + + + + + + + + + + + + + + + + + + + + + + @xcite focused on compressing function expansions after the problem was solved by means of reduced - set expansions .
that is , one first solves the full optimization problem at @xmath22 cost and subsequently one minimizes the discrepancy between the full expansion and an expansion on a subset of basis functions .
the exponent of @xmath23 arises from the fact that we need to compute @xmath24 kernels @xmath17 times .
evaluation of the reduced function set costs at least @xmath25 operations per instance and @xmath25 storage , since each kernel function @xmath26 requires storage of @xmath27 .
[ [ low - rank - expansions ] ] low rank expansions + + + + + + + + + + + + + + + + + + + subsequent work by @xcite and @xcite aimed to reduce memory footprint and complexity by finding subspaces to expand functions .
the key difference is that these algorithms reduce the function space _ before _ seeing labels .
while this is suboptimal , experimental evidence shows that for well designed kernels the basis functions extracted in this fashion are essentially as good as reduced set expansions .
this is to be expected .
after all , the kernel encodes our prior belief in which function space is most likely to capture the relevant dependencies between covariates and labels .
these projection - based algorithms generate an @xmath2-dimensional subspace : * compute the kernel matrix @xmath28 on an @xmath2-dimensional subspace at @xmath29 cost . *
the matrix @xmath28 is inverted at @xmath30 cost .
* for all observations one computes an explicit feature map by projecting data in rkhs onto the set of @xmath2 basis vectors via @xmath31}$ ] .
that is , training proceeds at @xmath32 cost . * prediction costs @xmath25 computation and @xmath25 memory , as in reduced set methods , albeit with a different set of basis functions .
note that these methods temporarily require @xmath33 storage during training , since we need to be able to multiply with the inverse covariance matrix efficiently .
this allows for solutions to problems where @xmath16 is in the order of millions and @xmath2 is in the order of thousands : for @xmath34 we need approximately @xmath35 gb of memory to store and invert the covariance matrix .
preprocessing can be parallelized efficiently .
obtaining a minimal set of observations to project on is even more difficult and only the recent work of @xcite provides usable performance guarantees for it .
[ [ multipole - methods ] ] multipole methods + + + + + + + + + + + + + + + + + fast multipole expansions @xcite offer one avenue for efficient function expansions whenever the dimensionality of the underlying space is relatively modest .
however , for high dimensions they become computationally intractable in terms of space partitioning , due to the curse of dimensionality .
moreover , they are typically tuned for localized basis functions , specifically the gaussian rbf kernel .
[ [ random - subset - kernels ] ] random subset kernels + + + + + + + + + + + + + + + + + + + + + a promising alternative to _ approximating an existing kernel function _ is to design new ones that are immediately compatible with scalable data analysis .
a recent instance of such work is the algorithm of @xcite who map observations @xmath11 into set membership indicators @xmath36 , where @xmath37 denotes the random partitioning chosen at iterate @xmath37 and @xmath38 indicates the particular set .
while the paper suggests that the algorithm is scalable to large amounts of data , it suffers from essentially the same problem as other feature generation methods insofar as it needs to evaluate set membership for each of the partitions for all data , hence we have an @xmath39 computational cost for @xmath2 partitions into @xmath5 sets on @xmath16 observations .
even this estimate is slightly optimistic since we assume that computing the partitions is independent of the dimensionality of the data . in summary , while the function class is potentially promising , its computational cost considerably exceeds that of the other algorithms discussed below , hence we do not investigate it further . [
[ random - kitchen - sinks ] ] random kitchen sinks + + + + + + + + + + + + + + + + + + + + a promising alternative was proposed by @xcite under the moniker of _ random kitchen sinks_. in contrast to previous work the authors attempt to obtain an _ explicit _ function space expansion directly .
this works for translation invariant kernel functions by performing the following operations : * generate a ( gaussian ) random matrix @xmath40 of size @xmath41 . * for each observation @xmath11 compute @xmath42 and apply a nonlinearity @xmath43 to each coordinate separately , i.e. @xmath44_i)$ ] .
the approach requires @xmath45 storage both at training and test time .
training costs @xmath46 operations and prediction on a new observation costs @xmath25 .
this is potentially much cheaper than reduced set kernel expansions .
the experiments in @xcite showed that performance was very competitive with conventional rbf kernel approaches while providing dramatically simplified code .
note that explicit spectral finite - rank expansions offer potentially much faster rates of convergence , since the spectrum decays as fast as the eigenvalues of the associated regularization operator @xcite .
nonetheless random kitchen sinks are a very attractive alternative due to their simple construction and the flexility in synthesizing kernels with predefined smoothness properties . [ [ fastfood ] ] fastfood + + + + + + + + our approach hews closely to random kitchen sinks .
however , it succeeds at overcoming their key obstacle the need to _ store _ and to _ multiply _ by a random matrix .
this way , fastfood , accelerates random kitchen sinks from @xmath4 to @xmath47 time while only requiring @xmath1 rather than @xmath4 storage .
the speedup is most significant for large input dimensions , a common case in many large - scale applications .
for instance , a tiny 32x32x3 image in the cifar-10 @xcite already has 3072 dimensions , and non - linear function classes have shown to work well for mnist @xcite and cifar-10 .
our approach relies on the fact that hadamard matrices , when combined with gaussian scaling matrices , behave very much like gaussian random matrices .
that means these two matrices can be used in place of gaussian matrices in random kitchen sinks and thereby speeding up the computation for a large range of kernel functions .
the computational gain is achieved because unlike gaussian random matrices , hadamard matrices admit fft - like multiplication and require no storage .
we prove that the fastfood approximation is unbiased , has low variance , and concentrates almost at the same rate as random kitchen sinks .
moreover , we extend the range of applications from radial basis functions @xmath48 to any kernel that can be written as dot product @xmath49 .
extensive experiments with a wide range of datasets show that fastfood achieves similar accuracy to full kernel expansions and random kitchen sinks while being 100x faster with 1000x less memory .
these improvements , especially in terms of memory usage , make it possible to use kernel methods even for embedded applications .
our experiments also demonstrate that fastfood , thanks to its speedup in training , achieves state - of - the - art accuracy on the cifar-10 dataset @xcite among permutation - invariant methods .
table [ tab : compcost ] summarizes the computational cost of the above algorithms .
.computational cost for reduced rank expansions .
efficient algorithms achieve @xmath50 and typical sparsity coefficients are @xmath51 . [ cols="<,<,<,<,<",options="header " , ] in figure [ fig : expansion ] , we show regression performance as a function of the number of basis functions @xmath2 on the cpu dataset .
as is evident , it is necessary to have a large @xmath2 in order to learn highly nonlinear functions .
interestingly , although the fourier features do not seem to approximate the gaussian rbf kernel , they perform well compared to other variants and improve as @xmath2 increases .
this suggests that learning the kernel by direct spectral adjustment might be a useful application of our proposed method . in the previous experiments ,
we observe that fastfood is on par with exact kernel computation , the nystrom method , and random kitchen sinks .
the key point , however , is to establish whether the algorithm offers computational savings . for this purpose
we compare random kitchen sinks using eigen and our method using spiral .
both are highly optimized numerical linear algebra libraries in c++ .
we are interested in the time it takes to go from raw features of a vector with dimension @xmath3 to the label prediction of that vector . on a small problem with @xmath52 and @xmath53
, performing prediction with random kitchen sinks takes 0.07 seconds .
our method is around 24x faster , taking only 0.003 seconds to compute the label for one input vector .
the speed gain is even more significant for larger problems , as is evident in table [ tab : improvements ] .
this confirms experimentally the @xmath47 vs. @xmath54 runtime and the @xmath1 vs.@xmath54 storage of fastfood relative to random kitchen sinks . in other words ,
the computational savings are substantial for large input dimensionality @xmath3 . to understand the importance of nonlinear feature expansions for a practical application , we benchmarked fastfood , random kitchen sinks on the cifar-10 dataset @xcite which has 50,000 training images and 10,000 test images .
each image has 32x32 pixels and 3 channels ( @xmath55 ) . in our experiments ,
linear svms achieve 42.3% accuracy on the test set .
non - linear expansions improve the classification accuracy significantly .
in particular , fastfood fft ( `` fourier features '' ) achieve 63.1% while fastfood ( `` hadamard features '' ) and random kitchen sinks achieve 62.4% with an expansion of @xmath56 .
these are also best known classification accuracies using permutation - invariant representations on this dataset . in terms of speed ,
random kitchen sinks is 5x slower ( in total training time ) and 20x slower ( in predicting a label given an image ) compared to both fastfood and and fastfood fft .
this demonstrates that non - linear expansions are needed even when the raw data is high - dimensional , and that fastfood is more practical for such problems . in particular , in many cases ,
linear function classes are used because they provide fast training time , and especially test time , but not because they offer better accuracy .
the results on cifar-10 demonstrate that fastfood can overcome this obstacle .
we demonstrated that it is possible to compute @xmath2 nonlinear basis functions in @xmath47 time , a significant speedup over the best competitive algorithms .
this means that kernel methods become more practical for problems that have large datasets and/or require real - time prediction .
in fact , fastfood can be used to run on cellphones because not only it is fast , but it also requires only a small amount of storage . note that our analysis is not limited to translation invariant kernels but it also includes inner product formulations .
this means that for most practical kernels our tools offer an easy means of making kernel methods scalable beyond simple subspace decomposition strategies .
extending our work to other symmetry groups is subject to future research . also note that fast multiplications with near - gaussian matrices are a key building block of many randomized algorithms .
it remains to be seen whether one could use the proposed methods as a substitute and reap significant computational savings .
i. bogaert , b. michiels , and j. fostier . computation of legendre polynomials and gauss legendre nodes and weights for parallel computing .
_ siam journal on scientific computing _ , 340 ( 3):0 c83c101 , 2012 .
b. boser , i. guyon , and v. vapnik .
a training algorithm for optimal margin classifiers . in d.
haussler , editor , _ proc . annual conf .
computational learning theory _
, pages 144152 , pittsburgh , pa , july 1992 .
acm press .
s. boyd , n. parikh , e. chu , b. peleato , and j. eckstein . distributed optimization and statistical learning via the alternating direction method of multipliers . _
foundations and trends in machine learning _ , 30 ( 1):0 1123 , 2010 .
a. das and d. kempe .
submodular meets spectral : greedy algorithms for subset selection , sparse approximation and dictionary selection . in l.
getoor and t. scheffer , editors , _ proceedings of the 28th international conference on machine learning , icml _ , pages 10571064 .
omnipress , 2011 .
a. dasgupta , r. kumar , and t. sarls . fast locality - sensitive hashing . in _ proceedings of the 17th acm
sigkdd international conference on knowledge discovery and data mining _ , pages 10731081 .
acm , 2011 .
f. girosi and g. anzellotti .
rates of convergence for radial basis functions and neural networks . in r.
j. mammone , editor , _ artificial neural networks for speech and vision _ , pages 97113 , london , 1993 .
chapman and hall .
s. matsushima , s.v.n .
vishwanathan , and a.j .
linear support vector machines via dual cached loops . in q.
yang , d. agarwal , and j. pei , editors , _ the 18th acm sigkdd international conference on knowledge discovery and data mining , kdd _ , pages 177185 .
acm , 2012 .
url http://dl.acm.org/citation.cfm?id=2339530 .
a. rahimi and b. recht .
random features for large - scale kernel machines . in j.c .
platt , d. koller , y. singer , and s. roweis , editors , _ advances in neural information processing systems 20_. mit press , cambridge , ma , 2008 .
n. ratliff , j. bagnell , and m. zinkevich .
( online ) subgradient methods for structured prediction . in _ eleventh international conference on artificial intelligence and statistics ( aistats ) _ , march 2007 .
a. j. smola and b. schlkopf . sparse greedy matrix approximation for machine learning . in
_ proceedings of the international conference on machine learning _ , pages 911918 , san francisco , 2000 .
morgan kaufmann publishers .
a. j. smola , b. schlkopf , and k .-
general cost functions for support vector regression . in t. downs , m. frean , and m. gallagher , editors , _ proc . of the ninth australian conf . on neural networks _ , pages 7983 , brisbane , australia , 1998
university of queensland .
a. j. smola , z. l. vri , and r. c. williamson .
regularization with dot - product kernels . in t.
k. leen , t. g. dietterich , and v. tresp , editors , _ advances in neural information processing systems 13 _ , pages 308314 . mit press , 2001 .
b. taskar , c. guestrin , and d. koller .
max - margin markov networks . in s. thrun , l. saul , and b. schlkopf , editors , _ advances in neural information processing systems 16 _ , pages 2532 , cambridge , ma , 2004 .
mit press . v. vapnik , s. golowich , and a. j. smola .
support vector method for function approximation , regression estimation , and signal processing . in m. c. mozer , m. i. jordan , and t. petsche , editors , _ advances in neural information processing systems 9 _ , pages 281287 , cambridge , ma , 1997 . mit press . c. k. i. williams .
prediction with gaussian processes : from linear regression to linear prediction and beyond . in m.
i. jordan , editor , _ learning and inference in graphical models _ , pages 599621 .
kluwer academic , 1998 .
christoper k. i. williams and matthias seeger .
using the nystrom method to speed up kernel machines . in t.
k. leen , t. g. dietterich , and v. tresp , editors , _ advances in neural information processing systems 13 _ , pages 682688 , cambridge , ma , 2001 . mit press .
r. c. williamson , a. j. smola , and b. schlkopf .
generalization bounds for regularization networks and support vector machines via entropy numbers of compact operators .
_ ieee trans .
inform . theory _
, 470 ( 6):0 25162532 , 2001 . | despite their successes , what makes kernel methods difficult to use in many large scale problems is the fact that storing and computing the decision function is typically expensive , especially at prediction time . in this paper
, we overcome this difficulty by proposing fastfood , an approximation that accelerates such computation significantly .
key to fastfood is the observation that hadamard matrices , when combined with diagonal gaussian matrices , exhibit properties similar to dense gaussian random matrices .
yet unlike the latter , hadamard and diagonal matrices are inexpensive to multiply and store .
these two matrices can be used in lieu of gaussian matrices in random kitchen sinks proposed by @xcite and thereby speeding up the computation for a large range of kernel functions .
specifically , fastfood requires @xmath0 time and @xmath1 storage to compute @xmath2 non - linear basis functions in @xmath3 dimensions , a significant improvement from @xmath4 computation and storage , without sacrificing accuracy .
our method applies to any translation invariant and any dot - product kernel , such as the popular rbf kernels and polynomial kernels .
we prove that the approximation is unbiased and has low variance .
experiments show that we achieve similar accuracy to full kernel expansions and random kitchen sinks while being 100x faster and using 1000x less memory .
these improvements , especially in terms of memory usage , make kernel methods more practical for applications that have large training sets and/or require real - time prediction . |
estimation of discrete structure , such as graphs or clusters , or variable selection is an age - old problem in statistics .
it has enjoyed increased attention in recent years due to the massive growth of data across many scientific disciplines .
these large datasets often make estimation of discrete structures or variable selection imperative for improved understanding and interpretation .
most classical results do not cover the loosely defined case of high - dimensional data , and it is mainly in this area where we motivate the promising properties of our new stability selection . in the context of regression , for example , an active area of research is to study the @xmath0 case , where the number of variables or covariates @xmath1 exceeds the number of observations @xmath2
; for an early overview see for example @xcite . in a similar spirit , graphical modelling with many more nodes than sample size
has been the focus of recent research , and cluster analysis is another widely used technique to infer a discrete structure from observed data .
challenges with estimation of discrete structures include computational aspects , since corresponding optimisation problems are discrete , as well as determining the right amount of regularisation , for example in an asymptotic sense for consistent structure estimation .
substantial progress has been made over the last years in developing computationally tractable methods which have provable statistical ( asymptotic ) properties , even for the high - dimensional setting with many more variables than samples .
one interesting stream of research has focused on relaxations of some discrete optimisation problems , for example by @xmath3-penalty approaches @xcite or greedy algorithms @xcite .
the practical usefulness of such procedures has been demonstrated in various applications .
however , the general issue of selecting a proper amount of regularisation ( for the procedures mentioned above and for many others ) for getting a right - sized structure or model has largely remained a problem with unsatisfactory solutions .
we address the problem of proper regularisation with a very generic subsampling approach ( bootstrapping would behave similarly ) .
we show that subsampling can be used to determine the amount of regularisation such that a certain familywise type i error rate in multiple testing can be conservatively controlled for finite sample size . particularly for complex , high - dimensional problems ,
a finite sample control is much more valuable than an asymptotic statement with the number of observations tending to infinity . beyond the issue of choosing the amount of regularisation
, the subsampling approach yields a new structure estimation or variable selection scheme . for the more specialised case of high - dimensional linear models , we prove what we expect in greater generality : namely that subsampling in conjunction with @xmath3-penalised estimation requires much weaker assumptions on the design matrix for asymptotically consistent variable selection than what is needed for the ( non - subsampled ) @xmath3-penalty scheme .
furthermore , we show that additional improvements can be achieved by randomising not only via subsampling but also in the selection process for the variables , bearing some resemblance to the successful tree - based random forest algorithm @xcite . subsampling ( and bootstrapping )
has been primarily used so far for asymptotic statistical inference in terms of standard errors , confidence intervals and statistical testing .
our work here is of a very different nature : the marriage of subsampling and high - dimensional selection algorithms yields finite sample familywise error control and markedly improved structure estimation or selection methods . in general ,
let @xmath4 be a @xmath1-dimensional vector , where @xmath4 is sparse in the sense that @xmath5 components are non - zero .
in other words , @xmath6 . denote the set of non - zero values by @xmath7 and the set of variables with vanishing coefficient by @xmath8 .
the goal of structure estimation is to infer the set @xmath9 from noisy observations . as a first supervised example , consider data @xmath10 with univariate response variable @xmath11 and @xmath1-dimensional covariates @xmath12 .
we typically assume that @xmath13 s are i.i . distributed .
the vector @xmath4 could be the coefficient vector in a linear model @xmath14 where @xmath15 , @xmath12 is the @xmath16 design matrix and @xmath17 is the random noise whose components are independent , identically distributed .
thus , inferring the set @xmath9 from data is the well - studied variable selection problem in linear regression . a main stream of classical methods proceeds to solve this problem by penalising the negative log - likelihood with the @xmath18-norm @xmath19 which equals the number of non - zero components of @xmath4 .
the computational task to solve such an @xmath18-norm penalised optimisation problem becomes quickly unfeasible if @xmath1 is getting large , even when using efficient branch and bound techniques .
alternatively , one can relax the @xmath18-norm by the @xmath3-norm penalty .
this leads to the lasso estimator @xcite , @xmath20 where @xmath21 is a regularisation parameter and we typically assume that the covariates are on the same scale , i.e. @xmath22 .
an attractive feature of lasso is its computational feasibility for large @xmath1 since the optimisation problem in ( [ lasso ] ) is convex .
furthermore , the lasso is able to select variables by shrinking certain estimated coefficients exactly to 0 .
we can then estimate the set @xmath9 of non - zero @xmath4 coefficients by @xmath23 which involves convex optimisation only .
substantial understanding has been gained over the last few years about consistency of such lasso variable selection @xcite , and we present the details in section [ subsec.randomlasso ] . among the challenges
are the issue of choosing a proper amount of regularisation @xmath24 for consistent variable selection and the fact that restrictive design conditions are needed for asymptotically recovering the true set @xmath9 of relevant covariates .
a second example is on unsupervised gaussian graphical modelling .
the data is assumed to be @xmath25 the goal is to infer conditional dependencies among the @xmath26 variables or components in @xmath27 .
it is well - known that @xmath28 and @xmath29 are conditionally dependent given all other components @xmath30 if and only if @xmath31 , and we then draw an edge between nodes @xmath32 and @xmath33 in a corresponding graph @xcite .
the structure estimation is thus on the index set @xmath34 which has cardinality @xmath35 ( and of course , we can represent @xmath36 as a @xmath37 vector ) and the set of relevant conditional dependencies is @xmath38 .
similarly to the problem of variable selection in regression , @xmath18-norm methods are computationally very hard and become very quickly unfeasible for moderate or large values of @xmath26 .
a relaxation with @xmath3-type penalties has also proven to be useful in this context @xcite .
a recent proposal is the graphical lasso @xcite : @xmath39 this amounts to an @xmath3-penalised estimator of the gaussian log - likelihood , partially maximised over the mean vector @xmath40 , when minimising over all nonnegative definite symmetric matrices .
the estimated graph structure is then @xmath41 which involves convex optimisation only and is computationally feasible for large values of @xmath26 .
another potential area of application is clustering .
choosing the correct number of cluster is a notoriously difficult problem . looking for clusters that are stable under perturbations or subsampling of the data can help to get a better sense of a meaningful number of clusters and to validate results .
indeed , there has been some activity in this area , most notably in the context of _ consensus clustering _ @xcite . for an early application see @xcite .
our proposed false discovery control can be applied to consensus clustering , yielding good estimates of the parameters of a suitable base clustering method for consensus clustering .
the use of resampling for purposes of validation is certainly not new ; we merely try to put it into a more formal framework and to show certain empirical and theoretical advantages of doing so .
it seems difficult to give a complete coverage of all previous work in the area , as notions of stability , resampling and perturbations are very natural in the context of structure estimation and variable selection .
we reference and compare with previous work throughout the paper .
the structure of the paper is as follows .
the generic stability selection approach , its familywise type i multiple testing error control and some representative examples from high - dimensional linear models and gaussian graphical models are presented in section [ sec.stable ] . a detailed asymptotic analysis of lasso and randomised lasso for high - dimensional linear models
is given in section [ sec.cons ] and more numerical results are described in section [ sec.numeric ] . after a discussion in section [ sec.disc ] ,
we collect all the technical proofs in the appendix .
stability selection is not a new variable selection technique .
its aim is rather to enhance and improve existing methods .
first , we give a general description of stability selection and we present specific examples and applications later .
we assume throughout this section [ sec.stable ] that the data , denoted here by @xmath42 , are independent and identically distributed ( e.g. @xmath43 with covariate @xmath44 and response @xmath45 ) . for a generic structure estimation or variable selection technique , we have a tuning parameter @xmath46 that determines the amount of regularisation .
this tuning parameter could be the penalty parameter in @xmath3-penalised regression , see ( [ lasso ] ) , or in gaussian graphical modelling , see ( [ glasso ] ) ; or it may be number of steps in forward variable selection or orthogonal matching pursuit @xcite or the number of iterations in matching pursuit @xcite or boosting @xcite ; a large number of steps of iterations would have an opposite meaning from a large penalty parameter , but this does not cause conceptual problems . for every value @xmath47 , we obtain a structure estimate @xmath48 .
it is then of interest to determine whether there exists a @xmath47 such that @xmath49 is identical to @xmath9 with high probability and how to achieve that right amount of regularisation .
we motivate the concept of stability paths in the following , first for regression .
stability paths are derived from the concept of regularisation paths .
a regularisation path is given by the coefficient value of each variable over all regularisation parameters : @xmath50 .
stability paths ( defined below ) are , in contrast , the _ probability _ for each variable to be selected when randomly resampling from the data . for any given regularisation parameter @xmath47 , the selected set @xmath49 is implicitly a function of the samples @xmath51 .
we write @xmath52 where necessary to express this dependence .
let @xmath53 be a random subsample of @xmath54 of size @xmath55 , drawn without replacement .
for every set @xmath56 , the probability of being in the selected set @xmath57 is @xmath58 the probability @xmath59 in ( [ pi ] ) is with respect to both the random subsampling ( and other sources of randomness if @xmath49 is a randomised algorithm , see section [ subsec.randomlasso ] ) . the sample size of @xmath55 is chosen as it resembles most closely the bootstrap @xcite while allowing computationally efficient implementation .
subsampling has also been advocated in a related context in @xcite . for every variable @xmath60
, the stability path is given by the selection probabilities @xmath61 , @xmath47 .
it is a complement to the usual path - plots that show the coefficients of all variables @xmath60 as a function of the regularisation parameter .
it can be seen in figure [ fig : stabpath ] that this simple path plot is potentially very useful for improved variable selection for high - dimensional data . in the remainder of the manuscript , we look at the selection probabilities of individual variables .
the definition above covers also sets of variables .
we could monitor the selection probability of a set of functionally related variables , say , by asking how often _ at least one _ variable in this set is chosen or how often _ all _ variables in the set are chosen .
left : the lasso path for the vitamin gene - expression dataset .
the paths of the 6 non - permuted genes are plotted as solid , red lines , while the paths of the 4082 permuted genes are shown as broken , black lines . selecting a model with all 6 unpermuted genes invariably means selecting a large number of irrelevant noise variables .
middle : the stability path of lasso .
the first 4 variables chosen with stability selection are truly non - permuted variables .
right : the stability path for the ` randomised lasso ' with weakness @xmath62 , introduced in section [ subsec.randomlasso ] .
now all 6 non - permuted variables are chosen before any noise variable enters the model .
_ , scaledwidth=95.0% ] we apply stability selection to the lasso defined in ( [ lasso ] ) .
we work with a gene expression dataset for illustration which is kindly provided by dsm nutritional products ( switzerland ) . for @xmath63 samples
, there is a continuous response variable measuring the logarithm of riboflavin ( vitamin b2 ) production rate of bacillus subtilis , and we have @xmath64 continuous covariates measuring the logarithm of gene expressions from essentially the whole genome of bacillus subtilis .
certain mutations of genes are thought to lead to higher vitamin concentrations and the challenge is to identify those relevant genes via a linear regression analysis .
that is , we consider a linear model as in ( [ eq : linear ] ) and want to infer the set @xmath65 .
instability of the selected set of genes has been noted before @xcite , if either using marginal association or variable selection in a regression or classification model .
@xcite are close in spirit to our approach by arguing for ` consensus ' gene signatures which assess the stability of selection , while @xcite propose to measure stability of so - called ` molecular profiles ' by the jaccard index .
to see how lasso and the related stability path cope with noise variables , we randomly permute all but 6 of the 4088 gene expression across the samples , using the same permutation to keep the dependence structure between the permuted gene expressions intact .
the set of 6 unpermuted genes has been chosen randomly among the 200 genes with the highest marginal association with the response .
the lasso path @xmath66 is shown in the left panel of figure [ fig : stabpath ] , as a function of the regularisation parameter @xmath24 ( rescaled so that @xmath67 is the minimal @xmath68-value for which the null model is selected and @xmath69 amounts to the basis pursuit solution ) .
three of the ` relevant ' ( unpermuted ) genes stand out , but all remaining three variables are hidden within the paths of noise ( permuted ) genes .
the middle panel of figure [ fig : stabpath ] shows the stability path .
at least four relevant variables stand out much clearer now than they did in the regularisation path plot .
the right panel shows the stability plot for randomised lasso which will be introduced in section [ subsec.randomlasso ] : now all 6 unpermuted variables stand above the permuted variables and the separation between ( potentially ) relevant variables and irrelevant variables is even better .
choosing the right regularisation parameter is very difficult for the original path .
the prediction optimal and cross - validated choice include too many variables @xcite and the same effect can be observed in this example , where 14 permuted variables are included in the model chosen by cross - validation . figure [ fig : stabpath ] motivates that choosing the right regularisation parameter is much less critical for the stability path and that we have a better chance to select truly relevant variables . in a traditional setting , variable selection would amount to choosing one element of the set of models @xmath70 where @xmath71 is again the set of considered regularisation parameters , which can be either continuous or discrete .
there are typically two problems : first , the correct model @xmath9 might not be a member of ( [ list ] ) .
second , even if it is a member , it is typically very hard for high - dimensional data to determine the right amount of regularisation @xmath24 to select exactly @xmath9 , or to select at least a close approximation .
with stability selection , we do not simply select one model in the list ( [ list ] ) .
instead the data are perturbed ( for example by subsampling ) many times and we choose all structures or variables that occur in a large fraction of the resulting selection sets . for a cutoff @xmath72 with @xmath73 and a set of regularisation parameters @xmath71 , the set of stable variables
is defined as @xmath74 we keep variables with a high selection probability and disregard those with low selection probabilities .
the exact cutoff @xmath72 with @xmath73 is a tuning parameter but the results vary surprisingly little for sensible choices in a range of the cutoff .
neither do results depend strongly on the choice of regularisation @xmath24 or the regularisation region @xmath71 .
see figure [ fig : stabpath ] for an example . before we present some guidance on how to choose the cutoff parameter and the regularisation region @xmath71 below , it is worthwhile pointing out that there have been related ideas in the literature on bayesian model selection .
@xcite show certain predictive optimality results for the so - called _ median probability model _ , consisting of variables which have posterior probability of being in the model of 1/2 or greater ( as opposed to choosing the model with the highest posterior probability ) . @xcite or @xcite are examples of more applied papers considering bayesian variable selection in this context .
when trying to recover the set @xmath9 , a natural goal is to include as few variables of the set @xmath75 of noise variables as possible .
the choice of the regularisation parameter is hence crucial .
an advantage of our stability selection is that the choice of the initial set of regularisation parameters @xmath71 has typically not a very strong influence on the results , as long as @xmath71 is varied with reason .
another advantage , which we focus on below , is the ability to choose this set of regularisation parameters in a way that guarantees , under stronger assumptions , a certain bound on the expected number of false selections .
[ def - not ] let @xmath76 be the set of selected structures or variables if varying the regularisation @xmath24 in the set @xmath71 .
let @xmath77 be the average number of selected variables , @xmath78 .
define @xmath79 to be the number of falsely selected variables with stability selection , @xmath80 in general , it is very hard to control @xmath81 , as the distribution of the underlying estimator @xmath82 depends on many unknown quantities .
exact control is only possible under some simplifying assumptions .
[ error control ] [ theo : error ] assume that the distribution of @xmath83 is exchangeable for all @xmath47 .
also , assume that the original procedure is not worse than random guessing , i.e. for any @xmath84 , @xmath85 the expected number @xmath79 of falsely selected variables is then bounded by @xmath86 we will discuss below how to make constructive use of the value @xmath87 which is in general an unknown quantity .
the expected number of falsely selected variables is sometimes called the per - family error rate ( _ pfer _ ) or , if divided by @xmath1 , the per - comparison error rate ( _ fcer _ ) in multiple testing @xcite . choosing less variables ( reducing @xmath88 ) or increasing the threshold @xmath72 for selection will , unsurprisingly , reduce the the expected number of falsely selected variables , with a minimal achievable non - trivial value of @xmath89 ( for @xmath90 and @xmath91 ) for the _
pfer_. this seems low enough for all practical purposed as long as @xmath92 , say .
the involved exchangeability assumption is perhaps stronger than one would wish , but there does not seem to be a way of getting error control in the same generality without making similar assumptions . for regression in ( [ eq : linear ] ) , the exchangeability assumption is fulfilled for all reasonable procedures @xmath93 if the design is random and the distribution of @xmath94 is exchangeable .
independence of all variables in @xmath75 is a special case .
more generally , the variables could have a joint normal distribution with @xmath95 for all @xmath96 with @xmath97 and @xmath98 .
for real data , we have no guarantee that the assumption is fulfilled but the numerical examples in section [ sec.numeric ] show that the bound holds up very well for real data . note also that the assumption of exchangeability is only needed to prove theorem [ theo : error ] .
all other benefits of stability selection shown in this paper do not rely on this assumption . besides exchangeability , we needed another , quite harmless , assumption , namely that the original procedure is not worse than random guessing .
one would certainly hope that this assumption is fulfilled .
if it is not , the results below are still valid with slightly weaker constants .
the assumption seems so weak , however , that we do not pursue this further .
the threshold value @xmath72 is a tuning parameter whose influence is very small . for sensible values in the range of , say , @xmath99 , results tend to be very similar . once the threshold is chosen at some default value , the regularisation region @xmath71 is determined by the desired error control .
specifically , for a default cutoff value @xmath100 , choosing the regularisation parameters @xmath71 such that say @xmath101 will control @xmath102 ; or choosing @xmath71 such that @xmath103 controls the familywise error rate ( fwer ) at level @xmath104 , i.e. @xmath105 . of course , we can proceed the other way round by fixing the regularisation region @xmath71 and choosing @xmath72 such that @xmath81 is controlled at the desired level . to do this ,
we need knowledge about @xmath77 . this can be easily achieved by regularisation of the selection procedure @xmath106 in terms of the number of selected variables @xmath107 .
that is , the domain @xmath108 for the regularisation parameter @xmath68 determines the number @xmath107 of selected variables , i.e. @xmath109 . for example , with @xmath3-norm penalisation as in ( [ lasso ] ) or ( [ glasso ] ) , the number @xmath107 is given by the variables which enter first in the regularisation path when varying from a maximal value @xmath110 to some minimal value @xmath111 .
mathematically , @xmath112 is such that @xmath113 . without stability selection ,
the regularisation parameter @xmath24 invariably has to depend on the unknown noise level of the observations .
the advantage of stability selection is that ( a ) exact error control is possible , and ( b ) the method works fine even though the noise level is unknown . this is a real advantage in high - dimensional problems with @xmath0 , as it is very hard to estimate the noise level in these settings .
[ [ pointwise - control . ] ] pointwise control .
+ + + + + + + + + + + + + + + + + + for some applications , evaluation of subsampling replicates of @xmath49 are already computationally very demanding for a single value of @xmath24 .
if this single value @xmath24 is chosen such that some overfitting occurs and the set @xmath49 is rather too large , in the sense that it contains @xmath9 with high probability , the same approach as above can be used and is in our experience very successful .
results typically do not depend strongly on the utilised regularisation @xmath24 .
see the example below for graphical modelling . setting @xmath114
, one can immediately transfer all results above to the case of what we call here pointwise control . for methods which select structures incrementally ,
i.e. for which @xmath115 for all @xmath116 , pointwise control and control with @xmath117 are equivalent since @xmath118 is then monotonically increasing with decreasing @xmath24 for all @xmath60 .
vitamin gene - expression dataset .
the regularisation path of graphical lasso ( top row ) and the corresponding point - wise stability selected models ( bottom row ) .
_ , scaledwidth=90.0% ] the same plot as in figure [ fig : eggs ] but with the variables ( expression values of each gene ) permuted independently .
the empty graph is the true model . with stability selection ,
only a few errors are made , as guaranteed by the made error control .
_ , scaledwidth=90.0% ] stability selection is also promising for graphical modelling . here
we focus on gaussian graphical models as described in section [ subsec.prelim ] around formula ( [ ggm ] ) and ( [ glasso ] ) .
the pattern of non - zero entries in the inverse covariance matrix @xmath119 corresponds to the edges between the corresponding pairs of variables in the associated graph and is equivalent to a non - zero partial correlation ( or conditional dependence ) between such pairs of variables @xcite .
there has been interest recently in using @xmath3-penalties for model selection in gaussian graphical models due to their computational efficiency for moderate and large graphs @xcite .
here we work with the graphical lasso @xcite , as applied to the data from 160 randomly selected genes from the vitamin gene - expression dataset ( without the response variable ) introduced in section [ subsec.examp1 ] .
we want to infer the set of non - zero entries in the inverse covariance matrix @xmath119 .
part of the resulting regularisation path of the graphical lasso showing graphs for various values of the regularisation parameter @xmath68 , i.e. @xmath120 where @xmath121 , are shown in the first row of figure [ fig : eggs ] . for reasons of display , variables ( genes ) are ordered first using hierarchical clustering and are symbolised by nodes arranged in a circle .
stability selection is shown in the bottom row of figure [ fig : eggs ] .
we pursue a pointwise control approach . for each value of @xmath68 , we select the threshold @xmath72 so as to guarantee @xmath122 , that is we expect fewer than 30 wrong edges among the 12720 possible edges in the graph .
the set @xmath123 varies remarkably little for the majority of the path and the choice of @xmath107 ( which is implied by @xmath68 ) does not seem to be critical , as already observed for variable selection in regression .
next , we permute the variables ( expression values ) randomly , using a different permutation for each variable ( gene ) .
the true graph is now the empty graph .
as can be seen from figure [ fig : eggsnull ] , stability selection selects now just very few edges or none at all ( as it should ) .
the top row shows the corresponding graphs estimated with the graphical lasso which yields a much poorer selection of edges .
stability selection demands to re - run @xmath124 multiple times . evaluating selection probabilities over 100 subsamples seems sufficient in practice .
the algorithmic complexity of lasso in ( [ lasso ] ) or in ( [ randomisedlasso ] ) below is of the order @xmath125 , see @xcite . in the @xmath126 regime , running the full lasso path on subsamples of size
@xmath127 is hence a quarter of the cost of running the algorithm on the full dataset and running 100 simulations is 25 times the cost of running a single fit on the full dataset .
this cost could be compared with the cost of cross - validation , as this is what one has to resort to often in practice to select the regularisation parameter . running 10-fold
cross - validation uses approximately @xmath128 as many computational resources as the single fit on the full dataset .
stability selection is thus roughly three times more expensive than 10-fold cv .
this analysis is based on the fact that the computational complexity scales like @xmath129 with the number of observations ( assuming @xmath130 ) .
if computational costs would scale linearly with sample size ( e.g. for lasso with @xmath131 ) , this factor would increase to roughly 5.5 .
stability selection with the lasso ( using 100 subsamples ) for a dataset with @xmath132 and @xmath133 takes about 10 seconds on a 2.2ghz processor , using the implementation of @xcite .
computational costs of this order would often seem worthwhile , given the potential benefits .
stability selection is a general technique , applicable to a wide range of applications , some of which we have discussed above . here , we want to discuss advantages and properties of stability selection for the specific application of variable selection in regression with high - dimensional data which is a well - studied topic nowadays @xcite .
we consider a linear model as in ( [ eq : linear ] ) with gaussian noise , @xmath134 with fixed @xmath16 design matrix @xmath12 and @xmath135 i.i.d .
the predictor variables are normalised with @xmath137 for all @xmath138 .
we allow for high - dimensional settings where @xmath139 .
stability selection is attractive for two reasons .
first , the choice of a proper regularisation parameter for variable selection is crucial and notoriously difficult , especially because the noise level is unknown . with stability selection ,
results are much less sensitive to the choice of the regularisation .
second , we will show that stability selection makes variable selection consistent in settings where the original methods fail .
we give general conditions under which consistent variable selection is achieved with stability selection .
consistent variable selection for a procedure @xmath93 is understood to be equivalent to @xmath140 it is clearly of interest to know under which conditions consistent variable selection can be achieved . in the high - dimensional context , this places a restriction on the growth of the number @xmath1 of variables and sparsity @xmath141 , typically of the form @xmath142 @xcite . while this assumption is often realistic , there are stronger assumptions on the design matrix that need to be satisfied for consistent variable selection . for lasso
, it amounts to the ` neighbourhood stability ' condition @xcite which is equivalent to the ` irrepresentable condition ' @xcite . for orthogonal matching pursuit ( which is essentially forward variable selection ) , the so - called ` exact recovery criterion ' @xcite is sufficient and necessary for consistent variable selection . here
, we show that these conditions can be circumvented more directly by using stability selection , also giving guidance on the proper amount of regularisation . due to the restricted length of the paper
, we will only discuss in detail the case of lasso whereas the analysis of orthogonal matching pursuit is just indicated .
an interesting aspect is that stability selection with the original procedures alone yields often very large improvements already .
moreover , when adding some extra sort of randomness in the spirit of random forests @xcite weakens considerably the conditions needed for consistent variables selection as discussed next .
the lasso @xcite estimator is given in ( [ lasso ] ) .
for consistent variable selection using @xmath143 , it turns out that the design needs to satisfy the so - called ` neighbourhood stability ' condition @xcite which is equivalent to the ` irrepresentable condition ' @xcite : @xmath144 the condition in ( [ irc ] ) is sufficient and ( almost ) necessary ( the word ` almost ' refers to the fact that a necessary relation is using ` @xmath145 ' instead of ` @xmath146 ' ) .
if this condition is violated , all one can hope for is recovery of the regression vector @xmath4 in an @xmath147-sense of convergence by achieving @xmath148 for @xmath149 .
the main assumption here are bounds on the sparse eigenvalues as discussed below .
this type of @xmath147-convergence can be used to achieve consistent variable selection in a two - stage procedure by thresholding or , preferably , the adaptive lasso @xcite .
the disadvantage of such a two - step procedure is the need to choose several tuning parameters without proper guidance on how these parameters can be chosen in practice .
we propose the randomised lasso as an alternative . despite its simplicity ,
it is consistent for variable selection even though the ` irrepresentable condition ' in ( [ irc ] ) is violated .
randomised lasso is a new generalisation of the lasso .
while the lasso penalises the absolute value @xmath150 of every component with a penalty term proportional to @xmath24 , the randomised lasso changes the penalty @xmath24 to a randomly chosen value in the range @xmath151 $ ] .
+ a proposal for the distribution of the weights @xmath152 is described below , just before theorem [ theo : randlasso ] .
the word ` weakness ' is borrowed from the terminology of weak greedy algorithms @xcite which are loosely related to our randomised lasso .
implementation of ( [ randomisedlasso ] ) is straightforward by appropriate re - scaling of the predictor variables ( with scale factor @xmath152 for the @xmath33-th variable ) .
using these re - scaled variables , the standard lasso is solved , using for example the lars algorithm @xcite or fast coordinate wise approaches @xcite .
the perturbation of the penalty weights is reminiscent of the re - weighting in the adaptive lasso @xcite . here
, however , the re - weighting is not based on any previous estimate , but is simply chosen at random ! as such , it is very simple to implement .
however , it seems nonsensical at first sight since one can surely not expect any improvement from such a random perturbation .
if applied only with one random perturbation , randomised lasso is not very useful .
however , applying randomised lasso many times and looking for variables that are chosen often will turn out to be a very powerful procedure . [
[ consistency - for - randomised - lasso - with - stability - selection . ] ] consistency for randomised lasso with stability selection .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + for stability selection with randomised lasso , we can do without the irrepresentable condition ( [ irc ] ) but need only a condition on the sparse eigenvalues of the design @xcite , also called the sparse riesz condition in @xcite . for any @xmath153 , let @xmath154 be the restriction of @xmath12 to columns in @xmath155 .
the minimal sparse eigenvalue @xmath156 is then defined for @xmath157 as @xmath158 and analogously for the maximal sparse eigenvalue @xmath159 .
we have to constrain sparse eigenvalues to succeed .
[ assum : sparse ] there exists some @xmath160 and some @xmath161 such that @xmath162 this assumption ( [ sparseassum ] ) is related to the sparse riesz condition in @xcite .
the equivalent condition there requires the existence of some @xmath163 such that @xmath164 compare with remark 2 in @xcite .
this assumption essentially requires that maximal and minimal eigenvalues , for a selection of order @xmath165 variables , are bounded away from 0 and @xmath166 respectively . in comparison ,
our assumption is significantly stronger than ( [ condzhang ] ) , but at the same time typically much weaker than the standard assumption of the ` irrepresentable condition ' necessary to get results comparable to ours .
we have not specified the exact form of perturbations we will be using for the randomised lasso in ( [ randomisedlasso ] ) . for the following , we consider the randomised lasso of ( [ randomisedlasso ] ) , where the weights @xmath152 are sampled independently as @xmath167 with probability @xmath168 and @xmath169 otherwise .
other perturbations are certainly possible and work often just as well in practice .
[ theo : randlasso ] consider the model in ( [ ymodel ] ) .
for randomised lasso , let the weakness @xmath104 be given by @xmath170 , for any @xmath171 , and @xmath172 .
let @xmath173 be a sequence with @xmath174 for @xmath149 .
let @xmath175 .
assume that @xmath92 and @xmath176 and that the sparse eigenvalue assumption [ assum : sparse ] is satisfied .
then there exists some @xmath177 such that for all @xmath178 , stability selection with the randomised lasso satisfies on a set @xmath179 with @xmath180 that no noise variables are selected , @xmath181 where @xmath182 with @xmath183 . on the same set @xmath179 , @xmath184 where @xmath185 .
this implies that all variables with sufficiently large regression coefficient are selected . under the condition that the minimal non - zero regression coefficient is bounded from below by @xmath186 as a consequence of theorem [ theo : randlasso ] , @xmath187 i.e. consistent variable selection for @xmath188 ( @xmath189 or @xmath190 ) in the sense of ( [ consvar ] ) even if the irrepresentable condition ( [ irc ] ) is violated .
if no such lower bound holds , the set of selected variables might miss variables with too small regression coefficients , which are , by definition , in the set @xmath191 .
[ rem : conserv ] theorem [ theo : randlasso ] is valid for all @xmath192 .
this is noteworthy as it means that even if the value of @xmath68 is chosen too large ( i.e. considerably larger than @xmath112 ) , no noise variables will be selected ( formula ( [ negative ] ) ) .
only some important variables might be missed .
this effect has been seen in the empirical examples as stability selection is very insensitive to the choice of @xmath68 .
in contrast , a hard - thresholded solution of the lasso with a value of @xmath68 too large will lead to the inclusion of noise variables .
thus , stability selection with the randomised lasso exhibits an important property of being conservative and guarding against false positive selections .
theorem [ theo : randlasso ] is derived under random perturbations of the weights .
while this achieves good empirical results , it seems more advantageous in combination with with subsampling of the data .
the results extend directly to this case .
let @xmath193 be the selection probability of variable @xmath194 , while doing both random weight perturbations and subsampling @xmath127 out of @xmath2 observations .
the probability that @xmath193 is above the threshold @xmath195 is bounded by a markov - type inequality from below by @xmath196 having used that @xmath197 as a consequence of theorem [ theo : randlasso ] .
if @xmath198 is sufficiently small in comparison to @xmath199 , this elementary inequality implies that important variables in @xmath200 are still chosen by stability selection ( subsampling and random weights perturbation ) with very high probability .
a similar argument shows that noise variables are also still not chosen with very high probability .
empirically , combining random weight perturbations with subsampling yields very competitive results and this is what we recommend to use .
there is an inherent tradeoff when choosing the weakness @xmath104 .
a negative consequence of a low @xmath104 is that the design can get closer to singularity and can thus lead to unfavourable conditioning of the weighted design matrix . on the other hand ,
a low value of @xmath104 makes it less likely that irrelevant variables are selected .
this is a surprising result but rests on the fact that irrelevant variables can only be chosen if the corresponding irrepresentable condition ( [ irc ] ) is violated . by randomly perturbing the weights with a low @xmath104
, this condition is bound to fail sometimes , lowering the selection probabilities for such variables .
a low value of @xmath104 will thus help stability selection to avoid selecting noise variables with a violated irrepresentable condition ( [ irc ] ) . in practice ,
choosing @xmath104 in the range of @xmath201 gives very useful results .
[ [ relation - to - other - work . ] ] relation to other work .
+ + + + + + + + + + + + + + + + + + + + + + + in related and very interesting work , @xcite has proposed ` bolasso ' ( for bootstrapped enhanced lasso ) and shown that using a finite number of subsamples of the original lasso procedure and applying basically stability selection with @xmath90 yields consistent variables selection under the condition that the penalty parameter @xmath68 vanishes faster than typically assumed , at rate @xmath202 , and that the model dimension @xmath1 is fixed . while the latter condition could possibly be technical
only , the first distinguishes it from our results .
applying stability selection to randomised lasso , no false variable is selected for all sufficiently large values of @xmath68 , see remark [ rem : conserv ] . in other words , if @xmath68 is chosen ` too large ' with randomised lasso , only truly relevant variable are chosen ( though a few might be missed ) . if @xmath68 is chosen too large with bolasso , noise variables might be picked up .
figure [ fig : ex3 ] is a good illustration .
picking the regularisation in the left plot ( without extra randomness ) to select the correct model is much harder than in the right plot , where extra randomness is added .
the same distinction can be made with two - stage procedures like adaptive lasso @xcite or hard - thresholding @xcite , where variables are thresholded .
picking @xmath68 too large ( and @xmath68 is notoriously difficult to choose ) , false variables will invariably enter the model .
in contrast , stability selection with randomised lasso is not picking wrong variables if @xmath68 is chosen too large . the stability paths for randomised lasso with stability selection using weakness parameters @xmath203 ( left panel identical to the original lasso ) and @xmath204 ( middle ) and @xmath62 ( right ) .
red solid lines are the coefficients of the first two ( relevant variables ) .
the blue broken line is the coefficient of the third ( irrelevant ) variable and the dotted lines are the coefficients from all other ( irrelevant ) variables .
introducing the randomised version helps avoid choosing the third ( irrelevant ) predictor variable .
_ , title="fig:",scaledwidth=32.0% ] the stability paths for randomised lasso with stability selection using weakness parameters @xmath203 ( left panel identical to the original lasso ) and @xmath204 ( middle ) and @xmath62 ( right ) .
red solid lines are the coefficients of the first two ( relevant variables ) .
the blue broken line is the coefficient of the third ( irrelevant ) variable and the dotted lines are the coefficients from all other ( irrelevant ) variables .
introducing the randomised version helps avoid choosing the third ( irrelevant ) predictor variable .
_ , title="fig:",scaledwidth=32.0% ] the stability paths for randomised lasso with stability selection using weakness parameters @xmath203 ( left panel identical to the original lasso ) and @xmath204 ( middle ) and @xmath62 ( right ) .
red solid lines are the coefficients of the first two ( relevant variables ) .
the blue broken line is the coefficient of the third ( irrelevant ) variable and the dotted lines are the coefficients from all other ( irrelevant ) variables .
introducing the randomised version helps avoid choosing the third ( irrelevant ) predictor variable .
_ , title="fig:",scaledwidth=32.0% ] we illustrate the results on randomised lasso with a small simulation example : @xmath205 and the predictor variables are sampled from a @xmath206 distribution , where @xmath207 is the identity matrix , except for the entries @xmath208 and their symmetrical counterparts .
we use a regression vector @xmath209 . the response @xmath11 is obtained from the linear model @xmath210 in ( [ eq : linear ] ) , where @xmath211 i.i.d .
@xmath212 . for @xmath213 , the irrepresentable condition in ( [ irc ] )
is violated and lasso is not able to correctly identify the first two variables as the truly important ones , since it always includes the third variable superfluously as well . using the randomised version for lasso , the two relevant variables are still chosen with probability close to 1 , while the irrelevant third variable is only chosen with much lower probability ; the corresponding probabilities are shown for randomised lasso in figure [ fig : ex3 ] .
this allows to separate relevant and irrelevant variables . and
indeed , the randomised lasso is consistent under stability selection .
an interesting alternative to lasso or greedy forward search in this context are the recently proposed forward - backward search foba @xcite and the mc+ algorithm @xcite , which both provably lead to consistent variable selection under weak conditions on sparse eigenvalues , despite being greedy solutions to non - convex optimisation problems
. it will be very interesting to explore the effect of stability selection on these algorithms , but this is beyond the scope of this paper . here
, we look instead at orthogonal matching pursuit , a greedy forward search in the variable space .
the iterative sis procedure @xcite , entails orthogonal matching pursuit as a special case .
we will examine the effect of stability selection under subsampling and additional randomisation . to have a clear definition of randomised orthogonal matching pursuit ( romp ) ,
we define it as follows .
+ a drawback of omp is clearly that conditions for consistent variable selection are quite strong .
following @xcite , the exact recovery condition for omp is defined as @xmath214 this is a sufficient condition for consistent variable selection .
if it is not fulfilled , there exist regression coefficients that cause omp or its weak variant to fail in recovery of the exact set @xmath9 of relevant variables .
surprisingly , this condition is rather similar to the irrepresentable @xcite or neighbourhood stability condition @xcite . in the spirit of theorem [ theo : randlasso ] , we have also a proof that stability selection for randomised orthogonal matching pursuit ( romp ) is asymptotically consistent for variable selection in linear models , even if the right hand side in ( [ erc ] ) is not bounded by 1 but instead by a possibly large constant ( assuming the weakness @xmath104 is low enough ) .
this indicates that stability selection has a more general potential for improved structure estimation , beyond the case for the lasso presented in theorem [ theo : randlasso ] .
it is noteworthy that our proof involves artificial adding of noise covariates . in practice
, this seems to help often but a more involved discussion is beyond the scope of this paper .
we will give empirical evidence for the usefulness of stability selection under subsampling and additional randomisation for orthogonal matching pursuit in the numerical examples below .
probability to select @xmath215 and @xmath216 important variables without selecting a noise variable with the lasso in the regression setting ( dark red bar ) and stability selection under subsampling ( light grey bar ) for the 64 different settings .
black crosses mark the result for stability selection with additional randomisation ( @xmath204 ) .
_ , scaledwidth=99.0% ] the equivalent plot to fig .
[ lasso ] for lasso applied to classification ( top two rows ) and omp applied to regression ( bottom two rows ) .
_ , title="fig:",scaledwidth=99.0% ] the equivalent plot to fig .
[ lasso ] for lasso applied to classification ( top two rows ) and omp applied to regression ( bottom two rows ) .
_ , title="fig:",scaledwidth=99.0% ] comparison of stability selection with cross - validation for the real datasets ( f ) and ( g ) . the cross - validated solution ( for standard lasso )
is indicated by a dot and the corresponding stability selection ( for randomised lasso , @xmath204 on the left and @xmath203 on the right ) by a red triangle , showing the average proportion of correctly identified relevant variables versus the average number of falsely selected variables .
each pair consisting of a dot and triangle corresponds to a simulation setting ( some specified snr and @xmath165 ) .
the broken vertical line indicates the value at which the number of wrongly selected variables is controlled , namely @xmath217 .
looking at stability selection , the proportion of correctly identified relevant variables is very close to the cv - solution , while the number of falsely selected variables is reduced dramatically .
_ , title="fig:",scaledwidth=47.5% ] comparison of stability selection with cross - validation for the real datasets ( f ) and ( g ) .
the cross - validated solution ( for standard lasso ) is indicated by a dot and the corresponding stability selection ( for randomised lasso , @xmath204 on the left and @xmath203 on the right ) by a red triangle , showing the average proportion of correctly identified relevant variables versus the average number of falsely selected variables .
each pair consisting of a dot and triangle corresponds to a simulation setting ( some specified snr and @xmath165 ) .
the broken vertical line indicates the value at which the number of wrongly selected variables is controlled , namely @xmath217 .
looking at stability selection , the proportion of correctly identified relevant variables is very close to the cv - solution , while the number of falsely selected variables is reduced dramatically .
_ , title="fig:",scaledwidth=47.5% ] to investigate further the effects of stability selection , we focus here on the application of stability selection to lasso and randomised lasso for both regression and the natural extension to classification
. the effect on omp and randomised omp will also be examined . for regression ( lasso and omp ) , we generate observations by @xmath210 . for classification , we use the logistic linear model under the binomial family . to generate the design matrices @xmath12 ,
we use two real and five simulated datasets , a. independent predictor variables .
all @xmath132 predictor variables are i.i.d .
standard normal distributed .
sample size @xmath133 and @xmath218 .
b. block structure with 10 blocks .
the @xmath132-dimensional predictor variable follows a @xmath206 distribution , where @xmath219 for all pairs @xmath220 except if @xmath221 , for which @xmath222 .
sample size @xmath223 and @xmath218 . c. toeplitz design .
the @xmath132-dimensional predictor variable follows a @xmath206 distribution , where @xmath224 and @xmath225 .
sample size @xmath223 and @xmath218 .
d. [ d ] factor model with 2 factors .
let @xmath226 be two latent variables following i.i.d .
standard normal distributions .
each predictor variable @xmath29 , for @xmath60 , is generated as @xmath227 , where @xmath228 have i.i.d .
standard normal distributions for all @xmath60 .
sample sizes are @xmath223 and @xmath218 , while @xmath132 .
e. identical to ( [ d ] ) but with 10 instead of 2 factors
. f. motif regression dataset . a dataset ( @xmath229 and @xmath230 ) about finding transcription factor binding sites ( motifs ) in dna sequences .
the real - valued predictor variables are abundance scores for @xmath1 candidate motifs ( for each of the genes ) .
our dataset is from a heat - shock experiment with yeast . for a general description and motivation about motif regression
we refer to @xcite .
g. the already mentioned vitamin gene expression data ( with @xmath64 and @xmath231 ) described in section [ subsec.examp1 ] .
we do not use the response values from the real datasets , however , as we need to know which variables are truly relevant or irrelevant . to this end
, we create sparse regression vectors by setting @xmath232 for all @xmath60 , except for a randomly chosen set @xmath9 of coefficients , where @xmath233 is chosen independently and uniformly in @xmath234 $ ] for all @xmath235 .
the size @xmath236 of the active set is varied between 4 and 50 , depending on the dataset . for regression ,
the noise vector @xmath237 is chosen i.i.d .
@xmath238 , where the rescaling of the variance with @xmath2 is due to the rescaling of the predictor variables to unit norm , i.e. @xmath239 . the noise level @xmath240 is chosen to achieve signal - to - noise ratios ( snr ) of @xmath241 and @xmath242 . for classification , we scale the vector @xmath4 to achieve a given bayes misclassification rate , either @xmath243 or @xmath244 .
each of the 64 scenarios is run 100 times , once using the standard procedure ( lasso or omp ) , once using stability selection with subsampling and once using stability selection with subsampling and additional randomisation ( @xmath204 for the randomised lasso and @xmath245 for randomised omp ) .
the methods are thus in total evaluated on about 20.000 simulations each .
the solution of stability selection can not be reproduced by simply selecting the right penalty with lasso , since stability selection provides a fundamentally new solution . to compare the power of both approaches , we look at the probability that @xmath246 of the @xmath165 relevant variables can be recovered without error , where @xmath247 . a set of @xmath248 variables
is said to be recovered successfully for the lasso or omp selection , if there exists a regularisation parameter such that at least @xmath249 variables in @xmath9 have a non - zero regression coefficient and all variables in @xmath250 have a zero regression coefficient .
for stability selection , recovery without error means that the @xmath251 variables with highest selection probability @xmath252 are all in @xmath9 .
the value @xmath112 is chosen such that at most @xmath253 variables are selected in the whole path of solutions for @xmath254 .
note that this notion neglects the fact that the most advantageous regularisation parameter is selected here automatically for lasso and omp but not for stability selection .
results are shown in figure [ fig : lasso ] for lasso applied to regression , and in figure [ fig : class_omp ] for lasso applied to classification and omp applied to regression again . in figure
[ fig : lasso ] , we also give the median number of variables violating the irrepresentable condition ( denoted by ` violations ' ) and the average of the maximal correlation between a randomly chosen variable and all other variables ( ` max cor ' ) as two measures of the difficulty of the problem .
stability selection identifies as many or more correct variables than the underlying method itself in all cases except for scenario ( a ) , where it is about equivalent . that stability selection is not advantageous for scenario ( a ) is to be expected as the design is nearly orthogonal ( very weak empirical correlations between variables ) , thus almost decomposing into @xmath1 univariate decisions and we would not expect stability selection to help in a univariate framework
. often the gain of stability selection under subsampling is substantial , irrespective of the sparsity of the signal and the signal - to - noise - ratio .
additional randomisation helps in cases where there are many variables violating the irrepresentable condition ; for example in setting ( e ) .
this is in line with our theory .
next , we test how well the error control of theorem [ theo : error ] holds up for these datasets . for the motif regression dataset ( f ) and the vitamin gene expression dataset ( g ) ,
lasso is applied , with randomisation and without . for both datasets
, the signal - to - noise ratio is varied between 0.5 , 1 and 2 .
the number of non - zero coefficients @xmath165 is varied in steps of 1 between 1 and 12 , with a standard normal distribution for the randomly chosen non - zero coefficients .
each of the 72 settings is run 20 times .
we are interested in the comparison between the cross - validated solution and stability selection . for stability selection , we chose @xmath255 and thresholds of @xmath256 , corresponding to a control of @xmath217 , where @xmath79 is the number of wrongly selected variables .
the control is mathematically derived under the assumption of exchangeability for the distribution of noise variables , see theorem [ theo : error ] .
this assumption is most likely not fulfilled for the given dataset and it is of interest to see how well the bound holds up for real data .
results are shown in figure [ fig : motifs ] .
stability selection reduces the number of falsely selected variables dramatically , while maintaining almost the same power to detect relevant variables .
the number of falsely chosen variables is remarkably well controlled at the desired level , giving empirical evidence that the derived error control is useful beyond the discussed setting of exchangeability .
stability selection thus helps to select a useful amount of regularisation .
stability selection addresses the notoriously difficult problem of structure estimation or variable selection , especially for high - dimensional problems .
cross - validation fails often for high - dimensional data , sometimes spectacularly .
stability selection is based on subsampling in combination with ( high - dimensional ) selection algorithms .
the method is extremely general and we demonstrate its applicability for variable selection in regression and gaussian graphical modelling .
stability selection provides finite sample familywise multiple testing error control ( or control of other error rates of false discoveries ) and hence a transparent principle to choose a proper amount of regularisation for structure estimation or variable selection .
furthermore , the solution of stability selection depends surprisingly little on the chosen initial regularisation .
this is an additional great benefit besides error control .
another property of stability selection is the improvement over a pre - specified selection method .
it is often the case that computationally efficient algorithms for high - dimensional selection are inconsistent , even in rather simple settings .
we prove for randomised lasso that stability selection will be variable selection consistent even if the necessary conditions needed for consistency of the original method are violated . and thus , stability selection will asymptotically select the right model in scenarios where lasso fails .
in short , stability selection is the marriage of subsampling and high - dimensional selection algorithms , yielding finite sample familywise error control and markedly improved structure estimation .
both of these main properties are demonstrated on simulated and real data .
an alternative to subsampling is sample splitting . instead of observing
if a given variable is selected for a random subsample , one can look at a random split of the data into two non - overlapping samples of equal size @xmath55 and see if the variable is chosen in both sets simultaneously .
let @xmath257 and @xmath258 be two random subsets of @xmath54 with @xmath259 for @xmath260 and @xmath261 .
define the simultaneously selected set as the intersection of @xmath262 and @xmath263 , @xmath264 define the simultaneous selection probabilities @xmath265 for any set @xmath153 as @xmath266 where the probability @xmath59 is with respect to the random sample splitting ( and any additional randomness if @xmath49 is a randomised algorithm ) .
we work with the selection probabilities based on subsampling but the following lemma lets us convert these probabilities easily into simultaneous selection probabilities based on sample splitting ; the latter is used for the proof of theorem [ theo : error ] .
the bound is rather tight for selection probabilities close to 1 .
[ lemma : bound ] for any set @xmath153 , a lower bound for the simultaneous selection probabilities is given by , for every @xmath267 , by @xmath268 _ proof .
_ let @xmath257 and @xmath258 be the two random subsets in sample splitting of @xmath54 with @xmath259 for @xmath260 and @xmath261 . denote by @xmath269 the probability @xmath270 .
note that the two events are not independent as the probability is only with respect to a random split of the fixed samples @xmath54 into @xmath257 and @xmath258 .
the probabilities @xmath271 are defined equivalently by @xmath272 , @xmath273 , and @xmath274 .
note that @xmath275 and @xmath276 it is obvious that @xmath277 . as @xmath278 , it also follows that @xmath279 .
hence @xmath280 which completes the proof .
@xmath281 the proof uses mainly lemma [ lemma : markov ] .
we first show that @xmath282 for all @xmath283 , using the made definitions @xmath284 and @xmath285 .
define furthermore @xmath286 to be the set of noise variables ( in @xmath75 ) which appear in @xmath287 and analogously @xmath288 .
the expected number of falsely selected variables can be written as @xmath289 . using the assumption ( [ btrg ] ) ( which asserts that the method is not worse than random guessing )
, it follows that @xmath290 .
putting together , @xmath291 and hence @xmath292 . using the exchangeability assumption
, we have @xmath293 for all @xmath283 and hence , for @xmath283 , it holds that @xmath294 , as desired .
note that this result is independent of the sample size used in the construction of @xmath49 , @xmath47 .
now using lemma [ lemma : markov ] below , it follows that @xmath295 for all @xmath296 and @xmath283 . using lemma [ lemma : bound ]
, it follows that @xmath297 .
hence @xmath298 , which completes the proof .
@xmath281 [ lemma : markov ]
let @xmath299 and @xmath49 the set of selected variables based on a sample size of @xmath300 . if @xmath301 , then @xmath302 if @xmath303 for some @xmath304 , then @xmath305 _ proof .
_ let @xmath306 be , as above , the random split of the samples @xmath54 into two disjoint subsets , where both @xmath307 for @xmath260 .
define the binary random variable @xmath308 for all subsets @xmath153 as @xmath309 denote the data ( the @xmath2 samples ) by @xmath310 .
the simultaneous selection probability @xmath311 , as defined in ( [ simult ] ) , is then @xmath312 where the expectation @xmath313 is with respect to the random split of the @xmath2 samples into sets @xmath257 and @xmath258 ( and additional randomness if @xmath49 is a randomised algorithm ) . to prove the first part
, the inequality @xmath314 ( for a sample size @xmath300 ) , implies that @xmath315 and hence @xmath316 therefore , @xmath317 using a markov - type inequality , @xmath318 thus @xmath319 , completing the proof of the first claim .
the proof of the second part follows analogously .
@xmath281 instead of working directly with form ( [ randomisedlasso ] ) of the randomised lasso estimator , we consider the equivalent formulation of the standard lasso estimator , where all variables have initially unit norm and are then rescaled by their random weights w. [ def : additionala ] for weights @xmath320 as in ( [ randomisedlasso ] ) , let @xmath321 be the matrix of re - scaled variables , with @xmath322 for each @xmath60 .
let @xmath323 and @xmath324 be the maximal and minimal eigenvalues analogous to ( [ phimin ] ) for @xmath321 instead of @xmath12 .
the proof rests mainly on the two - fold effect a weakness @xmath325 has on the selection properties of the lasso .
the first effect is that the singular values of the design can be distorted if working with the reweighted variables @xmath321 instead of @xmath12 itself .
a bound on the ratio between largest and smallest eigenvalue is derived in lemma [ lemma : trans ] , effectively yielding a lower bound for useful values of @xmath104 .
the following lemma [ lemma : boundedq ] then asserts , for such values of @xmath104 , that the relevant variables in @xmath9 are chosen with high probability under any random sampling of the weights .
the next lemma [ lemma:1/2 ] establishes the key advantage of randomised lasso as it shows that the ` irrepresentable condition ' ( [ irc ] ) is sometimes fulfilled under randomly sampled weights , even though its not fulfilled for the original data .
variables which are wrongly chosen because condition ( [ irc ] ) is not satisfied for the original unweighted data will thus not be selected by stability selection .
the final result is established in lemma [ lemma : nofalse ] after a bound on the noise contribution in lemma [ lemma : boundnoise ] .
[ lemma : trans ] define @xmath326 by @xmath327 and assume @xmath176 .
let @xmath320 be weights generated randomly in @xmath328 $ ] , as in ( [ randomisedlasso ] ) , and let @xmath321 be the corresponding rescaled predictor variables , as in definition [ def : additionala ] .
for @xmath329 , with @xmath330 , it holds under assumption [ assum : sparse ] for all random realisations @xmath320 that @xmath331 _ proof .
_ using assumption [ assum : sparse ] , @xmath332 where the first inequality follows by assumption [ assum : sparse ] , the equality by @xmath333 and the second inequality by @xmath334 .
it follows that @xmath335 now , let @xmath336 be again the @xmath337-diagonal matrix with diagonal entries @xmath338 for all @xmath60 and 0 on the non - diagonal elements .
then @xmath339 and , taking suprema over all @xmath336 with diagonal entries in @xmath328 $ ] , @xmath340 where the last step follows by a change of variable transform @xmath341 and the fact that @xmath342 as well as @xmath343 and thus @xmath344 for all @xmath345 with diagonal entries in @xmath328 $ ] . the corresponding argument for @xmath346 yields the bound @xmath347 for all @xmath348 .
the claim ( [ boundphiw ] ) follows by observing that @xmath349 for @xmath176 , since @xmath350 by assumption [ assum : sparse ] and hence @xmath351 .
@xmath281 [ lemma : boundedq ] let @xmath352 be the set @xmath353 of selected variables of the randomised lasso with weakness @xmath354 $ ] and randomly sampled weights @xmath320 .
suppose that the weakness @xmath355 .
under the assumptions of theorem [ theo : randlasso ] , there exists a set @xmath356 in the sample space of @xmath11 with @xmath357 , such that for all realisations @xmath358 , for @xmath359 , if @xmath360 , @xmath361 where @xmath191 is defined as in theorem [ theo : randlasso ] .
_ follows mostly from theorem 1 in @xcite . to this end ,
set @xmath362 in their notation .
we also have @xmath363 , as , by definition , @xmath333 , as in lemma [ lemma : trans ] .
the quantity @xmath364 in @xcite is identical to our notation @xmath365 .
it is bounded for all random realisations of @xmath358 , as long as @xmath355 , using lemma [ lemma : trans ] , by @xmath366 hence all assumptions of theorem 1 in @xcite are fulfilled , with @xmath367 , for any random realisation @xmath358 . using ( 2.20)-(2.24 ) in @xcite , it follows that there exists a set @xmath356 in the sample space of @xmath11 with @xmath368 for all @xmath359 , such that if @xmath369 , from ( 2.21 ) in @xcite , @xmath370 and , from ( 2.23 ) in @xcite , @xmath371 having used for the first inequality that , in the notation of @xcite , @xmath372 .
the @xmath373 factor was omitted to account for our different normalisation . for the second inequality
, we used @xmath374 .
the last inequality implies , by definition of @xmath191 in theorem [ theo : randlasso ] , that @xmath375 , which completes the proof .
@xmath281 [ lemma:1/2 ] set @xmath172 .
let @xmath138 and let @xmath376 be a set which can depend on the random weight vector @xmath320 .
suppose that @xmath377 satisfies @xmath378 and @xmath379 for all realisations @xmath358 .
suppose furthermore that @xmath380 for some @xmath381 implies that @xmath382 for all pairs @xmath383 of weights that fulfill @xmath384 for all @xmath385 , with equality for all @xmath386 .
then , for @xmath387 , @xmath388 where the probability @xmath389 is with respect to random sampling of the weights @xmath320 and @xmath390 is , as above , the probability of choosing weight @xmath104 for each variable and @xmath391 the probability of choosing weight 1 .
_ let @xmath392 be the realisation of @xmath320 for which @xmath393 and @xmath394 for all other @xmath395 .
the probability of @xmath396 is clearly @xmath397 under the used sampling scheme for the weights .
let @xmath398 be the selected set of variables under these weights .
let now @xmath399 be the set of all weights for which @xmath400 and @xmath401 for all @xmath386 , and arbitrary values in @xmath402 for all @xmath403 with @xmath404 .
the probability for a random weight being in this set is @xmath405 . by the assumption on @xmath155
, it holds that @xmath380 for all @xmath406 , since @xmath407 for all @xmath408 with equality for @xmath386 . for all weights @xmath406
, it follows moreover that @xmath409 using the bound on @xmath104 , it hence only remains to be shown that , if @xmath410 for all @xmath411 , @xmath412 since @xmath413 for any vector @xmath414 , it is sufficient to show , for @xmath415 , @xmath416 as @xmath417 is the projection of @xmath29 into the space spanned by @xmath418 and @xmath419 , it holds that @xmath420 . using @xmath421
, it follows that @xmath422 , which shows ( [ toshowww ] ) and thus completes the proof .
@xmath281 [ lemma : boundnoise ]
let @xmath423 be the projection into the space spanned by all variables in subset @xmath381 .
suppose @xmath92 .
then there exists a set @xmath424 with @xmath425 , such that for all @xmath426 , @xmath427 _ proof .
_ let @xmath428 be the event that @xmath429 as entries in @xmath430 are i.i .
@xmath136 distributed , @xmath431 for all @xmath432 .
note that , for all @xmath433 and @xmath434 , @xmath435 .
define @xmath436 as @xmath437 it is now sufficient to show that @xmath438 . showing this bound
is related to a bound in @xcite and we repeat a similar argument .
each term @xmath439 has a @xmath440 distribution as long as @xmath441 is of full rank @xmath442 .
hence , using the same standard tail bound as in the proof of theorem 3 of @xcite , @xmath443 having used @xmath444 for all @xmath445 in the last step and thus , using @xmath446 , @xmath447 which completes the proof by setting @xmath448 and concluding that @xmath425 for all @xmath445 . @xmath281 [ lemma : nofalse]let @xmath449 and @xmath450 be again the probability for variable @xmath33 of being in the selected subset , with respect to random sampling of the weights @xmath320 .
then , under the assumptions of theorem [ theo : randlasso ] , for all @xmath451 and @xmath92 , there exists a set @xmath179 with @xmath452 such that for all @xmath453 and @xmath454 , @xmath455 where @xmath191 is defined as in theorem [ theo : randlasso ] .
_ we let @xmath456 , where @xmath356 is the event defined in lemma [ lemma : boundedq ] and event @xmath424 is defined in lemma [ lemma : boundnoise ] . since , using these two lemmas , @xmath457 it is sufficient to show ( [ toshowif1 ] ) and ( [ toshowif2 ] ) for all @xmath458 .
we begin with ( [ toshowif1 ] ) .
a variable @xmath451 is in the selected set @xmath352 only if @xmath459 where @xmath460 is the solution to ( [ randomisedlasso ] ) with the constraint that @xmath461 , comparable to the analysis in @xcite .
let @xmath462 be the set of non - zero coefficients and @xmath463 be the set of regression coefficients which are either truly non - zero or estimated as non - zero ( or both ) .
we will use @xmath464 as a short - hand notation for @xmath465 .
let @xmath466 be the projection operator into the space spanned by all variables in the set @xmath464 . for all @xmath358 , this is identical to @xmath467 then , splitting the term @xmath468 in ( [ tmp1 ] ) into the two terms @xmath469 it holds for the right term in ( [ twoterms ] ) that @xmath470 looking at the left term in ( [ twoterms ] ) , since @xmath369 , we know by lemma [ lemma : boundedq ] that @xmath471 and , by definition of @xmath464 above , @xmath472 . thus the left term in ( [ twoterms ] ) is bounded from above by @xmath473 having used lemma [ lemma : boundnoise ] in the last step and @xmath474 .
putting together , the two terms in ( [ twoterms ] ) are bounded , for all @xmath475 , by @xmath476 we now apply lemma [ lemma:1/2 ] to the rightmost term .
the set @xmath464 is a function of the weight vector and satisfies for every realisation of the observations @xmath369 the conditions in lemma [ lemma:1/2 ] on the set @xmath377 .
first , @xmath477 .
second , by definition of @xmath464 above , @xmath478 for all weights @xmath479 .
third , it follows by the kkt conditions for lasso that the set of non - zero coefficients of @xmath480 and @xmath481 is identical for two weight vectors @xmath479 and @xmath482 , as long @xmath483 for all @xmath484 and @xmath485 for all @xmath486 ( increasing the penalty on zero coefficients will leave them at zero , if the penalty for non - zero coefficients is kept constant ) .
hence there exists a set @xmath487 in the sample space of @xmath320 with @xmath488 such that @xmath489 .
moreover , for the same set @xmath487 , we have @xmath490 .
hence , for all @xmath475 and , for all @xmath491 , the lhs of ( [ tmp1 ] ) is bounded from above by @xmath492 and variable @xmath451 is hence not part of the set @xmath493 .
it follows that @xmath494 with @xmath495 for all @xmath451 .
this completes the first part ( [ toshowif1 ] ) of the proof . for the second part ( [ toshowif2 ] )
, we need to show that , for all @xmath496 , all variables @xmath33 in @xmath9 are chosen with probability at least @xmath497 ( with respect to random sampling of the weights @xmath320 ) , except possibly for variables in @xmath498 , defined in theorem [ theo : randlasso ] . for all @xmath499 , however , it follows directly from lemma [ lemma : boundedq ] that @xmath500 .
hence , for all @xmath194 , the selection probability satisfies @xmath501 for all @xmath369 , which completes the proof .
@xmath281 since the statement in lemma [ lemma : nofalse ] is a reformulation of the assertion of theorem [ theo : randlasso ] , the proof of the latter is complete .
both authors would like to thank anonymous referees for many helpful comments and suggestions which greatly helped to improve the manuscript .
n.m . would like to thank fim ( forschungsinstitut fr mathematik ) at eth zrich for support and hospitality . | estimation of structure , such as in variable selection , graphical modelling or cluster analysis is notoriously difficult , especially for high - dimensional data .
we introduce stability selection .
it is based on subsampling in combination with ( high - dimensional ) selection algorithms . as such ,
the method is extremely general and has a very wide range of applicability .
stability selection provides finite sample control for some error rates of false discoveries and hence a transparent principle to choose a proper amount of regularisation for structure estimation .
variable selection and structure estimation improve markedly for a range of selection methods if stability selection is applied .
we prove for randomised lasso that stability selection will be variable selection consistent even if the necessary conditions needed for consistency of the original lasso method are violated .
we demonstrate stability selection for variable selection and gaussian graphical modelling , using real and simulated data . |
the need for a self - consistent , practical transport theory that can handle dynamic populations of relativistic particles extends across many disciplines in physics .
most obviously , astrophysical plasmas in the presence of strong gravitational and/or magnetic fields are often out of thermal equilibrium when the relevant dynamic time scale ( e.g. , associated with the process of accretion onto a compact object ) is short compared to the time required for the particles to interact internally with each other . situations in which this may occur include solar flares ( petrosian and liu 2003 ) , accretion onto supermassive black holes in the nuclei of active galaxies ( see , e.g. , melia and falcke 2001 ) , and re - acceleration of already relativistic particles in shocks produced during the merger of galaxies ( brunetti et al .
2004 ) . in all cases ,
the resultant photon spectrum produced by the energized particles is most simply explained in terms of a non - maxwellian distribution .
less directly , an argument can be made that the two dominant particle species ( say , electrons and protons in a fully ionized hydrogen plasma ) separate in energy space , leading to a situation in which the ion temperature is larger than that of the electrons ( see , e.g. , shapiro , lightman , and eardley 1976 ) .
however , whether such a two - temperature plasma may be maintained against collisional re - equilibration is still actively debated , and may be critical to the question of how a black hole accretes from its environment .
for example , there is now very good evidence that the gas orbiting the supermassive black hole , sgr a * , at the center of our galaxy attains a temperature in excess of @xmath0 k ( liu and melia 2001 ) , but it is still not understood whether this optically thin plasma maintains a single temperature .
in addition , nonthermal emission in sgr a * ( see , e.g. , ozel , psaltis , and narayan 2000 ; yuan , quataert , and narayan 2003 ) has been invoked to explain its low - frequency radio spectrum in accretion - dominated models .
these clear astrophysical applications should be aided by recent experiments with laser - heated plasmas , which are now probing the energy regimes of interest .
phenomena unique to relativistic distributions , such as their tendency to limit heat flux ( bell et al .
1981 ; rickard et al . 1989 ) , or the ability of rapid acceleration to create bi - maxwellian electron distributions ( guethlein et al . 1996 ) have now been observed in laboratory plasmas .
experimental constraints arising from both the astrophysical and laboratory contexts , acting in concert , should yield a solid , well - tested theory .
astrophysicists have had a theory for relativistic energy transport for some time .
dewar et al . (
1979 ) were followed by dermer and liang ( 1989 ) in calculating the energy exchanged between particles during a relativistic binary collision .
nayakshin and melia ( 1998 ) used these coefficients as the basis of a fokker - planck equation in energy .
but , as dewar et al . note , virtually all of the quantities that were invariant in the non - relativistic treatment cease to be so in the relativistic formulation .
much care ( must ) be taken to show how the various physical quantities ... transform . "
the resulting theory is not as transparent as one would like .
in contrast , the strategy we adopt here follows beliaev and budker ( 1955 ) in formulating kinetic theory in a covariant form valid in any frame of reference .
this approach is simpler , since the frame invariants are obvious .
it is also general : we present a solution that handles anisotropic or radically out - of - equilibrium distributions , and can be used to find electric and thermal conduction coefficients as well as energy exchange .
this approach also conveniently and elegantly reduces to the well - known rosenbluth ( 1957 ) theory in the non - relativistic limit .
relativistic distributions obey _ qualitatively _ different physics , since the covariant phase volume must incorporate energy to form an eight - dimensional phase space .
but the covariant theory is also _
quantitatively _ sufficiently different to shape observations .
we show here that a fully relativistic distribution equilibrates in a tenth to a hundredth the time given by non - relativistic theory .
the heat flux in a relativistic distribution , relative to that for the non - relativistic case , is inhibited by a factor of @xmath1 .
this is due to a drift velocity which approaches the speed of light asymptotically .
thus , situations where a nonthermal tail is likely to be present ( see , e.g. , blasi 2000 ; we will examine this situation in greater detail below ) require a covariant formulation .
solutions to covariant kinetic theory have existed for some time , too .
braams and karney ( 1987 ) made a spherical expansion of the kinetic kernel and found relativistic conductivities ( 1989 ) , but only in very limited circumstances .
this was followed by attempts at solution via a chapman - enskog expansion ( mohanty and baral 1996 ) , legendre expansion ( honda 2003 ) , grad expansion ( muronga 2004 ) , and numerical integration ( shoucri and shakarofsky 1994 ) .
but , to our knowledge , none of these solutions is six dimensional the ability to solve three dimensional integrals like ( eq .
37 below ) has only existed for a short time ( hahn 2004 ) .
manheimer et al .
( 1997 ) have called the self - consistent evaluation of even the simpler non - relativistic coefficients unfeasible " .
many approach the fokker - planck equation with implicit differencing , which is relatively difficult to modify with additional forces ( e.g. , in a particle - in - cell simulation ) .
instead , we follow the original suggestion of jones et al .
( 1996 ) and the subsequent application of habib et al .
( 2000 ) to model collisions as brownian motion .
a single particle s motion is observed at discrete times , and the affect of multiple collisions accumulated over this time is represented as dynamical friction and noise .
the distribution is approximated by some small fraction of the actual number of particles , all moving according to self - consistent kinetic forces .
the simulation thus solves single - body statistical dynamics , many times .
the desire for a covariant theory that is simple to apply and reduces naturally to the non - relativistic version is far from academic . as a concrete and important example , we show that our theory eliminates the possibility that stochastic acceleration is responsible for the hard x - ray emission in the coma cluster . before that , however , we present the relativistic kinetic theory in 2 , and then describe the method of solution in covariant form in 3 .
the application will be made in 4 .
we begin by discussing the development of the relativistic kinetic equation somewhat pedagogically , given the fact that its usage in the analysis of high - energy plasmas in astrophysics has been infrequent at best .
landau ( 1936 ) first approached the problem by showing that the boltzmann collision integral can be written as a flux in velocity space , @xmath2 thereby writing the boltzmann equation as a conservation equation for the single particle phase space distribution , @xmath3 where @xmath4 is the single particle phase space _ number _ density .
the poisson bracket @xmath5 represents a small change in the shape of a volume of phase space ( which nevertheless conserves number ) , the term @xmath6 a flux of particles out of it .
but what is @xmath7 ? to begin with , note that the number of collisions occurring in unit time between a particle with velocity @xmath8 and particles with velocity @xmath9 in the range @xmath10 is @xmath11 where @xmath12 is the amount of velocity exchanged . here
, @xmath13 , the probability that @xmath8 will be taken into @xmath14 , can be expressed evenly as @xmath15 .
this ensures that @xmath13 is symmetrical with respect to @xmath16i.e .
, that detailed balance is maintained .
the particle flux @xmath17 is found by subtracting the number of particles entering a volume of velocity space from the left , @xmath18 from that exiting to the right , @xmath19 this leads to an argument @xmath20 which can be expanded about small @xmath12 , giving the required @xmath7 , the so - called landau collision integral , @xmath21 b_{\alpha\beta}\ ; d^3v\;,\ ] ] where @xmath13 is expressed in terms of the collision cross section , @xmath22 and where @xmath23 finally , since momentum conservation dictates that the change @xmath16 is perpendicular to the relative velocity @xmath24 , it must be that @xmath25 and since @xmath26 can only depend on the vector @xmath24 , the tensor must take the form @xmath27\;,\ ] ] where @xmath28 thus , integrating equation ( 11 ) with a rutherford cross - section for @xmath29 , one gets @xmath30 where @xmath31 is the collision kernel , @xmath32 + and @xmath33 are cartesian indices . in
what follows , we define @xmath34 and @xmath35 for succinctness . in c.g.s .
units , @xmath36 . with one integration by parts
, the collision integral can be written + @xmath37\;,\ ] ] + where @xmath38 and @xmath39 f(\mathbf{v^\prime})\ ; d^3 v^\prime\;,\ ] ] + which is the fokker - planck equation seen commonly in the west .
equations ( 15 ) and ( 16 ) can be expressed in a simpler form by observing that the kernel @xmath31 obeys the relations + @xmath40 and @xmath41 + thus allowing the coefficients to be expressed as + @xmath42\;,\ ] ] and @xmath43\;.\ ] ] the advantage of this phrasing ( rosenbluth et al .
1957 ) is that the terms within the brackets contain elements in common with the poisson equation , for which sophisticated methods of solution exist ( e.g. , expansion by spherical harmonics and fft convolution ) .
when @xmath44 is a maxwellian , a harmonic expansion gives @xmath45 @xmath46\;,\ ] ] and @xmath47 where @xmath48 @xmath49 is the component of the diffusion tensor parallel ( perpendicular ) to the particle s velocity in the co - moving frame , @xmath50 is the error function , and @xmath51 .
since @xmath52 , the diffusion tensor fills out as @xmath53 + \frac{1}{2 } \delta_{\alpha \beta}\ , d_\bot\;.\ ] ] equations ( 21 ) thru ( 24 ) present a fairly complete non - relativistic kinetic theory .
the time @xmath54 required for a particle to be deflected by 90 degrees is a good measure of how long it will take a non - thermal injection to equilibrate ( spitzer 1962 ) .
this time is given by the condition @xmath55 thus , for particles whose
root mean square velocity is that of the group , @xmath56 , we have @xmath57 , @xmath58 ( m / s)@xmath59/s , and @xmath60 a slowing - down viscous time scale can be similarly defined
. one of our primary goals is to find a relativistically correct version of equation ( 26) the majority of problems ( even apparently difficult ones , such as nonthermal radiation from galaxy clusters ) can be resolved with a glance at this simple formula .
chandrasekhar ( 1957 ) approached the problem quite differently . consider replacing the truly discrete many - body potential @xmath61 with the smoothed version @xmath62 .
a distribution evolving only due to such a smooth , mean field force is ` collisionless'i.e .
, no particle ever leaves its volume of ( single particle ) phase space .
but in moving to an integrable charge distribution , we have lost track of small fluctuations in the number of particles in a given neighborhood . accumulated over a small time
, these fluctuations cause a random kink in a particle s path .
alone , this kink diffuses the probability of finding the particle with a particular velocity until all velocities are equally likely . since , however , a maxwellian distribution for this probability inevitably sets in , chandrasekhar reasoned that an associated viscous term must exist to bring diffusing particles back to the group , so that individual particles move through phase space according to the equations of motion + @xmath63 @xmath64 + where @xmath65 is a smooth mean field term ( plus any external forces ) , @xmath66 is dynamical friction that occurs when more particles hit the front end of a moving particle than the back , @xmath67 is the ` square root ' of the diffusion tensor , @xmath68 , and @xmath69 is a trivariate gaussian random vector with @xmath70 @xmath71 we show in appendix a why equations ( 2 ) and ( 28 ) are equivalent .
equations ( 27 ) and ( 28 ) reappropriate kinetic theory as a dynamics that is conceptually different than hamilton s : analytical dynamics follows the continuous motions of each of @xmath72 particles in a 2n - dimensional phase space ; it conserves energy , maintains the number of particles in a volume of phase space as a constant , and is reversible .
analytical dynamics has been described as a gradual unfolding of a contact transformation . " by contrast , statistical dynamics follows the motion of particles accumulated over some coarse time @xmath73indeed , since @xmath74 as @xmath75 , statistical dynamics is not even defined continuously .
statistical dynamics ignores all but the 6 phase space coordinates of the particle in question , representing the effect of the other degrees of freedom as friction and noise these terms cause a flux of particles through a surface of the _ single particle _ phase space .
statistical dynamics does not conserve energy , having at its heart the dissipative force @xmath76 , yet this force is precisely what is required to bring the assembly as a whole to a maxwellian distribution characterized by a constant energy .
statistical dynamics is irreversible .
chandrasekhar described it as the gradual unfolding of a markoff chain . " solving equation ( 1 ) by moving many particles according to the equations of motion ( 27 , 28 ) is sometimes called a ` monte carlo ' approach to kinetic theory .
it is important to stress the historical development here to avoid the tincture of a computational trick that this name implies : the langevin equation ( 28 ) is equivalent to the fokker - planck equation , and implies no further approximation provided collisions are short ( see eq . 30 ) .
it can be developed on physical grounds entirely separate from the boltzmann equation , and in fact was developed some 14 years before a successful solution of the fokker - planck equation appeared .
while we will use equations ( 27 ) and ( 28 ) to solve the evolution of the distribution , we return to equation ( 2 ) to derive the relativistically correct kinetic coefficients .
the crucial distinction between non - relativistic theory and the relativistic version that follows is that the function @xmath77 is _ not _ a relativistically invariant 4-scalar , since in different coordinate systems different particles will be considered to be located simultaneously in a given volume .
the integrals @xmath78 , and @xmath79 , however , _ do _ form a 4-vector : @xmath80 here @xmath81 is covariant , and if @xmath82 is covariant , than the phase space distribution @xmath4 must also be covariant .
rewriting equation ( 3 ) in the form @xmath83 we reason that , since the phase - space density , @xmath84 , and @xmath82 are covariant , @xmath85 must also be invariant .
it follows that the integral @xmath86 corresponding to equation ( 12 ) , must form a symmetric 4-tensor . here
@xmath87 is the velocity of one particle in the rest frame of the other . following a derivation similar to ( 10 - 13 )
, we find @xmath88 to be the desired manifestly covariant generalization of normal kinetic theory .
it is reasonable to neglect the zeroth component of ( 35 ) .
the reason for this is that the change in particle energy @xmath89 is of second order with respect to small scattering angle , and thus @xmath90 and @xmath91 are of third or fourth order . yet the fokker - planck equation is accurate only as far as second - order quantities .
leaving only the components of equation ( 8) corresponding to momentum flux , and re - expressing @xmath92 in a convenient form , we arrive at + @xmath93\;,\ ] ] where @xmath94 @xmath95 f(\mathbf{u^\prime})\ ; d^3 u^\prime\;,\ ] ] and where the kernel is now given by @xmath96\;.\ ] ] here , @xmath97 is the ratio of the momentum to the rest mass @xmath98 .
also , @xmath99 , @xmath100 , @xmath101 and @xmath102 in the non - relativistic limit , @xmath103 , @xmath104 and @xmath105 .
thus relativistic kinetic theory , which begins as manifestly covariant , also transitions to the non - relativistic limit at low velocity . from now on , when we refer to ` velocity ' in a relativistic context , we mean @xmath106 .
equation ( 36 ) is valid as long as bremsstrahlung is not so dominant as to make particle collisions inelastic , thus affecting the cross section .
provided this is the case , it is simple enough to calculate the bremsstrahlung losses as an integral over the electron distribution .
equations ( 37 ) and ( 38 ) may be solved with direct numerical integration . the langevin equation ( 28 )
can then be used to update a large number of phase - space points , rather than obtaining an implicit solution of the differential equation ( 6 ) .
besides the benefits of speed and the guaranteed conservation of particle number , this approach also allows various new acceleration mechanisms to be incorporated : as an example we include diffusion due to alfvnic turbulence and radiative losses within a single - species electron plasma in 3 and 4 .
it remains to show that the approach conserves energy and that the energy equipartitions correctly .
let us compare this result to that of nayakshin and melia ( 1998 , nm98 ) .
since nm98 construct a kinetic theory around the unitless energy @xmath107 , while covariant theory measures the friction and diffusion coefficients in the generalized velocity @xmath108 , direct comparison is not obvious . for example , in nm98 the energy exchange and viscous coefficients are equivalent , whereas covariant theory finds the energy exchange by integrating the flux of energy through a volume of phase space .
covariant theory gives diffusion coefficients both parallel and perpendicular to a particle s velocity ; the diffusion coefficients of nm98 , measured in the scalar @xmath109 , do not .
while the two theories measure different quantities , we may nevertheless place them in the same system of units for direct comparison .
the unitless diffusion coefficient is @xmath110 where the @xmath111 term gives the diffusion coefficient in units of @xmath112 , as in nm98 ( the kernel z is measured as @xmath113 , the constant a as @xmath114 ) . with a change of the variable of integration , @xmath115
the function @xmath116 is now the equivalent of equation ( 35 ) in nm98 .
as we shall see below , the difference between these two coefficients appears to be the reason why we do not recover blasi s ( 2000 ) results , since the method of nm98 is only valid in the highly relativistic regime ( see below ) . at first , the most glaring discrepancy is the absolute scale of the two theories : covariant theory ( along with nonrelativistic rosenbluth potentials ) is divergent ( fig .
1a ) , while that of nm98 ( fig .
1b ) is not .
there are also physically grounded discrepancies .
we show in the appendix that the fokker - planck and dynamical friction approaches are equivalent . in the latter theory ,
it is physically clear that all particles must be slowed by the viscous term , since a moving particle is always struck more frequently on its leading side , regardless of what relation its speed has to the rest of the group . to use an analogy with another @xmath117 potential :
a star is never accelerated by gravitational drag . yet the energy exchange term in nm98 ( fig .
2 ) changes sign , drawing all particles to the average ( scalar ) speed . in this case
, one must multiply the frequency of collisions by the fractional change in energy , which allows particles to both gain and lose energy as the result of multiple collisions .
thus , the covariant friction is positive - definite , while the ` friction ' term @xmath118 of nm98 is not .
perhaps the best way to compare the two theories , and to gauge their domain of validity , is to observe their predictions for the temporal evolution of a distribution .
crucially to the application of section 4 , we find that the theory of nm98 may not be applied in the extreme non - relativistic limit .
we follow the description of nm98 , beginning with a maxwellian distribution and solving the temporal evolution of the distribution via the chang - cooper implicit method recommended by petrosian and park ( 1996 ) , for a variety of transrelativistic temperatures @xmath119 ( so that @xmath120 is roughly @xmath121 k ) .
a fully relativistic distribution ( @xmath122 ) remains in equilibrium , conserving energy and particle number to around @xmath123 .
the kinetics of nm98 are therefore robust for highly energetic particles . as the temperature falls near the electron rest mass energy , however , the distribution deviates by between @xmath124 and @xmath125 . at these intermediate temperatures @xmath126 , an algorithm which explicitly conserves particle number ( as suggested in nm98 appendix a )
is required to resolve the coefficients .
neither the stochastic nor the implicit algorithms maintain a transrelativistic distribution in equilibrium . at @xmath127 ,
the distribution ceases to be relativistic , the function exp@xmath128 suffers from underflow errors , and the kinetics of nm98 do not maintain an equilibrium distribution . in blasi ( 2000 ) , the distribution prior to heating lies at @xmath129 ; it is therefore the failure of nm98 to correctly transition to the nonrelativistic limit that leads to the inaccuracy in blasi s ( 2000 ) formulation of the cluster s nonthermal emissivity .
figure 3 shows the kinetic coefficients calculated for a maxwellian distribution @xmath130 at a temperature @xmath131 k , compared to the known analytic versions given by equations ( 21 - 23 ) .
it s important to stress here that , while most of our results are presented as functions of the magnitude of the velocity , this quantity plays no role in the calculation itself . to make this figure , a large number ( here , one billion ) of particles are first initialized , scaled , and interpolated onto a grid , generally using the cloud - in - cell or triangle - shaped - cloud interpolants well - known from many - body mesh calculations .
this three dimensional grid represents a discrete version of the distribution @xmath132 on the right hand side of equations ( 15 ) and ( 16 ) . for any velocity @xmath9 ,
a subfunction interpolates values off the grid , and the result is an apparently continuous function .
we now concern ourselves with a single value of @xmath8the left hand side of equation ( 15)and carry out a monte carlo integration of the kernel ( eq .
10 ) using the vegas algorithm of the cuba integration library .
a maximum of 2500 function evaluations , beginning with 300 subdivisions and adding 500 new subdivisions in a refinement step gives the best speed - to - accuracy ratio .
for this one particular x , y , and z velocity , we sum over @xmath133 , resulting in six independent diffusion coefficients and three force coefficients comprising a structure .
we do this for some number of @xmath8 values , creating a second grid , each point of which is a structure containing the six @xmath134 or three @xmath135 .
two computational points are relevant here : first , we perform the sum over @xmath133 within the function call , rather than integrating three times and adding the answers .
second , the algorithm is about ten times faster when a vector of many functions is passed to the integrator in a single integration call , rather than defining the function and calling the integrator many separate times .
a lookup table is created so that in calling the integrator to function number one , say , the code automatically knows that it s looking for the @xmath136 value for @xmath137 m s@xmath138 . again , so long as the grid is sufficiently fine , a subfunction interpolates @xmath134 and @xmath135 off the grid for any @xmath8 .
we now have an apparently continuous function solving ( eqs .
19 - 20 ) .
the @xmath139-grid points are divided evenly among several processors using mpi .
each process accumulates its local distribution , which is then summed and re - broadcast to all processes .
a process solves the @xmath8 values belonging to it , and finally the full grid is gathered . in this fashion , one can avoid the passing of particles .
finally , for display purposes , we find the magnitude of each particle s velocity and friction , and we diagonalize its diffusion tensor .
that is , the coefficients in figure 1 play no role in evolving the distribution ; the solution makes no isotropic assumption and is naturally calculated in the lab frame . to evolve the distribution ,
each of the particles is updated using a first - order stochastic equation solver .
second order schemes exist ( qiang 2000 ) , but the multiple function evaluations are prohibitively expensive . the tensor q is found by diagonalizing d via jacobi rotation , taking the square root of the diagonal components , and then transforming back . making a transition to a relativistic maxwellian plasma
, we instead use the distribution function @xmath140 where @xmath141 is the modified hankel function of order 2 .
again , we are working with @xmath142 , the relativistic generalization of the velocity , and we use the term ` velocity ' exclusively with this meaning . with a substitution of @xmath143 in equation ( 37 ) , the resulting relativistic kinetic coefficients may be defined as @xmath144 @xmath145 for ease of use , we have expanded @xmath146 and @xmath147functions we call the _ enhancement _ factors , these being the multiplicative differences between the non - relativistic and relativistic expressions as @xmath148 and similarly for @xmath147 , where again the coefficients @xmath149 are expanded as functions of temperature , @xmath150 the resulting coefficients are given in table 1 ; the polynomial functions over @xmath151 fitted to each @xmath149 are shown in figure 4 . in figure 5
we show the actual enhancement factors @xmath146 and @xmath147 , together with the polynomials reconstructed from table 1 .
thus , with a few numbers , we can readily calculate the relativistic kinetic coefficients .
these enhancements can now be used to redefine the equilibration and viscous time scales , one of our primary goals in this transition to a covariant kinetic theory .
we have @xmath152 where the average velocity is evaluated from @xmath153 .
the factor @xmath154 is @xmath155 in the non - relativistic regime , but approaches @xmath156 when @xmath157 .
we may estimate it as the piecewise function @xmath158 now the length of time required for a relativistic distribution to relax to equilibrium may be measured against that for a non - relativistic distribution by the ratio @xmath159 , which is plotted in figure 6 .
( again , the definition of the spitzer time is valid at the average scalar velocity of the group .
the spitzer time is not particularly useful for calculations : rather , it is designed to give an intuitive grasp of the amount of time needed for equilibration , and we give its relativistic generalization as motivation for the re - examination of a wide class of problems . ) note that a fully relativistic plasma equilibrates in 1/100th of a spitzer time , @xmath160 .
the evolution of the non - relativistic and relativistic distributions at @xmath131 and @xmath161 k , respectively is shown in figures 8 and 9 , as a function of the spitzer time , @xmath160 . the non - relativistic distribution requires about two spitzer times to equilibrate , whereas the relativistic one reaches equilibrium in only @xmath162 .
when the energy is constant , or when the injection rate is known , we use normalization coefficients to maintain constant energy in the distribution ; an example of how these coefficients vary with time is plotted in figure 10 .
energy is conserved to an accuracy of @xmath163 , about the same as the precision of our integration .
however , when energy is not constant , this simple procedure is not feasible , and we must increase the grid size to a minimum of @xmath164 and the maximum number of function evaluations to 10,000 . in this case , normalization is no longer required . for example , in figure 11 , we allowed the distribution to relax for a full spitzer time before turning on stochastic acceleration , conserving energy all the while .
evidence for the presence of relativistic electrons in the intracluster medium ( icm ) is provided primarily by diffuse synchrotron emission at radio wavelengths . given the @xmath165 mpc - size of the synchrotron features , and
the fact that the radiative lifetime of the electrons appears to be orders of magnitude shorter than the time required to cover such distances , the presence of this emission also suggests that electrons are being accelerated out of the icm thermal pool .
typically , it is assumed that the radiating particles must be continuously re - accelerated on their way out these are ` primary ' models ( jaffe 1977 ; schlickeiser et al . 1987 , brunetti et al .
2001 , ohno et al .
however , protons could diffuse throughout the cluster without radiating , all the while colliding with thermal icm ions and cascading into the observed relativistic electrons these are ` secondary ' models ( dennison 1980 ) .
secondary models relieve the difficulties of continuously accelerating electrons against radiative and coulomb cooling , however blasi and colafrancesco ( 1999 ) suggest that protons could not provide enough secondary electrons to describe the radio emission without also producing @xmath166 gamma decays in excess of the number observed by egret .
coma s radio emission is accompanied by hard x - rays of similar spatial distribution ; this and its regular morphology classify the coma as a ` radio halo ' .
in contrast , ` radio relics ' are typically irregular and concentrated toward the cluster s periphery ( e.g. , feretti 2003 ) . interpreted as thermal bremsstrahlung , the observed x - rays imply an icm temperature between 8 and 9 kev .
coma s x - ray emission can not be described entirely as thermal bremsstrahlung , however .
the rossi x - ray timing explorer ( rephaeli and gruber 2002 ) and bepposax ( fusco - femiano et al .
2004 , hereafter ff04 ) have each made two observations of the coma , both claiming to find a hard x - ray tail ( hxr ) in excess of a thermal flux .
this is a controversial claim : the initial analysis of fusco - femiano et al .
( 1999 , ff99 ) was flawed
one of three spectra was counted twice leading to a 2.5 @xmath167 overstatement of the confidence level , according to rossetti and molendi ( 2004 , rm04 ) .
rm04 flatly state that the second bepposax observation shows no evidence of a hard tail .
ff04 , however , find a significant drop in the @xmath168 value for the fit to a hybrid thermal - power - law model ( @xmath169 for 7 dof ) as opposed to the purely thermal model ( @xmath170 for 9 dof ) .
this improvement can not be matched by a two - temperature model , since the second temperature ( @xmath171 kev ) is considered unrealistic .
rephaeli ( 1979 ) had predicted such a tail must exist when the synchrotron - emitting electrons inverse compton scatter with the cosmic background . in principle , this origin could be used as a probe for the icm magnetic field . yet , assuming that both the synchrotron radio and compton hxr are produced by the same population of electrons , ff04 infer a magnetic field of @xmath172g more than a factor of ten below the results inferred by faraday rotation ( @xmath173 g ) .
for this reason , nonthermal bremsstrahlung ( blasi 2000 ) was proposed as an alternative emission mechanism to inverse compton . here ,
nonthermal electrons are directly accelerated and directly radiate via @xmath174 bremsstrahlung .
petrosian ( 2001 ) has argued that bremsstrahlung is not sufficiently efficient to cool the distribution before an unacceptably large amount of energy is given to the icm , shifting the thermal body of bremsstrahlung above its observed flux . because this problem is simple ( it requires only stochastic acceleration , bremsstrahlung and coulomb cooling ) , and because it straddles the transrelativistic regime which our theory handles uniquely , we have chosen it as an example of the technique s power and simplicity .
our calculations agree with the analysis of petrosian .
they suggest that the source of confusion is the inability of nm98 and dl89 s kinetics to reduce to the nonrelativistic limit .
we find that the stochastically gained energy heats the body of the distribution , not just the tail , on a spitzer timescale of some tenths of a myr ; compare this to the tenths of gyr required for blasi s model .
similarly , the spitzer time requires that any merger event must have occurred within a few myrs , a vanishingly small window of time . with this motivation in mind ,
let us state the problem more precisely .
we use the pitch - angle averaged diffusion coefficient ( dermer , miller , and li 1996 ) , @xmath175 c^3 k_0 \beta_a^2 \eta_a ( r_b k_0)^{q-2 } p^q / \beta \;,\ ] ] to represent the resonant interaction of particles with alfvn waves . here
@xmath176 , @xmath177 , @xmath178 is the fraction of magnetic energy in alfvn waves , @xmath179 is the kolmogorov constant , @xmath180 kpc is the size of the galaxy , @xmath181 ( where @xmath182 is the alfvn velocity ) , and @xmath183 is the largest - scale wavenumber .
our point of departure is that of blasi s calculation ( @xmath184 @xmath185 , @xmath186 g , and @xmath187 kev ) , but we also look at a small range around these figures . in figure 10 , we show the evolution of the distribution , begun at 7.5 kev , with a window of 0.1 spitzer times between each curve .
one full spitzer time elapses before stochastic acceleration from a @xmath188 g field is turned on ( corresponding to the first curve on the left ) .
we see that rather than stochastic acceleration producing a high - energy tail , the high rate of thermalization provides energy to the whole distribution . the emissivity of this distribution is calculated self - consistently using particle number distributions and electron - proton bremsstrahlung cross section , @xmath189 where @xmath190 is the relativistic bremsstrahlung cross section averaged over all angles , and expanded to the @xmath191 th order .
including the elwert correction factor , this is ( haug 1997 ) @xmath192\times\nonumber\\ & \bigg[2ln ( \frac { \epsilon_o \epsilon_f + p_o p_f - 1}{k } ) - \frac{p_o p_f } { \epsilon_o \epsilon_f}\times\nonumber \\ & \bigg ( 1 + \frac{1 } { \epsilon_o \epsilon_f } + \frac{7}{20 } \frac { p_o^2 + p_f^2 } { ( \epsilon_o \epsilon_f)^3 } + ( \frac{9}{28 } k^2 + \frac{263}{210 } p_o^2p_f^2 ) \frac{1 } { ( \epsilon_o \epsilon_f)^3 } \bigg ) \bigg]\times\nonumber \\ & \frac{a_f}{a_o } \frac { 1-exp(-2 \pi a_o ) } { 1-exp(-2 \pi a_f ) } \;\;,\end{aligned}\ ] ] where @xmath193 is the fine structure constant , @xmath194 is the electron radius , @xmath195 is the photon energy in units of the electron mass energy , @xmath196 is the initial electron momentum in units of @xmath197 , @xmath198 is the total energy in rest mass units , @xmath199 , and @xmath200 in the energy regime considered , the error of this expansion is a few percent the contribution of @xmath201@xmath201 and @xmath201@xmath202 bremsstrahlung is of this order , so we exclude them from consideration .
once the cross section in equation ( 55 ) is found , we calculate the observed x - ray flux @xmath203 under the assumption of constant density and use the accepted volume @xmath204 and distance @xmath205 for coma to write @xmath206 , assuming a ( dimensionless ) hubble constant @xmath207 . the calculated spectrum is shown in figure 11 , together with the _ bepposax _ data , at spacings of @xmath208 . for completeness
, we also show the evolution for nine combinations of electron density and magnetic field . in the bottom
left corner
higher density and lower magnetic field a nonequilibrium tail never heats . in the top right lower density and higher magnetic field
the soft x - ray flux immediately moves above the observed limits . in none of these figures
do we address the other free parameter , the initial temperature . and
, while the calculation is done via the relativistically correct kernel , in no case is a relativistic tail produced before the nonrelativistic body heats to some excessively high energy .
all of these figures evolve over @xmath209 myr , some factor @xmath210 below that reported by blasi .
the corrected spitzer timescale is irreconcilable with the hxr observations ( fig .
11 ) , as in no case is a nonthermal tail accelerated before the icm equilibrates the added energy and reaches an unacceptably high temperature .
this is consistent with the result reported by petrosian ( 2001 ) : the erstwhile lack of a kinetic theory that is applicable throughout the transrelativistic regime was the source of discrepancy .
we have described a novel method for solving the covariant landau equation , unconstrained by assumptions of isotropy or low particle energy . using simple polynomial fitting formulae for the kinetic coefficients ,
we have described an efficient numerical technique for determining the temporal evolution of an arbitrarily relativistic particle distribution , which may also be subject to energy injection from the anomalous acceleration of particles , e.g. , in shocks or scattering with alfvn waves .
we anticipate that this method will find widespread application not only in astrophysics , but other physics disciplines as well . to demonstrate the power of our technique in understanding the behavior of high - energy plasmas in astrophysics , we have examined one of the currently considered models bremsstrahlung emission by a nonthermal tail for the hard x - ray emission in galaxy clusters , such as coma
our conclusion is that the time required by the underlying plasma to attain equilibrium is far too short compared to the stochastic acceleration time scale for any nonthermal high - energy extension to survive longer than @xmath211 myr .
it seems unlikely , therefore , that cluster mergers could be responsible for energizing the plasma turbulence required to sustain the nonthermal particle emissivity .
the application of covariant theory to relativistic electrons in the icm remains highly relevant .
as we suggest here , the time required for relaxation may vary by orders of magnitude from the currently used quantity . in a future paper , we will examine the production of @xmath166 gamma rays and the associated cascade of charged leptons during proton - proton collisions in the icm .
in addition , we have not yet completely resolved the complete set of differences between covariant theory and the kinetics of dl89 and nm98 , which amounts to better defining the range of validity for the latter approximate treatments .
we are performing a detailed comparison of their energy exchange and temporal evolution properties , and these results too will appear in a future paper .
this research was supported by nasa grant nag5 - 9205 and nsf grant ast-0402502 at the university of arizona .
we argue following zwanzig ( 2001 ) .
consider the stochastic equation @xmath212 where the noise @xmath213 is gaussian , with zero mean and delta - correlated second moment , @xmath214 let us now find the probability distribution @xmath215 , averaged over the noise .
now , @xmath215 is conserved : @xmath216 thus , substituting @xmath217 we arrive at the stochastic differential equation @xmath218 = -lf - \frac{\partial } { \partial v } \lambda(t ) f\;,\ ] ] where @xmath219 is an appropriately defined operator .
the solution to this equation is @xmath220 placing equation ( 61 ) back into equation ( 60 ) , we arrive at @xmath221 which is readily observed to be the fokker planck equation , @xmath222 once the average of @xmath215 is taken , and the zero average and unit correlation of @xmath223 is imposed .
berrington , r.c . ,
dermer , c.d .
2003 , apj , 594 , 709 beliaev , s.t . , & budker , g. i. 1955 , sov .
dokl . , 1 , 218 bell , a.r . , evans , g. , nicholas , d.j .
1981 phys . rev .
lett . , 46 , 243 blasi , p. 2000 , apj , 532 , l9 braams , b.j .
karney , c.f .
1987 , phys .
lett . , 59 , 1817 braams , b.j .
karney , c.f .
1989 , phys .
fluids , 1b , 1355 brunetti , g. , blasi , p. , cassano , r. , & gabici , s. 2004 , mnras , 350 , 1174b chandrasekhar , s. 1973 , _ stellar structure _
( new york , ny : dover ) dennison , b. 1980 , apj , 239 , l93 dermer , c. d. , & liang , e. p. 1989 , apj , 339 , 512 dermer , c.d . ,
miller , j.a . , and li , h 1996 , apj , 456 , 106 feretti , l. 2003 , in bowyer s. , hwang c .- y . , eds . asp conf .
301 , matter and energy in clusters of galaxies .
, san francisco , p. 143
fusco - femiano , r. , dal fuime , d. , feretti , l. , et al . 1999 ,
apj , 513 , l21 fusco - femiano , r. , et al .
2004 , apj , 602 , l73 guethlein , g. , foord , m.e . , price , d. 1996 , phys .
lett . , 77 , 1055 gabici , s. , blasi , p. 2003 , apj , 582 , 695 hahn , t. 2004 , arxiv : [ hep - ph/0404043 ] haug , e. 1997 , a&a , 326 , 417 honda , m. 2002 , phys
. plasmas , 10 , 4177 jaffe , w.j .
1977 , apj , 212 , 1 jones , m.e . , lemons , d.s . , mason , r.j . ,
thomas , v.a .
, winske , d. 1996 j. comp .
phys . , 123 ,
169 kuo , p .- h . ,
hwang , c .- y . , ip , w .- h .
2003 , apj , 594 , 732 landau , l.d .
1936 , physik .
sowjetunion , 10 , 154 lifshitz , e. m , & pitaevskii , l. p. 1981 , _ physical kinetics _
( boston , ma : butterworth - heinemann ) manheimer , w.m . ,
lampe , m. , joyce , g. 1997 , j. comp .
phys . , 138 , 563 mohanty , j. n. , baral , k.c .
1996 , phys .
plasmas , 3 , 804 muronga , a. 2004 , phys .
c , 69 , 034903 narayan , r. , yi , i. , & mahadevan , r. 1995 , nature , 374 , 623 nayakshin , s. , & melia , f. , 1998 , apj , 114 , 269 ( nm98 ) petrosian , v. , & park , b. , 1996 , apj , 103 , 255 petrosian , v. 2001 , apj , 557 , 560 petrosian , v. , & liu , s. 2003 , spd meeting 34 , 22.01 qiang , j. , ryne , r.d . ,
habib , s. , decyk , v. 2000 , j. comp .
phys , 163 , 434 qiang , j. , habib , s. 2000 , phys .
e 62 , 7430 rephaeli , y. , gruber , d. 2002 , apj , 579 , 587 rickard , g.j . ,
bell , a.r .
, epperlein , e.m .
1989 , phys .
lett . , 46 , 2687 rosenbluth , m.n . , macdonald , w.m .
, chuck , w. 1957 , phys .
, 107 , 350 rossetti , m. , molendi , s. 2004 , a&a , 414 , l41 schlickeiser , r. , sievers , a. , and thiemann , h. 1987 , a&a , 182 , 21 shapiro , s. l. , lightman , a. p. , & eardley , d. m. 1976 , apj , 204 , 187 shoucri , m. , shkarofsky , i. 1994 , comput .
, 82 , 287 spitzer , l. 1962 , _ physics of fully ionized gases _
( new york , ny : john wiley ) willson , m.a.g .
1970 , mras , 151 , 1 zwanzig , r. 2001 , _ nonequilibrium statistical mechanics _
( oxford : oxford university press ) crrrrrrrrrrr + @xmath224 & @xmath225 & @xmath226 & @xmath227 & @xmath228 & @xmath229 & @xmath230 & @xmath231 & @xmath232 & @xmath233 + @xmath234 & -1.61e-37 & 3.38e-35 & -2.92e-33 & 1.34e-31 & -3.50e-30 & 5.01e-29 & -2.97e-28 & -1.44e-27 & 3.76e-26 + @xmath235 & 2.72e-28 & -5.76e-26 & 5.08e-24 & -2.43e-22 & 6.87e-21 & -1.18e-19 & 1.20e-18 & -6.91e-18 & 9.48e-18 + @xmath236 & -7.04e-20 & 1.36e-17 & -1.08e-15 & 4.48e-14 & -1.07e-12 & 1.51e-11 & -1.38e-10 & 1.03e-9 & -1.02e-9 + @xmath237 & 1.74e-11 & -3.37e-9 & 2.64e-7 & -1.08e-5 & 2.46e-4 & -3.12e-3 & 2.10e-2 & -5.77e-2 & 1.0716 + + @xmath48 & @xmath225 & @xmath226 & @xmath227 & @xmath228 & @xmath229 & @xmath230 & @xmath231 & @xmath232 & @xmath233 + @xmath234 & 4.96e-38 & -1.06e-35 & 9.28e-34 & -4.31e-32 & 1.13e-30 & -1.61e-29 & 9.98e-29 & 1.10e-28 & -3.29e-27 + @xmath235 & -3.37e-29 & 7.51e-27 & -6.91e-25 & 3.36e-23 & -9.19e-22 & 1.35e-20 & -7.88e-20 & -3.66e-19 & 6.35e-18 + @xmath236 & 8.49e-21 & -2.15e-18 & 2.22 - 16 & -1.21e-14 & 3.74e-13 & -6.51e-12 & 5.54e-11 & -5.15e-11 & -3.65e-10 + @xmath237 & 1.55e-12 & -2.90e-10 & 2.17e-8 & -8.30e-7 & 1.73e-5 & -2.03e-4 & 1.60e-3 & 2.35e-2 & 1.00 + + @xmath238 & @xmath225 & @xmath226 & @xmath227 & @xmath228 & @xmath229 & @xmath230 & @xmath231 & @xmath232 & @xmath233 + @xmath234 & 1.09e-37 & -2.41e-35 & 2.24e-33 & -1.13e-31 & 3.39e-30 & -6.17e-29 & 6.74e-28 & -4.24e-27 & 1.39e-26 + @xmath235 & 5.80e-29 & -1.24e-26 & 1.10e-24 & -5.30e-23 & 1.51e-21 & -2.59e-20 & 2.66e-19 & -1.62e-18 & 7.51e-18 + @xmath236 & -7.43e-22 & 4.79e-20 & 7.39e-18 & -9.70e-16 & 4.49e-14 & -1.03e-12 & 1.27e-11 & -1.19e-10 & -7.18e-10 + @xmath237 & -5.87e-13 & 1.04e-10 & -7.20e-9 & 2.24e-7 & -1.73e-6 & -7.14e-5 & 1.45e-3 & 3.72e-2 & 9.92e-1 + | in this paper , we introduce a novel solution to the covariant landau equation for a pure electron plasma .
the method conserves energy and particle number , and reduces smoothly to the rosenbluth potentials of non - relativistic theory .
in addition , we find that a fully relativistic plasma equilibrates in only 1/100th of a spitzer time much faster than in the non - relativistic limit a factor of significant import to situations in which distortions to a maxwellian distribution are produced by anomalous methods of acceleration .
to demonstrate the power of our solution in dealing with hot , astrophysical plasmas , we use this technique to show that one of the currently considered models
continuous stochastic acceleration for the hard x - ray emission in the coma cluster actually can not work because the energy gained by the particles is distributed to the _ whole _ plasma on a time scale much shorter than that of the acceleration process itself . |
in the present work we study elliptic operators on manifolds with singularities in the situation , when the manifold is endowed with an action of a discrete group @xmath0 of diffeomorphisms .
more precisely , we consider operators equal to linear combinations of products of ( pseudo)differential operators and shift operators along the orbits of the group action .
such operators are called @xmath0-_operators _ , since they are associated with an action of group @xmath0 .
let us mention that the theory of @xmath0-operators on smooth manifolds is presently well developed : fredholm property and related analytic questions are studied ( e.g. , see @xcite ) , index formulas and related topolo gical problems are considered ( e.g. , see @xcite ) . in this paper
we study @xmath0-operators on manifolds with isolated singularities of conical type .
of course , the results of the paper can be extended to operators on manifolds with higher singularities : edges , corners , etc .
( e.g. , see @xcite ) . as usual in elliptic theory , fredholm property of an operator
is governed by the properties of its principal symbol .
we show that the principal symbol of a @xmath0-operator in our situation is a pair , consisting of the symbol on the main stratum ( interior symbol ) and the symbol at the conical point ( conormal symbol ) .
in addition , it turns out that ( unlike the interior symbol ) the conormal symbol _ is not _
an element of the crossed product and therefore the standard technique of proving the fredholm property of @xmath0-operators , which relies on crossed products ( see @xcite ) , does not apply .
one more feature of our theory is the fact that the conormal symbol at the conical point is an operator family exponentially depending on the parameter ( cf . the classical theory of parameter - dependent families @xcite ) . finally , let us note that in the proof of the fredholm property we systematically use the theory of @xmath1-algebras .
let us briefly describe the contents of the paper .
first , we give the statement of the problem ( section 1 ) .
then we formulate the finiteness theorem , which is the main result of the paper . to this end
, we compute the symbol of @xmath0-operators using the method of frozen coefficients . here , since we deal with nonlocal operators ( containing shift operators ) , one has to freeze the coefficients of the operator on the entire orbit , rather than at one point .
we also consider two examples : 1 ) @xmath2-operators on the sphere @xmath3 ; and 2 ) @xmath4-operators on the half - line @xmath5 . in these examples ,
the conditions , which give the fredholm property , are written out explicitly . then
( section 5 ) we give the proof of the fredholm property using the theory of @xmath1-algebras . as an important technical tool
, we construct the theory of elliptic parameter - dependent families of @xmath0-operators . at the end of the paper
, there is a small appendix , in which we give a direct computation of the interior symbol of @xmath0-operators using the method of frozen coefficients .
let us note that problems of this type were considered earlier in the works of a.b .
antonevich and his coworkers ( e.g. , see @xcite and the references there ) , however , most of the results obtained on this topic dealt with operators acting on @xmath6-spaces , whereas we consider operators in weighted sobolev spaces @xmath7 ( see @xcite ) .
the work was partially supported by rfbr ( projects nr .
12 - 01 - 00577 and 15 - 01 - 08392 ) , daad , and simons foundation .
we are grateful to prof .
elmar schrohe and institute of analysis at the leibniz university of hannover for hospitality .
in this section , we formulate the main problem of this work concerning elliptic operators associated with an action of a discrete group on a manifold with conical singularities . for a detailed exposition of elliptic theory on manifolds with conical singularities ( without group actions ) we refer the reader to @xcite , see also @xcite . [ [ g - manifolds - with - conical - points . ] ] @xmath0-manifolds with conical points .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + let @xmath8 be a manifold with a conical point denoted by @xmath9 .
this means that the complement @xmath10 is a smooth manifold and there exists a diffeomorphism of a punctured neighborhood of the conical point and the cartesian product @xmath11 here @xmath12 is a closed smooth manifold ( called the base of the cone ) , while @xmath13 corresponds to the conical point ( for this value of @xmath14 @xmath12 collapses to point ) .
_ a diffeomorphism @xmath15 _ is a mapping such that : 1 .
@xmath15 is a homeomorphism , which preserves the strata @xmath10 and @xmath9 and induces diffeomorphisms of these strata ; 2 . in a neighborhood of the conical point
the diffeomorphism has the form @xmath16 where @xmath17 is a real number and the diffeomorphism of @xmath12 is also denoted by @xmath18 for brevity .
let @xmath0 be a discrete group .
manifold @xmath8 is a _ @xmath0-manifold _
, if it is endowed with @xmath0-action .
to simplify the notation , we shall identify elements @xmath19 and corresponding diffeomorphisms @xmath15 .
[ [ weighted - sobolev - spaces . ] ] weighted sobolev spaces .
+ + + + + + + + + + + + + + + + + + + + + + + + recall the definition of weighted sobolev space @xmath20 of functions on a manifold with conical points .
the elements of this space are functions in the sobolev space @xmath21 outside a neighborhood of the conical point .
hence , it suffices to define the norm only for functions supported in a small neighborhood of the conical point .
the norm of function @xmath22 is defined as @xmath23 where @xmath24 is a nonnegative laplace operator on @xmath12 , or equivalently @xmath25 where @xmath26 is the mellin transform of function @xmath22 with respect to the radial variable @xmath14 ; @xmath27 is the weight line .
the space @xmath28 is just the space @xmath29 of square - integrable functions with respect to the measure @xmath30 where @xmath31 is a measure on @xmath12 .
the equivalence of expressions and follows from the properties of the mellin transform , in particular , from mellin plancherel equality ( e.g. , see @xcite ) : @xmath32 [ [ g - operators - on - manifolds - with - conical - points . ] ] @xmath0-operators on manifolds with conical points .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + let @xmath8 be a @xmath0-manifold .
the group action induces a representation of @xmath0 in sobolev spaces by shift operators : @xmath33 _ a conical @xmath0-operator _ on @xmath8 is an operator of the form @xmath34 where @xmath35 is a conical differential operator on @xmath8 of order @xmath36 and only finitely many operators @xmath37 are nonzero , so that the sum in is actually finite .
the aim of the present paper is to study ellipticity conditions , which guarantee fredholm property of operator .
as usual in elliptic theory , the conditions , under which one obtains fredholm property , are formulated in terms of the principal symbol of the operator
. we shall refer to the principal symbol as the `` symbol '' for brevity .
in our situation the symbol is a pair : the symbol on the main stratum ( interior symbol ) and the symbol at the conical point ( conormal symbol ) .
let us give the definitions of these symbols .
[ [ interior - symbol . ] ] interior symbol .
+ + + + + + + + + + + + + + + + to describe this symbol , we need to extend the group action to the cotangent bundle of the manifold .
denote the cotangent bundle of @xmath8 by @xmath38 ( e.g. , see @xcite ) .
recall the definition of @xmath38 .
let @xmath39 be the blowup of @xmath8 , which is a manifold with boundary equal to @xmath12 .
denote by @xmath14 and @xmath40 the normal and tangent variables at the boundary . by @xmath41
we denote the vector bundle over @xmath42 , whose sections are generated as a @xmath43-module by vector fields @xmath44 which are of unit length in the conical metric of the form @xmath45 .
by @xmath46 we denote the dual bundle ( with respect to the natural pairing of vectors and covectors ) .
one has the vector bundle isomorphism @xmath47 a diffeomorphism @xmath15 of the main manifold induces diffeomorphism of the cotangent bundle @xmath48 which is called the _ codifferential _ of @xmath18 . here
@xmath49 stands for the differential , while @xmath50 stands for the dual mapping .
let us define the symbol of operator at a point @xmath51 of the cotangent bundle off the zero section using the method of frozen coefficients .
note that the operator under consideration is essentially nonlocal : equation @xmath52 relates values of the unknown function @xmath53 and its derivatives on the entire _ orbit _ @xmath54 rather than at a single point @xmath55 .
for this reason , unlike the classical ( nonsingular ) case , we freeze the coefficients of the operator on the entire orbit of @xmath56 . performing such procedure , and applying fourier transform from @xmath57 to @xmath58 at each point of the orbit
, we can define the symbol as a function on the cotangent bundle off the zero section , with values in operators acting on functions on the orbit . a direct computation ( which we place in the appendix at the end of the paper )
gives the following expression for the interior symbol ( cf .
@xcite ) : @xmath59 here * the symbol acts on functions on the orbit @xmath60 , which is identified with @xmath0 using the mapping @xmath61 ; * @xmath62 is the right shift of functions on the group ; * the interior symbol of @xmath37 at a point @xmath63 is denoted by @xmath64 , while the expression @xmath65 is treated as an operator of multiplication of functions on @xmath0 : @xmath66 * the elements of the space @xmath67 are square summable functions @xmath68 with respect to the weight function on @xmath0 : @xmath69 hereinafter , we omit the dependence of the weight function on the variables @xmath70 for brevity . [ def1 ] the operator is the _ interior symbol _ of a @xmath0-operator @xmath71 at a point @xmath51 .
the expression has the following natural meaning .
namely , according to this formula : * the symbol of the shift operator @xmath72 is a right shift @xmath73 on the group ; * the symbol of differential operator @xmath37 is an operator of multiplication by the values of the symbol @xmath74 at the corresponding points on the orbit ; * finally , the symbol of @xmath75 is the sum of products of the symbols of @xmath37 and @xmath72 . in the general case ,
interior symbol depends on @xmath76 in a quite complicated way .
for example , it is frequently discontinuous in @xmath76 .
this is related with the fact that the structure of orbit can depend on an initial point on the orbit in a very complicated way .
[ prop-4 ] given two @xmath0-operators @xmath77 and @xmath78 , we have @xmath79 this follows directly from the fact that the symbol was obtained by the method of frozen coefficients .
however , the desired formula can also be obtained directly using only .
let us compute the restriction @xmath80 of the interior symbol to the boundary of the cotangent bundle , i.e. , for @xmath13 .
recall that near the conical point we have coordinates @xmath81 and dual coordinates @xmath82 . for the operator and @xmath83 we have @xmath84 where the weight function @xmath85 is equal to @xmath86 in this formula we used the special form of @xmath18 near the conical point .
in particular , it follows from that @xmath87 if @xmath88 is an isometry .
[ [ conormal - symbol . ] ] conormal symbol .
+ + + + + + + + + + + + + + + + to obtain the symbol of the operator at the conical point , we note that by assumption this point is fixed under the group action on the manifold . hence , to obtain the symbol it suffices to freeze the coefficients at this point , i.e. , to set @xmath13 in the coefficients .
we get @xmath89 now we omit the factor @xmath90 and apply mellin transform @xmath91 ( see ) . as a result in the coordinates
@xmath92 we obtain the operator family @xmath93 where @xmath94 , while @xmath95 is the shift operator on the base of the cone .
note that to obtain , we used the fact that the mellin transform takes operator @xmath96 to the operator of multiplication by @xmath97 , and the dilation operator @xmath98 to the operator of multiplication by @xmath99 , while the elements of the weighted space are taken to functions on the line @xmath100 , ranging in the space @xmath101 . the operator family is the _ conormal symbol of @xmath0-operator @xmath71 _ and is denoted by @xmath102 .
let @xmath103 be two @xmath0-operators of orders @xmath104 , respectively .
then their composition is a @xmath0-operator and one has @xmath105
this proposition is proved by a direct computation .
we note only that the shift of the argument of the conormal symbol in the latter formula is due to the presence of the factor @xmath90 in ( cf .
@xcite in the classical case ) .
[ [ ellipticity - and - fredholm - property . ] ] ellipticity and fredholm property .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + [ def - ell4 ] a @xmath0-operator @xmath71 ( see ) on a manifold with conical points is _ elliptic _ , if it has invertible 1 .
interior symbol @xmath106 for all @xmath107 ; 2 .
conormal symbol @xmath108 for all @xmath97 on the weight line @xmath109 .
[ th - finite1 ] if a @xmath0-operator @xmath110 is elliptic , then it is fredholm .
the proof of theorem [ th - finite1 ] uses the theory of @xmath1-algebras and is given in section [ sec - c1 ] .
the conormal symbol @xmath108 is a family of @xmath0-operators on @xmath12 , depending on the parameter @xmath97 . in section [ sec - c1 ]
we develop elliptic theory for such families and show that the family @xmath108 is invertible for @xmath94 large , provided the interior symbol @xmath111 is elliptic for the weight @xmath112 .
the results of this section extend to the case of pseudodifferential operators ( @xmath113do ) , i.e. , the case , when the operators @xmath114 , which appear in the definition of a @xmath0-operator , are pseudodifferential operators on @xmath8 .
concerning the theory of @xmath113do on manifolds with conical points see , for example , the monograph @xcite and the references there .
[ [ sphere - as - a - mathbbz - manifold . ] ] sphere as a @xmath2-manifold .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + consider the sphere @xmath3 as the extended complex plane with complex coordinate @xmath115 and treat the points of the sphere @xmath116 and @xmath117 as conical points denoted by @xmath118 and @xmath119 . on the sphere , we consider the action of group @xmath2 , defined by iterations of the diffeomorphism @xmath120 ( see figure [ fig1q ] . ) for definiteness , we assume that @xmath121 and @xmath122 .
.,width=188 ] we replace the complex coordinate @xmath115 by real coordinates @xmath123 using the change of variables @xmath124 , which defines the diffeomorphism ( blowup ) @xmath125 of the complement of the conical points and the infinite cylinder .
moreover , under this diffeomorphism neighborhoods of the conical points @xmath126 correspond to neighborhoods of the sections at @xmath127 , while the diffeomorphism @xmath18 is equal to : @xmath128 i.e. , it is just a helical motion of the cylinder .
[ [ symbols - of - g - operators . ] ] symbols of @xmath0-operators .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + on @xmath3 we consider conical @xmath2-operators @xmath129 with constant coefficients in @xmath130 ( here we omitted the factor @xmath90 right away ) , while the shift operator is equal to @xmath131 now @xmath71 acts in spaces with the weight @xmath132 , where the weight @xmath133 is at @xmath118 and the weight @xmath134 is at @xmath119 ( one can check that @xmath71 is a conical operator at the conical point @xmath135 using the inversion @xmath136 ) .
let us compute the interior and the conormal symbols of @xmath71 .
the interior symbol is equal to @xmath137 where @xmath138 and @xmath139 .
the conormal symbols at @xmath13 and @xmath135 are equal to @xmath140 by theorem [ th - finite1 ] operator @xmath71 is fredholm if its interior symbol @xmath141 is invertible whenever @xmath138 and the conormal symbol @xmath142 is invertible on the weight lines @xmath143 . to verify these invertibility conditions ,
let us compute the weight function @xmath85 in .
[ [ computation - of - the - weight - function . ] ] computation of the weight function .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + a computation shows that the measure @xmath144 is equal to @xmath145 we fix the metric @xmath146 on the cylinder . the weight function @xmath147 ( see ) up to equivalence does not depend on @xmath148 and is equal to 1 . at a point in the interior of @xmath149 : @xmath150 2 . at a point on the boundary @xmath151 : @xmath152 this proposition is proved by a direct computation using . [ [ examples . ] ] examples .
+ + + + + + + + + consider the operator @xmath153 \1 .
the interior symbol of this operator is equal to @xmath154 where the weight @xmath85 was defined in .
it is easy to algebrize operator .
let us formulate the corresponding result .
[ lemma-3 ] given @xmath155 , the fourier transform @xmath156 takes the space @xmath157 with the weight @xmath85 to the space of holomorphic functions in the annulus @xmath158 which are square integrable on the boundary . if @xmath159 , then the range of fourier transform is the space dual to the space of holomorphic functions in the annulus with outer radius @xmath14 and inner radius @xmath160 .
the proof is based on the fact that @xmath161 if and only if @xmath162 these conditions mean that the fourier transform of this function , i.e. , the laurent series @xmath163 is absolutely convergent in the mentioned annulus . by lemma [ lemma-3
] , we transform the symbol to the operator of multiplication @xmath164 acting in the space of analytic functions in the annulus .
hence , the interior symbol is invertible if and only if this function is invertible in the annulus .
the interior symbol is invertible if and only if @xmath165 .
the interior symbol is invertible for all @xmath166 if and only if @xmath167 the numbers @xmath168 run over the annulus with the outer radius @xmath169 with center at point @xmath170 ( see figure [ fig1 t ] ) . and the line @xmath171,width=264 ] the condition that the intersection of this annulus and the line @xmath171 is empty is precisely .
we finally get the invertibility condition for the interior symbol in the form @xmath172
the conormal symbol of the operator is @xmath173 where @xmath174 is the shift operator on the circle . for @xmath94 large
the conormal symbol is invertible ( under the assumption that the operator is interior elliptic ) .
expanding functions on the circle in fourier series , we obtain the ellipticity condition for the conormal symbol in the form : none of the equations @xmath175 has a solution @xmath97 such that @xmath143 .
solving transcendental equations , we can obtain the singular points of the conormal symbol , that is the values of the weight for which the problem is not fredholm .
note that the ellipticity condition can be effectively verified numerically : since eq .
has no solutions in the vertical stripe @xmath176 provided @xmath177 is large , all the solutions in the mentioned stripe are associated with a finite number of @xmath178 s and a bounded domain of @xmath97 s .
let us treat the half - line @xmath5 as a manifold with two conical points and consider the operator @xmath179 here we choose weights @xmath132 at @xmath13 and @xmath135 , correspondingly , and suppose that the numbers @xmath180 are _
incommensurable_. let us assume for definiteness that @xmath181 , @xmath155 .
let us obtain conditions under which the operator is fredholm . for simplicity we set @xmath182 and apply mellin transform .
we obtain the operator @xmath183 where @xmath184 is the vertical stripe in the complex plane , while @xmath185 is the mellin transform of the space @xmath186 .
the elements of this space admit holomorphic continuation inside this stripe ( see @xcite ) .
one can show that the operator is invertible in the space @xmath185 if and only if the function @xmath187 is nonzero for all @xmath188,$ ] @xmath189 $ ] .
it is easy to write out the invertibility condition for the latter function explicitly : given some @xmath112 , the sum of the first two terms runs over the circle in the complex @xmath115-plane with center at @xmath190 and radius @xmath191 , while the third summand runs over the circle with center at @xmath116 and radius @xmath192 . clearly , the function is invertible if and only if the mentioned circles do not intersect each other .
further , the intersection of two circles is nonempty if and only if the radii of the circles and the distance between their centers satisfy triangle inequalities ( see figure [ fig2a ] ) : @xmath193 [ width=6 cm ] ris02a.jpg more precisely , the intersection of two circles is nonempty if and only if all three inequalities are valid .
so , our operator is elliptic if for all @xmath188 $ ] one of the mentioned inequalities is not valid . for fixed @xmath112 the domain , in which one of inequalities is not valid ,
is shown in figure [ fig2b ] ( the domains , in which inequality one , two , three is not valid is denoted by i , ii , iii , correspondingly ) .
further , the intersection of all such domains over all @xmath188 $ ] is shown in figure [ fig3a ] ( we omit the corresponding elementary computations ) .
[ width=13 cm ] ris02b.jpg we see that the domain of parameters @xmath194 , for which the operator is invertible , consists of three components .
the meaning of this decomposition into three components is simple : the operator @xmath195 is equal to the sum of three terms and this operator is invertible if and only if one of the terms is dominating ( the first inequality corresponds to the domination of the identity operator , the second corresponds to @xmath196 , the third corresponds to @xmath197 ) .
[ width=13 cm ] ris03a.jpg
the aim of this section is to prove theorem [ th - finite1 ] , i.e. , to obtain fredholm property of elliptic @xmath0-operators . as usual ,
the fredholm property is proved by constructing an almost - inverse operator , i.e. , inverse up to compact operators . clearly , given @xmath0-operator ( where the sum is actually finite ) , a similar sum for an almost - inverse operator
is ( as a rule ) infinite and for this reason one has to work with @xmath0-operators equal to infinite sums .
unfortunately , a direct consideration of such sums in the general situation , which we consider , looks quite difficult .
however , there is a general approach , which enables one to overcome difficulties , related with infinite series .
this approach is based on systematic use of the language of @xmath1-algebras and their crossed products ( see @xcite ; below we follow the exposition in @xcite ) . [ [ c - algebra - of - psido - psim . ] ] @xmath1-algebra of @xmath113do @xmath198 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + let us briefly recall the structure of the @xmath1-algebra @xmath199 of zero order @xmath113dos on a manifold @xmath8 with conical singularities acting on the space @xmath200 , that is @xmath201 .
this algebra can be obtained ( e.g. , see @xcite ) as the norm closure of zero order operators with smooth interior symbols and smooth conormal symbols .
this algebra is endowed with the symbol mapping @xmath202 which takes a @xmath113do @xmath71 to its interior symbol @xmath111 ( continuous function on the cosphere bundle @xmath203 ) and conormal symbol @xmath102 ( family of @xmath113dos on the base of the cone @xmath12 , continuously depending on parameter @xmath204 ) . here
the interior symbol and the conormal symbol are subject to compatibility condition : the restriction of the interior symbol to the boundary @xmath205 is equal to the symbol of the conormal symbol , which is considered as a parameter - dependent family .
thus elements of the algebra @xmath198 can be treated as @xmath113dos on @xmath8 with continuous symbols .
[ [ g - operators - and - crossed - products . ] ] @xmath0-operators and crossed products .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + @xmath0-action on @xmath8 induces the unitary representation of @xmath0 on @xmath206 : @xmath207 @xmath208 where @xmath209 is the measure , which is used in the definition of the space @xmath210 .
the group @xmath0 acts on @xmath198 on the left as @xmath211 and one defines the maximal @xmath1-crossed product @xmath212 ( concerning the theory of crossed products see , e.g. , @xcite ) .
elements of the crossed product are families @xmath213 of @xmath113dos on @xmath8 , parametrized by the group .
a family is called finite , if it has only finitely many nonzero elements .
the set of finite families is denoted by @xmath214 and called the _ algebraic _ crossed product .
we shall use the following universal property of the crossed product @xmath215 of algebra @xmath195 and group @xmath0 , acting on @xmath195 by automorphisms : an arbitrary covariant representation , i.e. , a pair of representations @xmath216 ( the representation @xmath217 is assumed to be unitary ) induces a homomorphism of the crossed product : @xmath218 which on the elements of the algebraic crossed product has the form @xmath219 we now define a mapping , which takes a finite family @xmath220 to the @xmath0-operator @xmath221 by the universal property this mapping extends to the @xmath1-homomorphism @xmath222 here @xmath223 ( @xmath224 ) stands for the algebra of bounded ( compact ) operators acting on @xmath225 .
our aim in this section is to study fredholm property for @xmath0-operators . obviously , the fredholm property of a @xmath0-operator is equivalent to the invertibility of the corresponding element in the calkin algebra @xmath226 , i.e. , invertibility of the element @xmath227 for @xmath228 . to study the invertibility problem ,
we introduce the notion of symbol for @xmath0-operators and the corresponding quantization . since the conormal symbol of a @xmath0-operator is a family of @xmath0-operators with parameter ( see ) , let us study this notion . [ [ g - operators - with - parameter . ] ] @xmath0-operators with parameter .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + @xmath0 acts on the cosphere bundle @xmath203 and the boundary of this bundle , denoted by @xmath205 , is invariant with respect to this action .
hence , we can define the crossed product @xmath229 .
since the conical point has a punctured neighborhood diffeomorphic to the product @xmath230 , one has a diffeomorphism @xmath231 .
denote by @xmath232 the norm closure of the algebra of classical zero - order @xmath113dos on @xmath12 with parameter @xmath233 ( e.g. , see @xcite ) . here
@xmath234 stands for the algebra of continous bounded functions on the weight line @xmath235 ranging in bounded operators on @xmath236 . _
@xmath0-symbol with parameter _
@xmath97 is an element @xmath237 let us define quantization of @xmath0-symbols with parameter by the formula @xmath238 and @xmath239 is the unitary representation of @xmath0 on @xmath236 by shift operators on @xmath12 .
the formula defines a quantization mapping on the algebraic crossed product , which extends by the universal property to the @xmath1-crossed product as a @xmath1-homomorphism denoted by @xmath240 here @xmath241 is the ideal of continuous compact - valued functions vanishing at infinity .
a _ parameter - dependent family of @xmath0-operators _ is a family of operators @xmath242 such that @xmath243 for some symbol @xmath244 .
an element @xmath170 will be called the _ symbol of parameter - dependent family _ and denoted by @xmath245 .
the algebra of _ parameter dependent families of @xmath0-operators _ is defined as @xmath246 by construction we have the short exact sequence of @xmath1-algebras @xmath247 since the sequence is exact , we obtain the following corollary .
[ cor-7 ] if a parameter - dependent family @xmath248 is elliptic , i.e. , the symbol @xmath249 is invertible , then this family is fredholm for all @xmath250 and invertible at infinity .
[ [ symbols - of - g - operators - on - manifold - m . ] ] symbols of @xmath0-operators on manifold @xmath8 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + consider a @xmath0-operator equal to a finite sum @xmath251 on @xmath8 .
let us define its symbol @xmath252 as the pair @xmath253 here the interior symbol is denoted by @xmath111 and the conormal symbol by @xmath102 .
the algebra of symbols is defined as a subalgebra in the direct sum @xmath254 defined by the compatibility condition @xmath255 here @xmath256 stands for the restriction of the interior symbol to the boundary , and @xmath257 is the symbol of the parameter - dependent family ( see above ) .
now we can define the symbol mapping as the homomorphism of algebras : @xmath258 which extends the mapping .
this symbol mapping is well defined by the universal property of the crossed product .
[ [ quantization - of - symbols . ] ] quantization of symbols .
+ + + + + + + + + + + + + + + + + + + + + + + + let us define the quantization mapping @xmath259 which takes a symbol to a certain operator . to construct the mapping ,
consider an element @xmath260 , whose first component lies in the algebraic crossed product .
since the elements @xmath261 are nonzero only for finitely many diffeomorphisms @xmath19 , there exist neighborhoods @xmath262 of the conical point such that @xmath263 and in @xmath264 all such diffeomorphisms have the form .
let @xmath265 be smooth functions on @xmath8 , which are identically equal to zero outside @xmath264 and @xmath266 , respectively , and are identically equal to 1 in a smaller neighborhood of the conical point .
the symbol @xmath170 is quantized in two steps .
first , we use the mellin transform @xmath91 to define the operator @xmath267 with conormal symbol equal to @xmath268 ( this fact is verified directly ) .
second , by the compatibility condition , the interior symbol @xmath269 and the interior symbol @xmath270 of the operator @xmath271 are equal over @xmath205 .
hence @xmath272 i.e. , this difference is a symbol vanishing over @xmath205 .
therefore , to the interior symbol we naturally assign an operator on @xmath8 with zero conormal symbol . denote this operator by @xmath273 .
we finally define @xmath274 it is easy to show that the operator @xmath275 modulo compact operators does not depend on the choice of cut - off functions @xmath276 and @xmath113 .
[ [ main - theorem . ] ] main theorem .
+ + + + + + + + + + + + + [ th - cstar1 ] 1 .
the mapping @xmath277 ( see ) extends by continuity to a homomorphism of @xmath1-algebras and one has a commutative triangle of @xmath1-algebras and their homomorphisms @xmath278^\pi \ar[dr]_{\sigma } & & \mathcal{b}/\mathcal{k}\\ & \sigma\ar[ru]_{op } & } \ ] ] 2 .
( fredholm property ) suppose that an element @xmath279 has invertible symbol @xmath280 .
then the corresponding @xmath0-operator @xmath281 is fredholm .
let us first prove that the diagram is commutative in the algebraic case , i.e. , given @xmath282 with symbol @xmath283 , let us show that @xmath284 . on the one hand , we have @xmath285 on the other hand , the symbol of @xmath195 is equal to @xmath286 where @xmath287 , @xmath288 are the interior and the conormal symbols of @xmath289 .
quantization of this symbol gives the operator @xmath290 \widetilde t_h=\\ = \sum_h a_h \widetilde t_h.\end{gathered}\ ] ] here the last equality modulo compact operators follows from the fact that the operator in square brackets in and the operator @xmath289 have equal interior and conormal symbols ( this is proved by a direct computation ) and , hence , coincide up to compact operators .
let us now show that @xmath277 is a homomorphism on the subalgebra @xmath291 , consisting of symbols @xmath292 , where @xmath269 is in the algebraic crossed product .
indeed , given @xmath293 , we have @xmath294 , @xmath295 for some elements @xmath296 ( since the symbol mapping @xmath297 is surjective ) .
further , we have @xmath298 this gives the desired property .
it follows from item 2 .
that @xmath277 extends by the universal property of the crossed product from the dense subalgebra @xmath299 to the entire algebra @xmath300 . for this extension
the diagram remains well defined and commutative .
\4 . to prove the fredholm property
, it suffices to show that @xmath301 is invertible .
we claim that the inverse element is equal to @xmath302 .
indeed , we have @xmath303 here the first equality follows from the commutativity of diagram , while the second follows from the fact that @xmath277 is a homomorphism of algebras .
the equality @xmath304 is proved similarly .
the proof of the theorem is now complete .
[ [ invertibility - of - symbols . ] ] invertibility of symbols .
+ + + + + + + + + + + + + + + + + + + + + + + + + by theorem [ th - cstar1 ] a @xmath0-operator is fredholm , if its symbol is invertible as an element of a certain @xmath1-algebra . how to check this invertibility ?
the following proposition shows that the invertibility condition in theorem [ th - cstar1 ] can be equivalently written as invertibility of certain explicitly written operators .
to write out these conditions , we define a special set of representations @xmath305 of the crossed product in the algebra of bounded operators acting on the standard space @xmath306 on the group ( cf . ) .
given @xmath307 , this representation takes an element of the crossed product to its restriction on the orbit of @xmath63 .
[ prop-5 ] the symbol @xmath308 is invertible if and only if the following two conditions are satisfied : 1 .
the restriction @xmath309 of the interior symbol on the orbit of an arbitrary point @xmath307 is invertible ; 2 .
the conormal symbol @xmath310 is invertible for all @xmath250 .
since @xmath283 is a pair , it suffices to prove the invertibility of each of the components .
the interior symbol @xmath311 is in a crossed product . by a well - known result @xcite
an element of the crossed product is invertible if and only if its restrictions @xmath312 to the orbits of all points @xmath307 are invertible .
hence , by our assumptions the interior symbol is invertible in the crossed product .
let us prove the invertibility of the conormal symbol @xmath313 . by assumption ,
this symbol is invertible for each fixed @xmath97 .
we have to show that the norm of the inverse family is uniformly bounded ( continuity of the inverse family is obvious )
. now uniform boundedness on an arbitrary finite interval of @xmath314 follows from continuity , while for large values of the parameter uniform boundedness follows from the fact that the symbol @xmath315 is invertible , i.e. the parameter - dependent family @xmath316 is elliptic and by corollary [ cor-7 ] the inverse family is uniformly bounded for large @xmath97 .
the proof of the proposition is complete . [
[ proof - of - theoremth - finite1-fredholm - property - in - weighted - sobolev - spaces - hsgamma . ] ] proof of theorem [ th - finite1 ] ( fredholm property in weighted sobolev spaces @xmath7 ) .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + let @xmath317 be an elliptic @xmath0-operator of order @xmath36 on @xmath8 .
we want to prove that it is fredholm .
( reduction to operator on @xmath6 . )
first , we reduce @xmath71 to a zero - order operator acting on @xmath210 . to this end , consider the composition @xmath318 where @xmath319 stand for some elliptic @xmath113dos on @xmath8 , acting in corresponding weighted spaces .
this composition is a @xmath0-operator denoted by @xmath320 since @xmath321 are elliptic and by classical conical theory fredholm , the @xmath0-operators @xmath71 and @xmath322 are fredholm or not fredholm simultaneously .
hence , by the multiplicative property of the symbols we get @xmath323 hence , ellipticity of @xmath71 implies ellipticity of @xmath322 .
let us show that ellipticity of operator @xmath322 in the sense of definition [ def - ell4 ] implies invertibility of its symbol in the algebra of symbols @xmath324 indeed , invertibility conditions for such elements were obtained in proposition [ prop-5 ] .
let us show , that the conditions of this proposition are satisfied .
indeed , the conormal symbol @xmath108 is invertible by assumption for all @xmath250 .
it remains to prove the invertibility of the restriction @xmath325 to the orbit of an arbitrary point @xmath307 .
consider the diagram @xmath326^{\sigma(d')(x,\xi ) } \ar[d]^\simeq_{\sqrt{m_{x,\xi } } } & & l^2(g , m_{x,\xi } ) \ar[d]^\simeq_{\sqrt{m_{x,\xi } } } \\
l^2(g ) \ar[rr]_{\pi_{x,\xi}(\sigma ( d ' ) ) } & & l^2(g),}\ ] ] where the vertical isomorphisms are products with the square root of the weight function @xmath327 in ( we set @xmath201 ) .
this diagram is commutative ( this fact is easy to prove directly if we write operator @xmath322 as in ) .
now , since @xmath322 is elliptic in the sense of definition [ def - ell4 ] , it follows that the upper row in is an isomorphism .
since the diagram commutes , we obtain that the lower row is an isomorphism as well .
hence , the restriction @xmath328 of the symbol to the orbit of @xmath329 is invertible as desired .
the proof of theorem [ th - finite1 ] is now complete .
let us give the details of the computation of the interior symbol and the weight using the method of frozen coefficients .
let us consider a differential operator @xmath71 on @xmath8 , and @xmath330 .
by freezing the coefficients of @xmath71 at the points of the orbit @xmath54 and omitting lower order terms , we obtain an operator with constant coefficients in the space of functions on the orbit : , where @xmath266 is a vector space , is defined by a norm @xmath331 and measure @xmath332 on @xmath333 by the formula @xmath334 ] @xmath335 fourier transform @xmath336 at each point of the orbit takes this operator to an isomorphic operator : is the @xmath6-space of functions on a space @xmath337 square integrable with respect to the measure @xmath338 . ] @xmath339 as usual in elliptic theory , the symbol is considered off the zero section , i.e. , we replace the operator by @xmath340 operator acts on a direct sum of spaces , corresponding to different points of @xmath8 .
this is not quite convenient from the point of view of computations . to eliminate this drawback , we replace by an isomorphic operator acting on the space of functions at the point @xmath56 . to this end
, we note that for @xmath19 the codifferential of @xmath15 is an isomorphism of cotangent spaces @xmath341 hence , the corresponding shift operator is invertible and defines an isometric isomorphism of @xmath6-spaces : is a diffeomorphism , then the mapping @xmath342 defines an isometric isomorphism if and only if @xmath343 . ] @xmath344 where the weight function @xmath345 on @xmath346 is defined as @xmath347 which coincides with up to invertible factor @xmath348 . using isometric isomorphisms ,
the spaces in can be replaced by isomorphic subspaces of the form @xmath349 where the mapping ranges in the space of functions on @xmath346 with values in the space @xmath67 of functions on the group .
so , taking into account and , we see that the symbol of @xmath71 , which is obtained by freezing the coefficients of the operator on the orbit and applying the fourier transform , can be represented as the family of operators of multiplication @xmath350 by the values of the interior symbol at the corresponding points on the orbit of @xmath51 . a similar computation ( which we omit for brevity ) shows that the symbol of the shift operator @xmath72 is equal to the right shift operator on the group : @xmath351 formulas and give the desired formula for the symbol , and also the formula for the weight function .
v. nazaikinskii , a. savin , and b. sternin .
lliptic theory on manifolds with corners .
i. dual manifolds and pseudodifferential operators . in _ @xmath1-algebras and elliptic theory ii _ , trends in mathematics , pages 183206 .
birkhuser , basel , 2008 .
a. savin and b. sternin .
elliptic theory for operators associated with diffeomorphisms of smooth manifolds . in
_ pseudo - differential operators , generalized functions and asymptotics _ ,
volume 231 of _ operator theory : advances and applications _ , pages 126 .
birkhuser , 2013 . | in the present work we study elliptic operators on manifolds with singularities in the situation , when the manifold is endowed with an action of a discrete group @xmath0 . as usual in elliptic theory ,
the fredholm property of an operator is governed by the properties of its principal symbol .
we show that the principal symbol in our situation is a pair , consisting of the symbol on the main stratum ( interior symbol ) and the symbol at the conical point ( conormal symbol ) .
fredholm property of elliptic elements is obtained . |
more than a hundred young massive stars , mostly wolf rayet / o and b types , have been identified within a distance of @xmath14 from the massive black hole ( mbh ) in the galactic center ( gc ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
these young stars are empirically divided into two groups : ( 1 ) the majority of the young stars at a distance @xmath15@xmath16 from the mbh are located on coherent disk - like structures , i.e. , the clockwise rotating stellar ( cws ) disk and the possible counterclockwise rotating stellar ( ccws ) disk ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) ; and ( 2 ) the young stars within a distance of @xmath17 from the mbh ( denoted as gc s - stars ) , exclusively b - dwarfs , are spatially isotropically distributed and their orbital eccentricities follow a distribution of @xmath18 ( e.g. , * ? ? ?
* ; * ? ? ?
the existence of these young stars is quite puzzling as star formation in the vicinity of an mbh is thought to be strongly suppressed due to the tidal force from the mbh ( i.e. , the paradox of youth ; see * ? ? ?
* ; * ? ? ?
it is of great importance to address not only the formation of these stars but also the origin of their kinematics , which should encode fruitful information of the dynamical interplays between the central mbh and its environment .
young stars in the cws ( or ccws ) disk are probably formed in a previously existing massive gaseous disk due to instabilities and fragmentation developed in it ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
young binary stars in the disk(s ) may migrate or be scattered into the vicinity of the central mbh ( e.g. , * ? ? ? * ) and then be tidally broken up ( e.g. , * ? ? ?
* ; * ? ? ?
one component of a broken - up binary may be ejected out as a hypervelocity star ( hvs ) as discovered in the galactic halo ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and the other component may be captured onto a tighter orbit similar to that of the gc s - stars as proposed by @xcite .
if hvss were initially originated from a stellar structure like the cws disk , they may be spatially located close to the disk plane @xcite .
the current observations do show such a spatial correlation between the hvss and the cws disk , which suggests that majority of the hvss originate from the cws disk @xcite .
the hvss discovered in the galactic halo and the gc s - stars in the vicinity of the central mbh may naturally link to each other as they may both be the products of the tidal breakup of stellar binaries in the vicinity of the central mbh ( e.g. , * ? ? ?
therefore , it is interesting to simultaneously investigate the properties of the hvss in the galactic halo ( or the gc s - stars ) and their captured ( or ejected ) companions , and probability distribution of these properties . under the assumption
that both the hvss and gc s - stars are the products of tidal breakup of stellar binaries , the working hypothesis in this paper , we construct a number of monte carlo models to simulate the tidal breakup processes of stellar binaries in the gc and check whether these models can accommodate the current observations , and make further predictions on both the companions of hvss and that of gc s - stars for future observations .
this paper is organized as follows . in section [ sec :
overview ] , we overview the tidal breakup processes of stellar binaries in the vicinity of an mbh and the dynamical connection between the ejected and captured components . adopting relatively realistic initial conditions , we perform a large number of three - body experiments to realize the tidal breakup processes of stellar binaries in section [ sec : num_simu ] . assuming a constant injection rate of stellar binaries into the vicinity of the central mbh and adopting the results from the three - body experiments on the ejected and captured components , we use the monte carlo simulations to produce both the hvss and the gc s - stars . in section [ sec : orb_ev ] , we follow the orbital evolution of the captured stars to the present time by adopting the autoregressive moving average ( arma ) model @xcite , in which both the non - resonant relaxation ( nr ) and the resonant relaxation ( rr ) are included .
the simulated gc s - stars appear to be compatible with the observations of the gc s - stars . in section [ sec :
hvss ] , we investigate the effects of different binary injection models on the number ratio of the simulated hvss to gc s - stars .
the number ratio given by observations can be reproduced if the initial mass function ( imf ) of the primary components of stellar binaries is somewhat top - heavy . by calibrating the injection models with observations
, we estimate the number of the captured ( or ejected unbound ) stars , as the companions of hvss ( or gc s - stars ) , that could be detected in the future .
we also estimate the probability to have less massive stars captured on an orbit within that of s2 in section [ sec : innermost ] .
conclusions are given in section [ sec : conclusion ] .
for clarity , some notations of the variables that are frequently used in this paper are summarized in table [ tab : t1 ] .
given a physical variable @xmath19 ( e.g. , mass , velocity , semimajor axis , eccentricity ) , the distribution function of @xmath19 is denoted by @xmath20 so that @xmath21 represents the number of relevant objects with variable @xmath19 being in the range @xmath22 .
ll @xmath23 & mass of the central mbh + @xmath24 & mass of the primary component of an injecting stellar binary + @xmath25 & mass of the secondary component of an injecting stellar binary + @xmath26 & total mass of an injecting stellar binary , i.e. , @xmath27 + @xmath28 & @xmath29 + @xmath30 & initial semimajor axis of an injecting stellar binary + @xmath31 & initial pericenter distance of the mass center of the injecting stellar binary to the mbh + @xmath32 & initial semimajor axis of the orbit of an injecting stellar binary rotating around a central mbh + @xmath33 & initial velocity of the injecting stellar binary at infinity if the binary is on a hyperbolic orbit + @xmath34 & initial energy of the stellar binary + @xmath35 & tidal radius for the stellar binary + @xmath36 & orbital penetration parameter of the injecting stellar binary ( @xmath37 ) + @xmath38 & exponent of the power - law distribution of @xmath30 + @xmath39 & exponent of the power - law distribution of @xmath31 + @xmath40 & exponent of the power - law distribution of @xmath24 + & + @xmath41 & mass of the component that gains energy during the tidal breakup of a stellar binary + @xmath42 & mass of the component that loses energy during the tidal breakup of a stellar binary + @xmath43 & @xmath44 + @xmath45 & exchange energy between the two components during the tidal breakup of a stellar binary + & + @xmath46 & mass of the ejected star after the tidal breakup of a stellar binary + @xmath47 & mass of the captured star after the tidal breakup of a stellar binary + @xmath48 & velocity of the ejected component at infinity + @xmath49 & orbital semimajor axis of the captured component + @xmath50 & orbital semimajor axis of the captured component if the injecting binary is initially on a parabolic orbit + @xmath51 & orbital eccentricity of the captured component + @xmath52 & simulated total number of the ejected stars given a mass range + @xmath53 & simulated total number of the captured stars given a mass range + @xmath54 & simulated number of the detectable hvss at the present time for given selection criteria + @xmath55 & simulated number of the detectable captured stars at the present time for given selection criteria + @xmath56 & simulated fraction of the ejected stars that survive to the present time on the main sequence + @xmath57 & simulated fraction of the captured stars that survive to the present time on the main sequence + @xmath58 & simulated fraction of the captured stars that have already been tidally disrupted until the present time + @xmath59 & simulated fraction of the captured stars that can be detected at the present time for given selection criteria + [ tab : t1 ]
a stellar binary may be broken up if it approaches an mbh within a distance of @xmath60 , where @xmath61 is the mass of the mbh , @xmath62 is the semimajor axis of the binary , @xmath63 is the total mass of the binary , and @xmath64 and @xmath65 are the masses of the two components of the binary , respectively . during the breakup ,
one component of the binary , denoted as @xmath64 here , gains energy , and the other component @xmath65 loses energy . for an injecting stellar binary that is initially on a parabolic orbit relative to the mbh , the velocity of the binary mass center at its periapsis to the mbh ( @xmath66 ) is @xmath67 .
the component @xmath64 receives a velocity change on the order of @xmath68 if the eccentricity of the stellar binary is @xmath69 , and it gains energy @xmath70 .
the other component @xmath65 loses the same amount of energy @xmath45 .
if @xmath45 is sufficiently large , the component @xmath64 may manifest itself as an hvs with velocity at infinity @xmath71 if ignoring the deceleration due to the galactic gravitational potential .
the root mean square ( rms ) of @xmath48 is approximately @xmath72 where @xmath73 and @xmath74 is given by @xcite for injecting binaries on hyperbolic orbits with initial velocities at infinity of @xmath75 , i.e. , @xmath76 where the penetration parameter @xmath77 characterizes the minimum distance where the binary approaches the mbh , and @xmath31 is the initial pericenter distance of the binary .
the rms velocity @xmath78 apparently depends on the semimajor axis , the total mass and the mass ratio of the stellar binary , and the penetration parameter @xmath36 . the exact value of @xmath48 of the ejected component for any given stellar binary also depends on the relative orientation of the stellar binary orbital plane to the orbital plane of the binary rotating around the mbh and the orbital phases of the two components at the time of its breakup .
this dependence introduces a scatter of @xmath48 around the value @xmath79 given by equation ( [ eq : vhvs ] ) , as the orbital orientations of the injecting stellar binaries are probably random and the orbital phases of the two components are not fixed at the breakup time .
numerical simulations have shown that this scatter is approximately gaussian with a dispersion of @xmath80 @xcite , where the binary orbital orientations are assumed to be randomly distributed .
the symmetry of the orbital phases of the two binary components ( always at the opposite side to the mass center of the binary ) ensures the same probability of receiving energy for each star , which leads to the same ejection probability for both components if the injecting binaries are initially on parabolic orbits @xcite .
stellar binaries on orbits bound to the mbh may experience multiple close encounters with the mbh and the binary semimajor axes and eccentricities may be cumulatively excited to larger values until finally being broken up @xcite .
the distribution of @xmath48 for the ejected stars , produced during the first encounters of the binaries with the mbh , follows a fitting formula similar to equation ( [ eq : fd1 ] ) over @xmath81@xmath82 , i.e. , @xmath83 , as the initial bounding energy of the injecting stellar binaries is still significant ( for details , see * ? ? ?
* ) . for multiple encounters ,
the energy exchange @xmath45 between the two components is determined by the properties of the stellar binaries at the final revolutions .
our simulations show that @xmath78 of those ejected components for stellar binaries broken up within @xmath84 revolutions around the mbh still follows equation ( [ eq : vhvs ] ) , but @xmath74 is now best fitted by @xmath85 for @xmath86 . for stellar binaries on bound orbits ,
the light component has a larger probability to escape away from the mbh because the specific energy it could gain is generally larger than that of the heavy component in a counterpart case ( see equation [ eq : vhvs ] ) .
however , the difference in the ejection probability for the two components of the stellar binaries is significant only when the mass ratio of the massive ones to the light ones @xmath87 and @xmath88 is close to its initial bounding energy @xmath89 , where @xmath90 denotes the initial semimajor axis of the binary system composed of a stellar binary and the mbh ( see also * ? ? ?
* ; * ? ? ?
the component @xmath65 of a broken - up stellar binary loses energy by an amount of @xmath45 and it is captured onto a tighter orbit with semimajor axis @xmath91 . according to the energy conservation law , we roughly have mv_^2- e _ , [ eq : dk_b ] where @xmath92 is the initial energy of the stellar binary , and it is @xmath93 if the binary is initially on a hyperbolic orbit , or @xmath94 if on a bound orbit . the initial internal mechanical energy of the stellar binary @xmath95 is ignored in equation ( [ eq : dk_b ] ) .
we now have the general form for @xmath48 as @xmath96 . if @xmath97 , the semimajor axis of the captured star is @xmath98 where @xmath99 . for the cases considered in this paper , the injecting stellar binaries are either initially on hyperbolic orbits ( but close to parabolic ones ) or from stellar structures like the cws disk , and thus @xmath100 as approximately @xmath97 . equation ( [ eq : acap0 ] ) shows the connection between the properties of the captured stars left in the gc and that of their ejected companions in the galactic bulge and halo . for those hvss discovered in the galactic halo with @xmath101-@xmath102 , for those detected hvss
depends on the galactic potential model adopted , the values here are obtained from the galactic potential model given by @xcite . ]
their companions left in the gc may be initially on orbits with semimajor axis in the range of @xmath103-@xmath104 as stellar binaries with extreme mass ratios are rare . for the innermost s - star ,
i.e. , the s2 , of which the semimajor axis is @xmath105 , its companion ejected out should have @xmath106 if @xmath107 , and @xmath108 if @xmath109 , respectively .
for stellar binaries initially tightly bound to the mbh , we may have @xmath110 , the component gaining energy either remains bound to the mbh or is ejected out with low velocities .
the distribution of hvs properties is directly connected to the distribution of s - star properties ( note that here we do not mean that an observed hvs in the galactic halo is directly associated with an observed gc s - star as the products of the tidal breakup of the same binary star ) . according to equation ( [ eq : acap0 ] )
, the distribution of the semimajor axis of the captured stars @xmath111 is related to the distribution of the velocity of hvss at infinity @xmath112 if @xmath113 , i.e. , f _
^3/2f_()|_= , [ eq : va ] which suggests that any one of the two distributions above can be inferred from the other one .
the velocity distribution of the ejected stars are mainly determined by the initial sets on the distributions of @xmath30 and @xmath31 since ^1/2 ^ -1/2g(d ) , where @xmath74 denotes the dependence of the rms velocity @xmath114 on the penetration parameter @xmath36 , as shown in equations ( [ eq : fd1 ] ) and ( [ eq : fd2 ] ) for the cases of injecting binaries initially on hyperbolic orbits but close to parabolic ones and bound orbits like the stars in the cws disk , respectively .
the fitting forms of @xmath74 obtained from numerical experiments ( equations ( [ eq : fd1 ] ) and ( [ eq : fd2 ] ) ) are decreasing functions in the range of @xmath115 or @xmath116 , and thus @xmath74 may be approximated as a monotonically decreasing function .
we assume that the initial distribution of @xmath30 and @xmath31 are @xmath117 and @xmath118 , respectively ; and the probability of a stellar binary with semimajor axis @xmath30 broken up by the central mbh at a penetration distance @xmath36 is only a function of @xmath36 ,
i.e. , @xmath119 ( see * ? ? ?
if @xmath120 and ignoring the scatter of @xmath48 around @xmath121 ( i.e. , @xmath122 ) , the velocity distribution of the ejected components can be obtained as @xmath123 and this relation is valid only if @xmath74 is a monotonically decreasing function and it is independent of the detailed form of @xmath119 .
similarly , we also have f _ ( ) ^++1 , [ eq : fac ] which is consistent with the simple relation given by equation ( [ eq : va ] ) .
the estimated slope of @xmath124 ( or @xmath125 ) above is not affected by taking account of the gaussian - like scatter of @xmath48 around @xmath78 as the distribution is a power law . if considering of the various mass ratios among the injecting binaries ( see equations ( [ eq : vhvs ] ) and ( [ eq : acap0 ] ) ) , however , the resulted slope of @xmath124 may be somewhat flatter than the simple estimates above . note also that a larger @xmath39 may correspond to a slower migration / diffusion of stellar binaries into the low angular momentum orbits or the vicinity of the central mbh , and lead to fewer hvss at the high - velocity end and fewer captured stars in smaller distances to the mbh .
the periapsis of a captured star @xmath65 is roughly @xmath126 and the characteristic eccentricity of the captured star is |e_~1 - 1- ( ) ^1/3 .
[ eq : ecc ] the @xmath127 depends on @xmath43 and the mass ratio of the captured star to the mbh .
considering of the gaussian - like scatter in @xmath48 and correspondingly the scatter in @xmath91 , the probability that the breakup of a stellar binary with given semimajor axis and mass of each component results in a captured star with eccentricity @xmath128 is roughly @xmath129 , \label{eq : ecprob}\end{aligned}\ ] ] where @xmath127 is given by equation ( [ eq : ecc ] ) and @xmath130 . to capture an s2-like star ( i.e. , @xmath131 and @xmath132 ; see @xcite ; @xcite ) directly through the tidal breakup of stellar binaries , it is necessary to have @xmath133 ( @xmath134 ) and the probability is @xmath135 if @xmath136
according to equation ( [ eq : ecprob ] ) ( see * ? ? ?
the probability to capture stars onto orbits with @xmath137 is only @xmath138 if @xmath139 ; and @xmath140 even if @xmath141 ( see also @xcite ) . since the number of gc s - stars is only on the order of a few tens , it is difficult to produce
all the nine observed gc s - stars with eccentricities @xmath142 @xcite directly by the tidal breakup of stellar binaries .
in addition , the captured stars may initially remain on a disk plane if their progenitor binaries are originated from disk - like stellar structure(s ) as suggested by @xcite , which is different from the isotropic distribution of the gc s - stars .
therefore , additional physical mechanism is required to further make the captured stars evolve to orbits with lower eccentricities and spatially isotropically distributed if the gc s - stars are really originated from the tidal breakup of binary stars .
the processes , initially proposed by @xcite , may cause the captured stars dynamically evolving to their present orbits as discussed by a number of authors @xcite . in section [ sec : orb_ev ]
, we will take into account the relaxation processes , including rr , to approximately follow the dynamical evolution of each `` gc s - star '' after its capture due to the tidal breakup of stellar binaries ; we then check whether the eccentricity and spatial distributions of those surviving `` gc s - stars '' are compatible with current observations .
note here we use the quotes around the term gc s - stars to represent all of those captured stars with mass in the range of @xmath143-@xmath144 ; while the simulated gc s - stars ( without quotes ) represent those with mass @xmath143-@xmath144 surviving to the present time , which presumably correspond to the observed ones ( see section [ sec : orb_ev ] ) .
in this section , we first adopt monte carlo simulations to realize the tidal breakup processes and generate hvss and `` gc s - stars '' , and then we investigate in detail the connection between the simulated hvss and `` gc s - stars '' .
we use the code dorpi5 based on the explicit fifth ( fourth)-order runge
kutta method @xcite to calculate the three - body interactions between a stellar binary and the central mbh . for details of the numerical calculations ,
see @xcite .
the successive dynamical evolution of the captured stars in the gc and the kinematic motion of the produced hvss in the galactic potential will be discussed in sections [ sec : orb_ev ] and [ sec : hvss ] , respectively .
the mass of the central mbh is set to be @xmath145 throughout the numerical calculations in this paper @xcite . for the injecting stellar binaries ,
the initial conditions are set as follows : * the distribution of the semimajor axes @xmath146 follows the pik law , i.e. , @xmath147 ( e.g. , * ? ? ?
* ) . * the mass distribution of the primary stars @xmath24 follows a power law function , @xmath148 .
the distribution of the secondary star ( @xmath25 ) or the mass ratio @xmath149 can be described by two populations : ( 1 ) a twin population , i.e. , about 40% of binary stars have @xmath150 , and ( 2 ) the rest binaries , which follow a distribution of @xmath151 @xcite . * the initial eccentricity of the injecting binary
is assumed to be @xmath152 , as adopted in previous works ( e.g. , @xcite ; @xcite ) .
, the velocities of the resulted hvss from the four models are roughly smaller than those obtained for @xmath153 by @xmath154 .
] * the orientation of the inner binary orbital plane is chosen to be uniformly distributed in @xmath155 for @xmath156 .
for the orbits of the injecting stellar binaries relative to the central mbh , the initial conditions are set as follows : * the stellar binaries are assumed to be initially injected from either disk - like stellar structures ( similar to the cws disk ) or infinity . if they were from structures like the cws disk , the semimajor axes @xmath90 follows a power - law distribution proportional to @xmath157 in the range of @xmath15-@xmath16 according to current observations on the cws disk @xcite .
if they were from infinity , i.e. , unbound to the mbh , their initial velocities at infinity are set to @xmath158 . *
if the injecting binaries were from disk - like stellar structures , the orientations of their orbits relative to the mbh are assumed to satisfy a gaussian distribution around the central planes of the stellar disks with a standard deviation of @xmath159 ( cf .
* ; * ? ? ?
the planes of the host disks are assumed to be the same as the two planes that best fit the observations , i.e. , @xmath160 and @xmath161 , respectively , and these two planes are consistent with the cws disk plane and the plane of the northern arm ( narm ) of the mini - spiral in the gc ( or the outer warped part of the cws disk ; @xcite ) .
unbound hvss in the northern sky and they are still consistent with being located on two planes revealed by @xcite . that is , one of the disk planes is consistent with the cws disk plane , while the other disk plane is more consistent with the warped outer part of the cws disk and slightly deviates from the narm plane . in this paper
, we do not distinguish the narm plane from the warped outer part of the cws disk . ] the injection rates from these two disks are assumed to be the same .
* the periapsis that the injecting stellar binaries approach the mbh is simply assumed to follow a power law distribution , @xmath162 and @xmath163 .
a larger value of @xmath39 corresponds to a smaller fraction of the injecting stellar binaries that could approach the immediate vicinity of the central mbh .
it is still not clear which mechanism is responsible for the migration ( or diffusion ) of stellar binaries into the vicinity of the central mbh , although the secular instability developed in a stellar disk is proposed to be a viable one @xcite . instead of incorporating the detailed migration / diffusion process of the stellar binaries in the monte carlo simulations below , we choose to parameterize the migration / diffusion process qualitatively by different values of @xmath39 and a larger @xmath39 corresponds to a slower migration / diffusion process .
lcccc unbd - ms0 & -2.7 & 0 & @xmath164 & 250 + disk - ms0 & -2.7 & 0 & 0.04 - 0.5 & @xmath164 + disk - th0 & -0.45 & 0 & 0.04 - 0.5 & @xmath164 + disk - th2 & -0.45 & 2 & 0.04 - 0.5 & @xmath164 [ tab : t2 ] in this section , we perform monte carlo simulations by adopting four sets of initial conditions ( as listed in table [ tab : t2 ] ) . in the first model ,
the stellar binaries are assumed to be injected from infinity with initial velocity of @xmath165 .
for the primary components of the injecting binaries , we adopt the miller scalo imf ( e.g. , @xcite ) .
this model is denoted as `` unbd - ms0 '' . for the other three models ,
the stellar binaries are assumed to be originated from stellar structures like the cws disk , and the imf of the primary components is either set to be the miller scalo imf or a top - heavy imf with a slope of @xmath166 as suggested by recent observations of the disk stars ( see * ? ? ?
the slope of the initial distribution of the pericenter distance @xmath39 is set to be either @xmath69 or @xmath167 .
these models are denoted as `` disk - ms0 '' , `` disk - th0 '' , and `` disk - th2 '' , respectively .
the total number of three - body experiments is @xmath168 for each model with the initial settings described above . by comparing the results obtained from those different models ,
one may be able to distinguish the effects of different settings on the imf and the injection of stellar binaries .
if not specified , those ejected or captured stars with mass in the range of @xmath169-@xmath144 are recorded , thus both the hvss with mass @xmath169-@xmath170 and the captured stars with mass @xmath143-@xmath144 , corresponding to the currently detected ones , can be taken into account simultaneously .
the ejected or captured stars with mass in the range of @xmath171-@xmath3 are also considered for completeness . for other ejected or captured stars with mass out of the range of @xmath172-@xmath144 , they may be either too faint to be detected or too massive with too short lifetime and thus with too small probability to survive .
figure [ fig : f1 ] shows the distribution of the tidally broken - up stellar binaries in the @xmath48-@xmath91 plane , where @xmath48 is the velocity at infinity of the ejected component and @xmath91 is the semimajor axis of the captured component . as shown in panel
( a ) , the majority of the simulated @xmath48-@xmath91 pairs obtained from the unbd - ms0 model are close to the one estimated from equation ( [ eq : acap0 ] ) by setting @xmath173 ( solid line ) .
the main reasons for this are : ( 1 ) the majority ( @xmath174 ) of the injecting stellar binaries have mass ratios @xmath175 in the range of ( 1/2 , 2 ) under the assumption of two populations set for the stellar binaries ; and ( 2 ) all the injecting binaries have the same but negligible initial energy @xmath92 .
a small number of @xmath48-@xmath91 pairs , which apparently deviate significantly away from the solid line ( for @xmath176 obtained from equation [ eq : acap0 ] ) , are due to the breakup of the binaries with @xmath43 substantially larger or smaller than @xmath177 ( below or above the solid line ) . for the other three models ,
the simulation results do not deviate far away from the simple predictions by equation ( [ eq : dk_b ] ) ( for @xmath176 ) , except that fewer ejected stars at the high - velocity end are produced in the disk - th2 model than in the other models simply because not many stellar binaries can closely approach the mbh .
the scatters of @xmath48 around that predicted by equation ( [ eq : dk_b ] ) in panels ( b)-(d ) are more significant compared with that in panel ( a ) , which is caused by one or the combination of the effects as follows : ( 1 ) a distribution of the negative initial energy of the injecting stellar binaries originated from stellar structure like the cws disk ( panels ( b)-(d ) ) ; ( 2 ) relatively more progenitor binaries have @xmath43 substantially larger or smaller than @xmath177 in the cases with a top - heavy imf ( panels ( c ) and ( d ) ) ; and ( 3 ) fewer stellar binaries approach the immediate vicinity of the central mbh in the case of a large @xmath39 ( panel ( d ) ) . as seen from figure [ fig : f1 ] ,
if the observed gc s - stars , with semimajor axes @xmath178-@xmath179 , are produced by the tidal breakup of stellar binaries , their ejected companions are expected to have @xmath180-@xmath181 in the unbd - ms0 model and @xmath182-@xmath11 in the other models .
the disk - th2 model produces fewer ejected stars with @xmath48 substantially larger than @xmath102 compared with other models .
the captured companions of the detected hvss in the galactic halo are more likely to have @xmath183-@xmath184 in the unbd - ms0 model , which is consistent with the simple estimation by equation ( [ eq : acap0 ] ) , and have @xmath185-@xmath184 in the other models .
the travel / arrival time of the detected hvss from the gc to its current location is on the order of @xmath186 myr ( e.g. , * ? ? ?
* ) , which suggests that their companions were captured @xmath187 myr ago and the orbits of the captured companions may have been changed due to the dynamical interactions with its environment ( see section [ sec : orb_ev ] ) . as shown in figure [ fig : f1 ] , for those captured stars with @xmath188 , the probability that they have ejected companions with @xmath189-@xmath102 is negligible .
figure [ fig : f2 ] shows the distribution of those captured stars obtained from each model in the @xmath91 versus @xmath190 plane , where @xmath91 and @xmath191 are their semimajor axes and eccentricities achieved right after they were captured , respectively .
relatively more captured stars with low eccentricities are produced by the disk - ms0 model than by the unbd - ms0 model ( see panels ( a ) and ( b ) ) mainly because the injecting binaries can be broken up at relatively larger distance in the disk - ms0 model due to multiple encounters . and relatively more captured stars with low eccentricities and @xmath91 are produced in the disk - th model than that in the unbd - ms0 model ( see panels ( a ) and ( c ) ) because there are more injecting binaries with mass ratio @xmath43 substantially less than @xmath177 and @xmath192-@xmath193 for a top - heavy imf than that for the miller scalo imf .
the disk - th2 model produces relatively more captured stars with smaller eccentricities for any given @xmath91 than the disk - th0 model ( as shown in panels ( c ) and ( d ) ) , as those binaries are generally broken up at even larger distances in the disk - th2 model because fewer binaries can approach the very inner region due to the steepness of the adopted @xmath194
. however , the eccentricities of those captured stars , even produced in the disk - th2 model , are still statistically significantly higher than that of the observed gc s - stars .
the orbits of a number of gc s - stars , including s2 , can be directly produced in the disk - th0 model and the disk - th2 model if the injection rate of binaries is around a few times @xmath195 to @xmath196 as set for those models ( see similar rates obtained by @xcite ) .
according to figure [ fig : f2 ] , apparently it is extremely difficult to produce `` gc s - stars '' with @xmath137 directly through the tidal breakup mechanism of stellar binaries in the vicinity of the mbh .
note also that fewer captured stars with @xmath197 are produced in the disk - th2 model compared with those in other models because stellar binaries are harder to approach the innermost region than that in other models ( see panel ( d ) in figures [ fig : f1 ] and [ fig : f2 ] ) . according to the simulations above
, we find that @xmath198 of the detected hvss should have captured companions in the gc with mass @xmath169-@xmath170 as shown in the left panel of figure [ fig : f3 ] . and
similarly @xmath199 of the observed gc s - stars should have ejected companions with mass @xmath143-@xmath144 as shown in the right panel of figure [ fig : f3 ] . to find the possible counterparts of those current observed gc s - stars and hvss
, we will focus on the ejected stars with mass @xmath200-@xmath144 in the galactic bulge and halo and the captured stars with mass @xmath169-@xmath170 in the gc . for completeness
, we also count the ejected and captured stars with mass @xmath171-@xmath3 produced in all the models ( see table [ tab : t4 ] ) .
the orbits of the captured stars produced by the tidal breakup of stellar binaries may evolve due to dynamical interactions with the surrounding environments . in principle , two - body interactions between stars may cause exchanges of their angular momenta and energy . however , the timescale of the two - body relaxation ( @xmath201 yr ) is too long for it to be effective in changing the orbits of captured stars within their main - sequence lifetime ( e.g. , * ? ? ?
* ; * ? ? ?
the rr is an important dynamical process naturally resulted from the coherent torques between orbital averaged mass wires of stars moving in near - keplerian potential proposed by @xcite , which can lead to changes in both the eccentricities ( scalar rr ) and orientations ( vector rr ) of stars moving in the gc ( e.g. , * ? ? ?
the rr appears much more effective in changing the orbital configuration of the `` gc s - stars '' than the non - resonant two - body relaxation ( nr ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
therefore , the scalar and vector rr may be crucial in the follow - up dynamical evolution of the orbits of the captured stars .
the vector rr timescale is @xmath202-@xmath203 myr in the region hosting the `` gc s - stars '' , about one order of magnitude smaller than the scalar rr timescale @xcite , and thus the captured stars can evolve to an isotropic distribution on a timescale of @xmath7 myr @xcite even if they were originally on a plane - like structure . in this paper , we assume that the isotropic distribution of the gc s - stars can always be reproduced through the vector rr of the simulated captured stars within a timescale shorter than their lifetime . it has been suggested that the high - eccentric orbits of captured stars can dynamically evolve to that of the observed gc s - stars through the scalar rr within @xmath204 myr ( e.g. , * ? ? ?
* ) . however , previous studies assume a simple distribution of the initial eccentricities and semimajor axes of the captured stars ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . in this section ,
we adopt the distribution of @xmath191 and @xmath91 resulted from the injection models studied in section [ sec : num_simu ] .
we follow the evolution of @xmath191 and @xmath91 by taking both the rr and nr into account , and then compare the @xmath191 and @xmath91 distributions of the captured stars surviving to the present time with that of the observed gc s - stars .
we adopt the arma model first introduced by @xcite to perform monte carlo simulations of the long - term evolution of the captured stars . in the arma model , the rr phase and the nr phase
are unified , and the general relativistic ( gr ) precession of stars in the potential of the central mbh is also simultaneously included .
the arma model is characterized by the following three parameters : ( 1 ) the autoregressive parameter @xmath205 ; ( 2 ) the moving average parameter @xmath206 ; and ( 3 ) the parameter @xmath207 , which is the variance of a random variable @xmath208 following the normal distribution . in the arma model
, @xmath208 represents the random walk motion of the nr phase . at a time step of one orbital period of a star ,
the variation in the absolute value of its angular momentum is _ 1 j_t=_1_1j_t-1+_1^(1)_t-1 + ^(1)_t , and _ 1= , t_=f_min[t_prec(a , e),t_prec(a , ) ] ,
s= , given by ( * ? ? ?
* see their equation 33 ) , i.e. , the minus sign before @xmath209 was missing . ]
e_crit(a , e)= ( ) , _ 1=- , = a_e , _ 1=f _ , f_=0.52 + 0.62e-0.36e^2 + 0.21e^3 - 0.29 , where @xmath210 , @xmath211 , @xmath212 , @xmath213 , @xmath214 , @xmath215 , @xmath216 is the averaged mass of the field stars , @xmath217 and @xmath218 are the semimajor axis and orbital period of the star
, @xmath219 is the median value of the eccentricity of the field stars and it is @xmath220 for a thermal distribution , @xmath221 is the combined precession timescale for the newtonian precession and the general relativity precession ( see equations ( 25 ) , ( 27 ) , and ( 28 ) in * ? ? ?
* ) , and @xmath222 is the total number of stars within the radius equal to the semimajor axis of the captured star .
adopting a simple stellar cusp model , i.e. , @xmath223 , we have @xmath224 , where @xmath225 represents the radius within which the mass of stars equals the mbh mass and @xmath226 is the total number of field stars within @xmath225 . similar to @xcite
, we also assume a bahcall wolf cusp ( @xmath227 , @xcite ) unless otherwise stated and correspondingly @xmath228 .
the variable superscript ` ( 1 ) ' in the above equations means that the time step is one orbital period of the star being investigated . in the time step of @xmath229 period of the star ( @xmath230 )
, the model parameters ( @xmath231 , @xmath232 , @xmath233 ) can be obtained from parameters of one period ( @xmath205 , @xmath206 , @xmath207 ) . for further details of the arma model ,
see @xcite .
the arma model may not capture the exact dynamical physics of the system and give the exact kinematics of each individual star ; but for the purpose of our work and the addressing problems , it should be plausible and efficient to be applied here to obtain the evolution of the system in a statistical way .
the two - body nr is also taken into account in a way similar to that in @xcite . in a time step @xmath234 ,
the energy change of a captured star due to the nr is given by e = e ( ) ^1/2 , where @xmath235 is an independent normal random variable with zero mean and unit variance , and @xmath236 is the nr timescale in the gc given by t_nr = a_nr()^2 t_p .
the travel time of those hvss discovered in the galactic halo is @xmath237-@xmath238 myr if hvss were ejected from the gc @xcite . and
recent observations have also shown that the age of the detected hvss is on the order of @xmath187 myr , which is consistent with the gc origin @xcite . in the model of this paper
, we are unifying the formation of both the hvss and the gc s - stars by the tidal breakup of young stellar binaries originated from the young stellar disk(s ) in the gc . in order to be compatible with the above observations , we assume a constant injection rate of stellar binaries over the past @xmath238 myr myr or so ; @xcite )
, the observations have also shown that there are many b - dwarf stars in the disk region ( @xmath239 pc ) .
for example , @xcite find @xmath240 b - dwarfs in the disk region and the ages of these dwarfs could be substantially larger than @xmath241 myr .
note also that the observational bias on the detection of young but faint stars in the inner parsec is largely uncertain .
thus , current observations do not exclude the existence of less massive stars with ages much longer than those of the detected wolf rayet and o / b type disk stars in the gc .
we adopt the numerical results of the three - body experiments for each injection model in section [ sec : num_simu ] and calculate the energy and angular momentum evolution for each captured star by using the arma model . in these calculations , we also take account of the effects of the limited lifetime of the captured stars on the main sequence and the tidal disruption of those captured stars moving too close to the mbh .
we remove those captured stars if they move away from the main sequence or approach the mbh within a distance of @xmath242 , where @xmath243 and @xmath244 is the stellar radius . finally , we obtain the present - day semimajor axis and eccentricity distributions of the captured stars , which can be used to be compared to the observational distributions and constrain the models . the left panel of figure [ fig : f4 ] shows the cumulative distributions of @xmath91 of the captured stars surviving to the present time and that of the observed gc s - stars .
the thick lines represent the captured stars with mass @xmath143-@xmath144 surviving to the present time ( the simulated gc s - stars ) , roughly corresponding to the observed gc s - stars , and the thin lines represent the captured stars with mass @xmath169-@xmath170 , roughly corresponding to the captured companions of those hvss detected in the galactic halo .
although the lifetime of less massive stars on the main sequence is substantially longer and thus the dynamical evolution time is longer than that of the massive ones , the cumulative distribution of @xmath91 of the light captured stars is only slightly different from that of the massive ones ( see the left panel of figure [ fig : f4 ] ) .
the slope of the @xmath91 distribution is affected most by the @xmath194 distribution .
relatively fewer captured stars are produced in the inner region by the disk - th2 model compared with that obtained by the other models , and the fraction of captured stars with semimajor axes @xmath245 is proportional to @xmath246 for the disk - th2 model but to @xmath247 for the other three models .
for those models with @xmath248 , the @xmath91 distributions obtained from the numerical simulations is roughly consistent with the simple estimations from equation ( [ eq : fac ] ) ( e.g. , the slope is @xmath177 for @xmath147 and @xmath248 according to equation ( [ eq : fac ] ) ) . for the disk - th2 model , however , the @xmath91 distribution seems flatter than the simple expectation , i.e. , a slope of @xmath172 ( for @xmath147 and @xmath249 ) .
the flatter slope of @xmath250 resulted from the disk - th2 model may be due to the effect of various mass ratios of the injecting binaries , which is included in the numerical simulations but ignored in deriving equation ( [ eq : fac ] ) ( see panel ( d ) in figure ( [ fig : f1 ] ) ) .
the kolmogorov simirnov ( k - s ) tests find the likelihoods of @xmath251 , @xmath252 , @xmath253 , and @xmath254 that the @xmath91 distribution of the observed gc s - stars is the same as that of the simulated gc s - stars for the four models , respectively .
as shown in figure [ fig : f4 ] , most of the discrepancy between the observational distribution and the simulated one is apparently near the edges of those distributions .
since the anderson darling ( a - d ) test may be more effective than the k - s test and more sensitive to the distribution edges ( see * ? ? ?
* ) , we also adopt the a - d test here and find the likelihoods are @xmath253 , @xmath255 , @xmath256 , and @xmath257 for the four models , respectively .
these statistical tests suggest that the disk - ms0 model and the disk - th2 model may be more compatible with the observational @xmath91 distribution . for the majority of the simulated gc s - stars ( @xmath143-@xmath144 ) ,
the relative changes in their energy due to dynamical relaxation after their capture are less than @xmath25830% and their semimajor axes do not deviate much from the initial values right after their capture .
therefore , the distribution of the semimajor axis of currently observed gc s - stars can provide some information on their ejected companions ( see equations ( [ eq : dk_b ] ) and ( [ eq : acap0 ] ) ) .
for those captured stars with mass @xmath169-@xmath170 , however , their relative energy changes can be as large as @xmath177 mainly because of their longer lifetime and thus longer dynamical evolution time , and equation ( [ eq : acap0 ] ) is no longer reliable to provide estimations on the velocity of the ejected companions of those less massive captured stars by using their present - day semimajor axes .
the right panel of figure [ fig : f4 ] shows the cumulative eccentricity distributions obtained from different models .
as seen from figure [ fig : f4 ] , the @xmath191 distributions are only slightly different for different injection models because the initial @xmath191 are all close to @xmath177 in all the models . for those captured stars with mass @xmath169-@xmath170 , their present eccentricities are relatively lower than that of the observed gc s - stars because of their longer dynamical evolution time . for those simulated gc s - stars ( with mass @xmath200-@xmath144 ) resulted from any of the four injection models , their @xmath191 distribution is similar to that of the observed gc s - stars .
the k - s tests find a likelihood of @xmath259 that the eccentricity distribution of the observed s - stars is the same as that of the simulated gc s - stars for all the four models .
if alternatively adopting the a - d test , then the likelihoods are @xmath260 , and @xmath261 for the unbd - ms0 model , the disk - ms0 model , the disk - th0 model , and the disk - th2 model , respectively . according to these calculations
, the disk - th2 model may be more compatible with the observational @xmath191 distribution .
we note here that the simulations slightly over - produce the stars with high eccentricities ( @xmath191 close to @xmath177 ) with respect to the observations because the rr for those captured stars with extremely high eccentricities is quenched by the strong relativistic precession @xcite .
for this inconsistency , part of the reason might be the observational bias in detecting the gc s - stars , i.e. , the stars with high eccentricities are less likely to be detected at @xmath262 @xcite ; and part of the reason might be the limitation of the arma model .
but the main reason of the inconsistency does not appear to be due to ignoration of the `` bouncing effect '' demonstrated in figure 7 of @xcite , where the star starting from a low - eccentricity orbit and evolving close to a critical high - eccentricity orbit is then bounced back onto a low - eccentricity orbit due to the suppression of the rr by the fast gr precession at high - eccentricity orbits , as ( 1 ) the stars in our model are captured from tidal breakup of binary stars and they initially have eccentricities even higher than the critical eccentricities ; ( 2 ) during the simulation period , some of the stars have evolved onto low - eccentricity orbits as illustrated by the distribution at the low - eccentricity end in figure [ fig : f4 ] , and the simulated stars at the high - eccentricity end are those that evolve relatively slowly ; and ( 3 ) the suppression of the rr due to the fast gr precession has been modeled in our work , as mentioned before ( e.g. , see the definition of @xmath221 above ) .
the timescale of the rr process also depends on the mass of the field stars @xcite . to check this dependence ,
we perform additional simulations by setting the mass of field stars to @xmath263 or @xmath264 but with the total mass of the field stars fixed . according to the results of these simulations , we find that the simulated gc s - stars are on orbits with too high eccentricities compared with the observational ones if @xmath263 , or on orbits with too low eccentricities if @xmath264 . in the above calculations , a bahcall wolf cusp for the background stellar system in the gc
is adopted .
however , recent observations suggested that the background stellar distribution may be core - like rather than cusp - like ( e.g. , @xcite ) . similarly as done in @xcite
, we also adopt @xmath265 to mimic the effect of a core - like distribution , and the perturbation on the gc `` s - stars '' is assumed to be dominated by main - sequence stars ( e.g. , see @xcite ) .
in such a model , we find that the resulted s - stars on highly eccentric orbits with smaller pericenter distances are relatively more than those obtained from the cuspy model .
the reason is that the rr is less efficient in the core - like stellar distribution , and thus the evolution of the eccentricities of gc `` s - stars '' is slower .
the timescale of the rr process becomes much longer at the distance of the disks .
stars in this region are less affected by the relaxation processes and may well preserve some of their initial orbital configurations .
observations find many b - dwarfs in disk regions , with high eccentricities and more extended spatial distribution than disk stars . the resulted eccentricity distance distribution of these stars by the rr compared with the observations may provide useful constraints on the formation of the gc s - stars @xcite .
our simulations also produce many b - dwarfs in the disk region and their radial distribution is similar to the initial input ones for the injecting binaries .
however , there should also exist b - dwarfs in the disk region that are initially formed as single stars and the fraction of these stars is not clear yet .
a detailed dynamical study for the b - dwarfs in the disk region is complicated and beyond the scope of this paper .
the ejected stars move away from the gc after the breakup of their progenitor binaries and their velocities are gradually decelerated in the galactic gravitational potential .
some of them are unbound to the galactic potential and can travel to the galactic halo and may appear as the detected hvss if their main - sequence lifetime is long enough compared to the travel time ; while others with lower ejecting velocity may return to the gc . to follow the subsequent motion of the ejected components , we adopt the milky way potential model given by @xcite , which involves four components , including the contributions from the central mbh , the galactic bulge , the galactic disk , and the galactic halo , i.e. , = _ bh+_bulge+_disk+ _ halo , where _ bh = -gm_/r , _
bulge=- , _ disk = - , _ halo = - ( 1 + ) , respectively .
the model parameters for the last three components are @xmath266 , @xmath267 , the core radius @xmath268 , the scale length @xmath269 , and @xmath270 , where @xmath271 is the cosmic critical density , @xmath272 , @xmath273 is the cosmic fraction of matter , the virial radius @xmath274 , and the concentration @xmath275 .
the bulge , the disk , and the halo potentials adopted here are all spherical . if adopting non - spherical potentials , i.e. , a triaxial bulge / halo and a flattened disk potential , the bending effect due to the non - spherical component on the trajectories of ejected stars is important only for those with @xmath276 on a timescale of @xmath277 myr , but it is negligible for hvss with relatively high speeds ( see * ? ? ? * ) . note that the radial distribution of the ejected stars surviving to the present time ( and correspondingly the predicted number of the detectable hvss ) may be slightly different if adopting a different galactic potential model .
the total number of detectable hvss depends directly not only on how many stellar binaries can be injected into the immediate vicinity of the mbh , but also on the lifetime of these stars and the detailed settings on the imf , semimajor axis , and periapsis of the injecting stellar binaries ( see section [ sec : num_simu ] ) .
if the stellar binaries are injected from a far away region , the injection rate can be estimated through the loss - cone theory @xcite ; if the injecting stellar binaries originated from central stellar disks @xcite , the injection rate is difficult to estimate as the mechanism responsible for it is still not clearly understood ( cf .
@xcite ) . in principle , it is plausible to observationally calibrate the injection rate by the numbers of the detected hvss and gc s - stars . but this calibration becomes complicated if considering of the uncertainties in the settings of the distributions @xmath278 , @xmath194 , and imf of the injecting binaries , etc .
lccccccccc unbd - ms0 & -2.7 & 0 & 3.0 & 8.8 & 0.59 & 0.27 & 0.51 & 27 + disk - ms0 & -2.7 & 0 & 1.8 & 8.9 & 0.51 & 0.15 & 0.41 & 12 + disk - th0 & -0.45 & 0 & 0.22 & 10 & 0.47 & 0.19 & 0.54 & 1.4 + disk - th2 & -0.45 & 2 & 0.27 & 9.9 & 0.34 & 0.06 & 0.21 & 1.3 + disk - im0 & -1.6 & 0 & 0.61 & 9.5 & 0.50 & 0.16 & 0.47 & 4.0 + disk - im2 & -1.6 & 2 & 0.75 & 9.4 & 0.22 & 0.01 & 0.03 & 2.8 [ tab : t3 ] lccccccccccccccc unbd - ms0 & -2.7 & 0 & 13 ( 2.2 ) & & 504 ( 79)&189 ( 30 ) & 86 ( 13 ) & & 178 ( 28 ) & 58 ( 9 ) & 33 ( 5 ) & & 60 ( 9 ) & 33 ( 5 ) & 17 ( 3 ) + disk - ms0 & -2.7 & 0 & 7.0 ( 2.7 ) & & 177 ( 79 ) & 66 ( 29 ) & 82 ( 36 ) & & 62 ( 28 ) & 21 ( 9 ) & 32 ( 14 ) & & 29 ( 13 ) & 17 ( 8) & 17 ( 8) + disk - th0 & -0.45 & 0 & 1.8 ( 5.8 ) & & 23 ( 79)&8 ( 27 ) & 11 ( 37 ) & & 28 ( 97 ) & 8 ( 30 ) & 11 ( 39 ) & & 29 ( 103 ) & 19 ( 66 ) & 17 ( 60 ) + disk - th2 & -0.45 & 2 & 6.2 ( 23 ) & & 18 ( 79)&7 ( 31 ) & 8 ( 34 ) & & 16 ( 71 ) & 5 ( 23 ) & 9 ( 40 ) & & 20 ( 87 ) & 15 ( 64 ) & 17 ( 75 ) + disk - im0 & -1.6 & 0 & 3.1 ( 3.6 ) & & 61 ( 79 ) & 22 ( 29 ) & 28 ( 36 ) & & 37 ( 48 ) & 12 ( 15 ) & 18 ( 23 ) & & 29 ( 38 ) & 18 ( 23 ) & 17 ( 22 ) + disk - im2 & -1.6 & 2 & 60 ( 100 ) & & 48 ( 79 ) & 21 ( 35 ) & 32 ( 52 ) & & 24 ( 39 ) & 7 ( 11 ) & 18 ( 29 ) & & 20 ( 33 ) & 14 ( 23 ) & 17 ( 28 ) [ tab : t4 ] the @xmath172-@xmath170 hvss detected in the galactic halo and the gc s - stars should be linked to each other under the working hypothesis of this paper .
the total numbers of the simulated @xmath172-@xmath170 hvss and @xmath279-@xmath144 gc s - stars depend not only on the injection rate of stellar binaries but also on the detailed settings on the injection models .
however , the number ratio of the simulated @xmath172-@xmath170 hvss to the simulated @xmath279-@xmath144 gc s - stars may depend only on the details of the injection models described in section [ sec : num_simu ] , but not on the injection rate of stellar binaries .
any viable model should produce a number ratio compatible with the observations on the hvss and the gc s - stars , and thus this number ratio may provide important constraints on the models . we obtain both the numbers of the simulated hvss and the captured stars surviving to the present time and their number ratio by monte carlo simulations as follows . first , we obtain the total number of initially captured ( or ejected ) stars , @xmath53 ( or @xmath52 ) , with mass in the ranges of @xmath172-@xmath170 , @xmath280-@xmath3 , and @xmath279-@xmath144 , respectively . to do this , we assume a constant injection rate of binaries and randomly set the injection events over the past @xmath238 myr , and for each injection event we randomly assign it to a three - body experiment conducted in section [ sec : num_simu ] and adopt the results from the experiment .
second , we consider the limited lifetime of each ejected and captured star , the motion of each ejected star in the galactic potential , and the dynamical evolution of each captured star , and then obtain the fraction of the ejected stars ( @xmath281 ) that still remain on the main sequence at the present time or the similar fraction for the captured stars ( @xmath57 ) , and the fraction of the captured stars ( @xmath282 ) that have already been tidally disrupted until the present time .
third , we consider the kinematic selection criteria and obtain the fractions of those ejected and captured stars according to given observational selection criteria , i.e. , @xmath283 and @xmath59 , respectively .
the selection criteria are similar to those adopted in selecting the detected hvss and gc s - stars , i.e. , an ejected star is labeled as a _ detectable _ hvs if its heliocentric radial velocity in the galactic rest frame is @xmath284 or its proper motion in the heliocentric rest frame is @xmath285 , and its velocity at infinity is @xmath286 , and captured stars with semimajor axis @xmath287 are counted as _ detectable _ gc s - stars . for those simulated hvss with mass @xmath288-@xmath170
, we put an additional cut on their distances from the gc , i.e. , from @xmath289 to @xmath290 , in order to compare them to current observations .
finally , we obtain the number ratio of the _ detectable _ hvss to the _ detectable _ gc s - stars for each model as = .
the simulation results are listed in tables [ tab : t3 ] and [ tab : t4 ] for each model . here
we comment on a few factors that affect the predicted numbers of the simulated _ detectable _ hvss and gc s - stars and consequently the number ratio of these two populations .
( 1 ) the difference between the lifetime of the simulated hvss and gc s - stars : the detected hvss are in the mass range of @xmath291-@xmath170 , which are substantially smaller than that of the observed gc s - stars ( @xmath200-@xmath144 ) .
the lifetime difference leads to an enhancement in the number ratio of the _ detectable _ hvss to the gc s - stars roughly by a factor of @xmath203 under the assumption of a constant rate of injecting stellar binaries into the vicinity of the mbh over the past @xmath238 myr .
( 2 ) the place where the stellar binaries are originated : in the unbd - ms0 model , relatively more stars with high velocities ( e.g. , @xmath292 ) are generated than those in the other models .
( 3 ) the distribution of pericenter distance of those injecting stellar binaries , which is related to the speed of the migration or diffusion process of those binaries to the immediate vicinity of the central mbh : a change in this distribution may result in either a significant increase or decrease in both the number of hvss and that of captured stars , but their number ratio is not affected much .
the complete survey conducted by @xcite has detected @xmath293 unbound hvss with mass @xmath169-@xmath170 in a sky area of @xmath294 deg@xmath295 , and another massive hvs in the southern hemisphere with mass @xmath296 in the survey are not included in the number . ] and the total number of similar hvss in the whole sky should be @xmath297 .. and if assuming that the detected hvss were originated from two disk - like stellar structures , i.e. , the cws disk plane and the narm plane , as suggested by @xcite , the expected total number of hvss with mass @xmath169-@xmath170 is @xmath298
. the number ratios of the detected hvss to the gc s - stars become @xmath299-@xmath300 . ]
observations have revealed @xmath301 gc s - stars within a distance of @xmath302 from the central mbh .
the number ratio of the detected hvss to the gc s - stars is @xmath303-@xmath304 .
the number ratio resulted from any of the top four injection models listed in table [ tab : t2 ] is inconsistent with the observational ones .
both the unbd - ms0 model and the disk - ms0 model give a number ratio substantially larger than that inferred from observations , while the disk - th0 model and the disk - th2 model give too small number ratios .
@xcite suggest that the imf of young star clusters in the gc , i.e. , the arches cluster , may be top - heavy and the imf slope is @xmath305 , although some later studies argued that the imf of arches cluster may be still consistent with the salpeter one by assuming continuous star formation ( e.g. , * ? ? ?
considering of this , we adopt an imf with a slope of @xmath306 and perform two additional injecting models , i.e. , `` disk - im0 '' and `` disk - im2 '' , as listed in tables [ tab : t3 ] and [ tab : t4 ] .
our calculations show that the number ratios produced by the two models are close to the observational ones . obviously , the number ratio of the simulated hvss to gc s - stars is significantly affected by the adopted imf .
adopting a steeper imf may lead to a larger number ratio .
the velocity distribution of the detected hvss suggests a slow migration / diffusion of stellar binaries into the immediate vicinity of the central mbh ( see detailed discussions in * ? ? ?
the disk - im2 model can produce a velocity distribution similar to the observational ones , and the large @xmath39 adopted in this model also suggests a slow migration / diffusion of binaries into the vicinity of the central mbh . however , all the other models appear to generate too many hvss at the high - velocity end and thus a too flat velocity distribution compared with the observational ones
. this inconsistency could be due to many factors .
for example , ( 1 ) the velocity distribution of the detected hvss could be biased due to either the small number statistics and/or the uncertainties in estimating the observational selection effects ; ( 2 ) the uncertainties in the initial settings of the injecting binaries could also lead to some change in the velocity distribution , e.g. , the velocity distribution may be steeper if the semimajor axis distribution of stellar binaries is log - normal , rather than follow the pik law @xcite ; and ( 3 ) the injection of binaries to the vicinity of the central mbh may be quite different from the simple models adopted in this paper because of the complicated environment of the very central region , e.g. , the possible existence of a number of stellar - mass bhs or an intermediate - mass black holes ( bhs ) within @xmath307 . for each model , the injection rate of stellar binaries can be calibrated to produce the numbers of the observed hvss and gc s - stars . under the initial settings for each model described in section [ sec : num_simu ] ,
the numbers of the simulated hvss and the gc s - stars , similar to the detected ones , are listed in table [ tab : t4 ] .
the calibrated injection rates are also listed in table [ tab : t4 ] for each model and they are roughly on the order of @xmath308 to @xmath309 .
however , one should be cautious about that this injection rate depends on the initial settings on the distributions of the semimajor axes and the periapses of the injecting binaries . according to our simulations ,
many of the injected binaries are not tidally broken up , and the breakup rate is a factor of @xmath172-@xmath241 times smaller than the injection rate for those models studied in this paper , i.e. , about a few times @xmath310 to @xmath309 , which is consistent with the estimates by @xcite .
-@xmath144 and @xmath311 for the disk - im2 model .
the distribution is expressed by a hammer - aitoff projection in the galactic coordinates . in the top panel ,
the positions of the hvss are projected to infinity from the gc .
the solid red and blue curves represent the cws and narm disk planes , respectively , which are also projected to infinity .
the region above the green curve shows the area surveyed by @xcite in the northern hemisphere .
the open red and blue circles represent the unbound hvss originated from binary stars on the cws and narm disk planes , respectively . for illustration purpose ,
all the @xmath312 simulated hvss are shown in the top panel ( see table [ tab : t4 ] ) .
the injection rates are assumed to be the same for the injections from the cws and the narm planes and thus the number of hvss associated with the two planes are also the same . in the bottom panel ,
the positions of the hvss are not projected to infinity .
the solid red curves and the solid green curve are the same as those for the top panel .
the red crosses and blue plus symbols represent those unbound hvss with @xmath313 injected from the cws and narm disk planes , respectively .
the red open diamonds and the blue open squares represent the unbound hvss with proper motion @xmath314 injected from the cws and the narm disk planes , respectively . ]
the hvss discovered in the galactic halo are typically in the mass range @xmath291-@xmath170 .
hvss with other masses should also be populated in the galactic bulge and halo . in this section ,
we investigate the properties of the simulated hvss with mass @xmath143-@xmath144 , which are most likely to be the companions of the `` gc s - stars '' .
panel ( a ) of figure [ fig : f5 ] shows the cumulative distribution of the galactocentric distances of the high - mass ( @xmath143-@xmath144 ) hvss .
the predicted numbers of these hvss from different models are listed in table [ tab : t4 ] .
the total number of the simulated unbound hvss , as companions of the population of the gc s - stars , is @xmath315-@xmath316 according to the disk - im2 model , which is able to reproduce the numbers of the observed hvss and gc s - stars .
the majority of these unbound hvss are at distances of a few to a few tens @xmath317 from the gc , which are much closer to the gc than the detected @xmath172-@xmath170 hvss mainly because a high - mass star has a shorter main - sequence lifetime and thus the distance it can travel within the lifetime is small .
the close distances of these hvss from the gc suggest that the velocity vectors of many of them are not along our line of sight .
their three - dimensional velocities range from @xmath318 to @xmath319 in the galactocentric rest frame ( see panel ( b ) in figure [ fig : f5 ] ) ; and their heliocentric radial velocities in the galactic rest frame range from @xmath320 to @xmath11 ( see panel ( c ) in figure [ fig : f5 ] ) , where the negative and positive velocities represent moving toward and away from the sun , respectively . compared with the ejection velocities of the @xmath172-@xmath170 hvss , those of the @xmath279-@xmath144 hvss are relatively higher because of their higher mass ( see equation ( [ eq : vhvs ] ) ) .
most of the hvss have proper motions in the heliocentric rest frame as large as @xmath321 to a few tens @xmath321 ( see panel ( d ) in figure [ fig : f5 ] ) .
these hvss are bright enough to be detected at a distance less than a few tens @xmath317 by future telescopes , such as , the _ global astrometric interferometer for astrophysics _ spacecraft ( _ gaia _ ) , and their proper motions are also large enough to be measured .
figure [ fig : f6 ] shows the sky distribution of the simulated hvss with mass@xmath143-@xmath144 , which may represent the ejected companions of the `` gc s - stars '' , in the galactic coordinates by a hammer aitoff projection . in the top panel of figure [ fig : f6 ] , the positions of the hvss are projected to infinity from the gc , which are consistent with being located close to the cws disk plane and the narm plane ( also projected to infinity ) as expected . in the bottom panel of figure [ fig : f6 ] ,
the positions of the hvss are not projected to infinity .
as seen from the bottom panel , the simulated hvss with mass @xmath143-@xmath144 lie in the area below the projected curves of the cws disk plane and the narm plane , and most of these high - mass hvss reside out of the area surveyed by @xcite .
our calculations show that less than @xmath322 of the simulated unbound hvss with mass @xmath323 are located in the survey area
. this may be the reason that none of those high - mass hvss , possibly the companions of the `` gc s - stars '' , has been discovered in the survey area .
if the hvss are initially originated from the cws disk and the narm plane , the hvss with high radial velocities are also relatively rare in the direction close to the disk normals , i.e. , @xmath324 and ( @xmath325 ) while the hvss with high proper motions ( @xmath326 ) is relatively numerous in that direction because the velocity vectors of hvss are close to be perpendicular to the disk normals .
surveys of hvss in the southern sky with skymapper and others may find such massive hvss , as the possible companions of the `` gc s - stars '' , and provide crucial evidence for whether those gc s - stars are produced by the tidal breakup of stellar binaries . note that the spatial distribution of the hvss discussed in this section is directly related to the assumption that the injecting stellar binaries are originated from two disk - like stellar structures similar to the cws disk in the gc .
however , the other properties of the hvss or the gc s - stars discussed in this paper are affected by whether the injecting stellar binaries are bound to the central mbh or not , but not affected by whether they are initially on the cws disk plane or not .
the unbd - ms0 , disk - ms0 , disk - th0 , disk - th2 , disk - im0 , and disk - im2 models roughly produce @xmath327 , @xmath328 , @xmath329 , @xmath330 , @xmath331 , and @xmath332 captured stars surviving to the present time and with mass in the range of @xmath172-@xmath3 , less massive than that of the gc s - stars , within a distance of @xmath302 from the mbh ( see table [ tab : t4 ] and section [ sec : hvss ] ) .
the above numbers are obtained by calibrating the injection rate of the stellar binaries over the past @xmath238 myr to generate @xmath301 simulated gc s - stars similar to the observational number .
the captured stars with mass in the range of @xmath333-@xmath3 could be detected by the next generation telescopes , e.g. , the thirty meter telescope ( tmt ) or the european extremely large telescope ( e - elt ) .
these low - mass stars are potentially important probes for testing the gr effects near an mbh , if they are closer to the central mbh than s2 . in this section
, we estimate the probability distribution of the innermost captured low - mass stars ( @xmath169-@xmath3 ) by monte carlo realizations based on the calibrated injection rate .
the left panel of figure [ fig : f7 ] shows the probability distributions of the semimajor axis of the innermost captured star with mass @xmath169-@xmath3 resulted from different injection models .
for those models adopting @xmath248 , the resulted innermost captured star is typically on an orbit with semimajor axis @xmath334 , and the probability that its semimajor axis is less than that of s2 is @xmath335 . for the other models adopting @xmath249 ,
the resulted innermost captured star is on an orbit with semimajor axis of @xmath336-@xmath337 and the probability that its semimajor axis is smaller than that of s2 is @xmath199-@xmath174 .
the probability to capture a star within the orbit of s2 is larger for the @xmath248 models than for the @xmath249 models .
the reason is that relatively more stellar binaries can be injected into the immediate vicinity of the central mbh and thus more stars can be captured onto orbits with smaller semimajor axes ( see table [ tab : t4 ] ) in the models adopting @xmath248 than that adopting @xmath249 .
we conclude that the probability of a less massive star ( @xmath172-@xmath3 ) existing within the s2 orbit is at least @xmath338 and can be up to @xmath339 , which may be revealed by future observations and then offer important tests to general relativity .
the right panel of figure [ fig : f7 ] shows the probability distribution of the pericenter distance of the innermost captured low - mass star ( @xmath291-@xmath3 ) resulted from different injection models . for the injection models adopting @xmath248 , the pericenter distance distribution is concentrated within @xmath340 ; while for the other models adopting @xmath249 , the expected pericenter distance is broadly distributed over 10200au .
nevertheless , the probability that the pericenter distance of the innermost captured star with mass @xmath169-@xmath3 is less than that of s2 ( and s14 ) is still significant , i.e. , @xmath341 ( or @xmath342 ) .
the innermost captured star may have its semimajor axis and pericenter distance both significantly smaller than those of s2 , therefore , the gr effects on its orbit may be much more significant than that on s2 .
if taking into account the captured stars with even lower masses , e.g. , @xmath343 , the number of the expected captured stars surviving to the present time becomes much larger , especially for those models with large @xmath40 .
for example , the numbers of the captured stars with mass 17@xmath344 are @xmath345 , @xmath346 , @xmath347 , @xmath348 , @xmath349 , and @xmath350 for the six models , respectively . for those lower mass captured stars ,
the semimajor axis and the pericenter of the innermost one could be even closer to the central mbh . some stars may be transported to the vicinity of the central mbh by some mechanisms other than the tidal breakup of stellar binaries .
it is possible that some of these stars , with their origins different from the captured stars discussed above , exist within the s2 orbit , but which is beyond the scope of the study in this paper .
in this paper , we investigate the link between the gc s - stars and the hvss discovered in the galactic halo under the hypothesis that they are both the products of the tidal breakup processes of stellar binaries in the vicinity of the central mbh .
we perform a large number of the three - body experiments and the monte carlo simulations to realize the tidal breakup processes of stellar binaries by assuming a continuous binary injection rate over the past @xmath238 myr , and adopting several sets of initial settings on the injection of binaries . after the tidal breakup of a binary ,
we follow the dynamical evolution of the captured components in the gc by using the arma model ( see @xcite ) , which takes into account both the rr and nr processes , and we also trace the kinematic motion of the ejected component in the galactic gravitational potential . the properties of the ejected and captured components of the tidally broken - up binaries are naturally linked to each other as they are both the products of tidal breakup of binaries . for those hvss discovered in the galactic halo with mass @xmath169-@xmath170 and @xmath101-@xmath102 ,
their companions are expected to be captured onto orbits with semimajor axis in the range @xmath351-@xmath184 ; for the observed gc s - stars with semimajor axis @xmath351 - 4000@xmath352 in the gc , their companions are expected to be ejected out to the galactic bulge and halo with @xmath353-@xmath319 .
the energy of the captured stars evolves with time because of their dynamical interactions with the environment . for the captured stars with mass @xmath143-@xmath144 , the differences between their present - day energy and their initial ones are no more than @xmath354
; for the captured stars with mass @xmath291-@xmath170 , however , the difference can be by order of unity .
therefore , the current semimajor axis distribution of the gc s - stars may provide a good estimation on the velocity distribution of their ejected companions ( e.g. , equation ( [ eq : acap0 ] ) ) , but that of the captured stars with mass @xmath169-@xmath170 does not .
the eccentricities of the `` gc s - stars '' ( @xmath143-@xmath144 ) are close to @xmath177 right after the capture and may evolve to low values , and the eccentricity distribution of these simulated gc s - stars at the present time could be statistically compatible with the observational ones of the gc s - stars attributed to the rr processes . for those captured stars with mass @xmath169-@xmath355 , their eccentricities can evolve to even lower values at the present time compared with the high - mass gc s - stars ( @xmath143-@xmath144 ) because they interact with the environment for a longer time . to reproduce both the numbers of the detected hvss and
gc s - stars , the injection rate of binaries need to be on the order of @xmath310 to @xmath356 and the imf of the primary components is required to be somewhat top - heavy with a slope of @xmath357 . for the injection models that can reproduce the observational results on both the gc s - stars and the hvss , including the distributions of the semimajor axes and eccentricities of the gc s - stars , the spatial and velocity distributions of the detected hvss , and the number ratio of the hvss to the gc s - stars , the expected number of the @xmath169-@xmath3 captured companions is @xmath358 within a distance of @xmath302 from the central mbh .
future observations on the low - mass captured stars may provide a crucial check on whether the gc s - stars are originated from the tidal breakup of stellar binaries .
the companions of the hvss , which are captured by the central mbh , are usually less massive than that of the gc s - stars ( @xmath143-@xmath144 ) .
the semimajor axis of the innermost captured star with mass @xmath291-@xmath3 is @xmath336-@xmath6 , and the probability that it is smaller than that of s2 is @xmath359-@xmath360 for the @xmath249 models and @xmath335 for the @xmath248 models .
the pericenter distance of the innermost captured star with mass @xmath291-@xmath3 is @xmath361-@xmath8 and the probability that it is smaller than that of s2 ( or s14 ) is also significant , i.e. , @xmath341 ( or @xmath342 ) .
the existence of such a star will provide a probe for testing the gr effects in the vicinity of an mbh .
future observations by the next generation telescopes , such as , tmt or e - elt , will be able to investigate the existence of such a star , and provide important constraints on the nature of the central mbh if such a star is detected .
the number of the ejected unbound companions of the `` gc s - stars '' ( see the definition of the `` gc s - stars '' at the end of section [ sec : overview ] ) is roughly @xmath315-@xmath289 and the majority of these ejected stars are located within a distance of @xmath315 kpc from the gc .
the number of these ejected companions is substantially larger than the number of observed gc s - stars mainly because the observed gc s - stars are only a fraction of the `` gc s - stars '' and the rest of the `` gc s - stars '' were tidally disrupted and do not survive today ( see table [ tab : t4 ] ) .
their heliocentric radial velocities in the galactic rest frame range from @xmath362 to @xmath363 and their proper motions in the heliocentric rest frame can be as large as @xmath364 .
these high - mass ejected stars are bright enough to be detected at a distance less than a few ten kpc and their proper motions are also large enough to be measured by future telescopes , such as _ gaia_. the majority of the ejected companions of the gc s - stars lie outside the area surveyed by @xcite for our observers located at the sun .
we thank the referee and the scientific editor , eric feigelson , for helpful comments and suggestions .
we thank warren brown for useful comments on the paper and are grateful to ann - marie madigan for helpful communications on the arma model for the dynamical evolution of the captured stars .
this work was supported in part by the national natural science foundation of china under nos .
10973001 and 10973017 , and the bairen program from the national astronomical observatories , chinese academy of sciences .
* _ note added in proof .
_ * after the submission of this paper , the following two new observational results have been reported , which are relevant to this work .
( 1 ) meyer et al .
( 2012 , sci .
, 338 , 84 ) discovered a faint star , s0 - 102 , which is orbiting the mbh in the gc with shortest - known - period ( 11.5 yr ) .
the existence of such a star is consistent with our predictions shown in figure 7 .
( 2 ) lu et al .
( 2013 , apj , 764 , 155 ) estimate the initial mass function for stellar populations in the central 0.5 pc of the galaxy and find it is top - heavy with a slope of @xmath365 , which is consistent with the requirement by our model to re - produce the number ratio of hvss to gc s - stars ( see table 4 and section 5.1 ) . | in this paper , we investigate the link between the hypervelocity stars ( hvss ) discovered in the galactic halo and the galactic center ( gc ) s - stars , under the hypothesis that they are both the products of the tidal breakup of the same population of stellar binaries by the central massive black hole ( mbh ) . by adopting several hypothetical models for binaries to be injected into the vicinity of the mbh and doing numerical simulations ,
we realize the tidal breakup processes of the binaries and their follow - up dynamical evolution .
we find that many statistical properties of the detected hvss and gc s - stars could be reproduced under some binary injecting models , and their number ratio can be reproduced if the stellar initial mass function is top - heavy ( e.g. , with slope @xmath0 ) .
the total number of the captured companions is @xmath1 that have masses in the range @xmath2@xmath3 and semimajor axes @xmath4 and survive to the present within their main - sequence lifetime .
the innermost one is expected to have a semimajor axis @xmath5@xmath6 and a pericenter distance @xmath7@xmath8 , with a significant probability of being closer to the mbh than s2 .
future detection of such a closer star would offer an important test to general relativity .
the majority of the surviving ejected companions of the gc s - stars are expected to be located at galactocentric distances @xmath9 , and have heliocentric radial velocities @xmath10@xmath11 and proper motions up to @xmath12@xmath13 .
future detection of these hvss may provide evidence for the tidal breakup formation mechanism of the gc s - stars . |
the links between gravitation and thermodynamics are fundamental and exemplified in the four laws of black hole thermodynamics , first demonstrated by bardeen , carter and hawking @xcite . however , these authors initially considered this analogy with thermodynamics as being purely formal .
a fundamental idea was then proposed by jacob bekenstein @xcite in 1972 , when he claimed that the formal identity between the area of a black hole and entropy was really a physical identity .
bekenstein was strongly criticized by carter and hawking at the 1972 les houches summer school , where he had first presented his conjecture but , ironically , his expression for the entropy and the thermal emission of black holes was two years later demonstrated using a semiclassical calculation by stephen hawking himself @xcite .
more recently , strominger and vafa @xcite , by counting the quantum microstates of a macroscopic black hole , have verified the bekenstein - hawking entropy formula , and several derivations @xcite further confirmed its universality , which appears independent of the particular physical string theory used to derive it .
let us recall the main relations for this entropy and radiation mechanism .
for a ( non rotating and uncharged ) black hole of mass m , the internal energy is @xmath0 , and its entropy @xmath1 when m is expressed in units of planck mass . from this
, we can derive the expression for the black hole temperature : @xmath2 the evaporation time , for a black hole of mass m is @xmath3 , with m again expressed in planck mass units .
note that the bekenstein - hawking entropy expression strongly suggests that spacetime is discrete at the planck length scale : @xmath4
the motivation of this study was the argument , discussed initially with john bell @xcite and m.m .
nieto @xcite , that morrison s
antigravity @xcite antiparticles falling up" would a priori entail a vacuum instability .
but this is not necessarily a problem in itself . in fact , as just noted , in the presence of a black hole , the vacuum _ is _ unstable .
the question then becomes : what amount of antigravity @xmath5 between particles and antiparticles will mimic hawking s radiation , which should certainly be considered as an acceptable vacuum instability ? with surprise , it was found @xcite that morrison s antigravity , @xmath6 , just leads to the expression of hawking s temperature . but is antigravity compatible with the second law ?
in particular , would not antigravity allow a maxwell s demon to decrease the entropy of an isolated system by hiding photons in the vacuum ?
however , operating a maxwell s demon on curved spacetime has an information cost .
if a demon wants to follow continuously particles or photons on a curved manifold , he must pay the price of the dynamical kolmogorov - sinai entropy , which can be estimated using pesin s formula @xcite .
the characteristic distance for the divergence of the photons is obtained from the components of the riemann tensor , homogeneous to the inverse of the square of a curvature radius .
this leads to a characteristic distance @xmath7 in reference to the lyapunov exponent : @xmath8 when the demon is operating at a distance r from a massive body of mass m. in order to have a reasonable chance to suppress a photon in the box , of size at least @xmath9 where he is operating , our demon will have to pay a price , in bits : @xmath10 where @xmath11 is the fine structure constant and @xmath12 is the compton radius of the electron .
the first term is the minimal size of the enclosure where the demon is working , the second term is the number of collisions needed to absorb with a reasonable chance a photon in the fundamental state of the box , and the third term is the lyapounov length , after which the demon must pay a price of one bit .
obviously , the demon must realize his experiment at a distance of the center of attraction larger than the schwarzschild radius @xmath13 , and larger than the compton wavelength of the electron @xmath12 .
consequently , the demon is unable to operate efficiently enough to violate the second law of thermodynamics @xcite .
the motivation for reconsidering morrison s antigravity was the coincidence , noted by overhauser and chardin @xcite , and known by bell , that antigravity would explain cp violation in the neutral kaon system @xcite . but how could we justify antigravity ? to answer this question , let us study the kerr - newman solution , discovered by roy kerr @xcite and ezra newman @xcite , and incorporating three parameters : mass @xmath14 , specific angular momentum @xmath15 , and charge @xmath16 .
we will summarize briefly the properties of this solution , assuming a fast kerr solution , i.e. @xmath17 , a condition met by all elementary particles , with the notable exception of the neutral higgs boson @xcite .
the fast kerr geometry is particularly simple since it involves no horizon .
the angular momentum imposes an annular shape to the singularity , which appears naked but nevertheless almost invisible since the measure of initial conditions allowing to reach the ring singularity is zero .
brandon carter has studied the topology of this solution in the late sixties @xcite , noting the striking analogy that a kerr - newman electron " bears with real electrons . in particular , the gyromagnetic factor of the kerr - newman electron is @xmath18 and the geometric extension of the ring is of the order of the compton wavelength of the electron , giving it a spatial extension compatible with its cross - section . another interesting feature concerns the charge conjugation ( c ) properties of this solution . by crossing the interior of the ring ,
an observer will measure the charge and mass of the electron with a reversed sign . for a particle physicist
, this means that if we were initially looking at an electron , we can now see a positron by crossing the interior of the ring .
most surprisingly , this positron " has a repulsive gravity .
this results from the symmetry properties of the metric and electromagnetic field tensor form of the kerr - newman solution , which can expressed @xcite in boyer - lindquist coordinates : @xmath19 and : @xmath20 and where : @xmath21 and : @xmath22 another significant and surprising feature of the kerr - newman solution is the fact that it is possible to go backward in time by exploring the negative mass part @xmath23 of the solution .
this feature was also studied by brandon carter @xcite , and is known as the carter time machine " @xcite .
considered initially as an indication of inconsistency , this can now be considered as an incentive to consider the possibility that antimatter , i.e. matter going backwards in time , can be defined in general relativity . , a point b exists such that a signal emitted by a and reemitted by b comes back to a at zero time delay _ and _ with a direction identical to that of the initial signal .
the interaction between a and b is then diverging , and b appears to be at the position of a ]
an important generic property of gravitation is the existence of points of infinite magnification for the image of an object through the lensing " created by a massive object .
used in recent years to detect massive compact halo objects ( machos ) in our galactic neighborhood @xcite , this magnification , when it is infinite , has the consequence that the lensed object may appear infinitely more luminous and closer than its true position .
this property is even stronger for the fast kerr - newman geometry , where closed timelike curves ( ctcs ) exist between any two points . for a given point
a in the neighborhood of the ring , there exists a set of points b such that the radar interaction between a and b photons are emitted by a , scattered by b and received back by a is instantaneous .
the signal emitted comes back with zero time delay as seen by the emitter , and the b points will give an observer in a the impression that they are sharing the same position .
these points can be explicitly constructed in the kerr geometry in @xmath24 dimensions , where the spinning cosmic string is an exactly soluble model @xcite .
it is straightforward to demonstrate that the set of such points b lie on a portion of ellipse ( fig .
2 ) since the time pitch associated with a @xmath25 rotation around the spinning cosmic string can be written as @xmath26 , where _ a _ is the specific angular momentum per unit length of the string . from the existence of conjugate points in @xmath24 gravity , expected to be valid also in @xmath27 gravity from the existence of ctcs , there follows a definition of antiparticles in general relativity as the time - reversed _ image _ of particles observed through a kerr ring .
these kerr rings are probably present in all elementary particles , if they are string loops , and in the past singularity of the big bang .
this ( non - local ) definition of antimatter has the consequence that there exists a gravitational repulsion between matter and antimatter , defined relatively to each other and not in an absolute way .
the coupling of systems with opposite arrows of time is reminiscent of the dynamical systems studied by schulman @xcite . from the persistence of individual arrows of time for such weakly coupled systems , interactions of each system with the conjugate system
are expected to appear as noise .
during the sixties and early seventies , several attempts have been made , using notably a conjectured repulsion in strong interactions @xcite , which failed to justify the survival of significant matter and antimatter domains in a symmetric matter - antimatter universe .
the cosmological consistency of the previous definition of antimatter in general relativity has yet to be demonstrated . on the other hand
, it should be noted that the usual arguments invoked to exclude the existence of large domains of antimatter through the non observation of diffuse gamma - ray background @xcite are not applicable in the case where diffusion and annihilation at the border of matter and antimatter domains is prevented by gravitational repulsion . also , it is fascinating to note that the gravitational repulsion between matter and antimatter appears to lead to a natural explanation of the value of the cosmological constant observed in recent supernovae and cmb observations @xcite : @xmath28 where a nearly flat universe is composed of only @xmath29 of ordinary matter , with a dark matter density @xmath30 and an apparent cosmological density @xmath31 . to justify this statement ,
let us consider the expression for the deceleration parameter _
q _ as a function of the scale factor _ a _ and its derivatives : @xmath32 when observed on a scale larger than the matter or antimatter domains , this symmetric universe will appear flat with a parameter @xmath33 due to the repulsion of adjacent domains .
fundamentally , there exists no cosmological constant , but if we insist to parametrize the repulsive term by a cosmological constant , this then implies : @xmath34 although , at any given epoch , this equality will locally be verified , the evolution of @xmath35 with time has for consequence that the derived value of the effective cosmological constant " will vary according to @xmath36 since the matter density varies approximately in this way after recombination @xcite . therefore , there is no coincidence , only an ( incorrect ) assumption about the existence of a cosmological constant .
ripalda , in a different theoretical context , has also noted that repulsive gravity would lead to a cosmological constant of the correct order @xcite .
we have summarized the relations between gravitation and thermodynamics , which appear extremely strong and universal .
the bekenstein - hawking entropy relation , valid for a large class of string theories , can be considered as an indication that spacetime is discrete at the planck length scale and that gravitation is the master arrow of time asymmetry .
in addition , charge ( c ) and time - reversal ( t ) properties of the kerr - newman solutions suggest a natural definition of antimatter in general relativity .
this definition provides a parameter free explanation of the otherwise extraordinary coincidence of the cosmological constant energy density with the matter density observed in the supernovae sn1a and cmb data .
defined as the time - reversed image of matter through kerr wormholes , antimatter then provides with matter an explicit example of coupled systems with opposite arrows of time , as studied by schulman .
discussions with j - m .
rax , b. carter , d. connetable , g. esposito - farse , e. fischbach , t. goldman , m.m .
nieto , r. pain and a. tzella are gratefully acknowledged .
needless to say , these people are not responsible for the errors and omissions contained in this paper
. 0 j.m .
bardeen , b. carter , and s.w . hawking , _
phys . _ * 31 * , 161 ( 1973 ) .
j. bekenstein , _ phys .
rev . _ * d7 * , 2333 ( 1973 ) ; _ physics today _ * 33 * , 24 ( 1980 ) . s. w. hawking , _ nature _ ( london ) * 248 * , 30 ( 1974 ) ; _ comm . math
. phys . _ * 43 * , 199 ( 1975 ) .
a. strominger , and c. vafa , _ phys .
_ b*379 * , 99 ( 1996 ) . s. carlip , _ phys .
* 88 * , 241301 ( 2002 ) and references therein .
bell , in _ fundamental symmetries _ ,
p. bloch , p. pavlopoulos and r. klapisch , plenum press , 1987 .
m. m. nieto and t. goldman , _ phys .
rep . _ * 205 * , 221 ( 1991 ) and references therein .
p. morrison , _ am .
j. phys . _
* 26 * , 358 ( 1958 ) .
g. chardin , and j.m .
rax , _ phys .
b _ * 282 * , 256 ( 1992 ) .
g. chardin , _ nucl .
phys . a _ * 558 * , 477c ( 1993 ) ; _ hyperfine interactions _ * 109 * , 83 ( 1997 ) .
pesin , _ russ
. math . surveys _ * 32 * , 55 ( 1977 ) .
overhauser , physics department , purdue university , march 2000 , private communication .
j. h. christenson , j. w. cronin , v. l. fitch , and r. turlay , _ phys .
* 13 * , 138 ( 1964 ) .
r. p. kerr , _ phys .
lett . _ * 11 * , 238 ( 1963 ) .
e. t. newman , et al . ,
_ j. math .
phys . _ * 6 * , 918 ( 1965 ) . m. visser , _ lorentzian wormholes , from einstein to hawking _ , ( springer , new york , 1996 ) , chapter 12
. b. carter , _ phys .
_ * 141 * , 1242 ( 1966 ) ; phys . rev . *
174 * , 1559 ( 1968 ) .
b. oneill , _ the geometry of kerr black holes _ , ( a. k. peters , wellesley , 1995 ) .
c. alcock , et al
. _ nature _ * 365 * , 621 ( 1993 ) ; e. aubourg , et al .
_ nature _ * 365 * , 623 ( 1993 ) .
s. deser , r. jackiw , and g. t hooft , _ ann .
phys . _ * 152 * , 220 ( 1984 ) .
schulman , _ phys .
lett . _ * 83 * , 5419 ( 1999 ) .
r. omnes , _ phys
* 23 * , 38 ( 1969 ) ; _ phys .
rep . _ * 3 * , 1 ( 1970 ) .
cohen , a. de rujula , and s.l .
glashow , _ astroph .
j. _ * 495 * , 539 ( 1998 ) .
s. perlmutter , et al .
, _ nature _ * 391 * , 51 ( 1998 ) .
a. g. riess , et al .
, _ astroph .
j. _ * 116 * , 1009 ( 1998 ) .
a. benoit - lvy , and g. chardin , in preparation .
ripalda , gr - qc/9906012 . | the intimate links between gravitation and the second law are summarized and two less known relations between gravity and thermodynamics are studied .
firstly , the information cost required to operate a maxwell s demon on a curved spacetime is estimated using the kolmogorov - sinai entropy .
more importantly , the charge and time ( c and t ) reversal properties of the kerr - newman solution in general relativity show that this solution , similarly to the dirac equation , appears to represent both a particle and its antiparticle and suggests a definition of antimatter in general relativity .
this definition leads to a parameter free explanation of the cosmological constant term observed in the supernovae data .
the relation of this definition of antimatter with the coupled systems through opposite time arrows studied by schulman is also emphasized . |
solitons are a central paradigm of nonlinear physics @xcite .
they appear in systems as diverse as the ocean @xcite , nonlinear fiber optics @xcite , and bose - einstein condensates ( bec s ) @xcite .
such localized waves propagate without spreading .
they retain their form after interactions and are robustly stable with respect to perturbation .
in this sense they are particle - like . in fiber optics
, solitons have become the basis of transoceanic communication systems @xcite .
they are called _ bright _ when they represent a local maxima in the field , and dark or gray or kink when they represent a minima .
_ matter - wave _ bright solitons arise from a balance between nonlinearity and dispersion in the gross - pitaevskii equation ( gpe ) which describes the mean field of the bec .
such a macrosopic bound state of atoms is intrinsically quantum mechanical , in that its existence depends on quantum , rather than classical motion of the atoms .
matter - wave bright solitons therefore represent an intriguing combination of nonlinear and quantum physics .
investigations of bec s with attractive interactions have focused on their collapse @xcite . in the attractive gpe ,
which for the case of constant potential is also known as the focusing nonlinear schrdinger equation , it is well known that solutions collapse in finite time @xcite .
this collapse can be prevented , below a certain critical nonlinearity , by the addition of a spatially varying external potential .
intensive studies of collapse dynamics @xcite were initially inspired by experiments on the attractive bec @xcite and by the possibility of efficient implementation of feshbach resonances @xcite .
the latter use an external magnetic field to tune the @xmath0-wave scattering length which controls the strength of the nonlinearity governing the collapse : in particular , @xmath1rb has been the subject of recent experiments @xcite . in @xmath2li
there exists a feshbach resonance around 725 gauss for the state @xmath3 which is amenable to the study of _ stable _ phenomena in attractive bec s : the scattering length for @xmath4 is positive , and becomes slightly negative between 150 and 525 gauss , far from the resonance and thus far from the region in which three - body losses are significant .
@xmath2li was the first species used to create an attractive bec @xcite , and it is @xmath5li that has been used to create the first matter - wave bright solitons @xcite . the essential experimental method used in ref .
@xcite to transform a stable attractive bec into a matter - wave bright soliton was as follows .
a bec with positive scattering length was condensed into a harmonic trap .
the trap was adiabatically @xcite deformed into a cigar - shaped geometry , _ i.e. _ , @xmath6 , where @xmath7 and @xmath8 are the longitudinal and axial oscillation frequencies of the atoms in the trapping potential , respectively .
the scattering length was then tuned via a feshbach resonance to be small and negative .
the resulting condensate , a metastable macroscopic quantum bound state of about 5000 particles , was projected onto an _ expulsive _ harmonic potential in the longitudinal direction .
it was then observed to propagate , without spreading , over a distance of 1.1 mm , in contrast to an ideal gas , which is exploded exponentially by the potential .
thus it was the use of the expulsive potential which provided clear evidence of a self - trapped , soliton - like state . in this paper the stability and dynamics of a matter - wave bright soliton in an expulsive potential
are studied . the 3d gpe which describes the mean field of the bec
is written as @xmath9\right \ }
\psi(\vec{r},t ) = i\hbar\frac{\partial}{\partial t}\psi(\vec{r},t ) \ , , \end{aligned}\ ] ] where @xmath10 , @xmath11 is the @xmath0-wave scattering length tuned by the feshbach resonance , @xmath12 is the atomic mass , @xmath13 is the number of atoms in the condensate , and @xmath14 has been normalized to one .
equation ( [ eqn : gpe3d ] ) , in conjunction with variational studies , numerical integration , reduction to an effective one - dimensional problem , and the semi - classical wkb approximation , forms the basis of the work herein .
section [ sec : stability ] presents the size , energy , and critical parameters for a stable soliton .
section [ sec : nonlinear ] studies the dynamical effects particular to an expulsive potential : the equations of motion for the soliton parameters are derived ; weak excitations are shown to be exponentially damped ; higher order solitons created by sudden changes in the scattering length are shown to persist in the presence of the potential ; and an intriguing form of macroscopic nonlinear tunneling , which we call _ quantum evaporation _ , is described .
the latter eventually leads to explosion of the soliton .
a form of nonlinear tunneling similar to this has already been experimentally observed in the context of the relaxation of condensate spin domains @xcite . finally , in sec .
[ sec : conclusions ] the conclusions and experimental outlook are presented .
the first and most important question concerning the attractive bec is under what conditions it remains robustly stable .
the gpe , in general , has proven a sufficient theoretical tool for describing the vast majority of bec phenomena for @xmath15 . however , as a three - dimensional second - order nonlinear partial differential equation it is not amenable to analytic methods without approximation , of which the most commonly used for the repulsive , or positive scattering length case , is the thomas - fermi approximation @xcite , which neglects the kinetic energy term in eq .
( [ eqn : gpe3d ] ) . in the attractive case
one must find another method , since the kinetic energy and external potential terms must dominate the mean field in order to prevent collapse .
the variational method is the most immediate and useful approximation @xcite . a thorough variational study of the lagrangian dynamics of a purely gaussian ansatz has already been made by prez - garca _ et al .
however , this work did not focus on the case of bright soliton solutions in an expulsive potential . to this end
a hyperbolic secant rather than gaussian ansatz shall be utilized in the longitudinal direction . in limits germane to experiments this choice leads to closed form analytic expressions for the widths and energy of the condensate .
three separate cases are therefore considered : general three - dimensional axisymmetric confinement , which is the case of ref .
@xcite ; @xmath16 , which allows for full analytic solutions ; and finally the quasi-1d limit , which is the case of ref .
@xcite and the one likely to be relevant for future experiments on matter - wave bright solitons .
the gross - pitaevskii energy functional is defined as @xmath17 & = & \int d^3\vec{r}\ , \left\ { \frac{\hbar^2}{2m}|\nabla\psi(\vec{r}\,)|^2 + \frac{g\,n}{2}|\psi(\vec{r}\,)|^4 \right .
\nonumber \\ & + & \left . \frac{1}{2}m\left [ \omega_{\rho}^2 ( x^2+y^2)+\omega_z^2 z^2 \right ] |\psi(\vec{r}\,)|^2 \right\}\ .
\label{eqn : energy}\end{aligned}\ ] ] in the case of an expulsive potential @xmath18 .
a good stationary variational ansatz for a bright soliton solution is @xmath19 where @xmath20 and @xmath21 are the transverse and longitudinal widths which form the variational parameters . substituting eq .
( [ eqn : ansatz3d ] ) into eq .
( [ eqn : energy ] ) and rescaling all variables by the axial oscillator frequency @xmath8 , one finds @xmath22 where all quantities have been scaled to the transverse harmonic oscillator frequency @xmath8 or length @xmath23 : @xmath24 , @xmath25 , @xmath26 , @xmath27 , and @xmath28 . taking the partial derivatives with respect to @xmath29 and @xmath30 and assuming @xmath31 , one may simplify and obtain the equations for the extrema points : @xmath32 and @xmath33 whether these extrema are saddle points , minima , or maxima , is determined by the jacobean @xmath34 , which , using eq .
( [ eqn : root2 ] ) to eliminate @xmath30 , yields @xmath35 if eq .
( [ eqn : jac ] ) is negative for a given solution @xmath36 to eqs .
( [ eqn : root1 ] ) and ( [ eqn : root2 ] ) , @xmath37 is a saddle point ; if it is positive one must further examine @xmath38 or @xmath39 .
if these second partial derivatives are positive at @xmath36 the extremum is a minimum ; if they are negative the extremum is a maximum . as eqs .
( [ eqn : root1 ] ) and ( [ eqn : jac ] ) are high order polynomials , it is necessary to obtain @xmath36 numerically for given parameters @xmath40 and @xmath41 . for attractive interactions ,
such solutions are found to constitute a local minimum of the @xmath42 surface for a finite range of the parameters and constitute bright solitons , provided that the spatial extent of the condensate is smaller than the longitudinal oscillator length . a specific example relevant to the experiment of ref .
@xcite shall be presented in sec .
[ ssec : project ] .
the critical points for collapse caused by the mean field , and , in the case of @xmath18 , for explosion due to the expulsive potential , may be found by simultaneously solving eq .
( [ eqn : root1 ] ) and eq .
( [ eqn : jac ] ) for @xmath43 .
it is expedient to use the resultant to eliminate @xmath44 , which yields a single implicit equation in @xmath40 and @xmath41 .
the numerical solution to this polynomial equation is plotted in fig .
[ fig : crit ] ( solid line ) .
the outer branch signifies collapse ; the second , inner branch , which exists only for an expulsive potential , represents explosion . for @xmath45
the soliton can be made between the vertical axis and the collapse branch ; for @xmath18 it can be created in the region between collapse and explosion .
the cusp represents the doubly critical point beyond which a soliton can not be formed , and can be calculated analytically to be @xmath46 it is therefore clear that there is a limited parameter range under which bright solitons can be studied in harmonic potentials .
this range , as demonstrated in refs .
@xcite and @xcite , is experimentally realizable .
other features of interest not shown in fig [ fig : crit ] are the critical branch for collapse for oblate traps , @xmath47 , which approaches the vertical axis as @xmath48 , and the critical point for the isotropic case @xmath49 , which is found by the above methods to be @xmath50
. the isotropic case has been extensively studied , and is determined from a purely gaussian hypothesis @xcite to be @xmath51 . however , the value found from numerical studies @xcite of eq .
( [ eqn : gpe3d ] ) is @xmath52 . the filled circles in fig .
[ fig : crit ] mark critical points for collapse determined by imaginary time relaxation of eq .
( [ eqn : gpe3d ] ) , and differ by between 12.5% ( isotropic case ) and 17.5% ( zero longitudinal potential , following section ) from the variational predictions .
a more detailed description of our numerical methods is made in sec .
[ sec : nonlinear ] .
note that even lower critical values for collapse have been found in experiments for which the initial state differs from the equilibrium one @xcite : the variational studies presented herein make similiar predictions , as shall be illustrated in sec .
[ ssec : project ] . in the case of @xmath16 , eqs .
( [ eqn : root1 ] ) and ( [ eqn : jac ] ) factor to yield @xmath53 @xmath54 respectively .
@xmath55 , which was already discarded in the initial simplification leading to eqs .
( [ eqn : root1 ] ) and ( [ eqn : root2 ] ) , corresponds to the trivial linear solution for the non - interacting case @xmath56 , since @xmath57 .
the remaining factor in eq .
( [ eqn : fac1 ] ) , a polynomial of order three , may be solved analytically to yield a local minimum of the @xmath58 surface : @xmath59\right\}^{1/2 } \\ \label{eqn : length2 } & & \gamma_{z{\rm o } } = \frac{2}{\sqrt{3\,\zeta } } \cos\left[\frac{1}{3}\cos^{-1}(-\sqrt{3}\,\zeta)\right]\ , , \end{aligned}\ ] ] where @xmath60 is the dimensionality parameter , as explained below .
the shape of the @xmath58 surface is a bowl with a chute cascading down one side , a formation often found in high mountains : the bowl represent a stability basin and the chute , collapse . a second non - trivial root of eq . ( [ eqn : fac1 ] )
yields the saddle point at the start of the chute , while a third root is extraneous .
the critical point for collapse , which is also the @xmath61-intercept in fig .
[ fig : crit ] , is contained implicitly in the argument of the arccosines in eqs .
( [ eqn : length1 ] ) and ( [ eqn : length2 ] ) , @xmath62 this expression differs by only a few percent from the most extreme case for an expulsive potential , eq .
( [ eqn : crit1 ] ) . the value determined numerically by imaginary time relaxation of eq .
( [ eqn : gpe3d ] ) is @xmath63 , as illustrated in fig .
[ fig : crit ] . at the critical point
the soliton has size @xmath64 , from which it is clear that it is three - dimensional .
far from the collapse regime , _
i.e. _ , for @xmath65 , eqs .
( [ eqn : length1 ] ) and ( [ eqn : length2 ] ) may be taylor expanded to yield @xmath66\ , .\end{aligned}\ ] ] for @xmath65 , @xmath67 and the soliton is quasi - one - dimensional .
@xmath68 therefore characterizes the dimensionality of the system .
for @xmath65 the axial harmonic degrees of freedom are not populated and the system is essentially one - dimensional : as shown in sec .
[ ssec : free ] , @xmath69 and @xmath70 .
for strong transverse harmonic confinement it is therefore a reasonable approximation to assume a wavefunction of the form @xmath71 $ ] and integrate the gpe over the transverse degrees of freedom .
the result is called the quasi - one - dimensional gpe , to first order in @xmath68 @xmath72\phi(z , t ) \nonumber \\
= i\hbar\,\frac{\partial}{\partial t}\,\phi(z , t)\ , , \,\,\,\,\,\end{aligned}\ ] ] where @xmath73 is the renormalized quasi-1d coupling constant @xcite . as in eq .
( [ eqn : gpe3d ] ) , the wavefunction has been normalized to one .
the constant factor of @xmath74 is the shift in energy due to the tranverse harmonic confinement , @xmath75 . for @xmath76 , the analytically derived criterion used here for the system to be in the quasi - one - dimensional regime is equivalent to the intuitive criterion that the additional energy due to atomic interactions is much less than the characteristic energy scale of the transverse confinement , _ i.e. _ , @xmath77 , where @xmath78 is the chemical potential . consider the general stationary solitonic solution to eq .
( [ eqn : gpe1d ] ) , @xmath79 assuming the transverse portion of the 3d wavefunction @xmath14 remains in the linear ground state , eq .
( [ eqn : sol ] ) may be substituted into eq .
( [ eqn : gpe1d ] ) to yield @xmath80 and @xmath81
. then @xmath82 implies @xmath83 , so that @xmath84 .
the intuitive geometrical criterion @xmath85 obtains essentially the same result : @xmath86 , so that @xmath87 .
the same argument applies for @xmath88 provided that its effect on the soliton length parameter is small . in the case where the characteristic transverse confinement length approaches the @xmath0-wave scattering length @xmath11 , the gpe , eq .
( [ eqn : gpe1d ] ) , no longer models the system and a one - dimensional field theory with the appropriate effective coupling constant must be considered instead @xcite . since @xmath89 is on the order of angstroms for a stable attractive bec , this regime is not relevant to the present study .
equation ( [ eqn : sol ] ) is a good ansatz to obtain the stationary properties of a solitonic solution of eq .
( [ eqn : gpe1d ] ) .
consider the experimentally relevant case of @xmath90 , where @xmath68 is retained to order one .
then , using the same variational methods as described in sec .
[ ssec:3d ] , one obtains the following expression for the extrema : @xmath91 for @xmath45 there is a single real , positive root , which represents a minimum of the energy functional ; for @xmath18 there are two real , positive roots , which signify a minimum and a maximum .
the minimum in the latter case , the analytic form of which shall be needed in sec .
[ ssec : evaporation ] , is written explicitly as @xmath92 , where @xmath93 decreases monotonically from @xmath94 to 1 with increasing @xmath95 , and is defined as follows : @xmath96^{1/3}}\right\ } \left(\frac{\zeta}{|\lambda|}\right)^{2/3 } \nonumber \\ & & + \left\{\frac{[1+\sqrt{1-(64\pi^2/27)(|\lambda|/\zeta)^2}]^{1/3}}{\pi^{4/3}}\right\ } \left(\frac{\zeta}{|\lambda|}\right)^{4/3 } \ , .
\nonumber\end{aligned}\ ] ] one obtains a critical point for explosion , either from the jacobean or directly from the analytic form of the roots . to prevent this from occurring ,
the parameters must be chosen such that @xmath97 this quasi-1d prediction for stability against explosion is illustrated in fig .
[ fig : crit ] ( dashed line ) . since @xmath98
, this criterion implies that the longitudinal harmonic oscillator length for an expulsive potential must be at least about two times the soliton width . as shall be shown in sec .
[ ssec : evaporation ] , quantum evaporation places a further constraint on the available parameter space . in the case where @xmath99 is small , _
i.e. _ , far from the critical value for explosion , eq .
( [ eqn : q1droot ] ) may be taylor expanded to obtain a simple analytic expression for the soliton length : @xmath100 \ , .\ ] ] to fourth order in @xmath99 , this expansion is accurate to within 1% over half the permissible parameter range given by eq .
( [ eqn : crit4 ] ) .
the dynamics of a bright soliton in an expulsive harmonic potential are herein studied , to the authors knowledge , for the first time .
several intriguing new features arise .
firstly , sudden changes in trap parameters , such as was the case in ref .
@xcite , cause particle loss and shape oscillations . changing the strength of the nonlinearity via a feshbach resonance
has similar effects . in both cases
the excitations are demonstrated to damp exponentially for small changes .
secondly , it is shown that a large change in the nonlinearity can be used , with present experimental apparatus , to create a higher order soliton .
thirdly , the lifetime of a soliton in an expulsive potential is not infinite .
the effective potential experienced by an atom in the soliton is the external potential plus the potential well created by the mean field , itself a function of the number of atoms .
thus the wavefunction undergoes tunneling with a rate which depends on @xmath13 .
this nonlinear process accelerates , eventually leading to explosion .
the exponential damping of weak excitations , nonlinear tunneling , and eventual explosion of a soliton differ radically from the case of a constant potential . in the following sections these phenomena
are studied analytically and numerically . in the latter case the condensate wavefunction , which solves eqs .
( [ eqn : gpe3d ] ) and ( [ eqn : gpe1d ] ) in three and one dimensions , respectively , is expanded as @xmath101 .
the @xmath102 are used as a coordinate - discretized pseudo - spectral basis , and the @xmath103 are propagated in time with a fourth - order variable step runge - kutta algorithm .
these numerical techniques are standard @xcite . absorbing bounds , _
i.e. _ , imaginary potential wells , are added far from the soliton in order to simulate an unbounded system , with care being taken to insure there are no reflections @xcite .
the motion of the center of mass of the condensate is decoupled from the relative motion of its constituent atoms , as is generically the case for harmonic potentials .
the center of mass is unstable in an expulsive potential ; in the following only the relative motion of the atoms is considered .
thus the condensate wavefunction is symmetrized in our simulations in order to keep its center of mass at rest . in order to understand the time evolution of a soliton which is created in a non - equilibrium initial state
, one may begin with the equations of motion for the variational parameters of eq .
( [ eqn : ansatz3d ] ) . in general
, one adds a quadratic imaginary factor in @xmath61 , @xmath104 , and @xmath105 to the exponential in eq .
( [ eqn : ansatz3d ] ) and then performs a lagrangian variational analysis @xcite . using this method
, one finds @xmath106 @xmath107 where the derivative has been taken with respect to rescaled time @xmath108 .
these equations match those of ref .
@xcite , up to small numerical factors due to the asymmetry between the hyperbolic secant and gaussian functions in the ansatz .
the factor of @xmath109 in eq .
( [ eqn : ham2 ] ) has been made explicit in order to demonstrate the difference between the lagrangian analysis and simply taking the first derivative of eq .
( [ eqn : energy2 ] ) , which gives an identical result up to a slight difference in the effective mass .
one may then perform a linear perturbation analysis around the minimum and derive small excitation frequencies .
since this has already been done elsewhere @xcite it is not repeated here .
the essential limitations on changes in the soliton and trapping parameters to avoid collapse and/or explosion can be gleaned directly from the variational studies of sec .
[ sec : stability ] .
the shape of the energy surface @xmath110 is that of a long , narrow rift running along @xmath30 .
the sides of the rift slope rapidly upwards towards infinity .
towards increasing @xmath30 there is a saddle point beyond which explosion occurs ; towards decreasing @xmath30 the rift turns to point toward the origin and ends at a second saddle point , after which the condensate collapses .
the turning is due to the three - dimensional nature of the soliton near collapse .
let the lower of these two saddle points have an energy @xmath111 .
then the curve @xmath112 defines a closed loop which signifies the lower bound for creation of a soliton .
any soliton created with the initial condition @xmath113 and @xmath114 which has length and width within this loop , though its motion along the energy surface may be complicated , must , by conservation of energy , be stable against collapse and explosion . in fig .
[ fig : contour ] is shown an example of the energy surface for the case of ref .
the loop defined by @xmath115 is marked with a square , while the open and filled circles represent the minimum and the initial state , respectively .
the above variational arguments do not include phonon or particle emission .
thus they can not model the damping of excitations , which , as shall be illustrated in the following section is , for an expulsive harmonic potential , exponential .
bright solitons in fiber optics are well known to shed their excess power by radiation .
this is an example of the robustness which makes them so useful for long distance communications .
thus any pulse injected with an area of @xmath116 into a fiber optic results in a soliton with amplitude oscillations which decay as @xmath117 , where the governing one - dimensional focusing nonlinear schrdinger equation is conventionally written as @xmath118 without a fixed normalization @xcite . a sudden change in the scattering length or number of particles in a matter - wave bright soliton is , by a rescaling of the wavefunction amplitude , time , and position , mathematically identical to injecting a soliton of the form @xmath119 into the above equation , where @xmath120 with @xmath121 from eq .
( [ eqn : gpe1d ] ) .
the subscripts @xmath122 and @xmath123 signify the states just before and just after the abrupt change in the scattering length .
this case has been solved previously with the inverse scattering transform @xcite .
supposing @xmath124 with @xmath125 , the resulting number of particles @xmath126 emitted from a soliton with initial number of particles @xmath127 is @xmath128 supposing @xmath129 , where @xmath130 , one finds @xmath131 , from which it is apparent that losses are very small indeed .
an identical result can be obtained by the overlap integral between the initial and final soliton wavefunctions @xcite , as is now demonstrated .
let @xmath132 be the initial wavefunction of the soliton with parameter @xmath133 , and let @xmath134 be the wavefunction of the steady state soliton with parameter @xmath135 .
one can always split the initial wavefunction @xmath136 into a component parallel to @xmath137 and a component @xmath138 orthogonal to @xmath137 , so that @xmath139 . since @xmath140 for @xmath135 close to @xmath133 , @xmath138 has a small norm @xmath141 and one may calculate its temporal evolution by linearizing the gpe around the steady state solution @xmath137 .
the spectrum of linear excitations , also called the bogoliubov spectrum , is @xmath142 , where @xmath143 is the wavevector of the emitted wave and @xmath144 is the chemical potential of the depleted soliton at long times @xcite . as a consequence of this dispersion relation , the emitted wave @xmath138 will spread as an unconfined wavepacket , with a spatial width which scales as @xmath145 and a maximum amplitude as @xmath117 .
therefore the number of particles emitted by the soliton at long times is simply the number of particles in the component @xmath138 , that is , @xmath146 .
the overlap integral is @xmath147 to second order in @xmath148 .
it follows directly that @xmath149 , as was found above by expanding eq .
( [ eqn : num1 ] ) in @xmath148 . in ref .
@xcite the abrupt change in parameters in eq .
( [ eqn : gpe3d ] ) occurs in the longitudinal trapping frequency @xmath7 , rather than the strength of the mean field .
although this case is not covered by the inverse scattering transform above , one may generalize the considerations given by the overlap integral of eq .
( [ eqn : overlap ] ) .
one must recalculate the linear excitation spectrum in the presence of an expulsive harmonic potential . to obtain a discrete spectrum
the system is treated as enclosed in a box of length much larger than the soliton size ; periodic boundary conditions are chosen . neglecting quantum reflections ,
the eigenmodes may be treated as semi - classical plane waves .
the effect of the soliton on these waves is simply a shift in phase , similiar to the case of a free soliton @xcite , as may be seen by the following .
consider a semi - classical wave with wavevector @xmath143 emerging from the soliton .
the expulsive potential can be neglected over the width of the soliton provided that @xmath150 , i.e. , @xmath151 , so that the soliton is very far from explosion , as given by eq . ( [ eqn : crit4 ] ) .
near the edge of the soliton one must verify the applicability of the wkb approximation with the standard criterion @xmath152 .
given @xmath153 , @xmath154 .
since the soliton edge is being considered , @xmath155 , and since the emerging wave is created by a global change in the soliton shape , @xmath156 so that @xmath157 . this leads to the same condition as above , @xmath150 . in this case
the transmission coefficient is @xmath158 , where @xmath159 varies from @xmath160 to zero as @xmath161 varies from zero to infinity @xcite .
this shift in phase does not change the density of states in the limit where the box length tends to infinity .
one concludes that the expansion of linear excitations is unaffected by the soliton .
thus the spatial width of @xmath138 increases as @xmath162 and its amplitude decays as @xmath163 for times long as compared to @xmath164 .
equation ( [ eqn : overlap ] ) gives @xmath165 in fig .
[ fig : shed ] the cases of free and expulsive harmonic potentials discussed above as well as that of an attractive harmonic potential are illustrated .
numerical integration of the quasi-1d gpe shows that a 10% change in the value of @xmath166 results in irregular oscillations for the attractive potential , damping of the maximum density as @xmath117 for zero external potential , and damping as @xmath167 for the expulsive potential , where @xmath168 . in the latter case ,
[ fig : shed](c ) , @xmath169 , whereas the prediction from above is @xmath170 .
this enhanced loss is due to nonlinear tunneling , as shall be explained in sec .
[ ssec : evaporation ] . in simulations
@xmath126 is determined by evaluating the norm squared of the wavefunction as a function of time , since the absorbing bounds remove atoms outside the immediate region of the soliton .
thus , provided that the soliton is created with width and length dimensions that lie within the bounds obtained from variational studies of the final parameter set , one may expect it to shed in a time @xmath164 its excess mass and come to rest at the local minimum of the energy surface .
in the preceeding section , the matter - wave parallel to the injection of an optical soliton of the form @xmath171 with @xmath124 and @xmath125 was considered .
a soliton of this form is in fact the first of a denumerably infinite family of higher order soliton solutions to the one - dimensional focusing nonlinear schrdinger equation .
each such solution can be produced from the initial state @xmath172 , where @xmath173 is a positive integer , without loss of particles .
moreover , provided that @xmath125 , @xmath174 will always result in a soliton of order @xmath173 , with a fractional loss of @xmath175 , as derived in ref .
bright solitons with zero relative phase have an attractive interaction . as a soliton of the form
@xmath176 has less than @xmath173 times the energy of a soliton of form @xmath177 , a higher order soliton can be interpreted as tightly bound state of @xmath173 strongly overlapping solitons .
it is periodic in time in both its phase and density .
matter - wave _ higher order soliton can be created by a sudden shift in the scattering length .
in particular , the order two soliton requires that the scattering length be decreased by a factor of four : for @xmath2li this factor can be produced simply for the state @xmath3 by a shift in magnetic field from @xmath178 gauss to @xmath179 gauss , thereby shifting @xmath11 from @xmath180 to @xmath181 , where @xmath182 is the bohr radius .
as with the soliton of order one , higher order solitons persist in the presence of a sufficiently weak expulsive potential . one supposes two experimental approaches : in the first , a soliton , already formed in an expulsive potential , is subjected to a sudden change in the scattering length ; alternatively , the scattering length of an attractive condensate trapped in an attractive harmonic potential is simultaneously shifted by a factor of four with the projection of the soliton onto a weakly expulsive potential .
the latter may be easier for observing oscillations of the soliton shape , since the center of mass in unstable unless the relative position of the potential and the soliton are continuously adjusted . in either case
the initial scattering length must be chosen such that it satisfies eq .
( [ eqn : crit4 ] ) , lest the soliton explode as it returns periodically to the form of its initial state .
thus , to avoid both collapse and explosion on experimental time scales and simultaneously accomodate the factor of four change in @xmath166 , one should choose trapping potentials such that @xmath183 , as deduced from fig .
[ fig : crit ] . the order two soliton solution to eq .
( [ eqn : gpe1d ] ) with @xmath76 takes the form @xcite @xmath184\end{aligned}\ ] ] where @xmath185 , as above , with @xmath186 being the final scattering length .
thus the period is @xmath187 while the spatial width is on the order of @xmath188 .
figure [ fig : hos ] illustrates the creation of an order two soliton by abruptly changing the scattering length from @xmath189 to @xmath190 and simultaneously projecting the soliton from a constant to an expulsive potential of frequency @xmath191 hz .
this matches the analytic form of eq .
( [ eqn:2sol ] ) . a negligible number of particles @xmath192 are emitted due to the projection , in agreement with the prediction of eq .
( [ eqn : loss ] ) , here extended to the case @xmath193 .
the above simulated parameters are equivalent to an experiment for which @xmath194 hz , @xmath195 hz , @xmath196 , @xmath197 , and @xmath198 particles for @xmath2li . the shape variations of this time - dependent nonlinear structure should be observable _ in situ _ with a period of 19 ms . a bright soliton in an expulsive potential experiences an effective potential @xmath199 , as sketched in fig .
[ fig : pot ] .
this is a time - dependent potential well , the depth of which is a function of the number of particles .
the soliton therefore undergoes nonlinear tunneling , which we have called quantum evaporation . to estimate the rate of quantum evaporation one may consider linear tunneling in the external potential @xmath200 , where one neglects the explicit time dependence of the wavefunction
the semi - classical wkb approximation then predicts , for a one - dimensional system @xcite , @xmath201 where the first factor of 2 comes from the fact that the wavefunction can tunnel in two directions , @xmath202 is a characteristic oscillation frequency of the particles in the well , @xmath203 is the barrier length through which the wavefunction must tunnel , and @xmath204 , where @xmath78 is the chemical potential . since the effective potential well contains a single bound state ( see fig .
[ fig : pot ] ) , one may take heuristically @xmath205 , which is the height of the chemical potential above the minimum of the well . supposing that the wavefunction is modeled by the variational ansatz eq .
( [ eqn : sol ] ) , one may determine @xmath78 and the well depth @xmath206 directly from the methods outlined in sec .
[ ssec : q1d ] : @xmath207 is given by eq .
( [ eqn : sol ] ) , the length @xmath208 by the real positive root of eq .
( [ eqn : extrema1d ] ) for @xmath18 , and the chemical potential by subsequently minimizing the eigenvalue of eq .
( [ eqn : gpe1d ] ) . the semi - classical wkb tunneling rate , scaled to the frequency of the expulsive potential , is then given by @xmath209 where @xmath210 and @xmath211 here the integration variable @xmath212 and the parameters @xmath213 and @xmath214 are determined by @xmath215 , as illustrated in fig .
[ fig : pot ] .
the function @xmath216 , as defined by eq .
( [ eqn : q1droot ] ) , varies monotonically between 1 and 4/3 , the latter value occuring at the critical point for explosion . here
the additional chemical potential @xmath74 due to transverse confinement is neglected , since in the quasi-1d regime this is assumed to have no effect on the dynamics .
equation ( [ eqn : wkb2 ] ) is a function of the single parameter @xmath217 . in the case where the soliton is far from explosion , _
i.e. _ , @xmath218 , as given by eq .
( [ eqn : crit4 ] ) , the above reduces to the following simple limit : @xmath219 , @xmath220 , and @xmath221 .
the asymptotic tunneling rate is therefore @xmath222 a function of the single parameter @xmath223 . in fig .
[ fig : qe2](a ) is shown the complete and asymptotic wkb variational predictions of eq .
( [ eqn : wkb2 ] ) ( solid line ) and eq .
( [ eqn : wkb3 ] ) ( dashed line ) , respectively , and numerical data obtained by direct simulation of eq .
( [ eqn : gpe1d ] ) ( black diamonds ) . in fig .
[ fig : qe2](b ) the nonlinear tunneling of a soliton in an expulsive harmonic potential is shown explicitly .
the solid line shows the result given by numerical evolution of the quasi-1d gpe . in this simulation @xmath224 , or @xmath225 .
it is apparent that the tunneling accelerates and the soliton explodes , the latter occurring after about 5% of the particles in the soliton have been lost .
this differs by about 10% from the predicted critical point of eq .
( [ eqn : crit4 ] ) , pointing to a level of accuracy in the quasi-1d variational approximation similiar to that of the 3d variational approximation of sec . [ ssec:3d ] and
[ fig : crit ] .
the dashed line represents the numerical integration of @xmath226 , where @xmath227 is obtained by the wkb approximation as shown in fig .
[ fig : qe2](a ) .
the explosion time is similar to that of the simulation ; the critical particle number is as given by the variational approximation , as assumed in the above derivation . to experimentally observe quantum evaporation and subsequent explosion one faces the difficulty that the center of mass of the soliton is unstable .
there are two solutions : one may select parameters such that the tunneling rate is maximized and try to observe the soliton as it falls ; or one may use a soliton which is large enough to observe _ in situ _ and construct a feedback loop to continuously shift the peak of the expulsive potential so as to keep the soliton centralized at its maximum .
note that a soliton of order @xmath228 is expected to show the same kind of tunneling properties .
the size , energy , and stability regimes of a stable matter - wave bright soliton in a harmonic potential which is transversely attractive and longitudinally expulsive have been presented .
the critical branches for collapse and explosion have been delineated analytically .
the equations of motion of the soliton parameters were calculated , and the response of the soliton to sudden changes in trap parameters or the scattering length was shown to result in excitations which damp exponentially by emission of particles from the soliton .
it was shown that a higher order soliton can be created by a sudden change in the scattering length by a factor of four .
finally , an intriguing nonlinear dynamical effect was presented : a matter - wave bright soliton undergoes quantum evaporation and eventually explodes .
the parameter space for creation of a locally stable matter - wave bright soliton is accessible to present experiments , as already demonstrated in refs .
@xcite and @xcite .
data from the former shows that solitons are self - adjusting : when the scattering length of a condensate with @xmath22920,000 atoms was tuned to be negative , a large thermal cloud with a stable condensate core of @xmath2295,000 atoms was observed to remain .
projection onto an expulsive potential caused the observed thermal cloud to be rapidly expelled from the vicinity of the condensate , resulting in a nearly pure soliton .
the choice of a weaker expulsive potential and a less negative scattering length than that of ref .
@xcite could be used to create solitons observable _ in situ_. thus the experimental outlook for the investigation of matter - wave bright solitons is very positive .
an important aspect of solitons which could be investigated in future work is their interactions : this is an essential aspect of solitons which gives them their particle - like nature , in contrast to a solitary wave .
binary interactions between solitons depend on their relative phase : for @xmath230 they are attractive , while for @xmath231 they are repulsive @xcite .
such interactions could therefore measure phase decoherence times .
* acknowledgments : * we thank iacopo carusotto , gora shlyapnikov and the group of christophe salomon ( thomas bourdel , julien cubizolles , gabriele ferrari , lev khaykovich , and florian schreck ) for helpful discussions .
this work was supported by the distinguished international fellowship program of the national science foundation grant no .
mps - drf 0104447 .
laboratoire kastler brossel is an unit de recherche de lcole normale suprieure et de luniversit pierre et marie curie , associe au cnrs .
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 * , 97 ( 2000 ) .
y. castin , in _ coherent atomic matter waves _ , edited by r. kaiser , c. westbrook , and f. david ( edp sciences and springer - verlag , les ulis , france and berlin , germany , 2001 ) , pp .
1136 , e - print cond - mat/0105058 .
specifically , in one - dimensional simulations of eq .
( [ eqn : gpe1d ] ) , an external potential of form @xmath232 for @xmath233 is added to a box of length @xmath234 centered at the origin . for simulation parameters
@xmath235 , @xmath236 , and @xmath237 the absorbing bound parameters @xmath238 and @xmath239 were found to be sufficient . in the case
where @xmath240 , one must avoid the reflection of the low energy matter waves induced by the spatial variation of @xmath241 by adding a real potential drop far from the center of the soliton and before the absorbing bounds ; a sinusoidal function squared was used . | the stability regimes and nonlinear dynamics of bright solitons created in a harmonic potential which is transversely attractive and longitudinally expulsive are presented .
this choice of potential is motivated by the recent creation of a matter - wave bright soliton from an attractive bose - einstein condensate ( l. khaykovich _ et al .
_ , science 296 , 1290 ( 2002 ) ) .
the critical branches for collapse due to the three - dimensional character of the gas and explosion caused by the expulsive potential are derived based on variational studies .
particle loss from the soliton due to sudden changes in the trapping potential and scattering length are quantified .
it is shown that higher order solitons can also be created in present experiments by an abrupt change of a factor of four in the scattering length .
it is demonstrated that quantum evaporation occurs by nonlinear tunneling of particles out of the soliton , leading eventually to its explosion . |
the self - assembly and self - organization of large numbers of microorganisms and artificial microswimmers and the non - equilibrium phase transitions that result from their collective behaviour have recently become the focus of many theoretical and experimental studies @xcite .
thus , in a series of experiments , carried out with different types of artificial microswimmers , several collective phenomena have been reported such as dynamic clustering , phase separation , and swarming @xcite . in experiments with suspensions of motile living cells ( e.g. _
e.coli _ and _ b.subtilis _ bacteria or spermatozoa ) , a variety of regular and irregular large- and meso - scale density patterns has been found @xcite . with the typical body size of several @xmath0 m
, the colonies of motile cells exhibit arrays of circular vortices , swirls , and meso - scale turbulence with the correlation length of the collective motion ranging between @xmath1 m and @xmath2 m .
the emergence of large - scale coherent structures in systems composed of small - scale self - propelled objects is universal and independent of the mechanism of motility .
thus , density waves with @xmath3 m wavelengths are observed in an assay of @xmath4 m long actin filaments , driven by motor proteins @xcite .
stable networks of interconnected poles and asters are found in systems of microtubuli driven by kinesin complexes @xcite .
similar to the suspensions of sea urchins spermatozoa @xcite , highly coherent arrays of circular vortices were found in motile assays of microtubulus , propelled by surface - bound motor proteins @xcite . in order to explain the observed large- and meso - scale patterns , many theoretical models of interacting self - propelled particles
have been suggested and tested against the experimental findings .
one of the central questions of modelling is to determine the driving force and the minimal conditions for the emergence of each of the observed patterns .
historically , the first class of the developed models are the so - called _ dry _ systems , in which the motion of the embedding fluid medium is neglected @xcite .
in contrast , in _ wet _ systems , the motion and influence of the medium is considered as well ( see e.g. @xcite and references therein ) . in dry systems
the formation of density patterns is triggered by a linear instability of the homogeneous isotropic state ( i.e. , the trivial state ) .
the instability is caused by the direct interaction between the particles , which is not mediated or induced by a solvent . in the case of
electrically neutral and non - magnetic particles , the interaction mechanisms can be roughly divided into two categories .
the first category deals with steric effects such as the hard - core repulsion between colliding particles @xcite .
all other interaction types are due to long - range forces and fall within the second category .
they have to be introduced phenomenologically , such as the aligning or anti - aligning interaction in vicsek - type flocking models , @xcite or are due to hydrodynamic @xcite or phoretic interactions @xcite .
the physical mechanism of aligning interactions can be explained by collisions between swimmers with elongated bodies @xcite or by the bundling of flagellas of two colliding bacteria @xcite . the true origin of long - range anti - aligning interactions has not been properly explained yet . in the case of _ wet _ systems , the motion of the solvent medium gives rise to hydrodynamic interactions between the suspended particles @xcite .
several experimental and theoretical studies show that the inclusion of hydrodynamic interactions may destabilize the polar order at high densities , thus , it effectively acts as a long - range anti - aligning force @xcite .
presently , it is understood that the instability of a homogeneous suspensions of self - propelled particles can be induced by combining particle motility with either steric repulsion or with an aligning / anti - aligning interaction .
thus , it has been shown that at sufficiently high mean particle density , phase separation may occur in two - dimensional systems of repulsive finite - sized swimmers or self - propelled discs @xcite .
this result is explained by the self - trapping of colliding swimmers , i.e. , any two swimmers that collide and swim against each other remain in contact for a certain time span until their swimming directions have sufficiently changed .
recently , it was shown that a mixture of short - range aligning and long - range anti - aligning interactions between point - like active brownian particles leads to a rich variety of density and velocity patterns in dry systems without memory @xcite and with memory @xcite .
in contrast to previous studies , we demonstrate here that emergent collective dynamics in the form of persisting regular and irregular meso - scale density patterns can also be found in colonies of self - propelled particles that do not interact directly . to this end , we consider active brownian surfactant particles that move on the deformable surface of a thin liquid layer supported by a solid substrate .
the direction of swimming of each particle is assumed to have a non - zero vertical component , thus , leading to particles pushing against the film surface .
variations in the particle density give rise to large - scale film surface deformations that in turn induce flow in the layer of viscous fluid , which drives even more particles by advection towards denser regions and also rotates their swimming directions . particle diffusion and in particular the marangoni effect act stabilizing .
we assume that the swimmers act as a surfactant , i.e. , the local surface tension decreases with swimmer concentration .
this results in a soluto - marangoni effect , i.e. , marangoni forces due to concentration gradients act at the free surface of the liquid film - they are a direct consequence of entropic contributions to the free energy of the free interface plus surfactant system @xcite . in consequence , these marangoni forces act stabilizing by driving the liquid away from areas with increased particle concentration .
thus , in this system the interaction between the particles is indirect and only occurs when the liquid layer dynamically deforms .
the first model system of swimmers on a liquid carrier layer was introduced in ref.@xcite .
there it is assumed that the particles swim exclusively upwards at all times , and are not able to move along the film surface by self - propulsion .
it is shown that the resulting excess pressure onto the film surface may cause a long - wave deformational instability of the film .
the picture becomes more diverse , when one allows for lateral active motion of the swimmers as well @xcite .
then , a sufficiently large swimming velocity and a moderate rotational diffusion strength can suppress the long - wave instability due to the excess pressure discussed in ref .
@xcite . in ref .
@xcite the linear stability results are confirmed by hybrid ( multiscale ) discrete - continuous numerical simulations , but fully nonlinear results obtained with the continuum model become available only now . the paper is organized as follows : in section [ sec1 ] we derive the coupled long - wave evolution equations for the space- and time - dependent full swimmer density , which also includes swimmer orientation ( smoluchowski equation ) , and the space- and time - dependent film profile ( thin film equation ) .
next , we present a detailed linear - stability analysis of the trivial steady state , i.e. , of a homogeneous distribution of swimmers without preferred swimming direction ( orientation ) on the surface of a flat film .
different instability modes are discussed and located in a stability diagram spanned by rotational diffusion and self - propelling velocity . in section [ sec2 ] we discuss various spatio - temporal patterns that emerge in the nonlinear regime , when the swimmers self - organize into persisting non - uniform structures . in particular , we analyse stable standing and travelling density waves accompanied by film modulation waves , travelling waves that are modulated by large scale structures , and irregular patterns .
we discuss the multistability of several of these states in a certain region of the parameter space .
and with a single swimmer ( filled circle ) .
the instantaneous swimming direction is given by the vector @xmath5 that makes an angle @xmath6 with the local normal @xmath7 .
@xmath8 is the projection of @xmath5 onto the @xmath9 plane .
[ f1],scaledwidth=50.0% ] we consider a @xmath10-@xmath11 @xmath0 m thin liquid film on a solid plate with a time - dependent film thickness profile @xmath12 . the deformable liquid - gas interface ( shortly called `` free surface '' )
is covered by a colony of non - interacting active brownian particles .
besides being microswimmers the particles act as insoluble surfactants @xcite as often found for small particles @xcite . in this way the particles
are confined to move along the free surface and their density influences the interfacial energy of the free surface .
the lower bound for the average film thickness of @xmath13 m is dictated by the typical size of self - propelling mono - cellular organisms such as _ e.coli _ or _ the african trypanosome _
@xcite , or artificial swimmers such as phoretically driven janus particles ( see , for example , @xcite ) .
the typical size @xmath14 of such swimmers is of the order of @xmath15 m although recently much smaller swimmers of @xmath16 nm have also been created @xcite . in what follows
we assume a dilute limit , where the average separation distance between the swimmers is much larger than their size .
in this regime , direct two - particle interactions as well as hydrodynamic interactions can be neglected .
we start by deriving the equations of motion for an active brownian particle that moves along a two - dimensional time - dependent surface profile @xmath12 , as shown schematically in fig .
the three - dimensional position vector of the particle is given by @xmath17 .
the kinematic equation for the velocity reads @xmath18 note that for fixed surface shape the velocity vector @xmath19 and the unit vector normal to the surface @xmath20 are orthogonal to each other : @xmath21 . in the overdamped
limit the total velocity of the particle moving along the interface @xmath12 is given by the superposition of the local tangential components of the self - propulsion velocity @xmath22 , the local tangential fluid velocity @xmath23 of the film , the tangential component of the gravity force @xmath24 , and thermal noise @xmath25 , which results in diffusion along the free surface , @xmath26 here , @xmath27 denotes the mobility of the particle , @xmath28 its effective mass , which is reduced due to buoyancy effects for partly submerged particles , and the unit vector @xmath5 indicates the direction of swimming .
thermal noise is characterized by a gaussian random variable with zero mean and correlation function @xmath29 , where @xmath30 is the boltzmann constant , @xmath31 is the absolute temperature , and @xmath32 is a @xmath33 unit matrix .
note that the tangential component @xmath34 of any vector @xmath35 is given by @xmath36 furthermore , the fluid velocity @xmath37 at the free surface satisfies the standard kinematic boundary condition resulting from continuity @xcite @xmath38 we consider particles swimming upward against gravity and pushing against the film surface .
this already creates some polar order with a preferred vertical orientation of the swimmer bodies at the interface @xcite .
further reasons for such a polar order can be bottom - heaviness @xcite , the chemotactic response of bacteria swimming towards the surface in order to take up oxygen @xcite , or any mechanism at the interface that aligns the particles along the vertical . in the following , we will not present a full derivation of the orientational distribution at the interface . instead , for the distribution against the surface
normal we will assume that it always adjusts instantaneously compared to the slow dynamics of the film interface ( see below ) .
in what follows , we only take into account the long - wave deformations of the film surface , thus @xmath39 , with @xmath40 the wavelength of the surface deformations and @xmath41 the average film thickness . by noticing that the in - plane gradient @xmath42 is of order @xmath43
, we obtain for the surface normal @xmath44 and for any vector @xmath35 one has @xmath45 , as it follows from eq.([tau ] ) .
then , the langevin equation ( [ kinetic ] ) for the interfacial particle position becomes in leading order of @xmath43 , @xmath46 note that the tangential component of the gravity field vanishes in the long - wave limit , i.e. @xmath47 . the instantaneous orientation of swimmers is indicated by the three - dimensional unit vector @xmath5 , as shown in fig.[f1 ] . for swimmers in the bulk of the fluid
, the time - evolution of @xmath5 is well known : it is determined by the rotation due to local fluid vorticity , alignment against some external field such as gravity , and random rotation with the rate controlled by the rotational diffusivity @xmath48 . however , for swimmers at a free surface , the rate of change @xmath49 may be significantly modified as compared to bulk swimmers depending on the nature of the interaction between the swimmers and the free film surface .
for instance , a surface swimmer only partly submerged in the fluid and possibly with elongated body shape is easily rotated by local fluid vorticity within the film surface .
however , the rotational rate of @xmath5 against the interface normal is possibly reduced as the interaction energies change with orientation of the swimmers at the free surface similar to anchoring effects for liquid crystals . here
, we refrain from deriving the exact evolution equation for the orientation vector @xmath5 of partly submerged surface swimmers .
instead , we use the argument from above to decouple the vertical component @xmath50 from the in - plane component @xmath51 .
thus , we assume that the evolution of the film surface is slow and the equilibration of @xmath5 to a stationary distribution @xmath52 with respect to the vertical is fast . in this case , the mean vertical component of @xmath5 is given by @xmath53 whereas the in - plane component @xmath51 can vary according to the in - plane dynamics of @xmath5 , which couples to the temporal film evolution .
note that @xmath54 , which implies that the swimmers push on average against the liquid - gas interface . as a result
, the rotation of the in - plane component @xmath51 is described in terms of the polar angle @xmath55 ( see fig.[f1 ] ) @xmath56 where @xmath57 is the vertical component of the local fluid vorticity and @xmath58 is rotational noise with correlations @xmath59 .
furthermore , we introduce the mean in - plane velocity of an active particle , @xmath60^{1/2}$ ] and substitute @xmath61 in eq.([kinetic_xy ] ) by @xmath62 .
then , the smoluchowski equation for the particle probability density @xmath63 , equivalent to eqs.([kinetic_xy ] ) and ( [ rotation ] ) , reads @xmath64 where @xmath65 and the respective translational ( @xmath66 ) and rotational ( @xmath67 ) probability currents become @xcite @xmath68 the swimmers and the liquid - gas interface couple to each other through the local swimmer concentration @xmath63 that acts twofold .
first , each swimmer exerts the force @xmath69 in the direction normal to the surface @xcite .
so , the total pushing force @xmath70 of the swimmers per unit area is proportional to the direction - averaged local concentration of swimmers , @xmath71 , and becomes @xmath72 secondly , the swimmers act as a surfactant and change the local surface tension . assuming a relatively low concentration of swimmers
, the surface tension @xmath73 is known to decrease linearly with the local direction - averaged concentration @xmath74 @xcite , @xmath75 with the reference surface tension @xmath76 and @xmath77 . through eqs .
( [ push ] ) and ( [ marangoni ] ) the smoluchowski equation([smoluch ] ) and the thin film equation in the long - wave approximation @xcite for the local film thickness @xmath12 are coupled to each other @xcite , @xmath78\right ) -\gamma { \bm \nabla}\cdot\left(\frac{h^2}{2\mu}{\bm \nabla}\langle \rho \rangle\right)=0,\ ] ] where @xmath79 is the density of the fluid and @xmath0 is its dynamic viscosity .
the in - plane fluid velocity at the interface , @xmath80 , is determined by the film profile @xmath12 @xcite , @xmath81 and the vertical component of the vorticity is obtained as @xmath57 from eq.([surface_vel ] ) @xcite .
both , @xmath82 and @xmath83 enter the currents ( [ currents ] ) that determine the smoluchowski equation ( [ smoluch ] ) .
note , that without swimming along the interface and rotational diffusion , one can integrate this equation over @xmath55 to recover the model in ref .
@xcite with purely upwards pushing swimmers .
switching off the active swiming motion altogether one recovers the classical long - wave model for a dilute insoluble surfactant on a liquid film that may be written in a gradient dynamics form @xcite . in what follows we focus on the instability induced by the pushing force generated by swimmers that swim predominantly upwards . to this end , we neglect the stabilising effect of the hydrostatic pressure @xmath84 as compared with the typical pushing force per unit area @xmath85 .
experimentally , such a regime can be achieved by using , for example , bacteria - covered water films with a dense bacterial coverage . in the dilute limit
treated in this manuscript , one needs conditions of microgravity . to illustrate this example
, we present some estimates .
the maximal self - propulsion force of a unicellular bacterium is known to be of the order of several @xmath86 @xcite .
the maximal surface density @xmath87 is estimated as @xmath88 , where @xmath89 m is the typical size of the bacterial body .
the dilute limit corresponds to densities of at least one order of magnitude below @xmath90 .
consequently , the maximal pushing force per unit are is estimated as @xmath91 . on the other hand , @xmath92 for a @xmath13
m thick water film .
clearly , @xmath93 can be neglected against @xmath94 in case of @xmath95 . the possibility to experimentally detect the thin - film instability due to the pushing force @xmath96 exerted by self - phoretic particles
is further strengthened by recent experiments with @xmath97 nm small janus particles @xcite .
a much smaller particle size allows for larger surface particle densities and may give rise to larger excess pressure .
in fact , the maximal density increases as @xmath98 for decreasing particle size @xmath14 .
however , it remains unclear how the pushing force @xmath96 of a single self - phoretic particle scales with its size .
if the decrease of the pushing force with @xmath14 is slower than @xmath99 , the resulting excess pressure @xmath100 exerted by the particles onto the liquid - gas interface can be several orders of magnitude larger than the value of @xmath101 estimated before in the dilute limit of a bacterial carpet .
for all what follows , we non - dimensionalise the evolution equations for film thickness @xmath12 and swimmer density @xmath63 employing the scaling as in ref .
thus , we use @xmath41 as the vertical length scale , @xmath102 as the horizontal length scale , @xmath103 as the time scale , and the direction - averaged density of swimmers in the homogeneous state @xmath87 as the density scale .
this gives the relevant parameters of our model : the dimensionless in - plane self - propulsion velocity @xmath104 , the dimensionless in - plane rotational diffusivity @xmath105 , the translational surface diffusivity @xmath106 , and the excess pressure parameter @xmath107 .
furthermore , we introduce the effective in - plane diffusivity @xmath108 note that @xmath109 corresponds to the diffusion coefficient of a single self - propelled brownian particle moving along a flat two - dimensional surface @xcite . in appendix [ appendixa ] we summarize our non - dimensionalised dynamic equations .
we start by presenting more details of our stability analysis of the flat film as compared to our previous work @xcite including an analytical treatment and a more thorough discussion of the occuring dispersion relations .
we linearise the non - dimensionalized eqs .
( [ smoluch ] ) and ( [ thin_film ] ) about the homogeneous isotropic steady state given by @xmath110 , @xmath111 , using the ansatz @xmath112 where @xmath113 .
the linearised smoluchowski equation ( [ smoluch ] ) and the thin film equation ( [ thin_film ] ) become , respectively , @xmath114\frac{1}{2\pi } - d\delta \delta \rho - d \partial^2_\phi \delta \rho=0 .
\\ \label{thin_film_lin } \partial_t ( \delta h)+\frac{1}{3}\left[\delta^2 ( \delta h ) + \beta \delta \langle \delta \rho \rangle \right ] -\frac{1}{2}\delta \langle \delta \rho \rangle=0,\end{aligned}\ ] ] with @xmath115 .
the linearised surface velocity @xmath116 from eq .
( [ surface_vel ] ) reads @xmath117 \ , .\ ] ] next , we follow @xcite and fourier transform the perturbations @xmath118 and @xmath119 by using a continuous fourier transform in space and a discrete fourier transform in the angle @xmath55 . combining this with an exponential ansatz for the time evolution of the individual modes we have @xmath120 with the small dimensionless fourier amplitudes @xmath121 and @xmath122 , the wave vector of the perturbation @xmath123 , and the real or complex growth rate @xmath124 . substituting the expansions from eqs.([expansion ] ) into the linearized eqs.([smoluch],[thin_film ] ) , we obtain the eigenvalue problem @xmath125 with the eigenvector @xmath126 @xmath127 and the jacobi matrix @xmath128 , which corresponds to a banded matrix of the structure @xmath129 where @xmath130 , @xmath131 , @xmath132 .
the @xmath133 matrix @xmath134 in the upper left corner of @xmath128 is given by @xmath135 note that the matrix @xmath134 coincides with the jacobi matrix derived in ref.@xcite that encodes the linear stability in the special case of a flat film covered by autonomous purely upwards pushing motors that exert an excess pressure onto the liquid - gas interface .
in practice , we truncate the expansion in the angle @xmath55 and only take the first @xmath136 fourier modes into account .
then , the jacobi matrix @xmath128 is a @xmath137 matrix and the truncated eigenvector @xmath138 is @xmath139 dimensional . and @xmath140 .
in the stable lightly - shaded region re@xmath141<0 $ ] for all values of the wave number @xmath142 . the finite wave number instability sets in along the line marked by `` fw '' . the zero wave number instability sets in along the vertical dashed line @xmath143 , marked by `` zw '' .
the inset zooms into the region marked by the rectangle in the main panel : in the strongly shaded area marked `` mi '' , the mixed type fw /zw instability occurs .
panels ( b - d ) show re@xmath141 $ ] of the two leading eigenvalues for parameters taken ( b ) at point @xmath144 ( @xmath145 , @xmath146 , i.e. , @xmath147 ) , ( c ) at point @xmath148 ( @xmath145 , @xmath149 , i.e. , @xmath150 ) , and ( d ) at point @xmath151 ( @xmath152 , @xmath153 , i.e. , @xmath154 ) in panel ( a ) . dashed and solid lines indicate complex and real eigenvalues , respectively .
the inset in ( d ) zooms into the region marked by the rectangle .
[ f2],scaledwidth=95.0% ] the stability diagram of the homogeneous isotropic state , as computed for reduced excess pressure @xmath155 and translational diffusivity @xmath140 from eqs.([eig_lin ] ) is shown in fig.[f1](a ) in the parameter plane spanned by the reduced in - plane velocity @xmath156 and the effective diffusion constant @xmath157 of the surfactants .
we numerically compute the eigenvalues of the truncated jacobi matrix eqs.([jacobi ] ) for @xmath158 fourier modes and then check the results by doubling the number of the fourier modes to @xmath159 .
note that the choice @xmath155 and @xmath140 corresponds to a flat film that is unstable at zero in - plane velocity , @xmath160 as @xmath161 , the critical value for the onset of the long - wave instability @xcite , e.g. , @xmath162 for @xmath140 . in what follows
, we characterize the system by the set of parameters @xmath163 and also indicate the respective value of @xmath109 . for the system with non - zero in - plane velocity ( @xmath164 ) ,
we have earlier reported the existence of two different instability modes @xcite .
namely , for sufficiently large @xmath156 , there exists a wedge - shaped stability region , marked in fig.[f2](a ) by `` stable '' that separates regions where the two different instability modes occur .
the wedge opens at @xmath165 towards larger values of @xmath156 , i.e. , at any @xmath166 , there exists a window in the effective diffusivity @xmath109 for which the flat homogeneously covered film is stable ( note that @xmath167 depends on @xmath168 and @xmath169 ) . by crossing the two borders of the linearly stable region , the system changes stability via two distinct instability modes .
the first mode corresponds to an oscillatory instability with a finite wave number at onset , i.e. , a travelling wave instability . in this case , for parameters directly on the stability threshold , the leading eigenvalue @xmath170 of the jacobi matrix from eq .
( [ jacobi ] ) has a negative real part for all values of the wave number @xmath142 , except for the critical wave number @xmath171 , where @xmath170 has the form @xmath172 with some non - zero frequency @xmath173
. we will refer to this instability mode as the finite wave number instability ( fw ) .
the second mode ( @xmath174 ) corresponds to an instability with a zero wave number at onset .
this mode is characterized in detail in the next section . in the following
we present an approximate analytic expression for the zero wave number instability .
we start by introducing the fourier transformed fields @xmath175 and @xmath176 , according to @xmath177 and @xmath178 , into eq.([smoluch_dimless ] ) and obtain @xmath179 close to the threshold of the zero wave number instability , the amplitudes of all modes with the wave number @xmath180 rapidly decay with time .
therefore , in the limit @xmath181 , in eq.([smoluch_ft ] ) one may neglect the terms of orders @xmath182 and @xmath183 as compared to the ones of order @xmath184 and @xmath142 . in consequence ,
the density and film height equations decouple .
in fact , to this order the density equation describes a single self - propelled particle with rotational diffusivity @xmath185 and self - propulsion velocity @xmath156 but with neglected translational diffusivity . to proceed further
, we note that on length scales larger than the persistence length of an active particle , @xmath186 , the dynamics becomes purely diffusive .
to arrive at this result , one performs a multipole expansion of @xmath176 in the angle @xmath55 using only the monopol @xmath187 and the dipole moment @xcite .
the latter can be elimated in the dynamic equation for @xmath187 and from eq .
( [ smoluch_ft ] ) one arrives at @xmath188 which is coupled to the linearised thin film equation in fourier space @xmath189 + \frac{1}{2}k^2\hat{\langle \rho \rangle}=0.\end{aligned}\ ] ] here , @xmath190 is the effective diffusion constant of an active particle introduced earlier in eq .
( [ eff_diff ] ) .
the additional term results from the activity of the particle . equations ( [ smol_3 ] ) and ( [ thin_film_1 ] ) are identical to the linearised evolution equations found for the concentration field of the purely upwards swimming ( @xmath160 ) surfactant particles coupled to the thin film equation , as studied in ref.@xcite .
in fact , the results of the linear stability analysis of ref .
@xcite can be translated to the system of equations ( [ smol_3 ] ) and ( [ thin_film_1 ] ) by setting the translational diffusivity of the purely upwards swimming particles to be equal to @xmath109 .
the two eigenvalues @xmath170 resulting when introducing an exponential ansatz for the time dependence into eqs.([smol_3 ] ) and ( [ thin_film_1 ] ) are determined analytically ( cf . ref .
the jacobi matrix @xmath191 of the linearised eqs.([smol_3 ] ) and ( [ thin_film_1 ] ) is given by @xmath192 where we introduced @xmath193 .
the two eigenvalues @xmath194 are @xmath195 with @xmath196 $ ] and @xmath197 . )
( dotted lines ) .
( a ) @xmath145 , @xmath147 , ( b ) @xmath152 , @xmath154 .
the inset zooms into the region marked by the rectangle , where the analytic and numerical results are indistinguishable from each other . in the inset
we only show the analytically computed eigenvalues . , scaledwidth=95.0% ] in fig.[f3](a , b ) we compare the analytic eigenvalues given by eq.([a1_gamma ] ) with the two leading eigenvalues of the original non - reduced system , computed numerically , as described in section[linstabgen ] .
solid and dashed lines correspond to the numerically computed real and complex eigenvalues , respectively , dotted lines correspond to eq.([a1_gamma ] ) .
fig.[f3](a ) is obtained for the parameters corresponding to point @xmath144 in fig.[f2](a ) , i.e. , @xmath145 and @xmath147 , and fig.[f3](b ) corresponds to point @xmath151 in fig.[f2](b ) , i.e. , @xmath152 and @xmath154 . in both cases , the agreement is excellent up to @xmath198 .
next , we use eq.([a1_gamma ] ) in order to classify the zero wave number instability that sets in along the dashed vertical line marked by zw " in fig . [ f2](a ) .
analysing the eigenvalues in eq.([a1_gamma ] ) shows that the real part of the leading eigenvalue changes its sign at @xmath199 , or , equivalently at @xmath200 .
thus , for the value of @xmath155 used here , we obtain @xmath143 , in agreement with fig.[f2](a ) .
we emphasize that the above analytic results can only be applied in the limit of @xmath201 , where the approximation ( [ smol_3 ] ) applies .
we find that in the unstable region , i.e. , for @xmath202 , the fastest growing wave number @xmath203 , indicated in fig .
[ f3](a ) , always corresponds to a pair of two complex conjugate eigenvalues [ dashed line in fig .
[ f3](a ) ] . by locating the maximum of @xmath204 $ ] , we determine @xmath203 and its complex growth rate , @xmath205=\frac{3}{8}\eta^2,\,\,\,{\rm im}[\gamma(k_{\rm max})]=\pm\frac{\mid\eta \mid}{2}\sqrt{\frac{27}{2}\mid \eta\mid\left(\frac{1}{12}+\frac{d_{\rm eff}}{3}\right)-\frac{9}{16}\eta^2}.\end{aligned}\ ] ] for @xmath206 , the two leading eigenvalues are real , as indicated by the two solid lines in fig . [ f3](a ) . the wave number @xmath207 is @xmath208 this result implies that the character of the zero wave number instability is peculiar : directly at onset ( @xmath199 ) the leading two eigenvalues are real , however , already arbitrarily close above onset ( @xmath202 ) the band of unstable wavenumbers contains a region of real modes ( close to and including @xmath209 ) and a region of complex modes ( always including the fastest growing mode ) .
this behaviour is related to the existence of two conserved fields , @xmath210 and @xmath211 , that forces two real modes with growh rate zero at @xmath209 . in consequence
, the fastest growing wave number @xmath203 tends to zero when approaching the stability threshold from above . here
, we call this scenario a zero - wave number instability . the analytic results obtained in the previous section give for the zero wave number instability the threshold @xmath143 for @xmath155 .
this perfectly coincides with the numerically computed threshold as indicated by the left thick dashed line in fig.[f2](a ) . however
, these results also remain valid around @xmath143 for small values of @xmath212 deep in the unstable region .
this implies that regardless of the value of @xmath156 , the sign of the dispersion curve @xmath213 $ ] at very small wave numbers @xmath214 changes from negative to positive as @xmath109 is decreased past the critical value @xmath143 ( at @xmath155 ) .
however , there will always be an instability at a non - zero wave number , as we explain now . we observe a mixed - instability region , marked by " mi and heavily shaded in the inset of fig.[f2](a ) , where the system can be described as being unstable with respect to a mixed finite- and zero - wavelength instability . in this region ,
there exist two bands of unstable wave numbers : one with @xmath215 $ ] and another one with @xmath216 $ ] , with @xmath217 .
the two leading eigenvalues that correspond to the pure zero wave number , the pure finite wave number , and the mixed instabilities , are shown in figs.[f2](b ) , ( c ) , and ( d ) , respectively .
the respective parameters are ; point 1 : @xmath218 , i.e. , @xmath147 , point 2 : @xmath219 , i.e. , @xmath150 in the main panel , and point 3 : @xmath220 , i.e. @xmath154 in the inset of fig.[f2](a ) . dashed and solid lines correspond , respectively , to complex and real eigenvalues . when considering the type of dispersion curves , the transition from the finite wave number instability to the zero wave number instability can follow two different scenarios .
we identify them by keeping @xmath156 constant and gradually decrease @xmath109 . in the first scenario for @xmath221 , the maximum of @xmath222
$ ] in fig .
[ f2](c ) first becomes negative , i.e. , the finite wave number instability is stabilised while crossing the `` fw '' line in the stability diagram .
subsequently , the system crosses the " zw line and becomes unstable w.r.t . the zero wave number mode . in the second scenario for @xmath223 , the system
first crosses the line @xmath143 and enters the mixed instability region
. then the dispersion curve as in fig .
[ f2](d ) gradually transforms into the dispersion curve of the zero wave number instability , while leaving the region mi .
we also remark that the mixed instability region stretches from @xmath224 down @xmath160 . however , its horizontal width is negligibly small for @xmath225 .
the second scenario is visualised in fig .
[ f4 ] . in fig .
[ f4](a ) we fix @xmath152 and plot @xmath222 $ ] for three different @xmath109 : @xmath143 ( thick dotted - dashed blue line ) , @xmath226 ( dashed black line ) and @xmath227 ( thick dashed red line ) . in this case
, the transition from the mixed instability to the zero wave number instability occurs through the elevation of the local minimum of the dispersion curve , so that a single band of unstable wave numbers occurs starting at @xmath209 . in fig .
[ f4](b ) we fix @xmath228 what corresponds to the level of the cusp point marked by " c in the inset of fig .
[ f2](a ) .
we show @xmath222 $ ] for @xmath143 ( thick dotted - dashed blue line ) , @xmath229 ( dashed black line ) and @xmath230 ( thick dashed red line ) . in this case
, the transition from the mixed instability to the zero wave number instability occurs through the simultaneous elevation of the local minimum and the depression of the local maximum of the dispersion curve . in fig.[f4](c ) we fix @xmath231 and plot @xmath222 $ ] for @xmath143 ( thick dotted - dashed blue line ) , @xmath232 ( dashed black line ) and @xmath233 ( thick dashed red line ) . in this case
, the transition from the mixed instability to the zero wave number instability occurs through the depression of the local maximum of the dispersion curve .
all dispersion curves in fig .
[ f4 ] correspond to real eigenvalues only for a very narrow band of the wave numbers @xmath234 ( details are not shown here ) . by increasing @xmath185 ( decreasing @xmath109 ) . in each panel @xmath156
is fixed and three different values of @xmath185 are chosen .
the thick dotted - dashed ( blue ) lines correspond to @xmath143 , the thick dashed ( red ) lines represent the critical dispersion curve on the border of the mi region in fig .
[ f2 ] at @xmath235 , and the thin dashed ( black ) line is for a value of @xmath109 in between .
the transitions are shown along the lines : ( a ) @xmath152 , ( b ) @xmath228 corresponding to the cusp point c in the inset of fig.[f2](a ) ] , and ( c ) @xmath231 .
[ f4],scaledwidth=95.0% ]
in this section we address the system of nonlinear evolution equations for film hight @xmath210 and probability density @xmath236 , which we give in non - dimensional form in eqs .
( [ nondim ] ) and ( [ eq.several ] ) in appendix [ appendixa ] . in order to solve them numerically ,
we discretize both the film thickness @xmath12 and the density @xmath63 in a square box for the spatial coordinates with @xmath237 ) \times ( y \in [ -l/2,l/2])$ ] and in the interval @xmath238 $ ] for the orientation angle always using periodic boundaries .
we use @xmath239 or @xmath240 mesh points for each spatial direction to discretise space and @xmath241 fourier modes for the decomposition of the @xmath55-dependence of the density .
we adopt a semi - implicit pseudo - spectral method for the time integration , as outlined in the appendix and verify some of our results by using a fully explicit euler scheme with the time step of the order of @xmath242 . in order to quantify the spatio - temporal patterns in film hight and density , we introduce three global measures :
the mode type @xmath27 with @xmath243 characterizes if spatial modulations of the film surface @xmath12 and the average density @xmath244 are predominantly in - phase ( @xmath245 ) or predominantly in anti - phase ( @xmath246 ) ; the space - averaged flux of the fluid @xmath247 determined by @xmath248 -\frac{1}{2 } \left(h^2 { \bm \nabla \langle \rho\rangle}\right)\right)\,dxdy\end{aligned}\ ] ] allows us to distinguish between standing waves that correspond to @xmath249 and travelling or modulated waves that are characterized by a non - zero fluid flux ; and finally the space - averaged translational flux of the swimmers , @xmath250 indicates global surface motion of the swimmers . in the following
we indicate the richness in the dynamics of our system by giving examples of evolving patterns for specific parameter sets located in the stability diagram of fig .
[ f2](a ) .
mapping out a full state diagram is beyond the scope of this article .
we first study the parameters @xmath145 and @xmath149 ( @xmath150 ) in the unstable region of the phase diagram fig.[f2](a ) , close to the finite wave number instability threshold [ point @xmath148 in fig.[f2](a ) ] .
the corresponding dispersion curve is shown in fig.[f2](c ) .
the system size @xmath251 is set to be several times larger than the fastest growing wave length equal to @xmath252 , with @xmath203 denoting the wave number corresponding to the maximum of the dispersion curve . by numerically integrating eqs .
( [ nondim ] ) , we study the temporal evolution of the system from the homogeneous isotropic steady state @xmath110 and @xmath253 , i.e. , the trivial state .
the initial conditions are given by @xmath254 and @xmath255 , where the small amplitude random perturbations @xmath256 and @xmath257 represent two independent sources of white noise . , @xmath149 , @xmath140 , i.e. , @xmath150 [ the corresponding dispersion curve is shown in fig.[f2](c ) ] .
the snapshots are taken at @xmath258 : ( a ) average density @xmath259 ( grey scale map ) with average orientation field ( red arrows ) and ( b ) the film thickness @xmath260 .
@xmath261 is the angle between the main lattice direction and the horizontal axis .
plotted versus time are : ( c ) the mode type @xmath27 from eq.([int_mode ] ) , ( d ) zoom of the region in ( c ) marked by the red rectangle , and ( e ) local film thickness @xmath262 at a randomly chosen point on the surface .
[ f5],title="fig:",scaledwidth=95.0% ] , @xmath149 , @xmath140 , i.e. , @xmath150 [ the corresponding dispersion curve is shown in fig.[f2](c ) ] .
the snapshots are taken at @xmath258 : ( a ) average density @xmath259 ( grey scale map ) with average orientation field ( red arrows ) and ( b ) the film thickness @xmath260 .
@xmath261 is the angle between the main lattice direction and the horizontal axis
. plotted versus time are : ( c ) the mode type @xmath27 from eq.([int_mode ] ) , ( d ) zoom of the region in ( c ) marked by the red rectangle , and ( e ) local film thickness @xmath262 at a randomly chosen point on the surface .
[ f5],title="fig:",scaledwidth=85.0% ] .
shown is a small portion of the square domain in fig.[f5 ] over one oscillation cycle with period @xmath263 .
five snapshots of the film thickness @xmath12 ( in grey scale map ) are plotted together with a contour plot of the average density @xmath264 at the level of @xmath265 .
arrows indicated the shift vector of the pattern during the four phases of one oscillation cycle .
, scaledwidth=98.0% ] we found that after a transient phase of an approximate duration of @xmath266 time units , the system settles onto a stable time - periodic state .
it can be characterised as a regular standing wave , where stripe patterns change periodically between the two diagonal directions as https://www.dropbox.com/s/v4wm4l6a83dniee/v3d5_deff_6d175.mp4?dl=0 [ ] shows .
the mean fluid flux is zero , @xmath267 .
figure.[f5](a ) shows a snapshot of the average density @xmath259 in grey scale map together with the average orientation field @xmath268 shown by red arrows at time @xmath258 . the corresponding snapshot of the film thickness @xmath260 is shown in fig.[f5](b ) .
the patterns in figs.[f5](a ) and ( b ) are highly dynamic , with the shape of the surfaces @xmath210 and @xmath264 changing periodically in time , visualized in https://www.dropbox.com/s/v4wm4l6a83dniee/v3d5_deff_6d175.mp4?dl=0 [ ] and described in fig.[f6 ] .
[ f5](b ) indicates the moment in time when the orientiation of the stripes in the film profile changes from one diagonal to the other .
maxima in the film height @xmath12 occur , which are arranged in a perfect square lattice , tilted by the angle @xmath269 w.r.t .
the @xmath270-axis .
the time evolution of the mode type @xmath27 starting from the initially homogeneous state is plotted in fig .
[ f5](c ) .
after @xmath271 time units @xmath27 starts to oscillate periodically about the average of @xmath272 , as indicated in fig .
[ f5](d ) , where the zoom of the region marked by the rectangle in fig .
[ f5](c ) is shown .
as @xmath273 , the oscillations of the film thickness and the averaged density are in - phase .
the temporal period @xmath31 of the standing wave can be determined by observing the oscillations of the film thickness @xmath262 at a randomly chosen point , as fig.[f5](e ) demonstrates .
thus @xmath262 oscillates about the average film thickness @xmath110 as a perfect periodic function with the period @xmath263 .
remarkably , the temporal oscillations of @xmath27 in fig . [ f5](d ) are four times faster than the oscillations in film thickness .
this is due to the fact that a complete period of the standing wave consists of four phases , where the same spatial pattern reappears four times , each time shifted along one side of a square and rotated by @xmath274 . because @xmath27 is invariant under rotation and translation of the pattern , the oscillation period of @xmath27 is four times smaller then the overall period of the standing wave .
the four shifted and rotated patterns are clearly seen in https://www.dropbox.com/s/v4wm4l6a83dniee/v3d5_deff_6d175.mp4?dl=0 [ ] when concentrating on the cubic lattice formed by the maxima in the height profile . in fig .
[ f6 ] we illustrate the four phases by snapshots of the transient stripe patterns . during the first phase , the maxima of the average density @xmath275 shift along a straight line by a distance @xmath276 equal to half the spatial period @xmath277 .
the maxima of the film thickness follow the same path .
in each subsequent phase , the shift occurs along the direction that is orthogonal to the previous shift . after completing all four phases
, the maxima of @xmath275 and also of @xmath210 will have traveled along the sides of a square with the side length @xmath276 and have returned to the initial position . in between the square patterns , formed by the maxima in density , the film thickness assumes patterns of parallel ridges that during each quarter of the cycle decay into the square pattern , formed by the maxima in @xmath210 , and reappear rotated by @xmath278 .
one may say that during one cycle the pattern oscillates through several accessible patterns that are well known solutions for pattern forming systems on a square . in particular , they are known to occur as ( stable or unstable ) steady states in thin film equations that describe passive liquid layers , ridges and drops on homogeneous solid substrates @xcite .
the spatial period of the standing wave , @xmath277 , can be determined in real space by measuring the distance between two nearest maxima ( minima ) of the height profile @xmath260 taken at an arbitrary moment of time .
the maximal error in this procedure is of the order of @xmath279 where @xmath136 is the number of discretization points along the @xmath270 and @xmath280 axis and the factor @xmath281 reflects that the wave is directed along the diagonal of the domain .
we obtain @xmath282 for the square patterns in fig .
[ f5 ] using @xmath251 and @xmath239 to estimate the error . due to the periodic boundary conditions the measured @xmath277 of the patterns in fig .
[ f5 ] is slightly different from the fastest growing wave length found from the dispersion curve in fig.[f2](b ) as @xmath283 .
this difference is explained as follows . in order to fulfill periodic boundary conditions in a square domain of size @xmath284 , the periods of a wave projected , respectively ,
on the @xmath270 and @xmath280 axis are @xmath285 and @xmath286 , where @xmath28 and @xmath142 are some integers .
this restricts the possible rotation angles @xmath261 of a periodic pattern relative to the @xmath270 axis ( see fig .
they have to satisfy @xmath287 and the wave length of the pattern becomes @xmath288 .
thus , for the parameters used in fig .
[ f5 ] , the random initial conditions select the possible rotation angle @xmath289 .
this choice corresponds to @xmath290 in eq.([psi ] ) .
next , the integer @xmath291 is chosen in such a way that the resulting wave length of the pattern , @xmath292 , is close to the fastest growing wave length of @xmath293 . for later use
, we mention that the spatial period @xmath277 and the angle @xmath261 of a simulated periodic pattern can be determined by computing the time - averaged power spectral density of the film thickness profile @xmath12 according to @xmath294 where @xmath31 is the temporal period of oscillations and @xmath121 denotes the discrete fourier transform of @xmath12 .
the periodic boundary conditions for the square domain only allow for a discrete set of possible wave vectors forming a square lattice with lattice constant @xmath295 .
any periodic pattern in the height modulation @xmath260 gives a major peak of the power spectrum in eq.([spower ] ) , which is located at @xmath296 , @xmath297 , with the same integers @xmath28 and @xmath142 as in eq.([psi ] ) .
depending on the shape of the surface , secondary peaks ( higher harmonics ) might be present , but their strengths are typically orders of magnitude smaller compared to the major peak . in systems of active matter , stable square patterns have previously been found in vicsek - type models with memory .
it was shown that in the case of a ferromagnetic alignment between the self - propelled particles with memory in the orientational ordering the system settles to a perfectly symmetric state with a checkerboard arrangement of clockwise and anti - clockwise vortices @xcite . using our classification
, this checkerboard lattice corresponds to a square pattern with the main axis tilted by @xmath289 w.r.t .
the coordinate axes .
rectangular ( nearly quadratic ) positional order has also been reported as a state of collective dynamics in an active particle model with competing alignment interaction @xcite .
a similar oscillation between stripe and square patterns has been reported for a mesoscopic continuum model for an active filament - molecular motor system where the oscillation is described as alternating wave between aster - like states that form a square lattice and stripe states @xcite .
a related analysis of steady stripe and aster states is presented in @xcite .
other experiments with active matter find hexagonal patterns .
for instance , a hexagonal lattice of vortices was observed in suspensions of highly concentrated spermatozoa of sea urchins @xcite .
phenomenologically , the existence of hexagonal patterns is often studied using a swift - hohenberg ( sh ) equation for scalar fields @xcite . for such model equations
it is known that hexagonal structures can only be stable if the model equations are not invariant under inversion of the scalar field and that higher order gradient terms are needed to stabilize square patterns @xcite . in our case , inversion symmetry is broken , i.e. , eqs.([smoluch ] ) and ( [ thin_film ] ) are not invariant under the simultaneous transformations @xmath298 and @xmath299 . nevertheless ,
in our numerical simulations we did not find stable hexagonal patterns but find that square patterns dominate .
this could imply that higher order terms play an important role .
alternatively it may indicate that a sh equation is not the appropriate order parameter equation for our model that in contrast to standard variational sh equation has no gradient dynamics structure ( see discussion in section [ sec1 ] below eq .
( [ thin_film ] ) ) .
next , we study persisting patterns that emerge from the trivial state for parameters chosen far from the stability threshold .
thus , we set @xmath145 , @xmath300 , as in point @xmath301 in fig.[f2](a ) .
the steady state @xmath110 , @xmath253 is linearly unstable w.r.t . the finite wave number instability with a corresponding dispersion curve ( not shown ) similar to fig.[f2](c ) . in the square domain with side length @xmath251 , we start with the uniform state perturbed by small - amplitude random noise . after a transient of about @xmath241 time units
, a state evolves with an underlying square pattern , as demonstrated below , which is highly dynamic and strongly perturbed by irregular temporal and spatial variations ( see https://www.dropbox.com/s/gipwhnd7xi6d4sm/v3d5_deff_600.mp4?dl=0 [ ] ) . and @xmath302 , i.e. , @xmath300 for a domain size @xmath251 .
the temporal evolution started from the homogeneous state with random noise added .
snapshot is taken at @xmath303 .
[ f8 ] , scaledwidth=95.0% ] snapshots of the pattern in fig .
[ f8 ] show the film thickness @xmath260 and the average density @xmath304 at @xmath303 .
the latter varies between @xmath305 and @xmath306 , in a much larger range than for the regular pattern in fig .
one recognizes the underlying square pattern in the average density @xmath304 but the main lattice directions are tilted against each other .
the height profile @xmath260 looks even stronger perturbed .
still the elevated regions of the film surface ( drops ) are approximately arranged in a square lattice .
https://www.dropbox.com/s/gipwhnd7xi6d4sm/v3d5_deff_600.mp4?dl=0 [ ] shows how the tilted lattice planes in @xmath304 seem to split up and merge with their neighbors . the snapshot in fig .
[ f8 ] shows this scenario when going from left to right .
this gives the whole pattern a highly dynamic appearance . .
( a ) the mode type @xmath27 , ( b ) the spectral density @xmath307 of @xmath308 on the interval @xmath309 $ ] , ( c ) the magnitude of the fluid flux , @xmath310 , ( d ) the magnitude of the translational flux of the swimmers , @xmath311 , and ( e ) the time - averaged power spectral density of the height profile obtained with eq.([spower ] ) by averaging over the interval @xmath312 $ ] .
[ f9],title="fig:",scaledwidth=60.0% ] .
( a ) the mode type @xmath27 , ( b ) the spectral density @xmath307 of @xmath308 on the interval @xmath309 $ ] , ( c ) the magnitude of the fluid flux , @xmath310 , ( d ) the magnitude of the translational flux of the swimmers , @xmath311 , and ( e ) the time - averaged power spectral density of the height profile obtained with eq.([spower ] ) by averaging over the interval @xmath312 $ ] .
[ f9],title="fig:",scaledwidth=39.0% ] the temporal evolution of the pattern is visualized in fig . [ f9](a)-(d ) .
the mode type @xmath27 in plot ( a ) is positive and oscillates randomly about its average value of @xmath313 . in fig .
[ f9](b ) we plot the spectral density @xmath307 of @xmath308 calculated on the interval @xmath309 $ ] .
we find a clear maximum at the frequency @xmath314 with small width @xmath315 , which corresponds to a period of @xmath316 .
the frequency @xmath317 belongs to the pulsating pattern clearly recognizable in https://www.dropbox.com/s/gipwhnd7xi6d4sm/v3d5_deff_600.mp4?dl=0 [ ] . a second ,
broader peak is located at @xmath318 with width @xmath319 .
in addition , there exists a continuous background in @xmath307 , which gives the pattern its random dynamic appearance .
random oscillations of the magnitude of the fluid flux , @xmath310 , [ see fig . [ f9](c ) ] and of the magnitude of the translational flux of the swimmers , @xmath311 , [ see fig . [ f9](d ) ] , indicate global propagation of the pattern at each instance of time .
however , we find that the propagation direction randomly changes with time with no preferred direction as expected for square symmetry .
finally , to reveal the periodic structure of the pattern , we determined the time - averaged power spectral density from eq .
( [ spower ] ) averaged over the time interval @xmath312 $ ] .
as shown in fig .
[ f9](e ) , the spectral density has two major broad peaks : one is centered around @xmath320 , i.e. @xmath321 , @xmath209 , and the other one is centered around @xmath322 , i.e. @xmath323 and @xmath324 .
these peaks correspond to a square pattern with the main lattice directions aligned along the coordinate axes , i.e. @xmath325 or @xmath326 .
the dominating wave length or lattice constant @xmath277 of the pattern is @xmath327 .
the third peak with much less intensity at @xmath328 , i.e. @xmath291 and @xmath329 , can roughly be interpreted as a contribution from the sum of the two major wave vectors spanning the reciprocal lattice . in order to systematically study the occurence and stability of the two patterns studied in the previous sections , we follow these patterns in parameter space using a primitive `` continuation method '' .
namely , we take a snapshot of a converged ( time - dependent ) state at some parameter value and use it as initial condition for simulating the evolving pattern in a neighboring point in parameter space .
the technique allows us to follow states , which are linearly stable , and thereby identify multistability in parameter space .
depending on the initial condition different stable spatio - temporal stable patterns are obtained . for an overview of proper continuation methods , which are also able to follow unstable steady states and therefore to determine the complete bifurcation diagram , see refs .
. however , these methods are not readily available for time - periodic solutions of our pde system . during the continuation of the standing waves at fixed @xmath145 and @xmath251 .
standing waves that emerged from the homogeneous steady state at @xmath145 and @xmath150 are numerically continued by increasing @xmath109 .
the vertical lines mark times when @xmath109 is changed .
the respective values of @xmath109 for each interval are given by the numbers in each panel .
the thick red solid line shows the mode type , averaged over one oscillation period .
( b , c ) continuation of patterns with intermittent symmetry , emerged from the homogeneous steady state at @xmath145 and @xmath300 with @xmath251 . shown
are ( b ) the mode number @xmath27 and ( c ) the modulus of the fluid flux @xmath330 in dependence of time . the time axis is reversed .
[ f10],scaledwidth=95.0% ] first , we start with the regular standing wave pattern , which we explored in sec . [ subsec.standing ] and in fig .
[ f5 ] at fixed @xmath145 .
we follow the standing wave solution along the line connecting points @xmath148 and @xmath331 in fig .
[ f2](a ) by gradually decreasing @xmath185 ( or increasing @xmath109 ) in steps using four distinct values , namely @xmath332 . at each parameter point , we let the system settle into a stable state , which we identify by monotoring mode type @xmath308 and fluid flux @xmath333 . the resulting time evolution of mode type @xmath27 during the continuation schedule is shown in fig.[f10](a ) .
we find that the standing wave pattern keeps its main characteristics up to the largest value @xmath300 [ point @xmath331 in fig.[f2](a ) ] .
in particular , mode type @xmath27 shows regular oscillations .
the mean value of @xmath27 [ red solid line in fig.[f10](a ) ] and the oscillation amplitude of @xmath27 increase with @xmath109 .
they reach their respective maximal values of @xmath334 and @xmath335 at around @xmath336 , where the density and height variations are more pronounced compared to fig .
furthermore , the oscillation period monotonically increases with @xmath109 ( not shown ) .
interestingly , the spatial period of the pattern remains unchanged during the entire continuation schedule .
next , we start with the strongly perturbed square pattern from sec
. [ sucsec.strongly ] and fig.[f8 ] and follow the same path in fig.[f2](a ) but this time backward from point 4 to point 2 . for consistency , we use the same values of @xmath109 as in fig.[f10](a ) .
the time evolution of the mode type @xmath27 and the modulus of the fluid flux @xmath337 are plotted in figs .
[ f10](b ) and ( c ) with reversed time axis .
remarkably , the system reaches the regular standing wave pattern only for parameters close to the stability threshold , i.e. , when @xmath109 is decreased to the value in point @xmath148 in fig.[f2](a ) .
after some transient dynamics , visible in the time intervall from @xmath338 to 1100 in figs.[f10](b ) and ( c ) , the system settles on the stable standing wave pattern . for larger values of @xmath109 ,
i.e. , further away from the threshold , we found that the system is multistable . at @xmath339
stable regular standing waves still exist but also patterns similar to the strongly perturbed square pattern as shown in fig.[f8 ] are stable .
however , at @xmath340 , the dynamics of the square pattern becomes more regular but still keeps the feature of lattice planes splitting and merging with their neighbors .
this is demonstrated by https://www.dropbox.com/s/rlat7yr5fwtx8au/v3d5_deff_20.mp4?dl=0 [ ] and by the snapshots in fig .
[ f11](a ) and ( b ) , taken at @xmath338 during the numerical continuation in fig .
[ f10 ] . and
@xmath340 , during the numerical continuation of the patterns with intermittent symmetry in fig.[f10 ] .
snapshot taken at @xmath338 .
( c ) the components of the translational surface flux @xmath341 .
( d ) the spectral density of @xmath342 .
( e ) time - averaged power spectral density of the pattern , obtained with eq.([spower ] ) by averaging over the interval @xmath343 $ ] .
[ f11],scaledwidth=95.0% ] and @xmath340 , during the numerical continuation of the patterns with intermittent symmetry in fig.[f10 ] .
snapshot taken at @xmath338 .
( c ) the components of the translational surface flux @xmath341 .
( d ) the spectral density of @xmath342 .
( e ) time - averaged power spectral density of the pattern , obtained with eq.([spower ] ) by averaging over the interval @xmath343 $ ] .
[ f11],title="fig:",scaledwidth=55.0% ] and @xmath340 , during the numerical continuation of the patterns with intermittent symmetry in fig.[f10 ] .
snapshot taken at @xmath338 .
( c ) the components of the translational surface flux @xmath341 .
( d ) the spectral density of @xmath342 .
( e ) time - averaged power spectral density of the pattern , obtained with eq.([spower ] ) by averaging over the interval @xmath343 $ ] .
[ f11],title="fig:",scaledwidth=30.0% ] figure [ f11](d ) shows the @xmath270 and @xmath280 components of the translational flux of the swimmers .
one clearly recognizes directed motion , on average , into the negative @xmath270 and positive @xmath280 direction , which is also visible in https://www.dropbox.com/s/rlat7yr5fwtx8au/v3d5_deff_20.mp4?dl=0 [ ] .
the components of the fluid flux @xmath344 behave similarly .
furthermore , both flux components show a fast oscillation with a weak slow modulation superimposed . by taking the fourier transforms of @xmath345 [ fig .
[ f11](d ) ] , one identifies a dominant peak at @xmath346 corresponding to a period of @xmath347 of the fast oscillations .
they result from the pulsation in the square pattern as https://www.dropbox.com/s/rlat7yr5fwtx8au/v3d5_deff_20.mp4?dl=0 [ ] demonstrates , in particular , for the height profile .
the weak modulation generates a small peak in the power spectrum with frequency @xmath348 or period @xmath349 .
it is not really recognizable in the time evolution of https://www.dropbox.com/s/rlat7yr5fwtx8au/v3d5_deff_20.mp4?dl=0 [ ] .
otherwise , the continuous part of the spectrum as observed in fig .
[ f9](b ) for the strongly perturbed square pattern is missing here since the square pattern has a more regular dynamics .
finally , the time - averaged power spectral density @xmath350 , averaged over the interval @xmath343 $ ] , is given in fig .
[ f11](e ) .
the two major peaks correspond to the square pattern with spatial period of @xmath351 aligned along the coordinate axes .
the third , much weaker peak at @xmath352 with @xmath321 again roughly corresponds to a contribution of the two major wave vectors spanning the reciprocal lattice .
multistability of several persisting dynamic states under identical external conditions was also found in other systems .
for example , in experiments on groups of schooling fish @xcite it was observed that depending on the starting conditions and/or the nature of perturbations , as well as the group size , the fish group may exhibit two different dynamic states : the so - called milling state , which is characterized by fish swimming in a large circle , and the polarised state , which corresponds to fish swimming predominantly in one direction . , @xmath153 , ( @xmath154 ) , @xmath251 [ see point @xmath151 in fig . [ f2](a ) ] .
the corresponding dispersion curve is shown in fig .
[ f2](d ) .
( c ) mode type @xmath27 , ( d ) @xmath270 and @xmath280 component of the space averaged fluid flux @xmath247 , ( e ) the film height @xmath262 at a randomly chosen point on the film surface , and ( f ) its power spectrum @xmath350 .
( g ) power spectrum for the spatial modulation of the height profile [ see eq .
( [ spower ] ) ] averaged over the interval @xmath353 $ ] .
[ f12],title="fig:",scaledwidth=95.0% ] , @xmath153 , ( @xmath154 ) , @xmath251 [ see point @xmath151 in fig . [ f2](a ) ] .
the corresponding dispersion curve is shown in fig . [ f2](d ) .
( c ) mode type @xmath27 , ( d ) @xmath270 and @xmath280 component of the space averaged fluid flux @xmath247 , ( e ) the film height @xmath262 at a randomly chosen point on the film surface , and ( f ) its power spectrum @xmath350 .
( g ) power spectrum for the spatial modulation of the height profile [ see eq .
( [ spower ] ) ] averaged over the interval @xmath353 $ ] .
[ f12],title="fig:",scaledwidth=50.0% ] , @xmath153 , ( @xmath154 ) , @xmath251 [ see point @xmath151 in fig . [ f2](a ) ] .
the corresponding dispersion curve is shown in fig . [ f2](d ) .
( c ) mode type @xmath27 , ( d ) @xmath270 and @xmath280 component of the space averaged fluid flux @xmath247 , ( e ) the film height @xmath262 at a randomly chosen point on the film surface , and ( f ) its power spectrum @xmath350 .
( g ) power spectrum for the spatial modulation of the height profile [ see eq .
( [ spower ] ) ] averaged over the interval @xmath353 $ ] .
[ f12],title="fig:",scaledwidth=40.0% ] in the triangular parameter region in fig .
[ f2 ] , where the mixed instability occurs , one finds persisting patterns , i.e. , long - time stable spatio - temporal patterns that are characterized by a time - independent mode type and a constant non - zero fluid flux , i.e. , travelling waves .
figures [ f12](a ) and ( b ) give example snapshots for such a pattern obtained at point @xmath151 in fig.[f2](a ) at system size @xmath251 and at time @xmath354 .
[ ] reveals a pattern traveling approximately along the diagonal with a propagation speed estimated to be of the order of the self - propulsion velocity @xmath152 .
the maxima in the height profile form a rectangular lattice .
perturbations run along the lattice lines and shift the maxima by roughly half a lattice constant ; presumably , to match the periodic boundary condition for the whole square domain .
the time evolutions of mode type @xmath27 and of the components of the fluid flux , @xmath355 and @xmath356 , are shown in fig .
[ f12](c ) and ( d ) , respectively .
the height modulation at an arbitrary point plotted in fig .
[ f12](e ) looks rather irregular .
its power spectrum in fig .
[ f12](f ) reveals one peak at @xmath357 , which corresponds to one shift motion of a bump in the height profile .
frequencies at @xmath358 and @xmath359 belong to longer cycles of two or three shifts .
the power spectral density for the spatial modulations in the film height , averaged over the interval @xmath353 $ ] , is plotted in fig .
[ f12](g ) . in the upper and the lower halves of the snapshot in fig .
[ f12](b ) one can clearly see a rectangular pattern with the aspect ratio of @xmath360 .
remarkably , the aspect ratio of @xmath360 of the rectangular pattern is not compatible with the periodic boundary conditions of the square domain .
nontheless , the pattern is fitted into the square domain due to the presence of a defect - like modulation , seen at the center of the snapshot .
the defect disturbes the rectangular lattice dynamically , as it continuously moves along the lattice lines and shifts the elevations in the height profile ( see https://www.dropbox.com/s/tba6mi7rjvfkjlo/v2_deff_0d88.mp4?dl=0 [ ] ) . as a result , in the power spectrum in fig .
[ f12](g ) we find two major peaks located at @xmath361 and @xmath362 and one smaller peak located at @xmath363 .
the remaining two weaker peaks are again higher contributions from wave vectors in the reciprocal lattice .
persisting states characterized by propagating structures are well known for active matter systems .
thus , traveling density waves were found in experiments with an assay of actin filaments , driven by motor proteins @xcite . at the leading edge ( lamellipodium ) of a crawling cell , the alignment of actin filaments along the substrate leads to the forward translation of lamellipodium and thus , to cell motility @xcite . moving density stripes and propagating isolated density clusters
have been found in microscopic vicsek - type models and in continuum models of self - propelled particles @xcite . and @xmath364 in a system of size @xmath365 .
number of the fourier modes for the discretisation in space is @xmath240 .
[ f13],scaledwidth=95.0% ] when studying the nonlinear behaviour for @xmath145 and @xmath366 [ point @xmath144 in fig.[f2](a ) ] , we find truly random patterns emerging from the zero wave - number instability .
the corresponding dispersion curve is shown in fig.[f2](b ) and gives the fastest growing wave length as @xmath367 .
we set @xmath365 , @xmath240 and start the simulations at the trivial state . after a transient
, the system settles to an irregular pattern in space and time that oscillates randomly and locally travels in random directions .
typical snapshots are shown in fig .
[ f13 ] and https://www.dropbox.com/s/qb4mkxrx7gpzdjc/v3d5_deff_0d66.mp4?dl=0 [ ] illustrates the irregular spatio - temporal pattern . )
computed numerically by averaging the fourier transformed film thickness patterns in fig.[f13 ] over @xmath368 $ ] .
( b , c ) the mode type @xmath27 and the modulus of the fluid flux @xmath369 over time .
( d ) time evolution of @xmath262 at a randomly chosen point on the film surface .
( e ) spectral density @xmath350 of @xmath262 from ( d ) .
( f ) normalised radial autocorrelation function @xmath370 .
inset shows zoomed graph of @xmath370 around zero .
[ f14],title="fig:",scaledwidth=30.0% ] ) computed numerically by averaging the fourier transformed film thickness patterns in fig.[f13 ] over @xmath368 $ ] .
( b , c ) the mode type @xmath27 and the modulus of the fluid flux @xmath369 over time .
( d ) time evolution of @xmath262 at a randomly chosen point on the film surface .
( e ) spectral density @xmath350 of @xmath262 from ( d ) .
( f ) normalised radial autocorrelation function @xmath370 .
inset shows zoomed graph of @xmath370 around zero .
[ f14],title="fig:",scaledwidth=60.0% ] the randomness of the pattern is clearly visible in the time evolution of mode type @xmath27 and of the modulus of the fluid flux @xmath369 , which we plot in figs .
[ f14](a ) and ( b ) , respectively . in fig .
[ f14](a ) the power spectral density of the height profile averaged over the interval @xmath371 $ ] reveals a clear maximum at @xmath372 .
the spectral density is radially symmetric , which implies that spatial correlations in the pattern only depend on the distance @xmath373 between any two points on the film surface .
otherwise , it is continuous as expected for a random pattern . via the convolution theorem ,
the time - averaged power spectral density from eq .
( [ spower ] ) is directly related to the spatial height correlation function @xmath374[h({\bm w})-1]\,d{\bm w},\end{aligned}\ ] ] where @xmath375 and @xmath376 and the constant @xmath377 is chosen such that @xmath378 .
we plot the normalised radially - symmetric correlation function @xmath370 in fig .
[ f14](f ) over the length of half the system size , i.e. , @xmath379 . @xmath370 rapidly decreases with @xmath373 and drops by two orders of magnitude over the distance of @xmath380 , as shown in the inset of fig .
[ f14](f ) .
correlations become negligibly small at distances larger than 10 and the pattern looks random .
the oscillation at small distances corresponds to the maximum in the power spectral density .
they are caused by wave fronts traveling in random directions , which one recognizes in the orientation - averaged swimmer density in https://www.dropbox.com/s/qb4mkxrx7gpzdjc/v3d5_deff_0d66.mp4?dl=0 [ ] . finally , the film height at an arbitrary position changes randomly in time , as shown in fig .
[ f14](d ) . however , the spectral density @xmath350 plotted in fig .
[ f14](e ) has a clear peak at @xmath381 .
this implies that the temporal dynamics of the patterns can not be regarded as purely random .
highly dynamic random spatio - temporal patterns of active matter are known as quasi or mesoscale turbulence .
irregular turbulent states have been found in experiments with dense bacterial suspensions @xcite and in active microtubuli networks@xcite .
we have investigated the collective behaviour of a colony of point - like non - interacting self - propelled particles ( microswimmers ) that swim at the free surface of a thin liquid layer on a solid support .
in contrast to former work @xcite , where the motion of the particles was considered to be purely orthogonal to the free surface , here we have also allowed for active motion parallel to the film surface .
the resulting coupled dynamics of the swimmer density @xmath63 and the film thickness profile @xmath12 is captured in a long - wave model in the form of a smoluchowski equation for the one - particle density @xmath63 and a thin film equation for @xmath12 that allows for ( i ) diffusive and convective transport of the swimmers ( including rotational diffusion ) , ( ii ) capillarity effects ( laplace pressure ) including a marangoni force caused by gradients in the swimmer density , ( iii ) and a vertical pushing force of the swimmers that acts onto the liquid - gas interface .
first , we have extended the linear stability analysis of the homogeneous and isotropic state that was presented before in ref .
@xcite focusing , in particular , on the characteristics of the two instability modes ( one at zero wave number and one at finite wave number ) and their mixed appearance close to the border of the stable region in the stability diagram spanned by the swimmer speed @xmath156 and the effective diffusion constant @xmath382 . our linear stability analysis indicates that the onset and dispersion relation of the zero - wave number instability mode do not fit well into the classification scheme of cross and hohenberg @xcite .
the long - wave instability of the free film surface occurs at @xmath209 , where the zero - wave number mode has zero imaginary part .
however , arbitrarily close to onset in the unstable region , the fastest growing wave number @xmath203 corresponds to a pair of complex conjugate eigenvalues .
moreover , the entire unstable band of wave numbers , @xmath383 , does _ always _ contain a range of small @xmath384 ( with @xmath385 measuring the distance from the stability threshold ) , where the first two leading eigenvalues are real ( one @xmath386 and one @xmath387 ) , and a range @xmath388 , where the two leading eigenvalues form a complex conjugate pair .
the latter range always contains the fastest growing wave number @xmath389 .
this indicates that this zero - wave number instability is similar to a zero - frequency hopf bifurcation in dynamical systems @xcite and in the context of the instabilities of spatially extended systems , it might be called a zero - frequency type @xmath390 instability .
the behaviour at about @xmath209 has also important implications for a weakly nonlinear theory for the short - wave instability at @xmath391 .
such a theory would need to take into account that the slow complex modes around @xmath391 couple to the two unstable long - wave modes at @xmath392 with real eigenvalues .
we believe that such a coupling is responsible for the observed wave behaviour , where a travelling wave is perturbed by a long - wave modulation , as in fig.[f11 ] .
the two long - wave modes are a direct consequence of the existing two conserved quantities in the system : the mean film height and the orientation - averaged mean swimmer concentration .
simpler cases with one long - wave mode ( resulting from a single conserved quantity ) that couples to a short - wave mode have been considered in refs .
such an analysis is not feasible in our case , where the evolution equations capture the dynamics in two spatial dimensions _ and _ account for a fully @xmath55-dependent density @xmath63 .
however , a one - dimensional model system , where instead of rotational diffusion the swimmers can only flip between swimming to the left or right shows similar transitions and lends itself to a weakly nonlinear analysis .
such a simplified system is under investigation and will be presented elsewhere .
numerical simulations of the time evolution equations ( [ smoluch ] ) and ( [ thin_film ] ) reveal a rich variety of persisting dynamic states .
we have abstained from a comprehensive parameter study of the different persisting states .
instead , we have given an overview of the zoo of dynamic states that can be found , when solving the system of equations ( [ smoluch ] ) and ( [ thin_film ] ) .
in particular , for parameters chosen in the vicinity of the stability threshold of the finite wave number instability , we have found a highly regular dynamic standing wave pattern by starting the simulations with small random perturbations of the trivial state .
the standing wave pattern is characterised by a regular array of elevations of the free film surface that periodically transform and rearrange following a rather complex pathway .
thus , over one quarter of the oscillation cycle , the elevations transform from a perfect square lattice into an array of stripes followed by a new square lattice of elevations that is shifted w.r.t . the initial square lattice by exactly one half of the spatial wave period .
the transformation of the average density of swimmers follows the pattern of the film thickness profile .
the swimmers are arranged in a regular lattice with high and low density spots .
spatial variations of the film thickness profile and the orientation - averaged density profile are in - phase implying that high density spots sit approximately on top of the droplets .
the orientation of swimmers in each high density spot shows strong polar order with a hedgehog defect right at the maximum .
the space - averaged fluid flux of the square wave is zero at all times .
next , we have employed a primitive continuation method and followed the standing wave patterns through parameter space moving further away from the stability threshold .
our numerical results suggest that standing wave patterns exist and are stable possibly in the entire region , where the homogeneous state is linearly unstable w.r.t . the finite wave number instability ( i.e. for @xmath393 ) . on the reverse path , initial random perturbations of the homogeneous and isotropic state
develop into a strongly perturbed square pattern , where lattice lines continuously split and merge with their neighbors .
unlike for the regular standing wave pattern , the average fluid flux oscillates randomly about zero . however , the time - averaged power spectral density of the height profile reveals a clear square pattern .
further decreasing @xmath109 towards the stability threshold , the dynamics of the perturbed square lattice becomes regular .
the fluid flux in this state is periodic in time with a weak modulation superimposed .
the major frequency corresponds to a pulsation of the square pattern . by choosing the parameters in the mixed - instability region , we find a persisting traveling pattern characterised by constant space - averaged fluid flux .
elevations in the film surface are arranged in a rectangular lattice that travels in one direction with the speed of the order of the self - propulsion velocity , while perturbations continuously shift the hight elevations . at their positions
the swimmers form high - density spots with strong polar order around a hedgehog defect similar to the regular standing wave .
finally , choosing parameters from the region with the zero wave number instability , one finds a random spatio - temporal pattern with a correlation length much smaller than the system size .
our findings clearly show that a rich variety of persisting regular and irregular dynamic states can be found in an active matter system of self - propelled particles without direct interactions . in our model ,
the interaction between the swimmers occurs on a coarse - grained level and is mediated by large - scale deformations of the liquid film .
similar types of dynamic states found here were previously observed in other active matter systems of interacting particles as indicated at the respective ends of sections [ subsec.standing ] to [ patt - random ] .
for instance , stable square patterns were found in vicsek - type models with memory @xcite .
multistability of the system under identical external conditions was reported earlier in experiments with groups of schooling fish @xcite .
various traveling states occurred in experiments with motility assays of actin filaments driven by motor proteins @xcite , in microscopic vicsek - type models , and in continuum models of self - propelled particles @xcite .
finally , random or turbulent states were observed in experiments with dense bacterial suspensions @xcite and in active microtubule networks @xcite .
it is fascinating that active particles acting as surfactants at the surface of a thin liquid film provide a model system , where all these different dynamic patterns can be realized by tuning appropriate parameters .
possible extensions of the model include the incorporation of wettability effects by adding the derjaguin ( disjoining ) pressure to study swimmer carpets not only on films but also on shallow droplets and , in particular , interactions with ( moving ) contact lines .
one may also go beyond the approximation of point - like non - interacting particles by introducing finite size effects ( short - range interactions between particles ) as well as long - range ( hydrodynamic ) interactions .
the resulting smoluchowski equation would then contain non - local terms as in dynamical density functional theories for the diffusive dynamics of interacting colloids , polymers , and macromolecules @xcite .
for a consistent model also the film height equation would require additional terms that may be determined via the gradient dynamics formulation @xcite that has to be recovered in the limit of passive surfactant particles / molecules .
in the employed dimensionless quantities , the resulting coupled system consists of the reduced smoluchowski equation and the thin film equation .
it reads @xmath394 with the dimensionless fluid flux @xmath395 , the translational and rotational probability currents @xmath66 and @xmath396 , the surface fluid velocity @xmath397 and the vorticity of the fluid flow @xmath83 @xmath398 -\frac{1}{2}\left(h^2 { \bm \nabla \langle \rho\rangle}\right),\nonumber\\ { \bm j}_{\rm
t}&=&\left(v{\bm q}+{\bm u}_\parallel -d{\bm \nabla}\right)\rho,\nonumber \\ j_\phi&=&\frac{1}{2}\omega_z\rho- d \partial_\phi \rho,\nonumber\\ { \bm u}_\parallel&=&-h{\bm \nabla}\langle \rho\rangle + \frac{h^2}{2}{\bm \nabla}\left ( \delta h + \beta\langle \rho \rangle \right),\nonumber\\ \omega_z&=&\partial_x u_y -\partial_y u_x . \label{eq.several}\end{aligned}\ ] ] the coupled eqs.([nondim ] ) are solved numerically using the following version of the semi - implicit spectral method .
first , we average the density equation over the orientation angle @xmath55 .
this yields @xmath399 with the average translational current @xmath400 and @xmath401 .
it is worthwhile noticing that the only term in eq.([ap_eq1 ] ) that depends on the three - dimensional density @xmath63 is the average orientation vector @xmath402 .
all other terms in eq.([ap_eq1 ] ) , including the surface fluid velocity @xmath403 explicitly depend on the average density @xmath275 . at the next step ,
we single out the linear parts in all the terms in eqs.([ap_eq2 ] ) that explicitly depend on the average density @xmath275 .
this is done by linearising the current @xmath66 and the fluid flux @xmath395 about the trivial steady state given by @xmath110 and @xmath406 . finally ,
following the standard implicit time - integration scheme , we replace @xmath407 and @xmath408 by @xmath409 and by @xmath410 , respectively and take all linear terms at time @xmath411 and all nonlinear terms , including the term @xmath412 , at time @xmath413 . upon these transformations eqs.([ap_eq2 ] ) become @xmath414 with the nonlinear parts given by @xmath415 -\frac{1}{2 } { \bm \nabla}\left([h^2 - 1 ] { \bm \nabla \langle \rho\rangle}\right)\right]^t,\nonumber \\ ( \langle { \bm nl}_{\rm trans}\rangle)^t & = & \left[v \langle{\bm q}\rho\rangle + ( { \bm u})\langle \rho \rangle -\frac{1}{2}{\bm \nabla}(\delta h)-\left(\frac{\beta}{2}-1\right){\bm
\nabla}\langle \rho \rangle\right]^t.\end{aligned}\ ] ] after taking the discrete fourier transforms of eqs.([ap_eq3 ] ) , we find the updated fields @xmath416 and @xmath417 at the time step @xmath411 . with the update average density @xmath417 and the film thickness @xmath416 at hand , we find the updated surface fluid velocity @xmath418 and the updated vorticity @xmath419 .
these functions are then substituted into the three - dimensional density equation @xmath420 with the translational current @xmath421 and the rotational current @xmath422 . | we consider a colony of point - like self - propelled surfactant particles ( swimmers ) without direct interactions that cover a thin liquid layer on a solid support .
although the particles predominantly swim normal to the free film surface , their motion also has a component parallel to the film surface .
the coupled dynamics of the swimmer density and film height profile is captured in a long - wave model allowing for diffusive and convective transport of the swimmers ( including rotational diffusion ) .
the dynamics of the film height profile is determined by three physical effects : the upward pushing force of the swimmers onto the liquid - gas interface that always destabilizes the flat film , the solutal marangoni force due to gradients in the swimmer concentration that always acts stabilising , and finally the rotational diffusion of the swimmers together with the in - plance active motion that acts either stabilising or destabilising . after reviewing and extending the analysis of the linear stability of the flat film with uniform swimmer density , we analyse the full nonlinear dynamic equations and show that point - like swimmers , which only interact via long - wave deformations of the liquid film , self - organise in highly regular ( standing , travelling and modulated waves ) and various irregular patterns for swimmer density and film height . |
quantitative photoacoustic tomography is concerned with recovering quantitatively accurate estimates of chromophore concentration distributions , or related quantities such as optical coefficients or blood oxygenation , from photoacoustic images @xcite .
the source of contrast in photoacoustic tomography ( pat ) is optical absorption , which is directly related to the tissue constituents . by obtaining pat images at multiple optical wavelengths
, it may be possible to recover chemically specific information about the tissue .
however , such a spectroscopic use of pat images must consider the effect of the spatially and spectrally varying light fluence distribution . as
a photoacoustic image is the product of the optical absorption coefficient distribution , which carries information about the tissue constituents , and the optical fluence , which only acts to distort that information , the challenge in quantitative photoacoustic imaging is to remove the effect of the light fluence .
a common approach is to use a model of the unknown fluence and use it to extract the desired optical properties from the measured data .
this has been done analytically @xcite or numerically @xcite , often within a minimisation framework @xcite .
the majority of this literature uses the diffusion approximation to the radiative transfer equation to model the light distribution , which is accurate in highly scattering media and away from boundaries or sources @xcite . in pat ,
the region of interest often lies close to the tissue surface where the diffusion approximation is not accurate .
the radiative transfer equation ( rte ) , on the other hand , is widely considered to be an accurate model of light transport so long as coherent effects are negligible , which is the case here .
finite element discretisations of the rte have been developed @xcite and proposed for quantitative pat reconstructions @xcite , but due to the need to discretise in angle as well as space they quickly become computationally intensive and their applicability is limited to small and medium scale problems .
an alternative is monte carlo ( mc ) modelling @xcite , which is a stochastic technique for modelling light transport that converges to the solution to the rte .
the significant advantage of the mc approach is that it is highly parallelisable so scales well to the large - scale inversions that will be encountered in practice .
monte carlo models of light transport are popular in biomedical optics and have predominantly been applied in the planning of experimental measurements @xcite and in dosimetric studies for a range of light based therapies @xcite .
many of the applications are summarised by zhu et al .
one early mc model of light transport , mcml @xcite , computes the fluence in 3d slab geometry .
this model was later extended to simulate spherical inclusions in the tissue@xcite , and later to spheroidal and cylindrical @xcite inclusions .
mc modelling in 3d heterogeneous media has been shown both for voxelised media @xcite , which was later gpu - accelerated @xcite , and using a mesh - based geometry @xcite . although the rte is an equation for the radiance , which is a function of angle at every point , the quantity usually calculated by mc models is the fluence rate , which is the radiance integrated over all angles .
the reasons are practical : most measurable quantities are related to the fluence rate rather than the radiance , storing just the integrated quantity saves on computational memory , and the estimates for the fluence rate will converge sooner than the underlying estimates for the radiance . in photoacoustics ,
the measurable signal is related to the fluence ( the time - integrated fluence rate ) so current mc models can be used in the simulation of photoacoustic signals . however , as will be discussed in section 4 , the full angle - dependent radiance is required when tackling the inverse problem of estimating the optical coefficients , specifically the optical scattering
. in this paper , section [ sec : qpat ] introduces the inverse problem of quantitative pat .
sections [ sec : light_transport ] and [ sec : adjoint ] present forward and adjoint monte carlo models of the radiance employing a harmonic angular basis . in section
[ sec : gradients ] it is shown that this choice of basis allows the functional gradients for the inverse problem to be calculated straightforwardly .
inversions for absorption and scattering coefficient distributions are given in section [ sec : examples ] .
the inverse problem in qpat can be stated as the minimisation @xmath0 where the error functional is given by @xmath1 @xmath2 is the absorbed energy density and is the ` data ' for this problem .
it is related to the photoacoustic image by the grneisen parameter , which here is set to 1 .
additional regularisation terms or terms reflecting prior knowledge may also be added to @xmath3 .
gradient - based approaches to solving this problem require estimates of the gradients of the error functional with respect to the parameters of interest .
saratoon et al .
@xcite gives expressions for these gradients in terms of the forward and adjoint fields , @xmath4 and @xmath5 : @xmath6 and @xmath7 monte carlo models to calculate the radiance @xmath8 and adjoint radiance @xmath9 are given in the following two sections .
in pat , the optical and acoustic propagation times are so different that the optical propagation can be considered instantaneous and the time - dependence of the light transport can be neglected . the time - independent radiative transfer equation ( rte )
is given by @xmath10 where @xmath11 is the radiance , @xmath12 and @xmath13 are the absorption and scattering coefficients , respectively , @xmath14 is position , @xmath15 and @xmath16 are the original and scattered propagation directions , @xmath17 is the scattering phase function , @xmath18 is a source term and @xmath19 is used to indicate integration over angle in @xmath20 dimensions . to obtain approximations to the solutions to this equation , various flavours of mc
have been proposed @xcite .
the approach used here begins with launching a packet of energy , referred to herein as a ` photon ' , from a given position @xmath14 in an initial direction @xmath16 . after travelling a distance @xmath21 ) / { { \mu_{s}}}$ ] ( using the convention @xmath22 ) , where @xmath23)$ ] is a real uniform random variable on @xmath24 $ ] , a fraction of photon s ` weight ' @xmath25 is deposited in current voxel , where @xmath26 is the current weight ( or energy ) of the photon packet .
the photon weight is updated : @xmath27 .
scattering into a new direction @xmath15 in 2d involves sampling the scattering phase function , which describes the probability of a photon scattering from direction @xmath15 into direction @xmath16 .
the phase function used here was the 2d henyey - greenstein phase function , commonly used in biomedical optics @xcite , @xmath28 the parameter , @xmath29 , a property of the medium , is known as the anisotropy factor . sampling this equation for the scattering angle , @xmath30 , by solving for @xmath31 in the cumulative integral over angle yields @xmath32))\right ) .
\label{scattering_angle}\ ] ] a new step length , @xmath33 , is sampled and this process is repeated until the photon weight falls below some threshold value . by carrying out the above computation for many photons and adding the voxel weights will , for a sufficient number of photons , converge on a solution to the rte . by calculating photon paths through the medium
, the mc models presented in the literature @xcite do in fact simulate the radiance , but this typically integrated over angle upon deposition of the weights in the voxels . in order to simulate the radiance , a method of depositing the weight in the voxels without losing the angular information
is required . in order to compute the radiance using a mc model , angular as well as spatial discretisation
is required .
one approach is to use discrete ordinates , whereby the unit circle is divided equally into sectors and the weight deposited in a voxel is also assigned to the relevant angular sector .
the memory required will scale linearly with the number of sectors , and will slow convergence of the radiance estimate , compared with the fluence estimate , by a factor inversely related to the number of sectors . here , a harmonic angular basis was used because a sufficiently diffuse field is dense in such a basis , meaning the field can be represented using relatively few orders .
less memory will therefore be required . in 2d ,
the expansion for the radiance in a fourier basis is @xcite : @xmath34 where @xmath35 and @xmath36 are the coefficients associated with each harmonic and @xmath37 $ ] and is the angle of the photon direction @xmath16 relative to the z - direction ( i.e. @xmath38 ) .
( the equivalent expansion in 3d would be into spherical harmonics @xcite . ) for a given voxel , the weight is deposited into the relevant fourier coefficients according to @xmath39 where @xmath40 is the weight deposited by the @xmath41 photon traversing the @xmath42^th^ voxel . the algorithm was implemented in the julia programming language @xcite .
analytical solutions to the rte are available for the fluence for a range of geometries and source types @xcite , however there are few analytical solutions for the radiance , particularly in 2d . the rmc model was compared to one such analytic solution for an infinite , homogeneous 2d domain illuminated by an isotropic point source @xcite .
an isotropic point source was placed at the centre of a domain of size 15mm@xmath4315 mm , large compared to the transport mean free path in order to approximate an infinite domain .
the absorption and scattering coefficients were 0.01mm^-1^ and 10mm^-1^ respectively and the henyey - greenstein phase function @xcite was used used with @xmath29 set to 0.9 .
the pixel size was 0.05 mm @xmath43 0.05 mm , and 5 fourier harmonics were used .
[ fig : rmc_validation ] shows the good agreement between the analytical and rmc modelled radiance at radial distances of 2 mm and 3 mm from the source along the horizontal axis .
mm^-1^ , @xmath44mm^-1^ and @xmath45 .
results from an analytic method ( infinite domain ) and rmc simulations ( 15mm@xmath4315 mm square domain ) shown.,scaledwidth=100.0% ]
the adjoint equation to the rte is given by @xmath46 where @xmath47 is the adjoint radiance and @xmath48 is the adjoint source .
this was implemented numerically using the same mc scheme as for the forward rmc model ( section [ sec : radiance ] ) .
the principle difference is that the light sources @xmath49 typically used in pat are restricted to the boundary , but the adjoint source @xmath48 will not be , as a consequence of the fact that the ` data ' in qpat - the photoacoustic images - is volumetric .
the adjoint model was validated by checking it satisfied the condition : @xmath50 where @xmath51 and @xmath52 are the operators corresponding to the forward and adjoint rmc models , and @xmath53 and @xmath54 are the angle and position dependent source and detector .
three cases were tested : @xmath55 where @xmath56 and @xmath57 are the positions of the source and detector , @xmath58 are the spatial and angular sensitivity of the detector and source . substituting these into yields @xmath59 where @xmath60 and @xmath61 are the forward and adjoint radiances from computing @xmath62 and @xmath63 , respectively .
@xmath64 is the fluence , or angle - integrated radiance .
it can be seen from that the case where a pair of isotropic @xmath65-functions are used for @xmath66 and @xmath67 , that we expect the resulting fluence values at their respective positions , @xmath68 and @xmath69 , to be equal .
this is an intuitive result given the reciprocity of the rte and the angular indpendence of the source - detector combination .
simulations were performed using a 40mm@xmath4340 mm ( 101@xmath43101 pixel ) domain , and 10 fourier harmonics .
each source distribution emitted 10 ^ 6^ photons .
@xmath56 was set to be the centre of the domain with @xmath57 moved along the x - direction across the domain .
comparisons are shown in fig .
[ fig : isoiso_adjoint_test1 ] for case 1 with an isotropic source and detector , fig .
[ fig : isoaniso_adjoint_test2 ] for case 2 with an isotropic source and and anisotropic detector with @xmath70 , and fig.[fig : isoiso_adjoint_test3 ] for case 3 with the same @xmath71 but with the distributed @xmath72 shown in fig .
[ fig : isoiso_adjoint_test3](a ) .
good agreement was obtained in all cases , showing the the rmc adjoint model is an accurate representation of the rte adjoint . and @xmath73 to validate the adjoint model . @xmath74 and @xmath75 were isotropic point sources with @xmath74 at the centre of the domain and @xmath75 translated across the domain at y = 23.6mm.,scaledwidth=60.0% ] ; ( b ) plot of @xmath76 for validation of adjoint model .
plot was produced with @xmath77 as an isotropic point source at the centre of the domain .
@xmath78 was translated across the domain along a line at y = 23.6mm.,scaledwidth=100.0% ] ; ( b ) @xmath79 and @xmath80 to validate the adjoint model .
@xmath81 was an anisotropic point source emitting light over angle following @xmath82 .
@xmath81 was translated along a line across the domain at y = 23.6 mm , as shown by the grey line dashed line in ( a).,scaledwidth=100.0% ]
both the radiance and the adjoint radiance can be expressed as fourier series as in . by substituting these expressions into eqs .
[ absorption_gradient_main ] and [ scattering_gradient_main ] for the functional gradients , simple and easily computed expressions for the gradients can be obtained .
the fluence is simply given by the isotropic component of the field @xmath83 .
the other terms in the expressions for the functional gradients contain integrals of products of the radiance and its adjoint .
if @xmath84 , @xmath85 and @xmath86 are the fourier coefficients of the adjoint radiance , then the gradient with respect to absorption can be written as @xmath87d\theta.\end{aligned}\ ] ] by orthogonality , all terms for which @xmath88 integrate to zero and reduces to @xmath89d\theta,\\ & = -a_{0}(h^{meas}-{{\mu_{a}}}a_{0 } ) + \frac{1}{2\pi}a_{0}a_{0}^{*}+\frac{1}{\pi}\sum_{n=1}^{\infty}a_{n}a_{n}^{*}+\frac{1}{\pi}\sum_{n=1}^{\infty}b_{n}b_{n}^{*}. \label{grad_2d_abs_integral3}\end{aligned}\ ] ] this expression for the absorption gradient is computationally straightforward to evaluate due to the fact that it requires simply summing over products of fourier coefficients already loaded in memory .
the second term in is @xmath90 which contains the phase function given in and can be expanded using a fourier series in powers of @xmath29 @xcite : @xmath91 where @xmath92 .
thus we can write , @xmath93 \notag \\ \left[\frac{1}{2\pi}+\frac{1}{\pi}\sum_{l=0}^{\infty}g^{l}\cos(l(\theta-\theta'))\right ] \notag \\
\left[\frac{1}{2\pi}a_{0}^{*}+\frac{1}{\pi}\sum_{m=1}^{\infty}a_{m}^{*}\cos(m\theta)+\frac{1}{\pi}\sum_{m=1}^{\infty}b_{m}^{*}\sin(m\theta)\right ] d\theta d\theta',\end{aligned}\ ] ] where @xmath31 and @xmath94 are the angles between the z - axis and @xmath16 and @xmath15 , respectively . as such , the scattering angle between the previous direction @xmath15 into the new direction @xmath16 is given by @xmath95 .
it is possible to expand @xmath96 as @xmath97 which in turn allows us to employ orthogonality relationships to simplify the above integrals and write @xmath98 substituting this expression into , we can write the full expression for the functional gradient with respect to the scattering coefficient : @xmath99\left(1-g^{n}\right ) . \label{scattering_gradient3}\end{aligned}\ ] ] the ability to calculate these gradients is the first step to finding a computationally efficient way to solve the full qpat inversion using a monte carlo model of light transport .
the forward and inverse mc models of radiance described above were used with a gradient - descent ( gd ) scheme to estimate @xmath100 and @xmath101 from simulated pat images by minimising the error functional in .
as the adjoint source , @xmath102 , was independent of angle , photons were launched istropically with the launch position being spread out over the range of a source voxel using a randomly distributed number on the interval @xmath24 $ ] .
the initial photon weight was scaled according to the source strength with normalisation of the output quantity ( i.e. radiance , absorbed energy density , harmonic , etc . ) being @xmath103
. the adjoint source may be negative in some places , so the initial photon weight is negative and weight deposition is also negative .
the termination condition was therefore set to be the absolute value of the photon weight falling below the threshold value .
the gradients were calculated using eqs .
[ grad_2d_abs_integral3 ] and [ scattering_gradient3 ] .
a gd scheme was chosen for the minimisation because it is more robust to the mc noise in the functional gradients and error functional than techniques such as l - bfgs that use second - order information . a linesearch algorithm presented by hager and
zhang @xcite was used for the reconstruction of @xmath12 .
a backtracking linesearch was implemented for the reconstruction of @xmath13 .
this is described in section [ sec : mus - recon ] .
the termination condition used by the optimisation was @xmath104 where @xmath42 is the iteration number .
for all the reconstructions it was assumed that the data @xmath105 was given ; no noise was added to the data , but mc noise from the forward simulation of the data was present at about 0.7% ( evaluated by taking several runs of the forward model to estimate the average standard deviation over all positions across all model runs ) .
for each inversion , the forward and adjoint rmc simulations used 10 ^ 8^ photons and 10 fourier harmonics , and was executed on a dell 2u r820 32-core server .
the domain used in the estimation of the absorption coefficient consisted of a background absorption coefficient of 0.01mm^-1^ with a rectangular inclusion equal to 0.2mm^-1^ ( shown in fig .
[ recon - mua](a ) ) , and a background scattering coefficient of 5mm^-1^ with a rectangular inclusion equal to 15mm^-1^ ( shown in fig .
[ recon - mua](b ) ) .
the anisotropy was a homogeneously distributed value of 0.9 .
the measured data , @xmath105 , was formed using a mc simulation illuminated by a collimated line source on the boundary at z=0 mm and on the adjacent boundary at x=4 mm , consisting of 10 ^ 8^ photons .
the inversion for the absorption coefficient only was performed under the assumption that the scattering coefficient was known and the starting estimate of the absorption coefficient was a homogeneous value of 0.01mm^-1^. the termination condition in was satisfied after 11 iterations , having taken 4.1 hours to run , and is shown in fig .
[ recon - mua](b ) with profiles through the true and reconstructed distributions of @xmath12 shown in fig .
[ recon - mua](d ) .
very good agreement between the true and reconstructed absorption coefficient is observed , with a value of the error function after 11 iterations being 2.9@xmath4310 ^ -9^. the optimisation routine was stopped as the change in the error function on the 12^th^ iteration was below the function tolerance of 10 ^ -9^ , indicating convergence .
the average error in the estimate of the absorption coefficient @xmath106 , computed as @xmath107 , was 0.2% over the entire domain .
inversions for the scattering coefficient were performed using the same domain as above , shown in fig .
[ recon - mus](a ) and ( b ) , with the illumination and number of photons in the forward simulation also being the same . here
, it was assumed that the absorption coefficient was known and the starting estimate of the scattering coefficient was equal to the background value of 5mm^-1^. two modifications were necessary to achieve convergence in the optimisation for the scattering coefficient .
first , a custom gd algorithm was used in which a backtracking linesearch was implemented .
a backtracking linesearch @xcite starts with a large candidate step length and progressively reduces the step size whilst checking for a sufficient decrease in the error functional .
the sufficient decrease condition is expressed as @xmath108+\alpha^{(i)}p^{(i ) } ) \leq \epsilon([{{\mu_{a}}}^{(i)},{{\mu_{s}}}^{(i)}])+\nu\alpha^{(i)}\nabla\epsilon^{(i)t}p^{(i)},\ ] ] where @xmath109 was chosen to be 0.2 by inspection because this produced rapid convergence . in order to improve efficiency of the linesearch
, step sizes were bounded between @xmath110 $ ] ; it was found that this range yielded sufficiently large steps to ensure reasonably efficient progress in the minimisation .
second , the termination condition in was relaxed due to the much slower convergence of the scattering coefficient , and instead required a relative change in the error functional of 10 ^ -5^. this was satisfied after 35 iterations , and is shown in fig .
[ recon - mus](b ) with profiles through the true and reconstructed distributions of @xmath13 shown in fig .
[ recon - mus](d ) .
it can be seen from fig .
[ recon - mus](c ) and ( d ) that the inversion has partly reconstructed the inclusion in the scattering coefficient .
the inability to reconstruct edges of the inclusion in the scattering coefficient is expected , given the diffusive nature of the scattering .
however , the discrepancy in @xmath13 in the inclusion , evident from fig .
[ recon - mus](d ) , suggests premature termination of the optimisation .
this is due to the fact that the gradient with respect to scattering is small and prone to noise in the functional gradients .
this low snr in the gradients has the impact that that search directions in the optimisation routine are often sub - optimal , which results in little or no progress of the optimisation .
the progressive reduction in snr in the gradient means that non - descent steps are likely and can therefore trigger the termination condition .
in this paper a novel mc model of the rte was presented .
the model computes the radiance in a fourier basis in 2d and is straightforward to extend to 3d using a spherical harmonics basis .
the accuracy of the model was demonstrated by comparing the angle - resolved radiance at two positions in the domain to corresponding appropriate analytic solutions .
sections [ sec : gradients ] and [ sec : examples ] presented the application of the rmc algorithm to estimating the absorption and scattering coefficients from simulated pat images . in section [ sec : mua - recon ] it was observed that the absorption coefficient was estimated with an average error of 0.2% over the domain relative to the true value , when the scattering coefficient is known , and in the presence of 0.7% average noise in the data .
this is encouraging , particularly because noise is not only present in @xmath105 but in the fourier harmonics computed using the forward and adjoint rmc simulations , which is propagated to the estimates of the functional gradients .
consequently the search direction in the gd algorithm will always be sub - optimal .
furthermore , noise in @xmath111 , the estimate of the absorbed energy density at the @xmath112 iteration of the linesearch , will be also be propagated to the error functional , resulting in a non - smooth search trajectory for the linesearch because at every point @xmath113 , the error function will be corrupted by some different noise @xmath114 : @xmath115 . in practice
, this did not preclude reconstruction of the absorption coefficient since the calculated gradients remained descent directions despite the noise .
furthermore , the error functional in @xmath12 is sufficiently convex that the addition of some noise does not prevent the linsearch from yielding sufficiently large a step length to allow rapid convergence .
reconstruction of the scattering coefficent correctly located the scattering perturbation in the simulated image , however the peak value in the reconstruction was lower than the true value .
this is a direct consequence of the fact that the scattering coefficient is related to the absorbed energy distribution only through the optical fluence distribution .
consequently , the snr in @xmath116 is typically much less than that for absorption .
this causes termination of the algorithm before the peak magnitude of the parameter has been found in the search space .
the authors acknowledge the contribution of andre liemert who kindly provided radiance data used in the validation of rmc in section [ sec : validation ] .
the authors acknowledge the use of the ucl legion high performance computing facility ( legion@ucl ) , and associated support services , in the completion of this work .
10 j. laufer , b. cox , e. zhang , and p. beard , `` quantitative determination of chromophore concentrations from 2d photoacoustic images using a nonlinear model - based inversion scheme . , '' _ applied optics _ * 49 * , pp . 121933 , mar .
b. t. cox , s. r. arridge , and p. c. beard , `` estimating chromophore distributions from multiwavelength photoacoustic images .
, '' _ journal of the optical society of america . a , optics , image science , and vision _
* 26 * , pp .
443455 , feb .
t. tarvainen , a. pulkkinen , b. cox , j. kaipio , and s. arridge , `` bayesian image reconstruction in quantitative photoacoustic tomography .
, '' _ ieee transactions on medical imaging _ * 32 * , pp .
22872298 , aug .
2013 .
t. saratoon , t. tarvainen , b. t. cox , and s. r. arridge , `` a gradient - based method for quantitative photoacoustic tomography using the radiative transfer equation , '' _ inverse problems _ * 29 * , p. 075006
, july 2013 .
p. surya mohan , t. tarvainen , m. schweiger , a. pulkkinen , and s. r. arridge , `` variable order spherical harmonic expansion scheme for the radiative transport equation using finite elements , '' _ journal of computational physics _
* 230 * , pp .
73647383 , aug .
2011 .
d. boas , j. culver , j. stott , and a. dunn , `` three dimensional monte carlo code for photon migration through complex heterogeneous media including the adult human head , '' _ optics express _ * 10 * , pp . 15970 , feb . 2002 .
p. c. beard and t. n. mills , `` characterization of post mortem arterial tissue using time - resolved photoacoustic spectroscopy at 436 , 461 and 532 nm .
, '' _ physics in medicine and biology _ * 42 * , pp .
17798 , jan .
antonelli , a. pierangelo , t. novikova , p. validire , a. benali , b. gayet , and a. de martino , `` impact of model parameters on monte carlo simulations of backscattering mueller matrix images of colon tissue .
, '' _ biomedical optics express _ * 2 * , pp . 183651 , july 2011 .
n. manuchehrabadi , y. chen , a. lebrun , r. ma , and l. zhu , `` computational simulation of temperature elevations in tumors using monte carlo method and comparison to experimental measurements in laser photothermal therapy .
, '' _ journal of biomechanical engineering _
* 135 * , p. 121007
j. cassidy , v. betz , and l. lilge , `` treatment plan evaluation for interstitial photodynamic therapy in a mouse model by monte carlo simulation with fullmonte , '' _ frontiers in physics _ * 3 * , pp . 110 , feb .
2015 . v. periyasamy and m. pramanik ,
`` monte carlo simulation of light transport in tissue for optimizing light delivery in photoacoustic imaging of the sentinel lymph node .
, '' _ journal of biomedical optics _ * 18 * , p. 106008
, jan . 2013 .
v. periyasamy and m. pramanik , `` monte carlo simulation of light transport in turbid medium with embedded object spherical , cylindrical , ellipsoidal , or cuboidal objects embedded within multilayered tissues .
, '' _ journal of biomedical optics _ * 19 * , p. 045003
2014 .
a. sassaroli and f. martelli , `` equivalence of four monte carlo methods for photon migration in turbid media .
, '' _ journal of the optical society of america . a , optics , image science , and vision _
* 29 * , pp . 21107 , oct . 2012 .
r. hochuli , s. powell , s. arridge , and b. cox , `` forward and adjoint radiance monte carlo models for quantitative photoacoustic imaging , '' in _ proc . of spie ,
photons plus ultrasound : imaging and sensing _ , a. a. oraevsky and l. v. wang , eds . , * 9323 * , pp . 93231p10 ,
2015 .
a. liemert and a. kienle ,
`` analytical approach for solving the radiative transfer equation in two - dimensional layered media , '' _ journal of quantitative spectroscopy and radiative transfer _ * 113 * , pp .
559564 , may 2012 .
m. j. c. van gemert , a. j. welch , w. m. star , m. motamedi , and w .- f .
cheong , `` tissue optics for a slab geometry in the diffusion approximation , '' _ lasers in medical science _ * 2 * , pp .
295302 , dec . 1987 .
a. k. jha , m. a. kupinski , h. h. barrett , e. clarkson , and j. h. hartman , `` three - dimensional neumann - series approach to model light transport in nonuniform media . , ''
_ journal of the optical society of america .
a , optics , image science , and vision _ * 29 * , pp . 188599 , sept . 2012 . | forward and adjoint monte carlo ( mc ) models of radiance are proposed for use in model - based quantitative photoacoustic tomography . a 2d radiance mc model using a harmonic angular basis is introduced and validated against analytic solutions for the radiance in heterogeneous media . a gradient - based optimisation scheme is then used to recover 2d absorption and scattering coefficients distributions from simulated photoacoustic measurements .
it is shown that the functional gradients , which are a challenge to compute efficiently using mc models , can be calculated directly from the coefficients of the harmonic angular basis used in the forward and adjoint models .
this work establishes a framework for transport - based quantitative photoacoustic tomography that can fully exploit emerging highly parallel computing architectures . |
a _ classical link _ @xmath5 consists of finitely many , pairwise disjoint , piecewise smooth closed curves in @xmath6 or @xmath7 ; each individual closed curve is a _ component _ of @xmath5 .
a _ regular diagram _
@xmath0 of @xmath5 is obtained from a regular projection in the plane that is , a projection whose only singularities are double points called _ crossings _ by specifying the over - and under - crossing arcs at each crossing .
the projection itself is a 4-regular plane graph @xmath3 , the _ universe _ of @xmath0 . saying that @xmath3 is a _ plane graph _
means that it is given with a specific embedding in @xmath8 ; when the embedding is forgotten @xmath3 becomes an _
abstract graph_. we presume @xmath5 is given with orientations of its components ; these orientations make @xmath3 into a 2-in , 2-out digraph @xmath1 .
combinatorial structures associated with @xmath0 incorporate geometric information about its embedding in the plane in various ways for instance the classical checkerboard graph is defined using the connected components of @xmath9 , a rigid graph specifies for each vertex of @xmath3 the cyclic order of the incident edges , a signed gauss code specifies the cyclic order of the crossings on each link component along with underpassing - overpassing information at each crossing , and an atom includes an embedding of a graph on a surface .
most combinatorial structures associated to classical link diagrams must be adjusted when we try to extend them to _ virtual link diagrams _
each crossing in a virtual link diagram has three possible configurations .
in addition to the two classical configurations with different choices of under- and over - crossing arcs , there is the virtual configuration , in which two arcs of the diagram simply cross each other without intersecting .
the virtual crossings are not really there @xcite , in the same way that a plane representation of a graph may contain apparent edge - intersections that are not really there .
the universe of a virtual link diagram is an abstract 4-regular graph with vertices corresponding only to the classical crossings ; the virtual diagram gives a regular immersion of the graph in @xmath8 , not an embedding as in the classical case .
the virtual crossings naturally suggest an embedding of the universe on a closed , orientable surface @xmath10 of some genus , and the diagram can then be thought of as representing a link in the thickened surface @xmath11 . consequently one way to extend notions involving the plane geometry of classical link diagrams is to use the geometry of the surface @xmath10 ; for instance the classical checkerboard graphs extend in this manner to atoms associated with virtual diagrams .
( by the way we should make it clear that we use the terms _ link _ , _ virtual link _ and _ classical or virtual link _ interchangeably ; when we want to be restrictive we specify that a link is classical or non - classical . ) the kauffman bracket and jones polynomial of a classical link @xmath5 may be described in a variety of ways using combinatorial structures associated with regular diagrams of @xmath5 ; kauffman s basic description follows .
each crossing of a regular diagram @xmath0 has two _ smoothings _ , one denoted @xmath12 and the other denoted @xmath13 ; distinguishing between them is another example of the use of geometric information mentioned in the first paragraph above , as it involves the orientation of the plane in which @xmath0 is drawn . if @xmath0 has @xmath14 crossings then it has @xmath15 _ states _ , obtained by applying either the @xmath16 or the @xmath13 smoothing at each crossing .
given a state @xmath10 let @xmath17 denote the number of @xmath16 smoothings in @xmath10 , @xmath18 the number of @xmath13 smoothings in @xmath10 , and @xmath19 the number of simple closed curves in @xmath20 , including any crossing - free components that might appear in @xmath0 .
( these simple closed curves are called `` loops '' in @xcite , but we reserve this word for graph - theoretic use . )
then the _ kauffman bracket polynomial _ of @xmath0 is a sum indexed by states : @xmath21=\sum_{s}a^{a(s)}b^{b(s)}d^{c(s)-1}.\ ] ] [ ptb ] wifig1.ps almost immediately after the kauffman bracket was introduced , it was related to the tutte polynomial through the checkerboard graph construction @xcite .
a classical link diagram @xmath0 has an _ alternating _ orientation on its universe @xmath3 i.e. , the edges of @xmath3 may be directed in such a way that when one traverses any component of the link , one encounters forward- and backward - directed edges alternately .
such an alternating orientation arises naturally from the cycles that bound the complementary regions of @xmath0 , and is associated with the relationship between kauffman states of @xmath0 and spanning forests in the checkerboard graph that underlies the jones - tutte connection .
this connection between spanning forests and kauffman states is also part of a combinatorial discussion of the alexander polynomial @xcite . by the way ,
the relationship between eulerian circuits and spanning trees in regular directed graphs , known as the best theorem , appeared in the combinatorial literature long before the jones polynomial was introduced @xcite ; the special relationship between circuit partitions in a 4-regular plane graph and the tutte polynomial of an associated checkerboard graph also appeared first outside knot theory @xcite .
the kauffman bracket of a virtual link diagram is described in almost exactly the same way as that of a classical link diagram ; the only difference is that we can not say that @xmath19 denotes the number of _ simple _ closed curves in @xmath10 , as the closed curves need not be simple .
much of the jones - tutte machinery extends directly to virtual link diagrams which possess alternating orientations or ( equivalently ) orientable atoms ; see @xcite and also chapter iv of @xcite .
if the tutte polynomial is augmented with topological information to give an invariant of graphs embedded on surfaces as in @xcite , then the relationship between links in thickened surfaces and virtual links leads to a general relationship between this topological tutte polynomial and the jones polynomial @xcite . in @xcite zulli and
the present author proposed a different , purely combinatorial approach to the jones polynomials and kauffman brackets of classical and virtual knots ( one - component links ) that uses surprisingly little geometric information about knot diagrams , and does not involve either the ordinary tutte polynomial or the topological tutte polynomial . if @xmath0 is a regular diagram of a classical or virtual knot @xmath22 then the universe @xmath3 of @xmath0 has an interlacement graph defined with respect to the euler circuit obtained directly from @xmath22 , and the kauffman bracket @xmath23 $ ] can be obtained from this interlacement graph without using complementary regions , ordering the crossings , specifying over- and under - crossing arcs , ordering the edges incident on a vertex , distinguishing between smoothings , counting closed curves , or considering closed surfaces of any genus .
all that is required is to record whether each crossing s contribution to the writhe is + 1 or -1 ; this is done by attaching a loop to each vertex that corresponds to a crossing whose writhe contribution is -1 .
the resulting _ looped interlacement graph _ is denoted @xmath24 .
@xmath24 is obtained by considering @xmath3 as an abstract graph rather than an embedded one , so any virtual crossings are really not there ( rather than merely not really there ) as far as @xmath24 is concerned .
the analysis of @xcite describes the jones polynomial of a classical or virtual knot as the result of the composition of three functions .
the first function is the looped interlacement graph construction .
the second function associates a 3-variable graph bracket polynomial to an arbitrary graph , using either a recursion involving the local complement and pivot operations that have appeared in the theory of circle graphs and interlace polynomials @xcite or a formula involving matrix nullities over @xmath25 .
zulli @xcite proved that this formula , when applied to looped interlacement graphs , correctly assesses the number of closed curves in each kauffman state .
the third function modifies the 3-variable graph bracket polynomial so that the resulting graph jones polynomial is invariant under appropriate graph reidemeister moves .
@xmath26\qquad\mapsto\qquad v_{\mathcal{l}(d)}(t)$ ] ( virtual ) looped interlace-3-variable graph graph knot ment graph using bracket jones diagram euler circuit @xmath27polynomial polynomial the purpose of the present paper is to present an extension of this description to the jones polynomials of classical and virtual links .
each of the three steps must be modified to accommodate the extension .
@xmath28\qquad\mapsto\qquad v_{\mathcal{l}(d , c)}(t)$ ] ( virtual ) looped interlace-3-variable marked - graph linkment graph using marked - graph jones diagram an euler system @xmath29bracket polynomial polynomial the first step is again the construction of a looped interlacement graph .
an interlacement graph of a 4-regular graph or a 2-in , 2-out digraph is constructed by using a specific euler circuit , or if the original graph is not connected a specific _ euler system _ containing an euler circuit for each connected component .
a classical or virtual knot diagram has an euler circuit arising naturally from the diagrammed knot , but in general there is no canonical choice of an euler system in a universe @xmath3 associated to a diagram @xmath0 representing a multi - component link . consequently our discussion of this step in section 2 involves two complications that did not affect @xcite : @xmath3 need not be connected , and we must allow a variety of euler systems in @xmath3 .
the first of these complications is accommodated using a common graph - theoretic convention , which allows graphs to include free loops ; a free loop is not incident on any vertex , but does count as a connected component . if @xmath3 is a disconnected graph with @xmath30 connected components , we record this fact by inserting @xmath31 free loops into the interlacement graph @xmath4 .
this might seem unnatural at first , but it is necessary if we want to recover the kauffman bracket from @xmath4 ; without some such convention interlacement can not detect crossing - free unknotted components , or ( more generally ) distinguish connected sums from split unions .
[ ptb ] lbracfi2.ps the second complication is handled by adopting a convention to identify the euler system used to produce a looped interlacement graph @xmath4 .
we presume the link @xmath5 is given with orientations of its components , and we restrict our attention to directed euler systems of the directed universe @xmath1 .
each crossing of @xmath0 is then marked to record which of the two possible _ transitions _
( edge - pairings ) occurs at that crossing in the euler system @xmath2 used to produce @xmath4 .
( transitions provide a standard combinatorial description of circuits in 4-regular graphs and 2-in , 2-out digraphs ; see for instance @xcite . )
crossings at which @xmath2 follows the incident link component(s ) are left unmarked , and crossings at which @xmath2 is orientation - consistent with the incident link component(s ) but does not follow it ( or them ) are marked with the letter @xmath32 ; see figures [ lbracfi2 ] and [ lbracfi3 ] . the same convention is used to mark the vertices of @xmath3 and @xmath4 . according to results discussed in @xcite , different directed euler systems of @xmath1 are interconnected through _ transposition _ , a combinatorial operation .
it follows that changing the choice of @xmath2 affects @xmath4 through an appropriately modified version of the pivot operation , which we call a _ marked pivot_. [ tb ] lbracfi3.ps as in @xcite , the second step uses a formula involving matrix nullities over @xmath25 to associate a 3-variable bracket polynomial to an arbitrary graph ; but now we consider arbitrary graphs with marked vertices .
the marked - graph bracket polynomial of a looped interlacement graph @xmath33 is the same as the kauffman bracket polynomial of @xmath0 ; the relationship between the matrix nullity formula and the kauffman bracket is verified using an extension of a combinatorial equality due to cohn and lempel @xcite .
this equality is presented briefly in section 4 , and more fully in @xcite ; it incorporates the cohn - lempel equality along with results used in @xcite . in section 5
we define the 3-variable marked - graph bracket polynomial and deduce that @xmath34=[d]$ ] .
we also prove a very useful result : the bracket is invariant under the marked pivot operation mentioned above .
this invariance allows us to develop the theory of the marked - graph bracket using many of the results of @xcite , even though marked graphs did not appear there ; in essence , we often use marked pivots to simply push marked vertices out of the way .
for instance , in section 6 we show that the recursion given in @xcite extends to a recursive description of the marked - graph bracket polynomial involving the local complementation and pivot operations .
the recursion uses formulas that can only be applied to special configurations in which certain vertices have no marked neighbors ; such configurations can always be created using marked pivots .
the third step involves proving that the marked - graph bracket polynomial yields a marked - graph jones polynomial that is invariant under certain reidemeister moves on marked graphs ; these graph moves generate the `` images '' of the familiar reidemeister moves of knot theory under the construction of @xmath4 .
we say _ generate _ because we do not explicitly exhibit all the marked - graph reidemeister moves ; following an idea of stlund @xcite , we only exhibit enough basic moves to generate the rest through composition . as in @xcite , the fact that @xmath4 is defined by considering @xmath3 as an abstract graph rather than
a graph drawn on a plane or a surface is reflected in the fact that we can simply ignore the virtual reidemeister moves of @xcite . in sum ,
our results show that the kauffman bracket and jones polynomial can be defined for either of the following abstract ( non - embedded ) combinatorial structures .
* a 2-in , 2-out digraph @xmath1 , given with a partition of its edge - set into directed circuits and a partition of its vertex - set into two subsets * an undirected graph @xmath35 , given with a partition of its vertex - set into two subsets it would be interesting to understand how these two types of abstract combinatorial structures are related to other knot - theoretic notions . for instance , orientable and non - orientable atoms have been used for a variety of purposes , e.g. , to construct khovanov homology for virtual knots @xcite .
atoms involve graphs embedded on surfaces , so they contain more geometric information than the abstract combinatorial structures .
perhaps some constructions based on atoms do not require all of this geometric information , and can be modified to use one of the abstract combinatorial structures .
manturov has also recently introduced the idea of _ free _ knots and links , which are equivalence classes of 4-regular graphs under modified versions of the reidemeister moves @xcite ; they are related to the first kind of abstract combinatorial structure mentioned above , with the vertex - partition forgotten .
perhaps the combinatorial theory of circuit partitions in 4-regular graphs will turn out to be useful in defining invariants of these objects .
it would also be interesting to conduct a purely combinatorial investigation of the bracket as an invariant of either of these structures , apart from the special case in which @xmath1 is the universe of a diagram of an oriented link , the circuit partition of @xmath36 consists of the link s components , the partition of @xmath37 represents the distinction between positive and negative crossings , @xmath35 is the looped interlacement graph of @xmath1 with respect to an euler system @xmath2 , the partition of @xmath38 represents vestigial information regarding @xmath2 , and free loops represent all but one of the connected components of @xmath3 . in particular , we wonder whether there might be a useful relationship between the bracket polynomial and the interlace polynomials of @xcite or the multivariate interlace polynomial of @xcite .
the same comments apply to the jones polynomial , which is naturally regarded as a simplification of the bracket , modified so as to be invariant under a certain equivalence relation which is motivated topologically but may be described in purely combinatorial terms . before closing this introduction
we should thank l. zulli for many hours spent working together to formulate and make precise the ideas of @xcite , and also for his comments on early drafts of the current paper .
d. p. ilyutko was kind enough to bring @xcite to our attention ; the present work developed as we tried to understand the differences and similarities of the two approaches .
v. o. manturov s good advice improved the paper in several ways .
we are also grateful to lafayette college for its support .
in this section we discuss some terminology and results about eulerian graphs , and then apply these notions to link diagrams .
eulerian graphs have been studied for centuries ; it is not surprising that terminology has varied over the years , and some results have been rediscovered . for us
a _ graph _ may have loops or multiple edges , and each edge has two distinct _ directions _ ; in a _ directed graph _ a preferred direction has been chosen for each edge .
the terms _ adjacent _ and _ neighbor _ refer to non - loop edges only ; no vertex is adjacent to itself or a neighbor of itself . in order to provide appropriate notation for the two different directions of a loop , each edge
is technically regarded as consisting of two distinct half - edges , one incident on each end - vertex ; the preferred direction of a directed edge is expressed by designating one half - edge as _ _
initial _ _ and the other as _ terminal_. we will almost always abuse notation by leaving it to the reader to split edges into half - edges .
a _ path _ in a graph is a sequence @xmath39 ( technically @xmath40 where @xmath41 has the half - edges @xmath42 and @xmath43 ) such that for each @xmath44 , the vertices @xmath45 and @xmath46 are incident on the edge @xmath41 ; in a directed graph a _ directed path _ must respect all edge - directions , and an _ undirected path _ might not respect some edge - directions .
if @xmath47 the path is _ closed _ ; in this case it is customary but not mandatory to omit @xmath48 from the list .
a _ circuit _ is a closed path in which no edge appears more than once ; digraphs have _ directed circuits _ and _ undirected circuits_. an _ euler circuit _ in a connected graph is a circuit in which every edge of the graph appears , and an _ euler system _ in a graph is a set that contains one euler circuit for each connected component .
a connected , undirected graph has an euler circuit if and only if every vertex is of even degree ; a connected , directed graph has a directed euler circuit if and only if every vertex has equal indegree and outdegree .
we allow graphs to have _
free loops_. these do not interact in any way with the vertices and edges of the graph , and they do not affect anything we would otherwise say about the graph , except for the fact that each free loop is counted as a connected component . in particular , a free loop is not a kind of loop ; indeed if the terminology were not standard we would call them `` empty components . '' for example , if an undirected graph @xmath49 consists of a 4-cycle and two free loops then @xmath35 is 2-regular , and if @xmath50 are two edges in @xmath35 that are not incident on the same vertex then @xmath51 is bipartite and 1-regular ; @xmath35 has three connected components and @xmath52 has four . as a free loop is a connected component , an euler system for a graph must contain all the graph s free loops . in figures representing graphs ,
free loops are drawn as circles with no incident vertices .
let @xmath3 be an undirected 4-regular graph with @xmath53 .
it need not be the universe of any link diagram ; we use the letter @xmath3 only for notational consistency .
[ interlacement ] _ @xcite _ if @xmath2 is an euler system for @xmath3 then the _ interlacement matrix _
@xmath54 is the @xmath55 matrix over @xmath25 whose @xmath56 entry is @xmath57 if and only if @xmath58 and @xmath45 and @xmath59 are _ interlaced _ in @xmath2 , i.e. , when we follow @xmath2 starting at @xmath45 we encounter @xmath59 , then @xmath45 , then @xmath59 again before finally returning to @xmath45 .
[ intergraph ] if @xmath3 has @xmath60 connected components then the _ interlacement graph _
@xmath61 is obtained by adjoining @xmath31 free loops to the simple graph with vertex - set @xmath37 and adjacency matrix @xmath54 . definition [ intergraph ] differs from the corresponding definition of @xcite in its use of free loops . observe that if @xmath3 is disconnected then @xmath54 consists of diagonal blocks corresponding to the interlacement matrices of the connected components of @xmath3 .
the same occurs if @xmath3 is a `` connected sum , '' i.e. , if @xmath3 has two subgraphs @xmath62 and @xmath63 such that @xmath64 and @xmath65 .
the free loops in definition [ intergraph ] allow @xmath61 to distinguish these two situations : the interlacement graph of the disjoint union of @xmath62 and @xmath63 has one more free loop than the interlacement graph of a connected sum of @xmath62 and @xmath63 . _
@xcite _ the @xmath66__-transform _ _ @xmath67 of an euler system @xmath2 at a vertex @xmath68 is obtained by reversing one of the two @xmath68-to-@xmath68 paths within the euler circuit of @xmath2 in the connected component of @xmath3 containing @xmath68 .
the interlacement graph @xmath69 is obtained from @xmath61 by toggling ( i.e. , changing ) every adjacency involving neighbors of @xmath68 in @xmath61 ; that is , if @xmath70 are two distinct neighbors of @xmath68 in @xmath61 , then @xmath71 and @xmath72 are adjacent in @xmath69 if and only if they are not adjacent in @xmath61 .
if we apply a single @xmath66-transformation to a directed euler system of a 2-in , 2-out digraph @xmath1 then the result is no longer compatible with the edge - directions of @xmath1 .
suppose @xmath2 is a directed euler system in @xmath1 , and suppose @xmath68 and @xmath73 are two vertices of @xmath1 that are interlaced with respect to @xmath2 .
then @xmath2 contains a circuit @xmath74 .
the @xmath66-transform @xmath67 contains an undirected circuit @xmath75 ; the overbars indicate that the paths @xmath76 and @xmath77 have been reversed
. then @xmath78 contains @xmath79 and @xmath80 contains @xmath81 .
the lack of overbars indicates that this last is a directed circuit of @xmath1 .
arratia , bollobs and sorkin @xcite call the operation @xmath82 a _ transposition _ on the pair @xmath83 . comparing @xmath84 to @xmath81
, we see that @xmath85 is obtained from @xmath61 by interchanging the neighbors of @xmath68 and @xmath73 , and toggling every adjacency between vertices @xmath86 such that @xmath71 is adjacent to @xmath68 , @xmath72 is adjacent to @xmath73 , and either @xmath71 is not adjacent to @xmath73 or @xmath72 is not adjacent to @xmath68 .
that is , @xmath87 is obtained from the _ pivot _
@xmath88 by interchanging the neighbors of @xmath68 and @xmath73 .
( simply interchanging the names of @xmath68 and @xmath73 would have the same effect , but we prefer not to do this as it would complicate our discussion later ; see for instance lemma [ transform ] below . ) a natural way to identify an euler system @xmath2 in a 4-regular graph @xmath3 is to record , for each vertex of @xmath3 , which of the three possible _ transitions _ ( matchings of the incident half - edges ) is used by @xmath2 .
this gives a rather complicated description , listing a pair of incident half - edges at each vertex .
it is much simpler to use transitions to compare two euler systems @xmath2 and @xmath89 .
if we choose orientations for the euler circuits in @xmath2 then these orientations allow us to make @xmath3 into a 2-in , 2-out digraph @xmath1 : at each vertex @xmath68 , @xmath89 must have either the same transition as @xmath2 , the other transition that is compatible with the edge - directions in @xmath1 , or the transition that is incompatible with the edge - directions of @xmath1 .
we say @xmath89 _ follows _ @xmath2 at @xmath68 , @xmath89 is _ compatible with _
@xmath2 _ without following it _ at @xmath68 , or @xmath89 _ is incompatible _ with @xmath2 at @xmath68 , respectively , to describe the three cases .
the same description of the relationship between the transitions of @xmath2 and @xmath89 at @xmath68 will hold if some circuits of @xmath2 are oriented differently .
this simple description of the relationship between @xmath2 and @xmath89 is useful in proving the following result , due to kotzig @xcite , ukkonen @xcite and pevzner @xcite .
[ ktransform ] every two euler systems @xmath2 and @xmath89 of a 4-regular undirected graph @xmath3 are connected to each other through a sequence of @xmath66-transformations .
if there is no vertex at which @xmath89 is incompatible with @xmath2 , they are connected through a sequence of transpositions .
if @xmath2 and @xmath89 have the same transition at every vertex , then @xmath90 . proceeding by induction on the number of vertices at which their transitions differ , we have two cases . if their transitions are incompatible at a vertex @xmath68 then let @xmath91 .
@xmath92 has the same transitions as @xmath2 except at @xmath68 , where it shares the transition of @xmath89 , so the inductive hypothesis gives the result .
suppose instead that there is no vertex at which @xmath2 and @xmath89 have incompatible transitions .
let @xmath93 denote the subset of @xmath37 containing the vertices at which the transitions of @xmath2 and @xmath89 differ , and let @xmath94 have the shortest possible @xmath73-to-@xmath73 path within @xmath2 ; let the euler circuit appearing in @xmath2 that contains @xmath73 be @xmath95 , with @xmath96 no longer than @xmath97 .
if no vertex of @xmath93 appears on @xmath96 the transitions of @xmath89 are the same as those of @xmath2 at every vertex of @xmath98 except @xmath73 ; then the circuit @xmath98 appears in @xmath89 , an impossibility as @xmath98 is not an euler circuit .
the choice of @xmath73 implies that no vertex of @xmath93 can appear twice on @xmath96 ; hence there is a vertex @xmath99 that appears precisely once on @xmath96 .
that is , @xmath73 and @xmath100 are interlaced with respect to @xmath2 .
. then @xmath92 has the same transitions as @xmath2 at all vertices except @xmath73 and @xmath100 , where it is compatible with @xmath2 ; hence @xmath92 has the same transitions at @xmath73 and @xmath100 as @xmath89 , has different transitions from @xmath89 at two fewer vertices than @xmath2 , and is not incompatible with @xmath89 at any vertex .
the inductive hypothesis gives the result . just as using transitions to describe the relationship between two euler systems is simpler than using transitions to describe a single euler system , it is easier to describe an euler system in the universe of a link diagram than it is to describe an euler system in an arbitrary 4-regular graph .
suppose @xmath0 is a diagram of a classical or virtual link @xmath5 , given with orientations of its components ; these orientations make @xmath3 into a 2-in , 2-out digraph @xmath1 .
( any link component that is crossing - free in @xmath0 appears as a free loop in @xmath3 . ) at each crossing in @xmath0 , there are three transitions : one followed by the component(s ) of @xmath5 , one that does not follow the components of @xmath5 but is consistent with the edge - directions of @xmath1 , and a third that is inconsistent with these edge - directions . a directed euler system for @xmath1 will not involve any transitions of the third type .
a simple description of a directed euler system @xmath2 is incorporated into a link diagram @xmath0 by using the letter @xmath32 to mark each crossing at which @xmath2 is orientation - consistent with @xmath5 without following it .
( crossings at which @xmath2 follows @xmath5 are left unmarked . )
see figures [ lbracfi2 ] and [ lbracfi3 ] in the introduction .
we do not use the traditional notation for smoothings ( with @xmath102 instead of @xmath32 ) , in order to avoid confusion between notation that identifies an euler system in @xmath3 and notation that identifies a new diagram obtained by modifying @xmath0 . the same system is used to mark the vertices in the interlacement graph @xmath61 , resulting in the _ marked interlacement graph _ @xmath103 . [ h ] lbracfi4.ps as noted in the introduction , the famous jones - tutte connection of @xcite is related to a choice of edge - directions in @xmath3 which is _ alternating _ in the sense that as one traverses a component of the link , one encounters a reversal in edge - direction each time one passes through a vertex . if @xmath3 admits an alternating orientation as a 2-in , 2-out digraph then there is an alternating diagram @xmath104 with universe @xmath3 : at each classical crossing the arc with both edges directed inward is the underpassing arc in @xmath104 .
lemma 7 of @xcite then tells us that @xmath5 is checkerboard colorable ; @xcite also tells us that not all virtual links are checkerboard colorable , so restricting attention to euler systems that give alternating orientations would involve a significant loss of generality .
one way to avoid this loss of generality is to use nonorientable atoms ; see @xcite for instance .
our approach avoids loss of generality in a different way , by using edge - directions that are consistent with the orientations of the components of @xmath5 instead of using alternating edge - directions . by the way
, a simple induction on the number of virtual crossings shows that it is always possible to choose an alternating orientation for a virtual knot diagram , but the orientation may not correspond to a 2-in , 2-out digraph .
for instance every alternating orientation of the virtual knot diagram from @xcite indicated in figure [ lbracfi4 ] produces a vertex of indegree 4 and a vertex of outdegree 4 .
diagrams of multi - component virtual links need not have any alternating orientations at all ; for instance consider a diagram with only one virtual crossing , which involves two different link components .
suppose we are given a diagram @xmath0 of an oriented classical or virtual link @xmath5 , along with a directed euler system @xmath2 for @xmath105 .
then the _ _ looped interlacement graph _ _ @xmath4 is obtained from @xmath103 by attaching a loop at each vertex corresponding to a negative crossing . [ h ] lbracfi5.ps recall that if @xmath68 and @xmath73 are vertices of a graph @xmath35 then the _ pivot _ @xmath106 is the graph obtained from @xmath35 by toggling ( reversing ) the adjacency between every pair of vertices @xmath86 such that @xmath71 is a neighbor of @xmath68 , @xmath72 is a neighbor of @xmath73 , and either @xmath71 is not a neighbor of @xmath73 or @xmath72 is not a neighbor of @xmath71 .
( pivots have no effect on loops . )
we observed above that if a 4-regular graph @xmath3 has vertices @xmath107 that are interlaced with respect to an euler system @xmath2 , then @xmath108 is the graph obtained from @xmath109 by interchanging the neighbors of @xmath68 and @xmath73 .
this observation applies to looped interlacement graphs as follows .
[ transform ] suppose @xmath0 is an oriented link diagram with a directed euler system @xmath2 , and @xmath107 are adjacent in the looped interlacement graph @xmath4 .
then @xmath110 is the marked graph obtained from @xmath111 by toggling the marks on @xmath68 and @xmath73 ( i.e. , removing a @xmath32 at @xmath68 or @xmath73 if one is there , and adjoining a @xmath32 at @xmath68 or @xmath73 if one is not there ) , and interchanging the neighbors of @xmath68 and @xmath73 . at any vertex other than @xmath68 or @xmath73 , @xmath80 has the same transition as @xmath2 . at @xmath68 and @xmath73 ,
@xmath80 is compatible with @xmath2 without following it .
figure [ lbracfk1 ] illustrates some of the definitions given in section 2 .
each row of the figure has an oriented knot diagram on the left - hand side , the corresponding directed universe in the middle , and the resulting looped interlacement graph on the right .
the first two examples are different marked versions of a diagram of the figure - eight knot ; the transitions of the euler circuits are indicated by the patterns of dashes in the corresponding universes .
the third example is the well - known virtual knot of @xcite , and the fourth is a virtual unknot diagram ; their directed universes are isomorphic abstract graphs .
all euler circuits in the third and fourth directed universes give rise to the same looped interlacement graph , so we have not bothered to mark the diagrams or indicate euler circuits . note that definition [ intergraph ] does not require any free loops because all four diagrams are connected .
if we were to consider instead the split diagram consisting of the entire left - hand side of the figure , then the looped interlacement graph would include the right - hand side together with three additional free loops .
[ ptb ] lbracfk1.ps
cohn and lempel @xcite described the number of orbits in a finite set under a certain kind of permutation using the nullity of an associated binary matrix . when applied to a connected 2-in , 2-out digraph @xmath1 their equality uses the nullity of a submatrix of an interlace matrix to calculate the number of circuits that appear in a given partition of the edge - set of @xmath1 into directed circuits .
this equality was implicitly rediscovered as part of the combinatorial theory of the interlace polynomial ( theorem 24 of @xcite or @xcite ) , and its usefulness in knot theory was pointed out by soboleva @xcite .
other equalities similar to that of cohn and lempel have been used by knot theorists : zulli @xcite used a formula that always refers to @xmath55 matrices , and a more general version was stated by mellor @xcite and lando @xcite .
an extended version of the cohn - lempel equality that includes all of these and applies to arbitrary circuit partitions in arbitrary 4-regular graphs is proven in @xcite .
let @xmath3 be an undirected 4-regular graph with @xmath53 , and let @xmath2 be an euler system for @xmath3 .
choose an orientation for each circuit in @xmath2 , and let @xmath1 be the 2-in , 2-out digraph obtained from @xmath3 by directing all edges according to these orientations .
suppose @xmath112 is a partition of @xmath36 into undirected circuits ; @xmath112 should also contain every free loop @xmath3 might have .
suppose that we follow a directed edge @xmath113 of @xmath1 toward a vertex @xmath45 .
if the circuit of @xmath112 that contains @xmath113 leaves @xmath45 along the edge @xmath2 uses to leave @xmath45 after arriving along @xmath113 , we say @xmath112 _ follows _
@xmath2 _ through _ @xmath45 .
if the circuit of @xmath112 that contains @xmath113 leaves @xmath45 along the other edge directed away from @xmath45 , we say @xmath112 is _ orientation - consistent at _
@xmath45 _ _ but does not follow _ _ @xmath2 .
the last possibility is that the circuit of @xmath112 that contains @xmath113 leaves @xmath45 along the other edge directed toward @xmath45 ; in this case we say @xmath112 is _ orientation - inconsistent at _ @xmath45 .
( note that the description of @xmath112 at @xmath45 is unchanged if we begin with the other edge directed toward @xmath45 , and it is not affected by the choice of orientations for the circuits of @xmath2 or @xmath112 . ) a matrix @xmath114 is associated to @xmath112 : if @xmath112 follows @xmath2 through @xmath45 then the row and column of @xmath54 corresponding to @xmath45 are removed ; if @xmath112 is orientation - consistent at @xmath45 but does not follow @xmath2 then the row and column of @xmath54 corresponding to @xmath45 are retained ; and if @xmath112 is orientation - inconsistent at @xmath45 then the off - diagonal entries of the row and column of @xmath54 corresponding to @xmath45 are retained but their common diagonal entry is changed from @xmath102 to @xmath115 .
( extended cohn - lempel equality ) let @xmath3 be an undirected , 4-regular graph with @xmath60 connected components , and let @xmath112 be a partition of @xmath36 into undirected circuits ; @xmath112 must also contain every free loop of @xmath3 .
then the @xmath25-nullity of @xmath116 is @xmath117 we do not give a proof here ; the interested reader can find one in @xcite .
here is an example , though .
consider the complete bipartite graph @xmath118 , with vertices denoted 1 , 2 , 3 , 4 , 5 , 6 , 7 and 8 ; vertex @xmath119 is adjacent to vertex @xmath120 if and only if @xmath121 ( mod 2 ) .
let @xmath2 contain the euler circuit 1234567814725836 .
if @xmath112 follows @xmath2 at vertices 1 and 4 , is orientation - inconsistent at vertices 2 , 5 and 7 , and is orientation - consistent but does not follow @xmath2 at vertices 3 , 6 and 8 then @xmath122 so @xmath123 .
the one circuit in @xmath112 is the euler circuit 1278345236741856 .
the partition @xmath124 that disagrees with @xmath112 only by following @xmath2 at 2 and 6 corresponds to the matrix @xmath125 obtained by removing the first and fourth rows and columns of @xmath126 , so @xmath127 .
@xmath124 contains the circuits 1236 , 147658 and 254387 .
[ adjmatrix ] if @xmath35 is a graph with @xmath128 then the _ boolean adjacency matrix _ of @xmath35 is the @xmath55 matrix @xmath129 with the following entries in @xmath25 : @xmath130 if and only if @xmath45 is looped , and if @xmath58 then @xmath131 if and only if @xmath45 is adjacent to @xmath59 .
a _ marked graph _ is a graph @xmath35 given with a partition of @xmath38 into two subsets . in our discussion
we call the vertices in one cell of the partition _ unmarked _ , and vertices in the other cell _ marked _ ( or _ marked with the letter
_ @xmath32 ) .
[ adjt ] suppose @xmath35 is a marked graph with @xmath128 and @xmath132 .
let @xmath133 be the @xmath55 diagonal matrix whose @xmath134 diagonal entry is @xmath115 if and only if @xmath135 .
then @xmath136 is the submatrix of @xmath137 obtained by removing the @xmath134 row and column if @xmath45 is marked and the @xmath134 diagonal entry of @xmath137 is @xmath102 .
[ bracket ] the _ marked - graph bracket polynomial _ of a marked graph @xmath35 with @xmath138 free loops is @xmath139=d^{\phi}\cdot\sum_{t\subseteq v(g)}a^{n-\left\vert t\right\vert } b^{\left\vert t\right\vert } d^{\nu(\mathcal{a}(g)_{t})},\ ] ] where @xmath140 denotes the nullity of matrices with entries in @xmath25 .
a simple consequence of definitions [ adjt ] and [ bracket ] is that the marked - graph bracket is multiplicative on disjoint unions .
[ product ] if @xmath35 is the union of disjoint subgraphs @xmath141 and @xmath142 then @xmath143=[g_{1}]\cdot\lbrack g_{2}].$ ] @xmath35 has @xmath144 free loops . also , if @xmath132 and @xmath145 then @xmath146 so @xmath147 .
another simple property of the marked - graph bracket is that toggling loops in @xmath35 has the effect of reversing the roles of @xmath16 and @xmath148 in @xmath143 $ ] .
[ looptoggle ] let @xmath149 denote the marked graph obtained from @xmath35 by toggling all loops in @xmath35 ( i.e. , @xmath149 has loops at precisely those vertices where @xmath35 does not ) .
then @xmath150(a , b , d)=[g](b , a , d).$ ] if @xmath132 then @xmath151 , so the contribution of @xmath152 to @xmath143(b , a , d)$ ] is the same as the contribution of @xmath153 to @xmath150(a , b , d)$ ] .
recall that the kauffman bracket polynomial of a classical or virtual link diagram @xmath0 is a sum indexed by the states of @xmath0 : @xmath21=\sum_{s}a^{a(s)}b^{b(s)}d^{c(s)-1}.\ ] ] a kauffman state @xmath10 corresponds in an obvious way to a partition @xmath154 of @xmath36 into undirected circuits , with the proviso that every free loop of @xmath3 is included in both @xmath10 and @xmath154 .
if @xmath2 is a directed euler system for @xmath1 then we associate with each kauffman state @xmath10 of @xmath0 the subset @xmath155 consisting of the vertices corresponding to crossings where the state @xmath10 involves the @xmath13 smoothing .
observe that at an unmarked positive crossing , the @xmath13 smoothing corresponds to the transition that is orientation - inconsistent with @xmath2 , and the @xmath16 smoothing corresponds to the transition that is orientation - consistent with @xmath2 without following @xmath2 ; at an unmarked negative crossing the @xmath13 smoothing is orientation - consistent with @xmath2 without following it , and the @xmath16 smoothing is orientation - inconsistent with @xmath2 . at a positive crossing marked @xmath32 the @xmath13 smoothing
is orientation - inconsistent with @xmath2 and the @xmath16 smoothing follows @xmath2 ; at a negative crossing marked @xmath32 the @xmath13 smoothing follows @xmath2 and the @xmath16 smoothing is orientation - inconsistent with @xmath2 .
consequently if @xmath10 is a kauffman state of @xmath0 then @xmath156 is the same as the matrix denoted @xmath157 in the preceding section .
the extended cohn - lempel equality then tells us that @xmath158 as @xmath4 has @xmath159 free loops , we conclude the following . [ brackets ] suppose @xmath0 is a diagram of an oriented classical or virtual link @xmath5 , @xmath3 is the universe of @xmath0 and @xmath2 is a directed euler system for @xmath1 .
then the kauffman bracket polynomial of @xmath0 and the marked - graph bracket polynomial of @xmath4 are the same .
it follows that @xmath34 $ ] is independent of the choice of the particular directed euler system @xmath2 .
this independence actually follows from proposition [ ktransform ] , lemma [ transform ] and a more general result given in theorem [ marked pivot ] .
[ pivot ] suppose @xmath35 is a marked graph and @xmath107 are adjacent in @xmath35 .
the _ marked pivot _
@xmath160 is the marked graph obtained from @xmath106 by toggling the marks on @xmath68 and @xmath73 separately ( i.e. , removing a mark @xmath32 where there is one , and inserting a mark @xmath32 where there is none ) , and interchanging the neighbors of @xmath68 and @xmath73 . like an ordinary pivot
, a marked pivot has no effect on loops or free loops .
[ ptb ] lbracpiv.ps [ marked pivot ] suppose @xmath35 is a marked graph and @xmath107 are adjacent in @xmath35
. then @xmath143=[g_{c}^{vw}].$ ] for convenience we presume @xmath128 with @xmath161 and @xmath162 .
the proposition is proven by showing that the nullities of @xmath136 and @xmath163 are the same for every subset @xmath132 .
suppose the first two diagonal entries of @xmath137 are both 1 .
then @xmath164 for appropriate submatrices @xmath165 , where a bold numeral indicates a row or column whose entries are all equal . adding the first row of @xmath166 to every row in the third and fifth blocks of rows ( those containing @xmath167 and @xmath168 ) , and then adding the second row to every row in the third and fourth blocks of rows , we conclude that the nullity of @xmath136 is the same as that of @xmath169 which is @xmath163 up to permutation of the rows and columns .
suppose now that one of the first two diagonal entries of @xmath170 is 1 , and the other is 0 . if @xmath68 and @xmath73 are both unmarked then @xmath171 adding the first row of @xmath136 to every row in the third and fifth blocks of rows , and then adding the second row to every row in the third and fourth blocks , we conclude that the nullity of @xmath136 is the same as that of @xmath172 removing the first column and second row of this matrix does not change its nullity , and yields @xmath163 up to permutation of rows and columns . the argument can be reversed if @xmath68 and @xmath73 are both marked in @xmath35 , as they are then both unmarked in @xmath160 .
also , if precisely one of @xmath107 is marked in @xmath35 then the marked one will correspond to the diagonal entry 1 in one of @xmath173 and the diagonal entry 0 in the other of @xmath173 .
the argument just given applies again , possibly with the roles of @xmath35 and @xmath160 reversed .
suppose now that the first two diagonal entries of @xmath137 are both 0 , and suppose that precisely one of @xmath107 ( @xmath68 , say ) is marked in @xmath35
. then @xmath174 and @xmath175 for appropriate submatrices @xmath165 .
adding the first row of @xmath166 to every row in the second and third blocks of rows , and then adding the first column to every column in the second and third blocks of columns , we conclude that the nullity of @xmath136 is the same as that of @xmath176 which is @xmath163 up to a permutation of the rows and columns .
suppose now that the first two diagonal entries of @xmath137 are both 0 and that @xmath68 and @xmath73 are both unmarked in @xmath35
. then @xmath177 adding the first row of @xmath136 to every row in the third and fifth blocks of rows , and then adding the second row to every row in the third and fourth blocks , we conclude that the nullity of @xmath136 is the same as that of @xmath178 removing the first two rows and columns of this matrix does not change the nullity , and yields @xmath163 , up to a permutation of the rows and columns . if the first two diagonal entries of @xmath137 are both 0 and @xmath68 and @xmath73 are both marked the same argument applies , with the roles of @xmath35 and @xmath160 reversed .
theorem [ marked pivot ] will be helpful in the balance of the paper because it allows us to focus on a limited number of special cases .
for instance , we will see later that when applied to different configurations of marked vertices , the knot - theoretic reidemeister moves give rise to different marked - graph reidemeister moves .
theorem [ marked pivot ] allows us to ignore some of those configurations ( e.g. , ones involving adjacent marked vertices ) .
in this section we present a recursive description of the marked - graph bracket polynomial .
the recursion involves formulas that are only valid when certain vertices are without marked neighbors , so an arbitrary marked graph must be prepared for the recursion by using marked pivots to eliminate adjacencies between marked vertices .
[ tbh ] lbracl14b.ps if @xmath68 is a vertex of a ( marked ) graph @xmath35 then the _ local complement _
@xmath179 is obtained from @xmath35 by toggling every edge incident only on neighbor(s ) of @xmath68 .
recall that @xmath68 itself is not included among its neighbors , so edges incident on @xmath68 in @xmath35 are preserved in @xmath179 .
we should also point out that there is a different use of the term `` local complement , '' which is restricted to simple graphs and does not involve loops @xcite ; we follow @xcite instead , and toggle both loops and non - loop edges incident on neighbors of @xmath68 when constructing @xmath179 .
[ ptb ] lbraclc.ps [ recursion ] the marked - graph bracket polynomial of a marked graph @xmath35 can be calculated recursively using the following steps .
\(a ) if @xmath68 and @xmath73 are marked neighbors , replace @xmath35 with @xmath160 .
\(b ) suppose @xmath68 is unlooped and marked , and no neighbor of @xmath68 is marked .
then @xmath139=a[g - v]+b[g^{v}-v],\ ] ] where @xmath180 is obtained from @xmath35 by removing @xmath68 and every edge incident on @xmath68 .
\(c ) suppose @xmath68 is looped and marked , and no neighbor of @xmath68 is marked .
then @xmath139=b[g - v]+a[g^{v}-v].\ ] ] \(d ) if @xmath68 is looped and no neighbor of @xmath68 is marked , then @xmath139=a^{-1}b[g-\{v , v\}]+(a - a^{-1}b^{2})[g^{v}-v].\ ] ] here @xmath181 is obtained from @xmath35 by removing the loop at @xmath68 .
\(e ) let @xmath68 and @xmath73 be adjacent , unlooped , unmarked vertices .
if no neighbor of @xmath68 is marked then @xmath139=a^{2}[g^{vw}-v - w]+ab[(g^{vw})^{v}-v - w]+b[g^{v}-v].\ ] ] \(f ) suppose @xmath35 is a graph whose edges are all loops .
if @xmath35 has @xmath138 free loops , @xmath182 unlooped marked vertices , @xmath183 looped marked vertices , @xmath184 unlooped unmarked vertices and @xmath185 looped unmarked vertices then @xmath139=d^{\phi}(a+b)^{m_{1}+m_{2}}(ad+b)^{n_{1}}(a+bd)^{n_{2}}\text{.}%\ ] ] theorem [ marked pivot ] tells us that an application of part ( a ) does not change the marked - graph bracket polynomial , and the formula of part ( f ) follows directly from definition [ bracket ] .
as free loops are not affected by local complementation , pivots or marked pivots , the corresponding powers of @xmath186 can simply be factored out of the formulas of parts ( b)-(e ) ; hence it suffices to verify ( b)-(e ) for graphs without free loops . in part
( b ) @xmath68 is a marked , unlooped vertex with no marked neighbor .
let @xmath187 if @xmath188 then using elementary row operations , we see that @xmath189{ccc}% 1 & \mathbf{1 } & \mathbf{0}\\ \mathbf{1 } & m_{11 } & m_{12}\\ \mathbf{0 } & m_{21 } & m_{22}% \end{array } \right ) \\ & = \nu\left ( \begin{array } [ c]{ccc}% 1 & \mathbf{1 } & \mathbf{0}\\ \mathbf{0 } & \bar{m}_{11 } & m_{12}\\ \mathbf{0 } & m_{21 } & m_{22}% \end{array } \right ) = \nu\left ( \begin{array } [ c]{cc}% \bar{m}_{11 } & m_{12}\\ m_{21 } & m_{22}% \end{array } \right)\end{aligned}\ ] ] for appropriate matrices @xmath165 .
as @xmath68 has no marked neighbor , the toggling of diagonal entries in @xmath190 is compatible with definition [ adjt ] , and we conclude that @xmath191 this holds whenever @xmath192 , so @xmath193 $ ]
. as @xmath68 is marked , definition [ adjt ] has @xmath194 for every @xmath195 with @xmath196 .
it follows that @xmath143-s = a[g - v]$ ] .
part ( c ) follows from part ( b ) and proposition [ looptoggle ] .
we proceed to consider ( d ) .
suppose @xmath35 has a looped vertex @xmath68 whose neighbors are all unmarked .
if @xmath188 then @xmath197 and @xmath198 ; these equalities hold whether or not @xmath68 is marked .
let @xmath199 then @xmath143=ab^{-1}s+a^{-1}b\left ( [ g-\{v , v\}]-s\right ) $ ] , so @xmath143=a^{-1}% b[g-\{v , v\}]+(ab^{-1}-a^{-1}b)s$ ] . the nullity argument used in the proof of ( b ) shows that @xmath200=b[g^{v}-v]$ ] . turning to ( e ) , suppose that @xmath201 are adjacent , unlooped , unmarked vertices , and no neighbor of @xmath68 is marked .
let @xmath202 as @xmath68 has no marked neighbor the nullity argument given for part ( b ) applies ; it tells us that @xmath143-s_{1}-s_{2}=b[g^{v}-v]$ ] .
consider a subset @xmath132 with @xmath203 .
adding the first row to those in the third and fifth blocks of rows and adding the second row to those in the third and fourth blocks of rows , we see that @xmath189{cccccc}% 0 & 1 & \mathbf{1 } & \mathbf{1 } & \mathbf{0 } & \mathbf{0}\\ 1 & 0 & \mathbf{1 } & \mathbf{0 } & \mathbf{1 } & \mathbf{0}\\ \mathbf{1 } & \mathbf{1 } & m_{11 } & m_{12 } & m_{13 } & m_{14}\\ \mathbf{1 } & \mathbf{0 } & m_{21 } & m_{22 } & m_{23 } & m_{24}\\ \mathbf{0 } & \mathbf{1 } & m_{31 } & m_{32 } & m_{33 } & m_{34}\\ \mathbf{0 } & \mathbf{0 } & m_{41 } & m_{42 } & m_{43 } & m_{44}% \end{array } \right ) \\ & ~\\ & = \nu\left ( \begin{array } [ c]{cccccc}% 0 & 1 & \mathbf{1 } & \mathbf{1 } & \mathbf{0 } & \mathbf{0}\\ 1 & 0 & \mathbf{1 } & \mathbf{0 } & \mathbf{1 } & \mathbf{0}\\ \mathbf{0 } & \mathbf{0 } & m_{11 } & \bar{m}_{12 } & \bar{m}_{13 } & m_{14}\\ \mathbf{0 } & \mathbf{0 } & \bar{m}_{21 } & m_{22 } & \bar{m}_{23 } & m_{24}\\ \mathbf{0 } & \mathbf{0 } & \bar{m}_{31 } & \bar{m}_{32 } & m_{33 } & m_{34}\\ \mathbf{0 } & \mathbf{0 } & m_{41 } & m_{42 } & m_{43 } & m_{44}% \end{array } \right ) \\ & ~\\ & = \nu\left ( \begin{array } [ c]{cccc}% m_{11 } & \bar{m}_{12 } & \bar{m}_{13 } & m_{14}\\ \bar{m}_{21 } & m_{22 } & \bar{m}_{23 } & m_{24}\\ \bar{m}_{31 } & \bar{m}_{32 } & m_{33 } & m_{34}\\ m_{41 } & m_{42 } & m_{43 } & m_{44}% \end{array } \right ) = \nu(\mathcal{a}(g^{vw}-v - w)_{t}).\end{aligned}\ ] ] it follows that @xmath204 $ ] .
now , consider a subset @xmath132 with @xmath205 and @xmath206 .
adding the first row to those in the fourth and fifth blocks of rows , and the second row to those in the third and fourth blocks , we see that @xmath189{cccccc}% 0 & 1 & \mathbf{1 } & \mathbf{1 } & \mathbf{0 } & \mathbf{0}\\ 1 & 1 & \mathbf{1 } & \mathbf{0 } & \mathbf{1 } & \mathbf{0}\\ \mathbf{1 } & \mathbf{1 } & m_{11 } & m_{12 } & m_{13 } & m_{14}\\ \mathbf{1 } & \mathbf{0 } & m_{21 } & m_{22 } & m_{23 } & m_{24}\\ \mathbf{0 } & \mathbf{1 } & m_{31 } & m_{32 } & m_{33 } & m_{34}\\ \mathbf{0 } & \mathbf{0 } & m_{41 } & m_{42 } & m_{43 } & m_{44}% \end{array } \right ) \\ & ~\\ & = \nu\left ( \begin{array } [ c]{cccccc}% 0 & 1 & \mathbf{1 } & \mathbf{1 } & \mathbf{0 } & \mathbf{0}\\ 1 & 0 & \mathbf{1 } & \mathbf{0 } & \mathbf{1 } & \mathbf{0}\\ \mathbf{0 } & \mathbf{0 } & \bar{m}_{11 } & m_{12 } & \bar{m}_{13 } & m_{14}\\ \mathbf{0 } & \mathbf{0 } & m_{21 } & \bar{m}_{22 } & \bar{m}_{23 } & m_{24}\\ \mathbf{0 } & \mathbf{0 } & \bar{m}_{31 } & \bar{m}_{32 } & m_{33 } & m_{34}\\ \mathbf{0 } & \mathbf{0 } & m_{41 } & m_{42 } & m_{43 } & m_{44}% \end{array } \right ) \\ & ~\\ & = \nu\left ( \begin{array } [ c]{cccc}% \bar{m}_{11 } & m_{12 } & \bar{m}_{13 } & m_{14}\\ m_{21 } & \bar{m}_{22 } & \bar{m}_{23 } & m_{24}\\ \bar{m}_{31 } & \bar{m}_{32 } & m_{33 } & m_{34}\\ m_{41 } & m_{42 } & m_{43 } & m_{44}% \end{array } \right ) = \nu(\mathcal{a}((g^{vw})^{v}-v - w)_{t-\{w\}}).\end{aligned}\ ] ] the last equality is justified because @xmath68 has no marked neighbors in @xmath35 , so the toggling of diagonal entries in @xmath167 and @xmath207 is consistent with definition [ adjt ] .
it follows that @xmath208.$ ] one way to implement theorem [ recursion ] is to follow this simple outline : first use ( a ) to remove marks on adjacent vertices ; then use ( b ) and ( c ) to remove marked vertices ; then use ( d ) to remove loops ; and finally use ( e ) to remove the remaining edges .
typically , this will not result in the most efficient recursive calculation of the bracket polynomial of a given marked graph .
for one thing , an implementation of _ any _ branching algorithm can be simplified if one combines terms that correspond to isomorphic structures as they arise during the recursion .
also , in our case it is obviously convenient to use part ( f ) when @xmath35 has only isolated vertices , rather than first using parts ( c ) and ( d ) to deal with those vertices that happen to be looped .
more generally , proposition [ looptoggle ] will sometimes yield a more convenient graph ; for instance , if the vertices of @xmath35 are all looped it is certainly simpler to unloop all the vertices at once rather than using parts ( c ) and ( d ) of theorem [ recursion ] repeatedly .
in addition , an efficient computation would utilize proposition [ product ] .
the computational cost of an instance of part ( b ) , ( c ) , ( d ) or ( e ) varies according to the structure of the resulting graphs @xmath179 , @xmath106 and @xmath209 ; there are generally many choices of vertices at which parts ( b)-(e ) might be applied , and it is not easy to guess which of these choices might be most convenient .
finally , there are other identities which might be used in particular situations ( when two unmarked looped vertices are adjacent , when an unmarked looped vertex is adjacent to a marked unlooped vertex , etc . ) and might sometimes appear in more efficient computations than an implementation of theorem [ recursion ] .
two of these other identities are given in corollary [ reverse ] and proposition [ old recursion ] .
[ ptbh ] lbracfw.ps figure [ lbracfw ] illustrates a calculation of the kauffman bracket of a diagram of whitehead s link using theorem [ recursion ] .
the marked link diagram @xmath0 appears at the top left of the figure , with @xmath4 at the top right .
the two arrows directed downward from @xmath4 point to the two graphs that result when part ( d ) of theorem [ recursion ] is applied , with @xmath68 the only vertex of @xmath4 that is not adjacent to the marked vertex . part ( e ) of theorem [ recursion ] is then applied to the left - hand graph in the middle row of the figure , with @xmath68 again the only vertex of @xmath4 that is not adjacent to the marked vertex .
as indicated in the figure , the second term of the first step and the third term of the second step refer to the same graph @xmath210 .
the bracket polynomial of this graph is calculated using part ( c ) : @xmath211=a[(g^{v}-v)^{y}% -y]+b[g^{v}-v - y]$ ] , where @xmath72 is the marked vertex .
the result of the calculation is @xmath212=ab(ad+b)(a^{2}d+2ab+b^{2}d)+b^{2}(a+bd)(a^{2}d+2ab+b^{2}d)\\ + a(a(ad+b)^{3}+b(a^{3}d^{2}+3a^{2}bd+3ab^{2}+b^{3}d))\\ = d^{3}a^{5}+5a^{4}bd^{2}+10a^{3}b^{2}d+8a^{2}b^{3}+2a^{2}b^{3}d^{2}% + 5ab^{4}d+b^{5}d^{2}.\end{gathered}\ ] ] another calculation of the same bracket polynomial is indicated in figure [ lbracfw1 ] .
this time the first step is an application of part ( c ) , @xmath143=b[g - v]+a[g^{v}-v]$ ] with @xmath68 the marked vertex .
the second step is an application of part ( d ) , and the third step is an application of part ( e ) .
the result of the calculation is @xmath212=b[g - v]+a(ad+b)(a^{3}d^{2}+3a^{2}bd+3ab^{2}+b^{3}d)\\ = ba^{-1}b\cdot(a^{2}(ad+b)(a+bd)+ab(ad+b)^{2})\\ + ba^{-1}b\cdot b(a^{3}+3a^{2}bd+ab^{2}d^{2}+2ab^{2}+b^{3}d)\\ + b(a - a^{-1}b^{2})(a^{3}d^{2}+3a^{2}bd+3ab^{2}+b^{3}d)\\ + a(ad+b)(a^{3}d^{2}+3a^{2}bd+3ab^{2}+b^{3}d).\end{gathered}\ ] ] [ ptb ] lbracfw1.ps the recursive calculation of the kauffman bracket @xmath23=$ ] @xmath34 $ ] provided by theorem [ recursion ] has not appeared before , but the individual steps are all related to familiar properties of the kauffman bracket or jones polynomial .
the basic recursion of kauffman s bracket , @xmath23=a[d_{a}]+b[d_{b}]$ ] , is extended to marked graphs in parts ( b ) and ( c ) of theorem [ recursion ] .
as discussed in section 4 of @xcite , the formula of part ( d ) of theorem [ recursion ] corresponds to the jones polynomial s braid - plat formula ( denoted @xmath213 in @xcite ) and the kauffman bracket s switching formula ( denoted @xmath214 in @xcite ) , and the formula of part ( e ) corresponds to a double use of the basic recursion of the kauffman bracket , @xmath23=a^{2}[d_{aa}]+ab[d_{ab}]+b[d_{b}]$ ] .
in addition , theorem [ recursion ] implies the reversing property of the jones polynomial @xcite .
the proof is simple : consider a link diagram in which a certain component is incident on only one marked crossing , and apply corollary [ reverse ] to conclude that the kauffman bracket is not changed when the orientation of that component is reversed .
[ reverse]suppose @xmath68 is looped and marked in @xmath35 , and no neighbor of @xmath215 is marked .
then @xmath143=[g^{v}-\{v , v\}]$ ] .
this follows immediately from parts ( b ) and ( c ) of theorem [ recursion ] .
the recursion of theorem [ recursion ] is fundamentally different from the original recursive description of the classical jones polynomial @xcite , which involves reducing the unknotting number ( a quantity that is not generally well defined for virtual links ) .
the basic formula @xmath216 of @xcite corresponds to one of the other identities mentioned in the preceding section : [ old recursion ] if @xmath217 is looped and marked then @xmath139=ab^{-1}[g-\{v , v\}]+(b - a^{2}b^{-1})[g - v].\ ] ] if @xmath188 then @xmath218 and @xmath219 .
consequently , if @xmath220 then @xmath221=ab^{-1}s+a^{-1}b([g]-s)$ ] , so @xmath143=ab^{-1}[g-\{v , v\}]-(a^{2}% b^{-2}-1)s$ ] . as @xmath68 is marked , definition [ bracket ] tells us that @xmath222 $ ] .
observe that unlike the formulas given in theorem [ recursion ] , proposition [ old recursion ] allows @xmath68 to have marked neighbors .
if @xmath35 is a marked graph then the _ reduced marked - graph bracket _ @xmath224 is obtained from @xmath143 $ ] by the evaluations @xmath225 and @xmath226 .
[ 2movea ] suppose @xmath35 is a marked graph with two unmarked vertices @xmath68 and @xmath73 such that @xmath68 is looped , @xmath73 is not , and @xmath68 and @xmath73 are adjacent to the same vertices outside @xmath227
. then @xmath228 .
suppose for convenience that @xmath128 with @xmath161 and @xmath162 .
if @xmath229 then let @xmath230 , @xmath231 and @xmath232 . if @xmath68 and @xmath73 are not adjacent then using elementary row and column operations we see that @xmath189{cccc}% 1 & 0 & \mathbf{1 } & \mathbf{0}\\ 0 & 0 & \mathbf{1 } & \mathbf{0}\\ \mathbf{1 } & \mathbf{1 } & m_{11 } & m_{12}\\ \mathbf{0 } & \mathbf{0 } & m_{21 } & m_{22}% \end{array } \right ) = \nu\left ( \begin{array } [ c]{ccc}% 0 & \mathbf{1 } & \mathbf{0}\\ \mathbf{1 } & m_{11 } & m_{12}\\ \mathbf{0 } & m_{21 } & m_{22}% \end{array } \right ) , \\ \nu(\mathcal{a}(g)_{t_{1 } } ) & = \nu\left ( \begin{array } [ c]{cccc}% 0 & 0 & \mathbf{1 } & \mathbf{0}\\ 0 & 0 & \mathbf{1 } & \mathbf{0}\\ \mathbf{1 } & \mathbf{1 } & m_{11 } & m_{12}\\ \mathbf{0 } & \mathbf{0 } & m_{21 } & m_{22}% \end{array } \right ) = 1+\nu\left ( \begin{array } [ c]{ccc}% 0 & \mathbf{1 } & \mathbf{0}\\ \mathbf{1 } & m_{11 } & m_{12}\\ \mathbf{0 } & m_{21 } & m_{22}% \end{array } \right ) , \\
\nu(\mathcal{a}(g)_{t_{2 } } ) & = \nu\left ( \begin{array } [ c]{cccc}% 1 & 0 & \mathbf{1 } & \mathbf{0}\\ 0 & 1 & \mathbf{1 } & \mathbf{0}\\ \mathbf{1 } & \mathbf{1 } & m_{11 } & m_{12}\\ \mathbf{0 } & \mathbf{0 } & m_{21 } & m_{22}% \end{array } \right ) = \nu\left ( \begin{array } [ c]{cc}% m_{11 } & m_{12}\\ m_{21 } & m_{22}% \end{array } \right ) \text { \ and}\\ \nu(\mathcal{a}(g)_{t_{12 } } ) & = \nu\left ( \begin{array } [ c]{cccc}% 0 & 0 & \mathbf{1 } & \mathbf{0}\\ 0 & 1 & \mathbf{1 } & \mathbf{0}\\ \mathbf{1 } & \mathbf{1 } & m_{11 } & m_{12}\\ \mathbf{0 } & \mathbf{0 } & m_{21 } & m_{22}% \end{array } \right ) = \nu\left ( \begin{array } [ c]{ccc}% 0 & \mathbf{1 } & \mathbf{0}\\ \mathbf{1 } & m_{11 } & m_{12}\\ \mathbf{0 } & m_{21 } & m_{22}% \end{array } \right ) .\end{aligned}\ ] ] it follows that the image of @xmath233 under the evaluations @xmath225 and @xmath226 is 0 .
the image of the remaining term is the contribution of @xmath152 to the reduced bracket of @xmath234 , so @xmath235 .
if @xmath68 and @xmath73 are adjacent then @xmath189{cccc}% 1 & 1 & \mathbf{1 } & \mathbf{0}\\ 1 & 0 & \mathbf{1 } & \mathbf{0}\\ \mathbf{1 } & \mathbf{1 } & m_{11 } & m_{12}\\ \mathbf{0 } & \mathbf{0 } & m_{21 } & m_{22}% \end{array } \right ) = \nu\left ( \begin{array } [ c]{ccc}% 1 & \mathbf{1 } & \mathbf{0}\\ \mathbf{1 } & m_{11 } & m_{12}\\ \mathbf{0 } & m_{21 } & m_{22}% \end{array } \right ) , \\ \nu(\mathcal{a}(g)_{t_{1 } } ) & = \nu\left ( \begin{array } [ c]{cccc}% 0 & 1 & \mathbf{1 } & \mathbf{0}\\ 1 & 0 & \mathbf{1 } & \mathbf{0}\\ \mathbf{1 } & \mathbf{1 } & m_{11 } & m_{12}\\ \mathbf{0 } & \mathbf{0 } & m_{21 } & m_{22}% \end{array } \right ) = \nu\left ( \begin{array } [ c]{cc}% m_{11 } & m_{12}\\ m_{21 } & m_{22}% \end{array } \right ) , \\ \nu(\mathcal{a}(g)_{t_{2 } } ) & = \nu\left ( \begin{array } [ c]{cccc}% 1 & 1 & \mathbf{1 } & \mathbf{0}\\ 1 & 1 & \mathbf{1 } & \mathbf{0}\\ \mathbf{1 } & \mathbf{1 } & m_{11 } & m_{12}\\ \mathbf{0 } & \mathbf{0 } & m_{21 } & m_{22}% \end{array } \right ) = 1+\nu\left ( \begin{array } [ c]{ccc}% 1 & \mathbf{1 } & \mathbf{0}\\ \mathbf{1 } & m_{11 } & m_{12}\\ \mathbf{0 } & m_{21 } & m_{22}% \end{array } \right ) , \text { \ and}\\ \nu(\mathcal{a}(g)_{t_{12 } } ) & = \nu\left ( \begin{array } [ c]{cccc}% 0 & 1 & \mathbf{1 } & \mathbf{0}\\ 1 & 1 & \mathbf{1 } & \mathbf{0}\\ \mathbf{1 } & \mathbf{1 } & m_{11 } & m_{12}\\ \mathbf{0 } & \mathbf{0 } & m_{21 } & m_{22}% \end{array } \right ) = \nu\left ( \begin{array } [ c]{ccc}% 1 & \mathbf{1 } & \mathbf{0}\\ \mathbf{1 } & m_{11 } & m_{12}\\ \mathbf{0 } & m_{21 } & m_{22}% \end{array } \right ) .\end{aligned}\ ] ] the image of @xmath236 under the evaluations @xmath237 and @xmath226 is 0 , and the image of the remaining term is the contribution of @xmath152 to the reduced bracket of @xmath234 . [ 2moveb]suppose @xmath35 is a marked graph with two adjacent vertices @xmath68 and @xmath73 such that @xmath68 is looped and marked , @xmath73 is neither looped nor marked , and @xmath68 is the only neighbor of @xmath73 . if @xmath68 has no neighbor other than @xmath73 then @xmath238 , where @xmath239 is obtained from @xmath35 by adjoining a free loop .
otherwise , let @xmath240 be a neighbor of @xmath68 other than @xmath73
. then @xmath241 .
if @xmath68 has no neighbor other than @xmath73 then proposition [ product ] tells us that @xmath242\right\rangle \cdot\left\langle g - v - w\right\rangle $ ] , where @xmath243 $ ] is the subgraph of @xmath35 induced by @xmath227 .
a direct calculation shows that @xmath244=a^{2}% + 2abd+b^{2}$ ] , and hence @xmath245\right\rangle = -a^{2}-a^{-2}$ ] .
if @xmath68 has a neighbor @xmath240 other than @xmath73 then theorem [ marked pivot ] tells us that @xmath246 . in @xmath247 , @xmath68 and @xmath73
are adjacent , unmarked vertices , with @xmath68 looped and @xmath73 unlooped . when constructing @xmath247 we first construct @xmath248 , by toggling all adjacencies between vertices @xmath249 such that @xmath71 is adjacent to @xmath68 , @xmath72 is adjacent to @xmath240 and either @xmath71 is not adjacent to @xmath240 or @xmath72 is not adjacent to @xmath68 .
this includes toggling every adjacency between @xmath73 and a neighbor of @xmath240 other than @xmath68 .
as @xmath73 is adjacent to none of these vertices in @xmath35 , and not adjacent to any other vertex aside from @xmath68 , the neighbors of @xmath73 in @xmath248 are the same as the neighbors of @xmath240 . the construction of @xmath250 is completed by interchanging the neighbors of @xmath68 and @xmath240 , so the neighbors of @xmath73 in @xmath247 are the same as the neighbors of @xmath68 in @xmath247 .
proposition [ 2movea ] then tells us that @xmath251 .
[ 2movec ] suppose @xmath35 is a marked graph with two adjacent vertices @xmath68 and @xmath73 such that @xmath68 is looped and marked , @xmath73 is unlooped and unmarked , and @xmath68 and @xmath73 have precisely the same neighbors outside @xmath227 . if @xmath68 has no neighbor other than @xmath73 then @xmath252
otherwise , @xmath241 for any @xmath253 that is adjacent to @xmath68 .
if @xmath68 has no neighbor other than @xmath73 the preceding proposition applies .
suppose @xmath240 is a neighbor of @xmath68 other than @xmath73 ; then @xmath254 . in @xmath248 , the adjacency between @xmath73 and every vertex @xmath255 that is adjacent to only one of @xmath256 is toggled ; as @xmath73 has the same neighbors outside @xmath227 as @xmath68 has , it follows that @xmath73 has the same neighbors in @xmath248 as @xmath240 has .
the neighbors of @xmath68 and @xmath240 in @xmath248 are exchanged in @xmath247 , so proposition [ 2movea ] applies to @xmath68 and @xmath73 in @xmath247 . a _ _ marked - graph _ _ @xmath223 _ reidemeister move _ is constructed as follows . any finite sequence of marked pivots may be applied to a marked graph @xmath257 ; let @xmath35 denote the result .
@xmath35 is replaced with @xmath258 , @xmath259 or @xmath260 as in one of the three propositions above .
any finite sequence of marked pivots may then be applied .
the inverse of an @xmath223 move is also an @xmath223 move , as is the move obtained by first toggling all loops in @xmath257 , and then toggling all loops in the final graph .
proposition [ looptoggle ] and theorem [ marked pivot ] tell us that if @xmath261 is obtained from @xmath35 by using a marked - graph @xmath223 move to adjoin or remove two vertices , then @xmath262 . recall that if @xmath0 is a link diagram then an _ _
_ _ @xmath223 reidemeister move on @xmath0 involves replacing one of the two configurations pictured on the left or right of figure [ lbracfi7 ] with the configuration in the middle .
[ ptb ] lbracfi7.ps [ 2move ] if @xmath263 is obtained by applying an _ _ _ _ @xmath223 reidemeister move to a link diagram @xmath0 then there are euler systems @xmath2 and @xmath89 such that @xmath264 is obtained from @xmath4 by applying a marked - graph @xmath223 move .
suppose first that @xmath263 is obtained from @xmath0 by replacing the configuration on the left in figure [ lbracfi7 ] with the configuration that appears in the middle , and @xmath2 is an euler system for the directed universe @xmath1 of @xmath0 .
there are four possible arrangements of marks : either of @xmath107 could be marked , or neither , or both . in every one of these configurations the vertices of @xmath3 corresponding to @xmath68 and @xmath73
are interlaced with respect to @xmath2 , so if @xmath73 is marked we may replace @xmath2 with @xmath265 .
we proceed under the assumption that @xmath73 is not marked .
if neither @xmath68 nor @xmath73 is marked , the euler system @xmath2 must contain a circuit that leaves @xmath73 along the edge corresponding to the pictured arc @xmath113 , and whose next appearance in the pictured portion of @xmath3 occurs along the edge corresponding to the pictured arc @xmath266 .
then the crossings @xmath68 and @xmath73 correspond to adjacent , unmarked vertices @xmath267 such that such that @xmath68 is looped , @xmath73 is unlooped , and @xmath68 and @xmath73 are adjacent to the same vertices outside @xmath227 . removing @xmath68 and @xmath73 from @xmath4 is an instance of proposition [ 2movea ] , and results in @xmath268 , where @xmath89 is the euler system for @xmath263 obtained in the obvious way from @xmath2 . if @xmath68 is marked and @xmath73 is not , @xmath2 must instead contain a circuit @xmath269 that leaves the portion of @xmath0 pictured in figure [ lbracfi7 ] along @xmath113 and re - enters along @xmath270 . the crossings @xmath68 and @xmath73 again correspond to adjacent vertices @xmath267 such that @xmath68 is looped , @xmath73 is unlooped , and @xmath68 and @xmath73 are adjacent to the same vertices outside @xmath227 .
if @xmath269 passes through no crossing of @xmath0 only once on the way from @xmath113 to @xmath270 , then @xmath68 and @xmath73 have no neighbor in @xmath4 except each other .
removing @xmath68 and @xmath73 is then an instance of the first clause of proposition [ 2moveb ] , and @xmath271 where @xmath89 is the euler system for @xmath263 that contains the @xmath113-to-@xmath270 and @xmath272-to-@xmath266 portions of @xmath269 as separate circuits .
if @xmath269 passes through a crossing @xmath240 only once on the way from @xmath113 to @xmath270 , then lemma [ transform ] tells us that @xmath273 .
as @xmath68 and @xmath73 are both unmarked in @xmath274 , we may refer to the preceding paragraph .
suppose now that @xmath263 is obtained from @xmath0 by replacing the configuration on the right in figure [ lbracfi7 ] with the configuration in the middle .
it is impossible for both @xmath68 and @xmath73 to be marked , for this would result in an `` extra '' circuit incident on only @xmath68 and @xmath73 ; as @xmath2 is an euler system , it can not contain such a circuit .
if neither @xmath68 nor @xmath73 is marked , @xmath2 contains a circuit that leaves @xmath73 along the edge corresponding to @xmath272 and then re - enters the portion of @xmath0 pictured in figure [ lbracfi7 ] at @xmath113 .
then @xmath68 is looped , @xmath73 is unlooped , and @xmath107 are nonadjacent vertices with the same neighbors . removing @xmath68 and @xmath73 from @xmath4
is an instance of proposition [ 2movea ] , and the result is @xmath264 where @xmath89 is the euler system obtained from @xmath2 in the obvious way .
if @xmath68 is marked and @xmath73 is not then @xmath68 is the only neighbor of @xmath73 in @xmath4 . as in the third paragraph
, we have two cases .
if @xmath73 is the only neighbor of @xmath68 in @xmath4 then we refer to the first clause of proposition [ 2moveb ] .
@xmath275 where @xmath89 is obtained from @xmath2 by replacing the circuit @xmath269 that passes through @xmath68 and @xmath73 with two separate circuits , one including the @xmath270-to-@xmath113 portion of @xmath269 and the other including the @xmath272-to-@xmath266 portion of @xmath269 .
if instead @xmath68 has a neighbor @xmath253 in @xmath4 , then we replace @xmath2 by @xmath276 and refer to the preceding paragraph .
if @xmath73 is marked and @xmath68 is not then we toggle all loops , apply the preceding paragraph with the roles of @xmath68 and @xmath73 reversed , and then toggle all loops again .
[ 3movea ] suppose @xmath35 is a marked graph with three unmarked vertices @xmath278 such that @xmath278 are all adjacent to each other , @xmath279 is looped , @xmath68 and @xmath73 are unlooped , and every vertex @xmath280 is adjacent to either 0 or precisely two of @xmath278 .
let @xmath261 be the graph obtained from @xmath35 by removing all three edges @xmath281 , @xmath282 and @xmath227
. then @xmath283 .
each @xmath284 corresponds to eight subsets of @xmath38 , which make eight contributions to @xmath224 and eight contributions to @xmath285 . in each case , three of the eight cancel .
the proposition is true because the remaining five happen to equal each other . considering @xmath35 first , observe that the @xmath25-nullities of @xmath166 , @xmath286 and @xmath287 are ( respectively ) @xmath288{cccc}% 1 & 1 & 1 & \rho_{1}\\ 1 & 0 & 1 & \rho_{2}\\ 1 & 1 & 0 & \rho_{1}+\rho_{2}\\ \kappa_{1 } & \kappa_{2 } & \kappa_{1}+\kappa_{2 } & m \end{array } \right ) & = \nu\left ( \begin{array } [ c]{cccc}% 1 & 0 & 0 & \mathbf{0}\\ 0 & 1 & 0 & \rho_{1}+\rho_{2}\\ 0 & 0 & 1 & \rho_{1}\\ \mathbf{0 } & \kappa_{1 } & \kappa_{1}+\kappa_{2 } & m \end{array } \right ) , \\ \nu\left ( \begin{array } [ c]{cccc}% 0 & 1 & 1 & \rho_{1}\\ 1 & 0 & 1 & \rho_{2}\\ 1 & 1 & 0 & \rho_{1}+\rho_{2}\\ \kappa_{1 } & \kappa_{2 } & \kappa_{1}+\kappa_{2 } & m \end{array } \right ) & = \nu\left ( \begin{array } [ c]{cccc}% 0 & 0 & 0 & \mathbf{0}\\ 0 & 1 & 0 & \rho_{1}+\rho_{2}\\ 0 & 0 & 1 & \rho_{1}\\ \mathbf{0 } & \kappa_{1 } & \kappa_{1}+\kappa_{2 } & m \end{array } \right)\end{aligned}\ ] ] and @xmath289{cccc}% 0 & 1 & 1 & \rho_{1}\\ 1 & 1 & 1 & \rho_{2}\\ 1 & 1 & 0 & \rho_{1}+\rho_{2}\\ \kappa_{1 } & \kappa_{2 } & \kappa_{1}+\kappa_{2 } & m \end{array } \right ) = \nu\left ( \begin{array } [ c]{cccc}% 1 & 0 & 0 & \mathbf{0}\\ 0 & 1 & 0 & \rho_{1}+\rho_{2}\\ 0 & 0 & 1 & \rho_{1}\\ \mathbf{0 } & \kappa_{1 } & \kappa_{1}+\kappa_{2 } & m \end{array } \right)\ ] ] for appropriate row vectors @xmath290 , column vectors @xmath291 and a submatrix @xmath292 . consequently @xmath293 ; it follows that the contributions of @xmath152 , @xmath294 and @xmath295 to @xmath296 cancel each other . similarly , the contributions of @xmath152 , @xmath294 and @xmath295 to @xmath285 all cancel each other .
the @xmath25-nullities of @xmath297 and @xmath298 are @xmath288{cccc}% 1 & 1 & 1 & \rho_{1}\\ 1 & 1 & 1 & \rho_{2}\\ 1 & 1 & 0 & \rho_{1}+\rho_{2}\\ \kappa_{1 } & \kappa_{2 } & \kappa_{1}+\kappa_{2 } & m \end{array } \right ) & = \nu\left ( \begin{array } [ c]{cccc}% 1 & 0 & 0 & \mathbf{0}\\ 0 & 0 & 0 & \rho_{1}+\rho_{2}\\ 0 & 0 & 1 & \mathbf{0}\\ \mathbf{0 } & \kappa_{1}+\kappa_{2 } & 0 & m \end{array } \right ) \text { \
and}\\ \nu\left ( \begin{array } [ c]{cccc}% 1 & 0 & 0 & \rho_{1}\\ 0 & 1 & 0 & \rho_{2}\\ 0 & 0 & 0 & \rho_{1}+\rho_{2}\\ \kappa_{1 } & \kappa_{2 } & \kappa_{1}+\kappa_{2 } & m \end{array } \right ) & = \nu\left ( \begin{array } [ c]{cccc}% 0 & 1 & 0 & \mathbf{0}\\ 1 & 0 & 0 & \mathbf{0}\\ 0 & 0 & 0 & \rho_{1}+\rho_{2}\\ \mathbf{0 } & \mathbf{0 } & \kappa_{1}+\kappa_{2 } & m \end{array } \right)\end{aligned}\ ] ] respectively , so the contributions of @xmath299 to @xmath296 and @xmath285 are equal . similarly , @xmath300 makes equal contributions to @xmath296 and @xmath285 , and so does @xmath301
. also , the @xmath25-nullities of @xmath302 and @xmath303 are @xmath288{cccc}% 0 & 1 & 1 & \rho_{1}\\ 1 & 0 & 1 & \rho_{2}\\ 1 & 1 & 1 & \rho_{1}+\rho_{2}\\ \kappa_{1 } & \kappa_{2 } & \kappa_{1}+\kappa_{2 } & m \end{array } \right ) & = \nu\left ( \begin{array } [ c]{cccc}% 0 & 1 & 0 & \rho_{1}\\ 1 & 0 & 0 & \rho_{2}\\ 0 & 0 & 1 & \mathbf{0}\\ \kappa_{1 } & \kappa_{2 } & \mathbf{0 } & m \end{array } \right ) \text { \ and}\\ \nu\left ( \begin{array } [ c]{cccc}% 1 & 0 & 0 & \rho_{1}\\ 0 & 1 & 0 & \rho_{2}\\ 0 & 0 & 1 & \rho_{1}+\rho_{2}\\ \kappa_{1 } & \kappa_{2 } & \kappa_{1}+\kappa_{2 } & m \end{array } \right ) & = \nu\left ( \begin{array } [ c]{cccc}% 0 & 0 & 1 & \rho_{1}\\ 0 & 1 & 0 & \rho_{2}\\ 1 & 0 & 0 & \mathbf{0}\\ \mathbf{0 } & \kappa_{1 } & \kappa_{2 } & m \end{array } \right)\end{aligned}\ ] ] respectively , so the contributions of @xmath304 to @xmath296 and @xmath305 to @xmath285 are equal . similarly , the contributions of @xmath304 to @xmath306 and @xmath305 to @xmath224 are equal .
[ 3move ] a _ _ marked - graph _ _ @xmath277 _ reidemeister move _ is obtained by composing one of the moves described in proposition [ 3movea ] with any finite sequence of marked pivots and @xmath223 reidemeister moves .
the inverse of an @xmath277 move is also an @xmath277 move , as is the move obtained by first toggling all loops in the original graph , and then toggling all loops in the final graph .
theorem [ marked pivot ] , proposition [ 3movea ] and the results of section 8 imply that @xmath277 moves preserve the reduced marked - graph bracket . including composition with @xmath223 moves in definition [ 3move ]
is suggested by st - lund s observation that different - looking @xmath277 moves are interrelated through composition with @xmath223 moves @xcite .
the unmarked @xmath277 moves used in @xcite ( see figure [ lbracf8a ] ) are interrelated in the same way , but composition with @xmath307 moves was not useful in @xcite because the proof of the invariance of the reduced graph bracket under @xmath277 moves used induction on the number of vertices , and @xmath223 moves may increase the number of vertices .
moreover the inductive argument used to verify invariance under one @xmath277 move actually required invariance under different @xmath277 moves on smaller graphs .
the arguments we present here do not involve induction on the number of vertices , so stlund s observation is quite useful .
[ t ] lbracf8a.ps [ 3moveb ] suppose @xmath35 is a marked graph with three unmarked vertices @xmath278 such that the induced subgraph @xmath308 $ ] is isomorphic to one of the six 3-vertex graphs pictured in figure [ lbracf8a ] , and every vertex @xmath280 is adjacent to either 0 or precisely two of @xmath278 .
then there is a marked - graph @xmath277 reidemeister move whose effect is to toggle all the non - loop edges among @xmath278 .
the instance of the proposition involving the 3-vertex graph that appears at the top left of figure [ lbracf8a ] is verified in proposition [ 3movea ] . as was observed in @xcite ,
the other two instances that appear in the top row of figure [ lbracf8a ] can be obtained from this one through composition with @xmath223 moves .
( the same constructions given in @xcite apply here , because the @xmath223 moves of proposition [ 2movea ] coincide locally with the unmarked @xmath223 moves of @xcite , and do not require any special assumptions regarding the marked vertices of @xmath35 . )
the three instances pictured in the bottom row of figure [ lbracf8a ] follow , for if @xmath35 contains such an induced subgraph then we may toggle all loops in @xmath35 , apply the corresponding move pictured in the top row of figure [ lbracf8a ] , and then toggle all loops in the resulting graph .
corollary [ 3moveb ] states only that the moves illustrated in figure [ lbracf8a ] are examples of marked - graph @xmath277 moves .
as definition [ 3move ] includes composition with @xmath223 moves and marked pivots , it is not possible to give an exhaustive list of @xmath277 moves . if @xmath0 is a link diagram then an _ _ _ _ @xmath277 reidemeister move involves replacing a configuration like one of those in figure [ lbracfi9 ] with a configuration like the other .
the arcs appearing in the figure may be oriented in any way , and one may replace both configurations by their mirror images .
[ h ] lbracfi9.ps if @xmath263 is obtained by applying an _ _ _ _ @xmath277 reidemeister move to a link diagram @xmath0 then there are euler systems @xmath2 and @xmath89 such that @xmath264 is obtained from @xmath33 by applying a marked - graph @xmath277 move or a finite ( possibly empty ) sequence of marked pivots .
let @xmath2 be an euler system for the directed universe @xmath1 of the diagram @xmath0 which contains the left - hand side of figure [ lbracfi9 ] , and let @xmath263 be the link diagram that differs from @xmath0 only by containing the right - hand side of figure [ lbracfi9 ] .
suppose first that the crossings are all unmarked .
if the horizontal arc in the left - hand portion of figure [ lbracfi9 ] is oriented from left to right and the other two arcs are oriented from top to bottom , then the fact that the circuit of @xmath2 that passes through this portion of @xmath0 is an euler circuit implies that it is either @xmath309 or @xmath310 ; the euler circuit in the right - hand portion of figure [ lbracfi9 ] that arises naturally from @xmath2 will then be @xmath311 or @xmath312 , respectively .
all three crossings are positive , and each crossing outside the figure is interlaced with either 0 or precisely 2 of these ; for instance a crossing that appears once in @xmath96 and once in @xmath97 is interlaced with @xmath279 and @xmath73 in the first case , and @xmath68 and @xmath73 in the second case .
moreover each crossing in the left - hand portion of the figure is interlaced with the same outside crossings as the crossing on the right that has the same label ( @xmath279 , @xmath68 or @xmath73 ) .
it follows that the pictured @xmath277 move is the one that appears in the middle of the top row of figure [ lbracf8a ] .
if figure [ lbracfi9 ] is replaced by its mirror image and the orientations are the same as in the preceding paragraph , the same argument holds with all loops toggled . according to lemma 3 of @xcite , every other @xmath277 move represented by some oriented version of figure [ lbracfi9 ] or
its mirror image is a composition of @xmath223 moves and one of the two moves just discussed .
the result then follows from the two cases just discussed and proposition [ 2move ] .
suppose now that the portion of @xmath0 pictured on the left - hand side of figure [ lbracfi9 ] has a single marked vertex @xmath313 .
if @xmath71 is interlaced with any crossing @xmath72 outside the pictured portion of @xmath0 , then we may replace @xmath2 with @xmath314 , which has no marked vertices in the pictured portion of @xmath0 .
it remains to consider the possibility that precisely one of @xmath278 is marked and the marked element of @xmath315 is not interlaced with any crossing outside @xmath315 .
once again there are several special cases , but we need only discuss one explicitly .
for instance , suppose that the left - hand side is oriented as in the first paragraph of the proof , @xmath68 is marked and @xmath316 are not .
the circuit of @xmath2 incident on the left - hand side of figure [ lbracfi9 ] is @xmath317 or @xmath318 ; the obvious euler system for the right - hand side of figure [ lbracfi9 ] then includes the incident circuit @xmath319 or @xmath320 , respectively .
the assumption that @xmath68 has no neighbor outside @xmath282 implies that no crossing appears precisely once within @xmath96 ; it follows that @xmath268 is actually identical to @xmath4 .
if only @xmath73 is marked then the circuit of @xmath2 incident on the left - hand side of figure [ lbracfi9 ] is @xmath321 or @xmath322 , and the obvious euler system for the right - hand side of figure [ lbracfi9 ] includes the incident circuit @xmath323 or @xmath324 , respectively .
the assumption that @xmath73 has no neighbor outside @xmath281 implies that no crossing appears precisely once within @xmath96 ; it follows again that @xmath264 is identical to @xmath33 .
finally , if @xmath279 is the lone marked vertex in the left - hand portion of figure [ lbracfi9 ] then the incident circuit of @xmath2 is either @xmath325 ( in which @xmath279 is interlaced with @xmath73 ) or @xmath309 ( in which @xmath279 is interlaced with @xmath68 ) .
if @xmath279 is interlaced with @xmath68 then @xmath326 will have @xmath68 marked instead of @xmath279 , and if @xmath279 is interlaced with @xmath73 then @xmath327 will have @xmath73 marked instead of @xmath279 ; either way a transposition leads to one of the first two cases . as before ,
the mirror image is handled by toggling all loops , and other oriented versions of the left - hand side of figure [ lbracfi9 ] ( or its mirror image ) with precisely one marked vertex are handled by expressing the corresponding moves as compositions of @xmath223 moves with one of the two @xmath277 moves already considered .
suppose now that the portion of @xmath0 pictured on the left - hand side of figure [ lbracfi9 ] includes precisely two marked vertices .
if a marked crossing @xmath313 is interlaced with any crossing @xmath72 outside the pictured portion of @xmath0 , then we may replace @xmath2 with @xmath314 , which has only one marked vertex in the pictured portion of @xmath0 ; hence we may assume no marked crossing among @xmath315 is interlaced with any crossing outside @xmath315 .
the hypothesis that every crossing outside @xmath315 is interlaced with an even number of elements of @xmath315 then allows us to assume that no crossing outside @xmath315 is interlaced with any of @xmath278 .
moreover , if the two marked crossings are interlaced then the marks are removed by replacing @xmath2 with the euler system obtained by a transposition on these two vertices , so we may proceed assuming they are not interlaced .
suppose again that the left - hand side of figure [ lbracfi9 ] is oriented as above .
if @xmath328 are the two crossings that are marked then the incident euler circuit of @xmath2 must be @xmath329 .
there is then an euler circuit @xmath312 incident on the right - hand portion of figure [ lbracfi9 ] ; let @xmath89 be the euler system that coincides with @xmath2 except for this replacement of circuits incident on the pictured portions of @xmath0 and @xmath263 .
the assumption that no outside crossing is interlaced with any of @xmath278 with respect to @xmath2 tells us that every outside crossing appears twice in one @xmath330 ; hence no outside crossing is interlaced with any of @xmath278 with respect to @xmath89 either .
considering @xmath331 , we see that @xmath4 has a connected component whose only vertices are @xmath279 , @xmath68 and @xmath73 ; @xmath279 and @xmath68 are marked , nonadjacent neighbors of @xmath73 . considering @xmath312 , we see that @xmath264 has a connected component whose only vertices are @xmath279 , @xmath68 and @xmath73 ; @xmath279 and @xmath73 are marked , nonadjacent neighbors of @xmath68 .
that is , @xmath332 .
if @xmath279 and @xmath73 are the two that are marked then the euler circuit of @xmath2 incident of the left - hand portion of figure [ lbracfi9 ] must be @xmath309 ; there is then an euler circuit @xmath333 incident on the right - hand portion . if @xmath89 is the euler system that differs from @xmath2 only by including this latter circuit then @xmath334 .
finally , if @xmath68 and @xmath73 are the two that are marked then the euler circuit of @xmath2 incident on the left - hand portion of figure [ lbracfi9 ] must be @xmath335 ; hence an euler circuit incident on the right - hand portion is @xmath323 . if @xmath89 is the euler system that differs from @xmath2 only by including this latter circuit then @xmath336 .
finally , suppose all three of @xmath278 are marked with respect to @xmath2 .
any adjacency involving two of them allows a marked pivot that will eliminate two of the three marks , so we may assume no two of @xmath278 are interlaced with respect to @xmath2 .
this is not possible .
suppose @xmath0 and @xmath263 are diagrams of the same oriented virtual link type , with directed universes @xmath1 and @xmath337 .
let @xmath2 and @xmath89 be directed euler systems for @xmath1 and @xmath337 . then @xmath4 and @xmath264 can be transformed into each other through some finite sequence of marked pivots , marked - graph @xmath223 and @xmath277 reidemeister moves , and adjunctions and deletions of isolated , unmarked vertices .
@xmath0 and @xmath263 can be transformed into each other through some finite sequence of reidemeister moves of types @xmath338 , @xmath223 and @xmath277 and virtual reidemeister moves .
@xmath338 moves correspond to adjunctions and deletions of isolated , unmarked vertices .
virtual reidemeister moves have no effect on @xmath4 .
a _ reidemeister move _ on a marked graph is one of the following .
if @xmath35 has no free loops and no marked vertices then the bracket , reduced bracket and jones polynomial of @xmath35 coincide with those discussed in @xcite .
much of the discussion of @xcite is therefore still relevant . in particular , several important aspects of the classical theory of the jones polynomial
do not extend to this context ; for instance , we can not expect to calculate @xmath344 using a recursive algorithm that reduces @xmath35 to ungraphs in the same way that the jones polynomial of a classical link can be calculated by repeatedly using theorem 12 of @xcite ( the correct version of which is the formula @xmath216 ) to reduce @xmath5 to unlinks .
if @xmath346 consists of one isolated , unmarked , unlooped vertex then the disjoint union @xmath347 has @xmath348=(ad+b)\cdot\lbrack g]$ ] , so @xmath349 and @xmath350 , where @xmath14 and @xmath342 refer to @xmath35 .
a similar calculation shows that adjoining a single isolated , unmarked , looped vertex does not change @xmath351 .
an @xmath223 move has no effect on @xmath224 .
it adds @xmath352 or @xmath353 to the number of vertices while adding @xmath115 or @xmath354 ( resp . ) to the number of loops , so it also has no effect on @xmath355 .
in closing we observe that a special feature of the marked - graph bracket is that both definitions ( the nullity formula and the recursion ) are `` local '' in the sense that information from different connected components of @xmath35 is processed separately .
consequently there are multivariable forms of the marked - graph bracket and jones polynomials , in which different connected components of @xmath35 are associated with different sets of variables . for classical knots ,
disconnected interlacement graphs arise from connected sums , which are well understood @xcite ; on the other hand , the connected sum of virtual knot types is not uniquely defined @xcite .
whether there is any useful application of the multivariable bracket and jones polynomials remains to be seen .
99 r. arratia , b. bollobs , and g. b. sorkin , the interlace polynomial : a new graph polynomial , in _ proceedings of the eleventh annual acm - siam symposium on discrete algorithms _ ( association for computing machinery , new york , 2000 ) , pp .
237 - 245 .
r. c. read and p. rosenstiehl , on the gauss crossing problem , in _ combinatorics _ , keszthely ( 1976 ) , vol .
ii , colloq . math
jnos bolyai , * 18 * ( north - holland , amsterdam - new york , 1978 ) , pp . | let @xmath0 be an oriented classical or virtual link diagram with directed universe @xmath1 .
let @xmath2 denote a set of directed euler circuits , one in each connected component of @xmath3 .
there is then an associated looped interlacement graph @xmath4 whose construction involves very little geometric information about the way @xmath0 is drawn in the plane ; consequently @xmath4 is different from other combinatorial structures associated with classical link diagrams , like the checkerboard graph , which can be difficult to extend to arbitrary virtual links .
@xmath4 is determined by three things : the structure of @xmath1 as a 2-in , 2-out digraph , the distinction between crossings that make a positive contribution to the writhe and those that make a negative contribution , and the relationship between @xmath2 and the directed circuits in @xmath1 arising from the link components ; this relationship is indicated by marking the vertices where @xmath2 does not follow the incident link component(s ) .
we introduce a bracket polynomial for arbitrary marked graphs , defined using either a formula involving matrix nullities or a recursion involving the local complement and pivot operations ; the marked - graph bracket of @xmath4 is the same as the kauffman bracket of @xmath0 .
this provides a unified combinatorial description of the jones polynomial that applies seamlessly to both classical and non - classical virtual links
. _ keywords .
_ graph , virtual link , kauffman bracket , jones polynomial , euler circuit , circuit partition , trip matrix , interlacement , reidemeister move _ 2000 mathematics subject classification .
_ 57m25 , 05c50 |
partitioning of a linear system for its parallel solution typically aims at satisfying the two standard objectives : _ minimizing the communication volume _ and _ maintaining the load balance _ among different processors .
both of these requirements are motivated by considerations of the efficiency of the parallel matrix - vector multiplications , which lie in the heart of the iterative solution methods .
once the partitioning is performed , the obtained partitions are further used for constructing parallel preconditioners another crucial ingredient , contributing into the overall performance of the computational scheme .
however , the quality of the resulting preconditioner may depend significantly on the given partition , which , while targeting the efficiency of the parallel matrix - vector multiplication , ignores the nature of the employed preconditioning strategy .
the latter often leads to preconditioners of a poor quality , especially in the cases , where the coefficient matrices have entries with large variations in magnitudes . in the current work ,
we suggest to remove the requirement on the communication volume and , instead , consider partitionings , which favor the quality of the resulting preconditioner .
in particular , we focus on the additive schwarz ( as ) preconditioners , see , e.g. , @xcite , for symmetric positive definite ( spd ) linear systems , and present a partitioning algorithm , which aims at optimizing the quality of the as procedure by attempting to minimize the condition number of the preconditioned matrix , while maintaining the load balance . the problem of partitioning of a linear system @xmath0 is commonly formulated in terms of the adjacency graph @xmath1 of the coefficient matrix @xmath2 . here
, @xmath3 is the set of vertices ( nodes ) corresponding to the equations / unknowns of the linear system , and @xmath4 is the set of edges @xmath5 , where @xmath6 iff @xmath7 . throughout
, we assume that @xmath8 is spd , i.e. , @xmath9 , which , in particular , implies that the graph @xmath10 is undirected . the _ standard _ goal is to partition @xmath10 into @xmath11 `` nonoverlapping '' subgraphs @xmath12 , where @xmath13 and @xmath14 , such that @xmath15 imposing the additional constraint that the edge cut between @xmath16 is kept to a minimum , while the cardinalities of the vertex sets @xmath17 are approximately the same , i.e. , @xmath18 .
equations and unknowns with numbers in @xmath17 are then typically mapped to the same processor .
the requirement on the small edge cut between @xmath16 aims at reducing the cost of communications coming from the parallel matrix - vector multiplication .
the condition @xmath18 attempts to ensure the load balancing .
the solution of the above - described graph partitioning problem is np - complete .
however , there exist a variety of heuristics , which have been successfully applied to the problem ; see , e.g. , @xcite for an overview .
efficient implementations of relevant algorithms are delivered by a number of graph partitioning software packages , e.g. , chaco @xcite and metis @xcite .
we note that alternative approaches for partitioning of linear systems are known , e.g. , based on bipartite graph @xcite or hypergraph model @xcite , however , we do not consider them in this paper .
if the preconditioner quality becomes an objective of the partitioning , then along with the adjacency graph @xmath10 , it is reasonable to consider weights @xmath19 assigned to the edges @xmath20 , where @xmath19 are determined by the coefficients of the matrix @xmath8 .
the corresponding algorithm should then be able to take these weights into account and properly use them to perform graph partitioning .
an example of such an algorithm has been discussed in @xcite . indeed ,
if one considers partitioning as an `` early phase '' of a preconditioning procedure ( which , in the purely algebraic setting , is based solely on the knowledge of @xmath8 ) , then the use of the coefficients of @xmath8 at the partitioning step , e.g. , through the weights @xmath19 , represents a natural option .
this approach , however , faces a number of issues .
for example , given a preconditioning strategy , how does one assign the weights ?
what are the proper partitioning objectives
? how can the partitioning be performed in practice ? in this work ,
we address these three question for the case of an spd linear system and a simple domain decomposition ( dd ) type preconditioner
the _ nonoverlapping _ as . in particular , for a given @xmath8 , the proposed approach is based on the idea of constructing a ( bi)partition , which attempts to minimize an upper bound on the condition number of the preconditioned coefficient matrix over all possible balanced ( bi)partitions .
the resulting algorithm relies on the computation of eigenvectors corresponding to the smallest eigenvalues of generalized eigenvalue problems , which simultaneously involve the weighted and standard graph laplacians .
although the formal discussion is focused on the case of the _ nonoverlapping _ as , we show numerically that , in practice , adding several `` layers '' of neighboring nodes to the obtained sets @xmath17 in ( [ eqn : partn ] ) leads to decompositions of @xmath21 , which provide a good quality of the _ overlapping _ as preconditioning . the paper is organized as following . in section [ sec : bdiag ] , we recall several known results concerning the block - diagonal preconditioning for spd matrices .
these results motivate the new partitioning scheme , presented in section [ sec : cbs_partn ] .
the relation of the introduced approach to the existing graph partitioning schemes is discussed in section [ sec : other_spectral ] .
finally , in section [ sec : numer ] , we report on a few numerical examples .
let us consider @xmath8 as a block @xmath22-by-@xmath22 matrix , i.e. , @xmath23 where the diagonal blocks @xmath24 and @xmath25 are square of size @xmath26 and @xmath27 , respectively ; the off - diagonal block @xmath28 is @xmath26-by-@xmath27 .
let @xmath29 be a block - diagonal preconditioner , @xmath30 where @xmath31 , @xmath32 .
the dimensions of @xmath33 and @xmath34 are same as those of @xmath24 and @xmath25 , respectively .
since both @xmath8 and @xmath29 are spd , the convergence of an iterative method for @xmath0 , such as , e.g. , the the preconditioned conjugate gradient method ( pcg ) , is fully determined by the spectrum of the preconditioned matrix @xmath35 . if no information on the exact location of eigenvalues of @xmath35 is available , then the worst - case convergence behavior of pcg is traditionally described in terms of the condition number @xmath36 of @xmath35 ; @xmath37 with @xmath38 and @xmath39 denoting the largest and the smallest eigenvalues of the preconditioned matrix , respectively
. the question which arises is how we can bound @xmath36 for an arbitrary @xmath8 and a block - diagonal @xmath29 .
the answer to this question is given , e.g. , in ( * ? ? ?
* chapter 9 ) .
below , we briefly state the main result . in the subsequent sections , we will need this statement to justify the new partitioning algorithm .
[ def : cbs ] let @xmath40 and @xmath41 be finite dimensional spaces , such that @xmath8 in ( [ eqn : bk_sys ] ) is partitioned consistently with @xmath40 and @xmath41 .
the constant @xmath42 where @xmath43 and @xmath44 are subspaces of the form @xmath45 is called the cauchy - bunyakowski - schwarz ( cbs ) constant . in ( [ eqn : cbsa ] )
, @xmath46 denotes the standard inner product .
we note that @xmath47 can be interpreted as a cosine of an angle between subspaces @xmath43 and @xmath44 .
thus , since , additionally , @xmath48 , it is readily seen that @xmath49 .
also we note that @xmath47 is the smallest possible constant satisfying the strengthened cauchy - schwarz - bunyakowski inequality @xmath50 which motivates its name .
the cbs constant @xmath47 , defined by ( [ eqn : cbsa ] ) , turns to play an important role in estimating the condition number @xmath36 for the class of spd matrices @xmath8 and block - diagonal preconditioners @xmath29 .
[ thm : a1 ] if @xmath51 and @xmath52 in ( [ eqn : bk_prec ] ) , and @xmath8 in ( [ eqn : bk_sys ] ) is spd , then @xmath53 the bound given by theorem [ thm : a1 ] is sharp . in what follows ,
we use the cbs constants to construct a new approach for graph partitioning .
given decomposition @xmath54 of the set @xmath55 ( possibly with overlapping @xmath17 ) , we consider the as preconditioning for an spd linear system @xmath56 .
the preconditioning procedure is given in algorithm [ alg : as ] .
by @xmath57 we denote a submatrix of @xmath8 located at the intersection of rows with indices in @xmath58 and columns with indices in @xmath17 .
similarly , @xmath59 denotes the subvector of @xmath60 , containing entries from positions @xmath17 .
[ alg : as ] input : a , r , @xmath61 .
output : @xmath62 . 1 . for @xmath63 , do 2 .
set @xmath64 , @xmath65 , and @xmath66 .
3 . solve @xmath67 .
4 . set @xmath68 .
5 . enddo 6 .
@xmath69 . in this section ,
we focus on the case , where sets ( subdomains ) @xmath61 are nonoverlapping , i.e. , ( [ eqn : partn ] ) holds .
algorithm [ alg : as ] then gives a nonoverlapping as preconditioner , which is a form of the block jacobi iteration .
indeed , let @xmath70 be a permutation matrix , which corresponds to the reordering of @xmath21 according to the partition @xmath61 , where the elements in @xmath71 are labeled first , in @xmath72 second , etc .
then the as preconditioner @xmath29 , given by algorithm [ alg : as ] , can be written in the matrix form as @xmath73 where @xmath74 .
thus , algorithm [ alg : as ] results in the block - diagonal , or block jacobi , preconditioner , up to a permutation of its rows and columns . in the following subsection
we define an optimal _ bipartitioning _ for algorithm [ alg : as ] with two nonoverlapping subdomains .
let us assume that @xmath75 and consider a _ bipartition _ of @xmath76 , where @xmath77 and @xmath78 are nonempty , such that @xmath79 the following theorem provides a relation between a given bipartition and the condition number of the preconditioned matrix .
let @xmath80 in ( [ eqn : partitij ] ) be a bipartition of @xmath21 , where @xmath81 .
let @xmath29 be the as preconditioner for linear system @xmath0 with an spd matrix @xmath8 , given by algorithm [ alg : as ] , with respect to the bipartition @xmath80 .
then , @xmath82 where @xmath83 the spaces @xmath84 and @xmath85 are the subspaces of @xmath86 with dimensions @xmath26 and ( @xmath87 ) , respectively , such that @xmath88 according to ( [ eqn : reorder ] ) , for the given bipartition @xmath89 in ( [ eqn : partitij ] ) , the preconditioner @xmath29 , constructed by algorithm [ alg : as ] , is of the form @xmath90 where @xmath91 , @xmath92 , and @xmath70 is a permutation matrix corresponding to the reordering of @xmath21 with respect to the partition @xmath89 . in particular , for any @xmath93 , the vector @xmath94 is such that @xmath95 , i.e. , the entries of @xmath93 with indices in @xmath96 become the first @xmath26 components of @xmath97 , while the entries with indices in @xmath98 get positions from @xmath99 through @xmath100 .
we observe that the condition number of the matrix @xmath35 is the same as the condition number of the matrix @xmath101 , where @xmath102 and @xmath103 in ( [ eqn : prec2x2 ] ) . indeed , since a unitary similarity transformation @xmath104 preserves the eigenvalues of @xmath35 , we have @xmath105 , where @xmath106 .
the matrix @xmath107 represents a symmetric permutation of @xmath8 with respect to the given bipartition @xmath89 , and , thus , can be written in the @xmath22-by-@xmath22 block form , @xmath108 where @xmath91 , @xmath92 , and @xmath109 . since @xmath107 is spd and the preconditioner @xmath103 in ( [ eqn : prec2x2 ] ) is the block diagonal of @xmath107
, we apply theorem [ thm : a1 ] to get the upper bound on the condition number @xmath110 , and hence bound ( [ eqn : kappa ] ) on @xmath36 , where , according to definition [ def : cbs ] , the cbs constant @xmath111 is given by @xmath112 the matrix @xmath113 defines the permutation that is the `` reverse '' of the one corresponding to @xmath70 .
thus , the substitution @xmath114 and @xmath115 leads to expression ( [ eqn : cbsc])([eqn : wij ] ) for @xmath116 , where the @xmath84 and @xmath85 contain vectors , which can have nonzero entries only at positions defined by @xmath96 or @xmath98 , respectively . while no reasonably simple expression for the condition number of @xmath36 is generally available , ( [ eqn : kappa])([eqn : wij ] ) provides a sharp upper bound .
this suggests that instead of choosing the bipartition which directly minimizes @xmath36 , we look for @xmath80 , such that the _ upper bound _ in ( [ eqn : kappa ] ) is the smallest . the function @xmath117 is monotonically increasing .
therefore , the optimal value of @xmath118 in ( [ eqn : kappa ] ) is attained for a bipartition @xmath80 corresponding to the smallest value of the cbs constant @xmath116 .
since the targeted partitions are also required to be balanced , throughout this section , we request @xmath100 to be even , which guarantees the existence of _ fully balanced _
bipartitions @xmath119 in ( [ eqn : partitij ] ) , i.e. , such that @xmath120 .
the latter , however , will not be a restriction for the practical algorithm described below .
thus , we suggest an optimal bipartition @xmath121 for the nonoverlapping as preconditioner to be such that @xmath122 where @xmath84 and @xmath85 are the subspaces defined in ( [ eqn : wij ] ) . in the previous subsection
, we have shown that the cbs constant @xmath123 in ( [ eqn : cbsc ] ) provides a reasonable quantification of the quality of the bipartition @xmath89 in terms of the nonoverlapping as preconditioner with respect to the two subdomains . however
, finding an optimal bipartition @xmath121 ( possibly not unique ) according to ( [ eqn : opt ] ) , represents a challenging task .
therefore , we further construct bipartitions which _ attempt to approximate _ the minimizer @xmath124 of ( [ eqn : opt ] ) , rather than determine it exactly . in particular , we use the following approach . let @xmath125 , with @xmath116 in ( [ eqn : cbsc ] ) , be the objective function defined on all possible fully balanced bipartitions @xmath80 in ( [ eqn : partitij ] ) .
let @xmath126 be some other ( simpler ) function , which behaves similarly to @xmath127 , i.e. , the values of @xmath128 and @xmath116 change compatibly with respect to different bipartitions @xmath129 , and @xmath130 then , instead of @xmath131 , we attempt to minimize @xmath126 .
the constructed minimizer is used to define the bipartition for @xmath0 under the nonoverlapping as preconditioning , given in algorithm [ alg : as ] .
below , we suggest the choice of @xmath132 , and describe the resulting bipartitioning procedures . given a bipartition @xmath80 in ( [ eqn : partitij ] ) , @xmath120 , let us consider a set of pairs @xmath133 where @xmath134 denotes the unit vector with @xmath135 at position @xmath136 and zeros elsewhere . by ( [ eqn : cbsc ] )
, the computation of the cbs constant @xmath116 for this bipartition involves finding the maximum in @xmath137 and @xmath138 of the quantity @xmath139 instead , we suggest to evaluate ( [ eqn : qty ] ) on the pairs of the unit vectors @xmath140 in ( [ eqn : sample ] ) , and then find the mean of the values which result from this `` sampling '' , i.e. , define @xmath141 , such that @xmath142 we note that , in terms of the adjacency graph @xmath10 of @xmath8 , @xmath143 is equal to the edge cut between vertex sets @xmath96 and @xmath98 , i.e. , @xmath144 .
the expression above can be written as @xmath145 where @xmath146 is the weighted cut with @xmath147 denoting the weights assigned to the edges of @xmath10 .
thus , instead of the objective function @xmath125 , which according to ( [ eqn : opt ] ) , results in optimal bipartitions , we suggest to minimize @xmath126 in ( [ eqn : subopt ] ) , i.e. , find the minimizer @xmath148 of @xmath149 minimization ( [ eqn : min_subopt ] ) represents the problem of bipartitioning of the graph @xmath10 , which has the prescribed edge weights @xmath150 , with respect to the objective of minimizing the weighted cut normalized by the standard cut . since , by our assumption @xmath8 is spd , the weights @xmath19 are well - defined . the solution of ( [ eqn : min_subopt ] ) is then expected to approximate the optimal partition @xmath151 , i.e. , the minimizer of problem ( [ eqn : opt ] ) , which leads to the nonoverlapping as preconditioner of the optimal quality , in terms of minimizing the upper bound on the condition number of the preconditioned matrix .
let us reformulate optimization problem ( [ eqn : min_subopt ] ) in terms of bilinear forms involving graph laplacians .
first , we introduce the @xmath100-dimensional indicator vector @xmath152 with the components @xmath153 such that @xmath154 then , for the given bipartition @xmath80 , @xmath155 where @xmath156 is the weighted degree of the vertex @xmath157 ; @xmath158 denotes the set of vertices adjacent to the vertex @xmath157 .
the weighted cut @xmath159 can then be written as a bilinear form evaluated at the indicator vector @xmath152 , @xmath160 where @xmath161 is the weighted degree matrix , @xmath162 is the weighted adjacency matrix ( @xmath163 , @xmath164 ) of @xmath10 , and @xmath165 is the corresponding ( weighted ) graph laplacian .
similarly , for the same bipartition @xmath80 , we repeat the above derivations with @xmath166 to get the expression for the ( unweighted ) cut , i.e , @xmath167 where @xmath168 is the diagonal degree matrix , @xmath169 is the adjacency matrix ( @xmath170 , iff @xmath6 , and @xmath171 otherwise , @xmath172 ) of @xmath10 , and @xmath173 is the standard graph laplacian . using expressions ( [ eqn : wcut])([eqn : cut ] ) , minimization problem ( [ eqn : min_subopt ] ) can be written as @xmath174 where the minimum is searched over all possible indicator vectors @xmath152 .
the condition @xmath175 imposes the requirement that @xmath120 ; @xmath176 is @xmath100-vector of ones . in order to find an approximate solution of ( [ eqn : rq_discrete ] ) , we relax the requirement on @xmath152 to be the vectors of @xmath177 , and
embed the problem into the real space .
thus , instead of ( [ eqn : rq_discrete ] ) , we attempt to find @xmath178 , such that @xmath179 where @xmath165 and @xmath173 are the weighted and the standard graph laplacians of the adjacency graph of @xmath8 , respectively .
both @xmath165 and @xmath173 are symmetric positive semi - definite .
we assume for simplicity that the adjacency graph @xmath10 is connected , i.e. , the nullspace of @xmath173 is one - dimensional and is spanned by the vector @xmath180 .
we also note that @xmath180 is in the nullspace of @xmath165 .
problem ( [ eqn : rq ] ) then corresponds to the minimization of the generalized rayleigh quotient @xmath181 on the subspace @xmath182 .
since @xmath173 is spd on @xmath182 , the minimum in ( [ eqn : rq ] ) exists and is given by the smallest eigenvalue of the symmetric generalized eigenvalue problem on @xmath182 , @xmath183 the minimizer , i.e. , the eigenvector corresponding to the smallest eigenvalue of ( [ eqn : evp ] ) , can be viewed as an approximation to the minimizer of the discrete problem ( [ eqn : rq_discrete ] ) from the real vector space .
the solution of ( [ eqn : evp ] ) is delivered by eigensolvers , which can be applied to the problem , preferably , without factorizing the matrix @xmath173 , and which can be configured to perform iterations on the subspace @xmath182 . in our numerical tests in section [ sec : numer ] , we use the locally optimal block preconditioned conjugate gradient ( lobpcg ) method ; see @xcite .
a number of possible approaches can be used to define an indicator vector and , hence , the related partition , based on the given eigenvector @xmath184 corresponding to the smallest eigenvalue of problem ( [ eqn : evp ] ) .
for example , if the strict requirement on load balancing is enforced , i.e. , @xmath120 ( or @xmath185 if , in practice , @xmath100 is odd ) , then the set @xmath96 is formed by assigning the indices of @xmath186 smallest components of @xmath184 .
the indices of the remaining components form the set @xmath98 . in general
, however , the requirement on the full load balancing may be restrictive . for example , a slight imbalance in the cardinalities of @xmath96 and @xmath98 may lead to a significant improvement in the preconditioner quality .
in such cases , ignoring the explicit requirement on the load balancing , one can , e.g. , consider all the _ negative _ components of @xmath184 as approximations to @xmath187 , and assign their indices to the set @xmath96 .
the _ nonnegative _ components are then considered as approximations to @xmath135 , and their indices are put in @xmath98 .
thus , the sets @xmath96 and @xmath98 can be defined as @xmath188 similarly , @xmath96 and @xmath98 can be formed as @xmath189 where @xmath190 is a component of @xmath184 with the median value . a known generalization of this approach is to consider , say , @xmath191 `` candidate '' partitions @xmath192 , such that @xmath193 where the values @xmath194 are some chosen components of the vector @xmath195 , e.g. , @xmath196 ; @xmath197 .
the vector @xmath198 is obtained by sorting the eigenvector @xmath184 in the ascending order , and determines a linear search order .
all @xmath191 partitions @xmath192 are used to evaluate ( [ eqn : subopt ] ) .
the resulting bipartition @xmath80 is chosen to be the one , which delivers the smallest value of ( [ eqn : subopt ] ) , i.e. , corresponds to the minimizer of ( [ eqn : min_subopt ] ) over the @xmath191 `` candidate '' bipartitions in ( [ eqn : linorder_extr ] ) . in the partitioning algorithm below , we introduce a parameter _ loadbalance _ , which controls the sizes of the resulting sets @xmath96 and @xmath98 .
in particular , the parameter defines a smallest possible size , say @xmath199 , of the sets @xmath96 and @xmath98 , so that the indices of the @xmath26 smallest and @xmath26 largest components of the eigenvector @xmath184 are moved into the sets @xmath96 and @xmath98 , respectively .
the indices of the rest @xmath200 components are distributed among @xmath96 and @xmath98 similarly to ( [ eqn : linorder_extr ] ) , with @xmath201 , where @xmath202 and @xmath203 is a given parameter .
let @xmath80 be the bipartition of the set @xmath204 resulting from the approach , based on the eigenvalue problem ( [ eqn : evp ] ) , discussed in the previous subsection .
a natural way to construct further partitions is to apply the bipartitioning process separately to @xmath96 and @xmath98 , then to the resulting partitions and so on , until all the computed subpartitions are sufficiently small .
the obtained procedure is similar to the well - known recursive spectral bisection ( rsb ) algorithm , see , e.g. , @xcite , which is based on computing the fiedler vector @xcite , i.e. , the eigenvector corresponding to the smallest eigenvalue of @xmath205 each of the subgraphs associated with the sets @xmath96 and @xmath98 , which are determined by ( [ eqn : evp ] ) , can be , and often is , disconnected .
this constitutes an important difference between bipartitions delivered by eigenvalue problems ( [ eqn : evp ] ) and ( [ eqn : fiedler ] ) . at the same time , the assumption on the connectedness is crucial for eigenvalue problem ( [ eqn : evp ] ) to be well - posed for the given graph . indeed ,
if a graph has more then one ( trivial ) connected component , then the dimension of the nullspace of the corresponding graph laplacian @xmath173 is larger than one , i.e. , the nullspace of @xmath173 is no longer spanned only by the vector @xmath180 . in this case , @xmath173 becomes symmetric positive _ semidefinite _ on @xmath182 , and hence ( [ eqn : evp ] ) does not lead to a correct symmetric eigenvalue problem on @xmath182 .
the latter complication can be simply avoided by searching for the connected components in the subgraphs corresponding to @xmath96 and @xmath98 , and further treating the detected components as separate subpartitions in the suggested recursive procedure .
therefore , it is important to realize that the bipartitioning step corresponding to problem ( [ eqn : evp ] ) may result in more than two connected components .
finally , we note that if the weights @xmath206 take a single nonzero value , i.e. , are the same for all edges , then the weighted and the standard graph laplacians @xmath165 and @xmath173 represent the multiples of each other .
this happens for matrices @xmath8 with a very special , `` regular '' , behavior of entries , e.g. , as for the discrete laplace operator with constant ( diffusion ) coefficients . in such cases
, the eigenvector @xmath184 , which is used to define the bipartition , corresponds to the only nonzero eigenvalue of ( [ eqn : evp ] ) of multiplicity @xmath207 .
this observation implies that _ any _ bipartition can be expected , and , hence , the results of the approach based on ( [ eqn : evp ] ) are highly uncertain . in these situations , we simply replace ( [ eqn : evp ] ) , e.g. , by eigenvalue problem ( [ eqn : fiedler ] ) , or use any other ( bi)partitioning scheme , which targets to satisfy the standard criteria of minimizing the edge cut and maintaining the load balance .
let us now summarize the discussion into the following partitioning algorithm .
[ alg : partn1 ] input : @xmath10 , @xmath8 .
output : partition @xmath208 of @xmath209 . 1 .
assign weights @xmath147 to the edges of @xmath10 .
if @xmath19 are the same for all edges , then construct the bipartition @xmath80 using a standard approach , e.g. , based on the fiedler vector , and go to step @xmath210 with @xmath211 and @xmath212 .
2 . construct graph laplacians @xmath213 and @xmath214 .
3 . find the eigenpair corresponding to the smallest eigenvalue of problem ( [ eqn : evp ] ) .
4 . define the bipartition @xmath215 based on the computed eigenvector .
the sizes of @xmath96 and @xmath98 are controlled by the parameter _
loadbalance_. 5 .
find connected components @xmath216 in the subgraphs of @xmath10 corresponding to the vertex sets @xmath96 and @xmath98 .
6 . for all @xmath217 with @xmath218 , apply cbspartition(@xmath217,@xmath219 ) .
if all @xmath220 , then return @xmath208 .
the parameters _ loadbalance _ and _ maxsize _ in algorithm [ alg : partn1 ] are provided by the user .
the connected components can be detected by the standard algorithms based on the breadth - first search ( bfs ) or the depth - first search ( dfs ) ; see , e.g. , @xcite .
note that every weight @xmath19 assigned to the edge @xmath5 is the same at every level of the recursion in algorithm [ alg : partn1 ] , and , in practice , is assigned only once , i.e. , when constructing the adjacency graph @xmath10 of the whole matrix @xmath8 .
in the previous section , we have used the idea of minimizing the cbs constant @xmath116 to define the edge weights for the matrix adjacency graph and obtain objective ( [ eqn : min_subopt ] ) for graph partitioning , which aims at increasing the quality of the as preconditioner in algorithm [ alg : as ] .
below , we show that similar considerations can lead to the problems of graph partitioning with well - known objectives , such as the ( weighted ) cut minimization and the min - max cut @xcite . given a partition @xmath80 in ( [ eqn : partitij ] ) , @xmath120 ,
let us consider the set @xmath221 where @xmath134 denotes the unit vector with @xmath135 at position @xmath136 and zeros elsewhere . unlike ( [ eqn : sample ] ) , the set @xmath222 in ( [ eqn : sample_mincut ] ) contains _ all _ pairs @xmath223 with @xmath224 and @xmath225 , i.e. , including those , which correspond to @xmath226 . thus , instead of maximizing ( [ eqn : qty ] ) to compute the cbs constant @xmath116 in ( [ eqn : cbsc ] ) for the given bipartition @xmath89 , we sample ( [ eqn : qty ] ) at all @xmath227 pairs in ( [ eqn : sample_mincut ] ) and then find the mean value .
in other words , we define the quantity @xmath228 , such that @xmath229 where @xmath159 is the weighted cut , and @xmath147 are the weights assigned to the edges of the adjacency graph of @xmath8 .
thus , following the pattern of the previous subsection , instead of the objective function @xmath125 , which gives an optimal bipartition , we suggest to minimize @xmath230 in ( [ eqn : subopt_mincut ] ) , i.e. , find such a bipartition @xmath231 that @xmath232 it is readily seen that ( [ eqn : min_subopt_mincut ] ) represents the well - known problem of graph partitioning , which aims at finding equal - sized vertex sets @xmath96 and @xmath98 with the minimal ( weighted ) edge cut . in particular , repeating the derivations in subsection [ subsec : uv ] for this problem , we can define the bipartition @xmath89 according to the components of the eigenvector corresponding to the smallest eigenvalue of @xmath233 where @xmath165 is the _ weighted _ graph laplacian .
the recursive application of this procedure leads to the partitioning scheme , which is similar to the standard rsb algorithm with the difference that now edge weights @xmath234 are encapsulated into the graph laplacian .
let us further assume that the matrix @xmath8 is diagonally scaled , i.e. , @xmath235 in this case , the diagonal entries of @xmath8 are all equal to @xmath135 , and the off - diagonal elements are less than one . in particular , we note that the weights @xmath19 defined in the previous sections simply coincide with the entries of the scaled matrix . given a partition @xmath80 in ( [ eqn : partitij ] ) of @xmath204 ,
let us consider the set of pairs @xmath236 where @xmath237 and @xmath134 denotes the unit vector with @xmath135 at position @xmath136 and zeros elsewhere .
the vectors @xmath238 in @xmath239 are defined to have components @xmath240 , such that @xmath241 similarly , the vectors @xmath242 in @xmath243 are such that @xmath244 now , following the approach exploited throughout this paper , instead of maximizing ( [ eqn : qty ] ) to compute the cbs constant @xmath116 in ( [ eqn : cbsc ] ) for the given bipartition @xmath89 , we sample ( [ eqn : qty ] ) at all @xmath100 pairs in ( [ eqn : sample_mcut ] ) and find the mean value . in particular , recalling that the diagonal entries of @xmath8 are equal to @xmath135 after the diagonal scaling , we get @xmath245^{1/2 } } + \displaystyle \sum_{j \in j } \frac{\displaystyle \sum_{i \in i } |a_{ji}|}{\left[\displaystyle \sum_{k , l \in i } a_{kl } \mbox{sign}(a_{jl } ) \mbox{sign}(a_{jk } ) \right]^{1/2 } } \right ) \geq \\ \displaystyle \frac{1}{n } \left(\displaystyle \sum_{i \in i } \frac{\displaystyle \sum_{j \in j } |a_{ij}|}{\left[\displaystyle \sum_{k , l \in j } |a_{kl}| \right]^{1/2 } } + \displaystyle \sum_{j \in j } \frac{\displaystyle
\sum_{i \in i } |a_{ji}|}{\left[\displaystyle \sum_{k , l \in i } |a_{kl}| \right]^{1/2 } } \right ) \geq \frac{1}{n } \left ( \frac{\displaystyle \sum_{i \in i , j \in j } |a_{ij}|}{\displaystyle \sum_{k , l \in j } |a_{kl}| } + \frac{\displaystyle \sum_{j \in j , i \in i } |a_{ji}|}{\displaystyle \sum_{k , l \in i } |a_{kl}| } \right ) , \end{array}\ ] ] and define the quantity @xmath246 , such that @xmath247 where @xmath248 is the weighted cut between sets @xmath96 and @xmath98 with edge weights @xmath249 ; @xmath250 and @xmath251 .
thus , instead of the objective function @xmath125 , which gives an optimal bipartition , one can attempt to minimize @xmath252 in ( [ eqn : subopt_mcut ] ) , i.e. , find the minimizer @xmath253 of @xmath254 minimization ( [ eqn : min_subopt_mcut ] ) represents the problem of finding the so - called min - max cut ( mcut ) ; see @xcite .
we note that the explicit requirement @xmath255 has been dropped in ( [ eqn : min_subopt_mcut ] ) , however , the mcut is known to target balanced partitions .
the corresponding algorithm , which attempts to construct @xmath89 satisfying ( [ eqn : min_subopt_mcut ] ) , is based on finding the eigenpair corresponding to the smallest eigenvalue of the problem @xmath256 where @xmath165 is the weighted graph laplacian and @xmath257 is the diagonal weighted degree matrix .
the recursive application of this procedure delivers the actual partitioning scheme .
we remark that eigenvalue problem ( [ eqn : evp_mcut ] ) is also used to construct the normalized cuts ( ncuts ) , introduced in @xcite .
in this section , we present the results of graph partitioning with respect to the new objective ( [ eqn : min_subopt ] ) , and demonstrate the effects of the proposed partitioning strategy on the quality of preconditioning . in our numerical experiments , we apply algorithm [ alg : partn1 ] ( cbspartition ) , introduced in subsection [ subsec : recur ] , to a number of test problems with spd coefficient matrices .
the resulting partitions ( subdomains ) are passed as an input to the as preconditioner in algorithm [ alg : as ] , which is used to accelerate the convergence of the pcg method .
we refer to this solution scheme as pcg as . in order to assess the quality of the constructed as preconditioners , we consider the iteration counts of pcg as for different partitioning schemes . in particular ,
we compare pcg as with partitions resulting from algorithm [ alg : partn1 ] versus pcg as with the standard partitioning based on the rsb algorithm .
we also provide the comparisons for pcg as with partitioning schemes based on the ( weighted ) mincut and mcut objectives , discussed in section [ sec : other_spectral ] .
although , throughout the paper , the formal discussion has been concerned only with the case of the nonoverlapping as procedure , in some of our numerical examples , we skip this theoretical limitation and expand the obtained partitions with several `` layers '' of neighboring nodes .
this allows considering the effects of the partitioning strategies on the quality of the _ overlapping _ as preconditioners . in all the tests ,
the right - hand side and the initial guess vectors are randomly chosen .
we apply pcg as to the diagonally scaled linear systems , so that the corresponding coefficient matrices have @xmath135 on the diagonal ; see ( [ eqn : dscale ] ) . in this case , the weights @xmath19 , assigned to the edges of the adjacency graph , are equal to the entries of the scaled matrix . for all partitioning schemes ,
the parameter _ loadbalance _ , which controls the load balancing , is set to @xmath258 ; _ loadbalance _ @xmath259 .
the partitioning algorithms that we consider in the current paper are based on finding the eigenvectors of certain eigenvalue problems , i.e. , represent the _ spectral _ partitioning techniques .
we recall that algorithm [ alg : partn1 ] computes an eigenpair of problem ( [ eqn : evp ] ) .
the rsb algorithm targets the fiedler vector in ( [ eqn : fiedler ] ) .
the approaches based on the ( weighted ) mincut and mcut use the eigenvectors of problems ( [ eqn : evp_mincut ] ) and ( [ eqn : evp_mcut ] ) , respectively . in our numerical examples , as an underlying eigensolver , we use the lobpcg method , which allows to handle generalized symmetric eigenvalue problems , such as ( [ eqn : evp ] ) , without any factorizations of the matrices @xmath165 and @xmath173 , and can be easily configured to perform iterations on the subspace @xmath182 .
the lobpcg algorithm is a form of a ( block ) three - term recurrence , which performs the local minimization of the rayleigh quotient ; see @xcite for more details .
the method is known to be practical for large - scale eigenvalue computations , if a good preconditioner is provided to accelerate the convergence to the desired eigenpairs . for
all lobpcg runs , we construct preconditioners using incomplete cholesky ( ic ) factors with the drop tolerance @xmath260 of matrices @xmath261 ( for problems ( [ eqn : evp ] ) , ( [ eqn : evp_mincut ] ) , and ( [ eqn : evp_mcut ] ) ) and @xmath262 ( for problem ( [ eqn : fiedler ] ) ) .
the parameter @xmath263 is assigned with a small value , @xmath264 in our examples , to ensure that the ic procedure is correctly applied to the spd matrices . for problems ( [ eqn : fiedler ] ) , ( [ eqn : evp_mincut ] ) , and ( [ eqn : evp_mcut ] ) , we remove the orthogonality constraints on @xmath184 , and perform the block iteration , with the block size @xmath22 .
the solution is then given by the eigenpair corresponding to the second smallest eigenvalue .
for problem ( [ eqn : evp ] ) , however , we run a single vector iteration , choosing the initial guess from @xmath182 and ensure that the residuals are projected back to @xmath182 after the ic preconditioning . in all the tests ,
the lobpcg residual norm tolerance is set to @xmath265 . as a model problem
we choose the @xmath22d diffusion equation @xmath266 \times [ 0,1]\ ] ] with zero dirichlet boundary conditions .
the functions @xmath267 and @xmath268 are piecewise constants with jumps in the specified subregions . for all tests ,
we use the standard @xmath269-point fd discretization on @xmath270-by-@xmath270 uniform grid to obtain a ( diagonally scaled ) linear systems of size @xmath271 .
we consider two geometries for the location of jumps in coefficients @xmath267 and @xmath268 . in the first case
, the jumps occur in the subregion @xmath272 of the domain .
the second case is more complex , with jumps located in the `` checkerboard '' fashion ( `` @xmath269-by-@xmath269 black - white checkerboard '' ) . in the following example
, we assume that _ both _
coefficients @xmath267 and @xmath268 have jumps in the subregion @xmath272 of the problem s domain , such that @xmath273 figure [ fig : j_xy ] shows the result of a single step , i.e. , the bipartitioning , performed by algorithm [ alg : partn1 ] ( top left ) , the standard rsb approach ( top right ) , as well as the weighted mincut ( bottom left ) and the mcut ( bottom right ) algorithms .
+ we observe that , unlike the rest , the partition resulting from algorithm [ alg : partn1 ] does not perform `` cuts '' within the jump region .
in fact , this is consistent with the well - established computational experience , which suggests that the subdomains with different physics phenomena should be `` segmented '' out of the original domain .
similarly , we apply the bipartitioning step of all four approaches to the model problem , where the jump occurs only in the coefficient @xmath267 , i.e. , @xmath274 the resulting bipartitions are illustrated in figure [ fig : j_x ] .
we note that the partitions given by algorithm [ alg : partn1 ] ( top left ) , the weighted mincut ( bottom left ) and the mcut ( bottom right ) algorithms do not discard the mesh edges in the @xmath97-direction within the jump region .
the border between the subdomains is of a `` smoother '' shape for the mincut and mcut , which is preferable in terms of minimizing the communication volume .
however , as suggested by the numerical experiments below , the partitions based on algorithm [ alg : partn1 ] typically guarantee a smaller number of steps of the iterative scheme as the number of the desired subdomains becomes larger .
we also remark that , although independent of the matrix coefficients , the partitions resulting from the standard rsb approach may not be unique , see figures [ fig : j_xy ] and [ fig : j_x ] ( top right ) , if the targeted eigenvalue has multiplicity greater than one .
+ in figure [ fig : ev1 ] , we plot the components of the eigenvectors , corresponding to the smallest eigenvalues of ( [ eqn : evp ] ) , at the grid points . according to algorithm [ alg : partn1 ] ,
such eigenvectors are used to construct the bipartition . indeed , the components of the eigenvectors are well - separated , which allows to easily determine the partitions and detect the `` weak '' connection between the grid nodes to be discarded in order to obtain the resulting bipartition .
figure [ fig : j_x_cv ] demonstrates the effects of partitioning schemes on the quality of the nonoverlapping as preconditioners . in figure
[ fig : j_x_cv ] ( left ) , we consider the pcg as runs for the model problem with coefficients in ( [ eqn : j_xy ] ) , where the as preconditioners are constructed with respect to the partitions delivered by different algorithms .
the parameter _ maxsize _ , which determines the largest possible size of a single subdomain , is set to @xmath275 in this example . in figure
[ fig : j_x_cv ] ( right ) , we apply pcg as for the model problem with coefficients in ( [ eqn : j_x ] ) . here , the parameter _ maxsize _ ,
is set to @xmath276 .
this corresponds to a more realistic situation , where each subdomain is significantly smaller than the original domain .
the results in figure [ fig : j_x_cv ] suggest that pcg as with the partitioning scheme in algorithm [ alg : partn1 ] requires a noticeably smaller number of iterations to get the solution .
we remark that the quality of the as preconditioning with respect to the partitions delivered by algorithm [ alg : partn1 ] may often depend on the value of the parameter _
maxsize_. for example , in the case of the model problem with coefficients in ( [ eqn : j_xy ] ) , a small value of _ maxsize _ forces algorithm [ alg : partn1 ] to perform the partitioning inside the jump region , i.e. , inside the subdomain marked with `` o '' s in figure [ fig : j_xy ] ( top left ) .
this , clearly , can lead to less remarkable gains in the use of algorithm [ alg : partn1 ] compared to the other partitioning approaches . in the following pair of examples
, we assume the `` checkerboard '' pattern for the jump regions ( `` @xmath269-by-@xmath269 black - white checkerboard '' ) .
first , we consider jumps in both @xmath267 and @xmath268 in `` black '' positions , i.e. , @xmath277 in `` black '' , and @xmath278 in `` white '' .
the corresponding bipartitions resulting from different partitioning schemes are given in figure [ fig : j_xy_checker ] .
+ similarly , figure [ fig : j_x_checker ] demonstrates the bipartitions corresponding to the second case , with jumps only in @xmath267 , i.e. , @xmath279 in `` black '' , and @xmath280 in `` white '' , @xmath281 .
+ we recall that the bipartitioning step of algorithm [ alg : partn1 ] may deliver two _ disconnected _ subdomains , i.e. , two subgraphs which possibly contain more than one connected component .
each of these connected components is then processed separately in the recursive partitioning procedure .
this explains the presence of more than two subdomains in ( top left ) figures [ fig : j_xy_checker ] and [ fig : j_x_checker]the nodes of each connected component , resulting from a single step of algorithm [ alg : partn1 ] , are plotted as separate regions .
we also note that the recursive bipartitioning given by algorithm [ alg : partn1 ] may generate a number of small connected subdomains of sizes much smaller than the value of the _ maxsize _ parameter .
such subdomains should be treated with care when being assigned to parallel processors . in figure [ fig :
ev2 ] , we plot the components of the eigenvectors corresponding to the smallest eigenvalues of ( [ eqn : evp ] ) at the grid points for the `` checkerboard '' example .
it is possible to see that , as in the previous examples , both eigenvectors attempt to capture the discontinuities in the coefficients of the model problem . in figure
[ fig : j_x_checker_cv ] , we compare the convergence behavior of pcg as with different partitioning schemes .
figure [ fig : j_x_checker_cv ] ( left ) corresponds to the case of the jumps in both @xmath267 and @xmath268 in `` black '' positions .
we observe that for this relatively complex geometry of the jump locations , all partitioning schemes which use the information on the matrix coefficients result in the as preconditioners of a better quality . in this example
, the number of pcg as iterations is typically similar for the partitioning techniques in algorithm [ alg : partn1 ] , the weighted mincut , and mcut .
figure [ fig : j_x_checker_cv ] ( right ) demonstrates the runs of pcg as applied to the model problem with the jump only in @xmath267 in `` black '' positions . in this case
, the iterative scheme with partitions resulting from algorithm [ alg : partn1 ] gives the fastest convergence . in
both , figure [ fig : j_x_checker_cv ] ( left ) and figure [ fig : j_x_checker_cv ] ( right ) , the _ maxsize _ parameter has been set to @xmath276 . finally , we apply the partitioning schemes , discussed in this paper , to a set of test problems from the university of florida sparse matrix collection . in particular , we consider ill - conditioned spd matrices arising in structural engineering and computational fluid dynamics . in tables
[ tbl : ufl ] and [ tbl : ufl_ovlp ] , we report the numbers of iterations ( averaged after @xmath282-@xmath283 sample runs ) required by pcg as to reach the tolerance @xmath284 in the residual norm ( relative to the norm of the right - hand side vectors ) for _ nonoverlapping _ and _ overlapping _ as procedures , respectively .
.iteration numbers of the * nonoverlapping * pcg as with different partitioning schemes applied to test problems from the university of florida sparse matrix collection . [ cols="^,^,^,^,^,^ " , ]
in the present paper , we have shown that using matrix coefficients for graph partitioning allows to achieve a noticeable decrease in the number of iterations performed by an iterative scheme . for a class of spd matrices and as preconditioners , we have suggested an approach for assigning weights to the edges of the adjacency graph and formulated a new partitioning objective , which aims at approximately minimizing the cbs constant . the resulting partitioning algorithm is based on computing the eigenpairs corresponding to the smallest eigenvalues of the sequence of generalized eigenvalue problems , which involve both weighted and standard graph laplacians . in particular , this means that the proposed technique inherits all specificities of spectral partitioning algorithms , such as good quality of partitions , on the one hand , and the computational expenses related to finding eigenvectors , on the other hand .
thus , in order to obtain highly efficient graph partitioning schemes , it is important to study all aspects of the occurring eigencomputations , such as , e.g. , preconditioning , the use of alternative eigenvalue solvers , possible ways to replace the eigenvalue problems by linear systems .
other approaches for satisfying the suggested partitioning objective may be delivered , e.g. , by ( multilevel ) combinatorial graph partitioning techniques or by certain extensions of greedy algorithms .
we note that methods for reaching the new partitioning objective may be combined with the communications minimizing techniques .
as one could conclude from this work , it is likely that different choices of iterative methods and preconditioning strategies may require different schemes for graph partitioning with matrix coefficients . in the current paper , we have considered the case of pcg with the as preconditioning .
exploring the partitioning for other forms of parallel preconditioning ( e.g. , incomplete factorizations and multigrid ) is a natural continuation of the research in this direction .
constructing partitioning algorithms with matrix coefficients for nonsymmetric problems is also of a particular interest . , _
non - standard parallel solution strategies for distributed sparse linear systems _ , in parallel computation : proc . of acpc99 , a. u. p. zinterhof , m. vajtersic , ed . , lecture notes in computer science , berlin , 1999 , springer - verlag . | prior to the parallel solution of a large linear system , it is required to perform a partitioning of its equations / unknowns .
standard partitioning algorithms are designed using the considerations of the efficiency of the parallel matrix - vector multiplication , and typically disregard the information on the coefficients of the matrix .
this information , however , may have a significant impact on the quality of the preconditioning procedure used within the chosen iterative scheme . in the present paper ,
we suggest a spectral partitioning algorithm , which takes into account the information on the matrix coefficients and constructs partitions with respect to the objective of increasing the quality of the additive schwarz preconditioning for symmetric positive definite linear systems .
numerical results for a set of test problems demonstrate a noticeable improvement in the robustness of the resulting solution scheme when using the new partitioning approach . graph partitioning , iterative linear system solution , preconditioning , cauchy - bunyakowski - schwarz ( cbs ) constant , symmetric positive definite , spectral partitioning 15a06 , 65f08 , 65f10 , 65n22 |
it has been difficult to read the recent financial news without finding mention of collateralized debt obligations ( cdo s ) . these financial instruments
provide ways of aggregating risk from a large number of sources ( viz .
bonds ) and reselling it in a number of parts , each part having different risk - reward characteristics . notwithstanding
the role of cdo s in the recent market meltdown , the near future will no doubt see the financial engineering community continuing to develop structured investment vehicles like cdo s .
unfortunately , computational challenges in this area are formidable .
the main types of these assets have several common problematic features : * they pool a large number of assets * they tranche the losses .
the `` problematic '' nature of this combination is that the trancheing procedure is nonlinear ; as usual , the effect of a nonlinear transformation on a high - dimensional system is often difficult to understand .
ideally , one would like a theory which gives , if not explicit answers , at least some guidance . in @xcite
, we formulated a _ large deviations _
analysis of a homogeneous pool of names ( i.e. bonds ) .
the theory of large deviations is a collection of ideas which are often useful in studying rare events ( see @xcite for a more extensive list of references to large deviations analysis of financial problems ) . in @xcite ,
the rare event was that the notional loss process exceeded the tranche attachment point for an investment - grade tranche .
our interest here is heterogeneous pool of names , where the names can have different statistics ( under the risk - neutral probability measure ) .
there are several perspectives from which to view this effort .
one is that we seek some sort of _ homogenization _ or _ data fusion_. is there an effective macroscopic description of the behavior of the cdo when the underlying instruments are a large number of different types of bonds ? another is an investigation into the _ fine detail _ of the rare events which cause loss in the investment - grade tranches .
there may be many ways or `` configurations '' for the investment - grade tranches to suffer losses .
which one is most likely to happen ?
this is not only of academic interest ; it also is intimately tied to quantities like loss given default and also to numerical simulations .
we believe this to be an important component of a larger analysis of cdo s , particularly in cases where correlation comes from only a few sources ( we will pursue a simple form of this idea in subsection [ s : correlated ] ) .
we will find a natural generalization of the result of @xcite , where the dominant term ( as the number of names becomes large ) was a relative entropy . here
, the dominant term will be an integrated entropy , with the integration being against a distribution in `` name '' space .
our main result is given in theorem [ t : main ] and .
as in @xcite , we let @xmath0 $ ] . we endow @xmath1 with its usual topology under which it is polish ( cf .
@xcite ) . for each @xmath2 ,
the @xmath3-th name will default at time @xmath4 , where @xmath4 is an @xmath1-valued random variable . to fix things ,
our event space will be @xmath5 and , @xmath6 is the borel sigma - algebra of subsets of @xmath7 , and @xmath8 is the collection of probability measures on @xmath9 . ]
fix next @xmath11 ( which corresponds to a pool of size @xmath12 ) and @xmath13 and let @xmath14 be the associated expectation operator .. following @xcite , we define the notional and tranched loss processes as @xmath15}(\tau_n ) \qquad \text{and}\qquad { \bar l}^{(n)}_t { \overset{\text{def}}{=}}\frac{(l^{(n)}_t-\alpha)^+-(l^{(n)}_t-\beta)^+}{\beta-\alpha}\ ] ] for all @xmath16 , with @xmath17 , where @xmath18 and @xmath19 are the attachment and detachment points of the tranche ( since the @xmath4 s are all nonnegative , @xmath20 for @xmath21 ) .
our interest is then @xmath22}{{\mathbb{e}}_n[{\textbf{p}^{\text{prem}}}_n]}\ ] ] where @xmath23\ ] ] with @xmath24 being the interest rate , @xmath25 being the time horizon of the contract , and @xmath26 being the ( finite ) set of times at which the premium payments are due ( and such that @xmath27 for all @xmath28 ) .
we have assumed here , for the sake of simplicity , no recovery .
our interest specifically is in @xmath12 large .
let s now think about the sources of randomness in the names .
each name is affected by its own _ idiosyncratic _
randomness and by _ systemic _ randomness ( which affects all of the names ) .
assumedly , the systemic randomness , which corresponds to macroeconomic factors , is _ low - dimensional _ compared to the number of names .
for example , there may be only a handful of macroeconomic factors which a pool of many thousands of names .
we can capture this functionality as @xmath29 where the @xmath30 and @xmath31 are all independent random variables , and @xmath32 is some appropriate set in the product space of the sets where the @xmath33 s and @xmath31 take values .
our interest is to understand the implications of the structural model .
we are not so much concerned with specific models for the @xmath33 s , the @xmath31 , or the @xmath32 s but rather the structure of the rare losses in the investment - grade tranches .
we would also like to avoid , as much as possible , a detailed analysis of the parts of since in practice what we have available to carry out pricing calculations is the price of credit default swaps for the individual names ; i.e. ( after a transformation ) , @xmath34 .
thus we ca nt with certainty get our hands on the details of .
there may in fact be several models of the type which lead to the same `` price '' for the rare events involved in an investment - grade tranche .
if we can understand more about the structure of rare events in these tranches , we can understand which aspects of are important ( and then try to calibrate specific models using that insight ) .
regardless of the details of , we can make some headway .
the notional loss at time @xmath35 will be given by @xmath36 the definition of an investment - grade tranche is that @xmath37 is small .
guided by chebychev s inequality , lets define @xmath38 \qquad \text{and}\qquad \sigma^{(n ) } { \overset{\text{def}}{=}}\sqrt{{\mathbb{e}}\left[\left(l^{(n)}_{t-}-\mu^{(n)}\right)^2\right]}.\ ] ] if @xmath39 , chebychev s inequality gives us that @xmath40 in order for this to be small , we would like that @xmath41 be small ; this is the point of pooling . for any fixed value of @xmath42
, the conditional law of @xmath43 given that @xmath44 is the variance of @xmath45 ; thus the conditional variance of @xmath43 given that @xmath44 is at most of order @xmath46 .
hopefully , when we reinsert the systemic randomness , the variance of @xmath47 will still be small , and we will indeed have an investment - grade tranche .
in fact , we can do better than chebychev s inequality . by again conditioning on @xmath31 , we can write that @xmath48\ ] ] thus the tranche will be investment - grade if @xmath49 is small for `` most '' values of @xmath42 ( see remark [ r : systemic ] ) . as mentioned above , however , we know the law of @xmath43 conditioned on @xmath31 .
namely , @xmath50 this then clearly motivates a natural two - step approach .
our first step is to condition on the value of the systemic randomness ( which we may think of as fixing a `` state of the world '' or a `` regime '' ) and concentrate on how rare events occur due to idiosyncratic randomness ( i.e. , to effectively _ suppress _ the systemic randomness ) .
it will turn out that this is in itself a fairly involved calculation .
nevertheless , it is connected with a classic problem in large deviations theory_sanov
s theorem_. with this in hand , we should then be able to return to the original problem and average over the systemic randomness ( in subsection [ s : correlated ] ) .
some of the finer details of these effects of correlation will appear in sequels to this paper . here
we will restrict our interest in the effects of correlation to a very simple model ( which is hopefully nevertheless illustrative ) .
let s get started .
we want to consider the effect of a large number of names . for each @xmath12
, we suppose that @xmath4 ( for @xmath51 ) has distribution @xmath52 . to reflect our initial working assumption that the names are independent
, we thus let the risk neutral probability @xmath13 be such that only specifies the law of @xmath53 , not the law of the rest of the @xmath4 s , @xmath54 is not unique in @xmath55 . ]
@xmath56 for all @xmath57 and the random variables @xmath4 , and letting the probability measure @xmath54 depend on @xmath12 .
we could just as easily have fixed a common probability measure and let the default times be @xmath12-dependent .
given our later @xmath12-dependent measure change in section [ s : measurechange ] , we decided to have the measure be @xmath12-dependent from the start . ] .
[ ex : simpleexample ] fix distributions @xmath58 and @xmath59 on @xmath1 i.e. , @xmath58 and @xmath59 are in @xmath60 .
assume that for each @xmath12 , every third i.e. , @xmath61 name follows distribution @xmath58 and the others follow distribution @xmath59 ; i.e. , @xmath62 for all @xmath51 .
to be even more specific , one might let @xmath63 correspond to a bond with moody s a3 rating , and one might let @xmath64 correspond to a bond with moody s ba1 rating see @xcite .
although we could separately carry out the analysis of @xcite for the @xmath65 bonds and the @xmath66 bonds , we shall find that the combined cdo reflects a nontrivial combination of the calculations for each separate bond .
in particular , the losses in the cdo stem from a preferred _ combination _ of losses in both types of bonds .
see the ideas of example [ ex : twobonds ] . while the above example will give us insight into some calculations ,
another example along the lines of a merton - type model will be of more practical interest .
[ ex : merton ] assume that under @xmath54 the default likelihoods are given by merton - type models and of course , they are all independent . to keep the ideas and notation simple , let s assume that the companies have common risk - neutral drift @xmath67 , initial valuation @xmath68 , and bankruptcy barrier @xmath69 .
] , however , that under @xmath54 , @xmath3-th company has volatility @xmath70 , and that the @xmath71 s are approximately distributed according to a gamma distribution of scale @xmath72 and shape @xmath73 ; i.e. , for every @xmath74 , @xmath75 for each @xmath76 , let @xmath77 be given by @xmath78 dt\\ + { \left\ { } \newcommand{\rb}{\right\}}1-\int_{t\in ( 0,\infty)}\frac{\ln ( 1/k)}{\sqrt{2\pi \sigma^2 t^3}}\exp\left[-\frac{1}{2\sigma^2 t}\left(\left(\theta-\frac{\sigma^2}{2}\right)t+\ln \frac{1}{k}\right)^2\right ] dt\rb \delta_{\infty}(a ) .
\qquad a\in { \mathscr{b}}(i ) \end{gathered}\ ] ] we take @xmath79 .
we will frequently return to these two examples .
since the @xmath4 s are independent , @xmath43 is a sum of @xmath12 independent ( but not identically - distributed ) bernoulli random variables .
the central idea of collateralized debt obligations ( and structured finance in general ) is that by pooling together a large number of assets , one can use the law of large numbers to reduce variance and create derivatives which depend on tail events .
our assumption that the names are independent means that in some sense we have `` maximal '' randomness ; the dimension of idiosyncratic randomness is the same as the dimension of the number of names .
good bounds on tail behavior should thus result .
indeed , since the variance of a bernoulli random variable is less than @xmath80 , the variance of @xmath43 is at most @xmath46 .
we will exploit this calculation in lemma [
l : tail ] .
if the names are correlated , there is in a sense `` less '' randomness , so the variance should be larger . between our work here and that of @xcite , we have a number of tools which we can use when the degree of randomness is indeed comparable to the number of names in the cdo .
not surprisingly , we will need several assumptions . for the moment , we will phrase these in terms of the @xmath81 s .
later on , in section [ s : limitexists ] , we will find alternate assumptions if the @xmath81 s are samples from an underlying distribution on @xmath60 .
our first assumption is that the @xmath82 s have a certain type of limit ; some sort of assumption of this type is of course necessary if we are to proceed with an analysis for large @xmath12 .
note from @xcite that when the default times are identically distributed , the dominant asymptotic value of the protection leg depends only on the probability of default in time @xmath83
( i.e. , it does not depend on the structure of the default distribution within @xmath83 ) .
we will see the same phenomenon here . for each @xmath11 , define @xmath84}$ ] as @xmath85 note that since @xmath86 $ ] is polish and compact , so is @xmath87}$ ] ( * ? ? ?
thus @xmath88 has at least one cluster point .
we actually assume that it is unique ; [ a : limitexists ] we assume that @xmath89 exists .
in example [ ex : simpleexample ] , we would have that @xmath90 and in example [ ex : merton ] , we would similarly have that @xmath91 our next assumption reflects our interest in cases where where it is unlikely that the tranched loss process @xmath92 suffers any losses by time @xmath25 .
note here that @xmath93 = \frac1n\sum_{n=1}^n \mu^{(n)}_n[0,t)=\int_{p\in [ 0,1]}p{\bar { u}^{(n)}}(dp)\ ] ] for all @xmath11 .
also note that by the formula and the fact that the variance of an indicator is less than or equal to @xmath80 , we see that the variance of @xmath43 tends to zero as @xmath94 .
[ a : ig ] we assume that @xmath95}p{\bar { u}}(dp)<\alpha.\ ] ] assumption [ a : limitexists ] implies that @xmath96}p{\bar { u}}(dp ) = \lim_{n\to \infty}\int_{p\in [ 0,1]}p{\bar { u}^{(n)}}(dp ) = \lim_{n\to \infty}\frac{1}{n}\sum_{n=1}^n \mu^{(n)}_n[0,t).\ ] ] thus assumption [ a : ig ] is equivalent to the requirement that @xmath97 in the case of example [ ex : simpleexample ] , assumption [ a : ig ] is that @xmath98 and in the case of example [ ex : merton ] , assumption [ a : ig ] is that @xmath99 [ l : tail ] thanks to assumption [ a : ig ] , we have that @xmath100 . assumption [ a : ig ] is exactly that for @xmath11 sufficiently large , @xmath101<\alpha$ ] . thus by chebychev s inequality , @xmath102>\alpha-{\mathbb{e}}\left[l^{(n)}_{t-}\right]\rb\\ \le \frac{{\mathbb{e}}_n\left[\left(l^{(n)}_{t-}-{\mathbb{e}}\left[l^{(n)}_{t-}\right]\right)^2\right]}{\left(\alpha-{\mathbb{e}}\left[l^{(n)}_{t-}\right]\right)^2 } = \frac1{n^2 } \frac{\sum_{n=1}^n \mu^{(n)}_n[0,t){\left\ { } \newcommand{\rb}{\right\}}1-\mu^{(n)}_n[0,t)\rb}{\left(\alpha-{\mathbb{e}}\left[l^{(n)}_{t-}\right]\right)^2 } \\
\le \frac1{4n\left(\alpha-{\mathbb{e}}\left[l^{(n)}_{t-}\right]\right)^2}. \end{gathered}\ ] ] this implies the claimed result . thus the event that the cdo suffers losses is thus _
rare_. next , we need some bounds on `` certainty '' .
[ r : certainty ] suppose , for the sake of argument , that we take @xmath58 and @xmath59 in example [ ex : simpleexample ] so that @xmath103 and @xmath104 .
in other words , every third name is _ sure _ to default by time @xmath25 and default by time @xmath25 on the remaining bonds is _
impossible_. such a cdo would of course be of no practical interest .
however , we could envision a cdo where a third of the names are of junk status , and the remaining bonds are of impeccable quality .
our extreme example would thus be a natural first - order approximation in that case .
a moment s thought reveals that @xmath105 , so @xmath101=\frac{\lfloor n/3\rfloor}{n}$ ] .
thus if @xmath106 , assumption [ a : ig ] is satisfied .
in fact if @xmath12 is large enough , @xmath107 , so this is not a very interesting case .
there is simply too much certainty here .
note that assumption [ a : ig ] implies a bound on the number of bonds with certain default ; since @xmath108 for all @xmath109 $ ] , assumption [ a : ig ] implies that @xmath110}p{\bar { u}}(dp)<\alpha.\ ] ] the point of remark [ r : certainty ] is that if too many names can not default by time @xmath25 , then there is no way that @xmath43 can exceed @xmath18 ; we want to preclude this , and make sure that tranche losses are a rare , but possible , event .
[ a : nondegen ] we assume that @xmath111 .
the equivalent formulation of this assumption in terms of the @xmath112 s is that @xmath113 to connect this to our thoughts of remark [ r : certainty ] , note that if @xmath114 then @xmath115 in which case @xmath116 .
we thirdly need an assumption that ensures that defaults before time @xmath25 can occur _ right _ before time @xmath25 .
this is important for the precise asymptotics of theorem [ t : main ] ( and essential for the asymptotics of section [ s : has ] ) .
[ a : notflat ] we assume that @xmath117 if @xmath118 , then ( under @xmath54 ) the @xmath3-th name is `` default - free '' right before @xmath25 .
the point of this assumption is that this is default - free bonds are not `` too '' typical .
the requirement that we allow such a default - free structure for only @xmath18 ( in percent ) of the names is also natural .
if it is violated , then @xmath18 or more ( in percent ) of the names may be default - free just prior to @xmath25 ; there would be a nonvanishing ( as @xmath94 ) probability that the cdo suffers a loss due exactly to those names , and in that case , @xmath47 would be flat in a small region @xmath119 before @xmath25 ( one may further assume that @xmath119 is the maximal such interval ) . in this case , the analysis of section [ s : has ] would be a development of @xmath120 instead of @xmath121 ; this would then affect the results of theorem [ t : main ] .
lemma [ l : notflatlemma ] contains one framework for checking this assumption .
another way is the following result .
assume that there is a neighborhood @xmath122 of @xmath25 such that each @xmath123 is absolutely continuous with respect to lebesgue measure on @xmath124 with density @xmath125 and that furthermore the @xmath125 s are equicontinuous . if @xmath126 then assumption [ a : notflat ] holds .
first let @xmath127 be such that @xmath128 fix next @xmath129 such that @xmath130 and such that @xmath131 .
fix now @xmath132 and @xmath133 . if @xmath134 , then @xmath135 ; thus @xmath136 first let @xmath94 , then @xmath137 , then @xmath138 to see that assumption [ a : notflat ] holds
. our main result is an asymptotic ( for @xmath94 ) formula for @xmath139 $ ] and @xmath140 .
since the result will require a fair amount of notation , let s verbally understand its structure first .
the point of @xcite was that the dominant asymptotic of the price @xmath140 was a relative entropy term ; this entropy was that of @xmath18 relative to the risk - neutral probability of default . in @xcite ,
all bonds were identically distributed , so this amounted to the entropy of a single reference coin flip ( the coin flip encapsulating default ) .
here we have a distribution of coins , one for each name .
not surprisingly , perhaps , our answer again involves relative entropy , but where we average over `` name''-space , and where we minimize over all configurations whose average loss is @xmath18 .
to state our main result , we need some notation . for all @xmath141 and @xmath142 in @xmath143 , define @xmath144
\ln \frac{1}{1-\beta_2 } & \text{for $ \beta_1=0 $ , $ \beta_2\in [ 0,1)$ } \\
\infty & \text{else.}\end{cases}\ ] ] for each @xmath145 and @xmath146}$ ] , define @xmath147}\hbar(\phi(p),p){\bar { v}}(dp ) : \phi\in b([0,1];[0,1 ] ) , \int_{p\in[0,1]}\phi(p){\bar { v}}(dp)=\alpha'\rb.\ ] ] we will see in lemma [ l : increasing ] that @xmath148 .
our main claim is that as @xmath94 , @xmath149.}}\ ] ] [ r : explanation ] the minimization problem is fairly natural .
the asymptotic price of the protection leg depends upon how `` unlikely '' it is that the proportion of defaults exceeds the attachment point @xmath18 . when there is only one type of name ( e.g. @xcite ) , this is seen to depend on the relative entropy of the attachment point @xmath18 with respect to the risk - neutral probability that a reference bond defaults before time @xmath25 .
if there are several types of bonds ( cf .
example [ ex : simpleexample ] and the calculations of example [ ex : twobonds ] ) , there are a number of ways to get the total proportion of defaults to exceed @xmath18 .
namely , allow each bond type to default at a different rate , but require that the total default rate exceeds @xmath18 .
since the entropy is relative to the risk - neutral probability of default before time @xmath25 , we can organize these calculations around @xmath150 .
taking the minimum entropy of all such default configurations , we get exactly . to proceed a bit further
, we claim that we can explicitly solve . for @xmath109
$ ] and @xmath151 $ ] , set @xmath152}(p ) & \text{if $ \lambda=\infty$ } \\
\chi_{\{1\}}(p ) & \text{if $ \lambda=-\infty$.}\end{cases}\ ] ] some properties of @xmath153 are given in remark [ r : phiprops ] .
for @xmath145 , we define @xmath154 } : { \bar { v}}\{1\}\le \alpha'\le 1-{\bar { v}}\{0\},\ , { \bar { v}}\not = \mu^\dagger_{\alpha'}\rb \\ { \mathcal{g}}^{\text{strict}}_{\alpha ' } & { \overset{\text{def}}{=}}{\left\ { } \newcommand{\rb}{\right\}}{\bar { v}}\in { { \mathscr{p}}[0,1 ] } : { \bar { v}}\{1\ } < \alpha ' < 1-{\bar { v}}\{0\}\rb . \end{aligned}\ ] ] note that @xmath155 } : { \bar { v}}\{1\}=\alpha'= 1-{\bar { v}}\{0\}\rb = \{\mu^\dagger_{\alpha'}\}.\ ] ] the following result solves the minimization problem for @xmath156 in terms of @xmath153 .
[ l : finalitmin ] fix @xmath145 and @xmath146}$ ] .
if @xmath157 , there is a unique @xmath158 $ ] such that @xmath159}\phi\left(p,\lambda(\alpha',{\bar { v}})\right){\bar { v}}(dp ) = \alpha'.\ ] ] if @xmath160 , then @xmath161 .
we have that @xmath162}\hbar\left(\phi(p,\lambda(\alpha',{\bar { v}})),p\right){\bar { v}}(dp ) & \text{if $ { \bar { v}}\in { \mathcal{g}}_{\alpha'}$ } \\ 0 & \text{if $ { \bar { v}}=\mu^\dagger_{\alpha'}$ } \\
\infty & \text{else.}\end{cases}\ ] ] finally , @xmath163 is continuous on @xmath164 and @xmath165 is continuous on @xmath166 .
the proof of this result will be one of the main goals of appendix b. we note that assumptions [ a : ig ] ( recall ) and [ a : nondegen ] imply that @xmath167 .
one more final piece of notation is needed . for @xmath145 and @xmath157 , define @xmath168}\phi(p,\lambda(\alpha',{\bar { v}})){\left\ { } \newcommand{\rb}{\right\}}1-\phi(p,\lambda(\alpha',{\bar { v}}))\rb { \bar { v}}(dp).\ ] ] lemma [ l : sigmalim ] ensures that @xmath169 . [ t : main ] we have that @xmath170 = \frac{e^{-{\textsf{r}}t}\exp\left[-\lambda(\alpha,{\bar { u}})\left({\lceil n \alpha \rceil}-n \alpha\right)\right]}{n^{3/2}(\beta-\alpha)\sqrt{2\pi\sigma^2(\alpha,{\bar { u}})}}\\ \times { \left\ { } \newcommand{\rb}{\right\}}\frac{e^{-\lambda(\alpha,{\bar { u}})}}{(1-e^{-\lambda(\alpha,{\bar { u}})})^2 } + \frac{{\lceil n \alpha \rceil}-n \alpha}{1-e^{-\lambda(\alpha,{\bar { u } } ) } } + { \mathcal{e}}(n)\rb \exp\left[-n { \mathfrak{i}}(\alpha,{\bar { u}^{(n)}})\right ] \end{gathered}\ ] ] where @xmath171 . the organization of the proof is in section [ s : asympanal ] . as in @xcite ,
the granularity @xmath172 is unavoidable in a result of this resolution . as we had in @xcite , @xmath173=\sum_{t\in { \mathcal{t } } } e^{-{\textsf{r}}t}$ ] so the asymptotic behavior of the premium @xmath140 is given by @xmath174}{n^{3/2}(\beta-\alpha)\sqrt{2\pi\sigma^2(\alpha,{\bar { u}})}{\left\ { } \newcommand{\rb}{\right\}}\sum_{t\in { \mathcal{t } } } e^{-{\textsf{r}}t}\rb}\\ & \qquad \times { \left\ { } \newcommand{\rb}{\right\}}\frac{e^{-\lambda(\alpha,{\bar { u}})}}{(1-e^{-\lambda(\alpha,{\bar { u}})})^2 } + \frac{{\lceil n \alpha \rceil}-n \alpha}{1-e^{-\lambda(\alpha,{\bar { u } } ) } } + { \mathcal{e}}'(n)\rb \exp\left[-n { \mathfrak{i}}(\alpha,{\bar { u}^{(n)}})\right ] \end{aligned}\ ] ] where @xmath175 .
[ r : discretization ] although the dominant exponential asymptotics follows from theorem [ t : main ] , we can not replace @xmath176 in theorem [ t : main ] by @xmath177 ; the pre - exponential asymptotics of theorem [ t : main ] are at too fine a resolution to allow that .
a careful examination of the calculations of lemma [ l : ficont ] reveals that @xmath176 and @xmath177 should differ by something on the order of the distance ( in the prohorov metric ) between @xmath178 and @xmath150 . in general
, we should expect that this distance would be of order @xmath179 ; as an example consider approximating a uniform distribution on @xmath143 by point masses at multiples of @xmath179 .
then we would have that @xmath180 .
this @xmath181 term would contribute to the pre - exponential asymptotics of theorem [ t : main ] . to close this section ,
we refer the reader to section [ s : merton ] , where we simulate our results for the merton model of example [ ex : merton ] .
we also point out that it would not be hard to combine the calculations of sections [ s : measurechange ] and [ s : asympanal ] to get an asymptotic formula for the loss given default of the cdo .
the terms in front of the @xmath182 $ ] in theorem [ t : main ] would be a major part of the resulting expression for loss given default .
we hope to pursue this elsewhere .
we can now introduce a simple model of correlation without too much trouble .
assume that @xmath31 takes values in a finite set @xmath7 .
fix @xmath183 such that @xmath184 and @xmath185 for all @xmath186 ; we will assume that @xmath31 takes on the value @xmath42 with probability @xmath187 .
we can think of the set @xmath7 as the collection of possible states of the world .
if we believe in , we should then be in the previous case if we condition on the various values of @xmath31 . to formalize this , fix a @xmath188 .
for each @xmath11 , fix @xmath13 such that @xmath189 for all @xmath57 . to adapt the previous calculations to this case , we need the analogue of assumptions [ a : limitexists ] , [ a : ig ] , [ a : nondegen ] , and [ a : notflat ] .
namely , we need that the limit @xmath190 exists for each @xmath186 , we need that @xmath191}p{\bar { u}}(dp , x)<\alpha \qquad \text{and}\qquad \max_{x\in { \mathsf{x}}}{\bar { u}}(\{0\},x)<1-\alpha\ ] ] and we finally need that @xmath192 [ r : systemic ] the requirement that @xmath193 is a particularly unrealistic one .
it means that the tranche losses will be rare for _ all _ values of the systemic parameter . in any truly applicable model
, the losses will come from a combination of bad values of the systemic parameter and from tail events in the pool of idiosyncratic randomness ( i.e. , we need to balance the size of @xmath49 against the distribution of @xmath31 ) .
one can view our effort here as study which focusses primarily on tail events in the pool of idiosyncratic randomness .
any structural model which attempts to study losses due to both idiosyncratic and systemic randomness will most likely involve calculations which are similar in a number of ways to ours here .
we will explore this issue elsewhere . then @xmath170 = \frac{e^{-{\textsf{r}}t}}{n^{3/2}(\beta-\alpha)}\\ \times \sum_{x\in { \mathsf{x}}}\left(\frac{\exp\left[-\lambda(\alpha,{\bar { u}}_x)\left({\lceil n \alpha \rceil}-n \alpha\right)\right]}{\sqrt{2\pi\sigma^2(\alpha,{\bar { u}})}}{\left\ { } \newcommand{\rb}{\right\}}\frac{e^{-\lambda(\alpha,{\bar { u}}_x)}}{(1-e^{-\lambda(\alpha,{\bar { u}}_x)})^2 } + \frac{{\lceil n
\alpha \rceil}-n \alpha}{1-e^{-\lambda(\alpha,{\bar { u}}_x ) } } + { \mathcal{e}}_x(n)\rb \right.\\ \left.\times \exp\left[-n { \mathfrak{i}}(\alpha,{\bar { u}^{(n)}}_x)\right]p(x)\right ) \end{gathered}\ ] ] where @xmath194 for each @xmath186 .
similarly , we have that @xmath195}{\sqrt{2\pi\sigma^2(\alpha,{\bar { u}}_x)}}{\left\ { } \newcommand{\rb}{\right\}}\frac{e^{-\lambda(\alpha,{\bar { u}}_x)}}{(1-e^{-\lambda(\alpha,{\bar { u}}_x)})^2 } + \frac{{\lceil n \alpha \rceil}-n \alpha}{1-e^{-\lambda(\alpha,{\bar { u}}_x ) } } + { \mathcal{e}}'(n)\rb\right.\\ \left .
\times \exp\left[-n { \mathfrak{i}}(\alpha,{\bar { u}^{(n)}}_x)\right]p(x)\right ) \end{gathered}\ ] ] where @xmath196 for all @xmath186 .
if we further assume that there is a unique @xmath197 such that @xmath198 for @xmath11 sufficiently large , we furthermore have that @xmath199 & = \frac{e^{-{\textsf{r}}t}}{n^{3/2}(\beta-\alpha)}\\ & \qquad \times \frac{\exp\left[-\lambda(\alpha,{\bar { u}}_{x^*})\left({\lceil n \alpha \rceil}-n \alpha\right)\right]}{\sqrt{2\pi\sigma^2(\alpha,{\bar { u}})}}{\left\ { } \newcommand{\rb}{\right\}}\frac{e^{-\lambda(\alpha,{\bar { u}}_{x^*})}}{(1-e^{-\lambda(\alpha,{\bar { u}}_{x^*})})^2 } + \frac{{\lceil n \alpha \rceil}-n \alpha}{1-e^{-\lambda(\alpha,{\bar { u}}_{x^ * } ) } } + { \mathcal{e}}(n)\rb \\ & \qquad \times \exp\left[-n { \mathfrak{i}}(\alpha,{\bar { u}^{(n)}}_{x^*})\right]p(x^ * ) \\ s_n & = \frac{e^{-{\textsf{r}}t}}{n^{3/2}(\beta-\alpha){\left\ { } \newcommand{\rb}{\right\}}\sum_{t\in { \mathcal{t } } } e^{-{\textsf{r}}t}\rb}\\ & \qquad \times \frac{\exp\left[-\lambda(\alpha,{\bar { u}}_{x^*})\left({\lceil n \alpha \rceil}-n \alpha\right)\right]}{\sqrt{2\pi\sigma^2(\alpha,{\bar { u}}_{x^*})}}{\left\ { } \newcommand{\rb}{\right\}}\frac{e^{-\lambda(\alpha,{\bar { u}}_{x^*})}}{(1-e^{-\lambda(\alpha,{\bar { u}}_{x^*})})^2 } + \frac{{\lceil n \alpha \rceil}-n \alpha}{1-e^{-\lambda(\alpha,{\bar { u}}_{x^ * } ) } } + { \mathcal{e}}'(n)\rb\\ & \qquad \times \exp\left[-n { \mathfrak{i}}(\alpha,{\bar { u}^{(n)}}_{x^*})\right]p(x^ * ) \end{aligned}\ ] ] where @xmath171 and @xmath175 . note that we can use this methodology to approximately study gaussian correlations .
fix a positive @xmath200 and define @xmath201 for @xmath202 ; set @xmath203 .
define @xmath204dt \qquad x\in { \mathbb{r}}\ ] ] as the standard gaussian cumulative distribution function .
define @xmath205 if we have a pool of @xmath12 names and the risk - neutral probabilities of default of the @xmath3-th bond by time @xmath25 is @xmath206 and we want to consider a gaussian copula with correlation @xmath207 ( the case @xmath208 can be dealt with similarly ) , we would take the @xmath209 s such that @xmath210 this is related to the calculations of @xcite and @xcite ; those calculations are asymptotically related to our calculations .
we shall explore the connection with these two papers elsewhere .
we note , by way of contrast with @xcite and @xcite , that our efforts give a good picture of the _ dynamics _ of the loss process prior to expiry .
we also note that our model of is entirely comfortable with non - gaussian correlation .
note also that one could also ( by discretization ) allow the systemic parameter @xmath31 to be path - valued .
the starting point for our analysis is the random measure @xmath211 then @xmath212 $ ] . as in @xcite , we want to compute the asymptotic ( for large @xmath12 ) likelihood that @xmath213 .
we want to do this via a collection of arguments stemming from the theory of large deviations .
the value of the calculations in this section is that they naturally lead to a measure transformation ( cf .
section [ s : measurechange ] ) which will lead to the precise asymptotics of theorem [ t : main ] . for the moment , it is sufficient for our arguments to be formal ; it is sufficient to _ guess _ a large deviations rate functional for @xmath43 . in the ensuing parts of this paper
we will show that this guess is correct ( cf .
section [ s : asympanal ] ) .
define now @xmath214 for our calculations here in this section , we will assume that @xmath215 exists ( as a limit in @xmath216 ) .
see example [ ex : walk ] .
our approach is similar to that of @xcite ; we first identify a large deviations principle for @xmath217 , and then use the contraction principle to find what should be a rate function for @xmath43 .
we hopefully can identify the large deviations principle for @xmath217 by looking at the asymptotic moment generating function for @xmath217 and appealing to the grtner - ellis theorem .
the following result gets us started .
[ l : momgen ] for @xmath218 , @xmath219\right ] = \int_{\rho\in { { \mathscr{p}}(i)}}{\left\ { } \newcommand{\rb}{\right\}}\ln \int_{t\in i}e^{\varphi(t)}\rho(dt)\rb { u}(d\rho).\ ] ] to make this a bit clearer , let s first carry out these calculations for our test case .
for example [ ex : simpleexample ] , @xmath220\right ] = \lim_{n\to \infty}\frac{1}{n}\ln { \mathbb{e}}_n\left[\exp\left[\sum_{n=1}^n \varphi(\tau_n)\right]\right]\\ & \qquad = \lim_{n\to \infty}\frac{1}{n}\ln \prod_{n=1}^n { \mathbb{e}}_n\left[\exp\left[\varphi(\tau_n)\right]\right]\\ & \qquad = \lim_{n\to \infty}\frac{1}{n}{\left\ { } \newcommand{\rb}{\right\}}\left \lfloor \frac{n}{3}\right\rfloor \ln \int_{t\in i}e^{\varphi(t)}\check \mu_a(dt ) + \left(n-\left \lfloor \frac{n}{3}\right \rfloor\right ) \ln \int_{t\in i}e^{\varphi(t)}\check \mu_b(dt)\rb \\ & \qquad = \frac{1}{3 } \ln \int_{t\in i}e^{\varphi(t)}\check \mu_a(dt ) + \frac{2}{3}\ln \int_{t\in i}e^{\varphi(t)}\check \mu_b(dt ) .
\end{aligned}\ ] ] we can now prove the result in full generality . for every @xmath12 , @xmath221\right ] = \frac{1}{n}\ln { \mathbb{e}}_n\left[\exp\left[\sum_{n=1}^n \varphi(\tau_n)\right]\right]\\ & \qquad = \frac{1}{n}\ln \prod_{n=1}^n { \mathbb{e}}_n\left[\exp\left[\varphi(\tau_n)\right]\right ] = \frac{1}{n}\sum_{n=1}^n \ln \int_{t\in i}e^{\varphi(t)}\mu^{(n)}_n(dt)\\ & \qquad = \frac{1}{n}\sum_{n=1}^n \int_{\rho\in { { \mathscr{p}}(i)}}{\left\ { } \newcommand{\rb}{\right\}}\ln \int_{t\in i}e^{\varphi(t)}\rho(dt)\rb \delta_{\mu^{(n)}_n}(d\rho)= \int_{\rho\in { { \mathscr{p}}(i)}}{\left\ { } \newcommand{\rb}{\right\}}\ln \int_{t\in i}e^{\varphi(t)}\rho(dt)\rb { u}^{(n)}(d\rho ) . \end{aligned}\ ] ] now use remark [ r : continuity ] ; the claimed result thus follows .
we next appeal to the insights of large deviations theory .
we expect . ] that @xmath217 will be governed by a large deviations principle ( in @xmath60 ) with rate function , we treat @xmath60 as a subset of @xmath222 .
] @xmath223 by the contraction principle of large deviations , we then expect that @xmath224 should be governed by a large deviations principle ( in @xmath86 $ ] ) with rate function @xmath225 for all @xmath145 .
while all of this this looks very intimidating , there should be an entropy representation similar to that of @xcite .
[ ex : twobonds ] in example [ ex : simpleexample ] , we have that @xmath217 is equal in law to @xmath226 where @xmath227 is the empirical measure of @xmath3 i.i.d . random variables with law @xmath58 , and @xmath228 is the empirical measure of @xmath3 i.i.d
. random variables with law @xmath59 , and where the @xmath229 s and @xmath230 s are independent . by standard large deviations results
, we can see that @xmath231 is a @xmath232-valued random variable and , as @xmath94 , that it has a large deviations principle with rate function @xmath233 i.e. , @xmath234\ ] ] for `` regular '' subsets @xmath65 of @xmath232 .
the @xmath235 and @xmath236 in stems from the fact that @xmath237 is the sum of about @xmath238 point masses , while @xmath239 is the sum of about @xmath240 point masses ; on the other hand , the rate in is @xmath12 .
we thus have from the contraction principle that @xmath217 has a large deviations principle with rate function @xmath241
we can see this directly from ; @xmath242 here we have used the duality between entropy and exponential integrals see .
we would now like to appeal to a minimax result for lagrangians and switch the @xmath243 and @xmath244 .
note that @xmath232 is a convex subset of @xmath245 and that @xmath246 is convex on @xmath232 .
apart from the problems arising from the fact that @xmath247 and @xmath232 are infinite - dimensional , a minimax result thus looks reasonable .
let s see where this leads .
we should have that @xmath248 thus @xmath249 we can then use lemma 7.1 from @xcite to simplify things even further .
we have that @xmath250,\right.\\ & \qquad \qquad \left.\frac13 p_a[0,t)+\frac23 p_b[0,t)=\alpha ' , m_a[0,t)=p_a,\ , m_b[0,t)=p_b\rb \\ & = \inf{\left\ { } \newcommand{\rb}{\right\}}\frac13 \hbar(p_a|\check \mu_a[0,t ) ) + \frac23 \hbar(p_b|\check \mu_b[0,t ) ) : p_a , p_b\in [ 0,1],\ , \frac13 p_a[0,t)+\frac23 p_b[0,t)=\alpha'\rb .
\end{aligned}\ ] ] this leads to the following generalization . [
l : variational ] we have that @xmath251 where @xmath156 is as in .
we give the proof in appendix b. in example [ ex : simpleexample ] , we have that @xmath252 where @xmath253 satisfies @xmath254 in example [ ex : merton ] , we have that @xmath255 where @xmath253 satisfies @xmath256
let s start to reconnect our thoughts to our goal the asymptotic behavior of the protection leg .
namely , we want a formula which is the analog of theorem 4.1 of @xcite .
recall that the starting point of much of our analysis in section [ s : ld ] was lemma [ l : momgen ] and .
theorem 4.1 of @xcite on the other hand involves a change of measure for a finite number of the @xmath4 s . to set the stage for doing the same in our case ,
let s begin with a technical lemma .
[ l : increasing ] for @xmath12 large enough , @xmath257 and the map @xmath258 is continuous and nondecreasing on the interval @xmath259}p{\bar { u}^{(n)}}(dp),1-{\bar { u}^{(n)}}\{0\}\right).\ ] ] finally , @xmath260 the proof is in appendix b and uses assumptions [ a : ig ] and [ a : nondegen ] . although we wo nt explicitly use it , the fact that @xmath261 is increasing on @xmath262 is a natural requirement from the standpoint of large deviations .
a more precise form of would be that @xmath263\ ] ] as @xmath94 . from lemma
[ l : increasing ] , we get that @xmath264 see also proposition 3.6 in @xcite .
let s next reverse the arguments of section [ s : ld ] .
using lemma [ l : finalitmin ] to identify the minimizer @xmath176 in , we can reconstruct an @xmath265 similar to and which allows us to construct a near - optimal measure transformation ( by near - optimal we mean that the measure - transformation will be define by @xmath112 rather than @xmath150 ) . in order to introduce even more notation , for each @xmath11 and @xmath51 ,
let s set @xmath266)&\text{if $ { \mathfrak{u}}^{(n)}_n\in ( 0,1)$ } \\
\mu^{(n)}_n(a ) & \text{if $ { \mathfrak{u}}^{(n)}_n\in \{0,1\}$}\end{cases } \qquad a\in { \mathscr{b}}(i)\end{aligned}\ ] ] [ r : phiprops ] we will also need some facts about @xmath153 , so we here collect them .
we first note that @xmath267}{\left\ { } \newcommand{\rb}{\right\}}1-p+pe^\lambda\rb = \begin{cases } \inf_{p\in [ 0,1]}{\left\ { } \newcommand{\rb}{\right\}}1+p(e^\lambda-1)\rb & \text{if $ \lambda\ge 0$}\\ \inf_{p\in [ 0,1]}{\left\ { } \newcommand{\rb}{\right\}}1-p\left(1-e^\lambda\right)\rb & \text{if $ \lambda<0$}\end{cases } = \begin{cases } 1 & \text{if $ \lambda\ge 0 $ } \\
e^\lambda & \text{if $ \lambda<0 $ } \end{cases } = e^{\lambda^-}>0\ ] ] where @xmath268 ; thus the denominator of @xmath153 is always strictly positive for @xmath269 .
hence @xmath270 is in @xmath271 $ ] for all @xmath269 .
next , we note that if @xmath269 , then @xmath272 if and only if @xmath273 , and @xmath274 if and only if @xmath275 .
we can also take derivatives . for @xmath276 and @xmath269 , @xmath277 and thus @xmath278
finally , we have that @xmath279 is continuous on @xmath280 $ ] for each @xmath109 $ ] . in light of these thoughts
, we note that @xmath281 and if so , @xmath282 .
we can now make several calculations about .
first , @xmath283 with @xmath284}(t)&\text{if $ { \mathfrak{u}}^{(n)}_n\in ( 0,1)$ } \\ 1 & \text{if $ { \mathfrak{u}}^{(n)}_n\in \{0,1\}$}\end{cases}\ ] ] for all @xmath285 .
in light of , we have that each @xmath286 is finite and strictly positive .
let s also note that @xmath287}\phi(p,\lambda(\alpha,{\bar { u}^{(n)}})){\bar { u}^{(n)}}(dp ) = \alpha . \end{aligned}\ ] ] ( clearly @xmath288 if @xmath289 ; by , we also have that @xmath290 if @xmath291 ) .
[ t : measurechange ] we have that @xmath292 = i_ne^{-n{\mathfrak{i}}(\alpha)}\ ] ] for all positive integers @xmath12 , where @xmath293\chi_{\{\gamma_n>0\}}\right]\ ] ] where in turn @xmath294 \qquad a\in { \mathscr{f}}\\ \gamma_n&= \sum_{n=1}^n { \left\ { } \newcommand{\rb}{\right\}}\chi_{[0,t)}(\tau_n)-\alpha\rb = n(l_{t-}^{(n)}-\alpha)).\end{aligned}\ ] ] under @xmath295 , @xmath296 are independent and @xmath4 has law @xmath297 for @xmath51
. set @xmath298}(t ) & \text{if $ { \mathfrak{u}}^{(n)}_n\in ( 0,1)$ } \\ 0 & \text{if $ { \mathfrak{u}}^{(n)}_n\in \{0,1\}$}\end{cases } \qquad
t\in i\\ \gamma_n & { \overset{\text{def}}{=}}\sum_{n=1}^n \psi^{(n)}_n(\tau_n)- \sum_{n=1}^n \int_{t\in i}\psi^{(n)}_n(t)\tilde \mu^{(n)}_n(dt)\end{aligned}\ ] ] ( as we pointed out above , each @xmath299 is positive and finite on all of @xmath1 , ensuring that @xmath300 is well - defined ) .
then @xmath292 = \frac{{\mathbb{e}}_n\left[{\textbf{p}^{\text{prot}}}_n \exp\left[-\gamma_n\right]\exp\left[\gamma_n\right]\right]}{{\mathbb{e}}_n\left[\exp\left[\gamma_n\right]\right]}{\mathbb{e}}_n\left[\exp\left[\gamma_n\right]\right].\ ] ] some straightforward calculations ( recall ) show that @xmath301}\hbar(\phi(p,\lambda(\alpha,{\bar { u}^{(n)}})),p){\bar { u}^{(n)}}(dp)\\ & = n{\mathfrak{i}}(\alpha,{\bar { u}^{(n)}})\\ \exp\left[\sum_{n=1}^n \psi^{(n)}_n(\tau_n)\right]&= \exp\left[\sum_{n=1}^n\ln \frac{d\tilde \mu^{(n)}_n}{d\mu^{(n)}_n}(\tau_n)\right ] = \prod_{n=1}^n \frac{d\tilde \mu^{(n)}_n}{d\mu^{(n)}_n}(\tau_n ) \end{aligned}\ ] ] we chose @xmath300 exactly so that the following calculation holds : @xmath302\right]= e^{-n{\mathfrak{i}}(\alpha,{\bar { u}^{(n)}})}{\mathbb{e}}_n\left[\prod_{n=1}^n \frac{d\tilde \mu^{(n)}_n}{d\mu^{(n)}_n}(\tau_n)\right ] = e^{-n{\mathfrak{i}}(\alpha,{\bar { u}^{(n)}})}.\ ] ] we also clearly have that @xmath303\right]}{{\mathbb{e}}_n\left[\exp\left[\gamma_n\right]\right]}\\ = \frac{{\mathbb{e}}_n\left[\chi_a\exp\left[\sum_{n=1}^n \psi^{(n)}_n(\tau_n)\right]\right]}{{\mathbb{e}}_n\left[\exp\left[\sum_{n=1}^n \psi^{(n)}_n(\tau_n)\right]\right ] } = \tilde { \mathbb{p}}_n(a)\ ] ] for all @xmath304 .
the properties of @xmath295 are clear from the explicit formula .
finally , it is easy to check that @xmath305}(\tau_n)-\tilde \mu^{(n)}_n[t,\infty]\rb \\ & = \sum_{\substack{1\le n\le n \\ { \mathfrak{u}}^{(n)}_n\in ( 0,1 ) } } { \left\ { } \newcommand{\rb}{\right\}}\ln \frac{\tilde { \mathfrak{u}}^{(n)}_n}{{\mathfrak{u}}^{(n)}_n}-\ln \frac{1-\tilde { \mathfrak{u}}^{(n)}_n}{1-{\mathfrak{u}}^{(n)}_n } \rb { \left\ { } \newcommand{\rb}{\right\}}\chi_{[0,t)}(\tau_n)-\tilde \mu^{(n)}_n[0,t)\rb \\ & = \sum_{\substack{1\le n\le n \\ { \mathfrak{u}}^{(n)}_n\in ( 0,1 ) } } \ln \left(\frac{\phi({\mathfrak{u}}^{(n)}_n,\lambda(\alpha,{\bar { u}^{(n)}}))}{1-\phi({\mathfrak{u}}^{(n)}_n,\lambda(\alpha,{\bar { u}^{(n ) } } ) ) } \frac{1-{\mathfrak{u}}^{(n)}_n}{{\mathfrak{u}}^{(n)}_n}\right ) { \left\ { } \newcommand{\rb}{\right\}}\chi_{[0,t)}(\tau_n)-\tilde \mu^{(n)}_n[0,t)\rb .
\end{aligned}\ ] ] a straightforward calculation shows that for any @xmath276 and @xmath269 , @xmath306 recall now and note that if @xmath307 , then @xmath54-a.s .
@xmath308 , while if @xmath309 then @xmath54-a.s .
@xmath310 . thus @xmath54-a.s .
@xmath311 combining things together , we get that @xmath54-a.s .
, @xmath312 recall now . by and
, we see that @xmath313 is nonzero only if @xmath314 ; we have explicitly included this in the expression for @xmath315 .
we proceed now as in @xcite .
define @xmath316 ; then @xmath317 is the nonnegative collection of values which @xmath318 can take . for each @xmath12 ,
let @xmath319 $ ] be such that @xmath320\ ] ] on @xmath321 ( recall from that @xmath322 ; using this in , we have that @xmath323 $ ] )
. then @xmath324\right].\ ] ] the behavior of @xmath325 is very nice for large @xmath12 .
[ l : hasymp ] for all @xmath12 , we have that @xmath326 where @xmath327 section [ s : has ] is dedicated to the proof of this result .
we can also see that the distribution of @xmath318 is nice for large @xmath12 .
the proof of this result is qualitatively different than the corresponding proof of lemma 5.2 in @xcite .
[ l : probasymp ] we have that @xmath328 for all @xmath12 and all @xmath329 , where @xmath330 is as in and where @xmath331 the proof of this is the subject of section [ s : fourier ] ; the result is in some sense a statement of convergence in the `` vague '' topology .
we can now set up the proof theorem [ t : main ] .
for @xmath332 , define @xmath333{\left\ { } \newcommand{\rb}{\right\}}\frac{e^{-\lambda}}{(1-e^{-\lambda})^2 } + \frac{{\lceil n \alpha \rceil}-n\alpha}{1-e^{-\lambda}}\rb \end{aligned}\ ] ] then , as in @xcite , @xmath334{\left\ { } \newcommand{\rb}{\right\}}\frac{e^{-\lambda}}{(1-e^{-\lambda})^2 } + \frac{{\lceil n \alpha \rceil}-n \alpha}{1-e^{-\lambda } } + { \mathcal{e}}_3(\lambda , n)\rb\ ] ] where there is a @xmath335 such that @xmath336}{\lambda(1-e^{-\lambda})^2}\ ] ] for all positive integers @xmath12 and all @xmath332 .
we have that @xmath337 where @xmath338 \\ \tilde { \mathcal{e}}_2(n)&{\overset{\text{def}}{=}}\frac{1}{\sqrt{2\pi n\sigma^2(\alpha,{\bar { u}})}}\sum_{\substack{s\in { \mathsf{s}}_n\\s\le n^{1/4}}}h_n(s)e^{-\lambda(\alpha,{\bar { u}^{(n ) } } ) s}{\mathcal{e}}_2(s , n ) \\ \tilde { \mathcal{e}}_3(n)&{\overset{\text{def}}{=}}\frac{e^{-{\textsf{r}}t}}{n^{3/2}(\beta-\alpha)\sqrt{2\pi \sigma^2(\alpha,{\bar { u}})}}\sum_{\substack{s\in { \mathsf{s}}_n\\s\le n^{1/4}}}se^{-\lambda(\alpha,{\bar { u}^{(n ) } } ) s}{\mathcal{e}}_1(s , n ) \\ \tilde { \mathcal{e}}_4(n)&{\overset{\text{def}}{=}}\frac{e^{-{\textsf{r}}t}}{n^{3/2}(\beta-\alpha)\sqrt{2\pi \sigma^2(\alpha,{\bar { u}})}}\exp\left[-\lambda(\alpha,{\bar { u}^{(n)}})\left({\lceil n \alpha \rceil}-n \alpha\right)\right]{\mathcal{e}}_3(\lambda(\alpha,{\bar { u}^{(n)}}),n)\\ \tilde { \mathcal{e}}_5(n)&{\overset{\text{def}}{=}}\frac{e^{-{\textsf{r}}t}}{n^{3/2}(\beta-\alpha)\sqrt{2\pi \sigma^2(\alpha,{\bar { u}})}}{\left\ { } \newcommand{\rb}{\right\}}\tilde i_{2,n}(\lambda(\alpha,{\bar { u}^{(n)}}))-\tilde i_{2,n}(\lambda(\alpha,{\bar { u}}))\rb \end{aligned}\ ] ] keep in mind now the second claim of . we see that there is a @xmath339 such that for sufficiently large @xmath12 @xmath340 furthermore , we can fairly easily see that there is a @xmath341 such that @xmath342 for all sufficiently large @xmath12 ( note from and that @xmath343 is uniformly bounded in @xmath12 as long as @xmath344 is bounded away from zero from below ) . finally , we get that there is a @xmath345 such that @xmath346 for sufficiently large @xmath12 .
combine things together within the framework of theorem [ t : measurechange ] to get the stated result .
as an example of how the computations of section [ s : model ] work , let s delve a bit more deeply into example [ ex : merton ] . to be very explicit , let s assume that all the names are governed by the merton model with risk - neutral drift @xmath347 , initial valuation @xmath68 , and bankruptcy barrier @xmath348 .
we assume that expiry is @xmath349 .
assume that the volatility is distributed according to a gamma distribution with size parameter @xmath350 and shape parameter @xmath351 ; @xmath150 is then given by .
numerical integration shows that @xmath95}p{\bar { u}}(dp)=.0738.\ ] ] to understand how our calculations work , figure [ fig : fa ] is a plot of the function @xmath352}\phi(p,\lambda){\bar { u}}(dp).\ ] ] thus if the attachment point of the tranche is @xmath353 , we would have @xmath354 .
} \phi(p,\lambda){\bar { u}}(dp)$],width=480,height=288 ] let s next explicitly construct some @xmath355 s as in .
we do this as follows .
define @xmath356 for all @xmath357 ( using the fact that @xmath358 ) . for each @xmath12 , define @xmath359 for @xmath51 .
set @xmath360 for all @xmath51 .
then for every @xmath74 , @xmath361 we can then finally plot the `` theoretical '' cdo price against the number @xmath12 of names for several values of @xmath18 .
the results are in figure [ fig : fb ] for three values of @xmath18 . by `` theoretical '' ,
we mean the quantity @xmath362}{n^{3/2}(\beta-\alpha)\sqrt{2\pi\sigma^2(\alpha,{\bar { u}})}{\left\ { } \newcommand{\rb}{\right\}}\sum_{t\in { \mathcal{t } } } e^{-{\textsf{r}}t}\rb}\\ \times { \left\ { } \newcommand{\rb}{\right\}}\frac{e^{-\lambda(\alpha,{\bar { u}})}}{(1-e^{-\lambda(\alpha,{\bar { u}})})^2 } + \frac{{\lceil n \alpha \rceil}-n \alpha}{1-e^{-\lambda(\alpha,{\bar { u } } ) } } + { \mathcal{e}}'(n)\rb \exp\left[-n { \mathfrak{i}}(\alpha,{\bar { u}^{(n)}})\right ] \end{gathered}\ ] ] we have here set @xmath363 in .
figure [ fig : fb ] also removes the prefactor @xmath364 for several values of @xmath18,width=480,height=288 ]
we here study @xmath325 .
large sections of the proof will be similar to section 5 of @xcite . set @xmath365 on @xmath321 , @xmath366\cap[0,t)}e^{-{\textsf{r}}s}d{\bar l}^{(n)}_s.\ ] ]
the heart of lemma [ l : hasymp ] is the following result , the proof of which is at the end of this section .
[ l : ttimes ] we have that @xmath367\chi_{\{\gamma_n = s\}}=0.\ ] ] we assume that @xmath368 and that @xmath369 ( thus @xmath370 ) .
then as in section 7 of @xcite , we have that @xmath371\cap [ 0,t)}e^{-{\textsf{r}}s}d{\bar l}^{(n)}_s = e^{-{\textsf{r}}t}\frac{l^{(n)}_{t-}-\alpha}{\beta-\alpha } + { \textsc{\tiny e}}_n = \frac{e^{-{\textsf{r}}t}}{\beta-\alpha}\frac{\gamma_n}{n } + { \textsc{\tiny e}}_n\ ] ] where @xmath372 furthermore , we have that @xmath373 then for @xmath374 and @xmath329 such that @xmath375 , we have that @xmath376}{\frac{\gamma_n}{n}}\chi_{\{\gamma_n = s\ } } \le e^{{\textsf{r}}t}{\left\ { } \newcommand{\rb}{\right\}}\frac{1}{t}+{\textsf{r}}\rb \tilde { \mathbb{e}}_n[t-\tau^\alpha_n|\gamma_n]\chi_{\{\gamma_n = s\}}.\ ] ] combine and and lemma [ l : ttimes ] .
we now need to prove lemma [ l : ttimes ] . as in @xcite
, we will use the martingale problem as applied to a time - reversed martingale .
define @xmath377 for each positive integer @xmath3 ( note that the @xmath378 s are right - continuous , have left - hand limits , and are nonincreasing ) . also define @xmath379 for all @xmath380 and @xmath11 .
observe that @xmath381\ ] ] for all @xmath380 , @xmath11 , and @xmath51 , define @xmath382 } { z^{(n)}}_{(t - r)-}\mu^{(n)}_n(dr)\\ m^{(n , n)}_t&{\overset{\text{def}}{=}}{z^{(n)}}_t-{z^{(n)}}_0-a^{(n , n)}_t .
\end{aligned}\ ] ] several comments are in order concerning @xmath383 .
the integrand @xmath384){z^{(n)}}_{(t - r)-}$ ] is nonnegative ( but possibly infinite ) , so @xmath383 is well - defined ( but possibly infinite ) via the theory of lebesgue integration ; we can approximate @xmath385 $ ] from below via simple functions .
also , @xmath383 is negative , nonincreasing , and right - continuous . as we pointed out in @xcite , @xmath386 for all @xmath387 . thus @xmath388 & = \tilde { \mathbb{e}}_n\left[\int_{r\in ( 0,t ) } \frac{1}{\mu^{(n)}_n[0,r ] } { z^{(n)}}_{(t - r)-}\mu^{(n)}_n(dr)\right]\\ & = \int_{r\in ( 0,t ) } \frac{1}{\mu^{(n)}_n[0,r ] } \tilde { \mathbb{p}}_n\{\tau_n\le r\ } \mu^{(n)}_n(dr)\\ & = \begin{cases } \frac{\tilde { \mathfrak{u}}^{(n)}_n}{{\mathfrak{u}}^{(n)}_n } \mu^{(n)}(0,t ) & \text{if $ { \mathfrak{u}}^{(n)}_n\in ( 0,1)$ } \\
\mu^{(n)}_n(0,t ) & \text{if $ { \mathfrak{u}}^{(n)}_n\in \{0,1\}$ } \end{cases } \le \begin{cases } \tilde { \mathfrak{u}}^{(n)}_n & \text{if $ { \mathfrak{u}}^{(n)}_n\in ( 0,1)$ } \\ \mu^{(n)}_n(0,t ) & \text{if $ { \mathfrak{u}}^{(n)}_n\in \{0,1\}$ } \end{cases } \le 1.\end{aligned}\ ] ] thus @xmath389 is @xmath295-finite ( by tonelli s theorem ) . for every @xmath51
, @xmath390 is a @xmath295-zero - mean - martingale with respect to @xmath391 ; i.e. , for @xmath392 , @xmath393=m^{(n , n)}_s$ ] .
recall lemma 6.2 of @xcite and its proof .
measurability and integrability are clear ( use instead of ( 13 ) of @xcite ) .
define next @xmath394 : \mu^{(n)}_n[0,t - t)=0\rb \wedge t;\ ] ] then @xmath395 but @xmath396 for all @xmath397 .
we can thus use lemma 6.2 of @xcite to see that if @xmath398 , then @xmath393=m^{(n , n)}_s$ ] .
next , assume that @xmath399
. then @xmath400&= \tilde { \mathbb{e}}_n\left[\int_{r\in [ t - t , t - s)}\frac{1}{\mu^{(n)}_n[0,r]}\chi_{\{\tau_n\le r\}}\mu^{(n)}_n(dr)\right ] \\ & = \int_{r\in [ t - t , t - s)}\frac{1}{\mu^{(n)}_n[0,r]}\tilde { \mathbb{p}}_n\{\tau_n\le r\}\mu^{(n)}_n(dr)\\ \tilde { \mathbb{e}}_n\left[\left|{z^{(n)}}_t-{z^{(n)}}_s\right|\right ] & = \tilde { \mathbb{e}}_n\left[{z^{(n)}}_s-{z^{(n)}}_t\right ] = \tilde { \mathbb{p}}_n{\left\ { } \newcommand{\rb}{\right\}}t - t\le \tau_n < t - s\rb . \end{aligned}\ ] ] for any @xmath401 , we have that @xmath402 & \text{if $ { \mathfrak{u}}^{(n)}_n\in ( 0,1)$ } \\ \mu^{(n)}_n[0,r ] & \text{if $ { \mathfrak{u}}^{(n)}_n\in \{0,1\}$}\end{cases } \le \begin{cases } \frac{\tilde { \mathfrak{u}}^{(n)}_n}{{\mathfrak{u}}^{(n)}_n}\mu^{(n)}_n[0,t - t^*_n ) & \text{if $ { \mathfrak{u}}^{(n)}_n\in ( 0,1)$ } \\ \mu^{(n)}_n[0,t - t^*_n ) & \text{if $ { \mathfrak{u}}^{(n)}_n\in \{0,1\}$}\end{cases}=0.\ ] ] standard arguments thus imply that @xmath390 is @xmath295-a.s .
constant on @xmath403 , and so for any @xmath399 , we of course have that @xmath393=m^{(n , n)}_s$ ] .
finally , we claim that for any @xmath404 , @xmath405=0;\ ] ] if so , we can fairly easily conclude that @xmath406=m^{(n , n)}_s$ ] for any @xmath404 . by standard martingale - type arguments involving iterated conditioning
, this will finish the proof . to see
, we compute that @xmath407}\frac{1}{\mu^{(n)}_n[0,t - t^*_n]}\chi_{\{\tau_n\le t - t^*_n\}}\mu^{(n)}_n(dr).\ ] ] if @xmath408 , then @xmath409 and we note that @xmath54-a.s . ( and thus by absolute continuity @xmath295-a.s . )
@xmath410 on the other hand , assume that @xmath411
. then @xmath412 = \mu^{(n)}_n\{t - t^*_n\}>0 $ ] , and so @xmath54-a.s .
( and thus again by absolute continuity @xmath295-a.s . ) @xmath413 here we calculate that @xmath414 this proves and completes the proof .
let s now recombine things .
set @xmath415 for @xmath380 .
also observe that @xmath416 we next rewrite @xmath417 to be in reverse time .
set @xmath418 then , as in section 7 of @xcite , we know that @xmath419 .
fix now two parameters @xmath420 and @xmath421 .
we want to show ( this will occur in ) that it is unlikely that @xmath422 ; we want to do this by exploiting the equation @xmath423 assume now that in fact @xmath424 .
firstly , this implies that @xmath425 ( see figure 3 of @xcite ) .
thus @xmath426 on the other hand , we can combine and the second inequality of and the fact that the @xmath378 s are nonincreasing to see that for @xmath427 , @xmath428 also , @xmath429\le 1 $ ] , so some straightforward calculations show that @xmath430 where @xmath431 thanks to assumption [ a : notflat ] , we have that @xmath432 combining our above calculations going back to , we have that if @xmath433 , @xmath434 ( recall the first inequality of ) .
take conditional expectations of with respect to @xmath435 , and use the fact that @xmath43 is @xmath435-measurable .
thus if @xmath433 , @xmath436\rb.\ ] ] then ( heavily using the fact that the @xmath378 s are independent under @xmath295 ) , we get that @xmath437 \le \sqrt{\tilde { \mathbb{e}}_n\left[\left|\tilde m^{(n)}_\delta\right|^2\bigg|{\mathscr{g}^{(n)}}_0\right ] } \le { \left\ { } \newcommand{\rb}{\right\}}\frac{3}{n^2}\sum_{n=1}^n { \left\ { } \newcommand{\rb}{\right\}}2 + \tilde { \mathbb{e}}_n\left[\left|a^{(n , n)}_\delta\right|^2\bigg|{\mathscr{g}^{(n)}}_0\right]\rb \rb^{1/2}.\ ] ] we next compute that @xmath438 \le \tilde { \mathbb{e}}_n\left[\left|a^{(n , n)}_{t-}\right|^2\bigg|{\mathscr{g}^{(n)}}_0\right]\\ = \int_{r_1\in(0,t)}\int_{r_2\in ( 0,t)}\frac{1}{\mu^{(n)}_n[0,r_1]\mu^{(n)}_n[0,r_2]}\tilde { \mathbb{p}}_n{\left\ { } \newcommand{\rb}{\right\}}\tau_n\le r_1\wedge r_2\big|{\mathscr{g}^{(n)}}_0\rb \mu^{(n)}_n(dr_1)\mu^{(n)}_n(dr_2)\\ \le 2\int_{r_1\in(0,t)}\int_{r_2\in ( 0,r_1]}\frac{1}{\mu^{(n)}_n[0,r_1]\mu^{(n)}_n[0,r_2]}\tilde { \mathbb{p}}_n{\left\ { } \newcommand{\rb}{\right\}}\tau_n\le r_2\big|{\mathscr{g}^{(n)}}_0\rb \mu^{(n)}_n(dr_1)\mu^{(n)}_n(dr_2 )
. \end{gathered}\ ] ] if @xmath439 , then clearly @xmath440=0 $ ] .
assume next that @xmath441 .
for @xmath442 , @xmath443 thus for @xmath442 ( again using the fact that @xmath53 s are independent under @xmath295 ) we have that @xmath295-a.s . @xmath444}{\mu^{(n)}_n[0,t)}{z^{(n)}}_0 .
\end{gathered}\ ] ] thus @xmath445 \le \frac{2{z^{(n)}}_0}{\mu^{(n)}_n[0,t)}\int_{r_1\in ( 0,t)}\int_{r_2\in ( 0,r_1]}\frac{1}{\mu^{(n)}_n[0,r_1]}\mu^{(n)}_n(dr_2)\mu^{(n)}_n(dr_1)\\ \le\frac{2{z^{(n)}}_0}{\mu^{(n)}_n[0,t)}\int_{r_1\in(0,t)}\mu^{(n)}_n(dr_1 ) \le 2
. \end{gathered}\ ] ] summarizing thus far , we have that @xmath446 since @xmath447 , we have @xmath448 let s finally bound @xmath449 .
the above bound will show us that it is unlikely that @xmath422 . on the other hand , if @xmath450 , then in fact @xmath451 .
thus @xmath452 \le
\delta + \frac{t}{{\varepsilon}\delta_n({\varepsilon},\delta)}{\left\ { } \newcommand{\rb}{\right\}}\frac{\gamma^+_n}{n } + \frac{1}{n}+ \sqrt{\frac{12}{n}}\rb.\ ] ] in other words , @xmath453\chi_{\{\gamma_n = s\}}\le \delta + \frac{t}{{\varepsilon}\delta_n({\varepsilon},\delta)}{\left\ { } \newcommand{\rb}{\right\}}\frac{1}{n^{3/4 } } + \frac{1}{n}+ \sqrt{\frac{12}{n}}\rb.\ ] ] we now use . take @xmath94 , then @xmath137 , and finally @xmath138 .
let s start by representing @xmath454 as a fourier transform ; that will allow us to mimic various arguments from the central limit theorem . for @xmath11 and @xmath455 , define @xmath456 = \sum_{n=0}^n \exp\left[i\theta(n - n \alpha)\right]\tilde { \mathbb{p}}_n{\left\ { } \newcommand{\rb}{\right\}}\gamma_n = n- n \alpha\rb.\ ] ] thus @xmath457 = \sum_{n=0}^n e^{i\theta n}\tilde { \mathbb{p}}_n{\left\ { } \newcommand{\rb}{\right\}}\gamma_n = n- n \alpha\rb.\ ] ] thus for @xmath458 for some @xmath459 , @xmath460 e^{-i\theta n}d\theta \\
= \frac{1}{2\pi}\int_{\theta=-\pi}^{\pi } { \mathcal{p}}_n(\theta ) e^{-i\theta s}d\theta . \end{gathered}\ ] ] and so by a change of variables , @xmath461d\theta.\ ] ] this last representation is the same scaling as for the central limit theorem .
the advantage of using @xmath462 is that we can explicitly compute it .
we have that @xmath463\right ] = \prod_{n=1}^n \tilde { \mathbb{e}}_n\left[\exp\left[i\theta { \left\ { } \newcommand{\rb}{\right\}}\chi_{[0,t)}(\tau_n ) - \tilde { \mathfrak{u}}^{(n)}_n\rb\right]\right ] \\ & = \prod_{n=1}^n { \left\ { } \newcommand{\rb}{\right\}}\tilde { \mathbb{e}}_n\left[\exp\left[i\theta \chi_{[0,t)}(\tau_n)\right]\right]\exp\left[-i\theta \tilde { \mathfrak{u}}^{(n)}_n\right]\rb = \prod_{n=1}^n { \left\ { } \newcommand{\rb}{\right\}}\left(e^{i\theta}\tilde { \mathfrak{u}}^{(n)}_n + 1-\tilde { \mathfrak{u}}^{(n)}_n\right)\exp\left[-i\theta \tilde { \mathfrak{u}}^{(n)}_n\right]\rb\end{aligned}\ ] ] ( the part of the last equality due to @xmath3 for which @xmath289 is obvious ; for those @xmath3 for which @xmath291 we use ) we can now start to see the important asymptotic behavior of @xmath462 . before actually launching into the proof , we need to study @xmath464 of for a moment .
[ l : sigmalim ] we have that @xmath465 for each @xmath145 , the map @xmath466 is continuous and positive on @xmath166 .
we first observe that @xmath467}\phi(p,\lambda(\alpha,{\bar { u}^{(n)}})){\left\ { } \newcommand{\rb}{\right\}}1-\phi(p,\lambda(\alpha,{\bar { u}^{(n)}})\rb{\bar { u}^{(n)}}(dp ) = \sigma^2(\alpha,{\bar { u}^{(n ) } } ) . \end{gathered}\ ] ] if @xmath468 and @xmath469 in @xmath166 are such that @xmath470 , then we can write @xmath471}\left|\phi(p,\lambda(\alpha,{\bar { v}}_n)){\left\ { } \newcommand{\rb}{\right\}}1-\phi(p,\lambda(\alpha,{\bar { v}}_n))\rb -\phi(p,\lambda(\alpha,{\bar { v}})){\left\ { } \newcommand{\rb}{\right\}}1-\phi(p,\lambda(\alpha,{\bar { v}}))\rb\right|{\bar { v}}_n(dp)\\ + \left|\int_{p\in [ 0,1]}\phi(p,\lambda(\alpha,{\bar { v}})){\left\ { } \newcommand{\rb}{\right\}}1-\phi(p,\lambda(\alpha,{\bar { v}}))\rb{\bar { v}}_n(dp)-\int_{p\in [ 0,1]}\phi(p,\lambda(\alpha,{\bar { v}})){\left\ { } \newcommand{\rb}{\right\}}1-\phi(p,\lambda(\alpha,{\bar { v}}))\rb{\bar { v}}(dp)\right| \end{gathered}\ ] ] from remark [ r : phiprops ] and in a way similar to arguments in the proofs of lemmas [ l : lambdacont ] and [ l : ficont ] , we have that @xmath472}\left|\phi(p,\lambda(\alpha,{\bar { v}}_n)){\left\ { } \newcommand{\rb}{\right\}}1-\phi(p,\lambda(\alpha,{\bar { v}}_n))\rb -\phi(p,\lambda(\alpha,{\bar { v}})){\left\ { } \newcommand{\rb}{\right\}}1-\phi(p,\lambda(\alpha,{\bar { v}}))\rb\right|{\bar { v}}_n(dp)\\ \le \left|\lambda(\alpha,{\bar { v}}_n)-\lambda(\alpha,{\bar { v}})\right| \end{gathered}\ ] ] and we then use the continuity of lemma [ l : lambdacont ] in appendix b , and by weak convergence @xmath473}\phi(p,\lambda(\alpha,{\bar { v}})){\left\ { } \newcommand{\rb}{\right\}}1-\phi(p,\lambda(\alpha,{\bar { v}}))\rb{\bar { v}}_n(dp)=\int_{p\in [ 0,1]}\phi(p,\lambda(\alpha,{\bar { v}})){\left\ { } \newcommand{\rb}{\right\}}1-\phi(p,\lambda(\alpha,{\bar { v}}))\rb{\bar { v}}(dp).\ ] ] this proves the stated continuity . finally , if @xmath160 , then @xmath474 if and only if the integrand ( which is nonnegative ) in is @xmath469-a.s .
this occurs if and only if @xmath475 for @xmath469-a.e . @xmath109 $ ] , which , by remark [ r : phiprops ] , occurs if and only if @xmath476 .
but since @xmath160 , @xmath477 implying the desired positivity .
we also note that there are two positive constants @xmath478 and @xmath479 such that @xmath480 for all @xmath481 . indeed , the function @xmath482 is continuous and positive on @xmath483\setminus \{0\}$ ] , and @xmath484 .
a direct computation in particular thus shows that @xmath485 for all @xmath481 .
we will need two bounds in the proof of lemma [ l : probasymp ] .
the first bound is that @xmath486 is close to @xmath487 $ ] for @xmath67 not too large .
the second bound is that @xmath486 uniformly ( as @xmath94 ) decays in an integrable way in @xmath67 . by `` not too large '' we mean less than @xmath488
; we will use the fact that @xmath489 we first prove the desired asymptotics of @xmath462 . [ l : momgenas ] for each @xmath455 , @xmath490\ ] ] where there is a constant @xmath491 such that @xmath492 for all @xmath11 sufficiently large .
we would like to rewrite the last line of using exponentials of logarithms .
note that for all @xmath493 and all @xmath494 $ ] , @xmath495 thus @xmath496\ } \subset { \mathbb{c}}\setminus { \mathbb{r}}_-$ ] , so we can use the principal branch of the logarithm .
we thus have @xmath497 - 1\right)\right)- i\tilde { \mathfrak{u}}^{(n)}_n\frac{\theta}{\sqrt{n}}\rb \right]\ ] ] for all @xmath498 . for any fixed @xmath455
, @xmath499 is small for @xmath12 large enough ; we thus want to expand the logarithmic term near @xmath500 .
we want this approximation to be uniform in the @xmath501 s , however , so we need to be a bit careful . for @xmath493 and @xmath494 $ ] , implies that @xmath502 fix @xmath503 .
if @xmath504 and @xmath494 $ ] , then @xmath505 , and we can use the taylor expansion of the logarithm to see that @xmath506 where there is a constant @xmath339 such that @xmath507 for all @xmath504 and all @xmath494 $ ] .
recall next the standard fact that @xmath508 for all @xmath504 , where there is a @xmath341 such that @xmath509 for all @xmath504 . combining things together ,
we conclude that @xmath510 for all @xmath504 and @xmath494 $ ] , where there is a @xmath345 such that @xmath511 for all @xmath504 and @xmath494 $ ] .
collecting our calculations , we thus have that @xmath512 - 1\right)\right)- i\tilde { \mathfrak{u}}^{(n)}_n\frac{\theta}{\sqrt{n}}\rb = -\frac12 \sigma^2(\alpha,{\bar { u}^{(n)}})\theta^2 + \tilde { \mathcal{e}}_n(\theta)\ ] ] for all @xmath12 such that @xmath513 , where there is a @xmath514 such that @xmath515 for all @xmath481 and @xmath11 such that @xmath513 . the claimed result now easily follows .
we next prove the uniform bound on @xmath462 .
[ l : ubound ] there is a @xmath516 such that @xmath517\ ] ] for all @xmath498 and @xmath11 . for @xmath494 $ ] and @xmath518 , a calculation like shows that @xmath519 note that @xmath520 . secondly , note for any @xmath127 .
next check that @xmath521}\frac{\ln ( 1-x)}{x}<0 $ ] . to do so
, it suffices by continuity to check @xmath522 ; this can easily be done via lhpital s rule . ]
that there is an @xmath127 such that @xmath523 for all @xmath524 $ ] .
thus @xmath525 \le \exp\left[-2{\varkappa}{\varkappa}_- u(1-u ) \tilde \theta^2\right]\ ] ] for all @xmath493 .
consequently @xmath526\ ] ] for all @xmath498 and all @xmath11 .
the claimed result follows .
combining lemmas [ l : sopen ] and [ l : sigmalim ] , we know that @xmath527 for @xmath11 sufficiently large enough .
for such @xmath12 , @xmath528 where @xmath529d\theta \\ { \textsc{\tiny e}}_2(s , n ) & = \sqrt{\frac{\sigma^2(\alpha,{\bar { u}})}{2\pi}}\int_{|\theta|<n^{1/8}}{\mathcal{p}}_n\left(\frac{\theta}{\sqrt{n}}\right){\left\ { } \newcommand{\rb}{\right\}}\exp\left[-is\theta/\sqrt{n}\right]-1\rb d\theta \\ { \textsc{\tiny e}}_3(n ) & = \sqrt{\frac{\sigma^2(\alpha,{\bar { u}})}{2\pi}}\int_{|\theta|<n^{1/8}}{\left\ { } \newcommand{\rb}{\right\}}{\mathcal{p}}_n\left(\frac{\theta}{\sqrt{n}}\right)-\exp\left[-\frac12 \sigma^2(\alpha,{\bar { u}^{(n)}})\theta^2\right]\rb d\theta\\ & = \sqrt{\frac{\sigma^2(\alpha,{\bar { u}})}{2\pi}}\int_{|\theta|<n^{1/8}}\exp\left[-\frac12 \sigma^2(\alpha,{\bar { u}^{(n)}})\theta^2 \right]{\left\ { } \newcommand{\rb}{\right\}}\exp\left[\tilde { \mathcal{e}}_n(\theta)\right]-1\rb d\theta\\ { \textsc{\tiny e}}_4(n ) & = -\sqrt{\frac{\sigma^2(\alpha,{\bar { u}})}{2\pi}}\int_{|\theta|\ge n^{1/8}}\exp\left[-\frac12 \sigma^2(\alpha,{\bar { u}^{(n)}})\theta^2\right]d\theta\\ { \textsc{\tiny e}}_5(n ) & = \sqrt{\frac{\sigma^2(\alpha,{\bar { u}})}{2\pi}}\int_{\theta\in { \mathbb{r}}}\exp\left[-\frac12 \sigma^2(\alpha,{\bar { u}^{(n)}})\theta^2\right]d\theta-1 \\ & = \frac{\sqrt{\sigma^2(\alpha,{\bar { u}})}-\sqrt{\sigma^2(\alpha,{\bar { u}^{(n)}})}}{\sqrt{\sigma^2(\alpha,{\bar { u}^{(n ) } } ) } } \end{aligned}\ ] ] here
we have used the standard calculation that for all @xmath530 , @xmath531d\theta = \frac{1}{\sqrt{a}}\int_{\theta\in { \mathbb{r}}}\frac{\exp\left[-\frac12 a \theta^2\right]}{\sqrt{2\pi / a}}d\theta = \frac{1}{\sqrt{a}}\ ] ] ( namely , we use this calculation with @xmath532 ) . by lemma [ l : sigmalim ]
, we have that @xmath533 .
this directly implies that @xmath534 . similarly
to , we also have that for @xmath530 and @xmath11 @xmath535d\theta = \frac{2}{\sqrt{2\pi}}\int_{\theta = n^{1/8}}^\infty\exp\left[-\frac12 a\theta^2\right]d\theta \\
= \frac{2}{\sqrt{2\pi}}\exp\left[-\frac12 a n^{1/4}\right]\int_{\theta=0}^\infty\exp\left[-\frac12 a{\left\ { } \newcommand{\rb}{\right\}}\left(\theta+n^{1/8}\right)^2-n^{1/4}\rb \right]d\theta
\\ \le \frac{2}{\sqrt{2\pi}}\exp\left[-\frac12 a n^{1/4}\right]\int_{\theta=0}^\infty\exp\left[-\frac12 a\theta^2\right]d\theta = \frac{1}{\sqrt{a}}\exp\left[-\frac12 a n^{1/4}\right]{\left\ { } \newcommand{\rb}{\right\}}2\int_{\theta=0}^\infty\frac{\exp\left[-\frac12 a\theta^2\right]}{\sqrt{2\pi / a}}d\theta\rb \\ = \frac{1}{\sqrt{a}}\exp\left[-\frac12 a n^{1/4}\right ] .
\end{gathered}\ ] ] thus ( since @xmath536\right|\le 1 $ ] ) we have that @xmath537
\\ \left|{\textsc{\tiny e}}_4(n)\right|&\le \sqrt{\frac{\sigma^2(\alpha,{\bar { u}})}{\sigma^2(\alpha,{\bar { u}^{(n)}})}}\exp\left[-\frac12\sigma^2(\alpha,{\bar { u}^{(n)}})n^{1/4}\right ] .
\end{aligned}\ ] ] recalling , , and , we have that @xmath538 to finally bound @xmath539 , define @xmath540 which is fairly easily seen to be finite . again using
, we have that for @xmath11 sufficiently large @xmath541}{n^{1/8}}\sqrt{\frac{\sigma^2(\alpha,{\bar { u}})}{\sigma^2(\alpha,{\bar { u}^{(n)}})}}.\ ] ] combining things , the stated claim follows .
we have intentionally formulated our assumptions to reflect their usage . for a large @xmath12 , we can readily check in a given situation if @xmath542 furthermore , we can construct the measure @xmath112 of . for a finite but large @xmath12
, this would suggest that we use theorem [ t : main ] and to price the cdo .
our goal here is to take a slightly different tack and restructure our assumptions in the framework that the @xmath81 s are , in a sense , samples from an underlying distribution .
we would like to reframe our assumptions in terms of this underlying distribution .
our setup here is as follows .
we define @xmath82 as in , and we assume that holds .
[ ex : walk ] for example [ ex : simpleexample ] , we would have that @xmath543 and for example [ ex : merton ] , we would have that @xmath544 [ r : bho ] we also note that the relation between the @xmath81 s and @xmath545 can allow some complexities . for example , let @xmath546dt}{\int_{t\in [ 0,\infty)}\exp\left[-\frac{n(t-1)^2}{2}\right]dt}. \qquad a\in { \mathscr{b}}(i)\ ] ] for every @xmath12 and @xmath3 , @xmath81 is very nice .
however , it is fairly easy to see that @xmath547 , where the measure @xmath548 ( as an element of @xmath60 ) does not have a density with respect to lebesgue measure .
this suggests that in certain situations , there is value in stating regularity assumptions on the limiting measure @xmath545 , rather than on the approximating sequence of the @xmath81 s .
let s next define @xmath549{u}(d\rho ) \qquad t\in i\ ] ] by lemma [ l : distmeas ] , we know that @xmath550 is a well - defined cdf on @xmath1 ; informally , @xmath550 is the expected notional loss distribution ( see ) .
for example [ ex : simpleexample ] , we would have that @xmath551 + \frac23 \check \mu_b[0,t]\ ] ] and for example [ ex : merton ] , we would have that @xmath552\frac{\sigma^{{\varsigma}-1}e^{-\sigma/\sigma_\circ}}{\sigma_\circ^{\varsigma}\gamma({\varsigma})}d\sigma\ ] ] for each @xmath553 , define @xmath554 . by lemma [ l : meas ] , we know that @xmath555 is a measurable map from @xmath60 to @xmath86 $ ] .
let s then define @xmath556}$ ] as @xmath557\ ] ] for all @xmath558 .
let s now turn to our assumptions . if @xmath559 , then assumption [ a : limitexists ] holds and @xmath560 .
we first note that @xmath561 .
fix @xmath562 $ ] .
define @xmath563 \\|p_1-p_2|<\delta}}|\psi(p_1)-\psi(p_2)|.\qquad \delta>0\ ] ] since @xmath86 $ ] is compact , @xmath564 .
fix now @xmath565
. then ( using the notation of section [ s : proofs ] ) @xmath566}\psi(p){\bar { u}^{(n)}}(dp)-\int_{p\in [ 0,1]}\psi(p)(p_*{u})(dp)\right|=\left|\int_{\rho\in { { \mathscr{p}}(i)}}\psi(\rho[0,t)){u}^{(n)}(d\rho)-\int_{\rho\in { { \mathscr{p}}(i)}}\psi(\rho[0,t)){u}(d\rho)\right|\\ & \qquad \le \left|\int_{\rho\in { { \mathscr{p}}(i)}}{\left\ { } \newcommand{\rb}{\right\}}\psi(\rho[0,t))-\psi({\mathbf{i}}_{\psi^-_{t , m}}(\rho))\rb { u}^{(n)}(d\rho)\right|\\ & \qquad \qquad + \left|\int_{\rho\in { { \mathscr{p}}(i)}}\psi({\mathbf{i}}_{\psi^-_{t , m}}(\rho)){u}^{(n)}(d\rho)-\int_{\rho\in { { \mathscr{p}}(i)}}\psi({\mathbf{i}}_{\psi^-_{t , m}}(\rho)){u}(d\rho)\right| \\ & \qquad \qquad + \left|\int_{\rho\in { { \mathscr{p}}(i)}}{\left\ { } \newcommand{\rb}{\right\}}\psi({\mathbf{i}}_{\psi^-_{t , m}}(\rho))-\psi(\rho[0,t))\rb { u}(d\rho)\right| .
\end{aligned}\ ] ] by weak convergence , we have that @xmath567 for each @xmath565 .
by dominated convergence , we also have that @xmath568 thirdly , we calculate that for each @xmath569 @xmath570}{u}^{(n)}{\left\ { } \newcommand{\rb}{\right\}}\rho\in { { \mathscr{p}}(i ) } : \left|\rho[0,t)-{\mathbf{i}}_{\psi^-_{t , m}}(\rho)\right|\ge \delta\rb . \end{gathered}\ ] ] for every @xmath553 , @xmath571 , so by markov s inequality @xmath572 thus @xmath573 combine things together , take @xmath94 , them @xmath574 , and finally @xmath575 . if @xmath576 , then assumption [ a : ig ] holds .
we will use the equivalent characterization of assumption [ a : ig ] given in .
for each @xmath12 and @xmath577 in @xmath578 , we have that @xmath579 let @xmath94 to get that @xmath580 now let @xmath574 and use dominated convergence to see that @xmath581 this gives the desired claim .
we can also check assumption [ a : nondegen ] in our two favorite examples .
for example [ ex : simpleexample ] , we have that @xmath582 which is zero if @xmath583 and @xmath584 . for example [ ex : merton ] , we similarly have that @xmath585 we finally turn our attention to assumption [ a : notflat ] .
[ l : notflatlemma ] if @xmath586 then assumption [ a : notflat ] holds . for all @xmath420 and @xmath553 , @xmath587 $ ] so by lemma [
l : meas ] , we know that the map @xmath588 is a measurable map from @xmath60 to @xmath86 $ ] for each @xmath420 . secondly , for all @xmath589 , @xmath420 and @xmath11 , @xmath590
next , let @xmath591 be such that @xmath592 , @xmath593 is decreasing , @xmath594 if @xmath595 , and @xmath596 if @xmath597 . for each @xmath420 and @xmath565
, let @xmath598 be such that @xmath599 , @xmath600 if @xmath601 , and @xmath602 if @xmath603 .
we note that @xmath604 for all @xmath420 , @xmath565 , and @xmath553 , and that @xmath605 for all @xmath553 and @xmath420 .
fix @xmath420 , @xmath589 , and @xmath12 and @xmath577 in @xmath578 .
then @xmath606 take first @xmath94 .
we get that @xmath607 now let @xmath574 and then @xmath137 , and use dominated convergence in both calculations .
we get that @xmath608 now let @xmath138 to get the claim .
for example [ ex : simpleexample ] , we have that @xmath609 which is zero if either @xmath58 or @xmath59 is not flat at @xmath25 . for example [ ex : merton ] , we have that @xmath610
in this section we look more deeply into the variational problems which have appeared in our arguments .
most of this section is motivational ; the only results we need in the body of the paper are the regularity results of lemmas [ l : lambdacont ] , [ l : ficont ] , and [ l : sopen ] , and the proof of lemma [ l : increasing ] .
the remainder of the section is devoted to proving lemmas [ l : finalitmin ] and [ l : variational ] .
looking carefully at our arguments , we see that we could in fact _ define _ @xmath156 as in and proceed with the rest of our paper . nevertheless , we prove both lemma [ l : finalitmin ] and lemma [ l : variational ] so that we can have a fairly complete understanding of the calculations involved in identifying how the rare events are most likely to form . to begin our calculations , we first explore some regularity of the objects described in lemma [ l : finalitmin ] .
define @xmath611 } : { \bar { v}}\in { \mathcal{g}}_{\alpha'}\rb \\ { \mathcal{s}}^{\text{strict}}&{\overset{\text{def}}{=}}{\left\ { } \newcommand{\rb}{\right\}}(\alpha',{\bar { v}})\in ( 0,1)\times { { \mathscr{p}}[0,1 ] } : { \bar { v}}\in { \mathcal{g}}^{\text{strict}}_{\alpha'}\rb .
\end{aligned}\ ] ] also define @xmath612}\phi(p,\lambda){\bar { v}}(dp)\ ] ] for all @xmath151 $ ] and @xmath146}$ ]
. then we have [ l : lambdacont ] for each @xmath613 , the solution @xmath614 of exists and is unique . if @xmath615 , then @xmath161 .
thirdly , the map @xmath616 is continuous on @xmath617 as a map from @xmath618}$ ] to @xmath280 $ ] .
remark [ r : phiprops ] ensures that @xmath619 is strictly increasing on @xmath280 $ ] as long as @xmath620 .
fixing @xmath146}$ ] , the continuity of @xmath279 ( again using remark [ r : phiprops ] ) and dominated convergence imply that @xmath619 is continuous on @xmath280 $ ] . noting that @xmath621= \lim_{\lambda\to -\infty}{\mathbf{\phi}}(\lambda,{\bar { v}}),\ ] ] we can conclude that @xmath614 defined as in exists and is unique for @xmath622 .
we note that if @xmath623 , then @xmath624 , while if @xmath625 $ ] , then @xmath626 .
otherwise , @xmath161 .
let s next address continuity .
we begin with some general comments which we will at the end organize in several ways .
fix @xmath627 and @xmath628 in @xmath617 such that @xmath629 ( in the product topology ) .
assume also that @xmath151 $ ] is such that @xmath630 .
if @xmath269 , then @xmath631 let @xmath632 .
remark [ r : phiprops ] implies that @xmath633 . by weak convergence
, we have that @xmath634 .
combine all of these things to see that @xmath635 .
assume next that @xmath636 .
then there is a @xmath569 such that @xmath637<\alpha'-\delta$ ] , so by portmanteau s theorem , @xmath638\le { \bar { v}}[1-\delta,1]<\alpha'-\delta$ ] . since @xmath639 is increasing for each @xmath640 , we have that @xmath641}\phi\left(p,\lambda(\alpha'_n,{\bar { v}}_n)\right){\bar { v}}_n(dp ) \\
\le \phi\left(1-\delta,\lambda(\alpha'_n,{\bar { v}}_n)\right ) + { \bar { v}}_n[1-\delta,1 ] . \end{gathered}\ ] ] thus @xmath642 , so @xmath643 .
we similarly now assume that @xmath644 .
then there is a @xmath569 such that @xmath645<1-\alpha'-\delta$ ] , so by portmanteau s theorem , @xmath646\le { \bar { v}}[0,\delta]<1-\alpha'-\delta$ ] .
monotonicity of @xmath639 now implies that @xmath647}{\left\ { } \newcommand{\rb}{\right\}}1-\phi\left(p,\lambda(\alpha'_n,{\bar { v}}_n)\right)\rb{\bar { v}}_n(dp ) + \int_{p\in [ 0,\delta]}{\left\ { } \newcommand{\rb}{\right\}}1- \phi\left(p,\lambda(\alpha'_n,{\bar { v}}_n)\right)\rb { \bar { v}}_n(dp ) \\
\le { \left\ { } \newcommand{\rb}{\right\}}1-\phi\left(1-\delta,\lambda(\alpha'_n,{\bar { v}}_n)\right)\rb + { \bar { v}}_n[0,\delta ] . \end{gathered}\ ] ] thus @xmath648<1-\delta,\ ] ] so @xmath649 .
let s collect things together . if @xmath160 , then the previous two calculations imply that @xmath650 if @xmath344 is a cluster point of @xmath651 , then @xmath635 , so in fact @xmath652 . in other words , if @xmath160 , then @xmath653 .
next assume that @xmath654 ; then @xmath624 .
we know that @xmath649 . if @xmath655 is a cluster point of @xmath651 , then @xmath635 , which violates uniqueness of the definition of @xmath614 .
thus if @xmath654 , we must have that @xmath656 .
similarly , we next assume that @xmath657 .
then @xmath626 .
we at least know that @xmath658 .
if @xmath655 is a cluster point of @xmath651 , then again @xmath635 , again violating the uniqueness of the definition of @xmath614 .
thus if @xmath657 , we must have that @xmath659 . for each @xmath269 , we next define @xmath660 for all @xmath109 $ ] .
note that @xmath661 for @xmath662 and all @xmath269 .
[ r : hprops ] we have that @xmath663 for all @xmath276 and @xmath269 , and @xmath664 for all @xmath109 $ ] and @xmath269 .
thus @xmath665 for all @xmath109 $ ] and @xmath666 and @xmath667 in @xmath668 .
finally , remark [ r : phiprops ] implies that for @xmath269 and @xmath109 $ ] , @xmath669 where @xmath670 .
we now study the right - hand side of . to avoid confusion with @xmath156 of , define now @xmath671}{\mathbf{h}}(p,\lambda(\alpha',{\bar { v}})){\bar { v}}(dp)\ ] ] for all @xmath613 .
[ l : ficont ] we have that @xmath672 is continuous on @xmath673 . fix @xmath627 and @xmath628 in @xmath673 such that @xmath629 .
then @xmath674 .
we write that @xmath675}{\mathbf{h}}\left(p,\lambda(\alpha'_n,{\bar { v}}_n)\right){\bar { v}}_n(dp)-\int_{p\in [ 0,1]}{\mathbf{h}}\left(p,\lambda(\alpha',{\bar { v}})\right){\bar { v}}(dp)\right|\\ & \le \left|\int_{p\in [ 0,1]}{\left\ { } \newcommand{\rb}{\right\}}{\mathbf{h}}\left(p,\lambda(\alpha_n',{\bar { v}}_n)\right)-{\mathbf{h}}\left(p,\lambda(\alpha',{\bar { v}})\right)\rb { \bar { v}}_n(dp)\right|\\ & \qquad + \left|\int_{p\in [ 0,1]}{\mathbf{h}}\left(p,\lambda(\alpha',{\bar { v}})\right){\bar { v}}_n(dp)-\int_{p\in [ 0,1]}{\mathbf{h}}\left(p,\lambda(\alpha',{\bar { v}})\right){\bar { v}}(dp)\right|.\end{aligned}\ ] ] by remark [ r : hprops ] , we have that @xmath676}{\left\ { } \newcommand{\rb}{\right\}}{\mathbf{h}}\left(p,\lambda(\alpha_n',{\bar { v}}_n)\right)-{\mathbf{h}}\left(p,\lambda(\alpha',{\bar { v}})\right)\rb { \bar { v}}_n(dp)\right|\le \left|\lambda(\alpha_n',{\bar { v}}_n)+\lambda(\alpha',{\bar { v}})\right|\left|\lambda(\alpha_n',{\bar { v}}_n)-\lambda(\alpha',{\bar { v}})\right|,\ ] ] and by weak convergence that @xmath473}{\mathbf{h}}\left(p,\lambda(\alpha',{\bar { v}})\right){\bar { v}}_n(dp)=\int_{p\in [ 0,1]}{\mathbf{h}}\left(p,\lambda(\alpha',{\bar { v}})\right){\bar { v}}(dp).\ ] ] combining things together , we get the desired result
. we can now prove lemma [ l : increasing ] .
the following result will help us with the continuity claims .
[ l : sopen ] the set @xmath673 is open . furthermore , for each @xmath145 , @xmath166 is open .
fix @xmath615 and @xmath627 in @xmath618}$ ] such that @xmath629 in the product topology . by definition of @xmath673
, we have that there is a @xmath569 such that @xmath677 since @xmath678 and @xmath679 are closed subsets of @xmath86 $ ] , portmanteau s theorem implies that @xmath680 and @xmath681 .
thus for @xmath640 sufficiently large , @xmath682 .
hence @xmath673 is open .
fix next @xmath145 , @xmath160 , and @xmath468 in @xmath87}$ ] such that @xmath470 .
then @xmath615 , and @xmath683 .
since @xmath673 is open , we thus have that @xmath684 for all @xmath640 sufficiently large ; i.e. , @xmath685 for @xmath640 sufficiently large .
hence @xmath166 is indeed open .
we use lemma [ l : sopen ] to see that @xmath686 if @xmath11 is sufficiently large .
we use lemmas [ l : lambdacont ] and [ l : ficont ] to get the convergence claims of . by assumption
[ a : ig ] and [ a : nondegen ] , we get that there is an @xmath687 such that @xmath688}p{\bar { u}^{(n)}}(dp)<\alpha\ ] ] for all @xmath689 .
thus for @xmath689 , we have that ( use a calculation similar to ) @xmath690}p{\bar { u}^{(n)}}(dp)<1-\alpha+\alpha<1;\ ] ] thus for @xmath689 , @xmath691 , so in fact we have the following string of inequalities : @xmath692}p{\bar { u}^{(n)}}(dp)>{\bar { u}^{(n)}}\{1\}.\ ] ] thus for @xmath689 , @xmath693 .
lemma [ l : ficont ] thus ensures that @xmath261 is continuous on @xmath262 for @xmath694 . remark [ r : phiprops ] implies that @xmath153 is nondecreasing in its second argument , so @xmath695 must also be nondecreasing on @xmath262 . remark [ r : hprops ] ensures that @xmath696 is also nondecreasing in its second argument , so we can now conclude that @xmath261 is nondecreasing on @xmath262 . to finally understand the sign of @xmath253 , note that @xmath697}p{\bar { u}^{(n)}}(dp),{\bar { u}^{(n)}}\right),{\bar { u}^{(n)}}\right ) = \int_{p\in [ 0,1]}p{\bar { u}^{(n)}}(dp ) = { \mathbf{\phi}}(0,{\bar { u}^{(n)}});\ ] ] thus @xmath698}p{\bar { u}^{(n)}}(dp),{\bar { u}^{(n)}}\right)=0 $ ] . by , we know that @xmath699 for @xmath689 , so monotonicity implies that @xmath700}p{\bar { u}^{(n)}}(dp),{\bar { u}^{(n)}}\right)=0,\ ] ] and so @xmath701 . if @xmath702 , then @xmath703}\phi(p,\lambda(\alpha,{\bar { u}})){\bar { u}}(dp ) = \int_{p\in [ 0,1]}\phi(p,0){\bar { u}}(dp)\\ = \int_{p\in [ 0,1]}p{\bar { u}}(dp),\ ] ] which violates assumption [ a : ig ] .
thus @xmath704 , finishing the proof .
we next turn to the proof of lemma [ l : variational ] .
while lemma [ l : variational ] is not really needed in the paper , it does represent a key step in our chain of reasoning .
namely , the grtner - ellis theorem of large deviations tells us that the first step in studying rare events is to take the legendre - fenchel transform of a limiting logarithmic moment - generating function .
the background object of interest is the empirical measure , and the appropriate legendre - fenchel transform is given in .
the contraction principle tells us how to `` project '' a large deviations principle for @xmath217 onto @xmath705 ; that is .
this is the `` rigorous '' way to study the rare events leading to the losses in investment - grade tranches .
assumedly , they should lead to the intuitively - appealing rate function and agree with the fairly straightforward calculations of example [ ex : twobonds ] , both of which encapsulate the idea that there are many configurations leading to a loss , but we want the one which is least unlikely .
aside of intellectual curiosity , the value of a proof of lemma [ l : variational ] is that in the course of the calculations , a number of properties of extremals are identified ; these have direct implications for the rest of our calculations .
more exactly , they identify the measure change which we use in section [ s : measurechange ] . more generally ,
this measure change is closely related to importance sampling methods .
thus we believe that the extra effort needed to prove lemma [ l : variational ] is worthwhile . as a final comment before we begin , we note that @xmath706 an important part of the proof is the duality between entropy and exponential integrals . for any @xmath707 , @xmath708
also , for @xmath709 , let @xmath710 be the unique element of @xmath60 such that @xmath711 lemma [ l : distmeas ] ensures that the map @xmath712 is a measurable map from @xmath713 to @xmath60 .
let s first prove that @xmath714 fix @xmath715 such that @xmath716 .
fix also @xmath717 . for each @xmath553 , define @xmath718 as @xmath719 @xmath720 note that if @xmath721 is a sequence in @xmath60 converging ( in the weak topology on @xmath60 ) to @xmath553 , then for any @xmath593 and @xmath722 in @xmath247 @xmath723 thus @xmath724 is in @xmath725 .
we can now proceed .
we have that @xmath726 if @xmath727 is such that @xmath728 , then the supremum is @xmath729 . using this , we have that @xmath730 varying @xmath577 , we thus have that @xmath731 note that for any @xmath727 , @xmath732 we thus invoke lemma 7.1 from @xcite and see that @xmath733 let s next _ condition _ on the value of @xmath734 . since the map @xmath735 is a measurable map from @xmath60 to @xmath86 $ ] ( both of which are polish spaces ; see also lemma [ l : meas ] ) , there is a measurable map @xmath736 from @xmath86 $ ] to @xmath60 such that @xmath737}\check { u}_p(a)\psi(p){\bar { u}}(dp)\ ] ] for all @xmath738 and all @xmath739)$ ] .
fix now @xmath727 such that @xmath740 for each @xmath109 $ ] , define now @xmath741 then @xmath742;[0,1])$ ] .
clearly @xmath95 } \phi(p){\bar { u}}(dp ) = \int_{\rho\in { { \mathscr{p}}(i ) } } m\tilde ( \rho)[0,t){u}(d\rho)=\alpha'.\ ] ] convexity of @xmath743 in the first argument thus implies that @xmath744}{\left\ { } \newcommand{\rb}{\right\}}\int_{\rho\in { { \mathscr{p}}(i)}}\hbar(\tilde m(\rho)[0,t),p)\check { u}_p(d\rho)\rb { \bar { u}}(dp)\\ \ge \int_{p\in [ 0,1]}\hbar\left(\int_{\rho\in { { \mathscr{p}}(i)}}\hbar(\tilde m(\rho)[0,t),p)\check { u}_p(d\rho)\rb { \bar { u}}(dp ) = \int_{p\in [ 0,1]}\hbar(\phi(p),p){\bar { u}}(dp ) .
\end{gathered}\ ] ] this directly leads to let s now prove the reverse inequality ; i.e , that @xmath745 fix @xmath742;[0,1])$ ] such that @xmath746}\phi(p){\bar { u}}(dp)=\alpha'$ ] .
we can of course also assume that @xmath747}\hbar(\phi(p),p){\bar { u}}(dp)<\infty.\ ] ] for every @xmath553 , define @xmath748 if @xmath749 , and define @xmath750 if @xmath751 .
we first claim that @xmath752 for all @xmath553 . if @xmath749 , a direct calculation shows that this is in fact an equality . if @xmath751 , then @xmath753 .
thus @xmath95}\hbar(\phi(p),p){\bar { u}}(dp ) = \int_{\rho\in { { \mathscr{p}}(i)}}\hbar(\phi(\rho[0,t)),\rho[0,t)){\bar { u}}(dp ) \ge \int_{\rho\in { { \mathscr{p}}(i)}}h(m(\rho)|\rho){\bar { u}}(dp).\ ] ] fix next @xmath591 . by
, we thus have that @xmath472}\hbar(\phi(p),p){\bar { u}}(dp ) \ge \int_{\rho\in { { \mathscr{p}}(i)}}{\left\ { } \newcommand{\rb}{\right\}}\int_{t\in i}\psi(t)m(\rho)(dt ) - \ln \int_{t\in i}e^{\psi(t)}\rho(dt)\rb { u}(d\rho)\\ = \int_{t\in i}\psi(t)df_{{u}m^{-1}}(dt ) - \int_{\rho\in { { \mathscr{p}}(i)}}{\left\ { } \newcommand{\rb}{\right\}}\ln \int_{t\in i}e^{\psi(t)}\rho(dt)\rb { u}(d\rho ) . \end{gathered}\ ] ] note now that if @xmath749 , then @xmath754 . also , and
imply that if @xmath755 , then @xmath756 , and if @xmath757 , then @xmath758 .
thus @xmath759 thus @xmath760}\phi(p){\bar { u}}(dp)=\alpha'.\ ] ] thus @xmath95}\hbar(\phi(p),p){\bar { u}}(dp ) \ge
\sup_{\psi\in c_b(i)}{\left\ { } \newcommand{\rb}{\right\}}\int_{t\in i}\psi(t)df_{{u}m^{-1}}(dt ) - \int_{\rho\in { { \mathscr{p}}(i)}}{\left\ { } \newcommand{\rb}{\right\}}\ln \int_{t\in
i}e^{\psi(t)}\rho(dt)\rb { u}(d\rho)\rb\ ] ] and holds .
let s now turn to showing that the minimization problem is indeed solved by @xmath672 as stated in lemma [ l : finalitmin ] .
this will be a fairly involved proof .
again , this is not essential to the paper .
however , it is essential to understanding that does indeed give the optimal distribution of rare events leading to loss in the tranche ; i.e. , it explicitly solves .
we note before starting that for @xmath141 and @xmath142 in @xmath143 , @xmath761 observe that @xmath762 has singularities at @xmath763 and @xmath764 .
our first step is to solve when the singularities are more controlled .
fix now @xmath146}$ ] such that @xmath765 .
fix also @xmath145 .
our goal is lemma [ l : extremalscompactsupport ] ; to show that @xmath766 . along the way ,
corollary [ c : correctmin ] will require approximation of @xmath767 ; let @xmath768 be in @xmath143 such that @xmath769 . for @xmath421 , define @xmath770;[{\varepsilon},1-{\varepsilon } ] ) : \int_{p\in [ 0,1]}\phi(p){\bar { v}}(dp)=\alpha'_{\varepsilon}\rb \\ { \mathfrak{i}}_{\varepsilon}&{\overset{\text{def}}{=}}\inf{\left\ { } \newcommand{\rb}{\right\}}\int_{p\in [ 0,1]}\hbar(\phi(p),p){\bar { v}}(dp ) : \phi\in \mathcal{f}_{\varepsilon}\rb . \end{aligned}\ ] ] let @xmath771 be such that @xmath772 for all @xmath773 ( we use here the requirement that @xmath145 ) ; then for @xmath773 , we have that @xmath774 is in @xmath775 , so @xmath776 .
since @xmath746}\hbar(\alpha'_{\varepsilon},p){\bar { v}}(dp)<\infty$ ] ( the support of @xmath469 is a compact subset of @xmath143 , and @xmath777 is continuous on @xmath86\times ( 0,1)$ ] ) , we also know that @xmath778 .
then we have [ l : regularizedextremal ] fix @xmath773 .
the variational problem @xmath779 has a minimizer @xmath780 .
let @xmath781 be a sequence in @xmath775 such that @xmath746}\hbar(\phi^{({\varepsilon})}_n(p),p){\bar { v}}(dp)<{\mathfrak{i}}_{\varepsilon}+1/n$ ] .
clearly @xmath95}|\phi^{({\varepsilon})}_n(p)|^2{\bar { v}}(dp)\le 1,\ ] ] so @xmath782 is in the unit ball in @xmath783 $ ] .
thanks to alaoglu s theorem and the fact that @xmath783 $ ] is reflexive , we know that there is a subsequence @xmath784 and a @xmath785 $ ] such that @xmath786 weakly in @xmath783 $ ] . for any @xmath787 $ ] , @xmath788}\chi_a(p)\phi^{({\varepsilon})}_{n_k}(p){\bar { v}}(dp)-{\varepsilon}{\bar { v}}(a ) \ge 0 \\
\int_{p\in a}\{1-{\varepsilon}-\phi^{({\varepsilon})}(p)\ } { \bar { v}}(dp ) = ( 1-{\varepsilon}){\bar { v}}(a)-\lim_{k\to \infty}\int_{p\in [ 0,1]}\chi_a(p)\phi^{({\varepsilon})}_{n_k}(p){\bar { v}}(dp)\ge 0 \end{gathered}\ ] ] and @xmath95}\phi^{({\varepsilon})}(p){\bar { v}}(dp ) = \lim_{k\to \infty}\int_{p\in [ 0,1]}\phi^{({\varepsilon})}_{n_k}(p){\bar { v}}(dp)=\alpha'_{\varepsilon}.\ ] ] thus @xmath789 .
clearly @xmath790}\hbar(\phi^{({\varepsilon})}(p),p){\bar { v}}(dp ) \ge { \mathfrak{i}}_{\varepsilon}.\ ] ] since @xmath777 is convex in its first argument , we can also see that @xmath791}\hbar(\phi^{({\varepsilon})}_{n_k}(p),p){\bar { v}}(dp)\\ = \int_{p\in [ 0,1]}\hbar(\phi^{({\varepsilon})}(p),p){\bar { v}}(dp ) + \int_{p\in [ 0,1]}{\left\ { } \newcommand{\rb}{\right\}}\hbar(\phi^{({\varepsilon})}_{n_k}(p),p)-\hbar(\phi^{({\varepsilon})}(p),p)\rb { \bar { v}}(dp)\\ \ge \int_{p\in [ 0,1]}\hbar(\phi^{({\varepsilon})}(p),p){\bar { v}}(dp ) + \int_{p\in [ 0,1]}\frac{\partial \hbar}{\partial \beta_1}(\phi^{({\varepsilon})}(p),p){\left\ { } \newcommand{\rb}{\right\}}\phi^{({\varepsilon})}_{n_k}(p)-\phi^{({\varepsilon})}(p)\rb { \bar { v}}(dp ) .
\end{gathered}\ ] ] we next use the facts that @xmath780 takes values between @xmath792 and @xmath793 and that @xmath794 to ensure that @xmath795 is in @xmath783 $ ] . hence @xmath796}\frac{\partial \hbar}{\partial \beta_1}(\phi^{({\varepsilon})}(p),p){\left\ { } \newcommand{\rb}{\right\}}\phi^{({\varepsilon})}_{n_k}(p)-\phi^{({\varepsilon})}(p)\rb { \bar { v}}(dp)=0,\ ] ] and so @xmath797}\hbar(\phi^{({\varepsilon})}(p),p){\bar { v}}(dp).\ ] ] in combination with , this gives us the desired claim .
note here that the minimizer @xmath780 may not be unique ; in particular , we can change @xmath780 any way we want outside of the support of @xmath469 and we will still have a minimizer .
let s next study @xmath780 a bit more .
define the ( @xmath798$]-measurable ) sets @xmath799 : \phi^{({\varepsilon})}(p)={\varepsilon}\ } , \quad b^{\varepsilon}{\overset{\text{def}}{=}}\{p\in [ 0,1 ] : \phi^{({\varepsilon})}(p)\in ( { \varepsilon},1-{\varepsilon})\}\\ c^{\varepsilon}{\overset{\text{def}}{=}}\{p\in [ 0,1 ] : \phi^{({\varepsilon})}(p)=1-{\varepsilon}\}.\end{gathered}\ ] ] for convenience , let s also define @xmath800 : \phi^{({\varepsilon})}(p)\in [ \delta,1-\delta]\}\ ] ] for @xmath801 and note that @xmath802 as @xmath803 . also note that at the moment , we ca nt preclude that @xmath804 , @xmath805 , or @xmath806 are zero ( we will later , in lemma [ l : acsmall ] show that in fact @xmath807 is zero if @xmath792 is small enough ) .
[ l : lagrangianbasic ] fix @xmath773 .
there is a @xmath808 such that @xmath809 for @xmath469-a.e .
thus @xmath811 for @xmath469-a.e .
@xmath810 where @xmath153 is as in .
the result is of course trivially true if @xmath812 ; we thus assume that @xmath813 .
define the vector spaces @xmath814 : \eta\big|_{[0,1]\setminus b^{\varepsilon}}\equiv 0\rb \\ v_\delta & { \overset{\text{def}}{=}}{\left\ { } \newcommand{\rb}{\right\}}\eta\in b[0,1 ] : \eta\big|_{[0,1]\setminus b^{\varepsilon}_\delta}\equiv 0\rb .
\qquad \delta>{\varepsilon}\end{aligned}\ ] ] fix @xmath801 .
fix @xmath815 such that @xmath816}\eta(p){\bar { v}}(dp)=0.\ ] ] if @xmath817 is small enough , @xmath818 , so @xmath95}\hbar(\phi^{({\varepsilon})}(p)+\nu \eta(p),p){\bar { v}}(dp)\ge \int_{p\in [ 0,1]}\hbar(\phi^{({\varepsilon})}(p),p){\bar { v}}(dp).\ ] ] thus @xmath819}\frac{\partial \hbar}{\partial \beta_1}(\phi^{({\varepsilon})}(p),p)\eta(p){\bar { v}}(dp)=0.\ ] ] we next want to extend ths result to @xmath820 .
we first note that by continuity and the positivity assumption , @xmath821 .
thus there is a @xmath822 such that @xmath823 if @xmath824 .
fix now @xmath825 such that holds . for @xmath824 , define @xmath826 then @xmath827 and @xmath95}\eta_\delta(p){\bar { v}}(dp ) = \int_{p\in b^{\varepsilon}_\delta}\eta(p){\bar { v}}(dp)- c_\delta { \bar { v}}(b^{\varepsilon}_\delta)=0.\ ] ] hence @xmath95}\frac{\partial \hbar}{\partial \beta_1}(\phi^{({\varepsilon})}(p),p)\eta_\delta(p){\bar { v}}(dp)=0.\ ] ] note that @xmath828 } \le 2\|\eta\|_{b[0,1]}$ ] and that @xmath829 @xmath469-a.s . thus by dominated convergence , holds .
in fact , we have now proved that holds for all @xmath825 such that holds .
we finish the proof by arguments standard from the theory of lagrange multipliers .
we see that there is a @xmath830 such that @xmath95}{\left\ { } \newcommand{\rb}{\right\}}\frac{\partial \hbar}{\partial \beta_1}(\phi^{({\varepsilon})}(p),p)-\lambda^{\varepsilon}\rb \eta(p){\bar { v}}(dp)=0\ ] ] for all @xmath825 . from this an explicit computation
completes the proof .
let s now understand what happens at points where @xmath780 is either @xmath792 or @xmath793 . for convenience , define @xmath831 we have that @xmath832 and @xmath833 .
we use an argument by contradiction to show that @xmath832 .
assume that there is a sequence @xmath834 in @xmath835 such that @xmath836 for all @xmath640 and such that @xmath837 .
for all @xmath640 , @xmath838 and so @xmath839}\phi(p,\lambda_{{\varepsilon}_n}){\bar { v}}(dp ) \ge \varliminf_{n\to \infty}\{\alpha_{{\varepsilon}_n}-{\varepsilon}_n\ } \ge \inf_{{\varepsilon}\in ( 0,{\bar { \varepsilon}}_1)}\{\alpha_{\varepsilon}-{\varepsilon}\}>0.\ ] ] since @xmath840 , dominated convergence implies that @xmath473}\phi(p,\lambda_{{\varepsilon}_n}){\bar { v}}(dp ) = { \bar { v}}\{1\}=0,\ ] ] which is a contradiction . thus @xmath832 . similarly , to show that @xmath833 , assume that there is a sequence @xmath841 such that @xmath842 for all @xmath640 and such that @xmath843 .
then @xmath844 , so for all @xmath640 @xmath845 and so @xmath839}{\left\ { } \newcommand{\rb}{\right\}}1-\phi(p,\lambda_{{\varepsilon}_n})\rb{\bar { v}}(dp ) \ge
\varliminf_{n\to \infty}\{1-{\varepsilon}_n-\alpha_{{\varepsilon}_n}\ } \ge \inf_{{\varepsilon}\in ( 0,{\bar { \varepsilon}}_1)}\{1-\alpha_{\varepsilon}-{\varepsilon}\}>0.\ ] ] since here @xmath846 , we now have that @xmath473}{\left\ { } \newcommand{\rb}{\right\}}1-\phi(p,\lambda_{{\varepsilon}_n})\rb{\bar { v}}(dp ) = { \bar { v}}\{0\}=0.\ ] ] again we have a contradiction , implying that indeed @xmath833 .
we next disallow some degeneracies .
[ l : nosmallsets ] there is an @xmath847 such that @xmath848 and @xmath849 if @xmath850 .
we start with the fact that @xmath851}\phi^{({\varepsilon})}(p){\bar { v}}(dp ) = { \varepsilon}{\bar { v}}(a_{\varepsilon } ) + ( 1-{\varepsilon } ) { \bar { v}}(c_{\varepsilon } ) + \int_{p\in b_{\varepsilon}}\phi(p,\lambda_{\varepsilon}){\bar { v}}(dp).\ ] ] since @xmath852 , we have @xmath853 thus for @xmath773 , @xmath854 which gives us what we want .
let now @xmath855 be such that @xmath856 $ ] for all @xmath857 . for @xmath857 , we have that @xmath858 for @xmath469-a.e . @xmath859 and @xmath860 for @xmath469-a.e .
@xmath861 .
again fix @xmath801 .
fix also sets @xmath65 , @xmath66 , and @xmath862 in @xmath798 $ ] such that @xmath863 , @xmath864 , and @xmath865 . set @xmath866 then for @xmath867 , @xmath868 , and @xmath869 positive and sufficiently small , @xmath870 , so @xmath95}\hbar(\phi^{({\varepsilon})}(p)+\nu_1 \eta_1(p)+\nu_2\eta_2(p)+\nu_3\eta_3(p),p){\bar { v}}(dp)\ge \int_{p\in [ 0,1]}\hbar(\phi^{({\varepsilon})}(p),p){\bar { v}}(dp).\ ] ] differentiating with respect to @xmath867 , @xmath868 and @xmath869 , we conclude that @xmath472}\frac{\partial \hbar}{\partial \beta_1}(\phi^{({\varepsilon})}(p),p)\eta_1(p){\bar { v}}(dp)\ge 0 , \quad \int_{p\in [ 0,1]}\frac{\partial \hbar}{\partial \beta_1}(\phi^{({\varepsilon})}(p),p)\eta_2(p){\bar { v}}(dp)\ge 0\\ \int_{p\in [ 0,1]}\frac{\partial \hbar}{\partial \beta_1}(\phi^{({\varepsilon})}(p),p)\eta_3(p){\bar { v}}(dp)\ge 0.\end{gathered}\ ] ] in other words , @xmath871 letting @xmath872 , we see that these inequalities hold for any sets @xmath65 , @xmath66 , and @xmath862 in @xmath798 $ ] such that @xmath863 , @xmath873 , and @xmath865 . from the third equation of
, we see that @xmath874 is decreasing in its second argument .
thus for @xmath875 , we have that @xmath876 if @xmath857 .
fix now @xmath857 .
assume that @xmath877 .
by lemma [ l : nosmallsets ] , we have that either @xmath878 , or @xmath879 . in the first case
, we get from the first equation of that @xmath880 , and in the second case we get from the second equation of that @xmath881 .
similarly , we can next assume that @xmath879 .
by lemma [ l : nosmallsets ] , we have that either @xmath882 , or @xmath877 . in the first case , we get from the last equation of that @xmath883 , and in the second case we get from the second equation of that @xmath884 . finally , we have [ l : acsmall ] there is an @xmath885 such that @xmath886 for all @xmath857 .
fix @xmath857 such that @xmath887 .
some straightforward calculations show that if @xmath858 , then @xmath888 thus @xmath889\right ) \le { \bar { v}}\left[0,2{\varepsilon}e^{c_-}\right].\ ] ] similarly , if @xmath860 , then @xmath890 hence @xmath891\right ) \le { \bar { v}}\left[1-{\varepsilon}e^{c_+},1\right].\ ] ] since @xmath892 is a compact subset of @xmath143 , the claim now follows .
thus [ c : correctmin ] for @xmath893 , we have that @xmath894 and @xmath895 .
fix @xmath893 .
we have that @xmath896}\phi^{({\varepsilon})}(p){\bar { v}}(dp ) = \int_{p\in b_{\varepsilon}}\phi^{({\varepsilon})}(p){\bar { v}}(dp)=\int_{p\in b_{\varepsilon}}\phi(p,\lambda^{\varepsilon}){\bar { v}}(dp ) = \int_{p\in [ 0,1]}\phi(p,\lambda^{\varepsilon}){\bar { v}}(dp).\ ] ] by the uniqueness claim of lemma [ l : lambdacont ] , we thus have that @xmath897 .
similarly , @xmath898}\hbar(\phi^{({\varepsilon})}(p),p){\bar { v}}(dp ) = \int_{p\in b_{\varepsilon}}\hbar(\phi^{({\varepsilon})}(p),p){\bar { v}}(dp ) = \int_{p\in b_{\varepsilon}}\hbar(\phi(p,\lambda_{\varepsilon}),p){\bar { v}}(dp ) = { \mathfrak{i}}^*(\alpha_{\varepsilon}',{\bar { v}}).\ ] ] this implies the claimed statement .
we finally can show that @xmath766 agree ( under our current assumption that @xmath765 ) . in light of corollary
[ c : correctmin ] , this is informally tantamount to showing that @xmath899 .
here we also use the ability to approximate @xmath767 .
[ l : extremalscompactsupport ] we have that @xmath766 under the current assumption that @xmath765 .
clearly @xmath900 .
fix next @xmath569 and fix @xmath901 $ ] such that @xmath95}\phi(p){\bar { v}}(dp)=\alpha ' \qquad \text{and}\qquad \int_{p\in [ 0,1]}\hbar(\phi(p),p){\bar { v}}(dp)<{\mathfrak{i}}(\alpha',{\bar { v}})+\delta.\ ] ] for each @xmath421 , define @xmath902}\phi_{\varepsilon}(p){\bar { v}}(dp ) .
\end{aligned}\ ] ] note that @xmath903 .
thus , by dominated convergence @xmath904}\hbar(\phi_{\varepsilon}(p),p){\bar { v}}(dp)=\int_{p\in [ 0,1]}\hbar(\phi(p),p){\bar { v}}(dp ) \qquad \text{and}\qquad \lim_{{\varepsilon}\to 0}\alpha'_{\varepsilon}= \alpha'.\ ] ] by the first of these equalities , we see that there is an @xmath905 such that @xmath95}\hbar(\phi_{\varepsilon}(p),p){\bar { v}}(dp ) < { \mathfrak{i}}(\alpha',{\bar { v}})+2\delta\ ] ] for all @xmath906 .
thus for @xmath906 , @xmath907}\hbar(\phi_{\varepsilon}(p),p){\bar { v}}(dp ) \ge { \mathfrak{i}}_{\varepsilon}= { \mathfrak{i}}^*(\alpha'_{\varepsilon},{\bar { v}}).\ ] ] we have of course used here corollary [ c : correctmin ] to get the last equality , and we use to define the approximation sequence for @xmath767 .
take now @xmath908 and use the continuity result of lemma [ l : ficont ] ( note that @xmath628 and the @xmath909 s are all in @xmath673 ) .
then let @xmath910 and conclude that @xmath911 .
summarizing thus far our work since , we now know that @xmath766 if @xmath765 .
we now want to relax the restriction that @xmath765 .
[ l : extremalsnoboundary ] we have that @xmath766 for all @xmath145 and @xmath146}$ ] such that @xmath912 .
again , we clearly have that @xmath900 . to show the other direction , we must approximate .
as in the proof of lemma [ l : extremalscompactsupport ] , fix @xmath569 and @xmath901 $ ] such that @xmath95}\phi(p){\bar { v}}(dp)=\alpha ' \qquad \text{and}\qquad \int_{p\in [ 0,1]}\hbar(\phi(p),p){\bar { v}}(dp)<{\mathfrak{i}}(\alpha',{\bar { v}})+\delta.\ ] ] since @xmath913 , there is a @xmath914 such that @xmath915>0 $ ] for @xmath916 . for @xmath916 , define @xmath917)}{{\bar { v}}[{\varkappa},1-{\varkappa}]}. \qquad a\in { \mathscr{b}}[0,1]\ ] ] for @xmath916 , define @xmath918}\phi(p){\bar { v}}_{\varkappa}(dp ) = \frac{\int_{p\in ( 0,1)}\phi(p)\chi_{[{\varkappa},1-{\varkappa}]}(p){\bar { v}}(dp)}{{\bar { v}}[{\varkappa},1-{\varkappa}]}.\ ] ] then @xmath919 .
since @xmath920 , we have that @xmath921}\hbar(\phi(p),p){\bar { v}}_{\varkappa}(dp){\bar { v}}[{\varkappa},1-{\varkappa } ] \ge { \mathfrak{i}}(\alpha_{\varkappa},{\bar { v}}_{\varkappa}){\bar { v}}[{\varkappa},1-{\varkappa } ] = { \mathfrak{i}}^*(\alpha_{\varkappa},{\bar { v}}_{\varkappa}){\bar { v}}[{\varkappa},1-{\varkappa}]\ ] ] for all @xmath916 .
take now @xmath922 and use the continuity result of lemma [ l : ficont ] .
note that @xmath923 in the topology of @xmath87}$ ] ; as in the proof of lemma [ l : extremalscompactsupport ] , @xmath628 and the @xmath924 s are also all in @xmath673 .
we have that @xmath925 .
then let @xmath926 .
thirdly , we want to allow @xmath469 to assign nonzero measure to @xmath927 . before proceeding with this calculation ,
let s next simplify a bit .
namely , we remove from the admissible set of @xmath901 $ ] those for which @xmath746}\hbar(\phi(p),p){\bar { v}}(dp)$ ] is obviously infinite .
thus if @xmath928 , we can restrict the admissible @xmath929)$ ] to those with @xmath756 , and for such @xmath930 , we have that @xmath931 note that both of these equations also of course hold if @xmath932 .
similarly , if @xmath933 , we can restrict the admissible @xmath929)$ ] to those with @xmath758 , and for such @xmath930 , we have that @xmath934 again , both of these equations also hold if @xmath935 . combining our thoughts , we have that @xmath936 ) , \int_{p\in(0,1)}\phi(p){\bar { v}}(dp)=\alpha'-{\bar { v}}\{1\}\rb\ ] ] [ l : edgesupport ] we have that @xmath766 for all @xmath145 and @xmath937 .
assume first that @xmath160 .
then @xmath938 , and we define @xmath939}$ ] as @xmath940\ ] ] from and lemma [ l : extremalsnoboundary ] , we now have that @xmath941}\hbar(\phi(p),p){\bar { v}}_\circ(dp){\bar { v}}(0,1 ) : \phi\in \operatorname{\operatorname{hom}}([0,1 ] ) , \int_{p\in[0,1]}\phi(p){\bar { v}}_\circ(dp)=\frac{\alpha'-{\bar { v}}\{1\}}{{\bar { v}}(0,1)}\rb \\ = { \mathfrak{i}}\left(\frac{\alpha'-{\bar { v}}\{1\}}{{\bar { v}}(0,1)},{\bar { v}}_\circ\right ) { \bar { v}}(0,1 ) = { \mathfrak{i}}^*\left(\frac{\alpha'-{\bar { v}}\{1\}}{{\bar { v}}(0,1)},{\bar { v}}_\circ\right ) { \bar { v}}(0,1 ) ; \end{gathered}\ ] ] we used here the fact that since @xmath160 , @xmath942 note that @xmath95}\phi\left(p,\lambda\left(\frac{\alpha'-{\bar { v}}\{1\}}{{\bar { v}}(0,1)},{\bar { v}}_\circ\right)\right){\bar { v}}(dp ) = \int_{p\in [ 0,1]}\phi\left(p,\lambda\left(\frac{\alpha'-{\bar { v}}\{1\}}{{\bar { v}}(0,1)},{\bar { v}}_\circ\right)\right){\bar { v}}_\circ(dp){\bar { v}}(0,1 ) + { \bar { v}}\{1\}=\alpha'\ ] ] so in fact @xmath943 .
thus @xmath944 this proves the result when @xmath160 .
assume next that @xmath654 .
then @xmath945 ) , \int_{p\in(0,1)}\phi(p){\bar { v}}(dp)=0\rb
\\ = \int_{p\in ( 0,1)}\hbar(0,p){\bar { v}}(dp ) = \int_{p\in [ 0,1]}\hbar(\phi(p,-\infty),p){\bar { v}}(dp ) . \end{gathered}\ ] ] note that here @xmath624 . on the other hand , if @xmath657 , then @xmath946 , so @xmath945 ) , \int_{p\in(0,1)}\phi(p){\bar { v}}(dp)={\bar { v}}(0,1)\rb \\ = \int_{p\in ( 0,1)}\hbar(1,p){\bar { v}}(dp ) = \int_{p\in [ 0,1]}\hbar(\phi(p,\infty),p){\bar { v}}(dp ) . \end{gathered}\ ] ] here @xmath626 . by putting things together
, we can prove all of our extremal results .
the existence and uniqueness of @xmath947 is given in lemma [ l : lambdacont ] .
lemma [ l : edgesupport ] proves when @xmath157 .
if @xmath948 , then we note that @xmath95}p\mu^\dagger_{\alpha'}(dp)= \alpha'\ ] ] so @xmath949}\hbar(p , p)\mu^\dagger_{\alpha'}(dp ) = 0.\ ] ] next , let s look more closely at . if @xmath901 $ ] is such that @xmath950 then @xmath951 thus @xmath952 and @xmath953 , so in fact @xmath954 . in other words ,
if @xmath469 is not in @xmath955 , then the admissible set of @xmath930 s in is empty , implying that @xmath956 .
the continuity of @xmath947 and @xmath156 follows directly from lemmas [ l : lambdacont ] and [ l : ficont ] .
we here prove some of the really technical measurability results which we have used .
this is essentially for the sake of completeness .
we start with an obvious comment . for future reference , let s next define @xmath959 for all @xmath960 and @xmath565 .
then @xmath961 and @xmath962 are in @xmath247 , and @xmath963}\le \psi_{t , m}^+\ ] ] and pointwise on @xmath1 we have ( as @xmath574 ) @xmath964 and @xmath965}$ ] .
the value of these approximations , at least in the context of section [ s : limitexists ] is that convergence in the topology of @xmath216 directly allows us to pass to the limit only when integrating against an element of @xmath958 ( e.g. @xmath966 of remark [ r : continuity ] ) . to justify passing to the limit
when integrating against an element of @xmath967 , we must approximate .
[ l : distmeas ] fix @xmath972 .
the function @xmath973{v}(d\rho)\qquad t\in i\ ] ] is a well - defined cdf on @xmath1 i.e. , @xmath974,and @xmath975 is left - continuous and nondecreasing .
furthermore @xmath976 is the unique element of @xmath60 such that @xmath977 for all @xmath591 . finally , the map @xmath978 is a measurable map from @xmath216 to @xmath60 .
lemma [ l : meas ] immediately implies that the integral defining @xmath979 is well - defined .
it is fairly clear that @xmath979 is indeed a cumulative cdf on @xmath1 ( use dominated convergence to show right - continuity ) .
we define @xmath976 by setting @xmath980= f_{v}(t)$ ] ( by mapping @xmath1 to @xmath981 $ ] , it is sufficient by carathodory s extension theorem to see that this defines a measure on a semialgebra which generates @xmath982 ; see ( * ? ? ?
* section 12.2 ) ) .
standard approximation results ( viz
. , approximate @xmath593 by indicators ) then imply .
the right - hand side of uniquely defines @xmath975 .
finally , by remark [ r : continuity ] , we can easily see that if @xmath983 in @xmath216 , then for any @xmath591 , @xmath984 thus the map @xmath978 is continuous ( and thus measurable ) .
moody s public finance credit committee . the u.s .
municipal bond rating scale : mapping to the global rating scale and assigning global scale ratings to municipal obligations .
technical report , moody s investor services , 2007 .
huyn pham .
some applications and methods of large deviations in finance and insurance . in _
paris - princeton lectures on mathematical finance 2004 _ , volume 1919 of _ lecture notes in math .
_ , pages 191244 .
springer , berlin , 2007 . | we use the theory of large deviations to study the pricing of investment - grade tranches of synthetic cdo s . in this paper , we consider a heterogeneous pool of names . our main tool is a large - deviations analysis which allows us to precisely study the behavior of a large amount of idiosyncratic randomness .
our calculations allow a fairly general treatment of correlation . |
this study was supported in part by the global coe program through the nanoscience and quantum physics project of tokyo tech . and by a grant - in - aid for scientific research on priority areas `` physics of super - clean materials '' ( no . 20029009 and 17071009 ) and on innovative areas `` topological quantum phenomena '' ( no .
22103003 ) from mext of japan . 00
a. p. schnyder , s. ryu , a. furusaki , and a. w. w. ludwig : phys .
b * 78 * ( 2008 ) 195125 .
r. roy : arxiv:0803.2868v1 ( 2008 ) .
a. kitaev : aip conference proceedings , * 1134 * ( 2009 ) 22 .
x. l. qi , t. l. hughes , s. raghu , and s. c. zhang : phys .
* 102 * ( 2009 ) 187001 .
s. b. chung and s. c. zhang : phys .
* 103 * ( 2009 ) 235301 .
g. e. volovik : jetp lett .
* 90 * ( 2009 ) 587 .
g. e. volovik : jetp lett .
* 90 * ( 2009 ) 398 .
y. nagato , s. higashitani , and k. nagai : j. phys .
* 78 * ( 2009 ) 123603 .
y. tsutsumi , m. ichioka , and k. machida : arxiv:1010.3066 ( 2010 ) .
y. nagato , m. yamamoto , and k. nagai : j. low temp . phys .
* 110 * ( 1998 ) 1135 .
d. vollhardt and p. wlfle : the superfluid phases of helium 3 ( taylor and francis , london , 1990 ) .
w. p. halperin and e. varoquaux : helium three ( eds w. p. halperin and l. p. pitaevskii ) ( elsevier , amsterdam , 1990 ) .
s. murakawa , y. tamura , y. wada , m. wasai , m. saitoh , y. aoki , r. nomura , y. okuda , y. nagato , m. yamamoto , s. higashitani , and k. nagai : phys .
* 103 * ( 2009 ) 155301 .
y. nagato , m. yamamoto , s. higashitani , and k. nagai : j. low temp .
* 149 * ( 2007 ) 294 .
y. wada , s. murakawa , y. tamura , m. saitoh , y. aoki , r. nomura , and y. okuda : phys .
b * 78 * ( 2008 ) 214516 .
y. aoki , y. wada , m. saitoh , r. nomura , y. okuda , y. nagato , m. yamamoto , s. higashitani , and k. nagai : phys .
* 95 * ( 2005 ) 075301 .
m. saitoh , y. wada , y. aoki , s. murakawa , r. nomura , and y. okuda : phys .
b * 74 * ( 2006 ) 220505(r ) . | the superfluid @xmath0he b phase , one of the oldest unconventional fermionic condensates experimentally realized , is recently predicted to support majorana fermion surface states .
majorana fermion , which is characterized by the equivalence of particle and antiparticle , has a linear dispersion relation referred to as the majorana cone .
we measured the transverse acoustic impedance @xmath1 of the superfluid@xmath0he b phase changing its boundary condition and found a growth of peak in @xmath1 on a higher specularity wall .
our theoretical analysis indicates that the variation of @xmath1 is induced by the formation of the cone - like dispersion relation and thus confirms the important feature of the majorana fermion in the specular limit .
surface andreev bound states ( sabs ) of the superfluid @xmath0he b phase are receiving renewed attention as edge states " of a 3d time reversal invariant topological superfluid@xcite .
topological superfluids and superconductors are characterized by a non - trivial topological number in the gapped bulk state and a gapless edge state on their edges or surfaces .
sabs of the superfluid @xmath0he b phase can be regarded as majorana fermions as they satisfy the majorana condition , i.e. , a particle and its antiparticle are equivalent , because the degrees of freedom of the bound states are halved . when a surface is specular ,
linear dispersions are predicted for majorana fermions@xcite , which have been recently referred to as a majorana cone@xcite .
we measured the transverse acoustic impedance @xmath1 of the superfluid @xmath0he b phase changing its boundary condition up to practically specular scattering .
the observed variation of @xmath1 is well reproduced by theoretical analysis and is shown to be induced by the formation of the cone - like dispersion relation of sabs at higher specularities .
this provides an evidence of the existence of the majorana cone in superfluid @xmath2he b in the specular limit .
superfluid @xmath0he is a well established spin - triplet p - wave superfluid where the b phase is the realization of the balian - werthamer state , which breaks the relative spin - orbit symmetry@xcite .
the surface majorana states of the @xmath0he b phase should exhibit peculiar features such as a surface spin current and an anisotropic magnetic response@xcite . under the specular scattering boundary condition of the surface , sabs with a finite energy are not degenerate , and have a dispersion relation linear to the momentum , which takes the shape of a 2d cone for massless dirac fermions . in the spin - triplet p - wave pairing system ,
the annihilation operator of the negative energy state is equivalent to the creation operator of the positive energy state .
hence ,
only the positive energy cone , which is called a majorana cone , is physical@xcite . to date
, majorana surface fermion states have been discussed mostly on smooth surfaces ( specular scattering limit ) . however , intriguing problems are to elucidate how the majorana cone transforms as the roughness of the walls increases ( or the specularity @xmath3 decreases ) and how this transformation affects the experimental observations .
@xmath3 is a parameter introduced to specify the boundary condition for quasiparticle scattering .
@xmath4 corresponds to the specular scattering limit where quasiparticles conserve the parallel component of the momentum and only the perpendicular component is inverted after scattering off a wall .
in contrast , @xmath5 is the diffusive limit where quasiparticles are scattered in random directions . for a partially specular scattering condition @xmath6 ,
a fraction @xmath3 of the quasiparticles is specularly scattered , while the remaining fraction @xmath7 is scattered diffusely@xcite . in liquid
@xmath2he , @xmath3 can be controlled _ in situ _ by coating the wall with thin layers of @xmath8he@xcite .
@xmath8he atoms are selectively adsorbed onto the walls due to their larger mass and reduce the roughness of the wall .
figure 1a shows a theoretical calculation@xcite for the angle averaged surface density of states ( sdos ) @xmath9 for the b phase at various @xmath3 . in the diffusive limit where @xmath5 , a nearly flat sabs band with a very sharp edge at @xmath10 appears below the bulk superfluid gap @xmath11 .
the existence of @xmath10 was theoretically predicted many years ago@xcite . as @xmath3 increases , the edge @xmath10 increases and the low - energy spectral weight shifts toward the band edge forming a peak structure just below @xmath10 . for the specular limit where @xmath4 , sdos increases linearly with energy at low energy : @xmath12 .
this linear energy dependence reflects the linear dispersion of sabs or the majorana cone . as is shown in fig .
1b , the density of states at zero energy @xmath13 decreases rapidly as @xmath3 increases .
we measured the temperature dependence of the transverse acoustic impedance @xmath14 of the superfluid @xmath0he b phase at various @xmath3 values using ac - cut quartz transducers immersed in liquid @xmath0he . in this case
, the wall corresponds to the surface of the transducer .
@xmath15 is a complex quantity defined as the ratio of the shear stress @xmath16 of the liquid at the wall to the oscillation velocity @xmath17 @xmath18 @xmath1 can be obtained from the change in the resonance angular frequency @xmath19 and the quality factor @xmath20 of the quartz transducers , @xmath21 here @xmath22 is the measured angular frequency , @xmath23 is the acoustic impedance of the quartz transducer and @xmath24 is the harmonics number@xcite .
@xmath3 is determined independently by measuring @xmath14 in a normal fluid with @xmath25he coating .
the temperature dependence of @xmath1 is largest for pure @xmath0he without coating , where the boundary condition is supposed to be at the diffusive limit @xmath5 . the temperature dependence becomes smaller in the sample with @xmath8he coating due to the reduced transverse momentum exchange and @xmath26 at @xmath4 .
@xmath3 can be obtained by fitting @xmath14 with the fermi liquid theory@xcite .
details of the experimental procedure have already been reported@xcite .
we have demonstrated in earlier publications that @xmath1 is a unique surface probe for sabs in superfluid @xmath0he and have provided the spectroscopic details of sdos@xcite .
figure 2 presents the measured @xmath14 as a function of the normalized energy @xmath27 .
@xmath28 is the superfluid gap energy at @xmath29 obtained from the weak - coupling - plus model@xcite ; thus , a higher ( lower ) energy corresponds to a higher ( lower ) temperature .
@xmath30 is the normal state value of @xmath1 just above the transition temperature @xmath31 . to view data at various @xmath3 on the same scale ,
the vertical axis is the difference from the normal state value @xmath32 scaled by @xmath33 . in a pure @xmath2he system without @xmath25he coating ,
the boundary condition is supposed to be fully diffusive , i.e. , @xmath5 . in this system , we found a kink in the real part @xmath34 of the impedance and a peak in the imaginary part @xmath35 at a higher frequency , which are indicated by the downward arrows in fig .
2@xcite . theoretical analysis showed that these singularities are due to pair excitations of a sabs quasiparticle and a propagating bogolubov quasiparticle@xcite with frequency @xmath36 .
in fact , the kink structure in the real part @xmath34 is well reproduced by a joint density of states @xmath37 where @xmath9 is the sdos shown in fig .
1 . in the system with @xmath25he coating ,
@xmath10 increases in accordance with theory@xcite . in partially specular cases ,
new peaks grow in the low - energy region as @xmath3 increases@xcite , which are indicated by the upward arrows in fig .
2 . at the highest specularity @xmath38 we have achieved to date ,
the peaks become very distinctive .
the theoretical calculation reproduces the growth of the low - energy peak at higher @xmath3 .
figure 3 shows the theoretical results of the frequency dependence of @xmath1 for various values of @xmath3 .
the calculation was performed in the same manner as refs . and .
let us consider the origin of the new peak in partially specular cases .
we show in fig . 4a - e the angle resolved sdos at various @xmath3 .
the spectra are plotted as functions of the parallel component of the momentum @xmath39 where @xmath40 is the incident angle of the quasiparticle measured from the surface normal and @xmath41 is the fermi momentum .
the circles in figs .
4a - e indicate the peak positions in the spectra , and are plotted altogether in fig .
4f .
although the peak position is insignificant due to the very broad spectrum at @xmath5 , we plotted the maximum positions at every @xmath3 . at @xmath5 ( fig .
4a ) , states below @xmath10 appear for all incident angles@xcite and a cone - like structure is not recognized in the broad spectrum .
however , even at a relatively small specularity @xmath42 ( fig .
4b ) , the zero - energy weight is significantly suppressed , except around the normal incidence @xmath43 .
a cone - like structure is formed in this case , but the peak positions are still below the linear dependence , as found in fig .
4f . at @xmath44 ( fig .
4c ) , which is within the range of the experiment , a rather sharp cone is recognized and the peak positions follow a nearly linear dependence @xmath45 with @xmath46 cm / s as in the specular limit @xmath4 ( fig .
4f ) .
thus , the majorana cone like structure is realized at higher specularities .
it was shown that @xmath1 in low - energy region is dominated by the scattering of the thermally excited quasiparticles@xcite .
the low - energy peak in @xmath1 begins to grow at @xmath47 ( dot - dashed lines in fig .
3 ) as @xmath48 increases . here
, @xmath49 corresponds to the energy difference between the band edge and the propagating bogoliubov states with @xmath50 .
the growth in @xmath1 is a clear sign of the inelastic scattering of quasiparticles from the thermally occupied sabs around the sdos maximum to the propagating bogoliubov states .
@xmath1 falls around @xmath51 , since the low - energy sdos is suppressed by the formation of the majorana cone .
this is in contrast to the diffusive limit ( @xmath5 in fig .
3 ) . since the sdos in the diffusive limit remains finite down to zero energy , no double peak structure is found .
the growth of the low - energy peak is a distinct evidence of the formation of the surface majorana cone . in summary , the growth of low - energy peaks in the transverse acoustic impedance is observed in a topological superfluid , the superfluid @xmath0he - b phase .
this growth is reproduced theoretically by a self - consistent calculation .
the dispersion of sabs is calculated for various boundary conditions and a cone - like structure is clearly observed even on a partially specular wall .
the experimentally observed growth of the peaks is due to the formation of a surface majorana cone on a higher specularity wall , and is a strong indicator of the predicted majorana cone in the specular limit .
transverse acoustic impedance measurement is a promising technique for studying surface states and will provide more direct proof of the majorana nature as the anisotropic magnetic response of the majorana cone@xcite . |
the nearby common proper - motion pair 16 cyg a and b ( hd186408 and hd186427 , respectively ) have been of particular interest to stellar astronomers for a number of reasons . in their physical characteristics
, they are both very similar to the sun and as such have been labeled `` solar twins '' or `` solar analogs '' ( friel et al . 1993 , hereafter f93 ) . like any human twins , however
, these two stars clearly can not be exactly similar to the sun or even to each other , and it is their differences that have generated the most recent interest .
most notably , radial - velocity studies have revealed that 16 cyg b harbors a planetary - mass companion , while 16 cyg a apparently does not ( cochran et al .
this remarkable observation has itself been invoked by some to explain other unexpected chemical differences between these two stars ( cochran et al .
1997 ; king 1997 ; gonzalez 1998 , hereafter g98 ) their li abundances differ by a factor of five of more ( deliyannis et al . 2000 , hereafter d00 ) , while studies of the fe abundance of this pair have consistently suggested that 16 cyg a may be slightly more metal - rich ( @xmath20.05 dex ) than its companion ( g98 ) .
how are these differences achieved in such otherwise similar stars , which share a presumably common primordial environment ?
to what level can dynamic interactions with a planetary system affect abundances in a stellar photosphere ?
unfortunately , with the exception of lithium , the small differences in line depths between these two stars combined with the relatively large uncertainties in abundance determinations to date have prevented any conclusive statements to be made about deviations between 16 cyg a and b with regard to any other element , including iron ( g98 ; d00 ) ; within the uncertainties , the measured chemical differences are all effectively zero .
the primary goal of this work is to utilize a new differential line abundance analysis method to better constrain the relative [ fe / h ] values between 16 cyg a and b. such a method , as shown by langer et al .
( 1998 ) , eliminates non - trivial uncertainties in oscillator strengths , @xmath3 values .
further , we utilize a more extensive fe linelist covering a wider range in lower excitation potential ( @xmath4 ) than previous published studies . in 2 , we describe our observations and the data reduction , in 3 we describe our analysis methods and present results , and we discuss their significance in 4 .
our intentions from the start have been to perform a very careful spectroscopic analysis of 16 cyg a and b , optimized for differential analysis .
we determined at the outset to make every effort to minimize any systematic differences which might arise due to variations in our observations , reductions , and subsequent analyses of each star . with this goal in mind
, we obtained optical spectra of 16 cyg a and b within 30 minutes of each other on december 22 , 1999 , with the mcdonald observatory 2.7 m telescope .
we employed the coud echelle spectrograph ( tull et al .
1995 ) and a 2048x2048 tektronix ccd , and made no changes to the instrument between the two observations .
we achieved a resolving power of 58,000 ( as measured on a th - ar lamp spectrum ) with a signal - to - noise ratio ( s / n ) of @xmath2400 per pixel at 6700 .
spectra of a hot star were also obtained within one hour and at a similar airmass in order to compensate for telluric features in the spectra of 16 cyg a and b. data for 16 cyg a and b were reduced in as nearly an identical manner as possible , in both cases following the general method described in gonzalez ( 1997 ) .
each image was processed with the same bias and flat fields , and corrections for scattered light were made with the same fitting functions in both spectra .
in addition , continua of individual orders were normalized with exactly the same functions in both stars .
these steps were undertaken to minimize possible systematic differences between line depths which would ultimately impact our differential analysis .
our efforts yielded continuous one - dimensional spectra in the blue up to 5500 , with gaps thereafter up to roughly 10,000 .
to eliminate potential bias in measuring the equivalent widths ( ews ) of spectral features , one of us ( gg ) renamed each spectrum , and the other ( cl ) performed the measurements and analysis with no prior knowledge of which spectrum was associated with which star .
we employed a list of high - quality atomic lines used in our previous studies of planet bearing stars , and measured the ews of 60 fe i and 8 fe ii lines spanning broad ranges in @xmath5 from 0.09 to 6.22 ev and in ew from 4.7 to 119.8 m . in every case
the ew was measured for a line first in one star , then immediately afterwards in the second star before moving on to the next line .
this procedure was followed to minimize any systematic differences in ew measurements between the two stars .
telluric contamination in some fe lines was addressed by dividing the science object spectra by that of a hot star , and where performed , this division , as well as considerations of blending by nearby features , was treated identically for each line in each star .
our linelist and measured ew values are presented in table 1 .
we employed the measured ews from 3.1 and associated @xmath3 values taken from gonzalez et al .
( 2000 ) to derive a set of basic stellar parameters ( @xmath6 , @xmath7 , @xmath8 , and [ fe / h ] ) and associated uncertainties for 16 cyg a and b , using the methods described extensively in g98 and gonzalez & vanture ( 1998 ) .
briefly , all four of these parameters were systematically iterated with a recent version of the lte abundance code moog ( sneden , 1973 ) using kurucz model atmospheres ( 1993 ) until the mean fe i and fe ii abundances were equal and correlations of the individual fe i line abundances with both @xmath4 and the logarithm of reduced equivalent width ( rew ) were zero . following the method described in gonzalez & vanture ( 1998 ) , uncertainties in @xmath8
were determined from the standard deviation in the slope of a least - squares fit to the fe i vs. @xmath9 data .
this resulting uncertainty in @xmath10 was summed in quadrature with the standard deviation in the slope of a least - squares fit to the fe i vs. @xmath4 data to estimate the uncertainty in @xmath6 .
we combined theses uncertainties in @xmath4 and @xmath11 with the observed scatter in modeled fe i line abundances to calculate our final uncertainty in [ fe / h ] . level . ] our results using these standard methods are presented in table 2 , and we confirm previous studies : 16 cyg a and b are very similar to each other , as well as to the sun .
the pair are each slightly metal - enhanced relative to solar , however , with 16 cyg a presenting an 0.03 dex overabundance of fe relative to 16 cyg b , although this difference is not significant given our formal estimates of uncertainty . to reduce the uncertainties in our estimates from 3.2 , we reanalyzed our ew data from these two stars employing a technique similar to that described and effectively utilized by langer et al .
( 1998 ) in searching for small metallicity variations amongst red giants in m92 .
essentially , this method mirrors the standard " method discussed above , but instead of determining chemical abundances for each star by averaging the results determined from individual lines and subsequently comparing these averages between stars , it determines differential chemical abundances between stars by taking the differences in abundances calculated for each line _ individually_. this differential strategy effectively eliminates uncertainties in @xmath3 values , and allows one to take advantage of the increased precision in @xmath0[fe / h ] to further constrain differential values of the other stellar parameters .
we used our values for @xmath6 , @xmath7 , and @xmath8 from 3.2 to calculate [ fe / h ] for each line in each star , then determined differential fe i abundances ( @xmath0[fe / h ] ) for each line individually . '' values are of the form @xmath12 . ]
if the @xmath0@xmath6 and @xmath13 values from 3.2 were both correct , then plots of @xmath0[fe / h ] vs. both @xmath4 and @xmath9 would be uncorrelated .
our initial such plots using these `` @xmath0 '' values as inputs , however , showed slopes of @xmath14 dex / ev for @xmath0[fe / h ] vs. @xmath4 and @xmath15 dex/@xmath9 for @xmath0[fe / h ] vs. @xmath9 ( for comparison , these values in our standard analysis of 16 cyg a in 3.2 were @xmath16 and @xmath17 , respectively ) .
while these results are only slightly larger than their standard uncertainties , we note that they are both more precise and appreciably larger in magnitude than the final `` zero - slope '' conditions which resulted from our standard analysis .
such correlations indicate that the values of @xmath0 @xmath6 and @xmath13 calculated from the results in table 2 are off by some amount , and so we iterated @xmath0@xmath6 and @xmath18 by changing @xmath19 and @xmath20 until these correlations were forced to zero .
figure 1 presents a visualization of the effect of changing @xmath0@xmath6 and @xmath21 on the @xmath4 and @xmath9 slopes . for @xmath22 and 4.26 for 16 cyg a and b , respectively
, we find our best solution at @xmath0@xmath23k and @xmath24 km / s , where the stated uncertainties were calculated in precisely the same manner as described in 3.2 .
we note that these values are in very good agreement with our earlier , though less precise , results generated via the `` standard '' method .
figures 2 ( a ) and ( b ) show our final plots of @xmath0[fe / h ] vs. both @xmath4 and @xmath9 for our optimal values of @xmath0@xmath6 and @xmath13 ; immediately apparent is the fact that for the vast majority of lines analyzed , @xmath0[fe / h ] is positive .
we find a mean value of @xmath0[fe / h ] and its associated uncertainty of @xmath25 dex , where the stated uncertainty in @xmath0[fe / h ] was calculated via the method described in 3.2 , utilizing the uncertainties in @xmath0@xmath6 , @xmath13 , and the observed scatter of @xmath26 dex in individual line @xmath0[fe / h ] values .
uncertainty in @xmath0@xmath7 produced an effect of @xmath27 dex in @xmath0[fe / h ] , and was not considered in the stated error of the mean value of @xmath0[fe / h ] .
this uncertainty is in fact dominated by uncertainty in @xmath0@xmath6 , with a @xmath28k change effecting a @xmath29 dex change in @xmath0[fe / h ] .
we note that the atmospheric parameters for 16 cyg a and b calculated using the `` standard '' method stand in good agreement with previous published studies a heartening fact in light of discrepancies in ew measurements as well as variations in methodology among the various authors .
we believe our absolute values are more reliable than those of other recent studies due to our use of more fe i lines spanning a larger range in @xmath4 and ew ( d00 , for example , only employed 20 fe i lines with the smallest @xmath4 and ew values being 2.18 ev and 24 m , respectively ) .
furthermore , using our method of differential analysis , we are able to fine - tune the observed values of @xmath0@xmath30 and @xmath13 , and as a result have shown for the first time a statistically significant difference in [ fe / h ] between 16 cyg a and b. in figure 3 we compare our resulting @xmath0[fe / h ] values with those of previous studies .
the magnitude of this difference remains quite small , though , and while we have strived to eliminate possible systematic differences in our observations and analyses of these two stars , we must consider other potential mechanisms which might mimic true differences in chemical abundance before we can conclude that 16 cyg a is in fact enhanced in fe relative to 16 cyg b. most notably , measured values of [ fe / h ] are known to vary with chromospheric activity levels . a recent study of ca ii fluxes in planet - bearing stars by henry et al .
( 2000 ) , however , shows low values of @xmath31 for 16 cyg b , indicating little variability and surface activity .
our own spectra confirm the low levels of ca ii h and k emission , and show no signs of significant differences between 16 cyg a and b in this regard .
furthermore , _ hipparcos _
data sets the photometric variability of each of these stars at the 0.0007 magnitude level .
we therefore are confident that differences in chromospheric activity do not play a significant role in accounting for the measured values of @xmath0[fe / h ] .
in addition , we examined possible systematic errors which might be introduced due to assumptions in our atmospheric modeling , such as the possibility that a difference in the temperature minimum between the two stars might yield differing values of [ fe / h ] for the same @xmath6 ( wallerstein , 1972 ) . to test against this , we compared the central absorption in the mg i triplet near 5170 , and found no evidence supporting differences in temperature minima between 16 cyg a and b. we further argue that given the common ages of @xmath32 gyr ( g98 ) and similar values of @xmath6 , @xmath7 , and luminosity of these two stars confirmed independently from photometric analyses ( c.f .
f93 ) it is unlikely that any systematic errors in this strictly differential study might be due to more subtle model assumptions , such as that of lte . we therefore conclude that our measured value of @xmath0[fe / h ] is reliable to the level of our stated uncertainty , and that the photosphere of 16 cyg a _ is _ significantly more iron - rich than that of 16 cyg b. an obvious explanation for the observed value of @xmath0[fe / h ] is that it represents a primordial difference in the chemical composition of 16 cyg a and b. little high - resolution work exists on abundance differences between members of multiple star systems , however , and models of the formation of stellar systems are currently not sophisticated enough to address deviations in chemical abundance at the level of precision that we report here .
we intend to actively pursue the former shortcoming by employing our differential method on a wide sample of stellar systems ; however , until such primordial fluctuations are at least empirically , if not theoretically constrained , there is little more that can be said for or against this hypothesis .
if , on the other hand , the observed difference in [ fe / h ] is not entirely primordial , then it must be set by variations in the evolutionary history of these two stars .
a potential clue to the responsible mechanisms may lie in the aforementioned fact that the li abundances of 16 cyg a and b differ by a factor of at least five ( d00 ) . while ryan ( 2000 ) has noted that planet - bearing stars do not as a whole appear to differ from the field population in their values of [ li / h ] a finding supported by the larger sample discussed in gonzalez et al .
( 2000 ) it remains a challenge for standard theories of stellar atmospheres and evolution to explain the large difference in [ li / h ] between 16 cyg a and b given the considerable similarity of their physical characteristics and presumably shared environmental history . in response , various authors have discussed the possibility that the chemical composition of stellar photospheres can be affected by the presence of planetary companions .
cochran et al . ( 1997 ) and king ( 1997 ) proposed that dynamical interactions between a rotating star and its protoplanetary disk might significantly alter the angular momentum evolution of the star , and hence the rate at which the star depletes li .
this model is well supported by the recent investigations of li and be abundances reported by d00 , wherein these authors conclude that models of slow rotational mixing can explain the relative abundances of li and be in 16 cyg a and other anomolously li - rich f and g stars .
such a mechanism can not explain the more subtle variation in fe which we report here , however .
alternately , g98 proposed that the increased lithium content of 16 cyg a s photosphere may be the result of that star having consumed planetary material in its outer convection zone .
such self - pollution by materials of either chondritic or gas - giant composition would indeed produce an increase in the li abundance , while only slightly affecting the fe and be abundances ( g98 , d00 ) .
if our result indicating that 16 cyg a is slightly enhanced in fe relative to 16 cyg b is not a primordial effect , then it lends tentative support to this scenario . from the estimates presented by g98
, we calculate that the observed difference in [ fe / h ] between 16 cyg a and b can be explained by the accretion of 2.5 m@xmath33 of chondritic or 0.3 m@xmath34 of gas giant material .
the amount of li enhancement due to accretion of these bodies is difficult to determine , given the marked non - linear time , @xmath6 , and dynamical dependence of li depletion , along with the unkown timing of the putative accretion event(s ) .
nevertheless , it remains plausible that the observed difference in li between 16 cyg a and b is explainable by accretion at some intermediate age .
the authors are grateful to george wallerstein for helpful discussions and support through the kenilworth fund of the new york community trust , and to the anonymous referee whose helpful comments greatly increased the clarity of our manuscript .
this work has utilized the simbad database at cds , strasbourg , france , in addition to the abstract services of ads .
alexander , j. b. , 1967 , , 137 , 41 cochran , w. d. , hatzes , a. p. , butler , r. p. , & marcy , g. w. 1997 , , 483 , 457 deliyannis , c. p. , cunha , k. , king , j. r. , & boesgaard , a. m. 2000 , , 119 , 2437 ( d00 ) friel , e. , cayrel de strobel , g. , chmielewski , y. , spite , m. , lebre , a. , & bentolila , c. 1993 , , 274 , 825 ( f93 ) gonzalez , g. 1997 , , 285 , 403 gonzalez , g. & vanture , a. d. 1998 , , 339 , l29 gonzalez , g. 1998 , , 334 , 221 ( g98 ) gonzalez , g. , & laws , c. 2000 , , 119 , 390 gonzalez , g. , laws , c. , tyagi , s. , & reddy , b. e. 2001 , , 121 , 432 henry , g. w. , baliunas , s. l. , donahue , r. a. , fekel , f. c. & soon , w. 2000 , , 531 , 415 king , j. r. , deliyannis , c. p. , hiltgen , d. d. , stephens , a. , cunha , k. , & boesgaard , a. m. 1997 , , 113 , 1871 kurucz , r. l. 1993 , atlas9 stellar atmosphere programs and 2km / s grid cdrom vol . 13 , smithsonian astrophysical observatory langer , g. e. , fischer , d. , sneden , c. , & bolte , m. 1998 , , 115 , 685 perrin , m .-
& spite , m. 1981 , , 94 , 207 ryan , s. 2000 , , 316l , 35 sneden , c. 1973 , phd thesis , university of texas tull , r. g. , macqueen , p. j. , sneden c. , & lambert , d. l. 1995 , , 107 , 251 wallerstein , g. w. 1962 , , 6 , 407 wallerstein , g. w. 1972 , , 16 , 153 fe i & 5044.22 & 2.85 & -2.040 & 77.2 & 77.6fe i & 5247.06 & 0.09 & -4.930 & 70.2 & 70.7fe i & 5322.05 & 2.28 & -2.860 & 65.8 & 65.8fe i & 5806.73 & 4.61 & -0.900 & 60.2 & 59.2fe i & 5852.23 & 4.55 & -1.180 & 44.3 & 44.5fe i & 5853.16 & 1.48 & -5.280 & 9.4 & 9.6fe i & 5855.09 & 4.61 & -1.520 & 25.2 & 25.5fe i & 5956.71 & 0.86 & -4.550 & 55.1 & 56.5fe i & 6024.07 & 4.55 & -0.120 & 105.1 & 109.9fe i & 6027.06 & 4.07 & -1.090 & 69.3
& 68.0fe i & 6034.04 & 4.31 & -2.260 & 12.9 & 13.0fe i & 6054.08 & 4.37 & -2.200 & 11.6 & 12.6fe i & 6056.01 & 4.73 & -0.400 & 78.3 & 78.3fe i & 6089.57 & 5.02 & -0.860 & 40.0 & 39.2fe i & 6093.65 & 4.61 & -1.340 & 35.3 & 35.9fe i & 6096.67 & 3.98 & -1.810 & 41.4 & 41.9fe i & 6098.25 & 4.56 & -1.740 & 19.4 & 19.4fe i & 6151.62 & 2.18 & -3.290 & 55.0 & 55.3fe i & 6157.73 & 4.07 & -1.250 & 66.7 & 66.0fe i & 6159.38 & 4.61 & -1.870 & 14.2 & 14.2fe i & 6165.36 & 4.14 & -1.470 & 49.0 & 48.2fe i & 6180.21 & 2.73 & -2.610 & 58.0 & 59.7fe i & 6200.32 & 2.61 & -2.440 & 77.2 & 75.1fe i & 6213.44 & 2.22 & -2.660 & 87.6 & 88.2fe i & 6226.74 & 3.88 & -2.030 & 34.1
& 34.2fe i & 6229.23 & 2.84 & -2.820 & 42.3 & 42.3fe i & 6240.65 & 2.22 & -3.320 & 54.2 & 55.2fe i & 6265.14 & 2.18 & -2.570 & 90.7 & 89.7fe i & 6270.23 & 2.86 & -2.570 & 55.4 & 56.0fe i & 6380.75 & 4.19 & -1.320 & 57.7 & 57.5fe i & 6385.73 & 4.73 & -1.820 & 12.6 & 11.5fe i & 6392.54 & 2.28 & -4.010 & 20.6 & 20.5fe i & 6498.95 & 0.96 & -4.620 & 49.7 & 51.9fe i & 6581.22 & 1.48 & -4.660 & 27.9 & 28.2fe i & 6591.33 & 4.59 & -1.980 & 12.7 & 11.5fe i & 6608.04 & 2.28 & -4.010 & 19.9 & 21.2fe i & 6627.56 & 4.55 & -1.440 & 32.6 & 32.6fe i & 6646.97 & 2.61 & -3.850 & 12.0 & 11.3fe i & 6653.91 & 4.15 & -2.410 & 12.0 & 12.4fe i & 6703.58 & 2.76 & -3.010 & 42.2 & 42.8fe i & 6710.32 & 1.48 & -4.800 & 17.2 & 19.0fe i & 6725.36 & 4.10 & -2.180 & 20.8 & 21.3fe i & 6726.67 & 4.61 & -1.040 & 52.3 & 52.5fe i & 6733.15 & 4.64 & -1.450 & 30.8 & 30.7fe i & 6739.52 & 1.56 & -4.900 & 13.0 & 13.8fe i & 6745.98 & 4.07 & -2.770 & 9.1 & 8.7fe i & 6746.98 & 2.61 & -4.410 & 4.7 & 5.3fe i & 6750.16 & 2.42 & -2.620 & 79.4 & 80.4fe i & 6752.72 & 4.64 & -1.200 & 38.9 & 40.2fe i & 6786.86 & 4.19 & -1.950 & 31.2 & 30.5fe i & 6820.37 & 4.64 & -1.170 & 47.9 & 47.4fe i & 6839.84 & 2.56 & -3.360 & 36.8 & 38.2fe i & 6855.72 & 4.61 & -1.730 & 24.1 & 23.8fe i & 6861.95 & 2.42 & -3.800 & 23.5 & 24.3fe i & 6862.50 & 4.56 & -1.350 & 35.0 & 35.7fe i & 6864.32 & 4.56 & -2.300 & 7.7 & 9.0fe i & 7498.54 & 4.14 & -2.090 & 22.4 & 22.7fe i & 7507.27 & 4.41 & -1.050 & 64.7 & 64.5fe i & 7583.80 & 3.02 & -1.900 & 88.9 & 89.1fe i & 7586.03 & 4.31 & -0.180 & 117.6 & 119.8fe ii & 5234.63 & 3.22 & -2.200 & 92.5 & 89.5fe ii & 5991.38 & 3.15 & -3.480 & 38.7 & 35.0fe ii & 6149.25 & 3.89 & -2.700 & 42.4 & 40.4fe ii & 6247.56 & 3.89 & -2.300 & 60.7 & 57.0fe ii & 6369.46 & 2.89 & -4.110 & 26.2 & 23.0fe ii & 6442.95 & 5.55 & -2.380 & 6.2 & 5.3fe ii & 6446.40 & 6.22 & -1.920 & 5.3 & 4.3fe ii & 7515.84 & 3.90 & -3.360 & 17.7 & 15.8 | we utilize high - resolution , high signal - to - noise spectra to perform a differential analysis of fe abundances in the common proper - motion pair 16 cyg a and b. we confirm that both stars are slightly metal - rich compared to the sun , and we show for the first time that the primary is enhanced in fe relative to the secondary by a significant amount .
we find @xmath0[fe / h]@xmath1 .
this tends to support the `` self - pollution '' scenario proposed by gonzalez ( 1998 ) , though lack of a complete understanding of small primordial metallicity variations among binaries and open cluster members prevents a definitive conclusion . |
evidence for the @xmath1 meson was first seen in 1981 @xcite . for the next two decades , the argus experiment operating at desy in hamburg , germany and the cleo experiment operating at cesr at cornell university studied the properties of these mesons and made many important discoveries including the discovery of mixing .
the first examples of charmless hadronic @xmath1 decays were seen by cleo @xcite .
starting in 2000 , there have been many measurements of these decays with increasing precision by the babar experiment operating at pep - ii at slac in california and the belle experiment operating at kek in japan .
there are now nearly 100 of these decays that have been observed with a statistical significance of at least four standard deviations .
while the early measurements by cleo were groundbreaking , the errors are sufficiently large that they have little weight in the present world averages .
the cdf experiment at fnal also has measurements for these decays . while they are nearly as precise as babar and belle , there are only measurements for four decay channels .
thus in tables throughout this review , we concentrate on the copious measurements from belle and babar .
the theoretical study of weak nonleptonic decays of a heavy meson is difficult and involved due to the interplay of short- and long - distance qcd effects .
the effective weak hamiltonian at the quark level @xmath2 is theoretically well under control , where @xmath3 are four - quark operators and @xmath4 are the wilson coefficients which incorporate strong - interaction effects above the scale @xmath5 .
however , it is a difficult task to evaluate the hadronic matrix elements of the local operator @xmath3 reliably due to the nonperturbative qcd effects involved .
a simple and widely employed approach is based on the valence quark assumption and the vacuum - insertion ( or factorization ) approximation in which the hadronic matrix elements of two quark bilinear operators are saturated by the vacuum intermediate states . due to the experimental and theoretical efforts in the past decade ,
qualitative understanding of nonleptonic charmless @xmath1 decays has become possible .
since the @xmath1 meson is heavy , it is possible to describe the dynamics of hadronic decays by theories motivated by qcd rather than by phenomenological models .
a central aspect of those theories is the factorization theorem which allows us to disentangle short - distance qcd dynamics from nonperturbative hadronic effects . in the heavy quark limit
, matrix elements can be expressed in terms of certain nonperturbative input quantities such as light cone distribution amplitudes and transition form factors .
power corrections beyond the heavy quark limit generally give the major theoretical uncertainties . in this article
we give an overview of the experimental measurements and the theoretical understanding of the branching fractions and @xmath0-violating asymmetries of charmless @xmath1-meson decays .
we begin with the theoretical and experimental tools necessary for the study of hadronic @xmath1 decays .
this is followed with a discussion of 2-body , 3-body , quasi-2-body , and baryonic @xmath1 decays , and the status of time - dependent @xmath0-violation measurements and predictions .
we discuss the puzzles that remain unsolved .
in this section we will introduce various approaches that have been employed for studying the dynamics of hadronic @xmath6 decays . in the effective hamiltonian approach ,
the decay amplitude is given by @xmath7 where @xmath8 are cabibbo - kobayashi - maskawa ( ckm ) @xcite matrix elements , @xmath3 are four - quark operators and @xmath4 are the wilson coefficients which incorporate strong - interaction effects above the scale @xmath5 . a major theoretical issue is how to evaluate the matrix elements of the four - quark operators @xmath9 .
a widely used approximation is the so - called naive factorization " or vacuum - insertion approximation " under which the matrix element @xmath10 is approximated by @xmath11 or @xmath12 with @xmath13 being a bilinear current ; that is , the matrix element of a four - quark operator is expressed as a product of a decay constant and a form factor .
naive factorization is simple but fails to describe color - suppressed modes .
for example , the predicted ratio of @xmath14 is too small compared with the experimental value of 0.55 @xcite .
this is ascribed to the fact that color - suppressed decays receive sizable nonfactorizable contributions that have been neglected in naive factorization .
another issue is that the decay amplitude under naive factorization is not truly physical because the renormalization scale and scheme dependence of @xmath4 are not compensated by that of the matrix element @xmath15 . in the improved
generalized factorization " approach @xcite , nonfactorizable effects are absorbed into the parameter @xmath16 , the effective number of colors .
this parameter can be empirically determined from experiment .
since these early calculations have been replaced with improved ones discussed below , we will not discuss these early calculations . with the advent of heavy quark effective theory ,
nonleptonic @xmath1 decays can be analyzed systematically within the qcd framework .
there are three popular approaches available in this regard : qcd factorization ( qcdf ) @xcite , perturbative qcd ( pqcd ) @xcite and soft - collinear effective theory ( scet ) @xcite .
a detailed discussion of these theories goes beyond the scope of this review , and the interested reader is referred to the original literature .
basically , theories of hadronic @xmath1 decays are based on the factorization theorem " under which the short - distance contributions to the decay amplitudes can be separated from the process - independent long - distance parts . in the qcdf approach ,
nonfactorizable contributions to the hadronic matrix elements can be absorbed into the effective parameters @xmath17 @xmath18 where @xmath17 are basically the wilson coefficients in conjunction with short - distance nonfactorizable corrections such as vertex , penguin corrections and hard spectator interactions , and @xmath19 is the matrix element evaluated under the factorization ( or vacuum insertion ) approximation . for penguin operators @xmath20 , the current operators @xmath21 are replaced by the scalar or pseudoscalar densities . since power corrections of order @xmath22 are suppressed in the heavy quark limit , nonfactorizable corrections to nonleptonic decays are calculable . in the limits of @xmath23 and @xmath24 , naive factorization
is recovered in both qcdf and pqcd approaches .
power corrections are often plagued by the end - point divergence that in turn breaks the factorization theorem .
for example , the endpoint divergence occurs at the twist-3 level for the hard spectator scattering amplitude and at the twist-2 level for the annihilation amplitude . as a consequence ,
the estimate of power corrections is generally model dependent and can only be studied in a phenomenological way .
there is also an endpoint singularity in scet though this issue can be resolved after introducing the zero - bin subtraction procedure @xcite . in the pqcd approach ,
the endpoint singularity is cured by including the parton s transverse momentum . because a reliable evaluation of hadronic matrix elements is very difficult in general , an alternative approach is based on the diagrammatic approach .
it has been established that a least model - dependent analysis of heavy meson decays can be carried out in the so - called quark - diagram approach @xcite . in the diagrammatic approach ,
all two - body nonleptonic weak decays of heavy mesons can be expressed in terms of six distinct quark diagrams ( fig .
[ fig : qds ] ) : named in @xcite were originally denoted by @xmath25 , respectively , in @xcite . for the analysis of charmless @xmath1 decays , one adds the variants of the penguin diagram such as the electroweak penguin and the penguin annihilation . ]
@xmath26 , the color - allowed external @xmath27-emission tree diagram ; @xmath28 , the color - suppressed internal @xmath27-emission diagram ; @xmath29 , the @xmath27-exchange diagram ; @xmath30 , the @xmath27-annihilation diagram ; @xmath31 , the penguin diagram ; and @xmath32 , the vertical @xmath27-loop diagram .
it should be stressed that these quark diagrams are classified according to the topologies of weak interactions with all strong interaction effects included and hence they are _ not _ feynman graphs .
all quark graphs used in this approach are topological with all the strong interactions included , i.e. gluon lines are included in all possible ways .
the diagrammatic approach was applied to hadronic @xmath1 decays first in @xcite .
various topological amplitudes have been extracted from the data in @xcite after making some reasonable approximations , e.g. , su(3 ) symmetry .
the experimental measurements involve separation of small samples ( 102000 signal events ) from total samples of @xmath33 and light quark - pairs ( @xmath34 ) of several billion events .
the background is typically dominated by the copious @xmath34 production , where the background after preliminary sample selection is often 1000 times larger than the signal .
the separation of the signal from the large background relies on a number of variables that are common to all analyses . for signal
, the mass of a candidate , reconstructed from the charged and neutral tracks in the event , is equal to the @xmath1 mass , approximately 5.28 gev / c@xmath35 .
the center - of - mass energy of the candidate is equal to one - half of the @xmath36 rest energy since all @xmath37 @xmath1 experiments take advantage of the increased production rate of @xmath33 events at this resonance .
finally there are several variables that take advantage of the difference in the shape of the signal and background events . in the center - of - mass system ,
signal is typically spherical since the @xmath1 mesons are produced nearly at rest .
the @xmath34 background events are are characterized by back - to - back jets " of particles with small transverse momentum with respect to the direction of the leading quarks . for @xmath1
decays involving resonances in the final state ( @xmath38 , @xmath39 , @xmath40 , @xmath41 , @xmath42 , @xmath43 , etc ) , the mass of the daughter particles from the decay of these resonances is also used .
this is useful because the background often is dominated by events with combinatorial background , where the resonance candidate is not real but rather is composed of combinations of particles from the @xmath34 event that happen to have an invariant mass near that of the resonance . for spin-1 ( vector ) particles , the so - called helicity angle is also often useful since the decay of the daughters in the vector particle s rest frame is not typically uniform .
most analyses combine some of the above quantities into a maximum likelihood ( ml ) fit .
such a fit characterizes the signal and background with probability density functions ( pdfs ) that describe the distribution expected for each variable .
it is the difference between the shapes of the signal and background pdfs that allow extraction of the signal from the very large backgrounds .
the shape of these pdfs is typically determined from monte carlo ( mc ) simulation for signal and @xmath33 background events , with checks provided by
control " samples of other more copious decays .
the data itself is used to determine the pdfs for the @xmath34 background .
the free variables in the ml fit typically include the yields of the signal and backgrounds , the value of @xmath0 violation parameters where relevant , the longitudinal polarization fraction for decays with two particles with non - zero spin , and often some of the parameters that determine the @xmath34 background shapes .
the structure of the likelihood often assumes that the input observables are uncorrelated .
this assumption is tested with data and mc and correlations are typically below 10% .
the residual correlations may cause small signal biases ( @xmath4410% ) which are evaluated with mc and appropriate corrections are made .
the general expressions of topological amplitudes are @xmath45 for tree - dominated @xmath46 decays , @xmath47 for @xmath48 decays , and @xmath49 for @xmath50 decays , where @xmath51 and @xmath52 are color - allowed and color - suppressed electroweak penguin amplitudes , respectively , and @xmath53 is the penguin - induced weak annihilation amplitude .
we use unprimed and primed symbols to denote @xmath54 and @xmath55 transitions .
experimental results for branching fractions from babar and belle and theoretical predictions are summarized in fig .
[ f:2-body ] . here and in the following tables , the theoretical values and errors are from weighted averages of the various predictions with the errors divided by @xmath56 since the quoted theory errors indicate a range as parameters are varied , not @xmath57 errors . as mentioned in sec .
2.2 , endpoint divergences will occur in the qcdf approach in the penguin annihilation and hard spectator scattering amplitudes , which are often parametrized as @xcite @xmath58 where the parameters @xmath59 and @xmath60 are real and @xmath61 is a typical hadronic scale of order 500 mev .
since these parameters are unknown within qcd factorization , the central values of qcdf predictions correspond to @xmath62 and the power corrections due to annihilation effects give the largest theoretical uncertainties .
hence , for penguin - dominated modes , the central values predicted by qcdf are usually systematically below the measurements .
penguin annihilation and hard spectator scattering amplitudes are calculable in the pqcd approach as the endpoint singularity is overcome by the parton s transverse momentum .
the predicted central values by pqcd for penguin - dominated decays are normally higher than those of qcdf .
the predicted @xmath63 branching fraction of @xmath64 is too large , whereas @xmath65 of order @xmath66 is too small compared with experiment ( the ratio of the branching fractions is predicted much more precisely than one would infer from fig . [ f:2-body ] ) .
if the penguin amplitudes @xmath31 , @xmath51 , @xmath53 and the annihilation @xmath29 are neglected , a fit to the ratio of @xmath67 and @xmath68 rates will yield @xmath69 . in the short - distance ( @xmath70 ) factorization approach , @xmath71 .
it is thus a challenge to theorists to understand @xmath72 and @xmath63 decays . as for @xmath48 decays , @xmath73 and @xmath74
are dominated by the ckm - suppressed @xmath75 penguin , while @xmath76 proceeds only through weak annihilation .
hence , the branching fraction is expected to be of order @xmath77 for the former and @xmath78 for the latter , in agreement with the data .
the @xmath50 decays are dominated by penguin contributions because of @xmath79 and the large top quark mass . for the ratios defined by @xmath80 we have @xmath81 if the other quark - diagram amplitudes are negligible compared with @xmath82 .
the current experimental measurements give @xmath83 and @xmath84 .
there are two approximate sum rules for @xmath50 rates @xcite and rate asymmetries @xcite @xmath85 , \nonumber \\ { \ensuremath{\delta\gamma}}(k^+\pi^-)+{\ensuremath{\delta\gamma}}({\ensuremath{k^0}}\pi^+)&\approx & 2[{\ensuremath{\delta\gamma}}(k^+\pi^0)+{\ensuremath{\delta\gamma}}({\ensuremath{k^0}}\pi^0)],\end{aligned}\ ] ] based on isospin symmetry , where @xmath86 .
the rate sum rule is fairly well satisfied by experiment .
it is well known that it requires at least two amplitudes with nontrivial relative strong and weak phases to produce direct @xmath0 violation .
the direct @xmath0 asymmetry observed in @xmath87 decays , @xmath88 requires that the relative strong phase @xmath89 between @xmath90 and @xmath91 [ see eq .
( [ eq : ampbkpi ] ) ] be of order @xmath92 . in qcdf , the phase @xmath89 induced perturbatively from vertex corrections and penguin diagrams is too small and has a wrong sign @xcite .
therefore , it is necessary to include the contribution from the long - distance ( ld ) strong phase .
the nonperturbative strong phase induced from the weak decay @xmath93 followed by the final - state rescattering of @xmath94 can reproduce the experimental observation . in the pqcd approach @xcite , a large sd strong phase arises from the penguin - annihilation diagram @xmath95 ,
though this is in contrast to the conventional wisdom that strong phases are basically nonperturbative in nature . in fig .
[ f : acp ] we show @xmath96 for the most precisely measured decays , typically those with an average uncertainty @xmath970.10 or a deviation from zero of at least three standard deviations . of these only @xmath98 and @xmath99
have world averages for @xmath96 that are more than 4 standard deviations from 0 .
theory predictions are not included since the lack of knowledge of strong phases makes them very uncertain .
all of these decays are measured from charge asymmetries as suggested by eq .
[ eq : acpdef ] .
we defer until sec .
[ sec : td ] a discussion of the @xmath0 asymmetries of @xmath0-eigenstate modes such as @xmath68 that are measured with time - dependent techniques .
if the color - suppressed @xmath100 , color - allowed electroweak penguin @xmath101 and annihilation @xmath102 are negligible compared with the tree amplitude @xmath103 , it is clear from eq .
( [ eq : ampbkpi ] ) that the decay amplitudes of @xmath104 and @xmath105 will be the same apart from a trivial factor of @xmath106 .
hence , one will expect that @xmath107 , while they differ by 5.3 @xmath108 experimentally , @xmath109 .
since @xmath110 and @xmath111 with @xmath112 , the puzzle is resolved provided that @xmath113 is of order @xmath114 with a negative relative phase .
there are two possibilities for a large @xmath115 : either a large color suppressed @xmath100 or a large electroweak penguin @xmath101 .
we note that a global fit to @xmath116 , @xmath117 and @xmath118 data in the diagrammatic approach gives a relative strong phase of @xmath119 between @xmath100 and @xmath103 @xcite .
various scenarios for accommodating large @xmath100 @xcite or @xmath101 @xcite have been proposed .
it is evident from the discussion above that a large color - suppressed amplitude with a sizable relative phase to the tree amplitude can solve both the @xmath117 puzzle in @xmath0 violation and the enhancement of @xmath67 in rate .
based on su(3 ) flavor symmetry , direct @xmath0 asymmetries in @xmath117 and @xmath116 systems are related as @xcite : @xmath120 the first relation leads to @xmath121{\ensuremath{a_{\rm cp}}}(k^+\pi^-)\approx 0.37 $ ] , which is in good agreement with the current world average of @xmath122 .
three - body decays of heavy mesons are more complicated than the two - body case as they receive both resonant and nonresonant contributions .
they are generally dominated by intermediate vector and scalar resonances .
the analysis of these decays using the dalitz plot technique enables one to study the properties of various resonances .
moreover , the dalitz plot analysis of 3-body @xmath1 decays provides a nice methodology for extracting information on the unitarity triangle in the standard model .
the experimental results are summarized in fig .
[ f:3-body ] .
the agreement between belle , babar , and theoretical predictions is good in all cases . under the factorization approximation ,
vector and scalar meson resonances contribute to the two - body matrix elements @xmath123 and @xmath124 , respectively , with @xmath125 being a scalar density .
they can also contribute to the three - body matrix element @xmath126 .
resonant effects are generally described in terms of the usual breit - wigner formalism .
for example , @xmath127 where @xmath128 , @xmath129 and @xmath130 .
once the strong couplings for @xmath131 and the decay constants of @xmath132 and @xmath133 are known , we are ready to compute various resonant contributions to the 3-body decays of interest . using the narrow width approximation for resonance @xmath134 @xmath135 we can calculate the rates for the quasi - two - body decays @xmath136 and @xmath137 . for the details of theoretical calculations of charmless 3-body decays of @xmath1 mesons based on the factorization approach , see @xcite ; for a theoretical overview , see @xcite .
one of the salient features of 3-body decays is the large nonresonant contribution to penguin - dominated modes .
it is known that the nonresonant signal in charm decays is small , less than 10% @xcite . on the contrary ,
the nonresonant fraction is about @xmath44 90% in @xmath138 decays , @xmath139 in @xmath140 decays ( smaller in the @xmath141 decay ) , and @xmath44 14% in the @xmath142 decay .
hence , the nonresonant 3-body decays play an essential role in penguin - dominated @xmath1 decays .
nonresonant amplitudes in charm decays are usually assumed to be uniform in phase space
. however , this is no longer true in @xmath1 decays due to the large energy release in weak @xmath1 decays .
this makes the dalitz plot analysis of nonresonant contributions more difficult .
while both babar @xcite and belle @xcite have adopted the parametrization @xmath143 to describe the non - resonant @xmath138 amplitudes , they differ in the analysis of the nonresonant component in @xmath140 decays .
a detailed theoretical analysis in @xcite indicates two distinct sources of nonresonant contributions : a small contribution from the tree transition and a large one from the matrix elements of scalar densities , e.g. , @xmath144 , induced from the penguin transition .
this explains the dominance of the nonresonant background in @xmath138 decays , the sizable nonresonant fraction in @xmath145 and @xmath146 modes and the smallness of nonresonant rates in @xmath142 decays .
mixing - induced @xmath0 asymmetries of 3-body decays will be discussed in sec .
[ [ section ] ] among the 2-body @xmath1 decays , @xmath147 has the largest branching fraction , of order @xmath148 , while @xmath149 is only @xmath150 .
this can be roughly understood as follows .
let s express the @xmath41 and @xmath40 wave functions in the quark - flavor basis @xmath151 and @xmath152 @xmath153 where the mixing angle is extracted from data to be @xmath154 @xcite .
the interference between the @xmath155 amplitude induced by the @xmath156 penguin and the @xmath157 amplitude induced by @xmath158 is constructive for @xmath147 and destructive for @xmath159 . this explains the large rate of the former and the suppression of the latter .
however , the predicted rates are still somewhat smaller than the measurements ( see fig . [
f : eta ] ) .
many possible solutions to the puzzle for the abnormally large @xmath160 rate have been proposed : ( i ) a significant flavor - singlet contribution @xcite , ( ii ) a large @xmath161 form factor @xcite , ( iii ) a contribution from the charm content of the @xmath162 , ( iv ) an enhanced hadronic matrix element @xmath163 due to the axial u(1 ) anomaly @xcite , ( v ) a large chiral scale @xmath164 associated with the @xmath165 @xcite , ( vi ) a long - distance charming penguin in scet @xcite , and ( vii ) a large contribution from the two - gluon fusion mechanism @xcite . because the @xmath41 is dominated by the flavor - singlet component @xmath166 , it is plausible that a sizable flavor - singlet amplitude @xmath167 will account for the large @xmath160 rates .
when this contribution is included in the diagrammatic approach , a global fit to the @xmath168 data leads to a flavor - singlet amplitude @xmath125 about 3 to 4 times the magnitude of @xmath101 @xcite .
this flavor - singlet contribution is also supported by the consideration of the @xmath169 decays .
their topological quark decay amplitudes are @xmath170 where @xmath171 , @xmath172 , @xmath173 , @xmath174 , and we have assumed @xmath175 since this value is algebraically simple .
note that there is no tree contribution to the @xmath176 modes .
if @xmath177 , one will naively expect that @xmath178 .
qcdf predicts @xmath179 and @xmath180 @xcite .
the pqcd approach @xcite has very similar results . from the experimental measurements
@xmath181 and @xmath182 by babar @xcite and @xmath183 by belle @xcite , it appears that the current predictions of qcdf and pqcd may be too small for @xmath184 . in the presence of @xmath125 , @xmath184
is enhanced : @xmath185 @xcite .
since the two penguin processes @xmath158 and @xmath156 contribute destructively to @xmath186 , the penguin amplitude is comparable in magnitude to the tree amplitude induced from @xmath187 , contrary to the decay @xmath188 which is dominated by large penguin amplitudes .
consequently , a sizable direct cp asymmetry is expected in @xmath189 but not in @xmath190 @xcite .
indeed , the average of babar and belle measurements yields a 3@xmath108 effect , @xmath191 .
[ [ section-1 ] ] the experimental results are summarized in fig . [
the decays @xmath136 have been studied in qcd - motivated approaches such as qcdf , pqcd and scet ; see @xcite for comparison of the theory predictions in various approaches with experiment . for tree
dominated decays , the predictions of qcdf and pqcd are in agreement with experiment for @xmath192 and @xmath193 , while the scet calculations are smaller due to the smallness of both @xmath194 and @xmath195 form factors in scet . on the contrary , scet predicts larger rates than qcdf and pqcd for the color - suppressed decays such as @xmath196 because the hard form factor @xmath197 is comparable with the soft part @xmath198 ( @xmath199 ) and is enhanced by a large wilson coefficient . the predicted @xmath200 , of order @xmath201 by qcdf @xcite and of order @xmath202 by pqcd @xcite , are too small compared with the experimental average of @xmath203 while the scet calculations which rely on some input from experiment are consistent with experiment @xcite . for penguin - dominated @xmath136 decays ,
the predictions by qcdf are systematically below the measurements @xcite , while the data can be accommodated by fitting them to scet . as for @xmath204
, qcdf predicts a wrong sign for @xmath205 ) .
the pqcd approach predicts too large direct @xmath0 violation in many @xmath206 modes @xcite .
[ [ section-2 ] ] the underlying mechanism for @xmath207 is more complicated than @xmath206 and @xmath208 modes as it involves three polarization vectors .
the decay amplitude of @xmath207 can be decomposed into three components , one for each helicity of the final state : @xmath209 .
the transverse amplitudes defined in the transversity basis are related to the helicity ones via @xmath210 it is naively expected that the helicity amplitudes @xmath211 ( @xmath212 ) for @xmath213 respect the hierarchy pattern @xmath214 .
hence , charmless @xmath207 decays are expected to be dominated by the longitudinal polarization states and satisfy the scaling law , @xmath215 with @xmath216 and @xmath217 being the longitudinal , perpendicular , and parallel polarization fractions , respectively , defined by @xmath218 with @xmath219 .
the experimental measurements are summarized in fig .
[ f : polar ] .
the progress for these measurements is quite impressive , with the uncertainty for a handful of modes now @xmath220 . in sharp contrast to the @xmath221 case
, the large fraction of transverse polarization observed in @xmath222 and @xmath223 decays at @xmath1 factories is a surprise and poses an interesting challenge for theoretical interpretations .
various mechanisms such as sizable penguin - induced annihilation contributions @xcite , to 0.75 . ] final - state interactions @xcite , form - factor tuning @xcite and new physics @xcite have been proposed for solving the @xmath207 polarization puzzle .
two recent calculations @xcite indicate that nlo nonfactorizable corrections from vertex and penguin corrections and hard spectator scattering will render the positive - helicity amplitude of some @xmath224 modes comparable to the longitudinal one and hence will increase the transverse polarization .
for example , @xmath225 is naively expected to be @xmath226 in @xmath223 and @xmath227 decays .
however , nlo corrections decrease this expectation to @xmath228 and @xmath229 .
therefore , the polarization puzzle is alleviated to a large extent by the consideration of nlo effects .
the theoretical predictions in fig .
[ f : polar ] reflect these more recent calculations . according to the recent calculations based on qcdf @xcite
, there is a hierarchy pattern for the polarization fractions in @xmath222 decays : @xmath230 this pattern is compatible with measurements though only two , @xmath231 and @xmath232 , are well measured .
improved measurements of all of these decays are important in further testing the theoretical calculations .
even though @xmath225 can be substantially reduced in the presence of nonfactorizable corrections , the polarization anomaly is not fully resolved unless the rate is also reproduced correctly .
the experimental branching fraction measurements are summarized in fig .
[ f : vv ] . in most cases
the agreement between theory and experiment is quite good .
the recent measurements of the @xmath233 decays are well below the predicted average though this is somewhat misleading since the average is mostly from pqcd @xcite ; the qcdf prediction @xcite is about a factor of two smaller and in reasonable agreement with the data .
the @xmath223 rate predicted by qcdf ( pqcd ) is too small ( large ) compared with the data .
to improve the situation , qcd factorization and pqcd @xcite rely on penguin annihilation amplitudes , while scet invokes charming penguins @xcite , and the final - state interaction model considers final - state rescattering of intermediate charm states @xcite . in qcdf , the theoretical model predicts @xmath234 @xcite and @xmath235 @xcite , whereas experimentally the latter has a rate larger than the former .
one plausible possibility is that final - state interactions are important for @xmath236 but not for @xmath237 .
the @xmath236 amplitudes can be decomposed into the @xmath238 and 2 isospin states with isospin phases @xmath239 and @xmath240 , respectively .
when the isospin phase difference is sizable , the @xmath65 mode will be enhanced by the final - state rescattering of @xmath63 to @xmath65 ( this amounts to enhancing the color - suppressed amplitude @xmath28 , see also @xcite ) .
since @xmath241 has a rate much larger than @xmath242 , it is natural to expect that @xmath243 will receive large enhancement via isospin final - state interactions .
the fact that the branching fraction of this mode is rather small and is consistent with the theory prediction implies that the isospin phase difference of @xmath244 and @xmath245 must be negligible and so is the final - state interaction @xcite .
[ [ section-3 ] ] much less is known about scalar , axial - vector and tensor mesons , though there are already dozens of measurements involving these modes as decay products of @xmath1 mesons . in this section
we summarize the experimental measurements ( see fig . [
f : other ] ) and the theoretical expectations .
[ [ section-4 ] ] it is known that the identification of light scalar mesons is difficult experimentally and the underlying structure of scalar mesons is not well established theoretically .
it is hoped that through the study of @xmath137 , old puzzles related to the internal structure and related parameters , e.g. the masses and widths , of light scalar mesons can receive new understanding .
for example , it has been shown @xcite that if @xmath246 is a @xmath247 bound state , the predictions are @xmath248 and @xmath249 .
these exceed the current experimental 90% confidence level ( c.l . )
upper limits @xmath250 and @xmath251 @xcite , respectively , suggesting that the four - quark nature for the @xmath246 is favored .
one of the salient features of the scalar meson is that its vector decay constant @xmath252 defined by @xmath253 is either zero or small ( of order @xmath254 , @xmath255 ) .
therefore , when one of the pseudoscalar mesons in @xmath168 decays is replaced by the corresponding scalar , the resulting decay pattern could be very different .
for example , it is expected that @xmath256 and @xmath257 as the factorizable contribution proportional to the decay constant of the scalar meson is suppressed relative to the one proportional to the pseudoscalar meson decay constant .
this feature can be checked experimentally .
the decay @xmath258 is the first charmless @xmath1 decay into a scalar meson observed at @xmath1 factories @xcite .
it receives two different types of penguin contributions : one from @xmath259 and the other from @xmath158 . due to the large scalar decay constant @xmath260 of order 370 mev defined by @xmath261 that appears in the penguin amplitude ,
this decay is dominated by the @xmath158 penguin contribution with predictions of about @xmath262 @xcite .
the experimental measurements are for @xmath263 .
this product is of order @xmath264 for the @xmath265 and @xmath266 decays .
the theoretical predictions are consistent with experiment provided that @xmath267 . [ [ section-5 ] ] there are two distinct types of parity - even axial - vector mesons , namely , @xmath268 and @xmath269 .
the @xmath268 nonet consists of @xmath270 , @xmath271 , @xmath272 and @xmath273 , while the @xmath269 nonet has @xmath274 , @xmath275 , @xmath276 and @xmath277 .
the physical mass eigenstates @xmath278 and @xmath279 are mixtures of @xmath273 and @xmath277 states owing to the mass difference of the strange and non - strange light quarks . a prominent feature of the @xmath269 axial vector meson is that its axial - vector decay constant is small , vanishing in the su(3 ) limit .
this feature was confirmed by the babar observation @xcite that @xmath280 . by contrast , it is expected that @xmath281 since @xmath282
. the predicted branching fractions for the @xmath283 and @xmath284 modes are in good agreement with the babar measurements .
the comparison for @xmath285 and @xmath286 is sometimes good and sometimes not so good ( see fig . [
f : other ] ) ; improved measurements are needed .
[ [ section-6 ] ] decays to @xmath287 and @xmath288 final states have been systematically studied within the framework of qcd factorization @xcite .
the calculations indicate that some of the tree - dominated @xmath289 and @xmath290 modes have sizable rates .
for example , @xmath291 and @xmath292 .
likewise , the @xmath288 modes such as @xmath293 , @xmath294 , @xmath295 and @xmath296 are expected to have branching ratios of @xmath297 . of these , there are only a few upper limits from babar .
the preliminary result for @xmath298 is that the branching fraction is @xmath299 ( 90% c.l . ) .
the strong disagreement with the theoretical prediction is not understood .
a comparison of theory with the current data on @xmath300 , @xmath301 @xcite and @xmath302 @xcite seems to imply that penguin annihilation is small in penguin - dominated @xmath303 decays .
the prediction of @xmath304 in the absence of penguin annihilation @xcite agrees well with the experimental result of @xmath305 @xcite .
this indicates that it is the nlo correction rather than penguin annihilation that is responsible for pushing the longitudinal polarization fraction in @xmath306 down to the level of 0.5 .
[ [ section-7 ] ] many charmless @xmath1 decays with a tensor meson in the final state have been observed at @xmath1 factories ( see fig . [
f : other ] ) .
moreover , babar has measured @xmath225 in the decays @xmath307 and @xmath308 .
contrary to the penguin - dominated @xmath224 modes such as @xmath309 and @xmath310 , @xmath311 has @xmath312 .
intriguingly , the @xmath308 modes have @xmath225 consistent with 0.5 .
so far there are only two theoretical studies of the charmless @xmath1 decay to a tensor meson , both done with the generalized factorization approach @xcite .
neither of these calculations has a prediction for @xmath225 . at the moment
, the data for decays to tensor mesons is well ahead of the theory .
the experimental results for 15 decays involving light baryons are summarized in fig .
[ f : bary ] . in most cases , the agreement between experiment and theory is good .
a peak near the threshold area of the dibaryon invariant mass spectrum has been observed in many 3-body baryonic @xmath1 decays .
the so - called threshold effect indicates that the @xmath1 meson is preferred to decay into a baryon - antibaryon pair with low invariant mass accompanied by a fast recoil meson .
threshold enhancement was first conjectured by hou and soni @xcite .
they argued that in order to have substantial branching fractions for baryonic @xmath1 decays , one has to reduce the energy release and at the same time allow for baryonic ingredients to be present in the final state .
this is indeed the near - threshold effect mentioned above .
while various theoretical ideas @xcite have been put forward to explain the low mass threshold enhancement , this effect can be understood in terms of a simple short - distance picture @xcite . in the two - body
decays , one energetic @xmath247 pair must be emitted back to back by a hard gluon in order to produce a baryon and an antibaryon .
since this hard gluon is highly off mass shell , the two - body decay amplitude is suppressed by order of @xmath313 . in the three - body baryonic @xmath1
decays , a possible configuration is that the baryons in the pair @xmath314 are collinear and in the opposite direction from the meson . at the quark level
, the quark and antiquark emitted from a gluon are moving in nearly the same direction .
since this gluon is close to its mass shell , the corresponding configuration is not subject to the short - distance suppression .
this implies that the dibaryon pair tends to have a small invariant mass .
none of the two - body charmless baryonic @xmath1 decays have been observed so far and the present limit on their branching ratios has been pushed to the level of @xmath315 for @xmath316 and @xmath317 @xcite .
the fact that three - body final states have rates larger than their two - body counterparts , i.e. , @xmath318 is due to the threshold effect discussed above .
the study of 3-body decays is more complicated .
the factorizable contributions fall into two categories : ( i ) the transition process with a meson emission , @xmath319 , and ( ii ) the current - induced process governed by the factorizable amplitude @xmath320 .
the interested reader is referred to @xcite for further details .
naively it appears that the bremsstrahlung process will lead to @xmath321 , with @xmath322 being an electromagnetic fine - structure constant , and hence the radiative baryonic @xmath1 decay is further suppressed than the two - body counterpart , making its observation very difficult at the present level of sensitivity for @xmath1 factories .
however , there is an important short - distance electromagnetic penguin transition @xmath323 .
because of the large top quark mass , the amplitude of @xmath324 is neither quark - mixing nor loop suppressed .
moreover , it is largely enhanced by qcd corrections . as a consequence , the short - distance contribution due to the electromagnetic penguin diagram dominates over the bremsstrahlung .
the relatively large predictions of order @xmath325 @xcite have been confirmed with a measurement from belle for the decay @xmath326 @xcite .
measurement of angular distributions in the dibaryon rest frame will provide further insight to the underlying dynamics .
the sd picture and the pole model both predict a stronger correlation of the outgoing meson with the baryon than the antibaryon in the decay @xmath327 . this feature has been confirmed for @xmath328 and @xmath329 , but not for @xmath330 .
both babar @xcite and belle @xcite found that the @xmath331 in the latter decay prefers to be collinear with the @xmath332 in the @xmath333 rest frame , contrary to the above expectation .
this angular correlation puzzle indicates that either some long - distance effects enter and reverse the angular dependence or the dibaryon pair @xmath333 is produced from some intermediate state e.g. baryonium .
recently belle has made a new measurement of the angular distribution of @xmath334 @xcite .
naively , it is expected that the pion has no preference for its correlation with the @xmath335 or the proton as the dibaryon picks up energetic @xmath336 and @xmath337 quarks , respectively , from the @xmath338 decay .
however , the new belle measurement indicates a correlation between the pion and the @xmath339 . in short
, the correlation enigma has been found in the penguin - dominated modes @xmath330 and @xmath334 and it can not be explained by the sd @xmath340 picture .
this poses a great challenge to theorists .
the @xmath1 factories at kek in japan and pep - ii in california have asymmetric energies : the electron beam has an energy of @xmath341 gev while the positron beam is about 3 gev .
this asymmetry means that the center of mass of the @xmath36 is moving so that the produced @xmath342and @xmath343 mesons do not decay at the same point .
belle and babar take advantage of this to measure , with a precision of about 100@xmath344 m , the distance between the signal " @xmath1 decay and the tagged " @xmath1 decay .
the signal final state can come from either @xmath342 or @xmath343 ; the tagged @xmath1 is either a @xmath342 or @xmath343 with the flavor determined primarily by the charge of leptons or kaons in the event .
this measurement , together with the known properties of the motion of the @xmath36 system , allow measurement of the ( signed ) time difference @xmath345 between the @xmath342and @xmath343 decays .
then the time - dependent asymmetry of the decays to a final state @xmath346 is measured : @xmath347 where @xmath348 is the mass difference of the two neutral @xmath1 eigenstates , @xmath349 monitors mixing - induced @xmath0 asymmetry and @xmath350 measures direct @xmath0 violation ( belle uses @xmath351 ) . in 2001 ,
babar and belle used this technique to observe @xmath0 violation in the @xmath1 meson system for the first time , measuring the value of @xmath125 for the decay @xmath352 and similar @xmath353 decays .
since there is only one quark - level process involved in these decays , this is known to measure @xmath354 with an uncertainty of @xmath440.001 , where @xmath355 ( also called @xmath356 ) is one of the angles of the ckm triangle ( see fig .
[ fig : ut ] ) .
the ambiguity in extracting @xmath355 is resolved with other measurements so that the current world average is @xmath357 . in this section
we discuss how charmless hadronic @xmath1 decays can be used to measure the ckm angle @xmath358 and to search for physics beyond the standard model .
measurements of @xmath125 in the decay @xmath68 can be used to extract the angle @xmath358 .
however there are complications since the penguin amplitudes are not small ( penguin pollution " ) , so a second weak phase is introduced .
gronau and london showed that an isospin analysis @xcite involving also the decays @xmath67 and @xmath359 can be used to determine the amount of penguin pollution and hence extract the angle @xmath358 . while the measurements of @xmath125 and @xmath28 for the decay @xmath68 are fairly precise ( @xmath360 ) @xcite , the sensitivity to @xmath358 is rather poor since the penguin pollution is large in this decay .
the problem can be seen from the relatively large branching fractions for the penguin - dominated @xmath50 decays compared with @xmath46 ( fig .
[ f:2-body ] ) .
the situation is reversed for the @xmath224 decays where @xmath222 tends to be smaller than @xmath361 ( fig .
[ f : vv ] ) .
thus the most sensitive measurements of @xmath358 have come from measurements of @xmath125 for the decay @xmath362 @xcite and the isospin analysis involving also @xmath363 and @xmath364 .
some complications arise for these decays due to polarization and potential corrections to the isospin analysis due to @xmath365 mixing , electroweak - penguin amplitudes , and other isospin - breaking amplitudes .
the magnitude of all of these effects is small compared with the present precision .
while the exact numbers differ due to the statistical analysis used in the treatment of the results from belle and babar , current measurements yield a value for @xmath358 of about @xmath366 with an uncertainty of @xmath447@xmath367 .
the decay @xmath192 can also be used to constrain @xmath358 but this is much less precise with present data .
several methods have been employed for measurements of the ckm angle @xmath368 that are theoretically clean .
all of these methods use decays involving @xmath369 mesons such as @xmath370 and as such are beyond the scope of this review .
the best of these measurements from belle and babar find @xmath371 with uncertainties of @xmath4415@xmath367 .
charmless hadronic @xmath1 decays are also sensitive to @xmath368 and there is a long history of suggested methods , all of which are imprecise with current data .
the most successful methods have used global fits to many of the branching fraction and @xmath96 measurements discussed in this review .
a recent analysis of @xmath206 decays @xcite finds @xmath372 , though this method is generally regarded to be less robust due to flavor su(3 ) breaking and other theoretical uncertainties .
nevertheless , it is clear that measurements of the three angles sum to @xmath373 within the experimental uncertainties of @xmath374 .
possible physics beyond the standard model has been intensively explored through measurements of time - dependent @xmath0asymmetries in the penguin @xmath375 decays such as @xmath376 . in the sm , @xmath125 for these decays should be nearly the same as the value measured for the @xmath353 decays such as @xmath377 ; there is a small deviation _ at most _
in order to detect new physics unambiguously in the penguin @xmath375 modes , it is of great importance to understand sm predictions for the difference @xmath379 with @xmath380 ( @xmath381 ) for final @xmath0-even ( odd ) states .
the quantity @xmath349 has been estimated in various qcd - based approaches ; the results , together with the measured values , are summarized in table [ tab : deltas ] . since @xmath382 @xcite ,
it is clear that @xmath383 are predicted to be small and positive in most cases , while the experimental central values of @xmath383 are negative except for the @xmath384 and @xmath385 modes .
however the average of these measurements , @xmath386 @xcite , is less than one standard deviation below the value from @xmath387 decays .
@xcite & expt @xcite + @xmath388 & @xmath389 & @xmath390 & @xmath391 & @xmath392 + @xmath393 & @xmath394 & @xmath395 & @xmath396 & @xmath397 + @xmath398 & @xmath399 & @xmath400 & @xmath401 & @xmath402 + @xmath403 & @xmath404 & & @xmath405 & @xmath406 + @xmath407 & @xmath408 & & @xmath409 & + @xmath410 & @xmath408 & @xmath411 & @xmath412 & @xmath413 + @xmath414 & @xmath415 & & & @xmath416 + @xmath384 & @xmath417 & & & @xmath418 + @xmath385 & @xmath419 & & & @xmath420 + @xmath421 & @xmath422 & & & @xmath423 +
in this review , we have summarized branching fraction results for more than 100 charmless @xmath1-meson decays .
many of these decays have significant experimental signals .
we also have shown results for @xmath0-violating asymmetry measurements for nearly 50 of these decays .
this represents a truly impressive body of work , most of which have come from the belle and babar experiments in the last decade .
the global features of the branching fractions and cp asymmetries of these charmless @xmath1 decays are generally well described by the qcd - motivated theories such as qcdf , pqcd and scet .
the agreement between theory and experiment is generally satisfactory .
however , there remains some unsolved puzzles : ( i ) @xmath117 cp puzzle : it is naively expected that @xmath424 and @xmath87 have similar direct cp asymmetries , while they differ by 5.3 @xmath108 experimentally , @xmath425 ; ( ii ) the abnormally large @xmath147 rates : while the qualitative picture of the enormously large rate of @xmath188 over @xmath186 is understood , a precise quantitative prediction is still lacking ; ( iii ) branching fractions of @xmath426 : theory usually predicts too small rates for them and a reasonable one for @xmath363 ; ( iv ) the polarization puzzle in penguin - dominated @xmath207 decays : the transverse polarization fraction is not as small as naively anticipated ; ( v ) mixing - induced @xmath0 asymmetries : the predicted values of the effective @xmath427 for most of the @xmath158 induced decays are above the one obtained from @xmath352 , whereas experimentally the value of @xmath354 from the bulk of the decay modes is systematically below that of @xmath352 ; ( vi ) the angular correlation enigma in three - body baryonic @xmath1 decays : the short - distance picture of @xmath340 can not explain or accommodate the observed angular distributions in penguin - dominated decays such as @xmath428 and @xmath429 .
the aforementioned puzzles pose a great challenge to the @xmath1-physics community and their solutions need efforts from both theorists and experimentalists . either these enigmas can be resolved within the framework of the standard model provided that the hadronic matrix elements are under fair control or new physics effects already manifest themselves in some of the puzzles .
we thank rob harr for help with the plots and bill ford and hsiang - nan li for a careful reading of the manuscript .
this work was supported in part by the national science council of r.o.c . under grant nsc97 - 2112-m-001 - 004-my3 and us
departmrnt of energy under grant de - fg02 - 04er41290 .
aubert b , et al .
( babar collaboration ) .
_ 97:171805 ( 2006 ) aubert b , et al .
( babar collaboration ) .
d _ 75:012008 ( 2007 )
aubert b , et al .
( babar collaboration ) .
d _ 76:091102 ( 2007 )
aubert b , et al .
( babar collaboration ) .
arxiv:0807.4226 [ hep - ex ] ( 2008 ) abe k , et al .
( belle collaboration ) .
hep - ex/0610065 ( 2006 ) lin sw , et al .
( belle collaboration ) .
_ phys . rev .
_ 98:181804 ( 2007 ) lin sw , et al .
( belle collaboration ) .
_ 99:121601 ( 2007 ) lu cd , shen yl , wang w. _ phys .
73:034005 ( 2006 ) yoshikawa t. _ phys .
d _ 68:054023 ( 2003 ) ; buras aj , fleischer r , recksiegel s , schwab f. _ phys .
lett . _ 92:101804 ( 2004 ) ; wu yl , zhou yf .
d _ 72:034037 ( 2005 ) ; baek s , et al .
d _ 71:057502 ( 2005 ) ; baek s , london d. _ phys . lett .
b _ 653:249 ( 2007 ) ; feldmann t , jung m , mannel t. _ jhep _ 08:066 ( 2008 ) aubert b , et al .
( babar collaboration ) .
_ 93:181805 ( 2004 ) aubert b , et al .
( babar collaboration ) .
_ 95:011801 ( 2005 ) aubert b , et al .
( babar collaboration ) .
d _ 72:052002 ( 2005 )
aubert b , et al .
( babar collaboration ) .
_ phys . rev .
73:031101 ( 2006 ) aubert b , et al .
( babar collaboration ) .
74:032003 ( 2006 ) aubert b , et al .
( babar collaboration ) .
74:032005 ( 2006 ) aubert b , et al .
( babar collaboration ) .
74:051104 ( 2006 ) aubert b , et al .
( babar collaboration ) .
d _ 76:071104 ( 2007 )
aubert b , et al .
( babar collaboration ) .
_ phys . rev .
99:161802 ( 2007 ) aubert b , et al .
( babar collaboration ) .
. lett . _ 99:221801 ( 2007 ) aubert b , et al .
( babar collaboration ) .
_ phys . rev .
d _ 78:012004 ( 2008 )
aubert b , et al .
( babar collaboration ) .
d _ 78:052005 ( 2008 )
garmash a , et al .
( belle collaboration ) .
d _ 69:012001 ( 2004 )
garmash a , et al .
( belle collaboration ) .
d _ 71:092003 ( 2005 )
garmash a , et al .
( belle collaboration ) .
_ 96:251803 ( 2006 ) garmash a , et al . ( belle collaboration ) .
d _ 75:012006 ( 2007 )
cheng hy , chua ck , soni a. _ phys .
d _ 76:094006 ( 2007 )
aubert b , et al .
( babar collaboration ) .
_ 97:201802 ( 2006 ) aubert b , et al .
( babar collaboration ) .
_ 98:051802 ( 2007 ) aubert b , et al .
( babar collaboration ) .
d _ 75:111102 ( 2007 ) aubert b , et al .
( babar collaboration ) .
d _ 76:031103 ( 2007 )
aubert b , et al .
( babar collaboration ) .
_ 78:011107 ( 2008 ) ford wt ( babar collaboration ) .
arxiv:0810.0494 [ hep - ex ] ( 2008 ) chang p , et al .
( belle collaboration ) .
d _ 71:091106 ( 2005 )
sch " umann j , et al .
( belle collaboration ) .
97:061802 ( 2006 ) chang p , et al .
( belle collaboration ) .
d _ 75:071104 ( 2007 )
sch " umann j , et al .
( belle collaboration ) .
d _ 75:092002 ( 2007 )
wang ch , et al .
( belle collaboration ) .
d _ 75:092005 ( 2007 ) beneke m , neubert m. _ nucl .
b _ 651:225 ( 2003 )
aubert b , et al .
( babar collaboration ) .
_ 91:201802 ( 2003 ) aubert b , et al .
( babar collaboration ) .
d _ 69:011102 ( 2004 )
aubert b , et al .
( babar collaboration ) .
_ 93:051802 ( 2004 ) aubert b , et al .
( babar collaboration ) .
d _ 71:111101 ( 2005 )
aubert b , et al .
( babar collaboration ) .
74:011102 ( 2006 ) aubert b , et al .
( babar collaboration ) .
74:072008 ( 2006 ) aubert b , et al .
( babar collaboration ) .
_ 98:051803 ( 2007 ) aubert b , et al .
( babar collaboration ) .
d _ 76:011103 ( 2007 )
aubert b , et al .
( babar collaboration ) .
d _ 75:091103 ( 2007 )
aubert b , et al .
( babar collaboration ) .
d _ 76:071103 ( 2007 )
gordon a , et al .
( belle collaboration ) .
_ 542:183 ( 2002 ) chen kf , et al .
( belle collaboration ) .
_ 91:201801 ( 2003 ) chang p , et al .
( belle collaboration ) .
b _ 599:148 ( 2004 ) zhang j , et al .
( belle collaboration ) .
_ 94:031801 ( 2005 ) jen cm , et al .
( belle collaboration ) .
_ phys . rev .
d _ 74:111101 ( 2006 ) kusaka a , et al .
( belle collaboration ) .
d _ 77:072001 ( 2008 )
adachi i , et al .
( belle collaboration ) .
belle - conf-0818 ( 2008 ) lu cd , yang mz .
c _ 23:275 ( 2002 ) das pk , yang kc .
d _ 71:094002 ( 2005 ) ; hou ws , nagashima m. hep - ph/0408007 ( 2004 ) ; zou wj , xiao zj .
d _ 72:094026 ( 2005 ) ; chen ch , geng cq .
_ phys . rev .
71:115004 ( 2005 ) ; baek s , et al . _
d _ 72:094008 ( 2005 ) ; chang q , li xq , yang yd . _ jhep _ 0706:038 ( 2007 )
aubert b , et al .
( babar collaboration ) .
91:171802 ( 2003 ) aubert b , et al .
( babar collaboration ) .
arxiv:0901.3703 [ hep - ex ] ( 2009 ) aubert b , et al .
( babar collaboration ) .
_ 97:201801 ( 2006 ) aubert b , et al .
( babar collaboration ) .
arxiv:0901.3522 [ hep - ex ] ( 2009 ) aubert b , et al .
( babar collaboration ) .
d _ 76:052007 ( 2007 )
aubert b , et al .
( babar collaboration ) .
_ 99:201802 ( 2007 ) aubert b , et al .
( babar collaboration ) .
_ 100:081801 ( 2008 ) aubert b , et al .
( babar collaboration ) . _
d _ 78:071104 ( 2008 )
aubert b , et al .
( babar collaboration ) .
_ 101:161801 ( 2008 ) .
aubert b , et al .
( babar collaboration)._phys .
d _ 78:092008 ( 2008 )
aubert b , et al .
( babar collaboration ) .
arxiv:0901.1223 [ hep - ex ] ( 2009 ) zhang j , et al .
( belle collaboration ) .
_ 91:221801 ( 2003 ) zhang j , et al .
( belle collaboration ) .
_ 95:141801 ( 2005 ) somov a , et al .
( belle collaboration ) .
_ 96:171801 ( 2006 ) goldenzweig p , et al .
( belle collaboration ) .
_ 101:231801 ( 2008 ) zhu j , shen yl , lu cd . _ phys .
d _ 72:054015 ( 2005 )
aubert b , et al .
( babar collaboration ) .
_ 97:051802 ( 2006 ) aubert b , et al .
( babar collaboration ) .
_ 99:241803 ( 2007 ) aubert b , et al .
( babar collaboration ) .
99:261801 ( 2007 ) aubert b , et al .
( babar collaboration ) .
_ 100:051803 ( 2008 ) aubert b , et al .
( babar collaboration ) .
d _ 78:011104 ( 2008 )
aubert b , et al .
( babar collaboration ) .
arxiv:0808.0579 [ hep - ex ] ( 2008 ) abe k , et al .
( belle collaboration ) .
arxiv:0706.3279 [ hep - ex ] ( 2007 ) garmash a , et al .
( belle collaboration ) .
_ phys . rev .
d _ 65:092005 ( 2002 ) .
aubert b , et al .
( babar collaboration ) .
d _ 69:091503 ( 2004 )
aubert b , et al .
( babar collaboration ) . _
d _ 72:051101 ( 2005 )
aubert b , et al .
( babar collaboration ) .
d _ 76:092004 ( 2007 )
aubert b , et al .
( babar collaboration ) .
hep - ex/0608020 ( 2006 ) wang mz , et al .
( belle collaboration ) .
_ 90:201802 ( 2003 ) tsai yt , et al .
( belle collaboration ) .
d _ 75:111101 ( 2007 ) wang mz , et al .
( belle collaboration ) .
d _ 76:052004 ( 2007 )
wei jt , et al .
( belle collaboration ) .
b _ 659:80 ( 2008 )
chen jh , et al .
( belle collaboration ) .
_ 100:251801 ( 2008 ) adachi i , et al .
( belle collaboration ) .
belle - conf-0825 ( 2008 ) cheng hy , yang kc .
d _ 66:014020 ( 2002 )
aubert b , et al .
( babar collaboration ) .
d _ 76:071101 ( 2007 ) ; aubert b , et al .
( babar collaboration ) .
arxiv:0708.2097 [ hep - ex ] ( 2008 ) ; aubert b , et al .
( babar collaboration ) .
arxiv:0808.0700 [ hep - ex ] ( 2008 ) ; aubert b , et al .
( babar collaboration ) .
arxiv:0809.1174 [ hep - ex ] ( 2008 ) chen kf , et al .
( belle collaboration ) .
_ 98:031802 ( 2007 ) ; chao y , et al .
( belle collaboration ) .
d _ 76:091103 ( 2007 ) ; abe k , et al .
( belle collaboration ) .
arxiv:0708.1845 [ hep - ex ] ( 2007 ) ; fujikawa m , yusa y , et al .
( belle collaboration ) .
arxiv:0809.4366 [ hep - ex ] ( 2008 ) ; dalseno j , et al .
( belle collaboration ) .
arxiv:0811.3665 [ hep - ex ] ( 2008 ) | we give an overview of the experimental measurements and the theoretical understanding of the branching fractions and @xmath0-violating asymmetries of charmless @xmath1-meson decays .
most experimetal results are from the babar and belle experiments during the past decade .
the global features of these experimental results are typically well described by the qcd - motivated theories such as qcd factorization , pqcd and soft - collinear effective theory .
the agreement between theory and experiment is generally satisfactory , though there remain some unsolved puzzles that pose a great challenge to both theorists and experimentalists .
hadronic decay , cp violation , factorization , polarization , b meson |
the known meson spectrum contains three pseudoscalars [ @xmath11 , all with masses below @xmath12gev @xcite : @xmath13 ; @xmath14 ; and @xmath15 .
the lightest of these , the pion [ @xmath13 ] , is much studied and well understood as qcd s goldstone mode .
it is the basic degree of freedom in chiral effective theories , and a veracious explanation of its properties requires an approach to possess a valid realisation of chiral symmetry and its dynamical breaking . the @xmath14 is broad , with a width of @xmath17 to @xmath18mev . in the framework of constituent - quark models
it is usually interpreted as the pion s first radial excitation .
namely , the @xmath14 is pictured as : an @xmath19 @xmath20 meson , where @xmath21 denotes a constituent - quark ; and the first radially excited state of the @xmath13 on a @xmath20 @xmath22 trajectory , where @xmath23 is the `` principal quantum number '' @xcite . at first sight
it might appear natural to interpret the @xmath15 as the third state on the @xmath22 trajectory .
however , in comparison with @xmath14 , the @xmath15 is narrow , with a width of @xmath24mev , and has a decay pattern that may be consistent with its interpretation as a _ hybrid _
meson in constituent - quark models @xcite .
this picture has the constituent - quarks spins aligned to produce @xmath25 with @xmath26 obtained by coupling @xmath27 to a spin-@xmath28 excitation of the confinement potential .
it is legitimate to ask for a unified theoretical understanding of these states and , indeed , the entire trajectory of pseudoscalar mesons .
this is a topical question ; e.g. , refs.@xcite , and it is easy to identify at least one reason why . in the context of a constituent - quark model hamiltonian a subset , if not all , of the pseudoscalar mesons form a @xmath29 @xmath22 trajectory . in this framework
the support possessed at long - range by the bound state s wave function grows with increasing @xmath23 .
hence the properties of radially excited states become increasingly sensitive to the manner by which confinement is expressed in the potential . as we have already noted , in this same context
a definition and representation of hybrid mesons requires that explicit excitation of the confinement potential be included as an additional degree of freedom .
seen from this perspective one may anticipate that the properties of all the heavier pseudoscalar mesons are likely to be sensitive to the long - range part of the interaction between light - quarks in qcd , whether they be radial excitations or hybrid mesons .
this suggests that the study of their properties can provide a map of what might be called the confinement potential between light - quarks .
( nb . the information obtained thereby is complementary to that gathered in studies of axial - vector mesons @xcite , which in constituent - quark models are interpreted as orbital excitations of the @xmath30- and @xmath31-mesons . ) it is not possible to accurately describe pseudoscalar mesons using a framework that fails to respect the axial - vector ward - takahashi identity .
for example , chiral symmetry and its dynamical breakdown force the leptonic decay constant of every pseudoscalar meson , _ except _ the goldstone mode , to vanish in the chiral limit @xcite .
herein we therefore employ qcd s dyson - schwinger equations ( dses ) ( modern applications are reviewed in refs.@xcite ) for which a systematic , poincar covariant and symmetry preserving treatment of quark - antiquark bound states has been established @xcite . to provide exemplars we will focus primarily on the @xmath13 and the next - lightest pseudoscalar state .
nonetheless , the exact results will apply to all elements on the pseudoscalar meson trajectory .
it is noteworthy that in poincar covariant quantum field theory all bound states with given quantum numbers ; e.g. , ( @xmath32 , @xmath33 ) , are described by the same homogeneous bethe - salpeter equation ( bse ) .
this is kindred to the statement that all interpolating fields with the same quantum numbers are on - shell equivalent , a fact which is apparent in numerical simulations of lattice - qcd ; e.g. , ref.@xcite .
hence a given homogeneous bse yields the mass and bethe - salpeter amplitude of every bound state in the channel specified by ( @xmath32 , @xmath33 ) . in a confining theory a given @xmath34 trajectory will likely contain a countable infinity of bound states .
the lowest mass member of the trajectory is conventionally described as the ground state .
all other members may reasonably be described as excited states .
the radial excitation of a state with a given @xmath34 preserves this total - momentum@xmath35parity assignment .
however , it may be distinguished from the ground state by the pointwise behaviour of its bethe - salpeter amplitude , which when analysed appropriately exhibits a finite number of zeros . as in quantum mechanics
, the number of zeros can be associated with a principal quantum number @xmath23 .
studies of pseudoscalar mesons show that the ground state amplitude has no zeros and can therefore be associated with @xmath1 .
the amplitude of the next highest mass pseudoscalar possesses one zero and is therefore identified with @xmath36 ; e.g. , @xcite .
in simple models , this pattern continues @xcite .
it may be that hybrid mesons , if they exist , can likewise be identified through the pointwise behaviour of their bethe - salpeter amplitudes .
for example , a solution of the pseudoscalar bse , heavier than the first radial excitation , whose bethe - salpeter amplitude exhibits both : a pattern of zeros which does not match that associated with radial excitations ; and relationships between component functions in the bethe - salpeter amplitude different from those present in the lower mass solutions , would appear a reasonable hybrid candidate . of course , bethe - salpeter amplitudes are not themselves observable and the experimental categorisation of ground , and excited and putative hybrid states proceeds via analysis of their decay patterns . notwithstanding this
, the order in those decay patterns is determined in large part by the bethe - salpeter amplitudes pointwise behaviour .
we therefore anticipate that a natural distinction between straightforward radial excitations and hybrids may be possible without recourse to a constituent - quark model basis . in sec.[gapbse ] we recapitulate on aspects of the dses and truncation scheme that are relevant to our study .
the abelian anomaly features in sec.[exact ] , wherein exact results are derived regarding the coupling of pseudoscalar mesons to two photons .
we outline a renormalisation - group - improved model of the quark - antiquark scattering kernel in sec.[model ] .
it is used in that section to illustrate the exact results and explore effects of the model s realisation of light - quark confinement on , e.g. , bound state charge radii .
section [ epilogue ] is an epilogue .
a poincar covariant and symmetry preserving treatment of quark - antiquark bound states can be based on the homogeneous bethe - salpeter equation ( bse ) @xcite @xmath37_{tu } = \int^\lambda_q [ \chi(q;p)]_{sr}\ , k_{rs}^{tu}(q , k;p)\,,\ ] ] where : @xmath38 is the relative and @xmath39 the total momentum of the constituents ; @xmath40, ,@xmath41 represent colour , dirac and flavour indices ; @xmath42 @xmath43 ; and @xmath44 represents a poincar invariant regularisation of the integral , with @xmath45 the regularisation mass - scale @xcite .
( we shall subsequently describe regularisation explicitly . ) in eq.([bse1 ] ) , @xmath46 is the renormalised dressed - quark propagator and @xmath47 is the fully amputated dressed - quark - antiquark scattering kernel ; namely , it is the sum of all diagrams that can not be disconnected by cutting two fermion lines .
the product @xmath48 is a renormalisation point invariant .
hence , when the kernel is expressed completely in terms of renormalised schwinger functions , the homogeneous bse s solution is independent of the regularisation mass - scale , which may be removed ; viz .
, @xmath49 . in a given channel
the homogeneous bse only has solutions for particular , separated values of @xmath50 : @xmath51 , where @xmath52 is a bound state s mass , whereat @xmath53 is that bound state s bethe - salpeter amplitude . in the flavour nonsinglet pseudoscalar channel
the lowest mass solution is associated with the @xmath13 .
it is precisely qcd s goldstone mode @xcite , and we denote it by a value of @xmath1 .
the homogeneous bse next possesses a @xmath54 solution when @xmath50 assumes a value associated with the mass of the @xmath14 .
we label this state by @xmath36 . in the study of this meson in ref.@xcite
the tchebychev moments of the lorentz scalar functions that appear in the matrix - valued bethe - salpeter amplitude each exhibit a single zero .
it can therefore be described as a radially excited state .
hereafter the subscript @xmath23 is merely a counter labelling states of increasing mass : @xmath55 , etc . )
the pattern of isolated solutions continues so that in principle one may obtain the mass and amplitude of every pseudoscalar meson from eq.([bse1 ] ) .
herein we will exploit this in comparing properties of the two lowest - mass flavour - nonsinglet @xmath54 mesons just described .
the dressed - quark propagator appearing in the bse s kernel is determined by the renormalised gap equation @xmath56 wherein : @xmath57 is the dressed - gluon propagator , @xmath58 is the dressed - quark - gluon vertex , and @xmath59 is the @xmath45-dependent current - quark bare mass .
the quark - gluon - vertex and quark wave function renormalisation constants , @xmath60 , depend on the gauge parameter , the renormalisation point , @xmath61 , and the regularisation mass - scale .
a poincar invariant regularisation of the integral is essential and , since pseudoscalar mesons are our focus , we employ a pauli - villars scheme . that is implemented in eq.([gendse ] ) by considering the quarks as minimally anticoupled ( @xmath62 ) to additional massive gluons ( @xmath63 ) .
this effects a tempering of the integrand , which is expressed via a modification of the gluon propagator s ultraviolet behaviour : @xmath64 and regulates the integral s superficial linear divergence . the gap equation s solution has the form @xmath65 .
\label{sinvp } \end{aligned}\ ] ] it is obtained from eq.([gendse ] ) augmented by the renormalisation condition @xmath66 where @xmath67 is the renormalised ( running ) current - quark mass : @xmath68 with @xmath69 the lagrangian mass renormalisation constant . at one - loop order in perturbative qcd @xmath70 with @xmath71 , where @xmath72 is the number of active current - quark flavours , and @xmath73 is the renormalisation - point - invariant current - quark mass .
the chiral limit is unambiguously defined by setting @xmath74 @xcite , which is equivalent to the requirement @xmath75 the behaviour and features of the solution of qcd s gap equation are reviewed in refs.@xcite .
it is a longstanding prediction of dse studies that the dressed - quark propagator is strongly dressed at infrared length - scales , namely , @xmath76gev@xmath77 and that this is materially important in explaining a wide range of hadron properties @xcite .
indeed , an enhancement of the mass function , @xmath78 , is central to the appearance of a constituent - quark mass - scale and an existential prerequisite for goldstone modes .
the dse results have been confirmed in numerical simulations of lattice - regularised qcd @xcite and the conditions have been explored under which pointwise agreement between dse results and lattice simulations may be obtained @xcite .
the @xmath79 trajectory contains the pion , whose properties are fundamentally governed by the phenomenon of dynamical chiral symmetry breaking ( dcsb ) .
one expression of the chiral properties of qcd is the axial - vector ward - takahashi identity @xmath80 which we have here written for two quark flavours , each with the same current - quark mass : @xmath81 are flavour pauli matrices . in eq.([avwtim ] ) , @xmath82 is the axial - vector vertex : @xmath83_{tu } & = & z_2 \left[\gamma_5\gamma_\mu \frac{\tau^j}{2 } \right]_{tu}\\ & + & \int^\lambda_q [ \chi^j_{5\mu}(q;p)]_{sr } k_{tu}^{rs}(q , k;p)\ , , \label{avbse}\end{aligned}\ ] ] and @xmath84 is the pseudoscalar vertex @xmath85_{tu } & = & z_4 \left[\gamma_5 \frac{\tau^j}{2}\right]_{tu}\\ & + & \int^\lambda_q [ \chi^j_{5}(q;p)]_{sr } k_{tu}^{rs}(q , k;p)\ , .
\label{psbse}\end{aligned}\ ] ] the quark propagator , axial - vector and pseudoscalar vertices are all expressed via integral equations ; i.e. , dses .
equation ( [ avwtim ] ) is an exact statement about chiral symmetry and the pattern by which it is broken .
hence it must always be satisfied . since that can not credibly be achieved through fine tuning , the distinct kernels of eqs.([gendse ] ) , ( [ gensigma ] ) , ( [ avbse ] ) , ( [ psbse ] ) must be intimately related .
any theoretical tool employed in calculating properties of the pseudoscalar and pseudovector channels must preserve that relationship if the results are to be both quantitatively and qualitatively reliable . while a weak coupling expansion of the dses yields perturbation theory and satisfies this constraint , that truncation scheme is not useful in the study of bound states nor of other intrinsically nonperturbative phenomena ; such as confinement and dcsb .
fortunately at least one nonperturbative , systematic and symmetry preserving scheme exists .
( references@xcite give details . )
this entails that the full implications of eq.([avwtim ] ) can be elucidated and illustrated .
unless there is a reason for the residue to vanish , every isovector pseudoscalar meson appears as a pole contribution to the axial - vector and pseudoscalar vertices @xcite : @xmath86 viz .
, each vertex may be expressed as a simple pole plus terms regular in the neighbourhood of this pole , with @xmath87 representing the bound state s canonically normalised bethe - salpeter amplitude : @xmath88 \ ! ; \label{genpibsa } \end{aligned}\ ] ] and @xmath89 the residues expressed in eqs.([fpin ] ) and ( [ cpres ] ) , are gauge invariant and cutoff independent . for an elementary pseudoscalar meson , @xmath90 in eq.([genpibsa ] ) .
the first two of these functions can be described as characterising the pseudoscalar meson s pseudovector components ; and the last , its pseudotensor component .
the associated dirac structures necessarily occur in a poincar covariant bound state description : they signal the presence of quark orbital angular momentum .
equation([avwtim ] ) combined with eqs.([genavv ] ) ( [ cpres ] ) yields @xcite @xmath91 i.e. , an identity valid : for every flavour nonsinglet @xmath92 meson ; and irrespective of the magnitude of the current - quark mass @xcite . in the chiral limit additional information about the ground state pseudoscalar ( @xmath1 ) is available ; namely , an array of quark - level goldberger - treiman relations @xcite @xmath93 where @xmath94 , @xmath95 , @xmath96 are , respectively , the coefficient functions of @xmath97 , @xmath98 , @xmath99 in @xmath100 and @xmath101 equations ( [ bwti ] ) ( [ hwti ] ) are a pointwise consequence of dcsb and a pointwise expression of goldstone s theorem .
they can be used to show @xmath102 wherein @xmath103 is the vacuum quark condensate @xcite .
it is now plain from eq.([gmorgen ] ) that in the neighbourhood of @xmath74 @xmath104 viz . , the gell - mann oakes renner relation is a corollary of eq.([gmorgen ] )
to be concrete we will begin by considering the two - photon coupling as expressed via the renormalised triangle diagrams : @xmath105 where @xmath106 , the electric charge matrix @xmath107=e\,{\rm diag}[2/3,-1/3]$ ] , @xmath108 $ ] and @xmath109_{tu } = z_2 \left[\gamma_\mu \right]_{tu}\\ + \int^\lambda_q [ \chi^j_{\mu}(q;p)]_{sr } k_{tu}^{rs}(q , k;p)\ ] ] is the renormalised dressed - quark - photon vertex . the bare axial - vector vector vector vertex exhibits a superficial linear divergence and , as with all other schwinger functions , it must be rigorously defined via a poincar invariant regularisation scheme . in this case
an appropriate pauli - villars prescription corresponds to minimally anticoupling the photon to additional flavoured quarks with a large mass @xmath110 . to elucidate ,
we introduce @xmath111 wherein the current - quark - mass dependence is explicit , so that eq.([tmnr ] ) can rigorously be written as @xmath112 with @xmath113 as the last step in the calculation .
the dressed - quark propagators in eqs.([tmnr ] ) ( [ pvregd ] ) are understood to be calculated using the rainbow - truncation gap equation , which is defined by eq.([gendse ] ) with @xmath114 wherein @xmath115 is the free gauge boson propagator @xcite and @xmath116 will subsequently be specified .
the remaining element , the axial - vector vertex , is obtained from the ladder bethe - salpeter equation , whose kernel ( see eq.([bse1 ] ) , for example ) is defined by the dressed - quark propagators just specified and @xmath117_{ts } \ , \left[\gamma_\nu \frac{\lambda^a}{2}\right]_{ru } \!\!\!.
\label{ladderk}\end{aligned}\ ] ] in what follows it is important that the rainbow - ladder truncation is the first term in the systematic and symmetry preserving truncation scheme described in refs.@xcite and , furthermore , that with the choice @xmath118 the rainbow - ladder truncation is guaranteed to express the one - loop renormalisation group properties of qcd .
the axial - vector ward - takahashi identity depicted in fig.[figavwti ] is an analogue of @xmath119 it can be derived following the method in refs.@xcite if , and only if , every dressed - quark propagator that appears is obtained from the rainbow dse and the accompanying dressed vertices are determined from the ladder bethe - salpeter equation , both of which have just been defined . using the identity in fig.[figavwti ] it
can be shown @xcite that @xmath120 where @xmath121 .
this is an explicit demonstration that the triangle - diagram representation of the axial - vector two - photon coupling calculated in the rainbow - ladder truncation is a necessary and sufficient pairing to preserve the abelian anomaly . in general the coupling of an axial - vector current to two photons
is described by a six - point schwinger function , to which eq.([tmnr ] ) is an approximation .
the same is true of the pseudoscalar two - photon coupling and its connection with eq.([pmnr ] ) .
equation ( [ anomaly ] ) is valid for any and all values of @xmath122 .
it is an exact statement of a divergence relation between these two six - point schwinger functions , which is preserved by the truncation we will subsequently employ in illustrative quantitative studies . before providing those illustrations ,
however , we derive corollaries of eq.([anomaly ] ) that have important implications for the properties of pseudoscalar bound states .
if one inserts eqs.([genavv ] ) and ( [ genpv ] ) into eq.([anomaly ] ) and uses eq.([gmorgen ] ) , one finds that in the neighbourhood of each electric - charge - neutral pseudoscalar - meson bound - state pole @xmath123 in this equation , @xmath124 are nonresonant or _
contributions to the relevant schwinger functions , whose form is concretely illustrated herein upon substitution of @xmath100 and @xmath125 into eqs.([tmnr ] ) and ( [ pmnr ] ) , respectively .
moreover , @xmath126 is the six - point schwinger function describing the bound state contribution , which in rainbow - ladder truncation is realised as @xmath127 this schwinger function describes the direct coupling of a pseudoscalar meson to two photons .
the support properties of the bound state bethe - salpeter amplitude guarantee that the renormalised schwinger function is finite so that the regularising parameter can be removed ; i.e. , @xmath128 , in general and in our truncation , eq.([tpingg ] ) .
we note that owing to the @xmath129 ( euclidean lorentz ) transformation properties of each term on the l.h.s . in eq.([anomaly ] ) , one may write @xmath130 so that eq.([anomaly ] ) can be compactly expressed as @xmath131 it has been proven @xcite that in the chiral limit @xmath132 hence it follows from eq.([reganomaly ] ) that in this limit all pseudoscalar mesons , _ except _ the goldstone mode , decouple from the divergence of the axial - vector two - photon vertex .
( this is true unless @xmath133 diverges in the chiral limit , which is not the case , as we will see . ) in the chiral limit the pole associated with the ground state pion appears at @xmath134 and thus @xmath135 namely , outside the neighbourhood of the ground state pole the regular ( or continuum ) part of the divergence of the axial - vector vertex saturates the anomaly in the divergence of the axial - vector two - photon coupling .
on the other hand ,
in the neighbourhood of @xmath134 @xmath136 i.e. , on this domain the contribution to the axial - vector two - photon coupling from the regular part of the divergence of the axial - vector vertex combines with the direct @xmath137 vertex to fulfill the anomaly .
this fact was illustrated in ref.@xcite by direct calculation : eqs.([bwti ] ) ( [ hwti ] ) are an essential part of that demonstration .
if one defines @xmath138 in which case @xmath139 , then the physical width of the neutral ground state pion is determined by @xmath140 viz . , the second term on the l.h.s . of eq.([anomalypion ] ) evaluated at the on - shell points .
this result is not useful unless one has a means of estimating the contribution from the first term ; viz .
, @xmath141 .
however , that is readily done .
a consideration @xcite of the structure of the regular piece in eq.([genavv ] ) indicates that the impact of this continuum term on the @xmath142 coupling is modulated by the magnitude of the pion s mass , which is small for realistic @xmath41 and @xmath143 current - quark masses and vanishes in the chiral limit .
one therefore expects this term to contribute very little and anticipates from eq.([anomalypion ] ) that @xmath144 is a good approximation .
this is verified in explicit calculations ; e.g. , in ref.@xcite , which evaluates the triangle diagrams described herein , the first term on the l.h.s . modifies the result in eq.([anomalycouple ] ) by less than 2% .
there is no reason to expect an analogous result for pseudoscalar mesons other than the @xmath13 ; i.e. , the states which we denote by @xmath145 . indeed ,
as all known such pseudoscalar mesons have experimentally determined masses that are greater than @xmath146gev , the reasoning used above suggests that the presence of the continuum terms , @xmath141 and @xmath147 , must materially impact upon the value of @xmath148 .
this will subsequently be illustrated using the rainbow - ladder truncation .
we have stated that the rainbow - ladder truncation preserves the one - loop renormalisation group properties of qcd .
it follows that eq.([tpingg ] ) should reproduce the leading large-@xmath7 behaviour of the @xmath149 transition form factor inferred from perturbative qcd .
the qcd analysis has been performed for the ground state pion ( @xmath1 ) with the result @xcite @xmath150 and ref.@xcite verified that this is indeed the result contained in eq.([tpingg ] ) .
however , it is useful for our purposes to recapitulate on that derivation .
consider eq.([tpingg ] ) : the integral is finite and hence a shift in the integration variable is permitted , @xmath151 where @xmath152 and @xmath153 .
we assume that @xmath154 with @xmath155 and , because we do not restrict ourselves to ground state pseudoscalar mesons , assume besides that for the given @xmath23 under consideration @xmath156 . on this domain @xmath157 , @xmath158 , and it is valid at leading @xmath159-order in eq.([tpinggn ] ) to write @xcite @xmath160 so that @xmath161 since we are concerned with @xmath162 states , it follows that @xmath163 , \label{tanswer}\end{aligned}\ ] ] where eq.([tpinggna ] ) yields @xmath164 with @xmath165 $ ] . as we show in the appendix , on the large-@xmath7 domain ,
that part of @xmath166 which contributes to @xmath167 is @xmath168 @xmath169 , where @xmath170 is an anomalous dimension and @xmath171 is a mass - scale associated with the momentum space width of the meson s bethe - salpeter wave function .
similar reasoning exposes the leading contribution to eq.([tanswer ] ) from eq.([jres ] ) : @xmath172 @xmath169 . combining these results
one arrives at @xmath173.\label{enduv}\end{aligned}\ ] ] we emphasise that the coefficient of the leading @xmath174-term in eq.([enduv ] ) is exact and model - independent .
that is not true of the subleading @xmath8 term .
furthermore , with a given _ ansatz _ for @xmath175 in eqs.([rainbowdse ] ) and ( [ ladderk ] ) , eq.([expand ] ) is not sufficient to accurately determine the value of the coefficient of the @xmath8 term or the anomalous dimension because , for example , momentum - dependent dressing of the quark - photon vertex can contribute at this order .
nevertheless , our analysis highlights the existence of a nonzero subleading @xmath8 contribution whose strength is sensitive to features of the dynamics .
these observations were made previously for the ground state ( @xmath1 ) pion @xcite .
we can now return to one of the stated reasons for this analysis : eq.([enduv ] ) inserted in eq.([tgdef ] ) and combined with eq.([tpig ] ) reproduces the leading order result obtained in perturbative qcd , eq.([tpiuvqcd ] ) .
in fact , it provides more .
the perturbative result was only derived for the ground state pseudoscalar meson .
our analysis shows that for each meson on the pseudoscalar trajectory , identified herein by a value of @xmath23 , qcd predicts @xmath176 . \label{uvnot0}\end{aligned}\ ] ] it is now apparent from eq.([fpizero ] ) that @xmath177 @xmath178 namely , in the chiral limit the leading - order power - law in the transition form factor for excited state pseudoscalar mesons is o@xmath179 .
this result is model - independent .
furthermore , while we can not determine the qcd value of the coefficient @xmath180 in the present truncation , in general that coefficient is _ not _ proportional to @xmath10 , or some power thereof , for any value of @xmath23 .
we will see this clearly in the @xmath145 transition form factor for which , if that were the case , the @xmath8-term would be absent in the chiral limit . for all pseudoscalar states
there are mass - scales other than @xmath181 that are nonzero even in the chiral limit when chiral symmetry is dynamically broken .
in order to illustrate the results presented above and calculate other observables it is necessary to specify @xmath175 in eqs.([rainbowdse ] ) and ( [ ladderk ] ) .
we choose @xmath182 } \ , { \cal f}(s)\,,\ ] ] with @xmath183)]/s$ ] , @xmath184gev , @xmath185 , @xmath186 and @xmath187gev .
this form expresses the interaction as a sum of two terms .
the second guarantees eq.([calguv ] ) and therefore ensures that perturbative behaviour is correctly realised at short range ; namely , as written , for @xmath188gev@xmath77 , @xmath47 is precisely as prescribed by qcd . on the other hand ,
the first term in @xmath175 is a model for the long - range behaviour of the interaction .
it is a finite width representation of the form introduced in ref.@xcite , which has been rendered as an integrable regularisation of @xmath189 @xcite .
this interpretation , when combined with the result that in a heavy - quark heavy - antiquark bse the renormalisation - group - improved ladder truncation is exact @xcite , is consistent with @xmath175 leading to a richardson - like potential @xcite between static sources . the active parameters in eq.([calg ] ) are @xmath190 and @xmath191 , which together determine the integrated infrared strength of the rainbow - ladder kernel , but they are not independent . in fitting a selection of ground state observables @xcite , a change in one is compensated by altering the other ; e.g. , on the domain @xmath192\,$]gev , the fitted observables are approximately constant along the trajectory @xmath193 ( nb . the value of @xmath194 is typical of the mass - scale associated with nonperturbative gluon dynamics . ) herein , unless otherwise stated , we use @xmath195 equation ( [ calg ] ) defines a renormalisation - group - improved rainbow - ladder truncation .
this form , introduced in refs.@xcite , has been employed extensively in the calculation of properties of ground state pseudoscalar and vector mesons @xcite .
these applications are reviewed in ref.@xcite , from which it is apparent that the model describes a basket of thirty - one hadron observables with a rms error between calculation and experiment of @xmath196% .
the calculation of observables is now straightforward .
the kernel of the gap equation , eq.([rainbowdse ] ) , is completely specified .
thus a solution follows immediately upon fixing the current - quark mass : this sets the _
boundary condition _ , eq.([renorms ] ) .
we focus on the @xmath41-@xmath143 sector and assume isospin symmetry : @xmath197 with a result for the dressed - quark propagator in hand , the kernel of bethe - salpeter equations is also complete . the solutions of these equations yield : the bound state bethe - salpeter amplitudes ; the axial - vector and pseudoscalar vertices ; and the dressed - quark - photon vertex , all of which appear above . at this point
one has every element necessary for the calculation of an amplitude such as eq.([tpinggn ] ) and therewith experimental observables .
the numerical procedures are described in refs.@xcite .
small-@xmath7 behaviour of the @xmath149 transition form factor , defined in eq.([tpig ] ) , calculated with the current - quark mass in eq.([qmass ] ) .
the ground state s two - photon coupling suggested by eq.([anomalycouple ] ) is marked by `` @xmath198 '' .
, scaledwidth=48.0% ] figure [ fig : tpiggnmm ] depicts the small-@xmath7 behaviour of the @xmath149 transition form factor defined in eq.([tpig ] ) , calculated for the two lowest - mass @xmath199 states with @xmath200 ( recall that in this model the @xmath36 state is a radial excitation . )
it is notable that while @xmath201 , @xmath202 viz .
, it is negative on the entire kinematically accessible domain .
moreover , for nonzero current - quark mass we expect the sign of this form factor to duplicate the pattern set by the leptonic decay constant , which is @xmath203 @xcite .
nb . on the depicted domain and with the resolution in this figure there is no perceptible difference between these curves and those obtained in the chiral limit .
that is not true for larger @xmath7 , as will become apparent .
the coupling constants for decay into two real photons are presented in table [ table : couplings ] , as are the associated decay widths , calculated using @xmath204 it is evident from table [ table : couplings ] that eq.([anomalycouple ] ) is truly a good approximation for the @xmath13 .
l@ l@|l@ l@l@l@ & & @xmath205 & @xmath206 & @xmath207 & @xmath208 + & & ( gev ) & ( gev ) & ( gev)@xmath209 & ( ev ) + @xmath210 & @xmath211 & @xmath212 & @xmath213 & @xmath214 & + & @xmath73 , eq.([qmass ] ) & @xmath215 & @xmath216 & @xmath217 & @xmath218 + @xmath219 & @xmath211 & @xmath220 & @xmath221 & @xmath222 + & @xmath73 , eq.([qmass ] ) & @xmath223 & @xmath224 & @xmath225 & @xmath226 + the result for @xmath227 is , however , striking .
this coupling is negative because the @xmath219 s bethe - salpeter amplitude has a significant domain of negative support @xcite ; and while its magnitude is material , @xmath228 , it is finite even in the chiral limit .
the last fact demonstrates that the @xmath229 coupling is not inversely proportional to @xmath230 cf .
eq.([anomalycouple ] ) .
this confirms that the excited state decouples from the axial - vector two - photon vertex in the chiral limit , as described in connection with eq.([reganomaly0 ] ) .
consequently , the evolution with @xmath50 of the regular ( or continuum ) part of the divergence of the axial - vector two - photon vertex is smooth ; i.e. , @xmath231 and in addition @xmath232 _ { p^2= -m_{\pi_1}^2 } \approx \frac{1}{2}\,,\ ] ] with exact equality for @xmath211 . in fig.[fig :
uv01 m ] we depict the large-@xmath7 behaviour of the @xmath149 transition form factor obtained with the nonzero current - quark mass in eq.([qmass ] ) , for the two lowest mass pseudoscalars .
the ultraviolet behaviour anticipated for the ground state from perturbative qcd , eq.([tpiuvqcd ] ) , is evident .
this is a numerical verification of the argument associated with eqs.([tpinggn ] ) ( [ uvchiralnot0 ] ) ; viz .
, that the truncation we employ preserves leading - order qcd results . the analogous result for the first excited state , indicated by eq.([uvnot0 ] ) , is also conspicuous .
calculated large-@xmath7 behaviour of the @xmath149 transition form factor , eq.([tpig ] ) : _ diamonds _ ground state , @xmath1 ; and _ circles _ first excited state , @xmath36 .
the _ solid - lines _ are eq.([tpiuvqcd ] ) with either @xmath233 or @xmath230 from table [ table : couplings ] , as appropriate.,scaledwidth=48.0% ] for the ground state the behaviour of the transition form factor in the chiral limit is not markedly different from that found with @xmath73 in eq.([qmass ] ) and illustrated in fig.[fig : uv01 m ] . as evident in fig.[fig : uv1chiral ] , that is not the case for @xmath234 in the chiral limit .
while the form factor is initially negative , as may be anticipated from fig.[fig : tpiggnmm ] , it is positive for @xmath235gev@xmath77 and the asymptotic behaviour indicated in eq.([uvchiralnot0 ] ) is exhibited for @xmath236gev@xmath77 . with the model s parameter value specified in eq.([omegavalue ] ) , we find @xmath237 this mass - scale is commensurate with that set by the vacuum quark condensate .
the magnitude of @xmath238 depends on the model parameter .
so , too , does the precise location of the boundary between the domains on which the transition form factor has negative and positive support .
however , qualitative features , such as the existence of these domains , are robust .
it is noteworthy that while @xmath239 algebraically in the chiral limit , in practice there is always a numerical error .
hence , as is plain from eq.([uvnot0 ] ) , there will inevitably be a value of @xmath7 beyond which the erroneous nonzero value of @xmath230 , produced by the numerical error , will come to dominate the chiral - limit transition form factor . to obtain the value in eq.([fnvalue ] ) we estimated the magnitude of this pollution and subtracted it .
for this reason , within the accuracy of our numerical analysis , we can not provide reliable information on the @xmath240-modification .
the figure hints , however , at the presence in our model of such a modification to the @xmath8-behaviour .
large-@xmath7 behaviour of the @xmath234 transition form factor , eq.([tpig ] ) : _ diamonds _ the result obtained with @xmath73 in eq.([qmass ] ) ; _ circles _ our chiral limit calculation ( @xmath74 ) ; _ solid line _ the curve @xmath241.,scaledwidth=48.0% ] at leading order in the truncation scheme we are using , and in the isospin symmetric limit , the elastic electromagnetic form factor of a pseudoscalar meson is described by @xmath242 with @xmath243 .
each element that appears in the integrand is fully renormalised and the integral is finite .
the expression automatically satisfies @xcite @xmath244 and guarantees @xmath245 in ref.@xcite the model described in sec.[sec : rl ] was employed to calculate the electromagnetic form factor of the pion using eq.([piem ] ) .
the prediction was subsequently verified in a jlab experiment performed at intermediate @xmath7 @xcite .
we have calculated the charge radii of the two lowest mass pseudoscalars using the standard definition : @xmath246 our results appear in fig.[fig : emradii ] .
as promised in association with eq.([omegavalue ] ) , the ground state s properties are almost insensitive to the model s mass - scale , @xmath191 : in formulating the model , a path appeared in the @xmath247 parameter space along which vacuum and ground state properties vary little .
the orthogonality of the excited states with respect to the ground state means there is no reason to expect such insensitivity in properties of the excited states . and , indeed
, one observes that the charge radius of the first excited state changes rapidly with increasing @xmath191 , with the ratio @xmath248 varying from @xmath249 @xmath250 .
evolution of ground and first excited state pseudoscalar mesons electromagnetic charge radii with the model s scale parameter @xmath191 .
_ dotted line _ : @xmath251fm , which indicates the experimental value of the ground state s radius . we must estimate the derivative in eq.([usualradius ] ) numerically .
that is the primary source of the numerical error depicted in the figure , which corresponds to a relative error @xmath252% for @xmath1 and @xmath253% for @xmath36.,scaledwidth=48.0% ] this outcome can readily be interpreted .
the length - scale @xmath254 measures the range of strong attraction in our model : magnifying @xmath255 increases the range of strong attraction . in sec.[sec : intro ] we argued that the properties of radial excitations should be sensitive to the nature of the interaction between light quarks at long - range .
it is now apparent that this is true .
moreover , decreasing @xmath191 has the effect of increasing the active range of the confining piece of the interaction in eq.([calg ] ) .
this effectively strengthens the confinement force .
that compresses the bound state , as one observes in fig.[fig : emradii ] : @xmath256 decreases rapidly with decreasing @xmath191 ( increasing @xmath255 ) .
a similar result for the evolution of the mass was observed in ref.@xcite ; namely , the mass of the first excited state dropped rapidly with increasing @xmath255 . on the domain illustrated in fig.[fig : emradii ] , the mass of the ground state obtained with nonzero current - quark mass varied by only 3% while that of the first excited state changed by 14% .
it is natural to expect that an increase in the strength of the confinement force should increase the magnitude of the binding energy and hence reduce the mass , and that is precisely what occurs .
independent of the parameters , the ground state mass is identically zero in the chiral limit because the truncation is symmetry preserving .
dynamical chiral symmetry breaking , which has many consequences , is another reason why properties of the ground state pseudoscalar meson do not respond rapidly to modest parameter changes . )
it is natural to suppose @xmath257 ; namely , that a radial excitation is larger than the associated ground state .
however , our calculations illustrate that with the ground state pseudoscalar meson s properties constrained by goldstone s theorem and its pointwise consequences , eqs.([bwti ] ) ( [ hwti ] ) , it is possible for a confining interaction to compress the excited state with the consequence that @xmath258 .
an analysis of the @xmath191-dependence of @xmath259 indicates that a value of @xmath260gev may be obtained with @xmath261 @xcite .
however , quantitative difficulties connected with the behaviour of the dressed - quark propagator in the complex-@xmath262 plane @xcite currently prevent us from studying the excited state directly with @xmath263 in eq.([calg ] ) .
hence , we can not make a firm prediction for @xmath256 .
however , our results suggest @xmath264 , with a linear extrapolation giving @xmath265 naturally , we have also studied the evolution of @xmath266 with @xmath191 . on the domain illustrated in fig.[fig : emradii ] , @xmath267 varies by no more than 1% , whereas @xmath268 and @xmath269 , which is a variation over a range of @xmath270% . following the reasoning above , and taking account of the variation in @xmath259
, we conclude that it is likely @xmath271ev @xmath272 .
our best estimate is @xmath273 and linear extrapolation gives @xmath274
the strong interaction spectrum exhibits trajectories of mesons with the same spin@xmath275parity , @xmath34
. one may distinguish between the states on these trajectories by introducing an integer label @xmath23 , with @xmath1 denoting the lowest - mass state , @xmath36 the next - lightest state , etc .
the bethe - salpeter equation ( bse ) yields the mass and amplitude of every bound state in a given channel specified by @xmath276 .
hence it provides a practical tool for the poincar covariant study of mesons on these trajectories . in applying the bethe - salpeter equation to a study of pseudoscalar mesons we made use of the fact that at least one nonperturbative and symmetry preserving dyson - schwinger equation ( dse ) truncation scheme exists
this fact supports a proof that , in the chiral limit , excited state @xmath92 mesons do not couple to the axial - vector current ; viz .
, @xmath277 @xmath278 .
we demonstrated that the leading - order ( rainbow - ladder ) term in the dse truncation scheme , when consistently implemented , is necessary and sufficient to express the abelian anomaly .
it can therefore be used to illustrate the anomaly s observable consequences .
we capitalised on this to show that even though excited state pseudoscalar mesons decouple from the axial - vector current in the chiral limit , they nevertheless couple to two photons .
( nb . the strength of this coupling is materially affected by the continuum contribution to the abelian anomaly . )
hence the primakov process , as employed for example in _ primex _ at jlab @xcite , may be used as a tool for their production and study .
a renormalisation - group - improved rainbow - ladder truncation is guaranteed to express the one - loop renormalisation group properties of qcd .
we exploited this and thereby determined the leading power - law behaviour of the @xmath3 transition form factor .
when the current - quark mass is nonzero then , for all @xmath23 , this form factor behaves as @xmath279 at deep spacelike momenta . for all but the goldstone mode this leading order contribution vanishes in the chiral limit . in that case , however , the form factor remains nonzero and the ultraviolet behaviour is @xmath280 .
although only exposed starkly in the chiral limit for excited states , this subleading power - law contribution to the @xmath3 transition form factor is always present and in general its coefficient is not simply related to @xmath10 .
as one might rationally expect , the properties of excited ( @xmath145 ) states are sensitive to the pointwise behaviour of what might be called the confinement potential between light - quarks .
we illustrated this by laying out the evolution of the charge radii of the @xmath281 pseudoscalar mesons .
as it is shielded by goldstone s theorem , the ground state s radius can be insensitive to details of the long - range part of the interaction .
however , that is not true of @xmath256 , the radius of the first excited state , which is orthogonal to the vacuum .
an increase in the length - scale that characterises the range of the confining potential reduces @xmath256 .
this result states that increasing the confinement force compresses the excited state : indeed , it is possible to obtain @xmath282 .
however , our current best estimate is @xmath283 .
a detailed exploration of the properties of collections of mesons on particular @xmath34 trajectories offers the hope of exposing features of the long - range part of the interaction between light - quarks . in principle , this interaction can be quite different to that between heavy - quarks .
the pseudoscalar trajectory is of particular interest because its lowest mass entry is qcd s goldstone mode .
chiral current conservation places constraints on some properties of every member of this trajectory , whose study may therefore provide information about the interplay between confinement and dynamical chiral symmetry breaking .
acknowledgments we acknowledge profitable interactions with s.j .
brodsky , r.j .
holt and p.c .
this work was supported by : austrian research foundation _
fwf , erwin - schrdinger - stipendium _ no .
j2233-n08 ; department of energy , office of nuclear physics , contract nos .
w-31 - 109-eng-38 and de - fg02 - 00er41135 ; national science foundation contract no
. int-0129236 ; the _ a.v .
humboldt - stiftung _ via a _
bessel forschungspreis _ ; and benefited from the facilities of the anl computing resource center and the nsf terascale computing system at the pittsburgh supercomputing center .
we observed in sec.[exactuv ] that since we are concerned with @xmath162 states it follows that @xmath284 , \label{tanswera}\end{aligned}\ ] ] where eq.([tpinggna ] ) yields @xmath285 with @xmath286 $ ] . in this appendix
we work always on the ultraviolet domain where @xmath287 and @xmath158 .
the leading contribution to @xmath167 is obtained from eq.([iresa ] ) . that is apparent because @xmath288 and this result inserted into eq.([iresa ] ) , with eq.([fpin ] ) used to identify the residue , yields @xmath289 equation ( [ tpiuvqcd ] ) follows immediately .
herein we also want the subleading contribution .
consider eq.([iresa ] ) with the explicit @xmath290 factor , which we have already used , removed from the numerator : @xmath291 this term s contribution to eq.([tanswera ] ) can be written @xmath292 with @xmath293 inspection and consideration of eqs.([definechi ] ) , ( [ sinvp ] ) , ( [ genpibsa ] ) reveals that one may write @xmath294 = 4\ , p^2\,{\cal x}_{p^2}(\ell^2,(\ell\cdot p)^2),\end{aligned}\ ] ] where @xmath295 ; i.e. , @xmath50 is an eigenvalue , not a variable , and the result is a function of @xmath296 because @xmath54 .
( nb . for the following argument
it is not necessary to make explicit the colour and flavour structure . )
it is further apparent from eqs.([definechi ] ) , ( [ sinvp ] ) , ( [ genpibsa ] ) that when chiral symmetry is dynamically broken @xmath297 and , moreover , that @xmath298 is a smooth function of its arguments so that it may be written @xmath299 in addition , @xmath300 up to @xmath301-corrections , where @xmath302 is a mass - scale that characterises the momentum - space width of the pseudoscalar meson s bethe - salpeter wave function and @xmath303 is this wave function s anomalous dimension .
one can now return to eq.([finlo ] ) and use the information provided to determine the dominant contribution @xmath304 the angular integration is straightforward : @xmath305 from which it follows that ( recall @xmath158 ) @xmath306 where we have used eq.([xuv ] ) and neglected the anomalous dimension .
( nb . an inspection of eq.([anglekernel ] ) will reveal that the contribution to the integral from @xmath307 is finite and , as @xmath308 , suppressed with respect to the term we have exposed . ) the true exponent and width in the logarithmic modification to the power - law behaviour will be affected by , e.g. , dressing of the quark - gluon vertex ; i.e. , corrections to eq.([expand ] ) , and diagrams beyond the rainbow - ladder truncation
. herein we are satisfied merely to establish that the subleading power - law behaviour is @xmath8 and that , in general , a @xmath309-modification may be present .
anisovich , v.v .
anisovich and a.v .
sarantsev , phys .
d * 62 * , 051502 ( 2000 ) .
a. tang and j.w .
norbury , phys .
d * 62 * , 016006 ( 2000 ) .
t. barnes , f.e .
close , p.r .
page and e.s .
swanson , phys .
d * 55 * , 4157 ( 1997 ) .
celenza , b. huang and c.m .
shakin , phys .
c * 59 * , 1041 ( 1999 ) [ erratum - ibid .
c * 60 * , 039901 ( 1999 ) ] .
d. arndt and c .- r .
ji , phys . rev .
d * 60 * , 094020 ( 1999 ) .
t. barnes , `` radial excitations , '' hep - ph/9907259 . m.k .
volkov and v.l .
yudichev , phys .
nucl . * 31 * , 282 ( 2000 ) [ fiz .
chast . atom .
yadra * 31 * , 576 ( 2000 ) ] r. ricken , m. koll , d. merten , b.c .
metsch and h.r .
petry , eur .
j. * a9 * , 221 ( 2000 ) .
bakker , `` bound states in light - front dynamics , '' in _ light - cone physics : hadrons and beyond _ , edited by s. dalley , pp . 15 - 22 .
a. krassnigg and c.d .
roberts , fizika * b13 * , 143 ( 2004 ) .
a. krassnigg and c.d .
roberts , nucl .
* a737 * , 7 ( 2004 ) .
a. hll , a. krassnigg and c.d .
roberts , `` dses and pseudoscalar mesons : an aperu , '' in _ light - cone physics : hadrons and beyond _ , edited by s. dalley , pp . 158 - 165 , nucl - th/0311033 .
a. hll , a. krassnigg and c.d .
roberts , phys .
c * 70 * , 042203(r ) ( 2004 ) e.s .
ackleh , t. barnes and e.s .
swanson , phys .
d * 54 * , 6811 ( 1996 ) j.c.r .
bloch , yu.l .
kalinovsky , c.d .
roberts and s. m. schmidt , phys .
d * 60 * , 111502 ( 1999 ) c.j .
burden and m.a .
pichowsky , few body syst . * 32 * , 119 ( 2002 ) .
d. jarecke , p. maris and p.c .
tandy , phys .
c * 67 * , 035202 ( 2003 ) ; p. watson , w. cassing and p.c .
tandy , few body syst . * 35 * , 129 ( 2004 ) . c.d .
roberts and s.m .
schmidt , prog .
nucl . phys .
* 45 * , s1 ( 2000 ) .
r. alkofer and l. von smekal , phys .
rept . * 353 * , 281 ( 2001 ) .
p. maris and c.d .
roberts , int .
j. mod .
* e12 * , 297 ( 2003 ) .
a. bender , c.d .
roberts and l. von smekal , phys .
* b380 * , 7 ( 1996 ) .
a. bender , w. detmold , c.d .
roberts and a.w .
thomas , phys . rev .
c * 65 * , 065203 ( 2002 ) .
bhagwat , a. hll , a. krassnigg , c.d .
roberts and p.c .
tandy , phys .
c * 70 * , 035205 ( 2004 ) .
hedditch , d.b .
leinweber , a.g .
williams and j.m .
zanotti , nucl .
phys . proc .
suppl . * 128 * , 221 ( 2004 ) .
we employ a euclidean metric , with : @xmath311 ; @xmath312 ; @xmath313 ; @xmath314 ; and @xmath315 . for a timelike vector @xmath316 , @xmath317 .
p.maris , c.d .
roberts and p.c .
tandy , phys .
* b420 * , 267 ( 1998 ) .
p. maris and c.d .
roberts , phys .
c * 56 * , 3369 ( 1997 ) .
k. langfeld , h. markum , r. pullirsch , c.d .
roberts and s.m .
schmidt , phys . rev .
c * 67 * , 065206 ( 2003 ) .
zhang , p.o .
bowman , r.j .
coad , u.m .
heller , d.b . leinweber and a.g .
williams , phys .
d * 71 * , 014501 ( 2005 ) .
bhagwat , m.a .
pichowsky , c.d .
roberts and p.c .
tandy , phys .
rev . * c68 * , 015203 ( 2003 ) .
r. alkofer , w. detmold , c.s .
fischer and p. maris , phys .
rev . * d70 * , 014014 ( 2004 ) .
m.s . bhagwat and p.c .
tandy , phys .
rev . * d70 * , 094039 ( 2004 ) .
we use landau gauge in all calculations because it is a fixed - point of the renormalisation group .
ivanov , y.l .
kalinovsky and c.d .
roberts , phys .
d * 60 * , 034018 ( 1999 ) .
bicudo , phys .
c * 67 * , 035201 ( 2003 ) .
bicudo , s. cotanch , f. llanes - estrada , p. maris , e. ribeiro and a. szczepaniak , phys .
d * 65 * , 076008 ( 2002 ) .
p. maris and c.d .
roberts , phys .
c * 58 * , 3659 ( 1998 ) .
p. maris and p.c .
tandy , phys .
c * 65 * , 045211 ( 2002 ) .
p. kessler and s. ong , phys .
d * 48 * , 2974 ( 1993 ) .
d. kekez and d. klabuar , phys .
* b457 * , 359 ( 1999 ) .
roberts , fizika * b8 * , 285 ( 1999 ) .
tandy , fizika * b8 * , 295 ( 1999 ) .
yeh , phys .
d * 66 * , 014002 ( 2002 ) .
munczek and a.m. nemirovsky , phys .
* d 28 * ( 1983 ) 181 .
d.w . mckay and h.j .
munczek , phys .
d * 55 * , 2455 ( 1997 ) .
richardson , phys .
* b82 * , 272 ( 1979 ) .
p. maris and p.c .
tandy , phys .
c * 60 * , 055214 ( 1999 ) .
a similar form was used fruitfully in : p. jain and h.j .
munczek , phys .
d * 44 * , 1873 ( 1991 ) ; h.j .
munczek and p. jain , _
ibid_. d * 46 * , 438 ( 1992 ) ; p. jain and h.j .
munczek , _
ibid_. d * 48 * , 5403 ( 1993 ) .
p. maris and p. c. tandy , phys .
c * 62 * , 055204 ( 2000 ) .
p. maris and p.c .
tandy , phys .
c * 61 * , 045202 ( 2000 ) .
roberts , nucl .
* a605 * , 475 ( 1996 ) . j. volmer _ et al . _ [ the jefferson lab @xmath318 collaboration ] , phys .
* 86 * , 1713 ( 2001 ) .
a. hll , a. krassnigg , c.d .
roberts and s.v .
wright , `` on the complexion of pseudoscalar mesons , '' nucl - th/0411065 .
bhagwat , m.a .
pichowsky and p.c .
ibid_. , 054019 ( 2003 ) .
details on the primakoff experiment ( primex ) at thomas jefferson national accelerator facility can be found at http://www.jlab.org/primex/ . | ' '' '' the axial - vector ward - takahashi identity places constraints on particular properties of every pseudoscalar meson .
for example , in the chiral limit all pseudoscalar mesons , except the goldstone mode , decouple from the axial - vector current .
nevertheless , all neutral pseudoscalar mesons couple to two photons . the strength of the @xmath0 coupling , where @xmath1 denotes the goldstone mode , is affected by the abelian anomaly s continuum contribution .
the effect is material for @xmath2 .
the @xmath3 transition form factor , @xmath4 , is nonzero @xmath5 , and @xmath6 at large @xmath7 .
for all pseudoscalars but the goldstone mode , this leading contribution vanishes in the chiral limit . in this instance
the ultraviolet power - law behaviour is @xmath8 for @xmath2 , and we find numerically @xmath9 .
this subleading power - law behaviour is always present . in general its coefficient
is not simply related to @xmath10 .
the properties of @xmath2 pseudoscalar mesons are sensitive to the pointwise behaviour of the long - range piece of the interaction between light - quarks . |
the experimental success in realizing a bose - einstein condensate ( bec ) in dilute atomic gases @xcite has led to the strong experimental and theoretical activity in this field .
dilute gases turn out to be an ideal system for realizing and high accuracy manipulation of various quantum many - body phenomena .
the nonlinear nature of a bec in the mean field description allows investigations of e.g. vortex lattices @xcite , four - wave mixing @xcite , josephson - like oscillations @xcite and solitons in the system .
the latter have been realized experimentally for a quasi one - dimensional ( 1d ) atomic bec with repulsive effective interactions ( so - called dark solitons ) @xcite and with attractive interactions ( bright solitons ) @xcite .
there are also proposals to prepare experimentally solitons in a mixture of bosonic and fermionic atoms @xcite and in an atomic bec with a time - dependent scattering length @xcite . in the present publication we consider another type of solitonic solutions .
that is , parametric solitons , which in the context of nonlinear optics occur as a result of a coupling between fields in a nonlinear medium such that each field propagates as solitons @xcite . in the coupled atomic - molecular bec
, there is a nonlinear resonant transfer between atoms and molecules , as well as terms proportional to the densities .
the solitons of this type have been investigated in nonlinear optics @xcite and in the problem of the self - localization of impurity atoms in a bec @xcite . in the context of the atomic - molecular becs
the parametric solitons have been analyzed in ref .
the latter publications consider , in principle , solitons in free space in any dimension but they concentrate on the existence and stability of the soliton like states in 3d free space .
however , a profound analysis of the system in a quasi 1d trap is missing and methods for experimental preparation and detection are not considered . trapping an atomic bec in a quasi 1d potential allowed realizing remarkable experiments where bright soliton wave packets travelled in 1d space ( still with transverse confinement ) without spreading @xcite . in the case of the coupled atomic - molecular becs ,
experimentalists have to face a problem of atomic losses , which becomes significant at a feshbach resonance .
we show that to overcome this obstacle one has to deal with moderate particle densities , and consider the range of the system parameters where preparation and detection of the solitons are sufficiently fast to be able to compete with the loss phenomenon .
moreover , the soliton states can be prepared at the magnetic field slightly off the resonance value , where the number of created molecules is still considerable but the rate of atomic losses much smaller than at the resonance .
the paper is organized as follows . in sec .
ii we introduce the model of the system and in sec .
iii we analyze different solitonic states .
section iv is devoted to analysis of methods for experimental preparation and detection .
in order to obtain a complete model of a feshbach resonance in an atomic bec , the intermediate bound states ( molecules ) have to be included explicitly in the hamiltonian . at the resonance ,
the number of molecules becomes considerable and a second ( molecular ) bec is formed .
we take into account two body atom - atom , molecule - molecule and atom - molecule collisions , as well as the term responsible for the creation of molecules , i.e. transfer of pairs of atoms into molecules and vice versa @xcite : & = & ^3 r ( ^_a _ a + & + & ^_m _ m + & + & _ am ^_a _ a ^_m _ m + ) , [ hfr ] where @xmath0 is the atomic mass , @xmath1 , @xmath2 , @xmath3 are coupling constants for the respective interactions , and @xmath4 determines the strength of the resonance .
the @xmath5 parameter is a difference between the bound state energy of two atoms and the energy of a free atom pair @xcite , and it can be varied by means of a magnetic field . @xmath6 and @xmath7 stand for atomic and molecular trapping potentials , respectively .
we would like to point out that in our model an internal structure of molecules and all processes involving the internal structure are neglected and we consider only a single molecular state described by the operator @xmath8 .
in the present paper we consider the system parameters mainly related to @xmath9rb atoms .
that is , the mass @xmath0 , the coupling constant @xmath1 and the strength of the resonance @xmath4 correspond to @xmath9rb atoms and the feshbach resonance that occurs at the magnetic field of @xmath10 g @xcite .
however , because the precise values of the coupling constants @xmath11 and @xmath3 are unknown these constants have to be chosen arbitrary . for simplicity
we assume @xmath12 , where @xmath13 nm is the value of the atomic scattering length far from a feshbach resonance . in sec .
iii we show that this assumption is not essential because in a broad range of @xmath11 and @xmath3 values there exist stable soliton solutions .
the resonance strength parameter @xmath4 can be expressed in terms of @xmath14 , the resonance width , and @xmath15 , difference between magnetic moments of a molecule and a free atom pair , @xmath16 .
we have chosen for investigation the broad resonance which occurs at the magnetic field @xmath10 g where @xmath17 g and @xmath18 ( @xmath19 is the bohr magneton ) @xcite .
we focus on the case where the system is prepared in a harmonic trap with so strong radial confinement that the radial frequency of the trap exceeds the chemical potential of the system .
then , only the ground states of the transverse degrees of freedom are relevant and the system becomes effectively 1d .
the chosen trap frequencies are @xmath20 hz and @xmath21 hz . ) .
different frequencies in the longitudal direction do not introduce any noticeable changes in shapes of solitons as far as the widths of the soliton wavepackets are much smaller than the characteristic length of the traps . ]
the effective 1d coupling constants are obtained by assuming that atoms and molecules are in the ground states of the 2d harmonic trap of frequencies @xmath22 and @xmath23 , and by integrating the energy density over transverse variables . in the harmonic oscillator units , e_0&=&_a , x , x_0&= & , _
0&= & , [ units ] where @xmath24 , @xmath25 and @xmath26 are energy , length and time units , respectively , one obtains _
a=_m=_am 0.505 , and 41.0 .
the equations of motion for the atomic and molecular mean fields , corresponding to the hamiltonian ( [ hfr ] ) , in the 1d model , in the units ( [ units ] ) , are i & = & _ a + & + & _ m ^*_a i & = & _ m + & + & ^2_a , [ ev ] where @xmath27 is the total number of atoms in the system @xcite . the 1d detuning @xmath28 is modified with respect to the corresponding @xmath5 in the 3d case , i.e. @xmath29 . the wave - functions @xmath30 and @xmath31 are normalized so that x(|_a(x)|^2 + 2|_m(x)|^2)=1 .
[ norm ] the eqs .
( [ ev ] ) look like a pair of coupled gross - pitaevskii equations but with an extra term responsible for the transfer of atoms into molecules and molecules into atoms .
the presence of this term allows for solutions that reveal parametric soliton like behavior .
the mean field model ( [ ev ] ) can be used if the condensates are nearly perfect .
that is , if the depletion effects are negligible which is usually true even for small particle number in a system @xcite .
the model neglects also effects of particle losses .
the latter can be described by introducing terms proportional to densities with imaginary coefficients ( that can be estimated provided there is precise experimental analysis of the losses in the vicinity of the feshbach resonance ) in the square brackets of eqs .
( [ ev ] ) @xcite . ) for @xmath32 ( a ) and @xmath33 ( b ) .
dashed lines are the analytic solutions ( [ sol2 ] ) corresponding to the case without the interaction between particles and in the absence of the trapping potentials .
dotted lines are related to the gaussian ansatz ( [ gau ] ) for @xmath32 the states are almost the same as the exact ones .
the atomic wavefunctions @xmath34 , whereas the molecular ones @xmath35 .
the ground states of eqs .
( [ sev ] ) correspond to the detuning chosen so that @xmath36 , i.e. @xmath37 ( a ) and @xmath38 ( b ) . note that the widths of the states are much smaller than the width of the ground state of the harmonic oscillator , which indicates that even for a very small particle number the nonlinearities in eqs .
( [ sev ] ) determine the shapes of the states .
, width=325 ]
the time - independent version of eqs .
( [ ev ] ) reads _ a & = & _ a + & + & _ m _ a 2_m&= & _ m + & + & ^2_a , [ sev ] where @xmath39 is the chemical potential of the system , and we have assumed that the solutions are chosen to be real . the ground state of the system described by eqs .
( [ sev ] ) can be found numerically .
however , in the absence of the trapping potentials and for @xmath40 , there exists an analytical solution @xcite if = - , where l = ( ) ^1/3 .
that is , _
a(x ) & = & , _
m(x ) & = & - , [ sol2 ] where a & = & , & = & - , describe the parametric bright soliton of width @xmath41 , which can propagate in time without changing its shape . the corresponding time - dependent solution is _
a(x , t ) & = & _ m(x , t ) & = & - , where @xmath42 is the propagation velocity .
the analytical solution is known for a particular value of the detuning @xmath28 but solutions revealing soliton like character exist also for other values . by changing @xmath28 we can obtain solitons but with unequal population of atomic and molecular condensates .
( a ) and @xmath43 ( b ) .
the atomic wave - functions @xmath30 are odd functions while the molecular wave - functions @xmath31 are the even ones . note that populations of atomic and molecular condensates are unequal for the value of the detuning chosen , i.e. @xmath44 .
note that the widths of the states in panel ( a ) are much smaller than the width of the first excited state of the harmonic oscillator , which indicates that even for a very small particle number the nonlinearities in eqs .
( [ sev ] ) determine the shapes of the states . ,
width=325 ] in the presence of the ( repulsive ) interactions , the solutions for the ground states reveal wave - packets that , for increasing total number of atoms @xmath27 in the system , become wider and wider .
indeed , for @xmath45 , @xmath46 and @xmath47 and with increasing @xmath27 , the repulsive interactions are stronger ( see fig . [ ground ] ) . unlike the repulsive nature of the interactions , the term responsible for the transfer between atoms and molecules acts as an attractive potential . for @xmath27 not too high , the transfer term dominates over the interaction terms and it alone determines the shape of the wave - packets . in fig .
[ ground ] we can see that the ground state solution for @xmath32 is only slightly wider than the analytical solution ( [ sol2 ] ) . in fig .
[ ground ] we also present the results of variational analysis of ref .
@xcite ( including also nonzero coupling constants @xmath2 and @xmath3 ) where the solutions have been assumed to be given by gaussian functions , _ a(x)=ae^-ax^2 , & & _ b(x)=be^-bx^2 . [ gau ]
requiring the same fraction of atoms and molecules , for @xmath32 ( @xmath43 ) , we obtain @xmath48 , @xmath49 , @xmath50 , @xmath51 , @xmath52 ( @xmath53 , @xmath54 , @xmath55 , @xmath56 , @xmath57 ) which fits quite well to the exact solutions . even in the presence of the interactions the obtained solutions
do not loose their solitonic character , which can be verified by the time evolution of the states in 1d free space ( i.e. in the presence of the transverse confinement but with the axial trap turned off ) see sec .
iv for examples of the time evolution .
the existence of the solitons can be also verified employing the gaussian ansatz ( [ gau ] ) . indeed , in the absence of the axial traps we have checked that , for @xmath43 , @xmath58 and @xmath59 , the solutions exist with @xmath60 and @xmath61 in the range ( 1,20000 ) @xcite .
it shows also that the choice of equal coupling constants is not essential in order to deal with soliton like solutions .
all stationary states of the gross - pitaevskii equations ( [ sev ] ) have been obtained numerically by means of the so - called imaginary time evolution which implies they are stable solutions of these non - linear coupled schrdinger equations . in the absence of the trapping potential and for
@xmath62 one can find an asymptotic solution of eqs .
( [ sev ] ) that reveals two bright soliton structure , _
a(x ) & = & _
m(x ) & = & -- , valid for @xmath63 , where a & = & , & = & - , l & = & ( ) ^1/3&= & - . in the presence of the traps and the interactions we may look for similar solutions .
analysis of eqs .
( [ sev ] ) shows that due to the fact that the trapping potentials are even functions we may expect that eigenstate functions are either even or odd .
however , because of the presence of the @xmath64 term in the second of eqs .
( [ sev ] ) , an even function is the only possibility for @xmath65 . in fig .
[ 2sol ] there are examples of the two bright soliton solutions for @xmath32 and @xmath43 . ,
compared to the solution within the thomas - fermi approximation ( dashed lines ) .
panel ( a ) shows the atomic wavefunctions while panel ( b ) the molecular ones .
the insets show the same as the main panels but with plots enlarged around the trap center .
, width=325 ] with an increasing number of particles the shapes of the density profiles change . for large particle numbers
the healing length of the system becomes much smaller than the size of the system and one obtains density shapes that resemble the dark soliton state known for the gross - pitaevskii equation in the case of repulsive interparticle interactions @xcite the results for @xmath66 are shown in fig .
[ darksol ] . here
, however , only the atomic wave - function @xmath30 exhibits a phase flip .
we can approximate the exact eigenstates by ( see fig .
[ darksol ] ) _
a(x)&&^tf_a(x ) ( ) , _ m(x)&&^tf_m(x)|()| , where = , and ^tf_m(x)&= & - , ^tf_a(x)&= & , & = & ( - ) , are solutions for the ground state of the mean field equations within the thomas - fermi approximation , and the chemical potential @xmath39 can be found from the normalization condition ( [ norm ] ) .
finally we would like to mention that in the case of the harmonic trap , having any solutions @xmath67 and @xmath68 of the time - independent problem , eqs .
( [ sev ] ) , corresponding to the chemical potential @xmath39 , one can easily obtain time evolution of the initial wave - functions @xmath69 and @xmath70 , i.e. _ a(x , t)&=&_a0(x - q)e^-ite^i[q x - s(q ) ] , _
m(x , t)&=&_m0(x - q)e^-i2te^2i[q x - s(q ) ] , [ gal ] where s(q)=12 _
t_0^t dt , and + q=0 .
the proof can be done by direct substitution of ( [ gal ] ) into ( [ ev ] ) .
this indicates that , similarly as in the case of the gross - pitaevskii equation for a single condensate in a harmonic trap , time evolution of the translated stationary solutions reveals harmonic oscillations of the center of mass of the particle cloud .
to create experimentally a bright soliton state we propose to start with a purely atomic condensate in a quasi-1d trap with the magnetic field significantly higher than the resonance value . then the field should be slowly decreased until it reaches the resonance . actually , from the experimental point of view , it is better to end up slightly off the resonance because the rate of atomic losses is then much smaller than the rate at the resonance .
the final populations of atomic and molecular condensates are unequal in such case but they still reveal solitonic character . in the following we analyze different velocities of the sweeping of the field in order to find the optimal shape of the state .
( a ) and @xmath43 ( b ) .
the sweeping lasted @xmath71 ms ( a ) and @xmath72 ms ( b ) .
dashed lines show the exact ground states corresponding to the magnetic field at the end of the sweeping .
the detuning @xmath44 at the end of the evolution .
top curves , in each panel , are related to densities of atoms and bottom ones to densities of molecules.,width=302 ] we have done numerical calculations for @xmath27 in the range between 100 and 10000 .
assuming that the initial state is an almost purely atomic bec , i.e. the ground state of the system far from the resonance , the detuning @xmath28 has been varied linearly in time from 3331 to 0 ( @xmath73 corresponds to @xmath74 g ) . in order to obtain a nice solitonic state
, the evolution time has to be carefully chosen .
we have found that the square overlap between a final state and a desired eigenstate oscillates as a function of the evolution time .
therefore one has to find a proper value of the evolution time , corresponding to a maximum value of the square overlap . from the experimental point of view
, the shorter the evolution time the better because that allows one to reduce particle losses significant close to the resonance . in fig .
[ evol ] there are results for @xmath32 and @xmath43 corresponding to the optimized evolution time and compared with the ground states at the resonance .
one can see that despite quite short evolution times ( @xmath71 ms and @xmath75 ms for @xmath32 and @xmath43 , respectively ) , the final states reproduce the exact ground states quite well , i.e. the square overlaps are @xmath76 and @xmath77 . for @xmath78
the shortest evolution time that results in a reasonably high squared overlap ( i.e. @xmath79 ) is @xmath80 ms .
note that the population of the molecular bec is smaller than the population of the atomic bec because the final value of the detuning corresponds to the magnetic field slightly off the resonance . using the procedure described above it is possible not only to create a single bright soliton but also excite states that reveal multi - peak structure ( so - called `` soliton trains '' @xcite )
actually , deviations from the optimal sweeping time of the magnetic field result in multi - peak structure of the atomic and molecular densities .
an example of @xmath43 , where the soliton train is particularly well reproduced , is shown in the fig .
[ train1000 ] .
the calculations have been performed starting with the same initial state and the magnetic field value as in fig .
[ evol ] , but the sweeping time is now @xmath81 ms . after the sweeping of the magnetic field that lasted @xmath81 ms .
solid line indicates the atomic density while the dashed line the molecular density . , width=325 ] [ cols="^ " , ] in ref .
@xcite detection of time evolution of wave - packets without spreading in the presence of the transverse confinement but with the axial trap turned off ( or even in the presence of an inverted axial potential ) was an experimental signature of the bright soliton excitation in an atomic bec with attractive atom - atom interaction . in the present case
we can use the same test to check if the states obtained reveal stable soliton like character .
it turns out that indeed all the states presented do not spread when the axial trap is turned off during time evolution lasting even @xmath82 ms .
we do not attach the corresponding figures because they would show practically the same density profiles as the densities of the initial states . in ref .
@xcite the results of experimental creation of molecules from an atomic bec prepared not in a quasi 1d potential but in a 3d trap are presented . after producing the molecules ,
the experimentalists analyze the evolution of the system with the traps turned off but in the presence of an inhomogeneous magnetic field . they observe separation of the molecular cloud from the atomic one , which is due to the difference between atomic and molecular magnetic moments .
we will analyze the evolution of the atomic and molecular clouds in an inhomogeneous magnetic field but in the presence of the transverse confinement .
one may expect that when the solitonic states are created there is strong attractive coupling between atoms and molecules that will compete with the opposite effect resulting from the difference of the magnetic moments . in the magnetic field gradient , eqs .
( [ ev ] ) are modified . instead of the detuning @xmath28 there are terms proportional to the magnetic moments : @xmath83 in the first equation , and @xmath84 in the second one .
@xmath85 is a value of the magnetic field gradient and @xmath25 and @xmath24 are length and energy units ( [ units ] ) . by applying suitable unitary transformations ,
however , these terms can be eliminated from the first equation .
first , the coordinate transformation @xmath86 is applied , then the phases are adjusted according to : _ a & & _ a e^ ( ^2_a - ) t + _ m & & _
m e^i(^2_a - ) t , where @xmath87 .
finally , there are only two additional terms in ( [ ev ] ) ( both present in the second of eqs .
( [ ev ] ) ) : x & = & b_grad x , and a detuning : _ grad = - ( b_grad ) ^2 _ a , [ dtng_grad ] modified with respect to the case without gradient ( i.e. with only @xmath28 ) .
the behavior of the system in the presence of an inhomogeneous magnetic field depends on the value of the detuning @xmath88 . for @xmath89
the clouds do not separate but there is an efficient transfer of atoms into molecules . in fig .
[ gradtr ] we see that for @xmath43 , @xmath90 g / cm and @xmath89 , the number of molecules is increased after @xmath91 ms of time evolution . for @xmath92 ( corresponding to @xmath93 ) and for small particle number ( e.g. @xmath32 ) there is an opposite transfer , that is of molecules into atoms ( see fig .
[ gradn ] ) , and when molecules are absent the atomic wavepacket starts spreading . however , if @xmath27 is greater ( @xmath43 or more ) the wavepackets begin to split into several separated peaks ( see fig .
[ gradn1000 ] ) but there is still no separation between atomic and molecular clouds .
similar splitting of solitonic wavepackets has been analyzed in the case of a single component bec in the presence of a gravitational field @xcite . for greater field gradient than the value chosen in figs .
[ gradtr]-[gradn1000 ] or slightly different values of the coupling constants we observe qualitatively similar behaviour .
the longer creation and detection of the solitons lasts , the more serious becomes the problem of atomic losses in the vicinity of the feshbach resonance .
measurements of the losses for different feshbach resonances in @xmath9rb were done in a 3d trap ( frequencies : @xmath94 hz , @xmath95 hz , @xmath96 hz ) with an initial total number of atoms @xmath97 ( that corresponds to the density at the center of the atomic condensate of order of @xmath98 @xmath99 ) @xcite .
the observed loss rate , defined as a fraction of atoms lost during @xmath100 ms hold time in the trap , was @xmath101 for the resonance at @xmath10 g. in the present publication we have chosen very high frequency of the transverse traps ( i.e. @xmath102 hz ) in order to ensure that the 1d approximation is valid for all values of particle numbers considered .
it results in the peak density for solitonic states comparable to the experimental density in @xcite .
however , the density can be significantly reduced if one chooses weaker transverse confinement and a moderate particle number .
for example , for @xmath43 and for the transverse traps of @xmath103 hz , the peak soliton density is @xmath104 @xmath99 and the evolution time needed to obtain the final state with @xmath105 squared overlap with the desired eigenstate is @xmath106 ms .
the losses will be smaller if the magnetic field at the end of the sweeping corresponds to a value slightly above the resonance , where the solitons still form but the loss rate is smaller .
actually , solitons can be prepared and detected in the presence of the losses provided that the escaping atoms do not significantly excite the system of the remaining particles .
indeed , an experimental signature of the soliton excitation could be the time evolution of the system with the axial trap turned off , where the wavepackets are expected not to increase their widths .
importantly , when @xmath27 decreases , the influence of the atom - molecule transfer term ( which is responsible for the existence of the solitonic solutions ) on the character of the solutions grows , as compared to the influence of the interaction terms .
this is because the former depends on the particle number as @xmath107 while the latter as @xmath27 , see eq .
( [ ev ] ) .
consequently the normalized particle density should not increase its width even in the presence of the losses if a soliton state is prepared in an experiment .
we consider solitonic behaviour of coupled atomic - molecular bose - einstein condensates trapped in a quasi-1d potential . in the vicinity of a feshbach resonance
there are eigenstates of the system that reveal soliton - like character .
the profiles of the solitons depend on the total number @xmath27 of the particles in the system and on the values of the scattering lengths characterizing particle interactions . for moderate particle numbers and positive scattering lengths one can find two bright soliton solutions which for large @xmath27 turn into a state that resembles a dark soliton profile known for atomic condensates with the repulsive atom - atom interaction @xcite .
we analyze methods for experimental preparation and detection of the solitons in the atomic - molecular becs .
they can be obtained experimentally starting from a purely atomic bec and slowly decreasing the magnetic field down to a value slightly above the resonance .
evolution times necessary to create solitons are quite short which is very promising experimentally because it allows reducing atomic losses important in the vicinity of the resonance . to detect the solitons one can perform ( similarly as in refs .
@xcite ) the evolution of the system in the absence of the axial trap ( but still in the presence of the transverse confinement ) where the wave - packets are expected to propagate without spreading .
we have also considered creation of the bright soliton trains in a coupled atomic - molecular condensate .
soliton trains in bose - fermi mixtures obtained by temporal variations of the interactions between the two components has been recently proposed @xcite .
our system is different but its qualitative behaviour is to some extent similar . in both cases ,
there is a term in the hamiltonian that leads to an effective attractive interactions .
this term can dominate over repulsive interactions and stabilize bright solitons .
we are grateful to jacek dziarmaga and kuba zakrzewski for helpful discussions .
the work of bo was supported by polish government scientific funds ( 2005 - 2008 ) as a research project .
ks was supported by the kbn grant pbz - min-008/p03/2003 .
supports by aleksander von humboldt foundation and within marie curie tok project cocos ( mtkd - ct-2004 - 517186 ) are also gratefully acknowledged .
part of numerical calculations was done in the interdisciplinary centre for mathematical and computational modeling warsaw ( icm ) .
m. h. anderson , j. r. ensher , m. r. matthews , c. e. wieman , and e. a. cornell , science * 269 * , 198 ( 1995 ) . k. b. davis , m. -o .
mewes , m. r. andrews , n. j. van druten , d. s. durfee , d. m. kurn , and w. ketterle phys .
* 75 * , 3969 ( 1995 ) c. c. bradley , c. a. sackett , and r. g. hulet , phys .
* 78 * , 985 ( 1997 ) .
m. r. matthews , b. p. anderson , p. c. haljan , d. s. hall , c. e. wieman , and e. a. cornell phys .
lett . * 83 * , 2498 ( 1999 ) .
j. r. abo - shaeer , c. raman , j. m. vogels , and w. ketterle , science * 292 * , 476 ( 2001 ) .
f. dalfovo and s. stringari , phys .
a * 53 * , 2477 ( 1996 ) .
l. deng , e. w. hagley , j. wen , m. trippenbach , y. band , p. s. julienne , j. e. simsarian , k. helmerson , s. l. rolston , and w. d. phillips , nature * 398 * , 218 ( 1999 ) .
m. trippenbach , y. b. band , and p. s. julienne , phys .
a * 62 * , 023608 ( 2000 ) .
a. smerzi , s. fantoni , s. giovanazzi , and s. r. shenoy , phys .
lett . * 79 * , 4950 ( 1997 ) .
m. albiez , r. gati , j. flling , s. hunsmann , m. cristiani , and m. k. oberthaler , phys .
lett . * 95 * , 010402 ( 2005 ) . s. burger , k. bongs , s. dettmer , w. ertmer , k. sengstock , a. sanpera , g. v. shlyapnikov , and m. lewenstein , phys . rev
. lett . * 83 * , 5198 ( 1999 ) .
j. denschlag , j. e. simsarian , d. l. feder , c. w. clark , l. a. collins , j. cubizolles , l. deng , e. w. hagley , k. kelmerson , w. p. reinhardt , s. l. rolston , b. i. schneider , and w. d. phillips , science * 287 * , 97 ( 2000 ) .
l. khaykovich , f. schreck , g. ferrari , t. bourdel , j. cubizolles , l. d. carr , y. castin , and c. salomon , science * 296 * , 1290 ( 2002 ) .
k. s. strecker , g. b. partridge , a. g. truscott , and r. g. hulet , nature * 417 * , 150 ( 2002 ) .
p. g. kevrekidis , g. theocharis , d. j. frantzeskakis , and b. a. malomed , phys .
* 90 * , 230401 ( 2003 ) .
abdullaev , a. m. kamchatnov , v. v. konotop , and v. a. brazhnyi , phys .
lett . * 90 * , 230402 ( 2003 ) .
d. e. pelinovsky , p. g. kevrekidis , d. j. frantzeskakis , phys .
lett . * 91 * , 240201 ( 2003 ) . z. x. liang , z. d. zhang , and w. m. liu , phys . rev
94 * , 050402 ( 2005 ) .
m. matuszewski , e. infeld , b. a. malomed , and m. trippenbach , phys .
* 95 * , 050403 ( 2005 ) .
kivshar and g.p .
agrawal , _ optical solitons : from fibers to photonic crystals _ ( academic press 2003 )
. m. a. karpierz , opt . lett . * 16 * , 1677 ( 1995 ) .
h. he , m. j. werner , and p. d. drummond , phys .
e * 54 * , 896 ( 1996 ) .
l. berg , o. bang , j. juul rasmussen , and v. k. mezentsev , phys .
e * 55 * , 3555 ( 1997 ) . | we consider coupled atomic - molecular bose - einstein condensate system in a quasi - one - dimensional trap . in the vicinity of a feshbach resonance the system can reveal parametric soliton - like behavior .
we analyze bright soliton , soliton train and dark soliton solutions for the system in the trap and in the presence of the interactions between particles and find the range of the system parameters where the soliton states can be experimentally prepared and detected . |
in 1781 , gaspard monge @xcite raised the following problem . given two sets @xmath8 and @xmath0 of equal mass in @xmath2 , transport each mass unit of @xmath8 to a mass unit of @xmath0 at minimal cost .
more formally , given two measures @xmath9 and @xmath12 , find a map @xmath13 satisfying @xmath14 that minimizes @xmath15 where @xmath16 describes the cost of moving @xmath7 to @xmath17 . because of its obvious relevance to economics , and perhaps due to its mathematical challenge , this problem has received a lot of attention .
even with the euclidean distance as cost function @xmath18 it is not at all clear in which cases an optimal map @xmath13 exists .
progress by appell @xcite was honored with a prize by the academy of paris in 1887 .
kantorovich @xcite achieved a breakthrough by solving a relaxed version of monge s original problem . in 1975
, he received a nobel prize in economics ; see gangbo and mccann @xcite for mathematical and historical details . while usually known as the _ monge - kantorovich transportation problem _ , the minimum cost of a transportation is sometimes called _
wasserstein metric _ or , in computer science , _ earth mover s distance _ between the two measures @xmath9 and @xmath12 .
it can be used to measure similarity in image retrieval ; see rubner et al . @xcite .
if both measures @xmath9 and @xmath12 have finite support , monge s problem becomes the minimum weight matching problem for complete bipartite graphs , where edge weights represent transportation cost ; see rote @xcite , vaidya @xcite and sharathkumar et al .
@xcite .
we are interested in the case where only measure @xmath12 has finite support .
more precisely , we assume that a set @xmath0 of @xmath1 point sites @xmath19 is given , and numbers @xmath20 whose sum equals 1 .
a body @xmath8 of volume 1 must be split into subsets @xmath21 of volume @xmath22 in such a way that the total cost of transporting , for each @xmath23 , all points of @xmath21 to their site @xmath19 becomes a minimum . in this
setting , volume is measured by some measure @xmath9 , and transport cost @xmath16 by some measure of distance .
gangbo and mccann @xcite report on the cases where either @xmath24 with a strictly convex function @xmath25 , or @xmath26 with a non - negative , strictly concave function @xmath27 of the euclidean norm . as a consequence of deep results on the general monge - kantorovich problem
, they prove a surprising fact .
the minimum cost partition of @xmath8 is given by the additively weighted voronoi diagram of the sites @xmath19 , based on cost function @xmath18 and additive weights @xmath28 .
in this structure , the voronoi region of @xmath19 contains all points @xmath7 satisfying @xmath29 villani @xcite has more recently observed that this holds even for more general cost functions , that need not be invariant under translations , and in the case where both distributions are continuous .
figure [ cones - fig ] depicts how to obtain this structure in dimension @xmath30 . for each point site
@xmath31 we construct in @xmath32 the cone @xmath33 .
the lower envelope of the cones , projected onto the @xmath34plane , results in the voronoi diagram . independently , aurenhammer et al .
@xcite have studied the case where @xmath35 . here , a _ power diagram _ ( see @xcite ) gives an optimum splitting of @xmath8 .
in addition to structural results , they provide algorithms for computing the weights @xmath28 . regarding the case where the transportation cost from point @xmath7 to site @xmath19 is given by individual cost functions @xmath36 , it is still unknown how to compute the weights @xmath28 . in this paper
we consider the situation where the cost @xmath37 of transporting point @xmath7 to site @xmath19 is given by individual distance functions @xmath36 .
we require that the weighted voronoi diagram based on the functions @xmath38 is well - behaved , in the following sense .
bisectors are @xmath39dimensional , and increasing weight @xmath28 causes the bisectors of @xmath19 to sweep @xmath18space in a continuous way .
these requirements are fulfilled for the euclidean metric and , at least in dimension 2 , if the distance function @xmath40 assigned to each site @xmath5 is a translate of the same strictly convex distance function @xmath41 . we show that these assumptions are strong enough to obtain the result of @xcite : the weighted voronoi diagram based on the functions @xmath38 optimally solves the transportation problem , if weights are suitably chosen .
whereas @xcite derives this fact from a more general measure - theoretic theorem , our proof uses the minimization of a quadratic objective function and geometric arguments .
the purpose of our paper is to show that such a simple proof is possible . after stating some definitions in section [ defi - sec ] we generalize arguments from @xcite to prove , in section [ parti - sec ] , that @xmath8 can be split into parts of arbitrary given volumes by a weighted voronoi diagram , for a suitable choice of weights .
then , in section [ opti - sec ] , we show that such a partition is optimal , and unique . in section
[ unique - sec ] we show that if @xmath8 is connected , these weights are uniquely determined , up to addition of a constant .
let @xmath9 be a measure defined for all lebesgue - measurable subsets of @xmath2 .
we assume that @xmath9 and the @xmath18-dimensional lebesgue measure are mutually continuous , that is , @xmath9 vanishes exactly on the sets of lebesgue measure zero .
let @xmath0 denote a set of @xmath1 point sites in @xmath2 . for each @xmath4
we are given a continuous function @xmath42 that assigns to each point @xmath7 of @xmath2 a nonnegative value @xmath10 as the `` distance '' from site @xmath5 to @xmath7 .
for @xmath43 and @xmath44 we define the _ bisector _ @xmath45 and the _ region _ @xmath46 the sets @xmath47 are open and increase with @xmath48 .
now let us assume that for each site @xmath4 an additive weight @xmath49 is given , and let @xmath50 denote the vector of all weights , according to some ordering @xmath51 of @xmath0 .
then , @xmath52 is called the _ additively weighted bisector _ of @xmath19 and @xmath53 , and @xmath54 is the _ additively weighted voronoi region _ of @xmath19 with respect to @xmath0 .
it consists of all points @xmath7 for which @xmath55 is smaller than all values @xmath56 , by definition . as usual , @xmath57 is called the _ additively weighted voronoi diagram _ of @xmath0 ; see @xcite . clearly , @xmath58 and @xmath59 do not change if the same constant is added to all weights in @xmath60 . increasing
a single value @xmath28 will increase the size of @xmath19 s voronoi region . for increasing values of @xmath48 .
( ii ) an additively weighted voronoi diagram . ] * example .
* let @xmath30 and @xmath61 the euclidean distance .
given weights @xmath62 , the bisector @xmath63 is a line if @xmath64 , and a branch of a hyperbola otherwise .
figure [ hyperbolae - fig](i ) shows how @xmath65 sweeps across the plane as @xmath48 grows from @xmath66 to @xmath67 .
in fact , when @xmath68 , the bisector degenerates to a ray , and for larger @xmath69 , the bisectors are empty .
the bisector @xmath65 forms the boundary of the set @xmath70 , which increases with @xmath48 .
each bounded set @xmath8 in the plane will be reached at some point . from then on the volume of @xmath71 is continuously growing until all of @xmath8 is contained in @xmath70 .
given @xmath1 points @xmath53 with additive weights @xmath72 , raising the value of a single weight @xmath28 will cause all sets @xmath73 to grow until @xmath8 is fully contained in the voronoi region @xmath74 of @xmath19 .
figure [ hyperbolae - fig](ii ) shows an additively weighted voronoi diagram @xmath59 based on the euclidean distance .
it partitions the plane into 5 two - dimensional cells ( voronoi regions ) , and consists of 9 cells of dimension 1 ( voronoi edges ) and of 5 cells of dimension 0 ( voronoi vertices ) .
each cell is homeomorphic to an open sphere of appropriate dimension .
the next definition generalizes the above properties to the setting used in this paper .
[ admi - defi ] a system of continuous distance functions @xmath11 , where @xmath4 , is called _ admissible _ if for all @xmath75 , and for each bounded open set @xmath3 , there exist values @xmath76 and @xmath77 such that @xmath78 is continuously increasing from @xmath79 to @xmath80 as @xmath48 grows from @xmath76 to @xmath77 .
furthermore , @xmath81 if @xmath82 and @xmath83 if @xmath84 .
by symmetry we can assume w. l. o. g. @xmath85 and @xmath86 ; see figure [ hyperbolae - fig](i )
. we will need the following structural property , which essentially says that bisectors have measure 0 .
[ bisector - volume - lemm ] for a system of admissible distance functions @xmath11 , where @xmath31 , for any two points @xmath87 and any @xmath44 , and for any bounded open set @xmath88 , we have @xmath89 . by definition of @xmath70 and @xmath65 , for all @xmath90 and @xmath91 the following inequality holds : @xmath92 we let @xmath93 decrease to 0 . since , according to definition [ admi - defi ] , the function @xmath94 is continuous , we get @xmath95 this implies @xmath96
the following theorem shows that we can use an additively weighted voronoi diagram based on an admissible system of distance functions @xmath97 to partition @xmath8 into subsets of prescribed sizes .
[ partition - theo ] let @xmath98 and let @xmath38 , @xmath99 , be an admissible system as in definition [ admi - defi ] .
let @xmath8 be a bounded open subset of @xmath2 .
suppose we are given @xmath1 real numbers @xmath100 with @xmath101 .
then there is a weight vector @xmath102 such that @xmath103 holds for @xmath99 . if , moreover , @xmath8 is pathwise connected then @xmath60 is unique up to addition of a constant to all @xmath28 , and the parts @xmath104 of this partition are unique .
the function @xmath105 measures how far @xmath60 is from fulfilling the theorem . since each function @xmath106 is continuous by definition [ admi - defi ] , and because @xmath107 we conclude that the function @xmath108 is continuous , too .
let @xmath109 with @xmath77 as in definition [ admi - defi ] .
note that @xmath110 and @xmath111 together imply that @xmath112 is a positive constant .
on the compact set @xmath113^n$ ] , the function @xmath108 attains its minimum value at some argument @xmath60 . if @xmath114 we are done .
suppose that @xmath115 . since the volumes of the voronoi regions inside @xmath8 add up to @xmath80 , there must be some sites @xmath53 whose voronoi regions have too large an intersection with @xmath8 i.e. , of volume @xmath116 , while other region s intersections with @xmath8 are too small .
we now show that by decreasing the weights of some points of @xmath0 we can decrease the value of @xmath108 . for any point @xmath117
we consider the continuous `` excess '' function @xmath118 that indicates by which amount of volume the region of @xmath19 is too large or too small .
we have @xmath119 let @xmath120 be the set of points @xmath19 for which @xmath121 attains its maximum value , @xmath122 . by assumption ,
@xmath123 is neither empty nor equal to @xmath0 .
we will reduce the weights of the points in @xmath123 by the same amount @xmath124 , obtaining a new weight vector @xmath125 . note that , by construction , @xmath126 for any site @xmath127 , and @xmath128 , for any site @xmath129 .
let @xmath130 denote the weight vector @xmath131 where @xmath132 if @xmath133 , and @xmath134 otherwise .
the volume of the set of points of @xmath8 , that is assigned by @xmath135 to some point in @xmath123 and by @xmath136 to some point in @xmath137 is @xmath138 let @xmath139 and @xmath140 .
our goal is to choose @xmath124 in such a way that @xmath141 and to show that this choice decreases @xmath108 strictly .
since @xmath142 and @xmath143 is a monotone and continuous function , our aim is to solve by the intermediate value theorem . for all @xmath144
we observe that @xmath145 holds , since in this case the voronoi region @xmath146 of every point @xmath133 contains no point of @xmath8 .
now @xmath147 implies that we can indeed solve by the intermediate value theorem .
the latter inequality is evident if @xmath148 .
otherwise , it follows from @xmath149 . in the following
let @xmath150 , for short .
in order to prove the theorem we use the following reformulation of the statement @xmath151 : @xmath152 now , in order to prove inequality @xmath153 , we give an upper bound on the right - hand side of the inequality , which will be compared with a lower bound on the left - hand side of the inequality .
let @xmath154 denote the volume increase , in @xmath8 , of the region of site @xmath155 . by the choice of @xmath124
, we have @xmath156 , and @xmath157 we now give a lower bound on the left - hand side of @xmath153 . if @xmath158 denotes the volume decrease , in @xmath8 , of the region of site @xmath127 , then the corresponding calculation for the points in @xmath123 yields : @xmath159 where we have used the same arguments as above and also that @xmath160 , by definition , for all @xmath127 .
it is easy to verify the inequality @xmath161 for @xmath162 and @xmath98 .
hence inequality @xmath153 holds .
we have now shown the existence of a weight vector @xmath125 with @xmath163 . if @xmath125 belongs to @xmath113^n$ ] , we are done .
otherwise , suppose the maximum weight @xmath164 occurs for some point @xmath165 .
we add @xmath166 to all weights .
this does not change the voronoi diagram defined by @xmath125 , but now the maximum weight @xmath164 is exactly @xmath167 .
next we observe that if now the weight of some point @xmath53 is @xmath168 , then its voronoi region is empty , because @xmath8 is contained in the set @xmath169 , by definition of @xmath167 .
assigning weight @xmath79 to each of those points also results in empty voronoi regions for those points .
altogether , we now have @xmath170 for all @xmath117 .
we have shown the existence of a weight vector @xmath171^n$ ] that satisfies @xmath163 .
this contradicts the minimality of @xmath172 .
hence the minimum value of @xmath108 is @xmath79 .
uniqueness of the partition will be discussed in section [ unique - sec ] .
again , let @xmath173 , @xmath174 , be an admissible system as in definition [ admi - defi ] , and let @xmath8 denote a bounded and open subset of @xmath2 .
moreover , we are given real numbers @xmath100 whose sum equals @xmath80 . by theorem [ partition - theo ] there exists a weight vector @xmath102 satisfying @xmath175 now we prove that this subdivision of @xmath8 minimizes the transportation cost , i. e. , the average distance from each point to its site .
it is convenient to describe partitions of @xmath8 by @xmath9-measurable maps @xmath176 where , for each @xmath4 , @xmath177 denotes the region assigned to @xmath5 .
let @xmath178 denote the set of those maps @xmath13 satisfying @xmath179 for @xmath180 .
[ opti - theo ] the partition of @xmath8 into regions @xmath181 minimizes @xmath182 over all maps @xmath183 .
any other partition of minimal cost differs at most by sets of measure zero from @xmath184 . if the measure @xmath9 were a discrete measure concentrated on a finite set @xmath8 of points , the optimum partition problem would turn into the classical transportation problem , a special type of network flow problem on a bipartite graph .
the weights @xmath28 would be obtained as dual variables . the following proof mimics the optimality proof for the discrete case .
let @xmath185 be defined by @xmath186 for all @xmath23 , and @xmath187 be another map .
if we define @xmath188 , then the cost of map @xmath13 is @xmath189 where the inequality between lines @xmath190 and @xmath191 follows from the fact that @xmath192 by the definition of additively weighted voronoi diagrams .
here we have assumed for simplicity that the map @xmath185 assigns every point @xmath193 to _ some _ weighted nearest neighbor of @xmath5 in @xmath0 , according to some tie - break rule . since lemma [ bisector - volume - lemm ] implies @xmath194 , changing the assignment of those points has no influence on @xmath195 . both maps @xmath13 and @xmath185 partition @xmath8 into regions of volume @xmath22 , @xmath196 , i. e. , @xmath197 for all @xmath117 .
therefore , @xmath198 and the same value is obtained for @xmath185 : @xmath199 now the optimality claim follows from @xmath200 .
we still have to prove uniqueness of the solution , up to a set of measure 0 .
assume that @xmath13 differs from @xmath185 on some set @xmath201 of positive measure .
we are done if we show that the inequality between @xmath190 and @xmath191 holds as a strict inequality .
we can remove the points of @xmath59 from @xmath201 , since they have measure 0 . then holds as a strict inequality on @xmath201 .
the difference between @xmath190 and @xmath191 is then the integral of the positive function @xmath202 over a set @xmath201 of positive measure .
such an integral @xmath203 has always a positive value : the countably many sets @xmath204 and @xmath205 partition @xmath201 , and thus their measures add up to @xmath206 .
therefore , at least one of these sets has positive measure , and it follows directly that the integral is positive .
in contradistinction to theorem [ opti - theo ] , the weight vector @xmath60 and the resulting weighted voronoi partition in theorem [ partition - theo ] is not unique : if @xmath8 is disconnected , changing @xmath60 may move the bisectors between different connected components of @xmath8 without affecting the partition , or only changing it in boundary points of @xmath8 .
however , uniqueness can be obtained under the additional assumption that @xmath8 is pathwise connected .
then in theorem [ partition - theo ] , the weight vector @xmath207 that partitions @xmath8 into regions of prescribed size is unique up to addition of the same constant to all @xmath28 .
consequently , the parts in the weighted voronoi partition outside of @xmath8 are also uniquely determined .
the proof uses the following lemma .
[ decrease - lemm ] assume that @xmath8 is path - connected , in addition to being an open bounded set .
consider a partition of the point set @xmath0 in two nonempty sets @xmath123 and @xmath137 .
suppose , that , for a given weight vector @xmath60 , all regions @xmath208 have positive measure .
if we increase the weight of every point in @xmath123 by the same constant @xmath209 , then @xmath210 increases , for some @xmath127 . in this section , we will omit the reference to the set @xmath0 from the voronoi regions and simply write @xmath211 instead of @xmath212 , since the point set @xmath0 is fixed .
it is clear that the regions @xmath213 for @xmath127 can only grow , in the sense of gaining new points , but we have to show that this growth results in a strict increase of @xmath214 for some @xmath127 .
consider the continuous function @xmath215 and the function @xmath216 , which is defined analogously , and define @xmath217 these two sets partition @xmath2 , apart from some `` generalized bisector '' where equality holds .
roughly , @xmath218 is obtained by merging all voronoi regions @xmath219 for @xmath127 into one set , together with some bisecting boundaries between them .
since these boundaries have zero measure inside @xmath8 , we have @xmath220 let @xmath221 and @xmath222 , where @xmath223 if @xmath224 and @xmath134 otherwise .
it suffices to show that in going from @xmath60 to @xmath125 , @xmath225 increases , since then , by , one of the constituent voronoi regions @xmath226 must increase .
as mentioned , the set @xmath227 can only grow , but we have to show that this growth results in a strict increase of @xmath9 .
if @xmath228 , we are done because @xmath229 has increased to @xmath80 . otherwise , consider a path in @xmath8 connecting some point @xmath230 with some point @xmath231 .
we have , by definition , @xmath232 and @xmath233 .
hence , by the intermediate value theorem , we can find a point @xmath5 on the path with @xmath234 .
this point is an interior point of @xmath235 and @xmath236 , and hence there is a neighborhood of @xmath5 which adds a positive measure to @xmath237 when going from @xmath60 to @xmath125 .
to finish the proof of uniqueness in theorem [ partition - theo ] , suppose that two weight vectors @xmath238 do the job of producing the desired measures @xmath22 for the regions , but @xmath239 for any constant @xmath240 .
we may assume , by adding a constant to all entries , that @xmath241 , and @xmath242 for some @xmath23 .
we will now gradually change @xmath60 into @xmath125 by modifying its entries .
let @xmath243 . by assumption ,
@xmath123 is neither empty nor equal to @xmath0 .
we increase the weights of the points in @xmath123 , leaving the remaining values fixed .
the amount @xmath124 is chosen in such a way that @xmath244 for @xmath245 does not decrease below the remaining differences @xmath246 for @xmath247 . by lemma [ decrease - lemm ] , the measure of the region assigned to some point @xmath248 strictly increases in this process . when the limiting value @xmath124 is reached , the set @xmath123 of points where @xmath244 achieves its maximum is enlarged , and the whole process is repeated until @xmath249 .
during this process , @xmath250 will always be among the points whose weight is increased , and therefore , the measure of its region can never shrink back to the original value a contradiction .
we have given a geometric proof for the fact that additively weighted voronoi diagrams can optimally solve some cases of the monge - kantorovich transportation problem , where one measure has finite support .
surprisingly , the existence of an optimal solution the main mathematical challenge in the general case is an easy consequence of our proof . in remains to be seen to which extent our assumptions on the distance functions
@xmath251 can be further generalized .
d. gei , r. klein , and r. penninger .
optimally solving a general transportation problem using voronoi diagrams . in : w. didimo and g. liotta ( eds . ) , abstracts 28th europ .
workshop computational geom .
( eurocg12 ) , assisi , italy , ( march 2012 ) : 241-244 .
d. gei , r. klein , and r. penninger .
optimally solving a transportation problem using voronoi diagrams . to appear in j. gudmundsson , j. mestre , and a. viglas ( eds . ) , _ computing and combinatorics_. proc .
computing and combinatorics conf .
( cocoon ) , sidney , august 2012 .
notes comp .
, vol . 6842 , springer ( 2012 ) .
g. monge .
mmoire sur la thorie des dblais et de remblais .
histoire de lacadmie royale des sciences de paris , avec les mmoires de mathmatique et de physique pour la mme anne 29 ( 1781 ) : 666704 .
r. sharathkumar , and p. k. agarwal .
algorithms for the transportation problem in geometric settings .
proceedings of the twenty - third annual acm - siam symposium on discrete algorithms , ( soda 12 ) : 306317 , 2012 . | we consider the following variant of the well - known monge - kantorovich transportation problem .
let @xmath0 be a set of @xmath1 point sites in @xmath2 .
a bounded set @xmath3 is to be distributed among the sites @xmath4 such that ( i ) , each @xmath5 receives a subset @xmath6 of prescribed volume and ( ii ) , the average distance of all points @xmath7 of @xmath8 from their respective sites @xmath5 is minimized . in our model ,
volume is quantified by a measure @xmath9 , and the distance between a site @xmath5 and a point @xmath7 is given by a function @xmath10 .
under quite liberal technical assumptions on @xmath8 and on the functions @xmath11 we show that a solution of minimum total cost can be obtained by intersecting with @xmath8 the voronoi diagram of the sites in @xmath0 , based on the functions @xmath11 equipped with suitable additive weights . moreover , this optimum partition is unique , up to sets of measure zero . unlike the deep analytic methods of classical transportation theory ,
our proof is based directly on geometric arguments .
monge - kantorovich transportation problem , earth mover s distance , voronoi diagram with additive weights , wasserstein metric |
pq and is a dwarf nova with three known outbursts in the last @xmath0 70 years ( richter 1990 ) .
the last outburst on 21 march , 1988 reached a visual magnitude of 10 ( hurst 1988 ) .
it had many of the characteristics of the wz sge class of dwarf novae including a large outburst amplitude and strong balmer emission superimposed over broad absorption at quiescence ( howell et al . 1995 ) .
however , to be in the wz sge class , pq and must have an orbital period below the period gap .
time series optical spectra of pq and were obtained by schwarz et al .
( 2004 ) to determine the orbital period of the system .
they found a period of 1.7 hours from the fit to the radial velocity curve as determined from the convolution of double gaussians to the line profiles ( schafter 1985 ) .
unfortunately the derived period was similar in length to the total observing time and could nt be classified as a definite period due to undersampling .
nevertheless the h@xmath1 emission showed a clear periodic change in its blue and red components during the observing run which supported a short orbital period typical of wz sge systems .
an optically thin disk , showing the wd absorption , is indicative of a short period , low-@xmath2 system ( i.e. below the period gap ) .
likewise , the lack of an accretion disk in the optical spectrum placed a firm upper limit of 3 hours on the orbital period of pq and .
schwarz et al . also determined the effective temperature and surface gravity of the white dwarf ( wd ) using synthetic spectra from model atmospheres .
the best fits gave t@xmath3 = 12,000 @xmath4 1,000 k and log(@xmath5 ) = 7.7 @xmath4 0.3 ( cgs units ) which placed pq and in the region of the zz ceti instability strip ( bergeron et al .
1995 ; 2004 ) .
they noted that with its low accretion rate , pq and was an excellent candidate to search for non - radial oscillations which have recently been observed in other wz sge novae .
in this paper we present the results of a photometric campaign to search for non - radial pulsations in pq and .
section 2 provides details of the observations .
the analysis of the data are given in section 3 and our conclusions follow in section 4 .
our photometric observations were carried out using the orthogonal parallel transfer imaging camera ( optic , see howell et al , 2003 ) at the wiyn observatory 3.5-m telescope located on kitt peak .
optic is the only prototype orthogonal transfer ccd imager operating ( see tonry et al . , 2002 ) and
consists of two 2k by 4k ccid-28 otccds in a single dewar mounted adjacent to each other with a small gap in between the chips .
optic is controlled by standard sdsu-2 electronics running custom microcode and reads out the two otccds via 4 video channels , one located in each corner of the device .
optic has a read noise of @xmath64 electrons when read at a normal rate of 160 kpix / sec and a gain of 1.45 e / adu .
we used optic in conventional mode placing the target star and all comparison stars of interest in one of the ccds .
two time series data sets were obtained , the first was on the night of 14 september 2004 ut and the second on 13 october 2004 ut . the september observations consisted of @xmath0125 , 45 second observations using a johnson v filter .
the ccd was binned 2 x 2 ( the seeing on this night was 1.6 `` ) and the readout time was 8 seconds .
the october observations consisted of @xmath0100 60 second v - band integrations using 1 x 1 binning ( readout time was 24 seconds ) and the seeing was 0.5 '' .
the data were reduced using the standard iraf packages .
relative photometry was performed using four different background stars as references .
figure 1 shows a plot of the magnitude difference as a function of time for both nights .
the squares represent the difference between two of the reference stars while the triangles represent the difference between pq and and one of the reference stars .
the horizontal axis is the time since the start of observations for each night .
the light curves of both nights clearly show fluctuations of up to @xmath00.1 magnitudes .
all reference stars and pq and were of similar magnitude therefore the scatter in the reference star data gives an estimate of the errors in the magnitude change for pq and . from figure 1
we see that the magnitude errors are on order of 0.015 .
to search for any periodicities in the light curves the data were put through a fourier transform routine .
the resulting power spectra for each date are shown in figure 2 .
based on the length of time pq and was observed and the sampling rate of each night , we searched the frequency space from 0.5 to 30 hr@xmath7 . to determine the level of significance for peaks in the power spectrum we added a tracer signal to our data and ran it through the fourier routine until we were unable to detect it . from this we found that peaks with a power less than @xmath8 were not significant .
three strong peaks appear in the power spectra of both nights .
the strongest peak is found at a frequency of 5.6 hr@xmath7 , a period of 10.5 minutes , in both spectra .
we also see what appear to be harmonics of this peak at frequencies of @xmath02.8 hr@xmath7 and @xmath01.4 hr@xmath7 . at higher frequencies
the two power spectra differ slightly . in the september data we see no significant peaks beyond
the three already mentioned while the october power spectrum shows a few small peaks at higher frequencies .
these peaks are found at 7.8 , 8.4 , 9.2 , 22.5 , and 29 hr@xmath7 ( periods in the range of 2 - 8 minutes ) .
the difference between the two nights could be due to many things including sampling rates , seeing effects , and variability in the wd itself .
figure 3 shows the data phased to a period of 10.5 minutes and put in phase bins .
the error bars have been determined from the scatter in the light curve of the reference stars .
the data show an overall change in magnitude of about 0.1 .
the most significant period we find , 10.5 minutes , is much shorter than any expected orbital period for a cv .
several wz sge systems have recently been found with similar short period , low amplitude oscillations ( araujo - betancor et al 2005 ; townsley , arras & bildsten 2004 ; woudt & warner 2004 ; van zyl et al 2004 ) .
these periodic fluctuations have been attributed to zz ceti like non - radial pulsations of the wd primary . in the five systems reported to date , there appears to be a pattern in the periods found .
all of the stars show periods around 10 minutes and then several other periods on shorter time scales ( 3 - 6 minutes ) .
long term monitoring shows that these periods are variable between subsequent observations .
our findings for pq and match this general pattern . some of the other zz ceti systems in the literature have more complicated power spectra , showing peaks at longer frequencies that are multiples or combinations of the shorter frequency peaks .
we do nt see these patterns in our data for pq and however it could be that we have insufficient data to detect these complexities .
most of the other zz ceti systems have far more extensive data sets . the radial velocity analysis of schwarz et al .
( 2004 ) implied an orbital period of @xmath0102 minutes for pq and .
figure 4 shows the power spectrum of our data in the frequency range of 0.5 to 2 hr@xmath7 ( 120 to 30 minutes ) .
the strongest peak is the 2nd harmonic of the 10.5 minute pulsation period .
there is a smaller peak structure at about 0.7 hr@xmath7 ( 86 minutes ) but it does nt fit the previously derived orbital period and it is very close to our significance threshold .
it is disconcerting that the orbital period as determined from the radial velocities is not seen in the power spectrum of the photometry .
one explanation is that the pq and system is at a low inclination .
if this is the case then orbital motions would cause the observed periodicity in the h@xmath1 red and blue emission components but no photometric eclipses would be seen .
we can place an upper limit on the system inclination using the estimated mass ratio from schwarz et al .
( 2004 ) and the minimum inclination for primary eclipses ( see equations 2.5c and 2.64 from warner 1995 ) .
a @xmath9 places an upper inclination limit of about 74@xmath10 for pq and .
the available evidence points to pq and containing a zz ceti white dwarf .
the primary pulsation period we find is 10.5 minutes which is similar to those found in other wz sge type systems with zz ceti like white dwarf parameters and pulsations .
more extensive observations may reveal alternate pulsation modes as has been found in other zz ceti systems .
we find no evidence for the 1.7 hour orbital period reported by schwarz et al ( 2004 ) in our data .
van zyl , l. , warner , b. , odonoghue , d. , hellier , c. , woudt , p. , sullivan , d. , pritchard , j. , kemp , j. , patterson , j. , welsh , w. , casares , j. , shahbaz , t. , van der hooft , f. , & vennes , s. 2004 , , 350 , 307 | we have detected pulsations in time - series photometry of the wz sge dwarf nova pq and .
the strongest peak in the power spectrum occurs at a period of 10.5 minutes .
similar periods have been observed in other wz sge systems and are attributed to zz ceti type non - radial pulsations .
there is no indication in the photometry of an approximately 1.7 hour orbital period as reported in previous spectroscopic observations . |
variability with a few periods of the order of one day is frequently observed in main - sequence b - type stars ( e.g. * ? ? ? * ) .
objects exhibiting only these slow oscillations are now called * s*lowly * p*ulsating * b*-type stars ( spb ) and are found in the spectral - type range of b3-b8 .
it is commonly believed that this variability is due to excitation of high - order g - modes .
such modes are also invoked as an explanation of low - frequency variations observed in certain @xmath7 cephei stars .
linear non - adiabatic calculations predict excitation of high - order g - modes in a wide range of main sequence b - type stars .
however , difficulties in explaining in this way long - period variability in certain objects have been noted . in several stars , observations from space by the corot and _
kepler _ missions led to the detection of a large number of frequency peaks which may be associated with g - modes .
however , mode identification for individual peaks seems a formidable task . a method based on a search for equidistant spacing in period , that has been successfully used in the application to white dwarfs , was tried for the corot target hd50230 by @xcite .
the authors found a sequence of eight modes with a remarkably constant separation , much more so than in relevant models @xcite . in the same frequency range
there were over 100 frequencies ( some of much higher s / n than most of those forming the sequence ) for which no plausible interpretation was offered so far @xcite . in an attempt to explain
the oscillation spectrum of another corot target , hd 43317 , @xcite also could not account for the equidistant spacing reported by @xcite .
patterns in oscillation spectra apparently do not yield a key toward deciphering information on excited modes in spb stars .
we need to know more about mode selection in these stars .
perhaps we may learn something new about it from ground - based observations , which yield much fewer frequency peaks than the ones from space , but much more information about each of them that would enable mode identification . in this respect ,
the spb star @xmath0 eri appears to be the best choice .
a large number of three - passband photometric data for this star have been collected during two multisite campaigns ( msc ) aimed at the @xmath7 cep variable @xmath8 eri , 2002 - 2003 msc @xcite and 2003 - 2004 msc ( * ? ? ?
* hereafter j2005 ) .
the star has been used as one of two comparison stars in both campaigns .
the analysis of the 2002 - 2003 msc data by @xcite revealed a variability of @xmath0 eri with a frequency of 0.616 c / d .
however , the authors were not able to decide what is the origin of this variation : pulsation or rotational modulation .
the analysis of the extensive data set from both campaigns presented in j2005 led to the detection of five additional significant peaks in the g - mode frequency range typical for spb stars and an unambiguous classification of @xmath0 eri .
additional three - colour data were recently obtained by one of us ( gh ) . these and the msc data were employed in the mode identification presented in our work .
photometry with the most micro - satellite ( * ? ? ?
* hereafter j2013 ) was also used in frequency determination in section2.3 .
unfortunately , spectroscopic observations reported in the same paper yielded only the upper bound on radial velocity variations .
however , these observations were used to constrain parameters of the star ; this , as we will see later in the paper , is essential for mode identification .
all data used in our work are presented in the next section .
we also provide information on the mean physical parameters of the star and describe its models which will be used in the subsequent sections . in section3 , we confront frequencies of the peaks with instability ranges determined in one of the selected models .
we use the condition of mode instability as a requirement for its possible association with an observed frequency .
mode visibility is another factor limiting the set of modes for possible identification .
this matter is discussed in section4 .
section5 is devoted to an analysis of the data on the amplitudes and phases of the six peaks determined in three photometric passbands .
the aim is to determine the harmonic degree , @xmath9 , and the azimuthal order , @xmath10 , of the observed modes simultaneously with constraints on the stellar rotation rate , assuming that it is uniform .
the method , which employs the traditional approximation , has been described in detail by ( * ? ? ?
* hereafter d07a ) and here we give only its brief outline . we have not attempted to constrain a stellar model by fitting calculated frequencies to observations because it would not make sense in view of the approximation adopted in treating the effects of rotation and high density of the calculated oscillation spectra related to high radial orders of the excited modes .
our primary goal is to learn something about mode selection in spb stars .
we believe that this is essential for understanding complicated oscillation spectra from the space missions .
the discussion in the final section is focused on this issue .
@xmath0 eridani ( hd 30211 ) is a bright ( @xmath124 mag ) b5 iv star .
it is certainly not a slow rotator .
j2013 provide two independent estimates of the projected equatorial velocity of rotation , @xmath13130@xmath143 and 136@xmath146 km / s .
it has been known since a long time @xcite that the star is the primary in a single - lined spectroscopic binary system but only j2005 detected shallow eclipses .
this is a well detached system with the orbital period @xmath157.38090 d and eccentricity @xmath160.344 ( j2013 ) .
hence , we can neither assume synchronized rotation nor parallel rotation and orbital axes .
thus , mode identification must be done simultaneously with the determination of the inclination angle .
we adopted after j2013 @[email protected] and @[email protected] for the effective temperature and luminosity of @xmath0 eri .
these values were found from the mean strmgren photometric indexes and the revised value of the _ hipparcos _ parallax @xcite . from their own spectroscopic data , j2013 found a metal abundance close to solar .
they also determined the mean density of the star , @[email protected] g/@xmath20 from the orbital parameters .
the above - mentioned values of @xmath21 , @xmath22 and @xmath23 imply a mass @xmath246.0 m@xmath25 and a radius @xmath266.1 r@xmath25 , hence the break - up velocity amounts to @xmath27 430 km / s and the corresponding value of the critical rotational frequency to @xmath281.3 c / d .
thus , @xmath0 eri rotates at a speed of at least 30% of @xmath29 .
our models are spherically symmetric and include only the spherically symmetric effect of centrifugal force .
this is a crude but an acceptable approximation if @xmath30 , where @xmath31 and @xmath32 are the angular frequency of rotation and its critical value , respectively .
assuming for @xmath0 eri the radius of @xmath33 , we obtain @xmath34 for @xmath35 km / s and this number is the upper limit in our search for the mode identification .
the adopted lower limit is 130 km / s and it results from the value of @xmath36 km / s . in all our models
, rotation is assumed uniform .
all the stellar models in this work were calculated with the warsaw evolution code adopting opal opacities @xcite for @xmath37 = 0.28 , @xmath38=0.015 and the agss09 solar mixture of heavy elements @xcite .
parameters of the models adopted for mode identification in @xmath0 eri are listed in table1 . as may be seen in fig.1 , where positions of the models on the evolutionary tracks in the hr diagram are shown ,
these models satisfy the observational constraints discussed above . at a finite rotation rate ,
positions in the hr diagram depend on @xmath4 .
what is shown are aspect - averaged positions but the spread is small in comparison with the mean shift which we may asses by comparing tracks with the initial equatorial velocity of 200 and 0 km / s .
the three selected models were calculated adopting @xmath39200 km / s at the zams and the overshooting parameter , @xmath400.2 .
note in fig.1 that the track with @xmath400 enters the error box but only at the edge of the strip allowed by the mean density .
model 2 is on the same track as model1 but it is more evolved .
model 3 has a similar temperature to model 2 but has seven percent higher surface gravity and is less evolved .
all these models lie within the main sequence , close to its red boundary ( tams ) .
.models considered for @xmath0 eridani . [ cols="^,^,^,^,^,^,^ " , ] the modes we associated with @xmath41 are , in fact , unstable but for the tidal excitation this is irrelevant . what matters is only that the driving or damping rate is much lower than the frequency . however , it is important to stress that we end up with the same identification of the @xmath41 peak considering only unstable modes .
none of the @xmath42 values listed in tables3 - 5 comes close to 0.01 , which was set as the upper limit for acceptable identifications .
we can not exclude very different identifications if the equatorial velocity is above 280 km / s , which we adopted as the maximum only because of limitations of the method we used .
it is fortunate that consistent identifications were found well below this value . with @xmath43 km / s ,
the star rotates at less than 1/3 of the maximum rotation rate and the excess of the equatorial radius over the mean radius is below 4 percent .
the stabilization of retrograde modes was found by @xcite much closer to maximum equatorial velocity .
thus , there should be no reservation against our few identification with such modes .
using three - band photometric data , we determined angular numbers , @xmath9 and @xmath10 , of modes responsible for the six significant frequency peaks in the fast rotating spb star @xmath0 eri . simultaneously , narrow ranges for the allowed equatorial velocity of rotation were found .
we took into account modes with @xmath44 in three models with mean parameters strongly constrained by observational data .
the modes were calculated with a nonadiabatic version of _ the traditional approximation_. for the five frequency peaks only unstable modes were taken into account whereas for the one frequency @xmath45 this condition was relaxed . nevertheless , the identified modes for this frequency turned out to be unstable .
thus , we can not decide whether the @xmath41 peak arises from the tidal action exerted by the companion or from excitation by the @xmath46-mechanism operating in the @xmath38-bump zone .
fortunately , this uncertainty neither affects constraints on the rotation rate nor the mode identification .
the inferred equatorial velocity is slightly different for the three models but for all of them it is between 135 and 140 km / s .
one interesting difference between the models was found for the dominant peak , which was attributed to a zonal @xmath473 mode in two models ( models1 and 3 ) and to a prograde @xmath471 mode in one ( model2 ) .
these different identifications were found with models1 and 2 lying on the same @xmath48m@xmath25 evolutionary track and differing by less than 0.008 in @xmath21 . with model3 ( @xmath495.85 m@xmath25 , @xmath21 nearly the same as model2 )
all angular numbers were found to be the same as in model1 , except for one alternative identification for @xmath41 . with all our models , only the @xmath474 and @xmath500 numbers and only at the lowest equatorial velocity allowed by @xmath51 data were found acceptable for the second in the significance hierarchy peak ( @xmath52 ) .
this eliminated the @xmath471 and @xmath500 possibility for the dominant peak which otherwise would have been acceptable in the wide range of @xmath3 for all three models . with known @xmath3 , @xmath9 , and @xmath10 , we calculated amplitudes of the radial velocity variations and compared them with spectroscopic data , which yielded only an upper bound .
the @xmath473 , @xmath500 identification for @xmath53 leads to the amplitude exceeding the upper bound by at least a factor of three . with @xmath471 , @xmath501 ,
the excess is much lower .
it would be going too far to judge that model2 is better than the remaining two models because the spectroscopic and photometric data do not satisfy criteria of a reliable comparison , as specified in section5.3 .
nonetheless , the large difference in the calculated amplitudes points to the prospect for constraining model parameters with better data on the radial velocity amplitudes .
there are some differences in @xmath9 inferred with model2 and with the remaining two models . for the @xmath41 peak
, @xmath500 is found with all models but @xmath474 is inferred with models1 and 3 while @xmath476 with model2 . for the @xmath52 peak the @xmath474 , @xmath500 identification is found with all the models whereas for the @xmath54 peak we got @xmath474 , @xmath551 .
for the lowest amplitude peaks there are alternative identifications .
the inferred degrees range from @xmath471 to 6 .
it seems that there is no way to predict which of the huge number of unstable modes will reach the amplitude at a specified detection level . neither driving rate nor visibility is a reliable predictor .
perhaps the preference to the @xmath474 modes results from a compromise between visibility and the number of unstable modes in the frequency range of the peaks .
this number is much higher at @xmath474 than 1 .
pulsations in spb stars appear less predictable than the stochastic solar - like oscillations . without constraints on the angular numbers from multiband photometry
the task of mode identification in spb stars appears totally hopeless . in section1
, we stressed that an equidistant separations in period found by corot are most likely misleading .
the @xmath56 limit on the potential identifications is not justified . in the present work
, we considered modes up to @xmath476 but it may still be not enough .
the degree of excited modes even at the moderate rotation rates may only be determined jointly with azimuthal orders and the inclination of the rotation axis .
this complicates the mode identification but as a reward we obtain valuable information about rotation .
the result is sensitive to model parameters .
adequate data on the radial velocity amplitude lead to a new constraint on the model .
unfortunately , for @xmath0 eri we have only upper limits .
this is one of the sources of uncertainties of our results .
data on radial velocity amplitudes are very important but to be fully useful they have to come from an epoch not too distant from that of the photometric data and must be based on measurements of changes in the first moments of spectral lines which measure disc - averaged velocities .
we are indebted to the anonymous referee for pointing out that the frequency @xmath41 is a whole multiple of the orbital frequency .
the work of jdd and gh was supported by the polish ncn grants 2011/01/m / st9/05914 and 2011/01/b / st9/05448 .
wd was supported by the polish ncn grant dec-2012/05/b / st9/03932 .
99 aerts c. , dupret m .- a , 2012 , in shibahashi h. , takata m. , lynas - gray a. e. , eds , asp conf
vol 462 , progress in solar / stellar physics with helio- and asteroseismology .
pac . , san francisco , p. 103 asplund m. , grevesse n. , sauval a. j. , scott p. , 2009
, ara&a , 47 , 481 bildsten l. , ushomirsky g. , cutler c. , 1996 , apj , 460 , 827 breger m. , stich j. , garrido r. , martin b. , jiang s. y. , li z. p. , hube d. p. , ostermann w. , paparo m. , scheck m. , 1993 , a&a , 271 , 482 claret a. , 2000 , a&a , 363 , 1081 daszyska - daszkiewicz j. , dziembowski w. a. , pamyatnykh a. a. , goupil m - j . , 2002 , a&a , 392 , 15 daszyska - daszkiewicz j. , dziembowski w. a. , pamyatnykh a. a. , 2007a , acta astron . , 57 , 11 ( d07a ) daszyska - daszkiewicz j. , dziembowski w. a. , pamyatnykh a. a. , 2007b , eas publications series , 26 , 129 de cat p. , 2007 , coast , 150 , 167 de cat p. , daszyska - daszkiewicz j. , briquet m. et al .
, 2004 , in iau colloquium 193 , kurtz d. w. , pollard k. r. , eds , asp conf .
vol 310 , variable stars in the local group
, san francisco , p. 195
de cat p. , briquet m. , daszyska - daszkiewicz j. et al . , 2005 ,
a&a , 432 , 1013 degroote p. , aerts c. , baglin a. et al . , 2010 , nature , 464 , 259 dziembowski w. a. , daszyska - daszkiewicz j. , pamyatnykh a. a. , 2007 , mnras , 374 , 248 dziembowski w. a. , moskalik p. , pamyatnykh a. a. , 1993 , mnras , 265 , 588 frost e. b. , lee o. j. , 1910 , sci , 32 , 876 iglesias c. a. , rogers f. j. , 1996 , apj , 464 , 943 handler g. et al . , 2000 , mnras , 318 , 511 handler g. , shobbrook r. r. , jerzykiewicz m. et al . , 2004 , mnras , 347 , 454 jerzykiewicz m. , handler g. , shobbrook r. r. , pigulski a. , medupe r. , mokgwetsi t. , tlhagwane p. , rodrguez e. , 2005 , mnras , 360 , 619 ( j2005 ) jerzykiewicz m. et al .
, 2013 , mnras , 432 , 1032 ( j2013 ) kurucz r. l. , 2004 , http://kurucz.harvard.edu lee u. , 2010 , coast , 157 , 203 lee u. , saio h. , 1997 , apj , 491 , 839l ppics p. , briquet m. , baglin a. et al . , 2012 , a&a , 542 , 55 reese d. r. , prat v. , barban c. , vant veer - menneret c. , macgregor k. b. , 2012 , a&a , 550 , 77 savonije g. j. , 2005 , mnras , 443 , 557 savonije g. j. , 2007 , a&a , 469 , 1057 savonije g. j. , 2013 , a&a , 559 , 25 stellingwerf r. f. , 1978 , aj , 83 , 1184 szewczuk w. , daszyska - daszkiewicz j. , dziembowski w. a. , 2014 , in guzik j. a. , chaplin w. j. , handler g. , pigulski a. , eds , iau symposium 301 , precision asteroseismology .
cambridge university press , p. 109
townsend r. h. d. , 2002 , mnras , 330 , 855 townsend r. h. d. , 2003a , mnras , 340 , 1020 townsend r. h. d. , 2003b , mnras , 343 , 125 townsend r. h. d. , 2005 , mnras , 364 , 573 van leeuwen f. , 2007 , a&a , 474 , 653 | we present results of a search for identification of modes responsible for the six most significant frequency peaks detected in the rapidly rotating spb star @xmath0 eridani . all published and some unpublished photometric data are used in our new analysis .
the mode identification is carried out with the method developed by daszyska - daszkiewicz et al . employing the phases and amplitudes from multi - band photometric data and relying on the traditional approximation for the treatment of oscillations in rotating stars .
models consistent with the observed mean parameters are considered .
for the five frequency peaks , the candidates for the identifications are searched amongst unstable modes . in the case of the third frequency , which is an exact multiple of the orbital frequency ,
this condition is relaxed .
the systematic search is continued up to a harmonic degree @xmath1 .
determination of the angular numbers , @xmath2 , is done simultaneously with the rotation rate , @xmath3 , and the inclination angle , @xmath4 , constrained by the spectroscopic data on the projected rotational velocity , @xmath5 , which is assumed constant .
all the peaks may be accounted for with g - modes of high radial orders and the degrees @xmath6 .
there are differences in some identifications between the models .
for the two lowest amplitude peaks the identifications are not unique . nonetheless , the equatorial velocity is constrained to a narrow range of ( 135 , 140 ) km / s .
our work presents the first application of the photometric method of mode identification in the framework of the traditional approximation and we believe that it opens a new promising direction in studies of spb stars .
[ firstpage ] stars : early - type stars : oscillations stars : rotation stars : individual : @xmath0 eri |
actively accreting black holes have three fundamental properties mass ( @xmath2 ) , accretion rate ( @xmath3 ) , and angular momentum . measuring @xmath2 for active galactic nuclei at all redshifts has become possible due to reverberation mapping of low - redshift agn and the extrapolation of those results to high redshifts , via relations between @xmath2 and the widths of broad emission lines and the agn continuum luminosity .
accretion rate estimates have also been achieved for many agn , usually via the eddington ratio , @xmath4 .
the angular momentum , or spin ( @xmath5 ) , of active bhs is more elusive , as it requires probing the region near the inner edge of the accretion disk ( ad ) .
yet measurements of spin and spin evolution would provide valuable clues to the accretion history of active bhs and perhaps the evolution of the agn and host galaxies themselves . at present
, there are two primary methods for constraining the spin parameters of actively accreting bhs : ( 1 ) measuring the fe k@xmath0 emission line and/or a soft x - ray excess that some attribute to relativistic reflection ( e.g. * ? ? ?
* ; * ? ? ?
* ) , and ( 2 ) fitting the agn continuum emission ( cf ) ( e.g. * ? ? ?
* ; * ? ? ?
there are significant advantages and drawbacks to each method .
the fe k@xmath0 method is based on relativistic x - ray reflection .
it does not require prior knowledge of @xmath2 , the distance to the source , or the inclination of the disk , whereas these are all necessary ingredients for the cf method .
the main drawback , however , is that a very high - quality x - ray spectrum is required to properly model the continuum emission and the fe - k@xmath0 emission line , severely limiting the number of sources for which current technology allows a spin measurement . as a result , most agn with reflection measurements
are at a redshift less than 0.1 .
furthermore , the fe k@xmath0 emission line is present in just @xmath640% of bright , nearby type i agn @xcite , so some spin estimates are based on modeling just a soft x - ray excess ( e.g. * ? ? ?
* hereafter , r14 ) .
the cf method , on the other hand , can be applied to any agn where the continuum emission can be measured .
this vastly increases the number of agn for which a spin measurement can be made and has already been applied out to a redshift of @xmath61.5 @xcite .
the primary drawback is that wide wavelength coverage , sometimes exceeding the capabilities of a single observatory , is required to properly measure the shape of the sed .
furthermore , this method can not be applied effectively if the peak of the ad spectrum occurs in a wavelength regime inaccessible to current observatories , e.g. , the extreme uv ( where many agn spectra do indeed peak ) .
this method generally assumes a thin ad model , based on @xcite .
recent work has directly cast doubt on the x - ray reflection method .
@xcite find that the soft x - ray excess that some attribute to relativistic reflection is more likely due to warm comptonization .
similarly , @xcite is able to fit the fe k@xmath0 emission line for one of the agn with an x - ray reflection spin measurement without invoking relativistic reflection . for the cf method , while the standard thin ad model has been successful in fitting the uv - optical seds of many agn ( see e.g. * ? ? ?
* ; * ? ? ?
* ) , other work has found that the agn sed can be fit with the combination of a thermal disk component and a warm comptonization component @xcite , indicating the possibility of greater complexity in the continuum emission . with these two methods
now available and actively in use for the estimation of @xmath5 in agn , it is time to investigate whether these two methods give consistent results when applied to the same agn .
this is especially important given the uncertainties in both techniques and because neither method can probe the full agn population . in this work
, we compare the x - ray reflection and cf methods for two nearby agn h1821 + 643 and ngc 3783 .
ours is among the first attempts to make this comparison ( see also , * ? ? ?
both targets have a published spin estimate from the reflection method .
we perform the cf analysis and compare the results in detail . in 2 , we describe how we selected sources for this study and our search for appropriate archival data . in 3 , we describe the models and cf procedure ( based on @xcite ) .
4 and 5 describe our application of the cf method to the two agn , and we conclude in 6 with a discussion of our results and how the reflection and cf method compare for these two case studies .
we assume a @xmath7cdm model with @xmath8 , @xmath9 , and @xmath10 .
according to @xcite , there are currently 25 agn with spin estimates from the x - ray reflection method .
we use this list as a starting point to search for archival data to which the cf method can be applied .
the cf method is most effective when the `` turnover '' in the ad spectrum is probed .
this turnover occurs at shorter wavelengths for smaller black hole masses .
we therefore look first for existing high - quality uv spectroscopic observations of these agn . via the mast web portal , we identify four agn with high - level data products for hubble space telescope ( hst ) faint object spectrograph ( fos ) observations @xcite : fairall 9 , ngc 3783 , ngc 4151 , and h1821 + 643 . while the fos spectrum is sufficient for applying the cf method for h1821 + 643 , data at even shorter wavelengths is required for the lower@xmath2 seyfert galaxies .
we seek quasi - simultaneous data , and , for ngc 3783 , we identify observations from rosat taken on 1992 july 23 , just four days prior to the fos observation on 1992 july 27 that probe the appropriate wavelengths @xcite .
hence we proceed with two objects , h1821 + 643 and ngc 3783 , for our detailed spin comparison .
the fos spectra for h1821 + 643 and ngc 3783 are focused on the nucleus of the galaxy . for h1821 + 643
, we verify that the fos spectrum ( shown in fig .
[ fig : hspec ] ) is dominated by agn emission based on the broad - band star formation sed fit in @xcite .
similarly for ngc 3783 , the spectrum is at short enough wavelengths that the host galaxy contribution should be negligible @xcite .
therefore , we do not correct the fos spectra for stellar emission .
the only correction we make to the hst data is to divide out the galactic extinction , using the @xcite extinction law and the @xcite recalibration of the @xcite maps .
the rosat data have been analyzed ( * ? ? ?
* hereafter , t93 ) , and we make no further corrections in this work .
to apply the cf method , a model is required that can make specific predictions for the emitted radiation at each wavelength .
standard thin ad theory @xcite has been used for several decades to describe agn continuum emission .
newer models use this framework , but incorporate general relativistic corrections , comptonization in the disk atmosphere , and even disk winds ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
here we adopt the numerical code described in @xcite , assuming a viscosity parameter ( @xmath0 ) of 0.1 .
the shape and luminosity of the thin ad spectrum is mainly set by @xmath2 , @xmath3 , @xmath5 , and the inclination of the disk to our line - of - sight .
if we want to constrain @xmath5 , prior knowledge of the other parameters is necessary as the observed sed is not enough to break the parameter degeneracy of the models , where different combinations of these parameters can yield similar sed shapes .
additionally , any intrinsic reddening in the agn host galaxy will affect the observed sed shape .
we therefore adopt a bayesian approach that takes a large grid of models with varying values of @xmath2 , @xmath3 , @xmath5 , inclination , and reddening and maximizes the probability that any given model is a good representation of the data , while penalizing those models that are not consistent with the priors , which we establish for @xmath2 and @xmath3 @xcite .
this routine calculates a @xmath11 value for each model , using continuum windows along the observed sed . for the prior on @xmath2 , the reverberation mapping technique has been used to obtain @xmath2 for nearby agn ( e.g. , * ? ? ?
* ) , and these results have been extended to other agn , using the width of the broad emission lines and the continuum luminosity ( the single - epoch method ; e.g. , * ? ? ?
* ; * ? ? ?
a prior on @xmath3 can be estimated using @xmath2 and a measurement of the continuum luminosity at longer ( i.e. , optical or near - infrared ) wavelengths , assuming the canonical power law , @xmath12 @xcite . for the disk inclination ,
the only constraint we have is that our sample contains type-1 agn , so we can consider only inclinations where cos @xmath13 .
for intrinsic reddening , to limit the number of free parameters , we use a simple power - law curve , where @xmath14 mag .
we consider values of @xmath15 ranging from 0.0 to 0.50 mag .
h1821 + 643 is a brightest cluster galaxy ( bcg ) hosting a luminous agn at @xmath16 .
there are no direct reverberation mapping measurements for h1821 + 643 , but there have been several attempts to obtain @xmath2 via other methods .
these estimates range from @xmath61.2 to @xmath17 ( @xcite ; r14 ) , and there are theoretical arguments that the mass could be as high as @xmath18 m _
@xmath19 @xcite .
we adopt the most recent ` single - epoch ' measurement using the h@xmath20 emission line , @xmath21 , from @xcite , and we use their measurement of log @xmath22 46.1 ergs s@xmath23 for calculating @xmath3 .
we adopt errors of 0.3 and 0.2 dex , respectively , for @xmath2 and @xmath3 @xcite .
c|cc|cccc|c h1821 + 643 & 9.4 & 0.48 & @xmath24 & @xmath25 & @xmath26 & @xmath27 & @xmath28 + ngc 3783 & 7.47 & -1.9 & @xmath29 & @xmath30 & @xmath31 & @xmath32 & @xmath33 + ngc 3783 & 7.47 & -1.9 & @xmath34 & @xmath35 & @xmath36 & @xmath27 & + the best - fit ( i.e. , the most probable ) model is presented in fig . [ fig : hspec ] , and the full results are shown as probability contours in fig .
[ fig : hcont ] . from fig .
[ fig : hcont ] , it is clear that there is a strong preference for a large , positive spin parameter . in their analysis of the x - ray spectrum of h1821 + 643 ,
r14 obtain both a constraint on the spin parameter and a constraint on @xmath4 and the inclination .
applying these constraints to our cf routine , we obtain a similar probability distribution along the spin parameter axis as we did originally without these constraints .
ngc 3783 is a well - studied seyfert 1 , sba galaxy at @xmath37 .
the reverberation mapping technique has been applied to ngc 3783 , giving @xmath2 = @xmath38 m _ @xmath19 , with a corresponding continuum luminosity , log @xmath39 ergs @xmath40 , which we use to estimate @xmath3 . because ngc 3783 is in a lower @xmath2 regime than h1821 + 643 , the peak of the ad emission is in the extreme uv , a regime where we generally lack observations .
we can discriminate between different spin parameters only in the soft x - ray , where models with the highest spin parameters peak for lower - mass bhs .
ngc 3783 has a complex x - ray spectrum , with warm absorbers and a soft excess that appears and disappears @xcite .
we use rosat x - ray data ( see [ sec : data ] ) , in addition to the fos data , to apply the cf method to ngc 3783 .
we use the 1992 july 23 rosat observation , in particular , because it is nearly contemporaneous with the fos observation , and it extends to slightly lower energy ( down to 0.1 kev ) than more recent x - ray observations with chandra or xmm newton .
t93 fits the rosat data with several different power - laws based on different absorption models , ranging from a simple power - law model with @xmath41 = 2.22 to a warm absorber model with @xmath42 ( which is similar to the value found by @xcite of @xmath43 ) .
t93 also present a model with @xmath44 4.7 , which is much higher than other values in the literature , so we do not include it in our analysis .
a difficulty with using the x - ray spectrum of an agn for the cf method is that there is a known power - law component at x - ray wavelengths of unknown origin , in addition to possible emission from the ad .
hence , the x - ray data provides only an upper limit on the ad emission .
we therefore first alter our cf method to search through our model parameter space for the models with the highest spin parameter that give both a satisfactory fit to the fos spectrum ( @xmath45 ) and do not exceed the x - ray flux at 0.1 kev from the power - law fits to the rosat data . to be conservative in our upper limit , we adopt the warm absorber model power law ( @xmath42 ) from t93 .
we find models spanning the full range in spin parameter , including maximum spin , that can fit within the upper limit from the t93 warm absorber model power - law for the rosat data , as long as @xmath2 is at least as high as the @xcite @xmath2 estimate ( see , for example , the purple curve in fig .
[ fig : nspec ] ) . even for a maximally spinning black hole
, the thin ad emission does not directly contribute to the hard x - ray band ( i.e. , above 2 kev ; see fig .
[ fig : nspec ] ) .
hard x - ray observations of ngc 3783 give a less steep power law than in the soft x - ray band .
for example , t93 find @xmath46 when applying their warm absorber model to data from exosat .
if we assume that the excess emission indicated by a steeper powerlaw in the soft x - ray band is due to ad emission , we can subtract the hard x - ray powerlaw from the soft powerlaw at 0.1 kev to determine a continuum point for our regular bayesian cf procedure .
we use a value of 0.1 dex for the error on @xmath2 from @xcite .
the results of the cf routine are presented in the left panel of fig .
[ fig : ncont ] , and we find a median @xmath47 . using the x - ray reflection method , ( * ? ? ? * hereafter b11 ) determine a spin parameter @xmath48 at 90% confidence and @xmath49 at 99% confidence ( indicated by horizontal dotted and solid lines in fig .
[ fig : ncont ] ) .
ngc 3783 is known to have a warm absorber in its x - ray spectrum ( t93 ) , i.e. an outflow often presumed to originate from the ad of the agn ( e.g. , * ? ? ?
if this is the case for ngc 3783 , then the thin ad model must be modified , as the accretion rate would be reduced throughout the disk as material is ejected .
the @xcite thin ad code provides the option of adding a disk wind to the model .
we therefore rerun the cf routine using a model with a self - similar disk wind , where the mass outflow rate per decade of radius is constant .
the mass outflow rate for ngc 3783 has been estimated to be @xmath50 times the accretion rate ; however , much of this outflowing gas may come from beyond the accretion disk ( t93 ; * ? ? ?
in the absence of an empirical estimate of the mass outflow rate from the disk itself , we illustrate the affect of a massive disk wind by choosing a mass accretion rate at the outer part of the disc equal to three times the accretion rate at the innermost stable circular orbit ( isco ) .
the results are presented in the right panel of fig .
[ fig : ncont ] .
the main difference between these results and the results without the disk wind is that lower spin parameters ( @xmath51 ) are much less probable in the disk wind scenario .
this arises because the disk wind reduces the accretion rate in the inner part of the disk and thus suppresses the luminosity at short wavelengths . furthermore , while there is a high probability of @xmath49 both with and without a disk wind , there is clearly a lower probability of having @xmath48 if there is no disk wind ( there is a factor of @xmath61.6 difference in radiative efficiency between these two spin parameters ) .
there is also a positive correlation between the amount of intrinsic reddening and @xmath5 , with @xmath49 ruled out if there is close to zero reddening .
our aim in this work is to compare the derived spin parameters for the x - ray reflection and cf techniques for two `` case study '' agn .
table [ tab : table ] summarizes the results of the two methods , including values for @xmath4 , the disk inclination ( @xmath52 ) , and instrinsic reddening , as derived from the cf method . for h1821 + 643 , a bright agn with @xmath53 ,
r14 found @xmath54 using the reflection method .
for the cf analysis , the hst fos spectrum alone is sufficient , and while we do not obtain a very precise constraint on @xmath5 , we find a strong probability of a spin parameter that exceeds the lower limit from r14 , giving consistent results between the reflection and the cf method .
we emphasize here that r14 do not clearly detect an fe line , but instead fit excess continuum emission in the soft x - ray .
for some agn , physical processes other than relativistic reflection are the more likely cause of this soft excess ( * ? ? ? * and references therein ) , making this reflection spin measurement a tentative one ( see also 1 ) . for the cf method , from the posterior probability distribution ,
it is clear that if @xmath2 is higher , then @xmath5 would be constrained to the highest allowed values .
whereas , if @xmath2 is any lower , we would be unable to obtain a meaningful constraint on @xmath5 . for ngc 3783
, the fos spectrum lies along the power - law portion of the thin ad model spectrum , and only in the soft x - ray regime can models with different @xmath5 be distinguished .
fortunately , there is nearly contemporaneous fos and rosat data for ngc 3783 .
however , the x - ray data includes the known x - ray power - law emission that likely originates from above the ad ( often called the `` corona '' ) . using the x - ray flux as an upper - limit , we find that as long as @xmath2 is at least as high as the observed @xmath2 , any spin parameter could fit the data .
since there are other components besides the ad emission in the x - ray , if we assume that just the excess emission indicated by the steeper powerlaw slope in the soft x - ray , compared to in the hard x - ray , is due to the ad itself , applying the cf method to ngc 3783 gives a high probability for a high spin parameter , consistent with the 99% confidence lower limit from relativistic reflection ( b11 ) .
however , there is a low probability of @xmath5 exceeding the 90% confidence lower limit from b11 , unless we include a disk wind in the ad model .
the results of the cf method are , in general , consistent with the results of the reflection method for the two agn studied here . in particular
, the agreement is improved for ngc 3783 if we assume a disc wind , which we include based on the existence of a warm absorber in the x - ray spectrum . the disk wind analysis
, however , is tentative because it is unknown how much , if any , of the outflow originates from the inner part of the disk ( see e.g. * ? ? ?
if the outflow originates further out and therefore does not suppress the short wavelength thin ad emission , there is a slight tension between the two methods , as the reflection method suggests a slightly higher spin parameter than the cf method without a disk wind .
we also find that , without a disc wind , the highest spins are most probable for @xmath15 between 0.2 and 0.3 mag . while these reddening values are generally consistent with the constraints from broad emmission line measurements for ngc 3783 ( @[email protected] ; * ? ? ?
* ) , if the reddening is actually closer to 0.1 mag , then there is even greater tension between the reflection and cf results for @xmath5 . @xcite and
@xcite similarly find that the cf method suggests lower spin parameters for narrow - line seyfert 1s than the nearly maximal spin typically found for this agn subclass via x - ray reflection .
our study highlights one particular strength of the x - ray reflection method for nearby seyfert galaxies . for ngc 3783 , with a bh mass of @xmath6@xmath56 m _ @xmath19 , the inability to probe the extreme uv prevents us from obtaining a very precise estimate of @xmath5 .
however , we point out that recent work by @xcite casts doubt on whether the fe k@xmath0 line gives any information on @xmath5 for one of the agn in the reflection sample ( see also 1 ) .
nearly half ( 12 ) of the 25 agn with spin measurements from the reflection method have @xmath57 @xcite .
given that the cf method suggests lower spin for two cases with near - maximal reflection spin estimates ( 1h 0707@xmath58495 in @xcite and ngc 3783 presented here ) , these high - spin cases would be good candidates for further comparisons between the reflection and cf methods , especially those with even lower @xmath2 than ngc 3783 , whose ad seds would peak further into the soft x - ray .
there is also a new method proposed by @xcite , based on microlensing , that could be included in future comparisons of spin estimation techniques . as more and better x - ray measurements allow the reflection sample to grow and as better constraints on @xmath2 ( see e.g. , * ? ? ?
* ; * ? ? ?
* ) allow the cf method to more precisely determine @xmath5 for larger samples of agn , there will be a larger population where both methods can be properly applied and compared .
if such comparisons yield good agreement , then each method can be more confidently applied to the samples they are best suited for nearby seyferts for the x - ray reflection method and higher redshift quasars for the cf method .
if instead , these comparisons bring further tensions to light , then the assumptions underlying these methods may need to be revisited .
we thank the referee for helpful feedback .
we thank paulina lira , julie hlavacek - larrondo , and helen russell for useful discussion .
dmc and dh acknowledge support from a natural sciences and engineering research council of canada discovery grant and a fonds de recherche du qubec nature et technologies nouveaux chercheurs grant .
gwf acknowledges support from universit paris - saclay s idex program and loffice franco - qubcois pour la jeunesse . | angular momentum , or spin , is a fundamental property of black holes ( bhs ) , yet it is much more difficult to estimate than mass or accretion rate ( for actively accreting systems ) . in recent years , high - quality x - ray observations have allowed for detailed measurements of the fe k@xmath0 emission line , where relativistic line broadening allows constraints on the spin parameter ( the x - ray reflection method ) .
another technique uses accretion disk models to fit the agn continuum emission ( the continuum - fitting , or cf , method ) .
although each technique has model - dependent uncertainties , these are the best empirical tools currently available and should be vetted in systems where both techniques can be applied .
a detailed comparison of the two methods is also useful because neither method can be applied to all agn .
the x - ray reflection technique targets mostly local ( z @xmath1 0.1 ) systems , while the cf method can be applied at higher redshift , up to and beyond the peak of agn activity and growth . here
, we apply the cf method to two agn with x - ray reflection measurements . for both the high - mass agn , h1821 + 643 , and the seyfert 1 , ngc 3783
, we find a range in spin parameter consistent with the x - ray reflection measurements .
however , the near - maximal spin favored by the reflection method for ngc 3783 is more probable if we add a disk wind to the model .
refinement of these techniques , together with improved x - ray measurements and tighter bh mass constraints , will permit this comparison in a larger sample of agn and increase our confidence in these spin estimation techniques . |
in this paper , we consider a simple problem , namely the solution of the scalar wave equation @xmath0 subject to homogeneous initial conditions @xmath1 in the exterior of the unit sphere . here ,
@xmath2 denote the spherical coordinates of a point in @xmath3 with @xmath4 .
standard textbooks on mathematical physics ( such as @xcite ) present exact solutions for the time - harmonic cases governed by the helmholtz equation , but generally fail to discuss the difficulties associated with the fully time - dependent case .
as we shall see , it is a nontrivial matter to develop closed - form solutions , and a surprisingly subtle matter to develop solutions that can be computed without catastrophic cancellation . in this paper
, we restrict our attention to boundary value problems with dirichlet or robin conditions .
we consider the dirichlet problem first , and assume we are given data on the boundary of the unit sphere of the form : @xmath5 it is natural to begin by expanding both @xmath6 and @xmath7 in terms of spherical harmonics .
@xmath8 where @xmath9 @xmath10 is the standard legendre polynomial of degree @xmath11 , and the associated legendre functions @xmath12 are defined by the rodrigues formula @xmath13 we let @xmath14 and @xmath15 denote the laplace transforms of @xmath16 and @xmath17 : @xmath18 @xmath19 it is straightforward @xcite to see that @xmath14 satisfies the linear second order ordinary differential equation ( ode ) @xmath20 \hunm(r , s ) = 0,\ ] ] for which the decaying solution as @xmath21 is the modified spherical hankel function @xmath22 .
it follows that @xmath23 matching boundary data on the unit sphere , we have @xmath24 , and @xmath25 the remaining difficulty is that we have an explicit solution in the laplace transform domain , but we seek the solution in the time domain . for this , we write the right hand side of in a form for which the inverse laplace transform can carried out analytically .
first , from @xcite , we have @xmath26 where @xmath27 ( @xmath28 ) are the simple roots of @xmath29 lying on the open left half of the complex plane .
thus , @xmath30 where the second equality follows from an expansion using partial fractions and the coefficients @xmath31 are given from the residue theorem by the formula : @xmath32 substituting into , we obtain @xmath33 taking the inverse laplace transform of both sides , we have @xmath34 this involves the use of the convolution theorem and the formulas @xmath35 and @xmath36 , where @xmath37 is the heaviside function .
wilcox @xcite studied the solution of the scalar wave equation and derived formula in 1959 . in that short note , wilcox stated that the coefficients @xmath31 given by grew slowly based on the claim that @xmath38 as @xmath39 .
unfortunately , this estimate is incorrect .
in fact , even after multiplication by the exponentially decaying factor @xmath40 , the coefficients @xmath31 ( @xmath28 ) grow exponentially fast as @xmath39 . in the next section , we explain this growth in detail . as a result ,
even though is very convenient for the purpose of theoretical studies , it can not be used for numerical calculation due to catastrophe cancellation in carrying out the summation .
benedict , field and lau @xcite have recently developed algorithms for compressing the kernel , which they call the teleportation kernel , arising in sphere - to - sphere propagation of data both for the standard wave equation as well as wave equations arising in linearized gravitational theories .
for the wave equation their compressed kernels can be used to perform the same function as our solution of the dirichlet problem .
the largest value of @xmath11 considered in @xcite is @xmath41 .
it is as yet unclear if useful compressions for much larger values of @xmath11 can be constructed using their methods .
we first show that the coefficients @xmath31 ( @xmath28 ) defined in grow exponentially as @xmath39 , for fixed large @xmath42 .
indeed , lemma [ lem6.1 ] in section [ sec : num ] shows that the zeros @xmath27 of @xmath29 satisfy the estimates : @xmath43 for all @xmath44 and @xmath45 .
thus when @xmath42 is large , we have @xmath46 where the last line follows from stirling s formula @xmath47 .
we have computed @xmath48 for @xmath49 using , and plotted them in figure [ fig2.1 ] for @xmath50 , clearly exhibiting the exponential growth of @xmath51 .
we also plot @xmath52 as a function of @xmath44 for a fixed value of @xmath11 in fig .
[ fig2.2 ] . _ for increasing values of @xmath11 , with @xmath50 . ,
height=240 ] as a function of @xmath44 , for @xmath53 and @xmath50 .
, height=240 ] from the preceding analysis , it is clear that one can not use as stated , since the desired solution is @xmath54 and catastrophic cancellation will occur in computing @xmath55 from exponentially large intermediate quantities .
fortunately , even though @xmath56 grows exponentially as @xmath11 increases , we can rewrite in the form of a convolution , which involves much more benign growth : @xmath57 where the convolution kernel @xmath58 is defined by the formula @xmath59 if we write @xmath60 then from , we have @xmath61 where @xmath62 is the modified bessel function of the second kind .
the last expression follows from the fact that @xmath63 .
the convolution kernel @xmath64 and its laplace transform @xmath65 are plotted in figs .
[ fig2.3 ] and [ fig2.4 ] , respectively . as a function of @xmath66 for @xmath53 and @xmath50 .
the left - hand plot shows @xmath64 for @xmath67 $ ] , the middle plot shows the same function on @xmath68 $ ] , and the right - hand plot shows the function on @xmath69$].__,height=240 ] of the convolution kernel is plotted on the imaginary axis over the range @xmath70 $ ] for @xmath53 and @xmath50 .
the red ( lower ) curve corresponds to the real part of @xmath65 and the blue ( upper ) curve corresponds to its imaginary part.,height=240 ] the following lemma shows that the convolution kernel grows only quadratically as a function of @xmath11 at @xmath71 .
numerical experiments ( see fig .
[ fig2.3 ] ) suggest that @xmath64 is maximal in magnitude at @xmath71 .
thus , while the sum of exponential expression ( [ eq2.14 ] ) involves catastrophic cancellation , the function @xmath64 is , itself , well - behaved and we may seek an alternative method for the evaluation of the convolution integral .
let @xmath4 .
then @xmath72 by the initial value theorem for the laplace transform , @xmath73 the first equality in follows from . from @xcite ( formula 9.7.2 on page 378 ) , we have the asymptotic expansion @xmath74 where @xmath75 . substituting and into
, we obtain @xmath76 the result now follows from the fact that @xmath77 . despite the fact that @xmath56 grows exponentially with @xmath11 ,
shows that the _ sum of weights _
@xmath31 is only @xmath78 for fixed @xmath42 .
still , however , the formula can not be used in practice because of catastrophic cancellation in the summation @xmath79 thus , we will need a different representation for the convolution operator @xmath80 which is suitable for numerical computation . to obtain a stable formula , we note first that we may rewrite in the form : @xmath81 we then use to express @xmath82 as @xmath83 we can , therefore , compute @xmath84 recursively : @xmath85 and , finally , @xmath86 numerical experiments indicate that the above recursion is stable if the zeros @xmath27 of @xmath29 are arranged in ascending order according to their real parts , i.e. , @xmath87 is closest to the negative real axis and @xmath88 is closest to the imaginary axis .
alternatively , it is easy to show that the functions @xmath89 ( @xmath28 ) are the solutions to the following first order system of ordinary differential equations ( odes ) with zero initial conditions .
@xmath90 where @xmath91 is a column vector of length @xmath11 with the @xmath44th entry being @xmath89 , @xmath92 , @xmath93 are @xmath94 constant matrices defined by the formulas @xmath95 and @xmath96 is a column vector of length @xmath11 whose only nonzero entry is @xmath97 .
the ode system can actually be solved analytically .
that is , one may multiply both sides of by @xmath98 to obtain @xmath99 where @xmath100 .
it is clear that @xmath101 is a constant lower triangular matrix .
one could then diagonalize the system using the eigen - decomposition @xmath102 .
this , however , is numerically unstable since @xmath101 is a highly _ nonnormal _ matrix .
thus , even though the condition number of @xmath101 is not very high ( numerical evidence shows that @xmath103 ) , @xmath104 is extremely ill - conditioned .
in fact , more detailed analysis shows that this approach leads exactly to the formula .
nevertherless , the ode system itself can be solved numerically using standard ode packages , albeit less efficiently than the explicit recursive approach we present in section [ sec : num ] , especially for high precision .
in this section , we consider the robin problem for the scalar wave equation on the unit sphere : @xmath105 with homogeneous initial data @xmath106 and the boundary condition @xmath107 it should be noted that tokita @xcite extended wilcox s analysis of the dirichlet problem to the case of robin boundary conditions of the form @xmath108 , although he assumed that @xmath109 in his discussion .
we are primarily concerned with the case @xmath110 since it arises in the solution of the full maxwell equations @xcite . as in the analysis of the dirichlet problem , we first expand @xmath111 and @xmath112 in terms of spherical harmonics , perform the laplace transform in @xmath66 , match the boundary data and obtain @xmath113 and @xmath114 we turn now to a study the properties of the kernel in , letting @xmath115 and @xmath116 recalling from [ eq2.8 ] that @xmath117 , we have @xmath118 hence , @xmath119 in particular , for @xmath120 , we have @xmath121 obviously , the poles of @xmath122 are simply the zeros of @xmath123 .
those zeros have been characterized by tokita @xcite in the following lemma .
[ lem3.1 ] _ [ adapted from @xcite . ]
_ for @xmath124 , @xmath125 has @xmath126 simple roots denoted by @xmath127 .
all the roots lie in the open left half of the complex plane symmetrically with respect to the real axis .
furthermore , they satisfy the following estimates @xmath128 @xmath129 .
hence , there exists a positive number @xmath130 such that @xmath131 for all @xmath124 and @xmath132 . from the preceding lemma , for @xmath124 we have @xmath133
one could carry out a partial fraction expansion for the right hand side of to obtain @xmath134 where the coefficients @xmath135 are given by the formula @xmath136 this would yield @xmath137 unfortunately , the coefficients @xmath135 ( @xmath138 ) behave as badly as the coefficients @xmath31 defined in for the dirichlet problem .
that is , catastrophic cancellation in makes it ill - suited for numerical computation .
fortunately , as in section [ sec : recurrence ] , we can compute @xmath139 without catastrophic cancellation using the following recurrence ( @xmath140 ) : @xmath141 with @xmath142 we leave the derivation of the recurrence to the reader .
it is possible to write down a system of odes that is equivalent to .
we omit details since the derivation is straightforward and we prefer the recurrence for numerical purposes in any case .
in order to carry out the recurrences or , we first need to compute to compute the zeros of @xmath143 and @xmath144 .
the following lemma provides asymptotic approximations of the zeros of these two functions , which we will use as initial guesses followed by a simple newton iteration . in practice
, we have found that six newton steps are sufficient to achieve double precision accuracy for @xmath145 .
[ lem6.1 ] _
( asymptotic distribution of the zeros of @xmath143 and @xmath144 , adapted from @xcite ) ; see also the appendix . _ 1 .
the zeros of @xmath146 have the following asymptotic expansion @xmath147 uniformly in @xmath44 , where @xmath148 is defined by the formula @xmath149 @xmath150 is the @xmath44th negative zero of the airy function @xmath151 whose asymptotic expansion is given by the formula @xmath152 and @xmath153 is obtained from inverting the equation @xmath154 where the branch is chosen so that @xmath155 is real when @xmath156 is positive imaginary . in other words , @xmath153 lies on the curve whose parametric equation is @xmath157 where @xmath158 $ ] and @xmath159 is the positive root of @xmath160 .
the zeros of @xmath161 have the asymptotic expansion @xmath162 uniformly in @xmath44 , where @xmath163 is defined by the formula @xmath164 and @xmath165 is the @xmath44th negative zero of the first derivative of the airy function @xmath166 whose asymptotic expansion is given by the formula @xmath167 and @xmath168 is defined as in with @xmath155 replaced by @xmath169 .
figure [ fig6.1 ] shows the zeros of @xmath170 , @xmath171 , @xmath172 and @xmath173 . , @xmath171 , @xmath172 and @xmath173 .
the zeros of @xmath29 are marked by red @xmath174 , and the zeros of @xmath123 are marked by blue @xmath175 .
_ , height=288 ] we now present a high - order discretization scheme for computing @xmath84 and @xmath139 .
we will only discuss the computation of @xmath84 in detail , since the treatment of @xmath139 is analogous . recall that the relevant recurrence relations are and .
to proceed , we first introduce the auxillary functions @xmath176 then , becomes @xmath177 it is easy to check that @xmath178 satisfies the recurrence relation @xmath179 thus , we need only consider the calculation of the integral over @xmath180 $ ] . for this , we interpolate @xmath181 by a polynomial of degree @xmath182 with the shifted and scaled legendre nodes as interpolation nodes
. that is , @xmath183 where @xmath184 ( @xmath185 ) are the standard legendre nodes on @xmath186 $ ] and @xmath187 is the @xmath188 entry of the matrix converting function values to the coefficients of a legendre expansion .
substituting into the integral on the right side of and simplifying , we obtain @xmath189 where the coefficients @xmath190 ( @xmath185 ) are given by the formula @xmath191 substituting into , we obtain @xmath192 in order to be able to use , we need to calculate @xmath193 . for this , we can again apply the recurrence and obtain @xmath194 where the coefficients @xmath195 , for @xmath196 , are given by the formula @xmath197 in summary , the algorithm for computing @xmath198 proceeds in two stages : a precomputation stage and a time - marching stage . for spherical harmonic order @xmath11 , time step @xmath199 , and desired order of accuracy @xmath200 , precompute the coefficients needed in the marching scheme for @xmath84 . compute and
store the zeros @xmath27 ( @xmath28 ) of @xmath29 .
compute the legendre nodes @xmath184 ( @xmath185 ) and the @xmath201 matrix @xmath6 which converts function values to the coefficients of the corresponding legendre expansion .
compute the integrals @xmath202 . compute and store the coefficients @xmath203 defined in .
compute the integrals @xmath204 . compute and store the coefficients @xmath205 defined in . compute and
store @xmath206 .
compute and store @xmath207 . given @xmath11 , @xmath42 , @xmath208 , the desired order of accuracy @xmath200 , and the number of desired time steps @xmath209 ,
compute the spherical harmonic mode @xmath84 at @xmath210 defined by .
set @xmath211 .
set @xmath212 .
evaluate the boundary data @xmath213 by computing the spherical harmonic transform of the dirichlet data @xmath7 , and set @xmath214 .
use to compute @xmath215 use to update @xmath216 . set @xmath217 .
compute @xmath218 .
compute @xmath219 . for each spherical harmonic mode , it is easy to see that the precomputation cost is @xmath220 and the marching cost is @xmath221 , where @xmath200 is the desired order of accuracy and @xmath209 is the total number of time steps . thus , if we truncate the spherical harmonic expansion order at @xmath222 , then the precomputation cost is @xmath223 and the marching cost is @xmath224 .
the cost of computing the spherical harmonic transform of the boundary data at all times is @xmath225 and the cost of the inverse spherical harmonic transform at the final time is @xmath226 .
summing all these factors up , we observe that the total computational cost of our algorithm is @xmath224 .
we have implemented the above algorithm in fortran for both the dirichlet and robin problems governed by the scalar wave equation . to test its convergence and stability , we consider the exact solution @xmath227 with @xmath228 , @xmath229 , @xmath230 , @xmath231 , and @xmath232 , @xmath233 , @xmath234 , @xmath235 .
the numerical solution is computed on a sphere of radius @xmath236 at @xmath237 .
tables [ tab8.1]-[tab8.4 ] show the relative @xmath238 error of the numerical solution of the scalar wave equation for varying values of @xmath222 , the order of the spherical expansion and total number of time steps .
note that the solution is oscillatory in both space and time , so that finite difference or finite element methods would have difficulty computing the solution in the far field with high precision because of numerical dispersion errors . in tables
[ tab8.1 ] and [ tab8.3 ] , the order of temporal convolution is @xmath239 and we break the time interval @xmath240 $ ] into @xmath241 equispaced subintervals ( yielding a total of @xmath242 discretization points in time ) . in tables
[ tab8.2 ] and [ tab8.4 ] , we use @xmath243 terms in the spherical harmonic expansions .
these tables show that numerical solution converges spectrally fast to the exact solution .
.relative @xmath238 error of the numerical solution of the dirichlet problem with increasing spherical harmonic expansion order @xmath222 .
@xmath244 is the total number of discretization on the unit sphere .
since the discretization error is usually greater than the truncation error , @xmath245 is chosen to be @xmath246 .
thus @xmath247 .
the total number of discretization points in time is @xmath248 . [
cols="^,^,^,^,^,^,^",options="header " , ] .
the exact solution is of the same form as ( [ uexact ] ) - that is , induced by two sources in the interior of the unit sphere . [ fig8.2],height=192 ] for a true " scattering problem .
dirichlet boundary conditions are generated by two exterior sources placed on the @xmath156-axis , at @xmath249 and @xmath250 .
the left - hand plot shows the value of the boundary data at the north pole of the unit sphere as a function of time , and the right - hand plot shows the solution at the north pole of the outer sphere of radius @xmath236 .
[ fig8.3],height=192 ] for a true " scattering problem .
robin boundary conditions are generated by two exterior sources placed on the @xmath156-axis , at @xmath249 and @xmath250 .
the left - hand plot shows the value of the boundary data at the north pole of the unit sphere as a function of time , and the right - hand plot shows the solution at the north pole of the outer sphere of radius @xmath236 .
[ fig8.4],height=192 ] -plane within the annular region @xmath251 at @xmath252 . with boundary data as in fig .
[ fig8.3 ] .
note that the domain is approximately 50 wavelengths in size .
[ fig8.5],height=192 ]
we have presented an analytic solution for the scalar wave equation in the exterior of a sphere in a form that is numerically tractable and permits high order accuracy even for objects many wavelengths in size . aside from its intrinsic interest in single or multiple scattering from a collection of spheres , our algorithm provides a useful reference solution for any numerical method designed to solve problems of exterior scattering . at the present time , such codes are typically tested by fourier transformation after a long - time simulation and comparison with a set of single frequency solutions computed by separation of variables applied to the helmholtz equation . an exception is the work of sauter and veit @xcite , who make use of a formulation equivalent to that of wilcox to develop a benchmark solution for a time - domain integral equation solver which can be applied to scattering from general geometries .
exponential ill - conditioning is avoided by considering only low - order spherical harmonic expansions .
recently , grote and sim @xcite have also used an approach based on the local exact radiation boundary conditions proposed in @xcite to develop a new hybrid asymptotic / finite difference formalism for multiple scattering in the time domain .
the advantage of the grote - sim method is that spherical harmonic transformations are unnecessary and the evaluation formulas can be localized in angle .
however , they also restrict their attention to low - order expansions , and our preliminary experiments using their formulas indicate a loss of conditioning for @xmath11 large .
( the loss of conditioning presumably also applies to the radiation boundary conditions in @xcite . )
the method developed here should be of immediate use in both contexts as implemented above , our algorithm has @xmath253 complexity .
it is possible , however , to reduce the cost to @xmath254 .
this requires the use of a fast spherical harmonic transform ( see , for example , @xcite and references therein ) . with this fast algorithm ,
the cost of each spherical harmonic transform is reduced from @xmath226 to @xmath255 .
second , we believe that the convolution kernels can be compressed as in @xcite , so that they involve only @xmath256 modes for each @xmath11 for a given precision . we note that compressions for @xmath257 and various radii are reported in @xcite , both for the scalar wave equation considered here ( which they call the flat - space wave equation ) and for wave equations with zerilli and regge - wheeler potentials . in the latter cases ,
compressed kernels are also constructed for smaller values of @xmath11 , as the exact kernels do not have rational transforms .
tabulated coefficients required for implementing the compressed kernels may be found online @xcite .
for the extension of the present approach to the full maxwell equations , see @xcite .
software implementing the algorithm of the present paper will be made available upon request .
an alternative analysis of the instability phenomenon can be carried out using the uniform asymptotic expansions of the bessel functons due to olver @xcite .
we first recall the relationship between @xmath62 and the hankel function , @xmath259 : @xmath260 thus the residues we wish to estimate are given by @xmath261 to approximate these for @xmath262 we use ( see @xcite ) : @xmath263 which hold uniformly in @xmath264 ; thus in particular they hold in @xmath265 where we will be using them . here
@xmath155 is given by ( [ eq6.4 ] ) with the replacement @xmath266 .
i. : : @xmath267 has infinitely many zeros which lie on the negative real axis .
for @xmath44 large the jth zero , @xmath268 , of @xmath267 satisfies ( [ eq6.3 ] ) and the derivative satisfies @xmath269 ii .
: : for @xmath270 the function @xmath267 satisfies the asymptotic formula @xmath271 using ( i. ) we deduce that the poles , @xmath272 , are asymptotically given by ( [ eq6.1 ] ) and approximately lie on the curve @xmath273 where @xmath274 is defined in ( [ eq6.5 ] ) .
this is the curve for which @xmath275 is real and negative . to evaluate the residues we must calculate using ( [ ash]),([asdh ] ) @xmath276 where we have introduced @xmath277 obviously the scaling @xmath278 moves @xmath279 off the curve where the argument of the airy function is real and negative . thus using ( [ aiasy ] ) and ( [ eq6.4 ] ) we deduce that the asymptotic formula for @xmath258 contains an exponential term @xmath280 \nonumber\end{aligned}\ ] ] where @xmath281 here we have introduced @xmath282 . finally , we consider the real part of the expression in parentheses on the second line of ( [ resexp ] ) .
in particular we replace @xmath283 by a continuous variable @xmath284 traversing the scaled curve , @xmath274 , containing the approximate zeros .
then the function @xmath285 depends only on @xmath42 and the coordinate describing the curve ; in particular it is independent of @xmath11 and @xmath44 . in fig .
[ resasy2 ] we plot the real part of @xmath285 scaled by @xmath286 for @xmath50 .
this can be compared with fig .
[ fig2.2 ] by scaling both axes by @xmath53 and recognizing the vertical axis as the base ten logarithm .
we then observe good agreement with the numerical results .
the maximum value plotted in figure [ resasy2 ] is approximately @xmath287 , which is the predicted slope of the straight line plotted in fig .
[ fig2.1 ] .
again the agreement is good .
we note that increasing @xmath42 makes the problem somewhat worse ; the scaled maximum real part is approximately @xmath288 for @xmath289 and @xmath290 for @xmath291 . | we derive new , explicit representations for the solution to the scalar wave equation in the exterior of a sphere , subject to either dirichlet or robin boundary conditions .
our formula leads to a stable and high - order numerical scheme that permits the evaluation of the solution at an arbitrary target , without the use of a spatial grid and without numerical dispersion error . in the process , we correct some errors in the analytic literature concerning the asymptotic behavior of the logarithmic derivative of the spherical modified hankel function .
we illustrate the performance of the method with several numerical examples . 65m70 , 78a40 , 78m16 |
the first nucleus to exhibit ` real ' nuclear properties is the three - body nucleus .
there are only two stable @xmath14 nuclei , @xmath0 and @xmath1 ; neither member of this isospin doublet has any known excited state .
the ground state of @xmath0 has one proton and two spin - paired neutrons , while that of @xmath1 has one neutron and two spin - paired protons .
we refer to the neutrons in @xmath0 and the protons in @xmath1 as the ` even ' nucleons , and to their _ rms _ radii as the even radii for @xmath0 and @xmath1 , respectively .
similarly , the proton in @xmath0 and the neutron in @xmath1 are called ` odd ' nucleons and their _ rms _ radii shall be referred to as odd radii @xcite .
charge symmetry requires that , in the absence of coulomb effects and mass differences , these nuclei be identical .
this is because @xmath0 and @xmath1 are mirror nuclei and are related by a @xmath15 rotation in isospin space , or to put it more simply , one nucleus changes into the other by interchange of neutrons and protons .
there are four interactions available between the two charged pions and @xmath0 and @xmath1 target nuclei .
these are grouped conveniently in ` isospin ' pairs @xmath16 and @xmath17 , @xmath18 and @xmath19 which in turn are related to each other by the interchange of protons with neutrons and @xmath20 with @xmath21 .
_ in the absence of the coulomb force
_ , if the strong interaction is charge symmetric , scattering cross sections for the members of each of these pairs should be equal .
if we make the assumption that an incident pion interacts with a single target nucleon , then we can look at these interactions in terms of @xmath4-nucleon scattering amplitudes . at 180 mev ( near the peak of the @xmath22 resonance ) the @xmath23 ( @xmath24 ) scattering amplitude is about three times as large as the @xmath25 ( @xmath26 . ) consequently , the elastic cross sections for @xmath27 and @xmath28 are nine times as large as those for @xmath29 and @xmath30 .
the @xmath4-n scattering amplitudes can be broken down into _ spin - flip _ and _ non - spin - flip _ components . for a fixed target nucleon and pions of 180 mev ,
the _ spin - flip _ amplitude has a sine dependence on the scattering angle ; the _ non - spin - flip _ amplitude has a cosine dependence .
since no excited states are available , single _ spin - flip _ scattering from one of the even nucleons would leave both nucleons with the same spin and is thus forbidden by the pauli principle .
( see ref .
@xcite for further discussion on this point . )
however , at the large scattering angles @xmath31 for which our most recent data were obtained , kinematic considerations require that one consider double- or multiple - scattering effects
. this would open the door to double spin - flip scattering in the even nucleons which would not violate the pauli principle .
the description of the whole picture is discussed in the accompanying article of kudryavtsev _ et al . _
pion - nuclear amplitudes , discussed in @xcite , are a superposition of a single- and double - scattering of pions from the nucleons in the nucleus with fermi motion taken into account .
principal sources of the violation of charge symmetry in @xcite are ( i ) the coulomb interaction between the charged pions and the nuclei ( the external coulomb effect ) , ( ii ) n - p doublet and the mass splitting of the different charged states of the @xmath32-isobar , and ( iii ) the difference between the wave functions of @xmath33 and @xmath34 due to the additional coulomb repulsion between the two protons in the @xmath34 nucleus ( the internal coulomb effect . )
several ratios of scattering cross sections have been defined in conjunction with our earlier measurements @xcite in order to highlight certain features of the @xmath35 interaction .
the first two are the _ charge - symmetric _ ratios @xmath36 and @xmath37 the numerator and denominator of each pair are related by an exchange of protons with neutrons and @xmath20 with @xmath21 .
therefore , the strong interactions in the denominator and numerator are equal , and these ratios should be identically one at all energies and four - momentum transfers squared . however , since the form factor of @xmath1 is decreased owing to the coulomb repulsion between its protons , the cross section in the denominator is reduced and we expect that the ratios @xmath38 and @xmath39 will be somewhat greater than one @xcite . in @xmath38 ,
scattering from the odd nucleon dominates in the delta energy region , and so the scattering is a mixture of spin - flip and non - spin - flip . in @xmath39 , scattering from the even nucleon dominates , and so spin - flip scattering is suppressed .
the next ratio is @xmath13 , the `` superratio . ''
it is defined as the product of @xmath38 and @xmath39 , @xmath40 since @xmath38 and @xmath39 are charge symmetric , so is @xmath13 . finally , we define @xmath41 and @xmath42 these ratios are not charge symmetric , but , as shall be seen below
, several experimental uncertainties cancel when calculating @xmath43 and @xmath44 , and they can be used to derive the charge - symmetric @xmath45 where @xmath46 in section [ sec : expt ] , we briefly discuss the analysis of our data at energies spanning the @xmath22 resonance . the resulting charge - symmetric ratios @xmath38 and @xmath39 , non charge - symmetric ratios @xmath47 and @xmath48 , and superratio @xmath13
are presented in section [ sec : ratio ] . in section [ sec:90deg ] , we discuss the situation for angles greater than @xmath49 .
finally , we summarize our findings in section [ sec : concl ] .
the experimental details have been given by matthews _
et al . _
@xcite . here
, we briefly discuss the analysis of the data and the relevant experimental parameters for the determination of the scattering ratios .
[ tab : values ] [ cols="^,^,^,^,^,^,^,^",options="header " , ] the ratios @xmath38 and @xmath39 are extracted as @xmath50 and @xmath51 where @xmath52 @xmath53 refers to the scattering yield , and @xmath54 and @xmath55 are the number of scattering centers per unit volume in the @xmath0 and @xmath1 samples , respectively .
these latter were measured very accurately @xmath5 see ref .
@xcite for details .
elastic scattering yields from @xmath56 are used to scale the other yields to the known @xmath57 cross sections .
writing the ratio @xmath58 as @xmath59 , and defining @xmath60 as the yield per target nucleon @xmath61 @xmath62 , we have @xmath63 and similarly @xmath64 then @xmath65 all normalization quantities not related to @xmath54 and @xmath55 cancel in @xmath13 . in the definitions of @xmath38 and @xmath39 , it is the _ ratio _ of the @xmath57 yields and cross sections that appears . finally , we consider the ratios @xmath43 and @xmath66 . since the same charge of pion appears in both the numerator and denominator of each of these ratios , the non - target - related normalization quantities cancel here as well . then we have @xmath67 and @xmath68 so that @xmath69
values for all of the ratios measured in this experiment are given in table [ tab : values ] .
figure [ fig : r1r2 ] shows the angular distributions for the simple charge - symmetric ratios @xmath38 , @xmath39 , and the superratio @xmath13 at 142 , 180 , 220 , and 256 mev .
we see that @xmath38 is flat and structureless in the forward hemisphere but rises in the backward hemisphere where it remains high to @xmath15 .
the ratio @xmath39 shows structure at about @xmath70 in the forward hemisphere , approaches 1.0 in the backward hemisphere and stays there to @xmath15 .
most of the structure in r is therefore due to @xmath39 .
we note that , aside from the region near @xmath71 in the @xmath72 center - of - mass kinematics , which corresponds to @xmath73 in the @xmath74 center of mass where the non - spin - flip scattering amplitudes have zeros , the charge - symmetric scattering ratios do not have any sharp features . indeed ,
in the backward hemisphere they are fairly flat and quite smooth .
this is not surprising , since the @xmath4-nucleon amplitudes are smooth and have no zeros in this region .
the form factors for @xmath0 and @xmath1 , as measured with electron scattering , are smooth as well . finally , the interactions in the numerator and denominator for each ratio are approximately the same : primarily odd - nucleon in @xmath38 , and primarily even - nucleon in @xmath39 .
@xmath13 is the smooth product of two smooth functions . of more interest
is the general trend of the ratios as we progress from the forward to the backward hemispheres , see fig .
[ fig : r1r2cal ] .
the crossover of the two ratios , and the fact that @xmath38 is significantly different from unity but @xmath39 is consistent with unity is difficult to explain quantitatively .
it is interesting to note that the qualitative behavior of the simple ratios , including the crossover , were predicted by an optical - model calculation by gibbs and gibson @xcite before the back - angle data were taken .
one might speculate that as one passes to the backward hemisphere , with the diminishing importance of the single - scattering process , the double - scattering process provides a mechanism for spin - flip amplitudes to contribute in paired - nucleon scattering without violating the pauli exclusion principle . in the top plot of fig .
[ fig : r1r2cal ] , we assume that @xmath75 . in the bottom plot , we assume the 1.5% asymmetry suggested by smith _
et al . _
@xcite for @xmath76 , which means correcting the value of @xmath59 by 3% .
the effect is to _ increase _ the difference between @xmath38 and @xmath39 in the backward hemisphere .
figure [ fig : rprm ] shows results for @xmath43 and @xmath66 .
these ratios have very small error bars , owing to the cancellation of normalization quantities mentioned above .
the steep rise of @xmath66 and the fall below one of @xmath43 are indications that there is a steep rise in the 180-mev even - nucleon - dominated cross sections at large angles @xcite . however , the error bars for these ratios are much smaller than those for the cross sections .
these ratios are the most sensitive test of the large - angle scattering calculations ; in particular , these test the form factors used in the calculation .
figure [ fig : superratio ] shows @xmath13 at 180 mev .
the shape is derived from the two simple ratios , the bump at @xmath70 corresponding to the bump in @xmath39 and the steady rise in the backward hemisphere to the rise in @xmath38 .
figure [ fig : superratio ] also shows the calculations by kudryavtsev _ et al _
@xcite and by gibbs and gibson @xcite . in the latter the shape of the superratio was used to extract differences in the odd and even nucleon radii more precisely than is possible from existing electron scattering data .
gibbs and gibson s optical - model calculation shows a strong dependence on two parameters , the difference between the _ rms _ neutron radius in tritium and the _ rms _ proton radius in @xmath1 ( that is , the difference in the even - nucleon radii ) , and the difference between the _ rms _ proton radius in @xmath0 and the _ rms _ neutron radius in @xmath1 . in the approach of ref .
@xcite , a successful description of @xmath13 at t@xmath77 = 180 mev is based not only on the difference in the wave functions of @xmath0 and @xmath1 , but also on the @xmath32 mass splitting as well as on the inclusion of the double - scattering interaction of the pion with nucleons of the target nuclei . both models account for the role reversal of the @xmath78 and @xmath79 at back angles for the @xmath80 = 180 mev
. however , kudryavtsev _ et al .
_ fitted @xmath78 and @xmath79 to determine r , while gibbs and gibson fitted their calculations to r and inferred @xmath78 and @xmath79 prior to the time that the data were obtained .
although the leading terms included in the calculation of kudryavtsev _ et al .
_ reproduce the main features of the back - angle ratios at energies of 180 mev and above , yet another factor which might contribute to this remarkable role reversal of @xmath78 and @xmath79 at back angles is a two - step process consisting of the formation of a virtual delta by the interaction of the incident pion with a nucleon , followed by the scattering of this delta on the remaining correlated nucleon pair .
this process is clearly overshadowed by single scattering at angles near the non - spin - flip dip , but , like any two - step process , could become important as the momentum transfer increases . for @xmath78
, the dominant channels for delta formation would be @xmath81 in the numerator and @xmath82 in the denominator , whereas the reverse would be the case for @xmath79 .
[ for @xmath78 the correlated pairs would be ( nn ) in the numerator and ( pp ) in the denominator , and for @xmath79 they would be ( np ) pairs in both numerator and denominator . ] since the width of the @xmath83 is somewhat smaller than that of the @xmath84 ( about 5 mev out of a total width of about 120 mev ref .
@xcite ) , the lifetime of the former would be longer than that of the latter , and likewise its interaction cross section with the ( nn ) pair would be larger .
thus , at back angles , where this effect might become increasingly important , it would have the effect of increasing @xmath78 and decreasing @xmath79 , which corresponds to their observed behavior .
figure [ fig : r1r2back ] shows @xmath38 , @xmath39 , and @xmath13 at the large scattering angles ( @xmath85 ) for all pion energies plotted versus the four - momentum transfer squared . in fig .
5a , @xmath38 increases steadily , while in fig 5b , @xmath39 decreases slightly with -t .
@xmath13 displays a very slight increase with the four - momentum transfer squared , as shown in fig .
5c . from fig .
[ fig : r1r2back ] one can conclude that , since @xmath38 , @xmath39 , and @xmath13 , coming from different energy sets , fall on top of each other and follow the same general trend , these ratios are primarily functions of @xmath12 .
the behavior of @xmath43 and @xmath66 at the largest scattering angles versus @xmath86 can be analyzed with the help of table 1 . as is seen in the table
, @xmath87 decreases sharply from 142 to 180 mev , then rises slightly through 220 and 256 mev .
@xmath66 shows the opposite behavior , with a maximum at 180 mev .
the excitation functions for @xmath43 and @xmath88 for backward angles versus @xmath12 are shown in fig .
[ fig : rho1rho2back ] . as is the case for @xmath78 , @xmath79 , and r ,
the agreement of the overlaid data at different energies is also quite good .
we note that in the forward hemisphere , a model that assumes single @xmath74 scattering explains the behavior of elastic @xmath72 scattering quite nicely , and the reaction is completely defined by @xmath72 kinematics .
however , for angles greater than @xmath89 , kinematic considerations force us to consider two - step processes , especially double scattering to explain @xmath72 elastic scattering and all of the ratios are seen to be smooth functions of the momentum transfer .
several new measurements of the ratios @xmath38 , @xmath39 , and @xmath13 at energies @xmath86 = 142 , 180 , 220 , and 256 mev at backward scattering angles are presented .
these data complete our previous data sets at smaller angles and the same energies @xcite . where data sets overlap they are consistent with each other . at @xmath90 , the charge - symmetric ratios @xmath38 , @xmath39 , and @xmath13 are smooth functions of the scattering angle in the backward hemisphere . the ratios @xmath38 and @xmath39 cross each other at around @xmath91 ; @xmath38 becomes significantly different from 1.00 at backward angles while @xmath79 , which had been greater than 1.00 at forward angles , approaches unity .
deviation of the ratios @xmath38 , @xmath39 , and @xmath13 from unity gives evidence for charge - symmetry - violation effects in these reactions .
all three ratios , at energies of 180 , 220 , and 256 mev , are well described by the theoretical approach of ref .
additionally , the ratios at 180 mev match well with the results of a previous calculation @xcite that determines the value of the difference between the even and odd radii of @xmath1 and @xmath0 .
it also has been shown that all three ratios @xmath38 , @xmath39 , and @xmath13 ( as well as @xmath47 and @xmath48 ) at large scattering angles ( @xmath85 ) and for all of the incident pion energies studied can be described as a function of only the four - momentum transfer squared @xmath12 in the region where two - step processes such as double scattering dominate the @xmath72 elastic scattering .
the authors acknowledge the support of the us national science foundation , the us department of energy , and the gw research enhancement fund . we wish to thank r. boudrie , c. morris , and s. dragic for their assistance during the running of the experiment and a. e. kudryavtsev and i. i. strakovsky for a careful reading of the manuscript and for fruitful discussions .
w. j. briscoe , b. l. berman , r. w. caress , k. s. dhuga , s. dragic , d. knowles , d. macek , s. k. matthews , n. j. nicholas , m. f. taragin , d. b. barlow , b. m. k. nefkens , c. pillai , j. price , l. d. isenhower , m. e. sadler , s. j. greene , i. slaus , and i. supek , nucl .
* a553 * , 585 ( 1993 ) .
s. k. matthews , w. j. briscoe , c. bennhold , b. l. berman , r. w. caress , k. s. dhuga , s. n. dragic , s. s. kamalov , n. j. nicholas , m. f. taragin , l. tiator , d. b. barlow , b. m. k. nefkens , c. pillai , j. w. price , l. d. isenhower , m. e. sadler , i. laus , and i. supek , phys .
c * 51 * , 2534 , 1995 ; and s. k. matthews , ph.d .
thesis , the george washington university , 1992 ( unpublished ) .
g. r. smith , d. r. gill , d. ottwell , g. d. wait , p. walden , r. r. johnson , r. olszewski , r. rui , m. e. sevior , r. p. trelle , j. brack , j. j. kraushaar , r. a. ristinen , h. chase , e. l. mathie , v. pafilis , r. b. shubank , n. r. stevenson , a. rinat , and y. alexander , phys .
c * 38 * , 240 ( 1988 ) . | we have measured the elastic - scattering ratios of normalized yields for charged pions from @xmath0 and @xmath1 in the backward hemisphere . at 180 mev
, we completed the angular distribution begun with our earlier measurements , adding six data points in the angular range of @xmath2 to @xmath3 in the @xmath4 -nucleus center of mass .
we also measured an excitation function with data points at 142 , 180 , 220 , and 256 mev incident pion energy at the largest achievable angle for each energy @xmath5 between @xmath6 and @xmath7 in the @xmath8-nucleus center of mass .
this excitation function corresponds to the energies of our forward - hemisphere studies .
the data , taken as a whole , show an apparent role reversal of the two charge - symmetric ratios @xmath9 and @xmath10 in the backward hemisphere . also , for data @xmath11 we observe a strong dependence on the four - momentum transfer squared @xmath12 for all of the ratios regardless of pion energy or scattering angle , and we find that the superratio @xmath13 data match very well with calculations based on the forward - hemisphere data that predicts the value of the difference between the even - nucleon radii of @xmath0 and @xmath1 .
comparisons are also made with recent calculations incorporating different wave functions and double scattering models . |
motivated by problems arising from dna sequencing and the properties of circle graphs of 2-in , 2-out digraphs , arratia , bollobs and sorkin introduced a new family of graph invariants , the interlace polynomials , in @xcite .
these invariants may be defined either through recursive elimination of vertices or as sums indexed by subsets of vertices @xcite , much as the tutte polynomial may be defined either through recursive elimination of edges or as a sum indexed by subsets of edges .
ellis - monaghan and sarmiento @xcite have shown ( among other results ) that one of the one - variable interlace polynomials , @xmath3 , can be computed in polynomial time for bipartite distance hereditary graphs .
their proof depends on the corresponding result for the tutte polynomials of series - parallel graphs , first proved in @xcite , and the fact that bipartite distance hereditary graphs are circle graphs of euler circuits in medial graphs of series - parallel graphs . in this paper
we discuss several useful features of interlace polynomials that have been modified to incorporate vertex weights . after defining the weighted interlace polynomials in section 2
, we observe that a simple adjustment of weights makes it unnecessary to have the third term in the fundamental recursion @xmath4 of @xcite . in section 3
we present reduction formulas that can be used to eliminate a vertex that is a twin of another , or pendant on another .
these pendant - twin reductions are analogous to the series - parallel reductions of electrical circuit theory , in which two resistors wired in series ( resp .
parallel ) are replaced by one resistor with @xmath5 ( resp .
@xmath6 ) .
the pendant - twin reductions are used to extend the result of ellis - monaghan and sarmiento mentioned above to the two - variable interlace polynomials of looped , non - bipartite distance hereditary graphs . in section 4 we show that if @xmath7 is a looped graph obtained using the composition construction of cunningham @xcite then @xmath1 is equal to the interlace polynomial of a suitably re - weighted version of @xmath8 ; this generalizes results of @xcite that describe the @xmath3 polynomials of simple graphs constructed through substitution .
composition has proven useful in the study of circle graphs ( see for instance @xcite ) , so it is not surprising to see it appear in the theory of the interlace polynomials . in section 5
we discuss some elementary properties of the unweighted @xmath3 polynomial , focusing on simple ( unlooped ) graphs .
in section 6 we sketch a combinatorial description of the interlace polynomials of trees and forests introduced by anderson , cutler , radcliffe and the present author in @xcite .
this combinatorial description bears a striking resemblance to the activities description of the tutte polynomial , and in section 7 we extend it to arbitrary graphs using activities defined with respect to recursive interlace polynomial calculations .
we do not know whether or not these activities have a convenient combinatorial description in general .
we should observe that the idea of using vertex weights for interlace polynomials has appeared before , though our implementation of the idea is different from those we have seen elsewhere . in @xcite
, courcelle introduced a multivariate interlace polynomial that is more complicated than the polynomials we consider here , and involves assigning indeterminates to the vertices of a graph .
he used monadic second - order logic to show that it is possible to compute bounded portions of this polynomial ( and the entire unweighted interlace polynomial @xmath9 ) in polynomial time for graphs of bounded clique - width .
this technique is quite general but involves large built - in constants , so for pendant - twin reductions and compositions the formulas presented here are considerably more practical . also , blser and hoffmann @xcite use the idea of assigning indeterminates to vertices , along with the adjunction of two types of pendant - twin vertices , to show that evaluating interlace polynomials is generally @xmath10-hard for almost all values of the variables . in @xcite arratia , bollobs and sorkin
observe that there is a natural ( so natural it is `` practically a tautology '' ) correspondence between the kauffman bracket of an alternating link diagram and an interlace polynomial of an associated 2-in , 2-out digraph .
the situation is clear enough that we do not discuss it in detail , but it is worth mentioning that this correspondence may be extended to arbitrary link diagrams using vertex weights . the well - known relationship between the jones and tutte polynomials is similar , in that an edge - weighted or -signed version of the tutte polynomial conveniently incorporates crossing information when dealing with non - alternating links @xcite .
we recall terminology and notation of @xcite . _
graphs _ may have loops but not multiple edges or multiple loops .
the _ rank _ @xmath11 and _ nullity _ @xmath12 of a graph @xmath2 are those of its adjacency matrix considered over @xmath13 .
if @xmath14 then @xmath15 $ ] is the subgraph of @xmath2 induced by @xmath16 .
in addition , we say a graph is ( _ vertex- _ ) _ weighted _ by functions @xmath17 and @xmath18 mapping @xmath19 into some commutative ring with unity @xmath20 . for ease of notation
we prefer to denote a weighted graph @xmath2 rather than using the triple @xmath21 ; even when two graphs differ only in their weights we will denote them @xmath2 and @xmath22 rather than @xmath21 and @xmath23 .
also , if a graph is modified then unless otherwise stated , we presume that the weight functions @xmath17 and @xmath18 are modified in the most natural way . for instance , if @xmath24 then the other vertices of @xmath2 have the same vertex weights in @xmath25 as they have in @xmath2 .
if @xmath2 is a weighted graph then the _ unweighted version _ of @xmath2 is denoted @xmath26 ; it has the same underlying graph and the trivial weights @xmath27 .
[ 2varq]if @xmath2 is a vertex - weighted graph then the _ weighted interlace polynomial _ of @xmath2 is @xmath28)}(y-1)^{n(g[s])}.\ ] ] if @xmath2 is a vertex - weighted graph then the _ weighted vertex - nullity interlace polynomial _ of @xmath2 is @xmath29)}.\ ] ] if @xmath2 is a vertex - weighted graph then the _ weighted vertex - rank interlace polynomial _ of @xmath2 is @xmath30)}.\ ] ] the original , unweighted interlace polynomials of @xmath2 are recovered by using @xmath31 the three weighted interlace polynomials of a particular weighted graph @xmath2 may be substantially different from each other . considered as functions defined on the class of all weighted graphs , however , the three polynomials are essentially equivalent : if @xmath32 is obtained from @xmath2 by re - weighting @xmath33 using the indeterminates in the polynomial ring @xmath34 $ ] , then the value of any one weighted interlace polynomial on @xmath35 determines the values of all three polynomials on all weighted versions of @xmath2 . for instance , @xmath1 can be obtained from @xmath36 by substituting @xmath37 for @xmath38 , @xmath39 for @xmath40 and @xmath41 for @xmath42 .
( in contrast , the functions defined by the three unweighted interlace polynomials are rather different from one another , as there are many pairs of graphs distinguished by @xmath43 but not by @xmath3 @xcite . )
we mention all three weighted polynomials simply because one of them may be more convenient for some purposes than the others .
most of our results are stated for @xmath9 because it specializes to the others most readily . in fact
any one of @xmath44 , @xmath36 , @xmath45 determines @xmath2 up to isomorphism , because the vertex weights identify the contribution of each @xmath14 .
the looped vertices of @xmath2 appear in the 1-element subsets @xmath16 of rank 1 ( nullity 0 ) , some pairs of adjacent vertices of @xmath2 appear in the 2-element subsets @xmath16 that contain at least one unlooped vertex and have rank 2 ( nullity 0 ) , and the other pairs of adjacent vertices of @xmath2 appear in the 2-element subsets @xmath16 that contain two looped vertices and have rank 1 ( nullity 1 )
. moreover the same comment applies even if only one of @xmath46 is nontrivial , i.e. , if @xmath47 or @xmath48 .
this might make it seem possible to simplify our discussion by considering only @xmath17 or only @xmath18 , but we prefer to use both weights because they produce especially easy - to - read formulas , as noted in the discussions of theorems [ theorem1 ] and [ theorem3 ] below . setting one weight identically equal to 1 would not simply leave us with the other weight ; in essence , the remaining one would replace the ratio of the original two .
consequently , using only @xmath17 or only @xmath18 would entail unnecessary algebraic complications and losses of generality , because many formulas would require division .
for instance , with @xmath47 proposition [ prop1 ] would state that replacing @xmath49 with @xmath50 results in a graph @xmath22 with @xmath51 .
the special case @xmath52 would require a separate statement , and of course @xmath53 would be ruled out .
our first proposition follows immediately from definition [ 2varq ] .
[ prop1]suppose @xmath24 and @xmath54 .
let @xmath22 be obtained from @xmath2 by changing the weights of @xmath55 to @xmath56 and @xmath57
. then @xmath58 .
a fundamental property of the interlace polynomials is that they can be calculated recursively with the local complementation and pivot operations used by kotzig @xcite , bouchet @xcite and arratia , bollobs and sorkin @xcite .
( local complementation ) if @xmath55 is a vertex of @xmath2 then @xmath59 is obtained from @xmath2 by toggling adjacencies @xmath60 involving neighbors of @xmath55 that are distinct from @xmath55 . ( pivot )
if @xmath55 and @xmath61 are distinct vertices of @xmath2 then the graph @xmath62 is obtained from @xmath2 by toggling adjacencies @xmath60 such that @xmath63 , @xmath64 is adjacent to @xmath55 in @xmath2 , @xmath42 is adjacent to @xmath61 in @xmath2 , and either @xmath64 is not adjacent to @xmath61 or @xmath42 is not adjacent to @xmath55 .
note that local complementation includes loop - toggling ( when @xmath65 is a neighbor of @xmath55 distinct from @xmath55 ) but pivoting does not .
[ theorem1]if @xmath2 is a weighted graph then @xmath1 can be calculated recursively using the following properties .
\(a ) if @xmath55 is a looped vertex then @xmath66 \(b ) if @xmath55 and @xmath61 are loopless neighbors then @xmath67 \(c ) if @xmath2 has no non - loop edges then @xmath1 is @xmath68 the proofs of parts ( a ) and ( b ) of theorem [ theorem1 ] are essentially the same as the proofs of the corresponding formulas for the unweighted two - variable interlace polynomial @xcite . indeed ,
if we read @xmath17 as includes and @xmath18 as excludes then the weighted formulas serve as mnemonic devices to recall the proofs .
for instance part ( a ) is proven as follows .
the term @xmath69 reflects the fact that if @xmath70 then @xmath15=(g - a)[s]$ ] , so the contributions of @xmath16 to @xmath1 and @xmath71 differ only by a factor @xmath72 .
the term @xmath73 reflects the fact that if @xmath74 then @xmath75)=r \begin{pmatrix } 1 & \mathbf{1 } & \mathbf{0}\\ \mathbf{1 } & m_{11 } & m_{12}\\ \mathbf{0 } & m_{21 } & m_{22}% \end{pmatrix } = 1+r \begin{pmatrix } m_{11}^{c } & m_{12}\\ m_{21 } & m_{22}% \end{pmatrix } = 1+r((g^{a}-a)[s - a]),\ ] ] where bold numerals represent rows and columns and @xmath76 differs from @xmath77 in every entry . consequently the contributions of @xmath16 to @xmath1 and @xmath78 to @xmath79 differ by a factor @xmath80 .
the more complicated formula of part ( b ) reflects , among other things , the fact that each subset @xmath14 with @xmath81 contributes to all three terms ; the last two contributions cancel each other .
part ( c ) follows directly from definition [ 2varq ] .
it is not technically necessary to mention graphs with loops in ( c ) , as loops can always be removed using part ( a ) . however applying part ( a ) to
completely disconnected graphs is obviously inefficient .
other properties of the unweighted interlace polynomials also extend naturally to the weighted polynomials .
for instance , the second and third parts of theorem [ theorem2 ] below extend two properties discussed in section 3 of @xcite . the first part will be useful in section 4 , where we discuss substituted graphs .
[ theorem2](a ) if @xmath24 is loopless then @xmath82 \(b ) if @xmath83 are loopless neighbors then @xmath84 \(c ) if @xmath2 is the union of disjoint subgraphs @xmath85 and @xmath86 then @xmath87 . to prove ( a ) , note that if @xmath88 then the adjacency matrix of @xmath15 $ ] may be represented by @xmath89 adding the first row to each of those in the second group results in @xmath90 the adjacency matrix of @xmath91 $ ] . definition [ 2varq ] then tells us that @xmath92)}(y-1)^{n(g[s])}\\ & ~\\ & = \sum_{a\in s\subseteq v(g)}(% % tcimacro{\dprod \limits_{s\in s}}% % beginexpansion { \displaystyle\prod\limits_{s\in s } } % endexpansion \alpha(s))(% % tcimacro{\dprod \limits_{v\not \in s}}% % beginexpansion { \displaystyle\prod\limits_{v\not \in s } } % endexpansion \beta(v))(x-1)^{r(g^{a}[s])}(y-1)^{n(g^{a}[s])}\\ & ~\\ & = q(g^{a})-\beta(a)q(g^{a}-a).\end{aligned}\ ] ] the proofs of ( b ) and ( c ) are essentially the same as the proofs of the corresponding results in @xcite .
many properties of the unweighted interlace polynomials extend directly to the weighted polynomials , as we see in theorems [ theorem1 ] and [ theorem2 ] .
it may be a surprise that some properties of the unweighted interlace polynomials can be significantly simplified using vertex weights .
for instance , consider a computation tree representing a recursive implementation of theorem [ theorem1 ] .
the tree has two branches for each application of part ( a ) , three branches for each application of part ( b ) , and a leaf for each application of part ( c ) . as noted in the proof of theorem [ theorem1 ]
, the three terms of part ( b ) all incorporate contributions from the same subsets @xmath14 .
the recursive computation calculates these same contributions three times , on separate branches .
this inefficiency can be eliminated by rephrasing part ( b ) of theorem [ theorem1 ] so that no three - fold branches are necessary .
[ cor1]if @xmath55 and @xmath61 are loopless neighbors in @xmath2 then @xmath93 , where @xmath94 is obtained from @xmath95 by changing the weights of @xmath55 to @xmath96 and @xmath97 .
let @xmath98 be the graph obtained from @xmath95 by changing the weights of @xmath55 as in proposition [ prop1 ] , with @xmath99 and @xmath100 @xmath101
. then @xmath102 and @xmath103 , so @xmath104 = @xmath105 .
on the other hand , proposition [ prop1 ] tells us that @xmath106 if @xmath107 and @xmath108 are used instead of the weights given in corollary [ cor1 ] , then the resulting formula @xmath109 still has only two branches , and bears an interesting resemblance to part ( a ) of theorem [ theorem1 ] .
however using division to define @xmath110 may occasionally cause some algebraic difficulties , e.g. , it complicates the evaluation at @xmath111 , and it prohibits the use of rings in which @xmath112 is a divisor of zero .
corollary [ cor1 ] is one of several results in which vertex weights allow us to extend properties of the unweighted version of @xmath3 , seemingly the simplest kind of interlace polynomial , to the other interlace polynomials . in this instance
the extended property is that of possessing a recursive description represented by a binary computation tree .
another result of this type is corollary [ cor2 ] , which extends remark 18 of @xcite : if @xmath55 and @xmath61 are loopless neighbors in @xmath2 then the unweighted vertex - nullity interlace polynomials of @xmath2 and @xmath62 are the same .
[ cor2]let @xmath55 and @xmath61 be unlooped neighbors in @xmath2 , and let @xmath113 be the weighted graph obtained from @xmath62 by changing the weights of @xmath55 and @xmath61 to @xmath114 , @xmath115 , @xmath116 and @xmath117 .
then @xmath118 .
applying corollary [ cor1 ] to @xmath113 , with the roles of @xmath55 and @xmath61 reversed , tells us that @xmath119 = @xmath120 .
observe that @xmath121 has the underlying graph @xmath122 , and differs from @xmath25 only in the weights of @xmath61 , which are given by @xmath123 and @xmath124 . consequently @xmath125
= @xmath126 = @xmath127 .
as @xmath128 = @xmath129 , the result follows directly from corollary [ cor1 ] .
if we are confident that division by @xmath112 will cause no trouble , corollary [ cor2 ] may be restated with a simpler conclusion : using @xmath130 , @xmath108 , @xmath131 and @xmath132 yields @xmath133 .
other novel properties of the weighted interlace polynomials arise from a general observation .
suppose @xmath83 happen to have the property that for @xmath14 , the rank and nullity of @xmath16 are determined by the rank and nullity of @xmath134 , perhaps in different ways according to which of @xmath135 is contained in @xmath16
. then it may be possible to adjust @xmath49 and @xmath136 in such a way that the weighted interlace polynomial of the weight - adjusted version of @xmath137 incorporates all the information in @xmath1 .
the simplest instance of this observation occurs when @xmath55 and @xmath61 give rise to identical rows and columns in the adjacency matrix of @xmath2 .
two vertices @xmath135 of @xmath2 are _ _ identical twins _ _
if ( i ) either they are looped and adjacent or they are unlooped and not adjacent , and ( ii ) they have the same neighbors outside @xmath138 .
identical twins are called _ clones _ in @xcite , and unlooped identical twins are called _ duplicates _ in @xcite and _ false twins _ in @xcite .
we prefer the present terminology because the adjective _ false _ seems inappropriate , and because we do not know what we would call non - identical duplicates or clones in definition 7 below .
the following result extends proposition 40 of @xcite , proposition 4.14 of @xcite and section 3.1 of @xcite to vertex - weighted graphs .
[ theorem3]suppose @xmath55 and @xmath61 are identical twins in @xmath2 .
let @xmath22 be the graph obtained from @xmath137 by changing the weights of @xmath55 : @xmath139 and @xmath140
. then @xmath141 . if @xmath142 then @xmath143)$ ] = @xmath144)$ ] , because the adjacency matrices are the same .
if @xmath145 then @xmath143)$ ] = @xmath144)$ ] = @xmath146)$ ] = @xmath147)$ ] , because the only difference among the adjacency matrices is that the matrix of @xmath148 $ ] has two identical rows and columns , corresponding to @xmath55 and @xmath61 . as mentioned above , it is helpful to read @xmath17 as includes and @xmath18 as excludes so that ( for instance ) the appearance of @xmath149 in @xmath150 indicates that if @xmath151 then the summand of @xmath152 corresponding to @xmath16 includes the summand of @xmath1 corresponding to @xmath153 .
theorem [ theorem3 ] has the following inductive generalization .
suppose @xmath154 and @xmath155 are identical twins in @xmath2 . if @xmath22 is the graph obtained from @xmath156 by changing the weights of @xmath55 to @xmath157 then @xmath158 .
we refer to the process of combining several identical twins into a single vertex as an _ identical twin reduction _ no matter how many identical twins are combined .
[ theorem4]suppose @xmath61 is an unlooped degree - one vertex pendant on @xmath55 .
let @xmath22 be the graph obtained from @xmath137 by changing the weights of @xmath55 : @xmath159 and @xmath160
. then @xmath158 .
if @xmath151 then @xmath143)$ ] = @xmath144)$ ] , and if @xmath142 then @xmath143)$ ] = @xmath161)-2 $ ] = @xmath146)$ ] = @xmath144)$ ] .
if @xmath162 are unlooped and pendant on @xmath55 then @xmath163 are identical twins , so we can combine them into a single re - weighted vertex @xmath61 using theorem [ theorem3 ] and then remove @xmath164 using theorem [ theorem4 ] .
we refer to the removal of any number of unlooped vertices pendant on the same vertex as an _ unlooped pendant vertex reduction_. theorem [ theorem4 ] extends proposition 4.12 of @xcite and section 3.2 of @xcite , where collections of pendant vertices are called _
combs_. a recursive calculation that depends solely on theorem [ theorem1 ] and corollary [ cor1 ] is represented by a mixed binary - ternary computation tree ; if corollary [ cor1 ] is always used in place of part ( b ) of theorem [ theorem1 ] then the tree will be binary .
using identical twin and unlooped pendant vertex reductions makes the computation tree smaller , because each time one of these reductions is used , we avoid splitting the resulting portion of the calculation into branches .
similarly , a tutte polynomial computation that incorporates series - parallel reductions and deletion - contraction operations will generally result in a smaller formula than a computation that involves only deletion - contraction operations can provide .
this latter observation was made precise in @xcite : computing the tutte polynomial of a matroid @xmath165 using series - parallel reductions and deletion - contraction operations will result in an expression with at least @xmath166 terms , where @xmath166 is crapo s @xmath18 invariant @xcite ; moreover if @xmath167 the lower bound is attainable .
( this result is merely the extension to the tutte polynomial of the important theory of reliability domination ; see @xcite for expositions . ) for interlace polynomials , analogous lower bounds are derived from the unweighted vertex - nullity polynomial @xmath168 .
the coefficient of @xmath42 in @xmath169 is denoted @xmath170 as in @xcite , and the evaluation @xmath171 is denoted @xmath172 . [ theorem5]if part ( c ) of theorem [ theorem1 ] is applied only to loopless graphs then a computation of @xmath1 using corollary [ cor1 ] and theorems [ theorem1 ] , [ theorem3 ] and [ theorem4 ] is represented by a computation tree with no fewer than @xmath173 leaves .
if @xmath2 is a simple graph then the computation tree has no fewer than @xmath174 leaves .
we quickly review some elementary properties of @xmath169 from @xcite .
this polynomial is described recursively as follows .
if @xmath2 has a loop at @xmath55 then @xmath169 = @xmath175 + @xmath176 , if @xmath55 and @xmath61 are loopless neighbors then @xmath169 = @xmath176 + @xmath177 , and if @xmath178 is the edgeless @xmath179-vertex graph then @xmath180 .
the latter includes the empty graph @xmath181 with @xmath182 .
it follows by induction on the number of vertices that no graph has any negative coefficient in @xmath169 , and hence every graph has @xmath183 and @xmath184 . moreover , every nonempty simple graph has @xmath185 and every disconnected simple graph has @xmath186 .
theorem [ theorem5 ] is certainly true for @xmath178 , as @xmath187 . proceeding inductively ,
observe that if corollary [ cor1 ] or part ( a ) or ( b ) of theorem [ theorem1 ] is applied to a graph @xmath188 represented by a certain node of the computation tree , and @xmath189 and @xmath190 are the graphs represented by the ( first two ) resulting branch nodes , then @xmath191 = @xmath192 + @xmath193 ; certainly then @xmath194 = @xmath195 + @xmath196 .
( by the way , we call vertices of the computation tree _ nodes _ in order to distinguish them from vertices of @xmath2 . ) if theorem [ theorem3 ] or [ theorem4 ] is applied to remove a vertex @xmath61 from a graph @xmath197 then @xmath198 = @xmath199 , because the formulas of theorems [ theorem3 ] and [ theorem4 ] yield @xmath200 when @xmath49 = @xmath136 = @xmath201 = @xmath202 = @xmath203 , @xmath204 and @xmath205 .
it remains to consider the special case involving simple graphs .
we actually prove a slightly different result , namely : if @xmath2 is simple then the portion of the computation tree involving only nodes corresponding to connected graphs has no fewer than @xmath206 leaves .
if @xmath2 is disconnected then @xmath186 so @xmath2 satisfies the result trivially .
if @xmath207 is a connected , simple graph with @xmath208 then @xmath2 satisfies the result because @xmath209 .
proceeding inductively , observe that if corollary [ cor1 ] or theorem [ theorem1 ] ( b ) is applied to a connected graph @xmath188 represented by a certain node of the computation tree , and @xmath189 and @xmath190 are the graphs represented by the ( first two ) resulting branch nodes , then @xmath191 = @xmath192 + @xmath193 .
it follows that @xmath210 = @xmath211 + @xmath212 . if theorem [ theorem3 ] or theorem [ theorem4 ] is used to remove a vertex @xmath61 from a connected , simple graph @xmath188 with 3 or more vertices then @xmath213 , by corollary 4.17 of @xcite .
theorem [ theorem5 ] is of limited value because the computations discussed are not optimal .
the restriction that part ( c ) of theorem [ theorem1 ] is only applied to loopless graphs is an obvious inefficiency .
in addition , if some ( combinations of ) weights are 0 then it would be natural to simply ignore the corresponding parts of the computation .
there are also other useful twin reductions that do not fall under theorem [ theorem5 ] .
two vertices @xmath135 of @xmath2 are _
_ fraternal twins _
_ if ( i ) either they are looped and nonadjacent or they are unlooped and adjacent , and ( ii ) they have the same neighbors outside @xmath138 . unlooped fraternal twins are called _ true twins _ in @xcite , but we prefer the present terminology because the rows and columns of the adjacency matrix corresponding to fraternal twins are not quite the same . here is an extension of proposition 4.15 of @xcite to weighted graphs .
[ theorem6]suppose @xmath55 and @xmath61 are fraternal twins in @xmath2 .
let @xmath22 be the graph obtained from @xmath137 by changing the weights of @xmath55 : @xmath150 = @xmath214 and @xmath215 = @xmath216
. then @xmath158 .
if @xmath151 then @xmath143)$ ] = @xmath144)$ ] = @xmath217)$ ] , because the adjacency matrices are the same . if @xmath142 then @xmath143)$ ] = @xmath144)$ ] = @xmath161)-2 $ ] .
theorem [ theorem6 ] has the following inductive generalization .
suppose @xmath154 and @xmath155 are fraternal twins in @xmath2 . if @xmath22 is the graph obtained from @xmath156 by changing the weights of @xmath55 to @xmath218 then @xmath158 .
we refer to the process of combining several fraternal twins into a single vertex as a _ fraternal twin reduction _ no matter how many fraternal twins are combined .
in general , fraternal twin reductions are just as useful as identical twin reductions . however ,
as noted in proposition 38 of @xcite and corollary 4.16 of @xcite they have the effect of multiplying the unweighted vertex - nullity polynomial by powers of 2 .
( observe that if @xmath49 = @xmath136 = @xmath201 = @xmath202 = @xmath203 and @xmath204 then theorem [ theorem6 ] gives @xmath150 = @xmath215 = @xmath219 . )
consequently the lower bounds of theorem [ theorem5 ] are not valid for computations that utilize theorem [ theorem6 ] along with corollary [ cor1 ] and theorems [ theorem1 ] , [ theorem3 ] and [ theorem4 ] .
if we are given a reduction of a graph @xmath2 to disconnected vertices using twin reductions and unlooped pendant vertex reductions , then theorems [ theorem3 ] , [ theorem4 ] and [ theorem6 ] describe @xmath1 in linear time simply update the vertex weights at each step , and at the end refer to part ( c ) of theorem [ theorem1 ] .
if we are not given such a reduction then determining whether or not any such reduction exists , and finding one if possible , can be accomplished in polynomial time : as in corollary 5.3 of @xcite , simply search @xmath19 repeatedly for unlooped degree - one vertices and pairs of vertices @xmath135 with the same neighbors outside @xmath138 .
hence if @xmath2 has such a reduction then in polynomial time , theorems [ theorem3 ] , [ theorem4 ] and [ theorem6 ] provide a description of @xmath1 that completely avoids the branching formulas of corollary [ cor1 ] and parts ( a ) and ( b ) of theorem [ theorem1 ] .
this observation extends theorem 6.4 of @xcite from @xmath3 to @xmath9 and from simple graphs that can be analyzed without fraternal twin reductions to looped graphs that can be analyzed using all three types of reductions : if a graph @xmath2 can be reduced to a collection of disconnected vertices using unlooped pendant vertex reductions and the two types of twin reductions then theorems [ theorem3 ] , [ theorem4 ] and [ theorem6 ] provide a polynomial - time description of @xmath1 .
the theorem refers to a _ description _ rather than a _ computation _ because we have ignored the cost of arithmetic operations in the ring @xmath20 . in case @xmath220 arithmetic operations have low cost , and describing a weighted polynomial
@xmath1 is the same as computing an evaluation of the unweighted polynomial @xmath1 .
the full unweighted polynomial may be recovered from several evaluations by interpolation , so the theorem provides a polynomial - time computation of the unweighted polynomial . in more complicated rings like @xmath221
$ ] arithmetic operations may be so expensive as to prohibit polynomial - time computation of entire weighted polynomials . a thorough discussion of these matters is given by courcelle @xcite .
in this section we reformulate and extend some results of arratia , bollobs and sorkin @xcite regarding substituted graphs , using the following version of a construction introduced by cunningham @xcite . [ comp]a vertex - weighted graph @xmath2 is the _ composition _ of vertex - weighted graphs @xmath188 and @xmath8 , @xmath7 , if the following conditions hold .
\(a ) @xmath222 consists of a single unlooped vertex @xmath55 .
\(b ) the vertex @xmath55 is unweighted in both @xmath188 and @xmath8 , i.e. , @xmath223 in @xmath188 and @xmath8 .
\(c ) @xmath224 , and the vertices of @xmath2 inherit their weights from @xmath188 and @xmath8 .
\(d ) @xmath225 and @xmath226 . requiring @xmath55 to be unlooped and unweighted in both @xmath188 and @xmath8
guarantees that no information is lost when we remove @xmath55 in constructing @xmath2 .
[ h ] weighif.ps definition [ comp ] includes several other familiar notions .
if @xmath55 is isolated in @xmath188 or @xmath8 then @xmath227 is the disjoint union of @xmath228 and @xmath229 .
if @xmath55 is adjacent to every other vertex of @xmath188 and @xmath8 then @xmath227 is also denoted @xmath230 .
this is traditionally called a join but that term has recently been used for general compositions @xcite . if @xmath55 is adjacent to every other vertex of @xmath188 then @xmath227 is the graph obtained by substituting @xmath228 for @xmath55 in @xmath8 . if @xmath55 is adjacent to every other vertex of @xmath188 and @xmath228 is edgeless or complete then the ( un)looped vertices of @xmath228 are twins in @xmath227
. the following observation will be useful .
[ prop2 ] suppose two graphs @xmath231 and @xmath232 are identical except for the weights of a single vertex @xmath55 , and let @xmath233 be the graph that is identical to both @xmath231 and @xmath232 except for @xmath234 and @xmath235
. then @xmath236 .
let @xmath55 be an unweighted vertex of a simple vertex - weighted graph @xmath188 , and let @xmath189 and @xmath190 be the full subgraphs of @xmath228 induced by the neighbors ( resp .
non - neighbors ) of @xmath55 .
any weighted interlace polynomial @xmath237 may be analyzed in the following way .
eliminate all edges between vertices of @xmath228 using pivots and weight - changes as in corollary [ cor1 ] .
( in general there will be many different sequences of pivots that may be used ; this lack of uniqueness is not important in the analysis . ) the assignments of individual vertices of @xmath188 to @xmath189 and @xmath190 may change during this process , and vertex weights may also change ; but these reassignments and weight changes will be the same for all graphs @xmath8 . as in the proof of proposition 39 of @xcite
these pivots will not affect the internal structure of @xmath8 , because no two vertices of @xmath8 have distinct , nonempty sets of neighbors in @xmath188 . here
the phrase `` internal structure '' refers to vertex weights , the positions of loops , and adjacencies , including adjacencies between @xmath55 and other vertices of @xmath8 .
when step 1 is complete , @xmath237 is expressed as a sum in which each summand is the product of an initial multiplying factor and the weighted interlace polynomial @xmath238 of a graph in which every edge is incident on a vertex of @xmath239 .
if @xmath240 then @xmath241 is isolated , so the only effect of @xmath241 is to multiply that summand by @xmath242 .
this same effect is realized by removing @xmath241 and incorporating @xmath243 into the initial multiplying factor , so we may assume that every summand has @xmath244 . in a summand with @xmath245
the vertices of @xmath246 are all identical twins , and may be consolidated into a single vertex @xmath55 using theorem [ theorem3 ] . in a summand with @xmath247 a single vertex @xmath248 may be introduced with @xmath249 and @xmath250 ; this will not affect the value of the corresponding summand .
as in step 1 , these manipulations are the same for all @xmath8 . step 3 .
the weighted interlace polynomial @xmath237 is now expressed as a sum in which each summand is the product of an initial multiplying factor and a weighted interlace polynomial @xmath251 in which @xmath252 differs from @xmath8 only in the weights of @xmath55 . in each summand
we multiply the @xmath17 and @xmath18 weights of @xmath55 by the initial multiplying factor .
this has the effect of multiplying @xmath251 by that factor , so the summand is now simply @xmath251 .
proposition [ prop2 ] tells us that the sum is equal to a single weighted interlace polynomial @xmath251 , where @xmath49 and @xmath136 are obtained by adding together the @xmath17 and @xmath18 weights of @xmath55 in the various summands .
we deduce that there are weights @xmath49 and @xmath136 that depend only on @xmath188 and @xmath55 , and have the following property : in every instance of definition [ comp ] involving @xmath188 , the interlace polynomial @xmath237 equals @xmath251 , where @xmath252 is obtained from @xmath8 by using @xmath49 and @xmath136 as weights for @xmath55 .
finding explicit formulas for these weights is not difficult .
[ theorem8]let @xmath188 be a vertex - weighted simple graph with an unweighted vertex @xmath55 .
then every composition @xmath227 has @xmath253 , where @xmath252 is obtained from @xmath8 by using the following weights for @xmath55 .
@xmath254 with @xmath255 , we have @xmath256 = @xmath237 = @xmath251 = @xmath257 . with @xmath8 consisting of two adjacent , unlooped , unweighted vertices @xmath55 and @xmath241 we have @xmath258 = @xmath237 = @xmath259 = @xmath260 . the stated formulas for @xmath49 and @xmath136 follow .
in case @xmath261 might be a zero divisor in the ring @xmath20 , one can avoid any difficulty with the formulas of theorem [ theorem8 ] by first evaluating them in the polynomial ring @xmath262 $ ] , and then evaluating the resulting division - free formulas in @xmath20 .
theorem [ theorem8 ] concerns compositions @xmath227 in which @xmath188 is simple . if @xmath188 has looped vertices , a similar analysis requires two new steps .
begin by removing all loops in @xmath228 using local complementation as in part ( a ) of theorem [ theorem1 ] .
the result is a description of @xmath263 as a sum in which each summand is the product of an initial factor and an interlace polynomial @xmath238 or @xmath264 , with no loops in @xmath265 .
step 4 . after applying steps 1 - 3 to each of these summands ,
collect terms to obtain a formula @xmath266 in order to distinguish the two terms on the right - hand side we denote by @xmath267 the copy of @xmath55 in @xmath268 .
it might seem that we now have to determine four unknowns , namely the vertex weights @xmath49 , @xmath269 , @xmath136 and @xmath270 it turns out though that these unknowns are not independent .
there is an obvious isomorphism between @xmath271 and @xmath268 , and consequently @xmath272 may be obtained from @xmath273 simply by replacing @xmath274 with @xmath49 .
theorem [ theorem2 ] tells us that @xmath275 so @xmath276 may be obtained from @xmath273 by replacing @xmath274 with @xmath49 .
that is , the coefficient of @xmath49 in @xmath251 is precisely the same as the coefficient of @xmath274 in @xmath277 .
it follows that the sum @xmath251 + @xmath277 is unchanged if we replace @xmath49 by @xmath49 + @xmath274 and replace @xmath274 by @xmath278 . [ theorem9]let @xmath188 be a vertex - weighted graph with an unweighted , unlooped vertex @xmath55 .
then @xmath188 and @xmath55 determine weights @xmath49 , @xmath136 and @xmath270 such that every composition @xmath227 has @xmath279 , where @xmath252 is obtained from @xmath8 by using @xmath49 and @xmath136 as weights for @xmath55 and @xmath280 is obtained from @xmath268 by using @xmath281 and @xmath270 as weights for @xmath267 , the copy of @xmath55 in @xmath280 .
formulas for the three weights mentioned in theorem [ theorem9 ] may be derived from three instances of the theorem .
we use @xmath228 and @xmath188 as in theorem [ theorem8 ] , and also @xmath282 , the graph obtained from @xmath188 by attaching a loop at @xmath55 .
these correspond respectively to compositions of @xmath188 with graphs @xmath283 such that @xmath284 ; @xmath285 with @xmath241 an unweighted , unlooped neighbor of @xmath55 ; and @xmath286 with @xmath287 an unweighted , looped neighbor of @xmath55 . definition [ 2varq ] gives the following values .
@xmath288 @xmath289 @xmath290 we deduce these formulas .
@xmath291 @xmath292 @xmath293 separate formulas for @xmath136 and @xmath270 are derived in the obvious ways by adding and subtracting the last two .
as before , possible problems with denominators may be avoided by first evaluating the formulas in @xmath294 $ ] .
we close this section with different formulas for the weights @xmath49 , @xmath136 and @xmath270 that appear in theorems [ theorem8 ] and [ theorem9 ] .
suppose @xmath188 has an unweighted , unlooped vertex @xmath55 .
let @xmath295 denote the open neighborhood of @xmath55 , i.e. , the set containing the vertices @xmath296 that neighbor @xmath55 . given a subset @xmath297 let @xmath298 ( resp .
@xmath299 ) be the row ( resp .
column ) vector with entries indexed by @xmath300 whose @xmath301 entry is 1 or 0 according to whether @xmath302 or @xmath303 . also let @xmath304 be the adjacency matrix of @xmath305 $ ] .
note that @xmath306 and @xmath307{cc}% m & \kappa\\ \rho & 0 \end{array } \right ) , \;\;r\left ( \begin{array } [ c]{cc}% m & \kappa\\ \rho & 1 \end{array } \right ) \leq r(m)+2\ ] ] because adjoining a single row or column to a matrix raises the rank by 0 or 1 .
[ def9]the _ type _ of @xmath16 with respect to @xmath55 is defined as follows .
@xmath16 is of type 1 if @xmath308{cc}% m & \kappa\\ \rho & 0 \end{array } \right ) = r\left ( \begin{array } [ c]{cc}% m & \kappa\\ \rho & 1 \end{array } \right ) -1.$ ] @xmath16 is of type 2 if @xmath309{cc}% m & \kappa\\ \rho & 0 \end{array } \right ) = r\left ( \begin{array } [ c]{cc}% m & \kappa\\ \rho & 1 \end{array } \right ) .$ ] @xmath16 is of type 3 if @xmath308{cc}% m & \kappa\\ \rho & 1 \end{array } \right ) = r\left ( \begin{array } [ c]{cc}% m & \kappa\\ \rho & 0 \end{array } \right ) -1.$ ] every @xmath297 is of one of these types .
moreover , if @xmath55 has no looped neighbor then there is no @xmath297 of type 3 .
the fact that every @xmath297 is of type 1 , 2 or 3 appears in lemma 2 of @xcite .
suppose @xmath16 is of type 3 , so @xmath310{cc}% m & \kappa\\ \rho & 0 \end{array } \right ) .\ ] ] the row vector @xmath311 can not be a sum of rows of @xmath312 .
for if it were then @xmath313 would be the corresponding sum of rows of @xmath165 , and by symmetry @xmath314 would be the corresponding sum of columns of @xmath165 .
consequently it would follow that @xmath315{cc}% m & \kappa \end{array } \right ) = r\left ( \begin{array } [ c]{cc}% m & \kappa\\ \rho & 0 \end{array } \right ) .\ ] ] on the other hand , @xmath313 must be the sum of the rows of @xmath165 corresponding to the elements of some subset @xmath316 , for if it were not then it would follow that @xmath310{c}% m\\ \rho \end{array } \right ) = r\left ( \begin{array } [ c]{cc}% m & \kappa \end{array } \right ) = r\left ( \begin{array } [ c]{cc}% m & \kappa\\ \rho & 0 \end{array } \right ) -1.\ ] ] every such @xmath317 must contain an odd number of neighbors of @xmath55 , to avoid giving a sum of rows of @xmath312 equal to @xmath311 .
choose such a @xmath317 , and consider the induced subgraph @xmath318 $ ] .
as the sum of the rows of @xmath165 corresponding to elements of @xmath317 is @xmath313 , the sum of the rows of the adjacency matrix of @xmath318 $ ] is the row vector @xmath319 .
@xmath320 is odd , so the adjacency matrix of @xmath318 $ ] has an odd number of entries equal to 1 .
the matrix is symmetric , so at least one of these entries must appear on the diagonal .
that is , at least one vertex of @xmath318 $ ] is looped . for @xmath297 let @xmath321 be the matrix obtained from @xmath165 by toggling every entry @xmath322 that has @xmath323 .
then @xmath324 = @xmath325 because the first matrix is transformed into the second by adding the last column to every column corresponding to a neighbor of @xmath55 . similarly , adding the last column of the first matrix below to every column corresponding to a neighbor of @xmath55
tells us that @xmath326{cc}% m & \kappa\\ \rho & 1 \end{array } \right ) = r\left ( \begin{array } [ c]{cc}% m^{a } & \kappa\\ 0 & 1 \end{array } \right ) = 1+r(m^{a}).\ ] ] consequently , definition [ def9 ] may be restated using the relationship between @xmath327 and @xmath328 : if @xmath16 is of type 1 then @xmath329 , if @xmath16 is of type 2 then @xmath330 , and if @xmath16 is of type 3 then @xmath331 .
[ prop3]for @xmath332 let @xmath333)}(y-1)^{n((h - a)[s])}.\ ] ] then the following equations hold .
@xmath334 also , @xmath335 the first equality is obvious .
for the second , note that each @xmath336 gives rise to two subsets of @xmath337 , namely @xmath16 and @xmath338 ; these correspond to the adjacency matrices @xmath165 and @xmath339{cc}% m & \kappa\\ \rho & 0 \end{array } \right ) $ ] .
similarly , the third equality is derived by considering the contributions of two adjacency matrices for each @xmath297 , namely @xmath165 and @xmath339{cc}% m & \kappa\\ \rho & 1 \end{array } \right ) $ ] .
the last equality follows from the discussion preceding the proposition .
the weights mentioned in theorems [ theorem8 ] and [ theorem9 ] are @xmath340 , @xmath341 and @xmath342 .
in particular , @xmath343 if @xmath55 has no looped neighbor .
the corollary follows from proposition [ prop3 ] and the formulas given in theorem [ theorem8 ] and immediately after theorem [ theorem9 ] .
in this section we focus our attention on the unweighted vertex - nullity polynomial .
[ prop4]if @xmath2 is a connected , unweighted graph with at least one looped vertex then @xmath344 .
the proposition is certainly true if @xmath2 has @xmath208 vertices , as all three such graphs have @xmath345 .
the argument proceeds by induction on @xmath346 . recall that @xmath184 for every graph @xmath2 , and let @xmath55 be a looped vertex of @xmath2 . then @xmath347 , so @xmath348 , @xmath349 .
suppose @xmath55 is not a cutpoint of @xmath2 .
if @xmath2 has some looped vertex other than @xmath55 then the inductive hypothesis implies that @xmath350 .
if @xmath2 has no looped vertex other than @xmath55 then every neighbor of @xmath55 is looped in @xmath351 .
every component of @xmath351 contains at least one neighbor of @xmath55 , so the inductive hypothesis implies that @xmath352 for every component @xmath353 of @xmath351 . hence @xmath354 .
suppose now that @xmath2 has a looped cutpoint @xmath55 .
for each component @xmath353 of @xmath25 and each vertex @xmath355 that lies in some other component of @xmath25 , no edge connecting @xmath241 to an element of @xmath356 appears in @xmath25 .
consequently every such edge appears in @xmath351 , so all of @xmath357 is contained in a single component of @xmath351 .
as @xmath2 is connected , this implies that @xmath351 is also connected .
hence if any neighbor of @xmath55 is unlooped in @xmath2 , @xmath351 is a connected graph with a looped vertex and the inductive hypothesis implies that @xmath358 .
if instead every neighbor of @xmath55 is looped in @xmath2 , then every component @xmath353 of @xmath25 has a looped vertex and the inductive hypothesis implies @xmath359 .
[ cor4]an unweighted graph @xmath2 has @xmath360 if and only if @xmath207 has no nonempty simple component .
as noted in remark 20 of @xcite , every nonempty simple graph has @xmath185 .
it follows that every graph with a nonempty simple component also has @xmath185 .
on the other hand , proposition [ prop4 ] and the fact that @xmath361 together imply that every graph with no nonempty simple component has @xmath360 .
let @xmath2 be an unweighted graph , and let @xmath362 be its _ complement _
, i.e. , the graph with @xmath363 whose edges ( including loops ) are precisely the edges absent in @xmath2 .
then the following are equivalent : @xmath2 is simple , @xmath364 and @xmath365 .
suppose first that @xmath2 is simple , and let @xmath366 with @xmath55 the vertex of @xmath367 .
let @xmath8 be the two - vertex graph with @xmath368 in which @xmath287 is an unweighted , looped neighbor of @xmath55 .
theorem [ theorem8 ] tells us that @xmath369 on the other hand , the recursive description of @xmath9 @xcite tells us that @xmath370 recalling that @xmath371 = @xmath257 , we see that @xmath372 and consequently @xmath373 . recalling that @xmath374 = @xmath258 = @xmath260
, we see that latexmath:[\[q_{n}(g+e_{1}^{u})=q(g+e_{1}^{u})|_{x=2}=y\left ( \alpha(a)+\beta(a)\right ) now suppose that @xmath2 is not simple .
then @xmath2 has a looped vertex , and so does @xmath376 .
as @xmath376 is connected , corollary [ cor4 ] states that @xmath377 ; consequently @xmath378 .
a combinatorial description of the interlace polynomials of trees and forests is given in @xcite . in order to motivate the next section we sketch this description briefly here , omitting details and proofs .
recall that a tree @xmath317 is _ rooted _ by specifying a root vertex @xmath379 .
each non - root vertex @xmath380 then has a unique _ parent _ @xmath381 , a neighbor whose distance from @xmath382 is less than the distance from @xmath382 to @xmath241 .
the elements of @xmath383 are the _ children _ of @xmath381 , and the children of @xmath381 other than @xmath241 itself are _ siblings _ of @xmath241 .
an _ ordered _ tree is a rooted tree given with an order on the set of children of each parent vertex ; non - root vertices may then have _ earlier siblings _ and _ later siblings_. a set of vertices that contains no adjacent pair is _ independent _ , and a set of vertices _ dominates _ a vertex @xmath241 if it contains @xmath241 or contains some neighbor of @xmath241 .
an _ earlier sibling cover _ ( or _ es - cover _ ) in an ordered tree @xmath317 is an independent set @xmath384 that dominates @xmath382 and has the property that for every non - root vertex @xmath385 , every earlier sibling of @xmath241 is dominated by @xmath384 . for integers @xmath386 and @xmath387 the _ es - number _ @xmath388 is the number of @xmath386-element es - covers in @xmath317 whose non - root elements have @xmath387 different parents .
if @xmath389 is a subtree of @xmath317 and @xmath390 then we presume that the children of each parent vertex @xmath241 in @xmath389 are ordered by restricting the order of the children of @xmath241 in @xmath317 . with this convention , it is easy to verify that the earlier sibling covers in large trees arise from earlier sibling covers in subtrees .
[ recur ] let @xmath317 be an ordered tree with a leaf @xmath287 such that @xmath391 , all the siblings of @xmath287 are leaves , and @xmath287 has no later siblings . then @xmath392 and @xmath393 lemma [ recur ] is the key to an inductive proof that the terms in the definition of @xmath394 can be collected into sub - totals corresponding to earlier sibling covers .
let @xmath317 be an ordered tree with vertex weights , and let @xmath384 be an es - cover in @xmath317 .
let @xmath395 for each vertex @xmath241 define the @xmath384__-weight _ _ @xmath396 as follows : @xmath397 the product @xmath398 is the _ total weight _ of @xmath384 in @xmath317 , denoted @xmath399 .
[ treepoly ] if @xmath317 is an ordered tree with vertex weights then @xmath400 if @xmath317 has no more than one vertex of degree @xmath401 , it is not difficult to verify the theorem directly .
if @xmath317 has more than one vertex of degree @xmath401 then the theorem follows inductively from theorem [ theorem1 ] and corollary [ cor1 ] using lemma [ recur ] , as detailed in @xcite . setting @xmath48 and @xmath47
we deduce that the unweighted interlace polynomial of a tree is determined by a simple formula involving earlier sibling covers .
[ treeint]if @xmath317 is a tree then the unweighted interlace polynomial of @xmath317 is @xmath402 as the interlace polynomials are multiplicative on disjoint unions , these results extend directly to disconnected forests .
the tutte polynomial is a very useful invariant of graphs and matroids , which incorporates a great deal of information and can be defined in several different ways ; see @xcite , @xcite and @xcite for more detailed discussions than we provide here .
one of its definitions is given in definitions [ oldact ] and [ act ] .
[ oldact ] suppose @xmath2 is an unweighted graph with @xmath403 , and @xmath317 is a maximal spanning forest of @xmath2 .
an element @xmath404 is _ externally active _ with respect to @xmath317 if @xmath405 is the least index of an element of the unique circuit contained in @xmath406 .
an element @xmath407 is _ internally active _ with respect to @xmath317 if @xmath405 is the least index of an element of the unique cutset contained in @xmath408 .
the numbers of edges that are externally and internally active with respect to @xmath317 are denoted @xmath409 and @xmath410 , respectively .
theorem [ treepoly ] and corollary [ treeint ] bear a strong resemblance to this formula , with earlier sibling covers in rooted trees replacing maximal spanning forests in arbitrary graphs .
we do not know whether there is a combinatorial analogue of definition [ act ] that describes the interlace polynomials of an arbitrary graph , but there is an algorithmic analogue . before presenting it
, we recall another definition of the tutte polynomial . to recursively calculate @xmath423 one simply chooses an arbitrary edge of @xmath2 , and applies the appropriate part of definition [ delcont ] ; this process is repeated as many times as necessary .
such a computation is represented by a computation tree in which a node that represents an instance of part ( a ) has two children and a node that represents an instance of part ( b ) or ( c ) has only one .
[ actact ] let @xmath2 be a graph with @xmath411 , and consider the recursive implementation of definition [ delcont ] in which @xmath424 is removed first , then @xmath425 is removed in all branches , then @xmath426 is removed in all branches , and so on .
the leaves of the computation tree representing this implementation correspond to the maximal spanning forests of @xmath2 , with the leaf corresponding to @xmath317 resulting from the portion of the computation in which edges of @xmath317 are contracted and elements of @xmath427 are deleted .
a node of this portion of the computation tree represents the removal of an active edge if and only if it has precisely one child .
the proposition is implicit in the fact that definition [ delcont ] and definition [ act ] both yield @xmath423 , so it appears implicitly in just about every presentation of the tutte polynomial .
see theorem ix .
65 of @xcite or theorem x.10 of @xcite , for instance .
explicit discussions of the connection between activities and computation are less common in the literature , though there are some @xcite .
the proof is a direct induction on @xmath428 .
if @xmath424 is a loop then the maximal spanning forests of @xmath2 and @xmath429 coincide , and the computation tree for @xmath423 is obtained from the computation tree for @xmath430 by attaching a new root node of degree 1 , representing the removal of @xmath424 .
if @xmath424 is an isthmus then the maximal spanning forests of @xmath2 and @xmath431 correspond , and the computation tree for @xmath423 is obtained from the computation tree for @xmath432 by attaching a new root node of degree 1 . otherwise , the maximal spanning forests of @xmath2 that contain @xmath424 correspond to the maximal spanning forests of @xmath431 , and the maximal spanning forests of @xmath2 that do not contain @xmath424 are the maximal spanning forests of @xmath429 . the computation tree for @xmath423 consists of the root and two disjoint subtrees that are the computation trees for @xmath432 and @xmath430
. definition [ oldact ] defines active edges using the structure of @xmath2 , and this leads to definition [ act ] s description of @xmath423 as a generating function for maximal spanning forests .
proposition [ actact ] shows that we may also see activity from an algorithmic viewpoint : the external activity of a particular @xmath433 is revealed in the fact that one step of a calculation of @xmath423 involves removing @xmath434 using part ( b ) of definition [ delcont ] rather than part ( a ) .
this distinction affects the result of the computation , so it would be important even if activity could not be conveniently described using the structure of @xmath2 , or did not contribute to a convenient closed form for @xmath423 .
[ algact ] a node of a computation tree representing a recursive implementation of definition [ alg ] is _ active _ if it has precisely one child , i.e. , if it represents an application of part ( c ) or part ( d ) .
an activities formula for @xmath1 arises directly from a computation tree representing an implementation of definition [ alg ] .
the formula has one summand for each leaf of the computation tree ( each call to part ( e ) of definition [ alg ] ) , representing the product of the coefficients contributed by the nodes in the portion of the computation tree that gives rise to that leaf .
active and non - active nodes contribute different coefficients .
let @xmath317 be a rooted tree with root @xmath382 , and consider a recursive implementation of definition [ alg ] structured as follows .
if possible , apply part ( d ) of definition [ alg ] ; if not and there is a parent vertex other than @xmath382 then apply part ( b ) with a leaf of the type denoted @xmath287 in lemma [ recur ] as @xmath55 ; otherwise apply part ( b ) with the last child of @xmath382 as @xmath55 .
the leaves of the computation tree representing this implementation correspond to the earlier sibling covers of @xmath317 , with the es - cover @xmath384 corresponding to a given leaf constructed from the portion of the computation that gives rise to that leaf as follows : an occurrence of part ( d ) of definition [ alg ] contributes its @xmath55 to @xmath384 , and an occurrence of the @xmath441 branch of part ( b ) contributes its @xmath55 to @xmath384 .
if @xmath317 has no vertex other than @xmath382 then @xmath442 is the only es - cover in @xmath317 , and the computation consists simply of a single call to part ( e ) of definition [ alg ] . if @xmath317 contains no parent vertex other than @xmath382 and @xmath55 is the last child of @xmath382 , then the es - covers in @xmath317 include @xmath443 and the es - covers of @xmath444 .
the first step of the computation is an application of part ( b ) of definition [ alg ] with @xmath445 .
the inductive hypothesis applies to @xmath444 , and the branch of the computation corresponding to @xmath446 consists solely of calls to part ( d ) because every vertex of @xmath447 is isolated ; consequently the latter part of the computation tree contains only one leaf , corresponding to @xmath443 . if @xmath317 contains a parent vertex other than @xmath382 , then the first step of the computation is an application of part ( b ) of definition [ alg ] with @xmath448 as in lemma [ recur ] , and @xmath449 .
the computation tree contains a single node representing this first step and also two subtrees , one corresponding to @xmath450 and the other corresponding to @xmath451 .
the proposition follows inductively from lemma [ recur ] .
we do not know whether or not it is possible to reformulate definition [ algact ] so that it always refers to @xmath2 instead of a computation tree .
such a reformulation would certainly be valuable , as the resulting activities formulas would shed light on the combinatorial significance of the interlace polynomials
. definition [ algact ] extends directly to computation trees representing implementations of other recursions .
for instance , if definition [ alg ] is augmented by incorporating pendant - twin reductions then the resulting computation trees will have active nodes representing these reductions , in addition to active nodes representing parts ( c ) and ( d ) of definition [ alg ] .
arratia , r. , bollobs , b. , and sorkin , g. b. ( 2000 ) the interlace polynomial : a new graph polynomial . in _ proceedings of the eleventh annual acm - siam symposium on discrete algorithms _
( san francisco , 2000 ) , association for computing machinery , new york , ps .
237 - 245 .
blser , m. and hoffmann , c. ( 2008 ) on the complexity of the interlace polynomial , in : _
stacs 2008 : 25th international symposium on theoretical aspects of computer science _
( bordeaux , 2008 ) , ps .
97 - 108 , available at http://www.stacs - conf.org/. | the interlace polynomials introduced by arratia , bollobs and sorkin extend to invariants of graphs with vertex weights , and these weighted interlace polynomials have several novel properties .
one novel property is a version of the fundamental three - term formula @xmath0 that lacks the last term .
it follows that interlace polynomial computations can be represented by binary trees rather than mixed binary - ternary trees .
binary computation trees provide a description of @xmath1 that is analogous to the activities description of the tutte polynomial .
if @xmath2 is a tree or forest then these `` algorithmic activities '' are associated with a certain kind of independent set in @xmath2 .
three other novel properties are weighted pendant - twin reductions , which involve removing certain kinds of vertices from a graph and adjusting the weights of the remaining vertices in such a way that the interlace polynomials are unchanged .
these reductions allow for smaller computation trees as they eliminate some branches .
if a graph can be completely analyzed using pendant - twin reductions then its interlace polynomial can be calculated in polynomial time .
an intuitively pleasing property is that graphs which can be constructed through graph substitutions have vertex - weighted interlace polynomials which can be obtained through algebraic substitutions
. keywords .
interlace polynomial , vertex weight , pendant vertex , twin vertex , series , parallel , graph composition , graph substitution , join , computational complexity , tree , tutte polynomial , jones polynomial mathematics subject classification . |
given an open bounded connected lipschitz set @xmath1 and an exponent @xmath2 , we consider @xmath3 and @xmath4 the first nontrivial eigenvalues of the @xmath0laplace operator with dirichlet and neumann boundary conditions , respectively .
we remind that these can be variationally characterized by @xmath5 and @xmath6 since exact values of such quantities are known only for specific values of @xmath7 and special domains @xmath8 , it is important to give ( sharp ) estimates for these quantities in terms of ( simple ) geometric quantities such as measure , perimeter , diameter , relative isoperimetric constants and so on . in this direction , the reader could consult for instance @xcite and the references therein . with this respect
the most celebrated example is the _ faber krahn inequality _ ( see ( * ? ? ?
* chapter 3 ) for example ) which asserts that the following minimization problem @xmath9 is ( uniquely ) solved by @xmath10dimensional balls of measure @xmath11 . by taking advantage of the homegeneity properties of the functional @xmath12 ,
the previous can be summarized as @xmath13 where @xmath14 is now any @xmath10dimensional ball .
then @xmath3 can be bounded from below in a sharp way just in terms the measure of the set @xmath8 .
we point out that by using _ isoperimetric inequality _ or _ isodiametric inequality _ , from we can infer similar lower bounds for @xmath15 in terms of the perimeter or the diameter of @xmath8 .
observe that problem becomes trivial , when we replace @xmath3 with @xmath4 . indeed , the latter is actually zero each time @xmath8 is disconnected .
it turns out that the natural counterpart for @xmath16 is rather the maximization problem , i.e. @xmath17 again , this is generally expected to be solved by @xmath10dimensional balls of volume @xmath11 . unfortunately so far this problem has resisted all the attempts to be attacked with the unique exception of the case @xmath18 ( and partially of the limiting cases @xmath19 and @xmath20 , see @xcite and @xcite respectively ) .
the _ szeg
weinberger inequality _
@xcite states in fact that for @xmath18 problem is ( uniquely ) solved by @xmath10dimensional balls of measure @xmath11 . as before ,
the result can be rewritten in scaling invariant form as @xmath21 with equality holding if and only if @xmath8 is an @xmath10dimensional ball .
we recall that the proof of for @xmath18 crucially exploits some pecularities of the laplacian , like linearity and the knowledge of the explicit form of eigenfunctions on balls .
.2 cm a couple of comments on the szeg - weinberger result are in order .
first of all , inequality says that @xmath22 can be estimated _ from above _ just in terms of the measure of @xmath8 .
but differently from the case @xmath15 , now _ we can not directly infer _ similar upper bounds for @xmath22 in terms of perimeter or diameter .
then one may wonder whether such a kind of estimates hold true or not for every @xmath2 , at least for some particular classes of sets .
secondly , we notice that if @xmath14 is any @xmath10dimensional ball , using the fact that @xmath23 ( see @xcite or proposition [ prop : nodal ] below ) , from we can infer @xmath24 which can be see as a weak version of the szeg - weinberger inequality .
again , a natural question is whether inequality can be extended to the case of @xmath25 or not .
the last two questions are the starting point of our analysis . in this paper
we prove indeed sharp upper bounds on @xmath4 in terms of diameter , as well as generalizations of for @xmath26 , under the additional constraint that @xmath8 is a convex set .
then our main scope is to investigate the following shape optimization problem with convexity and diameter constraints @xmath27 of course , by homogeneity of the quantities involved the value @xmath28 has no bearing and could be replaced by any constant @xmath29 . in theorem [ p_cop ] we show that the previous upper bound is finite , then we compute it and show at the same time that _ this problem does not admit a solution_. notably , we show that for every admissible set @xmath8 there holds @xmath30 and we find a sequence @xmath31 of convex sets degenerating to a segment such that @xmath32 we refer to section [ sec:3 ] for more details .
as we will show , the previous result can be summarized by the following scaling invariant sharp inequality @xmath33 where @xmath14 is any @xmath10dimensional ball and inequality sign is strict .
the proof of is based on a clever choice of a special test function which reminds ideas exploited in @xcite . by joining and the _ isodiametric inequality _ , we immediately get ( see corollary [ coro : weaksw ] below ) @xmath34 which generalizes to @xmath25 for convex sets , as announced above . by keeping in mind the way such an estimate was proved for @xmath18 , the previous can be seen as _ the trace of a potentially existing szeg - weinberger inequality for the @xmath0laplacian_. for ease of completeness and in order to neatly motivate some of the studies performed in this paper , it is useful to recall at this point that the _ minimization _ problem @xmath35 highlights the same features as problem .
for example , here as well the infimum can be computed and is not attained .
more interestingly , a minimizing sequence is again given by a family of convex sets collapsing on a segment .
for @xmath18 this is a celebrated result by payne and weinberger ( see @xcite ) , recently generalized in @xcite and @xcite to @xmath25 .
the result can be summarized by the sharp inequality @xmath36 where the constant @xmath37 is defined by @xmath38 it is quite natural to wonder if similar conclusions can be drawn also in the case of the following generalized notion of first eigenvalues @xmath39 and @xmath40 where @xmath41 .
quite interestingly , it turns out that for @xmath42 one has the following picture : * one can prove the analogue of ; .2 cm * this estimate _ is not _
sharp ; .2 cm * the maximization problem @xmath43 now _ admits a solution _ ; .2 cm * a lower bound like is not possible ( and the infimum in is @xmath44 ) ; on the contrary , for @xmath45 all the previous statements have to be reverted .
in particular , we have @xmath46 and it is rather the minimization problem for @xmath47 which is now well - posed ( see section [ sec:4 ] for more details ) .
in section [ sec:2 ] we prove some basic results concerning properties of @xmath48 and @xmath49 .
section [ sec:3 ] is devoted to the investigation of problem . in section [ sec:4 ]
we consider the case @xmath50 . as a consequence of some the estimates proved in the paper , in section [ sec:5 ] we exhibit a nodal domain property for neumann eigenfunctions . roughly speaking
, this shows that for @xmath51 eigenfunctions associated to @xmath48 can not have a closed nodal line .
finally , appendix [ sec:1d ] deals with a one - dimensional variational problem and its extremals , whose properties are crucially exploited to prove optimality of the estimate .
the first author has been partially supported by the gaspard monge program for optimization ( pgmo ) , by edf and the jacques hadamard mathematical foundation , through the research contract macro .
part of this work has been done during some visits of the first author to napoli .
the departement of mathematics of the university of napoli and its facilities are kindly acknowledged .
we fix two exponents @xmath7 and @xmath52 such that @xmath2 and @xmath53 for every @xmath1 open bounded lipschitz set , we use the standard sobolev spaces @xmath54 and @xmath55 , the latter being the completion of @xmath56 with respect to the norm of @xmath57 .
we then define the two quantities @xmath58 and @xmath59 it is useful to recall that @xmath48 can be defined through the unconstrained minimization @xmath60 where we set @xmath61 . also , we have that if @xmath62 is such that the equation @xmath63 admits a nontrivial solution in @xmath55 , then @xmath64 .
we start with a preliminary result on the quantities @xmath47 and @xmath65 .
[ lm : comparison ] let @xmath2 and @xmath66 , then we have @xmath67 the result is a plain consequence of hlder inequality .
let us prove for example the second inequality : we pick @xmath68 a function minimizing the rayleigh quotient which defines @xmath69 .
we then define @xmath70 as the minimizer of @xmath71 thus we get @xmath72 which in turn gives the desired inequality , by minimality of @xmath73
. we will also need the following very simple geometric result for convex sets .
[ lm : normali ] let @xmath1 be an open convex set , and let @xmath74
. then @xmath75 where @xmath76 denotes the outer unit normal to @xmath77 at the point @xmath78 .
since @xmath8 is convex , given @xmath79 we have that @xmath80 i.e. the hyperplane orthogonal to @xmath76 and passing from @xmath78 is a supporting hyperplane for @xmath8 . in particular , since @xmath81 we get @xmath82 , which concludes the proof .
the following is the main result of the paper .
this shows that the nonlinear spectral optimization problem @xmath83 does not admit a solution , but a maximizing sequence is given by a family of convex sets suitably degenerating to a segment .
of course , the value @xmath28 for the diameter constraint plays no special role and could be replaced by any constant @xmath29 .
[ p_cop ] let @xmath1 be an open bounded convex set and @xmath2 .
then we have @xmath84 where @xmath14 is any @xmath10dimensional ball .
equality sign in is never achieved but the inequality is sharp . more precisely , there exists a sequence @xmath85 of convex sets such that : * @xmath86 , for every @xmath87 ; * @xmath88 converges to a segment of length @xmath89 in the hausdorff topology ; * we have @xmath90 where @xmath91 is an @xmath10dimensional ball having radius @xmath92 .
we split the proof into two parts : at first we prove , then we construct the sequence @xmath93 verifying .
.2cm__proof of .
_ _ first of all , we observe that inequality is in scaling invariant form . then without loss of generality
, we can confine ourselves to prove that @xmath94 where @xmath14 is the ball centered at the origin such that @xmath95 .
let us take @xmath96 the first dirichlet eigenfunction for the ball @xmath14 , normalized by the conditions @xmath97 this is radially symmetric and solves @xmath98 we then take two points @xmath99 such that @xmath100 , and we define the two caps @xmath101 which are mutually disjoint ( see figure [ fig : fig3 ] ) . and @xmath102 . ]
we then take the function @xmath103 where the costant @xmath104 is given by @xmath105 by using this function @xmath106 in the rayleigh quotient defining @xmath107 , we get @xmath108 where the strict inequality holds since @xmath106 _ can not be an eigenfunction _ achieves the minimal value @xmath48 , then @xmath106 would solve @xmath109 in a weak sense .
let us take @xmath110 , by picking a ball @xmath111 with radius @xmath112 sufficiently small so that @xmath113 , we would obtain that @xmath106 is a nonnegative solution of the equation above in @xmath111 .
then by harnack s inequality ( see ( * ? ? ?
* theorem 1.1 ) ) one obtains @xmath114 thus getting a contradiction . we point out that _ we are not using any unique continuation argument_. ] . by performing an integration by parts in the integrals at the numerator , we obtain on @xmath115 . ]
@xmath116 where we used the equation solved by @xmath117 .
observe that the first integral in the right - hand side is non - positive . indeed
@xmath117 is a radially decreasing function , then ( with a small abuse of notation ) we have @xmath118 and the claim follows from lemma [ lm : normali ] , since @xmath119 .
the same computations apply to the other terms appearing in the numerator , thus obtaining @xmath120 which concludes the proof of .
.2cm__optimality of _ _ .
let @xmath14 be a ball of diameter @xmath89 .
we now prove optimality of : for this we need to construct a sequence of convex sets @xmath93 , all sharing the same diameter @xmath89 , and such that @xmath121 for all @xmath122 and @xmath123 let us denote by @xmath124 and @xmath125 the left and right circular infinite cone in @xmath126 whose axis is the @xmath127axis , having vertex in @xmath128 , and whose opening angle is @xmath129 .
we set @xmath130 in dimension @xmath131 , @xmath88 is nothing but a rhombus of diagonals @xmath89 and @xmath132 . in higher dimension @xmath88
is obtained by gluing together the basis of two right circular cones of height @xmath92 and radii @xmath133 ( see figure [ fig : fig2 ] ) . of theorem [ p_cop ] .
] we claim that for this family inequality holds true .
we start observing that whenever @xmath134 then the rescaled function @xmath135 belongs to @xmath136 and we have @xmath137 and @xmath138 thus we obtain @xmath139 now we denote by @xmath140 a function which minimizes the rayleigh quotient defining @xmath141 and by @xmath142 the corresponding function which minimizes the functional defining @xmath143 . without loss of generality
we can assume that @xmath144 .
inequality implies that @xmath145 then there exists @xmath146 so that @xmath147 weakly in @xmath136 and strongly in @xmath148 . moreover we also have , we have @xmath149 which in turn gives @xmath150 by the arbitrariness of @xmath151 . ]
@xmath152 thus @xmath153 does not depend on the @xmath154 variable and we will write for simplicity @xmath155 with a slight abuse of notation . for all
@xmath156 $ ] we denote by @xmath157 the section of @xmath102 which is orthogonal to the @xmath127axis at @xmath158 and set @xmath159 .
then we get @xmath160 let us denote by @xmath161 the previous minimal value , then by lemma [ lm : apres ] a minimizer @xmath162 does exist and is a solution to the following boundary value problem @xmath163 still by lemma [ lm : apres ] we have that @xmath164 and hence @xmath162 solves @xmath165 finally , by reminding that @xmath166 for @xmath167 , if we set @xmath168 then this solves @xmath169 which means that the radial function @xmath170 is a dirichlet eigenfunction of @xmath171 of a @xmath172-dimensional ball of radius @xmath92 , namely @xmath14 .
hence @xmath173 and we get @xmath174 this concludes the proof .
from theorem [ p_cop ] and the isodiametric inequality @xmath175 we can infer the following upper bound on @xmath176 , in terms of the @xmath10dimensional measure . we recall that for @xmath18 this is indeed a consequence of szeg - weinberger inequality .
[ coro : weaksw ] let @xmath1 be an open bounded convex set and @xmath2 .
then we have @xmath177 where @xmath14 is any @xmath10dimensional ball .
in this section we discuss variants and extensions of theorem [ p_cop ] for the quantity @xmath47 when @xmath178 .
actually , with the very same proof of theorem [ p_cop ] we can prove the following upper bound .
this time , the resulting inequality _ is not sharp _ ( see remark [ oss : nonva ! ] below ) .
for this reason , though the argument is the same , we prefer to give a separate statement .
[ p_cop2 ] let @xmath1 be an open bounded convex set and @xmath180
. then we have @xmath181 where @xmath14 is any @xmath10dimensional ball .
we use the same notation as in the proof of theorem [ p_cop ] . with @xmath96
we now indicate the function achieving @xmath182 , normalized by the conditions @xmath183 by optimality it solves @xmath184 in @xmath14 , with homogeneous dirichlet boundary conditions . as before ,
we consider the two caps @xmath185 and @xmath102 and take @xmath103 where @xmath104 is chosen so to guarantee @xmath186 . by inserting @xmath106 in the rayleigh quotient defining @xmath48 and proceeding as in theorem [ p_cop ]
we now end up with @xmath187 the term on the right - hand side is of the form @xmath188 for @xmath179 the previous expression is maximal for @xmath189 .
such a maximal value is given by @xmath190 , we thus get @xmath191^{1-\frac{p}{q}}.\ ] ] we have @xmath192 and the sum of the two terms into square brackets is less than @xmath28 by , thus we can finally infer . as in the case @xmath193 ,
theorem [ p_cop2 ] implies the following generalization of corollary [ coro : weaksw ] .
let @xmath1 be an open bounded convex set and @xmath194 .
then we have @xmath195 where @xmath14 is any @xmath10dimensional ball .
[ oss : nonva ! ] this time , the estimate is not sharp . we keep the same notation as in the proof of theorem [ p_cop2 ] and still consider @xmath42 . by adding and subtracting the term @xmath182 on the right - hand side of , recalling and using the concavity of @xmath196 ,
we get @xmath197\\ & \le \lambda_{p , q}(b)+\frac{q - p}{q}\ , \lambda_{p , q}(b)\ , \left[\int_{\omega_0\cup \omega_1 } |u|^q\ , dx-\int_{b } |u|^q \ , dx\right]\\ \end{split}\ ] ] since @xmath117 is radially decreasing , a simple rearrangement argument finally gives @xmath198 where @xmath199 is the ball centered at the origin , such that @xmath200 .
now observe that by using the _ quantitative isodiametric inequality _
( see ( * ? ? ? * theorem 1 ) ) @xmath201 where @xmath202 is a dimensional constant and @xmath203 is the _ fraenkel asymmetry _ of @xmath8 , defined by @xmath204 now suppose that there exists a sequence of convex sets @xmath31 such that @xmath205 for @xmath42 .
then from one would obtain @xmath206 since @xmath207 in @xmath14 , this would imply @xmath208 and thus from @xmath209 the first condition in implies that @xmath210 converges sense , i.e. the characteristic functions @xmath211 converge in @xmath212 to @xmath213 . ] to a ball , in contrast with the fact that @xmath214 for a ball ( see figure [ fig : fig1 ] ) . and
@xmath102 can not cover the whole ball . ] as in the case @xmath193 , we can then ask whether the following shape optimization problem @xmath215 admits a solution or not .
quite surprisingly , this time we can infer existence of an optimal shape . [ teo : esistenza ] let @xmath180 , for every @xmath29 problem admits a solution , i.e. there exists an open and bounded convex set @xmath216 such that @xmath217 for every @xmath218 open and bounded convex set . by theorem [ p_cop2 ]
we already know that the suprem is finite .
let us call it @xmath219 and take a maximizing sequence of admissible sets @xmath220 .
of course we can assume @xmath221 since @xmath47 scales like a length to a negative power , we can also assume that @xmath222 finally , we can suppose that there exists a uniform constant @xmath223 such that @xmath224 goes to zero ( see remark [ oss : azzero ] below ) . thanks to the bound on the diameters , we can assume that the whole sequence @xmath93 is contained in a common compact set @xmath225 .
thus the sequence is relatively compact for the complementary hausdorff topology in @xmath226 : more precisely , there exists an open set @xmath227 such that @xmath88 ( up to a subsequence ) converges in the hausdorff complementary distance to @xmath8 ( see ( * ? ? ? * corollaire 2.2.24 ) ) .
moreover , @xmath8 is still convex and its diameter equals @xmath11 ( see ( * ? ? ? * section 2.2.3 ) ) .
we also observe that the characteristic functions @xmath228 converges to @xmath229 strongly implies a uniform bound on their perimeters and measures .
then it is sufficient to use the compact embedding @xmath230 , where @xmath231 is the space of functions with bounded variation . ] in @xmath232 and @xmath233weakly in @xmath234 . without loss of generality
, we can assume that @xmath8 contains the origin , since @xmath48 is not affected by translations .
we are now going to prove that @xmath235 at this aim , let us take @xmath236 a function attaining the infimum in the definition of @xmath237 .
since @xmath8 contains the origin , for every @xmath238 the set @xmath239 is such that @xmath240 then by hausdorff convergence for every @xmath238 there exists @xmath241 such that @xmath242 we also set @xmath243 then for every @xmath244 and every @xmath245 , we take @xmath246 such that @xmath247 we claim that the sequence @xmath248 in bounded uniformly in @xmath249 and @xmath244 , i.e. there exists @xmath250 such that @xmath251 and , we have @xmath252 and on the other hand @xmath253 where we used and the very definition fo @xmath254 . by keeping the two estimates together , we finally get .
thus we can suppose that @xmath255 converges ( up to a subsequence ) to @xmath256 as @xmath249 goes to @xmath257 , and @xmath258 is in turn uniformly bounded .
then we get @xmath259 for every @xmath244 , where we also used the @xmath233weak convergence of the characteristic functions , recalled above .
we now observe that @xmath260 where @xmath261 is an accumulation point of the net @xmath262 , and also @xmath263 thus it is now sufficient to take the limit as @xmath264 goes to @xmath44 in order to get , by arbitrariness of @xmath117 .
this finally gives that @xmath8 is a solution of .
[ oss : azzero ] for @xmath179 the quantity @xmath48 can not be bounded _ from below _ in terms of @xmath265 only .
in other words , for @xmath42 we have @xmath266 a minimizing sequence is given by any family of convex sets @xmath85 such that @xmath267 indeed , by applying lemma [ lm : comparison ] we get @xmath268 if we now apply theorem [ p_cop ] to the right - hand side , we gain @xmath269 thus for a sequence of convex sets verifying , we get that @xmath270 converges to @xmath44 . in this case
, we can show that an upper bound on @xmath47 like that of can not hold true and actually we have @xmath272 indeed , a maximizing sequence is given by any family of open convex sets @xmath273 such that @xmath274 actually , this is a consequence of estimate below .
let @xmath275 and @xmath1 be an open and bounded convex set .
then we have @xmath276 and @xmath277 where the constant @xmath37 is given by . again by lemma
[ lm : comparison ] with @xmath278 , we get @xmath279 by using the lower bound , we can obtain .
.2cmestimate is obtained by combining with the isodiametric inequality .
the estimate is the counterpart of theorem [ p_cop2 ] for the case @xmath45 .
thus this time it is the minimum problem @xmath280 that actually makes sense . by suitably adapting the proof of theorem [ teo : esistenza ] , one can see that the previous problem admits indeed a solution .
we leave the details to the interested reader .
if @xmath117 is a function achieving the infimum in the problem defining @xmath48 , then by _ nodal domain _ we mean every connected component of the ( open ) sets @xmath281 as a consequence of theorems [ p_cop ] and [ p_cop2 ] , in the case @xmath282 we have the following result .
[ prop : nodal ] let @xmath1 be an open and bounded convex set and @xmath194
. then @xmath283 moreover , every nodal domain of a function achieving @xmath48 has to intersect @xmath284 .
the proof of immediately follows by combining , the faber - krahn inequality @xmath285 and the isodiametric inequality . to prove the second assertion ,
let us argue by contradiction .
we take @xmath286 achieving @xmath48 and we assume that the open set @xmath287 has a connected component @xmath288 .
we can further suppose that @xmath289 , then @xmath290 and it solves @xmath291 so that @xmath292 thanks to the fact that @xmath293 and @xmath294 .
this yields @xmath295 . by using the strict monotonicity of @xmath49 with respect to set inclusion and
, we then get @xmath296 which gives the desired contradiction . when @xmath297 , the previous argument to infer that first nontrivial neumann eigenfunctions can not have a closed nodal line was originally due to pleijel ( see @xcite ) .
for the laplacian , inequality was conjectured by kornhauser and stakgold ( see @xcite ) and can be obtained ( again ) as a byproduct of the szeg weinberger inequality .
let @xmath298 and let @xmath299\to\mathbb{r}^+$ ] be the lipschitz continuous function defined by @xmath300 as in the proof of theorem [ p_cop ] .
we consider the variational problem @xmath301 in this section , we state and prove some properties of extremals of , needed to prove the sharpness of estimate .
[ lm : apres ] with the notation above , we have @xmath302 and problem admits a solution .
any optimizer @xmath162 is a weak solution of @xmath303 moreover , @xmath162 vanishes at @xmath304 only and thus is also a weak solution of @xmath165 in order to prove existence , we define as in the proof of theorem [ p_cop ] the convex set @xmath305 , obtained by gluing together the basis of two right circular cones of height @xmath92 and radii @xmath92 ( see figure [ fig : fig2 ] )
. then we recall that the function @xmath306 coincides with the measure of the @xmath307dimensional section @xmath308 we consider the poincar - type constant @xmath309 where @xmath310 is the closed subspace of @xmath136 , built up of functions depending on the first variable @xmath311 only . by a standard comptacness argument
, it is easy to see that @xmath312 and the infimum is attained .
moreover , by construction , we obtain that @xmath313 and the restiction to @xmath127axis of any minimizer @xmath117 of minimizes as well .
.2cmwe now prove the claimed properties of extremals .
we begin by noticing that any solution @xmath162 of solves as well @xmath314 where the set @xmath315 is given by @xmath316 for @xmath317 , we observe that @xmath315 is a @xmath318 manifold , thus we can apply the lagrange multipliers theorem .
this yields that @xmath162 has to satisfy @xmath319 for some multiplier @xmath320 . by choosing @xmath321 and by using that @xmath322
, we can conclude that @xmath323 , i.e. @xmath162 is a weak solution of . for @xmath324 some care
is needed , since this time @xmath315 is no more smooth . however , by proceeding as in @xcite ( see also ( * ? ? ?
* lemma 5.8 ) ) , we can prove again that @xmath162 has to be a weak solution of .
.2cmwe now prove that @xmath325 if and only if @xmath326 .
since @xmath162 is admissible in and @xmath327 is positive , @xmath162 has to vanish somewhere in the interval @xmath328 .
let us suppose that @xmath329 , with @xmath330 .
then we consider the new function @xmath331 this function is still admissible in , thanks to the symmetry of @xmath327 .
moreover , by construction @xmath332 is non - decreasing and such that @xmath333 thanks to the fact that @xmath329 . by using the equation in weak form and testing it against @xmath334 ,
we get @xmath335 by the properties of @xmath332 and eveness of @xmath327 , we thus obtain @xmath336 this shows that @xmath332 still solves . in order to reach a contradiction , for all @xmath238 we can define the increasing function @xmath337 by symmetry
@xmath338 is still an admissible function in . by using the convexity inequality @xmath339
we have @xmath340 for @xmath264 small enough , this contradicts the optimality of @xmath332 in , since @xmath341 by assumption .
thus we get @xmath342 and @xmath162 vanishes at the origin only .
l. esposito , b. kawohl , c. nitsch , c. trombetti , the neumann eigenvalue problem for the @xmath344laplacian , to appear on atti accad .
lincei rend .
lincei mat .
. , available at http://arxiv.org/abs/1405.3535 a. henrot , m. pierre , variation et optimisation de formes .
( french ) [ shape variation and optimization ] une analyse gomtrique .
[ a geometric analysis ] mathmatiques & applications [ mathematics & applications ] , * 48*. springer , berlin , 2005 . | in this paper we prove a sharp upper bound for the first nontrivial eigenvalue of the @xmath0laplacian with neumann boundary conditions .
this applies to convex sets with given diameter .
some variants and extensions are investigated as well . |
galaxy formation is one of the great unsolved problems in astrophysics .
we have clear evidence that many galaxies were in place at redshifts of @xmath7 but little knowledge of how or when they formed . using lyman - break imaging techniques
, @xcite discovered a large sample of @xmath8 @xmath9 3 galaxies with rest - frame ultraviolet luminosities @xmath10 @xmath11 10@xmath12 and comoving number density comparable to that of @xmath13 galaxies today .
these lyman - break galaxies ( lbgs ) exhibit surprisingly strong clustering for @xmath14 , with overdensities @xmath15 @xmath11 3 and comoving correlation length @xmath16 @xmath17 5@xmath18 mpc @xcite . in hierarchical models of structure formation such as @xmath19cdm
, the lbgs will undergo significant merging and should be even more clustered , rare massive galaxies today . the damped lyman @xmath0 absorption systems ( dlas ) , on the other hand , are thought to be the progenitors of typical present - day galaxies @xcite . we therefore must learn about the mass and environment of high - redshift dlas if we want to understand the formation of typical galaxies like the milky way .
the damped lyman @xmath0 absorbers are dense clouds of neutral hydrogen seen in absorption along the line of sight to distant quasars .
they appear to be drawn from the bulk of the protogalactic mass distribution ; this conclusion is supported by the agreement between the comoving mass density of neutral gas at @xmath20 and the mass density of visible stars in current galaxy disks @xcite .
however , the total gas content can not distinguish cdm scenarios in which dlas are numerous low - mass protogalaxies from passive evolution models in which the dlas are more rarely occurring massive objects . determining the mass of individual dlas is also critical for testing models of galaxy formation within the cdm paradigm .
this determination can not be performed using the absorption characteristics of the dlas , as the comoving density of dlas can not be determined without knowing their typical cross - sectional area .
one way to distinguish among models for the dlas is to measure the kinematics of the gas @xcite .
this paper describes an independent method which seeks to determine the mass of the dlas by investigating the extent to which they cluster with lbgs .
since dlas are predicted by hierarchical cosmologies to be rather weak in emission , and observation is consistent with this @xcite , it may be easier to constrain their mass using large - scale structure techniques than by searching for their emission ( see @xcite ) .
section [ sect : lss ] describes the theoretical basis for our investigation of the cross - correlation of dlas and lbgs .
imaging observations and photometric data reduction are described in
[ sect : imaging ] , and our multislit spectroscopy is presented in [ sect : spectroscopy ] .
we discuss the search for emission from the dlas themselves in
[ sect : emission ] , search for evidence of clustering with nearby lbgs in
[ sect : clustering ] , and discuss our results in
[ sect : conc ] .
hierarchical models of structure formation predict that the most massive galaxies tend to form in regions of unusually high density , whereas low - mass galaxies are more uniformly distributed in space .
this leads to an enhancement in the clustering of high - mass galaxies , referred to as bias , since they typically form near each other in rare overdense regions .
clustering thus provides a way to probe the mass distribution of dlas ; if there is a significant overdensity of lbgs near damped lyman @xmath0 absorbers it indicates that the dlas reside in massive dark halos .
if , on the other hand , the dlas are preferentially found in regions with an underdensity of lbgs , it would argue that some environmental effect , perhaps ionizing radiation from massive stars in the lbgs ( see @xcite ) , prevents the occurrence of large column densities of neutral hydrogen in overdense regions .
the expected result in hierarchical cosmologies is intermediate ; dlas are predicted to be low - mass objects and should be found in both the ( proto)clusters and the field with typically only a small enhancement of lbgs nearby .
while the cross - correlation of galaxies and clusters has been studied at low redshift ( e.g. @xcite ) , these objects can also be studied through their autocorrelation .
the inevitable sparse sampling of dlas and other absorption systems caused by only being able to detect them on the line of sight towards quasars or bright galaxies makes it very difficult to measure their autocorrelation .
hence , the cross - correlation is the best , and perhaps only , way to study the bias of dlas .
another advantage is that the cross - correlation of dlas and lbgs can be established without detecting emission from the galaxy causing the damped lyman @xmath0 absorption .
the probability of finding an lbg in a small volume @xmath21 a distance @xmath22 from a known dla is enhanced versus the average density of lbgs , @xmath23 \rho_{lbg } dv \ ; , \ ; \ ; \ ; \label{eq : density}\ ] ] where @xmath24 is the cross - correlation function of dlas and lbgs . on scales where linear bias is a good model , the cross - correlation of any two classes of objects
is determined by multiplying the dark matter correlation function @xmath25 by the product of the bias factors of each class : @xmath26 the bias of a set of collapsed objects of a given mass can be determined using the extended press - schechter method ( see @xcite ) .
the autocorrelation of lbgs ( @xmath27 ) is known at @xmath28 @xcite but has not yet been determined at @xmath29 ( * ? ? ?
* hereafter s99 ) , probably due to the increased difficulty of obtaining redshifts at @xmath30 .
the bias of other high - redshift objects such as quasars , radio galaxies , or lyman @xmath0 emitters can also be determined using the cross - correlation function .
we took deep bri images of the field around the @xmath31 quasar br 0951@xmath404 @xcite with the lris instrument @xcite at the keck observatory .
this field was chosen because there are two dlas along the line of sight towards br 0951@xmath404 , at @xmath32 and @xmath33 @xcite .
table [ tab : obs ] lists the exposure time and final depth in each filter .
images were reduced using a mix of standard iraf routines and customized code .
the final images cover a @xmath34 field .
photometry was performed on the r image using sextractor @xcite and then a custom code was used to extract photometry from the final b and i images using the exact apertures selected in r. prochaska et al .
( in preparation ) describe our data reduction methods in greater detail and make our photometric catalog and images public .
we modified the @xmath35 color selection criteria of s99 for our johnson - cousins bri filter set . )
shift bluewards in b and r , allowing the approximation ( b@xmath4r)@xmath36(@xmath37 ) . because our i filter is close to theirs in effective wavelength
, our ( r@xmath4i)@xmath38 colors should be roughly 1.6 times their ( @xmath39 ) colors .
] objects located in the region of color - color space bounded by ( b@xmath4r)@xmath40 , ( b@xmath4r)@xmath41(r@xmath4i)@xmath42 , and ( r@xmath4i)@xmath43 , have a high probability of being lyman - break galaxies at @xmath44 .
we have also used photometric redshifts ( see @xcite ) and have found at least one lbg that would have been missed using only the two - color region of s99 .
almost all of the color - selected candidates are identified by the photometric redshift technique as having a high probability of being at @xmath45 .
the photometric redshift method accounts for the individual photometric uncertainties in the b , r , and i fluxes of each object and compares the object fluxes with a set of template spectra rather than a single two - color region .
carefully designed seven - color photometric redshift schemes yield a scatter of @xmath46 at these redshifts @xcite .
it should therefore be impossible for photometric redshift selection to create a redshift distribution as narrow as the @xmath47 ( @xmath92000 km / s ) redshift spikes seen by @xcite .
we visually inspected all candidate galaxies in our images and rejected roughly half of the initial candidates for having suspect photometry .
many of the rejected objects lied near the edge of our imaging field in regions of lower signal - to - noise .
it is no surprise that such a high fraction of the initial candidates had suspect photometry , as the color selection region is sparsely populated and therefore the small fraction of low - redshift galaxies scattered into the color - selection region by photometric errors can easily comprise a large fraction of the candidates .
bright ( i@xmath48 ) objects with clearly non - stellar morphology were rejected as low - redshift interlopers , but bright stellar objects were retained in case they were new quasars or lensing - magnified lbgs ; most lbgs will look similar to point sources at the resolution of our images .
we performed longslit spectroscopy of two objects near br 0951@xmath404 and multislit spectroscopy of this field with four different slitmasks , all with 1.5@xmath49 width slitlets and the 150 line / mm grating ( 30 resolution ) on keck - lris .
table [ tab : obs ] lists details of our observations .
each slitmask targeted about 15 candidate objects .
the entire lris imaging field was covered using 2 slitmasks ( termed left and right ) that overlap in the region near the quasar .
these masks were redesigned for the march 2000 run to eliminate slits on objects whose redshifts were already determined and to add slits on additional candidate lbgs .
we used a gg495 dichroic filter to block second order light in december 1999 but decided against this in the subsequent run because wavelengths below 4950 can be used to check for the presence of the lyman limit .
the spectroscopic data were reduced using standard iraf techniques called by a customized version of the bogus codestern / homepage / bogus.html ] .
the functions _ response _ ( halogen lamp spectrum fitting ) , _ background _ ( sky subtraction ) and _ apall _ ( spectrum extraction ) were performed interactively following the multislit data reduction cookbook available at http://mamacass.ucsd.edu/people/gawiser/cookbooks.html .
redshifts were determined interactively using a customized pgplot routine which compared the measured spectra with templates of expected absorption ( and emission ) lines for lbgs @xcite and e+a galaxies @xcite .
gawiser et al .
( in preparation ) describe our spectroscopic data reduction in greater detail .
we obtained spectra of 35 candidate galaxies in the field of br 0951@xmath404 ( see table [ tab : q0951 ] ) .
eight candidates turn out to be galaxies at @xmath5 , thirteen appear to be low - redshift interlopers ( typically ellipticals at @xmath50 ) and fourteen have inconclusive spectra .
our color - selected sample with i@xmath51 is directly comparable to the objects observed by s99 .
we find 0.5 candidates per sq .
arcmin , twice the candidate density of s99 ; the difference is unlikely to be caused solely by poisson fluctuations given that there are 19 objects in this category in table [ tab : q0951 ] .
our rate of determining redshifts for objects observed spectroscopically is comparable to s99 , and the fraction of candidates identified as interlopers , i.e. at @xmath52 , is higher but very sensitive to small - number statistics at present .
we will revisit this comparison once a larger sample of bri - selected galaxies is available .
figure [ fig:2clr ] shows the two - color diagram of all 2254 objects with i@xmath53 detected in the field with the status of lbg candidates shown .
table [ tab : gals ] lists photometry and photometric redshift information on each @xmath5 galaxy with its best - fit redshift and uncertainty .
the photometric redshifts based on bri alone are impressively accurate for the confirmed high - redshift galaxies , with a dispersion of @xmath54 , including three cases where @xmath55 differs from the spectroscopic redshift by only 0.01 . however , adding in the catastrophic failures where @xmath56 galaxies were assigned photometric redshifts @xmath57 would drastically increase this dispersion .
adding near - infrared ( jhk ) or optical z - band photometry would be quite helpful in discriminating against these intrinsically - red low - redshift interlopers .
the increased fraction of interlopers among candidates selected primarily by photometric redshift confirms that the s99 color selection region is the most efficient place to find lbgs .
however , we differ from s99 in one major respect .
we do not assume that the spectra with insufficient signal - to - noise for redshift determination are drawn from the same population of objects as those with identified redshifts .
thus there is a factor of two uncertainty in the number abundance of color - selected @xmath5 galaxies , both here and in s99 , caused by almost half of the candidates lacking redshift identifications .
extrapolating from the observed color distribution of lbgs at @xmath28 , s99 predict that roughly half of the @xmath29 lbgs will lie outside of the color selection region .
the failure of photometric redshifts to find these lbgs implies either that this incompleteness has been overestimated or that the photometric redshift technique needs to be modified to better identify lbgs with unusual colors .
figure [ fig : spectra ] shows the rest - frame spectra of the confirmed @xmath5 galaxies .
these spectra show the typical mix of lyman @xmath0 emission ( g4,g6,g7 ) and lyman @xmath0 absorption ( g1,g2,g8 ) as well as two vague suggestions of emission within a broad absorption trough ( g3,g5 ) .
the signal - to - noise is generally quite poor , illustrating why i=25 is usually considered the limit for ground - based optical spectroscopy .
at @xmath58 , lyman - break galaxies found within a 10@xmath49 radius of the quasar may be the source of damped lyman @xmath0 absorption .
this radius corresponds to the maximum impact parameter ( @xmath59 kpc ) inferred from absorption line statistics in the passive evolution model @xcite .
if dlas represent formed disk galaxies as postulated by @xcite , we should be able to detect emission from up to @xmath60 of them at r@xmath61 @xcite .
however , in hierarchical models of structure formation , a much higher fraction of dlas are expected to be too dim and/or too close to the quasar to yield detectable emission @xcite .
our survey hopes to differentiate between these hypotheses ; if emission from dlas is found repeatedly at large impact parameters or high luminosity it will represent a serious conflict with the predictions of hierarchical cosmology .
we performed a complete survey for possible emission from dlas on the line of sight toward the quasar down to i@xmath62 at angular distances up to @xmath63 , although the psf of the quasar hampers our survey at angles less than @xmath64 where hierarchical cosmology predicts that the emission from most dlas will be found .
we identified three objects within @xmath63 of the qso line - of - sight that had a reasonable chance of being at high redshift ( our selection criteria were less stringent in this region ) .
one of these , g1 , meets the color selection criteria but has i@xmath65 and appears to have a redshift of @xmath66 .
although g1 is @xmath63 from the qso , it may be responsible for the @xmath67 absorption system seen in siiv and civ @xcite .
the second nearby object appears to be a low - redshift interloper , and the third has inconclusive redshift .
another color - selected galaxy , g2 at @xmath68 , lies @xmath6 ( @xmath69 kpc ) from the qso and may be responsible for the @xmath70 absorption system seen in lyman @xmath0 and civ in our keck - hires spectrum of br 0951@xmath404 @xcite
. it will be interesting to determine the fraction of galaxies detected in both emission and absorption as a function of distance from the quasar line - of - sight .
we have not detected emission physically associated with either of the damped lyman @xmath0 absorbers towards br 0951@xmath404 spectroscopically , but all of the models predict that a larger sample is needed before we should expect to find an lbg physically associated with a dla .
figure [ fig : zhist ] shows the redshift histogram of objects in the field of br 0951@xmath404 , with a binsize of @xmath71 ( @xmath72 mpc ) chosen because most of our redshift uncertainties are smaller than this value .
we chose to place the @xmath32 dla in the center of a bin in order to maximize the chance of finding an overdensity ( or underdensity ) of lbgs near the dla if one exists .
however , the results described below are not altered by shifting the bin locations .
there are eight confirmed lbgs in the br 0951@xmath404 field , so the search for clustering is plagued by small number statistics . before trying to determine the amplitude of a cross - correlation function , it makes sense to ask if any clustering has been observed that is inconsistent with poisson fluctuations about the expected redshift distribution of lbgs .
the expected redshift distribution refers to the multiplication of the true redshift distribution of lbgs on an average line of sight times the observational selection function .
while the observational selection function can be simulated ( see s99 ) , the true redshift distribution of lbgs is unknown , so it is necessary to determine the expected redshift distribution empirically .
we have not yet observed a sufficient number of lbgs to determine how our expected redshift distribution differs from that of s99 , who found @xmath73 .
we do expect a shift of 0.2 in redshift , so we choose to bracket the possibilities by showing three possible expected redshift distributions in figure [ fig : zhist ] .
the mean of all three is @xmath74 ; the dashed one is a gaussian with @xmath75 and is considered the most likely , the dot - dashed one is a gaussian with @xmath76 and is considered the narrowest possible , and the flat dotted one is the broadest possible ( a uniform distribution from @xmath77 to @xmath78 ) .
the presence of a bin in our redshift histogram that contains three lbgs is only a weak indication of autocorrelation of the lbgs , because poisson fluctuations of any of the three expected redshift distributions generate an appearance of clustering at least this strong over 10% of the time in our monte carlo simulations .
is below the mean redshift , the result is more significant for the gaussian expected redshift distributions , but this could easily be weakened if the true mean of the expected redshift distribution is @xmath79 .
we will revisit this issue when the expected redshift distribution is better determined by observations in additional fields . ] a kolmogorov - smirnov test of the unbinned cumulative redshift distribution versus the three expected distributions fails to find any inconsistencies at the 95% confidence level .
a similar analysis of the reported redshifts of s99 indicates that the redshift histograms of all 7 of their fields are consistent with poisson fluctuations of their expected redshift distribution .
we are interested primarily in the clustering environment of the dlas .
this allows us to select the bins known a priori to contain a dla and to ask if a statistically significant overdensity ( or deficit ) of lbgs has been found there .
finding zero lbgs in the bin containing the @xmath32 dla should occur at least 40% of the time for any of the expected redshift distributions .
finding zero lbgs in the bins containing the @xmath33 dla or @xmath31 quasar is at least as common .
even the apparent triple coincidence of finding no lbgs in any of those bins occurs in at least 30% of our monte carlo simulations , so we do not have good evidence for an anticorrelation of dlas and lbgs .
our results provide an upper limit on the amplitude of the dla - lbg cross - correlation function , but it is not a particularly interesting upper limit , as even the observed strong autocorrelation function of @xmath28 lbgs would be difficult to detect with so few objects .
a vast improvement in statistics will be required to measure the cross - correlation of lbgs and dlas .
the minimum redshift uncertainty of our lbgs is @xmath80 , so correlations on comoving scales of a few mpc and smaller can only be probed via angular clustering .
while 8 lbgs and 2 dlas are insufficient to measure the angular cross - correlation precisely , we can again test whether the observed distribution is consistent with being random .
the presence of 1 out of 7 lbgs from our i@xmath81 sample within 15@xmath49 of the qso line - of - sight is inconsistent with a random distribution at the 95% confidence level . of the quasar and was therefore excluded from this analysis . ]
since g2 is not particularly close to the qso or dlas in redshift this angular overdensity appears to be a coincidence rather than the projection of a three - dimensional structure .
we have not found significant evidence for autocorrelation amongst the 8 lbgs at @xmath5 in our current sample .
it is therefore not surprising that we find no evidence for a cross - correlation between lbgs and damped lyman @xmath0 absorbers , as the latter should be more difficult to detect if the dlas are indeed less massive than the lbgs .
our results at @xmath1 are comparable to the limited studies of the clustering environment of damped lyman @xmath0 systems that have been performed at @xmath82 .
@xcite found only one out of eleven lyman - break galaxies in the field of q0000 - 263 to be within @xmath83 of the @xmath84 dla .
@xcite found two out of 27 lbgs in the field containing q0201 + 1120 to be within @xmath85 of the @xmath84 dla .
this failure to detect clustering of lbgs with dlas at @xmath86 stands in contrast to the detection of strong autocorrelation of lbgs at @xmath82 @xcite , although the latter was determined from a much larger sample of galaxies . in order to obtain better statistics in the future ,
we have begun a systematic survey for lyman - break galaxies associated with damped lyman @xmath0 absorbers at @xmath14 .
the increased sky density of lbgs bright enough for spectroscopic identification at @xmath14 and reduced sky noise near 5000 should allow us to obtain significantly improved statistics .
the coincidence of lbg overdensity and damped lyman @xmath0 redshifts in a number of fields would argue in favor of strong biasing and hence large mass of the damped lyman @xmath0 absorbers , which is counter to the predictions of the @xmath19cdm model .
we see no positive indication of clustering of lbgs with the quasar br 0951@xmath404 , but this quasar is at too high a redshift for us to be sensitive to an lbg - quasar cross - correlation except in the unlikely case that our expected redshift distribution is uniform . observing fields with a quasar at lower redshift will allow us to measure the cross - correlation of quasars and lbgs , which will test the belief that quasars are highly biased markers of protoclusters @xcite . with better statistics , this observational program will measure the bias of dlas , quasars , and lyman - break galaxies , revealing much about the typical mass and environment of these high - redshift objects .
we wish to thank kurt adelberger , chuck steidel , and dan stern for invaluable conversations and jeff newman , andy bunker , max pettini , and the referee for their comments which improved this paper .
we thank the keck observatory staff for their patient and skillful assistance with the observations .
we used the bogus package written by dan stern , andy bunker , and adam stanford and cosmic ray removal routines written by mark dickinson to reduce our multislit data .
amw was partially supported by nsf grant ast 0071257 , and jxp acknowledges support from a carnegie postdoctoral fellowship .
@xmath87 a plus or minus sign indicates that a specific other redshift , z @xmath88 is also a good fit .
otherwise , the errors are approximately symmetrical and roughly correspond to 95% confidence . + @xmath89 defined as the integrated probability that the redshift is in the range @xmath45 .
the method corresponds to assuming a uniform prior as a function of redshift but then optimizing the likelihood at each redshift over a set of possible template spectra .
+ @xmath90 the redshift which maximizes the photometric redshift likelihood function .
+ @xmath91 g4 falls outside of the color selection region and was included on the basis of its photometric redshift .
the very blue @xmath92 color makes it fall directly on the evolutionary track of starburst galaxies at @xmath93 which are at too low redshift to drop out of the @xmath94 filter as fully as the color criteria require . + [ 0.9 ] 04 with i@xmath81 ( with the addition of object g1 ) .
objects that were selected as candidate lbgs based on their colors and/or photometric redshifts have been assigned large symbols as identified in the upper left corner .
objects that fall in the color selection region but are represented only by small points were found by visual inspection to have suspect photometry and were therefore not considered as candidate lbgs .
[ fig:2clr ] , title="fig : " ] [ 0.9 ] 04 .
redshifts of dlas and quasar are shown by vertical dotted lines .
the range of possible expected redshift distributions described in the text is also plotted ; the dot - dashed line has mean @xmath74 , @xmath76 , the dashed line has mean @xmath74 , @xmath95 , and the dotted line is a uniform distribution from @xmath77 to @xmath78 .
all are normalized to produce 8 galaxies .
[ fig : zhist ] , title="fig : " ] | we report the first observations of the clustering environment of damped lyman @xmath0 absorption systems at @xmath1 .
color selection and photometric redshifts were used to select 44 candidate lyman - break galaxies brighter than i@xmath2 from deep bri images of the 35 arcmin@xmath3 field containing the quasar br 0951@xmath404 .
multislit spectroscopy of 35 candidate galaxies was performed and 8 of these candidates have been confirmed as @xmath5 lyman - break galaxies . with only bri photometry ,
the photometric redshifts are quite accurate for the spectroscopically confirmed galaxies but have a high rate of misclassification due to color degeneracies between lyman - break galaxies and low - redshift ellipticals .
both of the @xmath5 galaxies found within @xmath6 of the quasar line - of - sight appear to be causing absorption systems in the quasar spectrum .
we use a battery of statistical tests to look for clustering in the redshift histogram of the @xmath5 galaxies but do not find measurable clustering of these lyman - break galaxies with the damped lyman @xmath0 absorbers . with a larger sample of galaxies , our method should determine the cross - correlation between these objects , which probes the bias and hence the mass of the damped lyman @xmath0 absorbers . |
graphene - related materials have been in the scientific limelight for the past few years due to the numerous applications envisaged for them@xcite . besides the huge potential for applicability , one key feature that makes graphene particularly popular
is the simplicity with which many of its physical properties can be described , primarily due to the simple dispersion relation for its electrons .
the linearity of this dispersion relation around the fermi level enables the description of graphene electrons in terms of massless dirac fermions@xcite .
this introduces a great level of mathematical transparency in the portrayal of their properties , which nevertheless is limited only to a narrow range of energies around the fermi level .
energy values outside this range are often needed , for example when gated graphene systems are considered@xcite or when calculation of a relevant physical quantity requires an integral over energy , but lack the mathematical transparency of those within the linear dispersion regime . in this communication
we show how this limitation can be circumvented by deriving a fully analytical closed - form expression for the single - electron green function of graphene in real space , a quantity that is instrumental in describing the behaviour of graphene electrons . because green functions are used in the study of several physical properties , improvements in their mathematical description will enable far more transparent and insightful expressions for the corresponding physical quantities .
this is particularly true for distant - dependent interactions across a graphene sheet , for example the effect of an impurity on the physical properties of the system at a certain distance away from where it is introduced .
another example is the interaction of two impurities embedded into the sheet . in both cases , for distances of more than a few lattice spacings ,
the interactions involved are mediated by the conduction electrons of the graphene host .
the green functions calculated in this paper describe the equilibrium properties of these electrons at low temperatures , allowing us to investigate the underlying interactions within a mathematically transparent framework .
such a methodology allows for the prediction of certain features of the interaction without recourse to numerical calculations .
the remainder of the paper is organised as follows .
the general method for calculating the green function required is introduced in section [ methods_sec ] .
the important directions of the graphene geometry , namely the _ armchair _ and _ zigzag _ directions are illustrated and an explicit calculation for the real - space off - diagonal green function element in each of these directions is performed in sections [ arm_dir ] and [ zig_dir ] , respectively .
the accuracy of our approach is demonstrated by comparison with a fully numerical calculation .
an extension of the method for arbitrary directions and inter - sublattice cases is discussed in section [ oth_dir ] .
the potential applications of our approach are discussed in section [ gen_app ] , before an explicit illustration is given for the case of the magnetic interaction in graphene in section [ rkky_app ] . here
a fully analytical method is used to derive the principal features of the interaction within the rkky approximation .
the general formula for the single - electron green function , within the nearest - neighbour tight - binding framework , in its eigenstate basis is @xmath0 where @xmath1 is the energy , @xmath2 is the eigenvector labeled by the wave vector @xmath3 and @xmath4 is the corresponding eigenvalue defined as @xmath5 the quantities @xmath6 and @xmath7 correspond to the lattice parameter of graphene and its nearest - neighbour electronic hopping@xcite , respectively , which are hereafter used as our units of distance and energy .
our choice is such that the @xmath8 ( @xmath9 ) direction is aligned to the zigzag ( armchair ) geometry of the graphene lattice as is shown schematically in fig .
[ schematic ] .
a simple unitary transformation defines the real - space basis @xmath10 , where the index @xmath11 labels the two - atom unit cell and the index @xmath12 refers to the intra - cell atoms corresponding to the two distinct sublattices of graphene , represented by filled and hollow circles in figure [ schematic ] .
when projected onto two different states @xmath10 and @xmath13 located in real space by the respective vectors @xmath14 and @xmath15 , the green function is written as @xmath16 where the integrals over @xmath17 and @xmath18 are performed over the first brillouin zone of graphene .
although we have selected two states that belong to the same sublattice ( @xmath19 ) , this constraint can be easily relaxed and generalized to describe the propagator between sites on different sublattices . before proceeding
, we will outline the basic steps taken to obtain the green function .
we tackle the first integral by analytical continuation to the complex plane , where it is subsequently solved using the residue theorem.@xcite the remaining integral can then be solved in the case of moderately large separation vectors by using the stationary phase approximation ( spa).@xcite we first consider the case of separation vectors @xmath20 along the ( armchair ) @xmath9-direction . by showing how to obtain the green function for this particular case we hope to illustrate the general method for calculating green functions for any direction .
note that the position vectors appear only as a difference and can be further simplified by defining @xmath21 . in this case , the integral is performed over the brillouin zone shown in fig .
[ constant_energy ] and the first integral , over @xmath18 , can be evaluated by extending @xmath22 to the realm of complex numbers and changing the integration contour from a straight line on the real axis to the boundaries of a semi - infinite rectangle in the upper half of the complex plane , with its base lying on the real axis between @xmath23 and @xmath24 . because the integrand vanishes in the limit @xmath25 \rightarrow \infty$ ] and
because the parts of the contour that are parallel to the imaginary axis cancel each other out , the @xmath18-integral can be evaluated by simply identifying the poles of the integrand lying inside the integration contour and finding their respective residues @xcite , that is , @xmath26 note that the scalar product @xmath27 is now more concisely expressed as @xmath28 and that @xmath29 defines the wave vector @xmath30 that comes out of the first integral . although eq.([poles ] ) provides a unique definition for @xmath31 , it does not specify the sign of @xmath30 uniquely .
its sign is selected by imposing that @xmath30 must necessarily lie within the integration contour of the @xmath18-integral . in reciprocal space for a few different energies .
horizontal and vertical axes are rescaled as @xmath32 and @xmath33 , respectively , so that they are plotted as dimensionless quantities .
the rectangular shaded area delimits the first brillouin zone over which the integrals for the armchair direction case are taken .
constant energy plots for two specific energies , @xmath34 ( solid ) and @xmath35 ( dashed ) , are drawn with thicker lines .
the corresponding stationary wave vectors @xmath36 for these energies are highlighted with arrows .
note that the sign of @xmath36 , shown as positive here for simplicity , must be chosen according to the conditions outlined in the text.,scaledwidth=40.0% ] the @xmath17-dependence contained in the wave vector @xmath30 , as seen in eq.([poles ] ) , means that the integrand in eq .
( [ remaining ] ) is an oscillatory function of @xmath17 that oscillates very rapidly for large values of the separation @xmath37 . in this case , the only non - vanishing contribution to the green function comes from regions for which the phase of the exponential function is stationary . to locate these stationary points
we must impose that @xmath38 , which leads to solutions of the form @xmath39 note that due to a topological change in the constant energy surfaces of the function @xmath40 at @xmath41 we separate the energy band into two separate regions , namely , the inner region defined by @xmath42 and the outer region defined by @xmath43 .
both stationary solutions written above are valid throughout the entire energy spectrum . however , outside those specified regions , albeit solutions , they give rise to complex values for the wave vectors @xmath30 . with complex wave vectors , the integrand of eq.([remaining ] )
tends to vanish for any sizable separation @xmath37 , meaning that the stationary values outside the ranges listed in eq.([tilde ] ) should have very little influence in the overall results for the green functions.@xcite the same can be understood from purely geometrical arguments applied to the constant energy surfaces of @xmath40 in reciprocal space , depicted in fig .
[ constant_energy ] . when searching for stationary solutions for @xmath30 , which in this case lie parallel to the @xmath9-axis , the two solutions resulting from eq.([tilde ] ) are the only possible ( real ) ones within the rectangular brillouin zone of the hexagonal lattice .
the tilde symbol ( @xmath44 ) will hereafter be used to refer to the values of @xmath17 and @xmath30 satisfying the stationary condition .
therefore , the wave vector @xmath30 when expanded in a taylor series around the stationary value @xmath45 has no linear component and , up to second order , is approximated by @xmath46 where @xmath47 and @xmath48 note that the sign of @xmath49 must be chosen to ensure that @xmath30 lies within the @xmath18 integration contour as before .
the sign of @xmath50 is determined by its correspondence to the curvature of @xmath30 at @xmath45 .
if we now insert eq.([q ] ) into eq.([remaining ] ) and make use of the fact that @xmath17 and @xmath30 will not vary very much around their respective stationary values @xmath45 and @xmath36 , we are left with a much simplified expression for the green function , which now reads @xmath51 the remaining integral is a well known gaussian integral whose solution gives @xmath52 this can be rewritten in a more transparent fashion using the definitions in eqs .
( [ tilde ] ) , ( [ c1 ] ) and ( [ c2 ] ) to provide a completely analytical expression for the off - diagonal green function matrix element between two graphene sites separated by a distance @xmath37 along the armchair direction . for positive values of energy ( @xmath53 ) ,
we find @xmath54 where @xmath55 .
we again consider the distinct cases @xmath56 and @xmath57 and note that the only occasion when both stationary points contribute is at energies very near @xmath58 @xcite .
[ figarm ] compares both the real and imaginary parts of the green function for the case of @xmath59 obtained by the analytical expression above with those obtained through a numerical evaluation of eq.([above ] ) .
for @xmath60 we note that the real ( imaginary ) part of the green function is an odd ( even ) function of energy , as can be seen in fig .
[ figarm ] , and make use of the relations @xmath61 and @xmath62 . as a function of energy ( in units of @xmath63 ) for the case of @xmath64 .
top ( bottom ) panel shows the real ( imaginary ) part of the green function .
lines correspond to the results evaluated by eq.([sol_x ] ) whereas points are the result of brute - force numerical calculations of eq.([above ] ) .
, scaledwidth=40.0% ] of the bandwidth for which the error between the analytical and brute force numerical results is below 1% , plotted as a function of the separation @xmath37 ( in units of @xmath6 ) .
the circles ( triangles ) correspond to separations in the armchair ( zigzag ) direction.,scaledwidth=40.0% ] at first glance one might think that the replacement of the well established linear dispersion approximation with another that is valid only for asymptotically large values of separation is unlikely to improve the range of validity of the overall result .
however , as seen in fig .
[ figarm ] there is hardly any noticeable difference between the analytical and numerical results across the entire energy band .
because the analytical expression relies on the stationary phase argument , this remarkable agreement is likely to regress as the separation ( @xmath37 ) is reduced . to test how good an approximation eq.([sol_x ] )
is , in fig .
[ figerror ] the fraction @xmath65 of the bandwidth for which the relative error between the numerical and analytical evaluations is less than @xmath66 is plotted as a function of the separation .
the plot with circular points corresponds to the armchair direction . even for small separations ( @xmath67 ) , the energy range
for which the green functions are very accurately described by our analytical expression exceeds 90% of the bandwidth . in other words
, there is only a very narrow energy range in which the disagreement exceeds @xmath66 .
most remarkably , as the separation is increased this small range decreases very rapidly , indicating that eq.([sol_x ] ) is capable of accurately describing the green function across almost the entire energy spectrum for separation values larger than a few lattice parameters .
this is a major advantage when compared to the narrow fraction of the bandwidth that meets the linear dispersion criterion .
we now turn our attention to the case of separation vectors @xmath20 along the ( zigzag ) @xmath8-direction . by following the same approach described for the armchair direction
the relevant green function can be similarly calculated .
the major difference between the calculations is that the ordering of the integrals is swapped . for the zigzag direction ,
we first perform a contour integral over @xmath17 before making use of the spa to solve the remaining integral over @xmath18 .
we make a different choice of brillouin zone , as shown in fig .
[ fig_zz_energies ] , which will simplify the selection of stationary points later . by performing the first integral similarly to before , we arrive at an expression analogous to eq .
( [ remaining ] ) for the off - diagonal green function in the zigzag direction @xmath68 where @xmath30 now represents the poles coming out of the @xmath17 integral , which are given by @xmath69 it should be noted that in the zigzag case there are two contributions to the integral arising from the two possible sign choices of the poles in the definition above . the correct overall sign for @xmath30 in each case is selected as before by ensuring that @xmath30 lies within the integration contour . the contributions from each pole
must then be summed to give the final result .
as before , we assert that the only non - vanishing contributions to the integral in eq .
( [ remaining_zz ] ) occur when the phase of the exponential term is stationary .
imposing @xmath70 we find the stationary solution @xmath71 . as before with the brillouin zone for the zigzag case now illustrated by the shaded area .
the thick lines once more refer to constant energy plots for @xmath34 ( solid ) and @xmath35 ( dashed ) .
the corresponding stationary wave vectors @xmath36 for these energies are highlighted with arrows .
note that the sign of @xmath36 , shown as positive ( solid ) or negative ( dashed ) only for simplicity , must be chosen according to the conditions outlined in the text .
for clarity the arrows are shifted slightly in the vertical direction , but it should be noted that all the stationary points occur at @xmath72 .
, scaledwidth=45.0% ] unlike the stationary points calculated for the armchair direction , the zigzag direction stationary points are independent of energy .
this fact is evident when the constant energy plots in fig .
[ fig_zz_energies ] are examined from the perspective of stationary values of @xmath18 .
the separation of the energy band into two separate regions is not necessary in this case as the stationary points for both regions occur for the same value of @xmath18 .
the wavevector @xmath30 is now taylor expanded as before and we find expressions for @xmath49 and @xmath50 analogous to eqs .
( [ c1 ] ) and ( [ c2 ] ) @xmath73 and @xmath74 the superscript sign in the expressions for @xmath49 and @xmath50 refer to the choice of sign in eq .
[ zigzagpoles ] .
using these expressions the integral once more reduces to a gaussian integral whose solution gives @xmath75 where the sum over @xmath76 includes the contributions arising from the choice of sign for the poles .
( [ c1_zz ] - [ sol_zz ] ) can be combined as before to provide a single fully analytical expression for the off - diagonal green function matrix element between two graphene sites separated by a distance @xmath37 in the zigzag direction .
this is found to be @xmath77 where @xmath78 .
here we have a single expression that describes the green function across the entire band . in fig .
[ fig_zz_comp ] we compare the expression for the green function calculated here with the result of a fully numerical calculation , for the case of @xmath79 . as with the armchair direction , an excellent match is found across the entire band .
the plot in fig .
[ figerror ] ( triangular symbols ) shows the discrepancy between the numerical and analytical results as a function of distance .
once more we find that beyond a couple of lattice spacings there is only a very narrow energy range in which the disagreement exceeds @xmath66 . as a function of energy ( in units of @xmath63 ) for the case of @xmath79 .
top ( bottom ) panel shows the real ( imaginary ) part of the green function .
lines correspond to the results evaluated by eq.([sol_zz ] ) whereas points are the result of brute - force numerical calculations of eq.([above ] ) .
, scaledwidth=45.0% ] having presented the derivation of the green function for the separation vector along the armchair and zigzag directions , it is straightforward to generalize it to other cases . for arbitrary directions , although the procedure is similar , we shall see that the identification of the poles or stationary solutions may result from high order polynomial equations that are not always analytically solvable . in the armchair ( zigzag ) case , the expression for the stationary point @xmath80 ( @xmath81 ) is given by an easily solvable expression of the form @xmath82 .
this expression arises from the decision to take the contour integral along the @xmath83-space direction parallel to the separation vector @xmath20 , which results in a phase term equal to the pole of the contour integral .
since the expressions for the poles in the armchair and zigzag directions , eqs .
( [ poles ] ) and ( [ zigzagpoles ] ) respectively , are easily found from eq .
( [ graphene ] ) , the calculation of all the necessary quantities is quite straightforward . to extend this approach to an arbitrary separation vector
, we must first rewrite eq .
( [ graphene ] ) in terms of @xmath83-space vectors @xmath84 and @xmath85 which are parallel and perpendicular respectively to the required separation vector . following this
, we must perform the contour integral over @xmath84 to get an expression for the green function analogous to eq .
( [ remaining ] ) .
however , the expression for the poles of this contour integral will depend strongly on the separation vector chosen and will usually result from a high order polynomial equation that may need to be solved numerically .
it is important to note that this equation will only depend on the direction , and not the length , of the separation vector , so that once the green function for a particular direction has been constructed it is valid for any required distance across the graphene lattice in that direction .
a similar methodology holds for the case of green functions between sites on the different sublattices of graphene . in this case
( [ above ] ) must be altered slightly to read @xmath86 where now we have @xmath87 , with @xmath88 .
the integral can now be split into two parts with different phase terms coming from the two components of @xmath89 .
these can then be solved individually using the approach described above to give the required green function .
the usefulness of having an analytical expression for the real space green function , valid throughout the entire electronic energy band , becomes obvious when physical properties of graphene involving energy scales outside the linear dispersion region are investigated .
this is particularly advantageous when such properties carry size or position dependence because in this case eq.([sol_x ] ) for the armchair direction , or eq.([sol_zz ] ) for the zigzag direction , can be more concisely expressed in the form @xmath90 so that the @xmath1- and @xmath37-dependences are clearly distinguished .
furthermore , even in the case when the functional form of the coefficient @xmath91 is not particularly simple , the expressions in eqs.([sol_x ] ) and ( [ sol_zz ] ) can be used to expand the green function in a polynomial series , which is undoubtedly far simpler and more treatable than the original expression in eq.([above ] ) .
we anticipate that the ability to clearly isolate the distance dependence in the green function will allow a more transparent treatment of some of the more eagerly investigated properties of graphene . this approach will be shown more clearly in the next section when our formalism is used to investigate the magnetic interaction between two magnetic moments in graphene .
this type of interaction is perfectly suited for investigation using our approach since it is mediated by the conduction electrons of the graphene host . however , there are many other interesting physical properties that can be explored .
the interaction between precessing magnetic moments is one area of particular interest . within the random phase approximation
this dynamic coupling can be described by an integral over a complex function involving the convolution of green functions .
initial numerical results in carbon nanotubes , which are closely related to graphene , suggest that the range of the dynamic interaction may be greater than that of the more familiar static case@xcite .
we anticipate that an extension of the description provided below for the static case may be useful in attempting to solve the required integral analytically and understand the distance dependence of the dynamic coupling .
the ability of our approach to obtain green functions over a very large fraction of the energy band becomes increasingly important in this case due to the convolution of green functions of different energies that appears in the expression .
another topic that lends itself to our approach is the effect of disorder@xcite and in particular , the effect of an impurity , on the properties of graphene .
friedel oscillations , occuring in the local density of states as a function of distance from an introduced impurity , have been studied within the linear dispersion regime using a green function formalism@xcite .
although the local density of states is associated with the diagonal term of the green function , the distance dependence of the oscillations will be determined solely by the off - diagonal term calculated here .
similarly , the signatures of magnetic adatoms in graphene when probed by scanning tunneling spectroscopy have also been investigated using a theoretical approach@xcite .
this method again avails of green function methods within the linear dispersion regime .
we anticipate that our approach may allow for an extension of such studies to energies beyond the linear dispersion regime .
to demonstrate the power and applicability of our new formalism , we turn our attention to the magnetic interaction in graphene .
this interaction determines the relative orientation of magnetic moments embedded in graphene and has been the subject of many recent papers@xcite , as an understanding of this interaction is a major step in the implementation of graphene devices in the field of spintronics .
when the linear dispersion approximation is used , a cut - off function is required to prevent the result diverging due to high energy contributions .
there has been some debate about the effect of the cutoff function chosen on the resultant interaction calculated@xcite . other approaches to circumvent this problem
involve numerical calculations@xcite which can lack the transparency of an analytical solution . here
we shall show that the decay rate and oscillation period of the interaction as a function of distance emerge naturally from a simple calculation using our formalism and without resorting to an energy cut - off or a restriction to the linear dispersion regime .
the exchange energy , @xmath92 , within the rkky approximation@xcite is proportional to the static susceptibility , @xmath93 , which can in turn be written in terms of green functions , allowing us to write @xmath94 for two moments occupying like - sites in the graphene lattice separated by a distance @xmath37 , where @xmath95 is the fermi function .
this quantity relates to the energy difference between the ferromagnetic and antiferromagnetic alignment of the moments , with its sign giving the energetically favourable alignment . using the expression in eq .
( [ concise ] ) we write @xmath96 where @xmath97 , @xmath98 , t being the temperature and @xmath99 the boltzmann constant .
the integral in eq ( [ j_integral ] ) can be solved by replacing it with a contour integral in the energy upper - half plane . in this case
the poles are given by the zeroes of the denominator of the fermi function , namely the matsubara frequencies , @xmath100 where @xmath101 is an integer labelling the poles .
writing the coefficient @xmath102 as a taylor series , and the wavevector @xmath103 as a first order expansion , around the fermi energy gives @xmath104 using @xmath105 ( @xmath106 ) to denote the @xmath107th derivative of @xmath108 ( @xmath49 ) evaluated at the fermi energy .
this can be rewritten as @xmath109 in the low temperature limit , @xmath110 , this expression simplifies to one of the form @xmath111 in this form the oscillation period and decay rate of the interaction at different fermi energies can be easily extracted .
the decay rate in the asymptotic limit@xcite is determined by the leading term in eq ( [ j_simple ] ) , namely @xmath112 , suggesting that , in general , @xmath113 . however , at @xmath114 , it is straightforward to determine from eq .
( [ gf_et ] ) for the armchair direction , and from eq .
( [ gf_zz_et ] ) for the zigzag direction , that the coefficient @xmath115 vanishes and the decay rate is in fact determined by the first surviving term , @xmath116 , resulting in @xmath117 for undoped graphene , as reported elsewhere@xcite .
also , at @xmath114 , the oscillation period is perfectly commensurate with the graphene lattice spacing and thus oscillations are masked .
when @xmath118 , the leading term does not vanish , and the oscillation period is no longer commensurate with the lattice spacing , leading to the observed oscillatory interaction@xcite that decays as @xmath113 .
note that these conclusions are reached for values of @xmath119 regardless of whether they lie within the linear dispersion regime .
the correct decay rate and oscillatory behaviour for the rkky interaction in graphene have emerged naturally and in a mathematically transparent fashion from our formalism , without resorting to the linear response approximation or the need for a cut - off function .
as far as the authors are aware , this is the first time this has been performed within a fully analytical framework .
in summary , we have derived closed - form expressions for the single - electron green function of graphene in real space that does not rely on the linearity of its dispersion relation near the fermi level . the full derivation of this quantity for the two principal directions investigated on the graphene lattice has been presented along with a discussion for extending the methodology to other cases .
the expressions are valid across a very large fraction of the energy band and yet remain mathematically transparent .
the newly - acquired simplicity with which we can describe the electronic properties of graphene will lead to insightful new ways of studying the physical properties of this material at energy scales well beyond the linear dispersion regime .
the approach described here for the magnetic interaction can be modified straightforwardly to extend the validity of expressions derived in topics including , but not limited to , modelling the dynamic magnetic coupling and the effects of adatoms in graphene systems .
the authors wish to acknowledge support received from science foundation ireland and the irish research council for science , engineering and technology under the embark initiative .
msf thanks a. h. castro neto and v. m. pereira for fruitful discussions during his stay in boston and the condensed matter theory visitor s program at boston university for support .
a. k. geim and k. s. novoselov , nature materials , * 6 * , 183 ( 2007 ) a. h. castro neto , f. guinea , n. m. r. peres , k. s. novoselov and a. k. geim , rev
. mod . phys . * 81 * , 109 ( 2009 ) the electronic hopping parameter of graphene is commonly quoted in the literature as approximately @xmath120 . within this work
, the energy , @xmath1 , is expressed in units of @xmath63 . however , whenever @xmath7 appears in an expression it is important that a negative quantity is used .
it is common practice to set @xmath121 when dealing with expressions similar to eqs .
( [ gf_et ] ) and ( [ gf_zz_et ] ) .
however due to the sign choices implicit in the calculation of these quantities this may lead to an erroneous result .
if renormalising @xmath7 , a negative value will return the correct green function .
t. costa , d. f. kirwan and m. s. ferreira , phys .
b * 72 * , 085402 ( 2005 ) m. l. boas , _ mathematical methods in the physical sciences _ , wiley ( 1983 ) a. a. abrikosov , _ fundamentals of the theory of metals _ , north holland ( 1988 ) accounting for both stationary solutions improve the result only for energies very near the van hove singularities at @xmath58 .
other than that , the excluding conditions at eq.([tilde ] ) are sufficient . s. saremi , phys . rev .
b * 76 * , 184430 ( 2007 ) l. brey , h. a. fertig and s. das sarma , phys . rev
. lett . * 99 * , 116802 ( 2007 ) m. a. h. vozmediano , m. p. lopez - sancho , t. stauber and f. guinea , phys .
b * 72 * , 155121 ( 2005 ) v. k. dugaev , v. i. litvinov and j. barnas , phys . rev .
b * 74 * , 224438 ( 2006 ) a. m. black - schaffer , phys .
b * 81 * , 205416 ( 2010 ) j. e. bunder and h. h. lin , phys . rev .
b * 80 * , 153414 ( 2009 ) m. sherafati and s. satpathy , arxiv:1008.4834 ( unpublished ) m. a. ruderman and c. kittel , phys . rev . *
96 * , 99 ( 1954 ) t. a. kasuya , prog .
* 16 * , 45 ( 1956 ) k. yosida , phys
. rev . * 106 * , 893 ( 1957 ) d. f. kirwan , c. g. rocha , a. t. costa and m. s. ferreira , phys .
b * 77 * , 085432 ( 2008 ) a. t. costa , r. b. muniz and m. s. ferreira , new j. phys . *
10 * , 063008 ( 2008 ) f.
s. m. guimares , d. f. kirwan , a. t. costa , r. b. muniz , d. l. mills and m. s. ferreira , phys .
b * 81 * , 153408 ( 2010 ) f. culchac , a. latge and a. t. costa , arxiv:1007.3285 ( unpublished ) v. m. pereira , j. m. b. lopes dos santos and a. h. castro neto , phys . rev .
b * 77 * , 115109 ( 2008 ) c. bena , phys . rev . lett . * 100 * , 076601 ( 2008 ) b. uchoa , l. yang , s .- w .
tsai , n. m. r. peres and a. h. castro neto , phys .
lett . * 103 * , 206804 ( 2009 ) | among the many interesting features displayed by graphene , one of the most attractive is the simplicity with which its electronic structure can be described . the study of its physical properties is significantly simplified by the linear dispersion relation of electrons in a narrow range around the fermi level .
unfortunately , the mathematical simplicity of graphene electrons is only limited to this narrow energy region and is not very practical when dealing with problems that involve energies outside the linear dispersion part of the spectrum . in this communication
we remedy this limitation by deriving a set of closed - form analytical expressions for the real - space single - electron green function of graphene which is valid across a large fraction of the energy spectrum . by extending to a wider energy range the simplicity with which graphene electrons are described
, it is now possible to derive more mathematically transparent and insightful expressions for a number of physical properties that involve higher energy scales .
the power of this new formalism is illustrated in the case of the magnetic ( rkky ) interaction in graphene . |
intermediate polars ( ips ) have been the focus of numerous studies since they first emerged as a separate class of cataclysmic variables ( cvs ) .
possessing observational characteristics of both dwarf novae and magnetic cataclysmic variables , the current model for intermediate polars paints them as a fusion of these two types of systems .
an asynchronously rotating magnetic white dwarf primary accretes magneto - hydrodynamically controlled material from a truncated accretion disk which is fed via an accretion stream by a red dwarf secondary star @xcite .
the magnetically controlled accretion produces observational properties similar to those of a polar , while the visible accretion disk creates its signature double peaked spectral lines and other typical cv properties .
the different components of the intermediate polar system : the secondary star , accretion stream , accretion disk , accretion region on the white dwarf surface , and the white dwarf itself , separately dominate in wavelength regions ranging from infrared to x - ray .
while this makes ips very complicated objects to study , it also provides the opportunity to examine a multitude of astrophysical phenomena : from hot plasmas and accretion processes in the presence of high gravity and temperature to stellar astrophysics and binary evolution .
an opportunity for an in - depth observational study of an intermediate polar arose when we were awarded one million seconds of time on the _ extreme ultraviolet explorer satellite _ ( _ euve _ ) to observe ex hydrae . in collaboration with other researchers
, we used a number of other satellites to obtain simultaneous observations during the _ euve _ observation .
x - ray data were obtained with _
chandra _ ( spectroscopy ) , the _ rossi x - ray timing explorer _ ( _ rxte _ , spectroscopy and photometry ) , and the unconventional stellar aspect ( usa ) experiment ( photometry ) . along with _ euve _
, ultraviolet data were taken with the _ far ultraviolet spectroscopic explore _
( _ fuse _ , spectroscopy ) and the _ hubble space telescope _ ( _ hst _ , spectroscopy ) .
several ground based observatories were used to obtain optical data : apache point observatory ( apo , spectroscopy ) , rosemary hill observatory ( rho , photometry ) , and braeside observatory ( bo , photometry ) .
infrared spectra were obtained at the united kingdom infrared telescope ( ukirt ) .
this paper is the first in a series which will detail the observations obtained during our campaign .
ex hya has been the object of many previous observational studies for two main reasons : it is very bright ( @xmath2 ) and , with an inclination of @xmath3 @xcite , ex hya is one of only two x - ray eclipsing intermediate polars ( the other is xy ari , hellier 1997 ) .
however , some difficulties arise when interpreting observational data due to the fact that the relatively long white dwarf spin period of @xmath4 minutes is roughly two - thirds that of the relatively short binary orbital period of @xmath5 minutes @xcite . masses for the white dwarf in ex hya have been determined using various methods and have yielded a range of values : measurements of the accretion stream shock height and temperature give @xmath6 @xcite , radial velocities derived from optical time resolved spectra have led to slightly higher values for the white dwarf mass , @xmath7 @xcite , and a mass as high as @xmath8 has been calculated using binary star parameters @xcite .
the accretion curtain theory of ips , developed from high energy studies , suggests that modulation observed on the white dwarf spin period is due to varying optical depths through curtains of accreting material along our line of sight to the emission regions .
however , according to recent x - ray observations @xcite , it is more likely that photoelectric absorption combines with an occultation of one of the accretion poles to give the observed spin modulation .
maximum ( minimum ) of the spin phase in ex hya occurs when the upper magnetic pole points away from ( toward ) the observer .
prior to this campaign , the most detailed euv study of ex hya was that performed by @xcite , who obtained 180 ks of data with the _ euve _ satellite .
they found the spin phased light curve to be sinusoidal in shape and modulated by a factor of @xmath9 , with a maximum count rate of @xmath10 occurring at @xmath11 .
the binary phased light curve exhibits a bulge eclipse which extends over @xmath12 @xcite .
this bulge eclipse has a non - zero flux , being @xmath13 of the non - bulge eclipsed flux averaged over the white dwarf spin phases , and has an intrinsic column density determined to be @xmath14 @xcite .
there is also a feature attributed to an eclipse by the secondary star at @xmath15 @xcite with a centroid which remains constant to within @xmath16 s for all white dwarf phases implying an emission region located within @xmath17 of the white dwarf .
the ingress and egress times of the secondary eclipse were used to constrain the euv emitting region size to be @xmath18 at 68% confidence @xcite .
the euv spectra presented in @xcite contain highly ionized iron and neon emission lines which are indicative of a @xmath19 k plasma and a volume emission measure of @xmath20 . using the volume emission measure and the characteristic cooling time for a @xmath19 k plasma
, they determined the accretion rate onto the white dwarf to be @xmath21 .
ex hya was observed with the _ euve _ deep survey ( ds ; @xmath22 ) instrument and short wavelength spectrometer ( sw ; @xmath23 , @xmath24 , bowyer 1995 ) continuously for six weeks from 2000 may 5 - june 10 , totaling approximately one million seconds of on target time given a @xmath25 duty cycle . in spite of two safe holds caused by a battery failure and very high he ii @xmath26airglow , the data quality is very good . about @xmath27 of the ds photometer data and a bit less of the sw spectrometer data
were rejected during periods of high background / airglow since the deadtime corrections became very large ( @xmath28 ) .
typical ds deadtime corrections for the remaining data were around @xmath29 ( i.e. , the detector was too busy to process new photons @xmath30 of the time ) .
two types of products were produced : ds light curves and sw spectrometer spectra , and each is discussed below in detail .
photometry was performed from the ds images using a circular region of radius 13 pixels centered on the source and an annular background region with an inner radius of 50 pixels and outer radius of 100 pixels ( 1 pixel = @xmath31 , sirk et al .
the photon counts were then binned on a time interval of 10 seconds and barycentric corrections were applied to the mid - time of each bin . from this photometric data set , phase folded light curves were created for the binary period , the white dwarf spin period , and for a few combinations of these two periods ( see [ complcs ] ) .
figure [ splc ] shows the _ euve _ photometric data folded on the white dwarf spin ephemeris , @xmath32 @xcite , while figure [ bplc ] presents the _ euve _ photometric data folded on the binary ephemeris , @xmath33 @xcite .
the binary ephemeris phase zero is based on optical binary eclipse timings , while spin phase zero is based on optical spin maxima timings .
we will denote the spin phase as @xmath34 and the binary phase as @xmath35 .
the spin phased light curve ( figure [ splc ] ) , with resolution @xmath36 or @xmath37 s , can be fit with the sinusoidal function , @xmath38 , where @xmath39 counts s@xmath40 , @xmath41 counts s@xmath40 , and @xmath42 .
maximum occurs at @xmath43 and minimum occurs at @xmath44 ; both values differ from the expected values of @xmath45 ( maximum ) and @xmath46 ( minimum ) , and the maximum phase also differs from a previously determined value of @xmath47 @xcite .
the shift in maximum phase is real as the errors quoted for the spin ephemeris and period @xcite combine to give an expected error of @xmath48 within the light curve ; indicating a possible change in accretion geometry .
the count rate at spin maximum is @xmath49 the spin minimum count rate ; the non - zero minimum count rate value ( @xmath50 ) implies that either part of the euv emitting region is still in view during the minimum phases , or that there is euv flux within the binary system not confined to a small occulted region .
two theories have been advanced to explain the spin modulation in intermediate polars : absorption by an accretion curtain @xcite and occultation of the euv emitting region by the white dwarf @xcite . while both theories often seem to explain the observed phenomena in different ips , the euv flux in ex hya at minimum is less than half that at maximum , suggesting that if one of the poles is occulted , the other may be partially absorbed or the poles may have very different magnetic field strengths ( which could manifest as accretion poles with different temperatures . ) for reference , we refer to phases @xmath51 as spin maximum , phases @xmath52 correspond to spin decline , phases @xmath53 are spin minimum , and phases @xmath54 are spin rise , as illustrated in figure [ splc ] .
euv modulation seen in the binary light curve ( figure [ bplc ] , with resolution @xmath55 or @xmath56 s ) is the result of the inclination angle of ex hya , rather than emission over the binary orbit .
the fluctuations and dips seen throughout the light curve are due to eclipses of varying natures ; the large broad dip , or bulge dip , centered at @xmath57 , is caused by an increase in optical depth along our line of sight to the euv emitting region as viewed through the bulge created by the impact of the accretion stream with the outer edge of the accretion disk .
the temporal extent of this dip , from @xmath58 , implies that the bulge physically extends over half of the accretion disk .
the dip at @xmath59 has generally been attributed to an eclipse of the euv emitting region by the white dwarf , while the dip at @xmath60 has not been observed before in euv or other high energy photometry of ex hya ( e.g. , rosen et al .
1991 ; hurwitz et al .
1997 ; mauche 1999 ) . to extract the _ euve _ sw spectra , a spectral region 13 pixels tall centered on the spectrum , and two background strips each 100 pixels tall were chosen
( 1 pixel = @xmath61 , abbott et al .
the background spectrum was fit with a smooth legendre polynomial before subtracting .
utilizing this method ensures that no additional poisson noise is added to the spectrum @xcite .
spectra were created for a variety of binary and white dwarf spin phases .
figure [ spec ] presents spin phased spectra , but see figure 2 of belle , howell , & sirk ( 2002b ) for binary phased spectra .
the euv spectra presented here are not significantly better in signal - to - noise per spectrum than those presented in @xcite , even though our observations were @xmath62 longer .
assuming equal background levels during both observations , the signal - to - noise ratio in each of our _ four _ phase bins should be roughly @xmath63 greater than the signal - to - noise ratio in each of the two phase bins shown in @xcite . however , due to higher solar activity during our observations which increased the _
euve _ background level ,
the gain in signal - to - noise per spectrum is marginal .
the spin maximum spectrum exhibits the greatest flux , both in the continuum and emission lines , while the spin minimum spectrum has a continuum level consistent with zero .
the spin decline and spin rise spectra have average flux levels that are roughly equal to each other , in between the flux levels of the spin maximum and spin minimum spectra .
discerning between a continuum level and a forest of low signal - to - noise emission lines is difficult for these spectra , thus we only identify lines as probable that extend above the @xmath64 error level ( adopting a @xmath65 criterion for the lines virtually rules out all line detections . )
table [ t2 ] gives the observed wavelengths , fluxes , and the log of the temperature at which the emission line has its greatest power ( @xmath66 ) for the @xmath64 lines identified in the four spin phased spectra along with a few other unblended lines which are used for density determinations ( see [ sect_spec ] ) .
gaussians were fit to the emission lines in order to determine line parameters and the continuum was taken to be zero in each of the four spectra .
the error reported for each @xmath67 measurement is a @xmath68 error and does not include the rms error of the sw spectrometer , @xmath69 . to aid us in the selection of lines , each listed line was also identified in the summed spectrum .
the sw spectrometer has a moderate wavelength resolution of 0.5 , causing many of the emission lines to be blended together
. for the lines we believe are blended , both lines are listed in table [ t2 ] for the identification at a single wavelength measurement .
emission lines appearing in the euv spectra of ex hya are due to highly ionized iron and neon and have formation temperatures of @xmath70 k.
the exact interpretation of the phenomena present in the euv binary light curve of ex hya is dependent upon the accuracy of the binary ephemeris .
previous euv and high energy binary light curves of ex hya have shown a dip at @xmath59 ( or @xmath71 , mukai et al .
1998 ) which has been attributed to a stellar eclipse of the euv emitting region or white dwarf @xcite . however , there is no direct evidence for a stellar nature of this dip , other than the dip occurs _
close _ to the expected phase of @xmath72 . with a new dip appearing at @xmath60 , similar in nature to the @xmath59 dip , we are lead to question which , if any , dip is stellar in nature
, what could be the cause of a non - stellar dip , and why the dip at @xmath60 has not been observed before .
the binary phasing , @xmath73 , first determined by @xcite ( ephemeris constants were updated by hellier & sproats 1992 ) , and still used today , is based on @xmath74 and @xmath75 optical light photometric eclipse timings .
we note that the error reported in the @xcite phasing propagates to only @xmath76 today .
photometric eclipses seen in optical light in cataclysmic variables with accretion disks are often due to eclipses of the hot spot by the secondary star . the optical eclipse depth in ex hya
was seen to be shallow and variable over time @xcite and during outburst becomes broad and shallow @xcite implying that the accretion disk is eclipsed rather than the white dwarf .
thus , photometric phase zero according to the @xcite binary ephemeris for ex hya is likely to be a hot spot eclipse and not true binary zero ( i.e. , inferior conjunction of the secondary star ) .
how does the binary ephemeris photometric phase zero compare with spectroscopic phase zero and true binary zero ?
a stellar eclipse of the hot spot ( or bulge ) denotes photometric phase zero , while spectroscopic phase zero is taken to occur at the red - to - blue crossing of the optical emission line radial velocity curve ; the latter theoretically occurring at inferior conjunction of the secondary star .
however , due to a bias in the ( disk ) emission lines from the hot spot , the red - to - blue crossing can occur approximately when the hot spot is moving perpendicular to the observer s line of sight .
for example , @xcite showed that for cataclysmic variables with a strong optical emitting hot spot , a phase offset exists between spectroscopic and photometric phase zero ; a stellar eclipse of the hot spot would occur prior to the hot spot being perpendicular to our line of sight .
the measured radial velocity curves of the h@xmath77 and h@xmath78 optical emission lines of ex hya exhibit a red - to - blue crossing which lags photometric binary phase zero by approximately @xmath79 in phase @xcite .
the optical emission lines are likely to be biased by the motion of the hot spot @xcite so that the red - to - blue crossing may be interpreted as occurring when the hot spot is traveling approximately perpendicular to the observer s line of sight .
thus , traveling along the binary orbit , true binary phase zero would occur first , followed by photometric phase zero , and then spectroscopic phase zero , i.e. , an eclipse of the white dwarf by the secondary star ( if present ) would be seen before photometric phase zero
. applied to ex hya , this implies that an eclipse of the white dwarf by the secondary would be seen at a phase prior to photometric binary zero , such as @xmath59 .
however , the minimum depth of a stellar eclipse is likely to remain constant and not be dependent upon the spin phase during which it is observed
. we will show in the following section that a constant minimum is not the case for this `` eclipse '' .
therefore , although the phasing of the dip at @xmath59 may be correct for a stellar `` eclipse '' , its classification as such is still questionable .
the identification of the origin of the dips seen at @xmath59 and @xmath60 is facilitated by examining the binary phased light curve extracted at different spin phases ( figure [ wdcompare ] ) .
an intriguing feature of the euv light curves is the relative intensity of the dips at phases @xmath59 and @xmath60 . table [ countrate ] provides the unscaled mid - dip count rates for both the @xmath59 and @xmath60 dips .
the @xmath59 dip is shallowest during the spin maximum phase and not noticeably different from the other small dips appearing throughout these light curves .
the count rate at dip minimum in the spin maximum light curve ranges from @xmath80 greater than the mid - dip count rate in the spin minimum light curve to @xmath81 greater than the mid - dip count rate in the spin decline light curve .
the difference between mid - dip count rates for the @xmath60 dip is not as extreme , with the spin maximum mid - dip count rate being @xmath82 greater than the spin minimum mid - dip count rate , ranging to @xmath81 greater than the spin decline value .
gaussians were fit to the @xmath59 and @xmath60 dips in the averaged and all four spin phased binary light curves .
the results of the gaussian fits are given in table [ times ] .
the errors on the measured midpoint values fall within the light curve resolution of @xmath83 .
the midpoint of the @xmath59 dip appears to be consistent between the spin maximum and spin rise phases , at @xmath84 , and between the spin minimum and decline phases at @xmath85 , however , within the resolution of @xmath83 , these values are the same .
the @xmath60 dip midpoint is also consistent in phase at @xmath86 .
fwhm and fwzi values are a bit more scattered .
the fwhm values for both dips are the same , within errors , between spin maximum and minimum .
the spin rise and decline fwhm values are scattered and show no coherent pattern as are the fwzi values for each dip over all spin phases .
the short duration of both dips implies that the region being eclipsed is most likely small in size and close to the white dwarf surface , i.e. the inner accretion curtain close to the accretion pole where the euv radiation is emitted .
other asymmetries between the binary euv light curves of ex hya in figure [ wdcompare ] are seen within the broad dip extending from @xmath87 , which is attributed to photoelectric absorption of the accretion region by the bulge or hot spot @xcite .
this broad dip feature has been seen in the light curves of previous euv and x - ray studies of ex hya @xcite .
@xcite show a broad dip which extends for approximately @xmath88 in the @xmath89 kev band and has an asymmetric shape with a gradual decline and rapid rise while the eclipse seen in @xcite ( @xmath90 kev ) is a little wider in phase .
the euv bulge dip as shown in @xcite and @xcite extends from @xmath91 and shows the same gradual decline and rapid rise as the binary light curve presented in this paper . in a typical cataclysmic variable or intermediate polar system with an accretion disk , the hot spot , the region where the mass transfer stream impacts with the accretion disk , is centered around binary phase @xmath92 and does not extend `` backwards '' past the line of centers .
two things make the bulge in ex hya atypical ; it physically extends over half of the outer edge of the accretion disk ( see also hellier et al . 2000 who present a @xmath75-band light curve of ex hya with an orbital hump extending from @xmath93 ) , and it extends past binary phase zero .
however , we do note that some non - magnetic cataclysmic variables also exhibit an extended hot spot region ( e.g. , z cha , wood et al .
1986 ) .
the bulge dip appears slightly different when compared between each spin phase as shown in figure [ wdcompare ] .
several differences between the bulge dips in the different light curves are immediately obvious : for example , during the maximum phase , the bulge dip is shallower and has a more gradual decline .
however , the temporal extent of the bulge dip appears to be the same in all spin phases , with the exception of an extra shallow dip , preceding the bulge dip , from @xmath94 in the spin minimum light curve .
one could argue that the reason for the more rapid decline in flux in the spin minimum light curve is simply due to the fact that there is already a higher optical depth along the line of sight ( the accretion curtain ) to the accretion pole during the minimum phase .
however , this absorption effect would act to absorb the emission by the same amount throughout the entire binary orbit rather than causing extra absorption at a particular binary phase . in the early phases of the bulge dip ,
@xmath95 , the spin maximum count rate is @xmath96 greater than the spin minimum count rate . comparing this value to the count rate difference in the averaged spin light curve ( @xmath49 , figure [ splc ] ) , it is obvious that there must be extra absorption in addition to the accretion curtains during the bulge dip phases .
a change in bulge size is likely to manifest itself observationally in wavebands additional to the euv , and indeed , we do see a difference when comparing our optical data to the optical light curves from previous studies .
for example , our @xmath75-band light curve compared with the @xmath75 light curve of @xcite shows a decreased eclipse depth and very prevalent optical sidebands ( see figure 7 of belle et al .
2002b for a comparison of the light curves ) .
speculating that these sidebands might be seen in the euv photometry of ex hya , we phased our euv photometric data on common optical orbital sidebands ( or beat frequencies ) usually seen in intermediate polars : @xmath97 , @xmath98 , and @xmath99 , where @xmath100 is the spin frequency and @xmath101 is the binary orbital frequency .
coherent modulation , to some degree , was seen on all of these beat frequencies . in order to investigate the modulations seen on the orbital sidebands and other periods
, we searched our euv photometry for periods using the phase dispersion minimization ( pdm ) routine @xcite .
pdm provides statistical testing for a given period via a two - sided f - test .
figure [ power ] presents the results of this analysis , in the form of a thetagram , with the y - axis representing @xmath102 which expresses frequency significance , with lower values of @xmath102 indicating more statistically significant frequencies .
the photometric data was searched for periodicity over time intervals of 10 min ( 144 cycles day@xmath40 ) to 20 days ( 0.05 cycles day@xmath40 ) , consistent with the available sampling window provided by our long data set and the nyquist sampling limit . for the data set shown in figure [ power ] , which contains 97,274 samples , a value of @xmath103 defines the 99% confidence level for period believability .
the 95% confidence level is near @xmath104 .
the occurrence of sampling aliases was lowered by increasing the bin size over which the data were searched , as recommended by @xcite .
we see that in figure [ power ] essentially all the periods found are real , they are very well defined , and we have identified a few frequencies of interest on the figure , including that for the _ euve _ satellite orbital period ( @xmath105 min ) .
there are approximately 50 frequencies with @xmath102 values above the 95% confidence level ( @xmath104 ) appearing in the thetagram .
we have identified all of these frequencies as beats between the spin , orbit , and satellite frequencies , but have only labeled the more common beat frequencies on figure [ power ] , and frequencies that have been identified in previous photometry of ex hya .
the @xmath106 day@xmath40 frequency is strong with @xmath107 and many of the frequencies appearing in the range @xmath108 day@xmath40 are satellite beat frequencies , but we have not labeled these frequencies in order to avoid confusion on the plot .
the strongest beat frequencies are those between the satellite and orbit or spin frequency , rather than between the orbit and spin frequencies .
we note that the orbital frequency , @xmath101 , is much stronger in the euv than in our optical and x - ray data ( @xmath101 and @xmath100 have roughly equal strengths in the euv , while @xmath100 is @xmath81 stronger than @xmath101 in the optical , and @xmath109 stronger in the x - ray ) , which is the cause of the numerous beat frequencies visible in the figure [ power ] .
figure [ pdm ] is a time - sampled thetagram , where the gray scale indicates @xmath102 as in figure [ power ] .
this plot was made by a modified version of the pdm code allowing for time - sampled thetagrams to be produced .
here we used 5000 data points per segment , and then stepped through the entire data set by increments of 500 data points ( data point refers to a 10 sec bin and the sets of data points do not necessarily have equal time sampling ) .
we thus produced the trailed thetagram seen in figure [ pdm ] , which covers the time period from 2000 may 5 to june 10 , the entire _ euve _ observation .
along the y - axis , one time step refers to a bin of 5000 data points .
the known periods and sidebands , including the satellite beat frequencies , are fairly constant and strong throughout the temporal sampling covered in the plot .
however , there are some interesting frequencies around 25 day@xmath40 and 33 day@xmath40 which appear suddenly in the middle of the data set .
these frequencies are likely to be beat frequencies , but their appearance during only part of the observation is intriguing and suggests that one or more of the beating frequencies is not constant throughout the data set , i.e. , the geometry and/or physical conditions of ex hya are evolving , to some extent . the frequencies appearing at @xmath110 day@xmath40 are associated with the frequencies @xmath111 and @xmath112 but it is interesting to note that what appears to be the @xmath111 frequency wanders back and forth throughout the entire data set .
the appearance of the wandering and intermittent periods in our high energy data probably indicates that there are additional or changing sources of reprocessing of the euv radiation
. these may always be present in the euv or may be manifestations of the variable nature of ex hya . a detailed study of the periodicities seen in our contemporaneous observations of ex hya in the x - ray , euv , uv , and optical will be presented in a future work .
figure [ spec ] presents euv spectra extracted from the four white dwarf spin phases , as described in figure [ splc ] , and table [ t2 ] presents the stronger lines we have identified in each spectrum , along with their fluxes and @xmath113 .
the emission lines seen in these spectra are from highly ionized neon and iron , with the coolest lines ( neon ) having @xmath114 k and the hottest lines ( iron ) having temperatures of @xmath115 k. only five lines appear ( above the @xmath64 error ) in each of the four spin phased spectra , four of which are blended : @xmath116 , @xmath117/ @xmath118 , and @xmath119/ @xmath120 .
these lines have @xmath121 6.3 , 6.9 , 6.7 , 7.1 , and 6.9 k , respectively . using the flux ratios of these lines
, we can further confine the temperature range of the emitting plasma . according to the optically thin plasma model of mewe , gronenschild , & van den oord ( 1985 ) , the flux ratios of the lines at @xmath122 , @xmath123 , and @xmath124
match those expected for a @xmath125 k plasma .
the flux ratio determined temperature is in agreement with the @xmath66 values of the individual lines .
however , if we consider the flux ratio of the two ne lines we have identified in the spin maximum and decline spectra , / @xmath126 to @xmath127 , the plasma temperature determination is significantly lower , near @xmath128 k , for both spin phases .
line ratios may also be used to determine plasma densities according to @xcite , table iva , using lines that have @xmath129 k. only two of the four spectra have emission lines near this temperature which are unblended : spin maximum and spin decline .
the _ spin maximum _ line ratio , @xmath130/ @xmath131@xmath132 , gives a density of @xmath133 ( note that @xmath130 actually has @xmath134 ) .
the line ratio of @xmath135/ @xmath136@xmath137 in the _ spin decline _ spectrum gives a lower density , @xmath138 ( @xmath139 when considering the lower limit determined from the uncertainty ) .
these values may indicate an observed density range , although poorly constrained .
we would like to note that the density determined from the @xmath140 line ratio does have a problem , as the ratio uses lines from two different charge states and depends strongly on the plasma temperature .
for example , if we were to initially assume a low density plasma , table iv of @xcite shows that the value of the @xmath140 ratio measured here matches well with a @xmath141 k. considering that our spin decline line ratio finds a low density , it is likely that assuming a constant temperature of @xmath142 k for the @xmath140 ratio does not adequately take into account the strong dependence on temperature of the different charge states , and that the lines are likely to be formed in a lower density plasma . using only the reliably determined density above ( @xmath1 ) , our value does not agree with the electron densities which have been previously determined for ex hya .
@xcite found an electron density of @xmath143 using the ratio of the / blend to the emission lines in their _ euve _ spectra .
@xcite found a value of @xmath144 from our _ chandra _ spectrum using the @xmath145 line ratio .
we note that while these two values are self - consistent , the @xcite value was determined from blended lines in different charge states and the @xcite ratio is affected by photoexcitation .
the appearance of the / lines at @xmath146 in each spectrum ( except during mid - bulge dip , see figure 2 of belle et al .
2002b ) suggests that a high temperature plasma ( @xmath19 k ) is constantly in view during the times in which we see other euv emitting regions of lower temperature and lower density .
the absence of the @xmath146 line , and also the @xmath147line , during the mid - bulge dip phase suggests that these lines are formed interior to the bulge and that their emitting region is roughly confined to the orbital plane .
the emission line centers appear to shift over the four spin phases .
each spin phase is comprised of many orbital phases , so radial velocities of these lines will be a complex sum of both motions . also , a radial velocity curve over spin phase will not have much meaning when only containing four points .
however , it is interesting to note that the / blend at @xmath146has a fairly constant negative velocity at all spin phases with an average value of @xmath148 .
the other two strong lines which appear in each of the four spectra also have negative velocities over all four spin phases .
the @xmath116 emission line has its most blue - shifted velocity of @xmath149 during spin maximum and its lowest blue - shifted velocity of @xmath150 during spin decline .
the / blend at @xmath147 is not blue - shifted to the same extent and has its greatest blue - shift of @xmath151 during spin minimum .
the sw spectrometer rms error of @xmath69 @xcite , which has not been added to the @xmath68 errors given for the line velocities above , translates to a velocity error of @xmath152 at @xmath153 , @xmath154 at @xmath155 , and @xmath156 at @xmath157 ; even assigning a systematic maximum uncertainty to the line velocities , we note that the emission lines are generally still blue - shifted .
( the @xmath158velocity of ex hya has not been subtracted from these values due to the fact that it is not well constrained . ) taken at face value , the phase - independent blue - shifts of the emission lines in the _ euve _ spectra tell us that the lines may be associated with some outflowing material . however , we note that @xcite reported that the @xmath159emission lines had a wavelength offset consistent with zero ( though the error from the wavelength accuracy they quote does allow these lines to be blue - shifted . ) the accretion pole regions would be a likely site for outflowing material in the form of a wind or jet .
the disappearance of the @xmath146 line during the bulge dip would mean that the poles must be highly inclined with respect to the spin axis if the cause of its blue - shift is due to jets .
we believe instead that this outflow may originate as a wind from the inner edge of the accretion disk , a feature common in high mass transfer binaries , possibly associated with shocks in the accretion disk ( see below )
. such an emission site would constantly provide outflowing material directed at the observer , while also being absorbed during the bulge dip phases . in an upcoming paper discussing our _ hst _ data @xcite , we show that blue - shifted absorption components ( p cygni profiles ) , features which are indicative of outflowing material , are seen in several of the uv emission lines .
the current intermediate polar model has the white dwarf evenly accreting matter from an annular accretion disk via accretion curtains onto both magnetic poles ( e.g. , patterson 1994 ) .
however , @xcite have recently shown that ex hya exists in a spin equilibrium state such that this model may not hold true .
they first note that although a disk may form in any intermediate polar , it is generally far from keplerian due to the influence of the magnetic field on the accretion flow in the disk . for ex hya , where the spin period is close to the binary orbital period ( @xmath160 for ex hya compared with @xmath161 for other ips ) , the equilibrium condition is such that the corotation and the circularization radii are roughly equal to the distance to the inner lagrange point @xmath162 .
this condition is true for a special class of intermediate polars which possess a white dwarf with a magnetic moment of @xmath163 ( and @xmath164 hr ) , corresponding to a magnetic field strength of @xmath165 mg for the white dwarf in ex hya .
the euv data presented in this paper support the model alluded to by the @xcite study in which non - uniform accretion from the disk onto the white dwarf magnetic poles occurs .
the extended bulge dip seen in the binary light curve ( figure [ bplc ] ) , the absorption dips seen throughout the binary light curves of figure [ wdcompare ] , along with the differing absorption amounts seen between these same light curves are all evidence of non - uniform accretion .
we propose that the magnetic field in ex hya is strong enough to begin pulling some material vertically from the outer edge of the disk , thereby forming a large accretion curtain which would produce the very broad and extended bulge .
material after binary phase 0 would accumulate as it encounters the bulge in the accretion disk .
this concept is depicted in figure [ curtain ] ( although we have drawn accretion curtains which extend only from the inner or outer disk , it is more likely that there would be a continuous curtain extending between the two disk edges . ) matter attaching to the magnetic field lines at the outer edge of the disk would form accretion streams which would act to absorb radiation emitted from the accretion poles and would produce dips within the binary light curves , such as those seen at @xmath59 and @xmath60 . these dips are reminiscent of dips seen in polar light curves ( e.g. , belle , howell , & mills 2000 ) . inspection of figure [ wdcompare ] shows that the dips at @xmath59 and @xmath60 have varying depths ( values are given in table [ countrate ] ) , with the dips during spin minimum having the greatest amount of absorption . at spin and binary phases such that spin minimum and the bulge eclipse are coincident ( @xmath166 and @xmath167 ) , material in the bulge , already extending above the plane of the disk , may be controlled by the magnetic field and produce denser streams .
if the dips seen in the binary phased light curve are caused by material attached to the field lines , why are the dips seen along the binary phase rather than the spin phase ?
the reality of the magnetic field attachment to the outer edge of the disk is that the field lines will be affected by magnetic torque such that the lines will become twisted . while the white dwarf magnetic pole has already started to rotate away from the bulge
, the distortion of the magnetic field caused by the torque will allow the field lines to remain attached at the bulge . as long as the field lines remain attached
, matter will be kept from falling back down to the accretion disk . in this way
, the dips would be seen along the binary phase rather than the spin phase .
if the scenario just described is true for ex hya , the strength of the magnetic field needed to pull material from the outer edge of the disk can be estimated . using the equation , @xmath168 to calculate the radius of the magnetosphere @xcite and setting @xmath169 equal to the outer accretion disk radius
, we can solve for @xmath170 .
we use white dwarf masses of @xmath171 @xcite and @xmath172 @xcite to represent the range of primary masses which have been determined for ex hya .
optical spectroscopic emission line velocities ( our optical data will appear in a future paper ) , and the white dwarf mass range , give the outer disk radius as @xmath173 or @xmath174 . with an accretion rate of @xmath175 @xcite , we find a magnetic field strength of @xmath176 mg .
we would like to note that using the optical emission line velocities to determine the disk radius assumes that some of the disk is in keplerian motion .
the far - reaching magnetic field has implications other than producing an extended bulge .
as stated previously , the circularization radius is roughly equal to the corotation radius @xcite .
this intimates that part or possibly all of the accretion disk is rotating with the white dwarf .
although we do not know the latitude of the magnetic poles on the white dwarf surface , the modulation seen on the spin period suggests that the poles are moderately inclined ( @xmath177 , kim & beuermann 1995 , 1996 ) with respect to the rotation axis of the white dwarf .
the accretion disk , then , would not feel a constant magnetic force , but rather would feel the highest field strength at the regions of the disk closest to the poles .
at these locations along the disk , the disk material would be more affected by the magnetic field strength and hence would become controlled by the magnetic field .
thus , we have a situation in which some amount of material in two chunks , or pie slices ( one facing each pole ) , along the accretion disk are rotating with the white dwarf , while the rest of the accretion disk may ( or may not ) be rotating with the white dwarf .
the outer disk velocities measured from the optical spectra are around @xmath178 , while the velocity of the magnetic field sweeping through the disk is calculated at the corotation radius ( @xmath179 cm ) to be @xmath180 , again enforcing the statement that the outer disk is corotating with the white dwarf magnetic field .
an issue pertaining to the corotating pie slices in the accretion disk is the presence of shocks which may form as matter of different velocities interacts in the accretion disk .
modeling shocks is beyond the scope of this paper , however , we argue that the similar velocities of the white dwarf and accretion disk ( as discussed previously ) would produce very low energy shocks .
high energy emission lines are already seen in the spectra and because the bulge dip eclipses all euv emission ( i.e. , we can not distinguish between the white dwarf or accretion disk being eclipsed ) , it is not possible to discern if these lines originate in a shock from the accretion disk , or from the shocks and high energy plasmas associated with regions near the magnetic accretion poles .
however , if the shocks produce winds , they may be responsible for the highly blue - shifted lines seen in the euv spectra .
another question to answer is why strong power is not seen at the @xmath181 frequency if the magnetic field is coupled to the accretion disk ( though we note that while this frequency is not strong , it is above the 95% confidence level . )
this might be expected if there was a bright euv point source situated on the edge of the disk , however , the outer edge of the disk is vertically extended and cool and will emit most strongly at lower energies , such as optical wavelengths .
we will show in a future paper describing our optical data that strong power _ is seen _ on the @xmath181 frequency .
we should also consider the possibility that an overflow accretion stream may be responsible for the small dips seen at @xmath59 and @xmath60 , instead of magnetically controlled streams causing these dips .
we argue that this is not likely for several reasons .
first , there are many dips seen throughout the bulge dip , starting at @xmath182 and continuing through to @xmath183 , which makes it likely that these dips are caused by material associated with the bulge .
second , the overflow stream is unlikely to extend far enough above the orbital plane in order to be along the line of sight to the accretion poles .
third , this overflow stream would follow along its ballistic trajectory and would not be in a position to cause a dip occurring after binary phase 0 .
the photometric and spectroscopic data presented here have provided us with some intriguing results for ex hya .
this system clearly does not conform to the traditional intermediate polar model . a broad bulge dip ,
numerous smaller dips throughout the binary light curve , and the inconsistencies between binary light curves extracted at different spin phases all indicate that ex hya may be accreting material from the outer edge of the accretion disk thereby producing an extended bulge and large accretion curtain . for material at the outer edge of the disk to become magnetically controlled
, the magnetic field of the white dwarf must be on the order of @xmath184 mg .
these extended regions may also be reprocessing sites of euv radiation , as can be seen from the many strong beat frequencies appearing in the single thetagram ( figure [ power ] ) and time - series thetagram ( figure [ pdm ] ) .
iron spectral line ratios give a temperature of @xmath185 k for the emitting plasma but provide only a poorly constrained density of @xmath138 .
it is likely that regions of varying temperature and density are viewed simultaneously throughout the spin orbit .
future papers will investigate the physical meaning of the strong frequencies in greater detail , along with relating the euv data to our other wavelength observations .
we would like to thank c. mauche for his contributions and comments on this work and the anonymous referee for very valuable comments .
this research was partially supported by nasa grant nag5 - 8644 to sbh .
99 abbott , m. j. et al .
1996 , , 107 , 451 allan , a. , hellier , c. , & beardmore , a. 1998 , , 295 , 167 belle , k. e. , et al .
2002a , in preparation belle , k. e. , howell , s. b. , & mills , a. 2000 , , 112 , 343 belle , k. e. , howell , s. b. , & sirk , m. 2002b , in asp conf .
264 , continuing the challenge of euv astronomy : current analysis and prospects for the future , ed .
s. howell , j. dupuis , d. golombek , f. walter , & j. cullison ( san francisco : asp ) , 96 beuermann , k. & osborne , j. p. 1988
, , 189 , 128 bowyer , s. 1995 , proc .
spie , 2517 , 97 crdova , f. a. , mason , k. o. , & kahn , s. m. 1985 , , 212 , 447 fujimoto , r. & ishida , m. 1997 , , 474 , 774 hellier , c. 1997 , , 291 , 71 hellier , c. , kemp , j. , naylor , t. , bateson , f. m. , jones , a. , overbeek , d. , stubbings , r. , & mukai , k. 2000 , , 313 , 703 hellier , c. , mason , k. o. , rosen , s. r. , & crdova , f. a. 1987 , , 228 , 463 hellier , c. & sproats , l. n. 1992 , ibvs , 3724 hurwitz , m. , sirk , m. , bowyer , s. , & ko , y .- k .
1997 , , 477 , 390 kaitchuck , r. h. , hantzios , p. a. , kakaletris , p. , honeycutt , r. k. , & schlegel , e. m. 1987 , , 317 , 765 king , a. r. & wynn , g. a. 1999 , , 310 , 203 kim , y. & beuermann , k. 1995 , , 298 , 165 kim , y. & beuermann , k. 1996 , , 307 , 824 mason , e. et al .
2000 , , 318 , 440 mauche , c. w. 1999 , , 520 , 822 mauche , c. w. , liedhal , d. a. , & fournier , k. b. 2001 , , 560 , 992 mewe , r. , gronenschild , e. h. b. m. , & van den oord , g. h. j. 1985 , , 62 , 197 mukai , k. , ishida , m. , osborne , j. , rosen , s. , & stavroyiannopoulos , d. 1998 , in asp conf .
137 , wild stars in the old west , ed
. s. howell , e. kuulkers , c. woodward ( san francisco : asp ) , 554 mumford , g. s. 1967 , , 15 , 1 patterson , j. 1994 , , 106 , 209 reinsch , k. & beuermann , k. 1990 , , 240 , 360 rosen , s. r. , mason , k. o. , & crdova , f. a. 1988 , , 231 , 549 rosen , s. r. , mason , k. o. , mukai , k. , & williams , o. r. 1991 , , 249 , 417 siegel , n. , reinsch , k. , beuermann , k. , van der woerd , h. , & wolff , e. 1989 , , 225 , 97 singh , j. & swank , j. 1993 , , 262 , 1000 sirk , m. m. , vallerga , j. v. , finley , d. s. , jelinsky , p. , & malina r. f. 1997 , , 110 , 347 stellingwerf , r. f. 1978 , , 224 , 953 warner , b. 1995 , cataclysmic variable stars ( cambridge : cambridge univ . press ) wood , j. , horne , k. , berriman , g. , wade , r. , odonoghue , d. , & warner , b. 1986 , , 219 , 629 lcccccccc @xmath186 ( 6.3 ) & @xmath187 & @xmath188 & @xmath189 & @xmath190 & @xmath191 & @xmath192 & @xmath193 & @xmath194 + @xmath195 ( 5.8 ) & @xmath196 & @xmath197 & @xmath198 & @xmath199 & @xmath200 & @xmath201 & & + @xmath202 ( 6.9)/ + @xmath203 @xmath204 ( 6.7 ) & @xmath205 & @xmath206 & @xmath207 & @xmath208 & @xmath209 & @xmath210 & @xmath211 & @xmath212 + @xmath213 ( 5.8)/ + @xmath203 @xmath214 ( 5.7 ) & @xmath215 & @xmath216 & @xmath217 & @xmath218 & & & & + @xmath219 ( 7.0)/ + @xmath203 @xmath220 ( 7.0 ) & @xmath221 & @xmath222 & & & & & @xmath223 & @xmath224 + @xmath225 ( 6.9 ) & @xmath226 & @xmath227 & & & & & @xmath228 & @xmath229 + @xmath230 ( 7.0)/ + @xmath203 @xmath231 ( 6.9 ) & @xmath232 & @xmath233 & @xmath234 & @xmath235 & & & & + @xmath236 ( 7.0 ) & @xmath237 & @xmath238 & & & & & & + @xmath239 ( 7.0 ) & @xmath240 & @xmath241 & @xmath242 & @xmath243 & & & & + @xmath244 ( 7.0 ) & & & @xmath245 & @xmath246 & & & & + @xmath247 ( 7.0 ) & & & @xmath248 & @xmath249 & & & & + @xmath250 ( 7.1)/ + @xmath203 @xmath251 ( 6.9 ) & @xmath252 & @xmath253 & @xmath254 & @xmath255 & @xmath256 & @xmath257 & @xmath258 & @xmath259 + ccccccc averaged & @xmath269 & @xmath270 & 0.024 & @xmath271 & @xmath272 & 0.044 + maximum & @xmath273 & @xmath274 & 0.019 & @xmath275 & @xmath272 & 0.041 + minimum & @xmath276 & @xmath277 & 0.025 & @xmath271 & @xmath272 & 0.031 + rise & @xmath273 & @xmath270 & 0.030 & @xmath275 & @xmath278 & 0.045 + decline & @xmath276 & @xmath279 & 0.030 & @xmath271 & @xmath274 & 0.025 + | on 2000 may 5 , we began a large multi - wavelength campaign to study the intermediate polar , ex hydrae .
the simultaneous observations from six satellites and four telescopes were centered around a one million second observation with _
euve_. although ex hydrae has been studied previously with _
euve _ ,
our higher signal - to - noise observations present new results and challenge the current ip models .
previously unseen dips in the light curve are reminiscent of the stream dips seen in polar light curves . also of interest
is the temporal extent of the bulge dip ; approximately 0.5 in phase , implying that the bulge extends over half of the accretion disk .
we propose that the magnetic field in ex hydrae is strong enough ( a few mg ) to begin pulling material directly from the outer edge of the disk , thereby forming a large accretion curtain which would produce a very broad bulge dip .
this would also result in magnetically controlled accretion streams originating from the outer edge of the disk .
we also present a period analysis of the photometric data which shows numerous beat frequencies with strong power and also intermittent and wandering frequencies , an indication that physical conditions within ex hya changed over the course of the observation .
iron spectral line ratios give a temperature of @xmath0 k for all spin phases and a poorly constrained density of @xmath1 for the emitting plasma .
this paper is the first in a series detailing our results from this multi - wavelength observational campaign . |
numerical simulations and analytic calculations have shown that the gravitational growth of tiny density perturbations in the @xmath8cdm cosmological model leads to a wealth of structures over cosmic time .
the spatial distribution of gravitationally bound structures follows a web - like pattern composed of dense clusters , filaments , sheets and otherwise near - empty void regions ( see e.g. * ? ? ?
* ; * ? ? ?
this `` cosmic web '' is the weakly nonlinear manifestation of the large - scale tidal field and its formation can be readily understood in the first order lagrangian perturbation theory of the zeldovich approximation @xcite .
recent large scale galaxy surveys have observationally confirmed the presence of the cosmic web in the universe ( e.g. * ? ? ?
* ; * ? ? ?
since galaxies form and evolve together with the cosmic web in which they are embedded , it is a question of high relevance whether the properties of galaxies depend on scales beyond the most immediate vicinity and thus reflect the effect of large - scale tidal fields on the assembly of the galaxy .
general ellipsoidal density perturbations collapse sequentially along their three axes leading to matter accreting first onto a sheet , then collapsing to a filament before matter finally streams along the filaments into the densest regions . as a consequence of these processes , the large - scale morphology of the cosmic web might leave its imprint on the orientations of galaxies through the anisotropy of accretion and mergers .
recently , evidence has accumulated that galaxies are indeed not oriented randomly with respect to one another . on small scales ,
several studies report a radial alignment of satellite galaxies towards the centre of mass of the group or cluster in which they reside ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
such an alignment is predicted by n - body simulations of dark matter halos ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
it is commonly believed that tidal torques are responsible for this radial alignment ( e.g. * ? ? ?
* ) , as tidal forces on these relatively small scales ( @xmath9 mpc ) are sufficiently strong to affect halo shapes on short time - scales .
more surprisingly , however , observational results indicate that the orientations of galaxies are correlated up to scales of order @xmath1 times larger than the virial radii of their host haloes .
@xcite , @xcite and @xcite detect an alignment of galaxy shapes for the luminous red galaxies ( lrgs ) in the sdss survey out to the largest probed scales ( @xmath10 mpc ) .
@xcite report a similar result for redshifts @xmath11 . for lower luminosity galaxies , in particular for @xmath12
, no alignment has been detected yet ( see also * ? ? ?
* ) although some results using n - body simulations indicate spin alignment around voids @xcite . correlations of the spin of galaxies have been the subject of several analytical models ( see e.g. * ? ? ? * for a review , and references therein ) . compared with more local effects ,
the relative weakness on galaxy scales of tidal forces arising from the @xmath13 scales requires that such effects maintain spatial coherence over rather long time scales for them to be able to affect galaxy properties .
also it is not clear how they are counteracted by local non - linear effects in the most immediate vicinity of the galaxy : non - linear gravitational effects as well as ram pressure exerted on the disk gas could potentially lead to a reorientation of galactic angular momentum . furthermore ,
the impact of hot in contrast to cold mode accretion separated by a halo mass of @xmath14 of gas onto galaxies ( cf . * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) on the orientation of the resulting disks is not clear .
while understanding the formation of galaxies in the context of their host large - scale structure is a key question for galaxy formation theory , a precise knowledge of correlated galaxy orientations is of high importance also for future high - precision weak lensing studies of dark energy .
weak lensing signals can be severely affected by systematic contaminations due to large - scale alignments in the spatial distribution of cosmic matter on scales from a few to more than a hundred mpc ( due to galaxy ellipticity - ellipticity alignment : @xcite ; and due to the much more problematic shear - ellipticity alignment : @xcite ) .
approaches that have been suggested so far to remove the contamination due to shear - ellipticity alignment suffer from several limitations and uncertainties : the statistical down - weighting of the effect proposed by @xcite relies on precision determinations of photometric galaxy redshifts ; other techniques , such as proposed by e.g. @xcite , depend heavily on toy models ( e.g. * ? ? ?
* ) which do not include any ( as yet unknown ) dependence of alignments on galaxy luminosity , mass , or redshift .
several studies using @xmath15-body simulations however find evidence for a dependence of the strength of halo - lss alignment on mass and redshift ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
unfortunately , it is currently still computationally unfeasible to use hydrodynamical simulations of galaxy formation of the necessary resolution in a sufficiently large volume to investigate the correlation function of galaxy shapes and spins . to circumvent this barrier ,
some authors have used large - volume @xmath15-body simulations together with a prescription to assign galactic disk orientations to the dark matter haloes which typically either assumes perfect alignment with the dm halo spin or some random misalignment angle between the two ( e.g. * ? ? ?
also analytic models based on the halo model have been developed that include prescriptions for small scale satellite - central alignments @xcite .
probability distributions for the angles between halo and galaxy spins have been determined by various authors ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
accepting the current unfeasibility of large volume hydrodynamic simulations , we propose a different approach .
combining local measurements of the alignment of galaxies with the large - scale structure as has been investigated by @xcite and @xcite in pure @xmath15-body simulations circumvents the need for large volumes but can still provide constraints on the degree of galaxy alignment to be expected .
the results of @xcite on the basis of four sph simulations of the formation of single isolated galaxies indicate an alignment of the spin of these galaxies with the intermediate principal axis of the tidal field at turnaround that persists to a slightly weaker degree also to @xmath2 . in this paper , we study the orientation of a sample of @xmath1 galactic discs with respect to their larger scale environment in a hydrodynamical cosmological simulation of the formation of a large filament .
we have chosen this particular large - scale environment as it provides an environment where the gravitational shear is expected to be relatively large so that weak lensing shear should have good signal - to - noise .
any intrinsic alignment of galaxies on these scales will thus impact the measurements through the gi - lensing term .
prior investigations in dark matter cosmological simulations have shown evidence for alignment of halo spin with filaments . however , considering the gas component rather than dark matter , the filament is a high pressure environment so that hydrodynamic effects , such as ram - pressure , are expected to play a role in the evolution of the simulated galaxies .
this paper is organised as follows : we describe the details of our numerical simulation , the method to identify galaxies and the selection of galaxies for our analysis in section [ sec : galform_methods ] .
we first study the alignments between the gas , stellar , dark matter angular momenta and the total halo spin in section [ sec : galform_aligninternal ] . then
, we investigate in section [ sec : galform_align_lss ] the alignment of the various components with the cosmic large - scale structure , quantitatively described by the tidal field . in section [ sec : numerical_artefacts ] we discuss the numerical convergence of our results , before a comparison of our results to observations is given in section [ sec : galform_observations ] .
finally , we summarise and provide our conclusions in section [ sec : galform_summary ] .
simulated stellar light composite images of two well - resolved disk - galaxies using the @xmath16 , @xmath17 and @xmath18 band light and taking into account dust absorption .
the white circles correspond to the radius containing 95 per cent of the stellar mass .
the yellow line corresponds to the stellar angular momentum vector direction , the red line to the gas angular momentum.,title="fig:",scaledwidth=35.0% ] + simulated stellar light composite images of two well - resolved disk - galaxies using the @xmath16 , @xmath17 and @xmath18 band light and taking into account dust absorption .
the white circles correspond to the radius containing 95 per cent of the stellar mass .
the yellow line corresponds to the stellar angular momentum vector direction , the red line to the gas angular momentum.,title="fig:",scaledwidth=35.0% ] we simulate the formation of a large cosmic filament and the galaxies embedded in it using the amr code ramses @xcite to evolve dark matter , gas and stars .
gas dynamics is computed using a second - order unsplit godunov scheme , particles ( dark matter and stars ) are evolved using the particle - mesh method .
the simulation includes standard recipes for star formation @xcite , supernova feedback and chemical enrichment @xcite .
we do not repeat details of these sub - grid models but kindly refer the reader to the mentioned papers .
the filament was chosen by visual inspection of the environment of the most massive haloes in a dark - matter - only preflight simulation .
we then chose a filament that is very prominent ( and thus relatively massive ) and not criss - crossed by too many smaller scale filaments .
initial conditions were generated using the grafic-2 tool from a power spectrum computed with linger .
we adopt cosmological parameters as obtained from the wmap 5th year data release @xcite , i.e. we use the density parameters @xmath19 , @xmath20 , @xmath21 , a power spectrum normalisation of @xmath22 and a spectral index @xmath23 for the primordial spectrum , as well as a hubble constant of @xmath24 with @xmath25 .
the filament is simulated using the `` zoom - in '' technique as a high - resolution region of @xmath26 diameter at redshift @xmath2 inside of a comoving @xmath27 cosmological box .
the effective initial resolution of our simulation is @xmath28 particles and grid points inside this high - resolution region , dropping in steps of 8 to @xmath29 in the remainder of the box .
additional refinements are confined to the high - resolution region only .
we adopt a refinement strategy based on density threshold so that the number of particles per cell ( and the baryon mass per cell ) remains roughly constant in time .
since the filament moves slightly in space , we introduce a `` colour '' field to mark the lagrangian volume of the filament .
this field is passively advected during the formation of the structures and thus follows the evolution of the filament perfectly . in order to save memory , we used as colouring variable the metallicity of the gas , with a very low ( but non - zero ) value of @xmath30 solar in the high resolution region and exactly zero outside .
additional refinements are only allowed where the metallicity exceeds one tenth of this initial value .
in addition , cooling and star formation are also switched off in non - coloured regions .
the maximum resolution of our simulation is _ physical _ @xmath31 at all times .
in addition , our lagrangian volume excludes the most massive galaxy groups ( halo mass above @xmath32 at @xmath2 ) in order to speed up computation as we are only interested in smaller objects in this study . the two panels of figure [ fig : thefilament ] give a visual impression of the filament at @xmath2 .
we show two panels , the first illustrating the distribution of dark matter , the second is a composite image representing the density , pressure and metallicity distribution in the gas .
note that the main filament is entirely wrapped by an accretion shock .
it is thus clearly a high pressure environment .
the less massive filaments remain cold except at node points where the more massive haloes reside .
an inspection of the time - evolution of the simulation reveals that this accretion shock emerges first from the most massive haloes along the filament but then grows along the filament boundaries until the entire filament is enclosed by a shock . in the two panels of figure [ fig : thegalaxies ] , we show two examples of well - resolved disk - galaxies at @xmath2 in the simulation .
the images represent a simulated stellar light composite image combining the i , r and g bands into an rgb image taking into account dust reddening along the line of sight .
the images also show the orientation of the stellar and gaseous angular momentum and indicate that these are perpendicular to the disks and well aligned with one another . in order to identify galaxies in the simulated data ,
we directly use the amr mesh structure and label contiguous volumes exceeding a density threshold .
first , we pick the highest refinement level which exceeds @xmath33kpc resolution in order to smooth out galactic substructure .
next , we keep only those cells on this level , which exceed a threshold @xmath34 in baryonic overdensity .
this particular choice of @xmath35 has been found to be in excellent agreement with a visual identification of the galaxies . in a final step
we label all contiguous regions that are left .
this labelling is performed using a parallel implementation of the hoshen - kopelman cluster finding algorithm @xcite .
finally , all star particles and gas cells ( also on higher refinement levels ) are assigned to the mesh cluster @xmath36 , i.e. the `` galaxy '' , in which they reside .
galaxy stellar centres are then determined by iteratively computing the stellar centre of mass within shrinking spherical apertures and moving the aperture to the new centre in each iteration .
the procedure is stopped once less than 50 particles are contained within the aperture .
the centre of the gaseous disk is defined as the densest gas cell in the galaxy and the centre of the inner dark matter halo is defined as the most bound dark matter particle when evaluating the potential for all dark matter particles that are contained in the galaxy patch . in order to determine halo masses and virial radii , we grow spheres around each galaxy centre until the enclosed matter overdensity falls below the overdensity @xmath37 of a virialised spherical perturbation in a @xmath8cdm cosmology as given by the fitting formula of @xcite . in a next step ,
we determine equivalence classes of galaxies that are contained within the virial radii of each other . the galaxy with the highest stellar mass in each equivalence class we call the `` central galaxy '' , all other class members are `` satellite galaxies '' .
we do not consider these satellite galaxies further in this study .
finally , unresolved galaxies are identified by tracking the dark matter particles that end up in galaxies back to the initial conditions and requiring that none of them started in a region that was not initially marked for cooling and refinement .
these unresolved galaxies are then removed from the galaxy catalogue . the central galaxies in our samples have stellar radii @xmath38 ( which is the radius that contains 95 per cent of the stellar mass ) that are at most 0.11 at @xmath5 , 0.095 at @xmath39 and @xmath40 at @xmath2 of the virial radius of their parent halo .
this makes us confident that the identified galaxies are not contaminated by other satellites in the halo and thus constitute clean samples of central galaxies . for our analysis , we keep only the best resolved galaxies with at least 25000 star particles , corresponding to a minimum stellar mass of @xmath41 . for these galaxies , @xmath42 , the radius containing 80% of the stellar mass is resolved by at least 8 resolution elements .
furthermore , this requirement implies that we can compare our results to a lower resolution run ( see appendix [ sec : numerical_artefacts ] ) for which the minimum star particle number for galaxies of equal mass is then a factor of 8 lower , but still above 3000 .
we split the sample of well - resolved galaxies according to their host halo masses into 3 different halo groups .
the high - mass sample is defined as all halos above the non - linear mass @xmath43 .
prior results using pure @xmath15-body simulations have indicated that @xmath44 might be a characteristic scale for the spin - alignment behaviour of dark matter haloes ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the intermediate sample is for halo masses smaller than this non linear mass , but larger than the `` cold accretion limit '' of @xmath45 ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
haloes above this critical mass are believed to host a hot pressurized atmosphere that can exert ram pressure effects on the central galaxy .
the third mass sample is for halo masses smaller than the critical mass .
note that for the two higher redshifts @xmath5 and @xmath39 , we merge together the high and intermediate mass samples to improve statistical significance .
detailed information about the galaxy samples is given in table [ tab : halo_samples ] .
the distribution of stellar masses in the samples is shown in figure [ fig : massdistributions ] .
rccc & lm & mm & hm + + @xmath46 / @xmath47 & @xmath48 & @xmath49 & @xmath50 + median @xmath51 / @xmath47 & @xmath52 & @xmath53 & @xmath54 + number of galaxies & 70 & 39 & 12 + + @xmath46 / @xmath47 & @xmath48 & - & @xmath55 + median @xmath51 / @xmath47 & @xmath52 & - & @xmath56 + number of galaxies & 48 & - & 28 + + @xmath46 / @xmath47 & @xmath48 & - & @xmath55 + median @xmath51 / @xmath47 & @xmath52 & - & @xmath57 + number of galaxies & 58 & - & 31 + in this study , we are interested in the orientation of disk galaxies . since we exclude the most massive ( large group - scale and above ) haloes from refinement , indeed all galaxies considered are disky .
we can thus simply quantify the disk orientation by the angular momenta of the various disk constituents : gas , stars and dark matter .
we compute the angular momentum for each galaxy @xmath36 by evaluating the sum @xmath58 where @xmath59 is the stellar centre of the @xmath60th galaxy inner region ( see previous section ) , and @xmath61 is the centre of mass velocity . the sum is evaluated separately for all star particles that are part of an identified galaxy @xmath36 to yield the stellar angular momentum @xmath62 , as well as for all dm particles that are contained within a sphere of the radius of the most distant star particle in the galaxy @xmath63 .
finally , we also compute the gas angular momentum @xmath64 by summing over all leaf cells of the amr tree that are contained in the galaxy ( and are thus effectively bounded by an iso - density surface
see section [ sec : galfind ] for details ) . for gas
, @xmath59 refers to the densest cell in the galaxy patch and @xmath65 to the centre of mass velocity computed over all leaf cells in @xmath36 , while for dark matter @xmath59 refers to the most bound dark matter particle and @xmath61 to the centre of mass velocity of the dark matter particles in the sphere containing @xmath36 ( see also section [ sec : galfind ] ) . in order to determine the total halo angular momentum , we evaluate eq .
( [ eq : angmom ] ) summing over all components and up to the virial radius to obtain @xmath66 . in analogy to the approach followed by @xcite
, we use the eigenvectors of the tidal field tensor smoothed on some scale @xmath67 to probe alignments with the large - scale structure .
this approach has two main advantages : ( 1 ) as demonstrated in @xcite , the tidal field provides a natural dynamical quantification of the geometry of the large - scale structure ; and ( 2 ) any local alignments with the tidal field directly translate into shear - ellipticity alignments in weak lensing @xcite . in addition , when supplemented with auto - correlation functions of the non - linear tidal field ( see * ? ? ?
* ) , local alignments with the tidal field can be translated into correlation functions .
we will investigate this approach in future work .
we determine the tidal field by first constructing the full matter overdensity field @xmath68 of the entire simulation volume on a @xmath69 grid .
this grid is subsequently smoothed with a gaussian kernel of varying filtering radius @xmath67 . finally , we solve poisson s equation using the fast fourier transform ( fft ) to obtain the potential @xmath70 , evaluate the hessian matrix @xmath71 by finite differencing the potential and interpolating via cloud - in - cell to the galaxy positions .
diagonalising @xmath72 at each galaxy s position , we find the three eigenvalues @xmath73 and the corresponding unit eigenvectors @xmath74 as a function of the applied filtering scale @xmath67
. note that we define @xmath72 to be simply the hessian of the potential ,
i.e. the signs of the eigenvalues might be reversed with respect to some other studies .
the eigenstructure of the hessian of the gravitational potential is intimately connected to the large - scale dynamics of the cosmic web .
filamentary regions tend to have one negative and two positive eigenvalues when smoothed on large - enough scales . in particular ,
we find e.g. average values @xmath75 , @xmath76 and @xmath77 on scales of comoving @xmath78 where the averages are taken over the sample of our galaxies at @xmath39 .
units are such that @xmath79 , where @xmath80 is the matter overdensity .
the signature of these eigenvalues indicates that the galaxies are , on average , indeed embedded in a filamentary environment .
the signs of the eigenvalues reflect the large - scale expansion along the filament corresponding to the direction indicated by the eigenvector of the negative eigenvalue @xmath81 and gravitational contraction in the perpendicular plane corresponding to the plane spanned by the eigenvectors @xmath82 and @xmath83 .
the connection between the eigenstructure of @xmath72 and dynamics follows readily from the zeldovich theory of pancake collapse @xcite . in figure
[ fig : quiver_web ] , we show the orientations of the eigenvectors with respect to the cosmic web structure . to summarize , @xmath84 indicates the overall direction of a filament , while @xmath82 and @xmath83 span the perpendicular plane .
a more detailed account is given in @xcite .
.[tab : galaxycomponentalignment ] ( upper half ) median alignment angles between the gas ( g ) , stellar ( s ) and dark matter ( dm ) component of the galactic disk at redhifts @xmath2 , @xmath85 and @xmath86 .
( lower half ) median alignment angles between the three components of the galactic disk and the total angular momentum of the host halo .
errors are non - symmetric uncertainties on the median ( see text for more details ) .
values are given for the entire sample of well - resolved central galaxies since no mass dependence was found . [ cols="<,<,<,<",options="header " , ] in table [ tab : galaxycomponentalignment_lowres ] , we give the median angles between the stellar , gas and dark matter component within the galaxies and with the total halo angular momentum at @xmath2 for the low - resolution run . comparing these angles with those of the high - resolution run in table [ tab : galaxycomponentalignment ] , we find that the median alignment angles obtained at lower spatial resolution are marginally larger , the corresponding vectors thus less aligned . given the width of the distributions , the results can however be considered consistent at high significance . in figure
[ fig : projspins_lowres ] , we show the orientations of stellar angular momentum vectors for the low - resolution simulation .
this should be compared with the corresponding figure [ fig : projspins ] for the high - resolution run .
visually , we see a stronger clustering of spin vectors around the cartesian mesh basis vectors than for the high - resolution run .
spurious grid alignment is thus present at @xmath2 in both simulations to some degree . in figure
[ fig : spintide_lowres ] we show the scale dependence of the alignment between the galaxy stellar , gas and dark matter angular momentum and the three eigenvectors of the tidal field for the low - resolution run .
these results should be compared with results for the high - resolution simulation given in figure [ fig : spintide00439 ] .
we find that the results are perfectly consistent for all samples except the low - density low - mass bin where we find a slightly stronger anti - alignment with the intermediate principal axis of the tidal field in the low - resolution run .
also , the weak anti - alignment with the third principal axis is not seen in the high - resolution run .
it is likely that this difference for low - mass galaxies in low density regions is due to the anisotropy induced by the cartesian mesh which is expected to be stronger for the low - resolution run .
we conclude that the resolution of our high resolution run is still insufficient for galaxy orientations to be not affected by spurious grid relaxation in low density regions at low redshifts .
it is thus not clear whether the alignment of spins with the intermediate axis that we find at higher redshifts would persist in low - density environments also down to @xmath2 in the absence of numerical effects .
this question has to be addressed with even higher resolution simulations than the ones we used in our analysis .
the results of @xcite , who find a survival of alignment with the intermediate axis for the four isolated galaxies they simulated , is however indicative that such alignment should persist to some degree . | we use a @xmath0 resolution hydrodynamic amr simulation of a cosmic filament to investigate the orientations of a sample of @xmath1 well - resolved galactic disks spanning two orders of magnitude in both stellar and halo mass .
we find : ( i ) at @xmath2 , there is an almost perfect alignment at a median angle of @xmath3 , in the inner dark matter halo regions where the disks reside , between the spin vector of the gaseous and stellar galactic disks and that of their inner host haloes .
the alignment between galaxy spin and spin of the entire host halo is however significantly weaker , ranging from a median of @xmath4 at @xmath5 to @xmath6 at @xmath2 .
( ii ) the most massive galaxy disks have spins preferentially aligned so as to point along their host filaments .
( iii ) the spin of disks in lower - mass haloes shows , at redshifts above @xmath7 and in regions of low environmental density , a clear signature of alignment with the intermediate principal axis of the large - scale tidal field .
this behavior is consistent with predictions of linear tidal torque theory
. this alignment decreases with increasing environmental density , and vanishes in the highest density regions .
non - linear effects in the high density environments are plausibly responsible for establishing this density - alignment correlation .
we expect that our numerical results provide important insights for both understanding intrinsic alignment in weak lensing from the astrophysical perspective and formation and evolution processes of galactic disks in a cosmological context .
[ firstpage ] cosmology : theory , large - scale structure of universe galaxies : formation , evolution methods : numerical |
ngc4388 is an active galaxy situated near the core of the virgo cluster .
the spiral stellar disk ( sb(s)b pec ; phillips & malin 1982 ) is viewed nearly edge - on ( inclination @xmath6 ) with a dust lane across near the nucleus .
the distance of the galaxy is assumed to be 19.7 mpc ( sandage & tammann 1984 ) throughout this paper , giving an angular scale of 95.5 pc arcsec@xmath7 .
the blue and far infrared luminosities are @xmath8 and @xmath9 , assuming this distance ( david , jones & forman 1992 ) .
optical spectra of the nucleus show emission lines characteristic of a seyfert type 2 galaxy ( phillips & malin 1982 ; phillips , charles & baldwin 1983 ) .
an anisotropic ionization field is indicated by the cones of highly - ionized line emission extending above and below the stellar disk ( colina et al 1987 ; pogge 1988 ) .
gas emitting [ oiii]@xmath10 extends as far as 50 arcsec ( @xmath115kpc ) from the nucleus ( pogge 1988 ) .
the ionization state of the gas is best described as due to photoionization by a powerful nonthermal source ( pogge 1988 ; colina 1992 ) .
collimation of the nuclear radiation is strongly supported by the vla images ( stone , wilson & ward 1988 ; hummel & saikia 1991 ) .
the detection of a weak broad wing to h@xmath12 ( fwzi @xmath13 ) reported by filippenko & sargent ( 1985 ) suggests the presence of a seyfert 1 nucleus obscured from our direct view .
shields & filippenko ( 1988 , 1996 ) also find off - nuclear broad h@xmath12 emission ( fwzi @xmath144000 ) arising within the biconical high - ionization regions observed by pogge ( 1988 ) .
a plausible origin for these properties is the scattering of line emission from a hidden broad line region .
a strong hard x - ray source in ngc4388 has been detected by the birmingham x - ray coded mask telescope ( sl2 xrt : hanson et al 1990 ) , the _ ginga _ lac ( takano & koyama 1991 ) and sigma ( lebrun et al 1992 ) , supporting the hypothesis of a powerful hidden nucleus .
the results suggest the presence of a power - law x - ray source of photon index , @xmath15 , absorbed by a column density @xmath11 a few times @xmath16 . the absorption - corrected , 210 kev luminosity = @xmath2 , is typical for a low luminosity seyfert 1 galaxy .
the origin of weak soft x - ray emission seen from the source is , on the other hand , controversial the soft x - ray luminosity , @xmath17 , first detected by the _ einstein observatory _ ( forman et al 1979 ) , is @xmath111 per cent of the absorption - corrected luminosity of the hard x - ray source and consistent with the possibility that the soft x - rays are scattered nuclear radiation .
however , the _ rosat _ hri image of ngc4388 shows the 0.12.4 kev emission to be clearly extended with a radius of 45 arcsec and luminosity @xmath18(matt et al 1994 ; see also fig .
the radial profile shows that half of the soft x - ray emission originate beyond the central 15 arcsec .
matt et al ( 1994 ) claim that the large extent is evidence against a scattering origin for the soft x - ray emission , since the x - ray luminosity of the central source implied by the hard x - ray observations is insufficient to photoionize the necessary column density .
they propose as alternative explanations either thermal emission from a starburst or a collection of discrete sources .
the _ rosat _
pspc spectrum shows a steep spectral slope , @xmath1923 ( rush & malkan 1996 ) which may be consistent with this interpretation
. however , the spectral resolution of these instruments does not provide any conclusive evidence .
we report here the first _
asca _ results on ngc4388 ( see tanaka , inoue & holt 1994 for a brief discussion of the capabilities of _ asca _ ) .
the superior spectral resolution of _ asca _ over the 0.510 kev band enables us to identify several different components in the spectrum .
ngc4388 was observed in two different pointings at adjacent elliptical galaxies in the virgo cluster , m86 ( ngc4406 ; _ asca _ results of this galaxy appear in awaki et al 1994 and matsushita et al 1994 ) and m84 ( ngc4374 ) during the _ asca _ performance verification ( pv ) phase .
these observations were carried out in series between 1993 july 3 and 4 , each with an effective exposure time of about 20 ks .
ngc4388 was off axis by 18 and 10 arcmin , respectively .
in the first observation , ngc4388 was outside the field of view of the solid state imaging spectrometer ( sis ) detector so no sis data was obtained .
data reduction was made using the _ asca _ standard softwares , xselect and ftools .
one dataset of effective exposure time 17 ks for both sis ( s0 and s1 ) , and two datasets of each 20 ks were obtained for both gas imaging spectrometers ( gis : g2 and g3 ) . the full band ( 0.710 kev ) gis image from the first observation is shown in fig .
the two elliptical galaxies , m86 and m84 are present in the field of view as well as ngc4388 .
distortion of the x - ray image of ngc4388 is due to the off - axis response of the x - ray telescope ( xrt ) and is consistent with the point spread function at that position ( the elongation of the image aligning with the galaxy major axis is a coincidence ) .
the sis image obtained at a small off - axis position does not show evidence for significant extended emission within the spatial resolution of asca .
3 indicates that ngc4388 is the only strong source in the energy band above 3 kev but is faint in the softer band .
significant x - ray emission at the position of ngc4388 is , however , detected both in the gis and sis in the energy band below 3 kev as an excess above the diffuse emission spreading out from the m86 direction .
energy spectra were extracted from an elliptical - shaped region with the major axis aligned with the elongation of the image ( 3 @xmath20 1.6 arcmin for the sis ; 5 @xmath213 arcmin for the gis ) .
background subtraction must be done carefully , both because of the weakness of ngc4388 itself and the complex underlying diffuse emission in this busy field .
since ngc4388 lies close to the core of the virgo cluster there is significant emission from the intra - cluster medium ( icm ) .
the temperature of the icm , @xmath22 kev , has been measured with _ ginga _
( koyama , takano & tawara 1991 ) and _ rosat _ ( bhringer et al 1994 ) .
in addition , there may be some extended emission from m86 kev and abundance @xmath23for the total x - ray emission of m86 after fitting with a raymond - smith thermal spectrum model . ] which makes the background around ngc4388 complicated .
we have checked the background data by dividing it into four distinct regions .
the data taken from the far side ( sw of ngc4388 ) of m86 and m84 are weaker than in the other directions and show a different spectral shape .
this implies that the diffuse emission has a gradient in intensity and spectrum across the source - extracted region of ngc4388 .
we thus took background data from a circular , annulus of width 5 arcmin for the sis and a circular or an elliptical annulus ( depending on off - axis angle ) with a radius of 68 arcmin for the gis .
the background used here should be a good approximation of the true background , since taking data from the region surrounding the source should smear the gradient out .
the background - subtracted count rate in each detector is shown in table 1 .
the source fraction of the total counts in the source - extraction region for each detector is about 70 per cent for the sis data , 60 per cent for the first gis data , and 68 per cent for the second gis data .
we find no obvious change in observed flux during each observation .
any possible flux change is less than @xmath24 per cent during each @xmath25 ks long observation ( net exposure time is @xmath1120 ks in each ) .
differences of count rate between the two gis observations ( table 1 ) are owing to differences of effective area of the photon collecting region at the given off - axis positions .
the effective area obtained through the xrt and gis responses in the first observation is @xmath26 per cent of that in the second observation .
this means that there is no significant change in source flux between the two observations after the correction .
.the background - subtracted count rate in each detector .
there is no sis data in the first observation .
ngc4388 is observed at off - axis angle @xmath27 arcmin in the first observation , and @xmath28 arcmin in the second observation .
the effective area of the source - photon collected region of ngc4388 in the first observation is @xmath26 per cent of that in the second observation . [ cols="^,^,^,^,^ " , ] we detected strong absorption @xmath29 for the hard x - ray component , with a relatively small uncertainty compared with previous measurement ( see table 2 ) .
the constraint on the photon index is weak owing to the limited energy range and strong absorption .
the best - fit @xmath30 is consistent with previous results ( hanson et al 1990 ; takano & koyama 1991 ) .
a strong fluorescence iron k@xmath12 line is found for the first time in this object .
the equivalent width ( ew ) is quite large despite the relatively narrow line width .
there is no evidence for the line broadening in the gis data whereas there is weak evidence ( @xmath31 kev ) in the sis data .
when we assume that the pl@xmath32 has an identical @xmath33 to the pl@xmath34 , some extra absorption , @xmath35 , is required ( see table 2 ) .
this means that the soft x - ray spectrum around 2 kev is extremely flat .
the observed fluxes are summarized in table 3 .
the flux in the 310 kev band is slightly smaller than that observed by the sl2 xrt ( @xmath36 ; hanson et al 1990 ) .
the soft x - ray flux is roughly consistent with previous measurements from the _ rosat _ pspc ( @xmath37 , rush & malkan 1996 ) and the _ rosat _ hri ( @xmath38 , matt et al 1994 ) , extrapolating our best - fit model down to 0.1 kev , although the _ rosat _ flux depends on the assumed model .
lcc @xmath39 & sis & gis + + total & 1.6 & 1.4 + + total & 6.4 & 6.5 +
r - s & 2.9 & 4.4 + pl@xmath32 & 3.5 & 2.1 + + total & 1.6 & 1.4 + pl@xmath34 & 1.3 & 1.1 + pl@xmath32 & 0.11 & 0.15 +
we find clear evidence for strong obscuration along the line of sight .
the column density @xmath29 implied from the _ asca _ spectrum is similar to those observed in some other classical seyfert 2 galaxies ( e.g. , awaki et al 1991 ; iwasawa et al 1994 ) . the photon index ( @xmath30 )
is also consistent with that of seyfert galaxies .
our best - fit value can be justified by the fact that a _ sigma _ observation in the 40150 kev band ( lebrun et al 1992 ) is in agreement with an extrapolation of the result of hanson et al ( 1990 ) , unless any extra component , e.g. compton reflection component , has a significant contribution at the hard x - ray band ( the reflection contribution should be small if the inner disk is highly inclined to our line of sight ) .
the absorption - corrected 210 kev flux of the hard power - law component is @xmath40 , corresponding to a luminosity , @xmath41 .
this is within the range of seyfert 1 nuclei .
colina ( 1992 ) predicted that a uv luminosity , @xmath42 , is required to explain the high ionization extended nebula .
if ngc4388 has the ratio , @xmath43 , generally appropriate for active galaxies , then the intrinsic x - ray luminosity of the central source is reasonable .
this agreement may imply that the radiation of the central continuum is not significantly anisotropic , because the x - ray measurement is for line - of - sight radiation while the calculation of the uv luminosity is made for the ionized nebulae extending perpendicular to the line - of - sight .
the implied heavy obscuration is opaque to the optical / uv continuum from the central source , unless the dust to gas ratio is unusually low . even in the near - infrared band , hei@xmath44
m and pa@xmath45 m do not show any broad component ( ruiz , rieke & schmidt 1994 ) .
the displacement between the optical nucleus and the radio/10@xmath46 m peak ( stone et al 1988 ) is likely due to this obscuration effect , and the true nucleus should be at the radio/10@xmath46 m nucleus position .
even though deficits of hi and co gas in this galaxy have been reported ( chamaraux , balkowski , & gerard 1980 ; kenney & young 1986 ) , plenty of cold gas remains close to the central x - ray source .
this can form an obscuring torus such as proposed in the unification scheme of seyfert galaxies ( e.g. , antonucci 1993 ) .
the anisotropic radio source extending perpendicular to the galaxy ( stone et al 1988 ; hummel & saikia 1991 ) and the conical high - ionization regions observed in the optical band ( pogge 1988 ; colina et al 1987 ) are compatible with the idea of collimation of radiation by the torus .
we note that the symmetrical axes of the torus and the stellar disk of ngc4388 are aligned as well as ngc4945 ( iwasawa et al 1993 ) , while they are misaligned in most cases ( e.g. , ngc1068 , antonucci & miller 1985 ; ngc5252 , tadhunter & tsvetanov 1989 ) .
an alternative source of the strong x - ray absorption is the interstellar medium in the nearly edge - on galaxy disk itself .
recent near - infrared imaging has revealed that this galaxy has a boxy bulge ( mcleod & rieke 1995 ) .
this and the dust lane crossing in front of the nucleus can be responsible for some or all of the absorption . in this case , the flat spectrum between 1 and 3 kev is possibly explained by transmission of scattered radiation from the central source seen through the stellar disk . since the disk is tilted by @xmath11 18 degrees from being exactly edge - on , there may be less absorption along these lines of sight , so making a flat spectrum in the softer band in addition to the strongly absorbed component .
we note there will be some scattering in the obscuring medium with @xmath47 leading to a hard power - law component in the 13 kev band ( yaqoob 1996 ) .
this , however , is at least an order of magnitude fainter than the observed power - law in that band , unless the absorbing medium is partially ionized .
a strong iron k@xmath12 emission line is detected at 6.4 kev .
the large equivalent width ( @xmath48 ev ) of the 6.4 kev line is one of the signatures of reprocessing in cold matter .
various calculations of x - ray spectra emerging from a central power - law source surrounded by a torus do predict a large ew ( awaki et al 1991 ; krolik , madau & ycki 1994 ; ghisellini , haardt & matt 1994 ) .
however , the observed ew is larger than the value ( @xmath49 300 ev ) expected from the work of awaki et al ( 1991 ) with @xmath47 , taking an opening angle of the torus of about 1/3 , as deduced from the optical extended nebula ( pogge 1988 ) .
the excess ew could be due to an enhanced iron abundance .
although the present data can not provide any meaningful restriction on the depth of any iron k absorption edge at 7.1 kev , and hence on the iron abundance , a deep edge due to a supersolar abundance of iron could partly explain the flat hard x - ray spectra , found in some _ ginga _ seyfert 2 spectra ( awaki et al 1991 ; smith & done 1996 ) , especially when the column density exceeds @xmath50 .
a similar iron k line feature has been found in _
spectra of another seyfert 2 galaxy mrk 3 ( iwasawa et al 1994 ) .
mrk 3 showed a significant intensity change of the iron k line in response to the continuum variation between two observations of 3.6 yr apart .
this fact suggests that the line is produced within @xmath111 pc from the central source .
we do not of course detect any variability within the present one - day observing run of ngc4388 and therefore can not obtain any constraint about the line emitting region from this observation .
there is a flat spectral component detected in the intermediate energy range between the hard component and the thermal component .
such a flat component has been observed in _ asca _ spectra of seyfert 2 galaxies ngc1068 ( ueno et al 1994 ) and ngc6552 ( fukazawa et al 1994 ) , and the eclipse spectrum of x - ray binary vela x-1 ( nagase et al 1994 ) .
these are considered to be scattered continuum of the central source .
it is therefore plausible that this component in ngc4388 is also scattered radiation from the central obscured source , although the possibility that it is radiation from the central source escaping through a partially covering absorber as mentioned in section 3.1.2 .
is also viable . as the broad h@xmath12 detected from the off - nuclear region ( shields & filippenko 1988 ; 1996 ) suggests , scattered light
may be observed in the optical band .
however , spectropolarimetry in the optical blue band does not detect any such broad - line region ( kay 1994 ) .
one can expect that several emission lines produced in the photoionized gas or cold gas which is responsible for the scattering whould be observable on the x - ray continuum , as seen in the _ asca _ spectra of mrk 3 ( iwasawa et al 1994 ) and ngc6552 ( fukazawa et al 1994 ; reynolds et al 1994 ) .
it is , however , hard to detect such lines in our data , because of poorer statistics and contamination by the thermal emission .
the detected absorption column density of a few times @xmath51 imposed on the power - law is consistent with the narrow - line region reddening e(b
v)@xmath52 mag deduced from the [ sii ] doublet ( malkan 1983 ) , although a recent estimate of the reddening from pa@xmath53/h@xmath54 gives a greater value , e(b
v)@xmath55 mag than this ( ruiz et al 1994 ) .
we do however detect a marginally significant iron line due to hydrogenic iron in the sis spectrum at @xmath56 kev of equivalent width @xmath57 ev against the absorbed hard continuum or @xmath58 kev against an extrapolation of the scatttered continuum .
the strength of this line is difficult to predict as it is probably due to resonant scattering of the continuum ( matt , brandt & fabian 1996 ) , but it does corroborate the scattering hypothesis . to be so highly ionized , the scattering medium is probably within a pc of the nucleus for the observed x - ray luminosity of the central source .
we note that a reasonable fit to the whole sis spectrum can be obtained with a partially - covered power - law spectrum , with a gaussian line at 6.4 kev and an edge at 0.9 kev .
this last feature could be due to a oxygen warm absorber , such as is commonly observed in the spectra of seyfert 1 galaxies , along the line of sight to the scattering medium .
this model is rejected because the _ rosat _ hri image shows that the scattering medium is much more extended than can be plausibly ionized by the central source ( matt et al 1994 ) .
other explanations for the flat spectrum are transmission through the edge - on stellar disk , as discussed in the previous section , and a population of high mass x - ray binaries .
this last possibility is unlikely given the expected massive star population in the galaxy .
the ratio of far infrared luminosity to blue luminosity can be an indicator of relative excess of young population ( o stars and high mass x - ray binaries ; david et al 1992 ) .
since this ratio in ngc4388 , @xmath59 , is a factor of 10 lower than in m82 , it is unlikely that an intensive starburst like that in m82 is taking place in ngc4388 . a collection of x - ray binaries is expected to emit a much smaller x - ray luminosity than that observed ( @xmath60 ) . the _ asca _ spectrum , combined with the extended image ( fig .
1 ) resolved by the _ rosat _
hri ( matt et al 1994 ) , provides evidence that the soft x - ray emission below 1kev is of thermal origin . such thermal emission has been observed in the _ asca _ spectra of other nearby spiral galaxies which contain an obscured seyfert nucleus ( e.g. , ngc1068 , ueno et al 1994 ; ngc4258 , makishima et al 1994 ) .
the temperature is @xmath61 kev and the abundance is extremely low ( @xmath62 ) .
the 0.53 kev luminosity of this thermal emission is = @xmath63 , consistent with the _ einstein _ ipc measurement ( forman et al 1979 ) .
our spectral fits shows most of the observed x - rays below 1.5 kev come from the thermal component ( see fig . 4 and fig . 5 ) .
this is compatible with the fact that most of the x - ray emission detected by the _
hri is extended over the nucleus ( matt et al 1994 ) , taking account for the _ rosat _ bandpass .
the extended thermal gas could be the stripped interstellar medium of the galaxy .
starburst activity is not strong in ngc4388 , as discussed in section 4.3 , but consistent with that in normal galaxies with respect to the @xmath64 correlation ( david et al 1992 ) .
ngc4388 is known to be located close to the virgo cluster core , and to move at high velocity of 1311 relative to the cluster ( corbin et al 1988 ) in the deep cluster potential the deficiency of hi and co in this galaxy has been explained in terms of ram - pressure stripping by interaction with the virgo icm ( chamaraux et al 1980 ; giovanelli & haynes 1983 ; kenney & young 1986 ) .
this stripping mechanism may also explain the x - ray plume of the nearby elliptical galaxy m86 ( forman et al 1979 ; white et al 1991 ; rangarajan et al 1995 ) which also has similar high radial velocity relative to the cluster ( they may be part of an infalling group ) .
the lower temperature and lower abundance in ngc4388 are different from the x - ray plume of m86 , however .
high quality soft x - ray imaging , such as will be available with _ axaf _ , are needed to settle this issue .
finally , we note that at the radius of the thermal component ( 1.44.5 kpc , matt et al 1994 ) the gas density is only @xmath65 @xmath66 , assuming bremsstrahlung emission since the abundance is so low , and any scattered x - ray emission will be negligible .
the bremsstrahlung emission from any scattering medium at a temperature @xmath67 k would swamp the scattered flux .
the radiative cooling time of the gas is @xmath68 yr and the recombination time ( of say highly ionized oxygen ) is @xmath69 yr . the flow time at a velocity of @xmath70 is @xmath71 yr . the gas may therefore be due to an outburst in the last ten million yr , or a more general wind from the nuclear regions .
the mass flow rate is then about _
m_@xmath72yr@xmath7 where @xmath73 is a wind velocity in unit of 300 and @xmath74 is the radius of the thermal emission region in units of 2 kpc , which is at the highest end of the values for superwind galaxies ( _ m_@xmath111100yr@xmath7 , heckman , armus & miley 1991 ) .
this large mass injection is unlikely for ngc4388 since it shows no evidence for a strong starburst such as in superwind galaxies .
the low measured abundance ( though error is rather large ) is also puzzling .
it is however measured mainly for iron - l in our spectrum .
as the abundance of iron is usually lower than that of other elements in starburst galaxies , the low abundance may not be unusual if the thermal emission is produced through a recent starburst .
we have identified several problems in the interpretation of the x - ray data of ngc4388 : a ) the large soft x - ray extent , if thermal , means a large mass loss rate from the bulge which can only plausibly last a few million yr , b ) a thermal model requires a very low abundance in the extended gas and c ) the iron line is high for the observed obscuration ( and in the case of an iron line from reflection assuming that the solid angle subtended by this and thicker gas at the nucleus is from a toroidal geometry aligned with the disk of the galaxy ) .
these unusual propoerties can be solved if we assume that the nucleus has not always had its present luminosity , @xmath75 , but had a higher luminosity , @xmath76 a thousand years or more ago when gas at the radius of the soft x - ray extent ( @xmath77 kpc ) was being irradiated .
the warm absorber solution in section 4.3 requires that the ionization parameter , @xmath78 where the density of gas at radius @xmath79 is @xmath80 , is @xmath81 .
then @xmath82 where @xmath83 is the fraction of the volume within @xmath79 being irradiated and @xmath84 is the thomson cross - section .
we then find @xmath85 , the thomson depth of the region is then @xmath86 and the density @xmath87 .
provided that @xmath83 is not small the luminosity is plausible for a central engine of more than @xmath88 .
it is possible that the optical line emission only shows up one side of a very wide cone so @xmath89 . with this solution
, all of the x - ray emission below 3 kev can be due to scattering by partially - ionized gas within a few kpc of the nucleus .
the feature in the spectrum around 1 kev is then due to absorption by oviii .
the oviii absorption edge would be produced in a region close to the nucleus as found in mcg6 - 30 - 15 ( otani et al 1996 ) .
the presence of dust in the ionized gas implied from the optical / near infrared extinction is not peculiar but similar to the dusty warm absorber observed in iras13349 + 2438 ( brandt , fabian & pounds 1996 ) as well as mcg6 - 30 - 15 .
the nucleus must have been 100 or more times more luminous in the past .
enhanced iron k - line emission could be produced by gas in the bulge and disk of the galaxy ( see fabian 1977 ) .
a continuous decrease in x - ray luminosity over a long time scale has been observed in some seyfert galaxies such as ngc 2992 ( weaver et al 1996 ) , mrk 3 ( iwasawa et al 1994 ) and ngc 1275 ( a. edge , private communication ) .
the seyfert 1.9 galaxy ngc2992 , for example , has declined in luminosity by more than a factor of 20 over the last 17 yr . the luminosity variation is similar to that found for those seyfert galaxies , except that in ngc4388 we require a larger variation over a longer timescale .
as discussed for mrk 3 ( iwasawa et al 1994 ) and ngc 2992 ( weaver et al 1996 ) , a reverberation effect makes the ew of the iron k line large if the x - ray luminosity decreases continuously .
the main problem is the lack of observed optical polarization
. this could however be low due to the small extraction window used by kay ( 1994 ) , which mainly sampled the region obscured by the galaxy disk , and to a wide opening angle for the emerging radiation , which reduces the predicted level of polarization .
further spectropolarimetric observations of a wider region are required .
we find from the _ asca _ observations that the spectrum of the seyfert 2 galaxy ngc4388 has many components
. the hard x - ray emission is strongly absorbed , and a strong iron k line ( @xmath90 440730 ev ) is found at 6.4 kev .
the hard continuum has a photon index @xmath30 and is absorbed by a column density @xmath29 .
evidence for a thermal spectrum is found in the soft x - ray band , which can be identified with the spatially extended emission observed by _
hri ( matt et al 1994 ) .
a flat spectrum joins these two components in the band around 2 kev
. it may be the scattered continuum of the obscured central source .
however , several difficulties still remain ; ( 1 ) an extremely low abundance is required for the thermal emission model ; ( 2 ) an unlikely high mass flow rate is required to explain the soft x - ray extent ; and ( 3 ) the iron - k line stronger than expected from the torus model , with the observed column density .
these can be overcome if the central source was a hundred times brighter a thousand or more years ago .
all the soft x - ray emission below 3 kev can then be due to the continuum scattered in a partially ionized medium irradiated by the luminous source .
the iron k line produced by gas in the bulge and disk in the galaxy then remains strong , because of the larger size .
this will be tested by a detailed imaging spectroscopy with future missions like _ axaf _ and _ astro - e_.
we thank all the members of the _ asca _ pv team , giorgio matt and david white for useful discussion and niel brandt for helping to make the _ rosat _ hri image .
the optical image of ngc4388 was obtained through the _ sky view _
facility operated by heasarc at nasa / gsfc .
this research has made use of data obtained through the high energy astrophysics science archive research center online service , provided by the nasa - goddard space flight center .
acf and ki thank the royal society and pparc , respectively for support . | we present results on the seyfert 2 galaxy ngc4388 in the virgo cluster observed with _ asca _ during its performance verification ( pv ) phase .
the 0.510 kev x - ray spectrum consists of multiple components ; ( 1 ) a continuum component heavily absorbed by a column density @xmath0 above 3 kev ; ( 2 ) a strong 6.4 kev line ( equivalent width @xmath1 ev ) ; ( 3 ) a weak flat continuum between 1 and 3 kev ; and ( 4 ) excess soft x - ray emission below 1 kev .
the detection of strong absorption for the hard x - ray component is firm evidence for an obscured active nucleus in this seyfert 2 galaxy .
the absorption corrected x - ray luminosity is about @xmath2 .
this is the first time that the fluorescent iron - k line has been detected in this object ; the large ew is a common property of classical seyfert 2 nuclei .
the flat spectrum in the intermediate energy range may be a scattered continuum from the central source
. the soft x - ray emission below 1 kev can be thermal emission from a temperature @xmath3 kev , consistent with the spatially extended emission observed by _ rosat _
hri .
however , the low abundance ( @xmath4 ) and high mass flow rate required for the thermal model and an iron - k line stronger than expected from the obscuring torus model are puzzling .
an alternative consistent solution can be obtained if the central source was a hundred times more luminous over than a thousand years ago .
all the x - ray emission below 3 kev is then scattered radiation . 0_h_@xmath5 0_q_@xmath5 |
lunar - based astronomical observations have advantages over both ground - based and space - based observations .
1 ) it has been known for decades that the moon is surrounded by an extremely tenuous atmosphere ( @xmath2 at night , and a much higher value @xmath3 in the daytime , see heiken et al .
1991 ; stern 1999 for a review ) .
this means that atmospheric opacity , atmospheric scattering / emission and , the atmospheric turbulence are absent .
2 ) unlike space - based observations , the moon provides a large stable platform for maintaining astronomical instruments in permanently stable configurations .
3 ) the sky s diurnal motion on the moon is @xmath4 , 27 times slower than on the earth .
this allows long term monitoring for 13 days without interruption .
4 ) the temperature in the permanent shadow regions ( pdrs ) at both poles of the moon could permanently be as low as 30k .
pdrs are therefore ideal places for infrared observations .
5 ) observations at very low frequency ( @xmath510mhz , vlf ) are feasible on the moon , but not on earth , because vlf electromagnetic waves can not penetrate the earth s ionosphere . the apollo-16 mission performed far - ultraviolet observations on the lunar surface in 1972 for the first time in history ( page & carruthers 1977 ; carruthers & page 1977 ; carruthers & page 1972 ) .
the observations were carried out with a 3-inch schmidt telescope equipped with a far - ultraviolet camera / spectrograph operating in the wavelength range from 100 - 160 nm . the field - of - view was about 20 , and the limiting magnitude was 11 .
the telescope was placed in the shadow of the lunar module to avoid heating by the sun . in total ,
178 images were obtained by the telescope , and delivered to the earth by the astronauts .
in addition to the earth s geocorona , the observed objects included star clusters and the magellanic clouds .
these observations , however , contributed little to astronomical knowledge because of the low level of the technology used at that time .
the chinese change - iii mission is designed to deploy a dedicated , robotic lunar - based ultraviolet telescope ( lut , cao et al .
2011 ) on the surface of the moon .
lut will perform astronomical observations in near - ultraviolet ( nuv ) band .
the main scientific goal of lut is to monitor variable stars and active galaxies in the nuv band for more than a dozen days
. objects with large brightness variations in the nuv band include cataclysmic variable stars , large/ small mass binaries and novae , quasars and blazars , dme stars , and lyr rr stars .
the variations of the stellar temperature , radius , and accretion rate of these compact objects can be studied , useful to verify current stellar atmosphere models and to investigate the origin of the instability .
lut will additionally perform low - galactic - latitude sky surveys as a complement to the nasa galex ( galaxy evolution explorer ) mission ( martin et al . 2005 ) whose onboard mcps prevent from observing the bright objects with nuv@xmath510mag ( morrissey et al . 2005 ) . at present , lunar - based observations are only feasible in the moon daytime mainly due to the lack of electronic power supply at night .
the indian chandrayaan-1 payload chandra s altitude composition explore ( chace ) recently claimed a lunar exosphere pressure of @xmath6torr ( sridharan et al .
2010a , b ) , higher than that found by the apollo missions by 2 orders of magnitude ( see section 3 for more details ) .
because of these new results , this paper wishes to estimate the effect of this potentially enhanced lunar exosphere on the signal - detection performance of lut .
the sunlight can be scattered and/or re - radiated by the particles of the lunar exosphere .
these processes mainly include rayleigh scattering , resonant scattering , and emissive photodissociation .
the paper is organized as follows . after describing the basic properties of lut in section 2 , section 3
briefly summarizes previous measurements of the lunar exosphere .
the sky background brightness is calculated for the lunar exosphere in section 4 . the estimated brightness
is then compared with the brightness of a point source with ab=13 mag .
section 5 calculates the count rates at the lut focal plane , and estimates the final signal - to - noise ratio ( s / n ) for an observation of a point source .
as part of the scientific payload of the change - iii mission , lut will land on the moon and work in the lunar daytime ( cao et al .
a limiting magnitude of ab=13 mag is designed for an exposure time of 30 seconds .
the optical system of lut is a f/3.75 ritchey - chrtien telescope with an aperture of 150 mm .
a pointing flat mirror mounted on a two - dimensional gimbal is used to point and track a given object .
an ultraviolet - enhanced ccd e2v47 - 20 , manufactured by the eev company , is chosen as the detector mounted at the nasmyth focus .
table 1 tabulates the designed performance parameters on which our subsequent calculations are based .
we refer the readers to cao et al .
( 2011 ) for more details on the mission s concept and design .
.the designed performance parameters of lut [ cols="<,^,^,^",options="header " , ] [ table1 ] [ figure1 ]
we estimate the near - ultraviolet sky background brightness of the lunar exosphere .
our calculations show that the sky brightness is @xmath0 within the wavelength range 245 - 340 nm .
the signal - to - noise analysis indicates that the detection performance of lut can be degraded by the sky background emission in most cases .
an ab=13 mag point source can be detected by lut at a signal - to - noise ratio above 8 when the oh concentration is less than @xmath1 .
however , the effect on the performance is not ignorable when the exosphere is as dense as suggested by chace .
the authors thank the anonymous referee for his / her comments that improve the quality of the paper .
special thanks go to paul wesselius , the scientific editor , for his / her great effort in improving the manuscript .
the study is funded by the national science and technology major project .
jw is supported by the national science foundation of china ( under grant 10803008 ) .
lc is supported by the national science foundation of china ( grant nos .
10978020 and 10878019 ) .
jw , jsd , ylq and jyw are supported by the national basic research program of china ( grant 2009cb824800 ) .
the authors would like to thank f. tian , h. zhao and prof . j. y. hu for valuable discussions and suggestions .
we would like to thank james wicker for help with the language .
arnold , j. r. ice in the lunar polar regions .
jgr 84 , 5659 - 5668 , 1979 behara , n. t. , jeffery , c. s. lte model atmospheres with new opacities . i. methods and general properties .
a&a 451 , 643 - 650 , 2005 bertaux , j. l. the uv bright spot of water vapor in comets .
a&a 160 , l7-l10 , 1986 cao , l. , ruan , p. , cai , h. b. , et al .
lut : a lunar - based ultraviolet telescope .
scchg 54 , 558 - 562 , 2011 carruthers , g. r. , page , t. apollo 16 far - ultraviolet camera / spectrograph : earth observations .
science 177 , 788 - 791 , 1972 carruthers , g. r. , page , t. apollo-16 far - ultraviolet spectra in the large magellanic cloud .
apj 211 , 728 - 736 , 1977 chamberlain , j. w. , hunten , d. m. theory of planetary atmosphere , academic press , newyork , 1987 clark , r. n. detection of adsorbed water and hydroxyl on the moon .
science 326 , 562 - 564 , 2009 crider , d. h. , vondrak , r. r. hydrogen migration to the lunar poles by solar wind bombardment of the moon .
adspr 30 , 1869 - 1874 , 2002 crovisier , j. the photodissociation of water in cometary atmospheres .
a&a 213 , 459 - 464 , 1989 dalgarno , a. , williams , d. a. rayleigh scattering by molecular hydrogen .
apj 136 , 690 - 692 , 1962 dyar , m. d. , hibbitts , c. a. , orlando , t. m. et al .
mechanisms for incorporation of hydrogen in and on terrestrial planetary surfaces .
icar 208 , 425 - 437 , 2010 feldman , p. d. , mccandliss , s. r. , morgenthaler , j. p. , et al .
galaxy evolution explorer observations of cs and oh emission in comet 9p / tempel 1 during deep impact .
apj 711 , 1051 - 1056 , 2010 fukugita , m. , ichikawa , t. , gunn , j. e. , et al .
the sloan digital sky survey photometric system .
aj 111 , 1748 - 1756 , 1996 gullixson , c. a. two dimensional imagery .
aspc 23 , 130 - 159 , 1992 heiken , g. vaniman , d. , french , b. m. lunar sourcebook : a user s guide to the moon i. cambridge university press , 1991 hibbitts , c. a. , dyar , m. d. , orlando , t. m. , et al
. thermal stability of water and hydroxyl on airless bodies .
lpi 41 , 2417 - 2418 , 2010 hodges , jr .
r. r. helium and hydrogen in the lunar atmosphere .
jgr 78(34 ) , 8055 - 8064 , 1973 hodges , jr .
r. r. , hoffman , j. h. , yeh , t. t. j. , et al .
orbital search for lunar volcanism .
jgr 73(23 ) , 7307 - 7317 , 1972 hunten , d. m. , sprague , a. l. origin and character of the lunar and mercurian atmospheres .
adspr 19 , 1551 - 1560 , 1997 johnson , r. e. , baragiola , r. lunar surface - sputtering and secondary ion mass spectrometry .
georl 18 , 2169 - 2172 , 1991 martin , d. , et al . the galaxy evolution explorer : a space ultraviolet survey mission , apjl 619 , 1 - 6 , 2005 page , t. l. , carruthers , g. r. apollo 16 far - ultraviolet imagery and spectra of the large magellanic cloud .
1977 , space research xvii ; proceedings of the open meetings of working groups on physical sciences , june 8 - 19 , 1976 and symposium on minor constituents and excited species , philadelphia , pa . , june 9 , 10 , 1976 .
( a78 - 18101 05 - 42 ) oxford and new york , pergamon press , 749 - 755 , 1977 killen , r.m . ,
ip , w .- h .
the surface - bounded atmospheres of mercury and the moon .
geophys 37(3 ) , 361 - 406 .
, 1999 killen , r. m. , potter , a. , fitzsimmons , a. , et al .
sodium d2 line profiles : clues to the temperature structure of mercury s exosphere .
p&ss 47 , 1449 - 1458 , 1999 killen , r. , shemansky , d. , mouawad , n. expected emission from mercury exospheric species , and their ultraviolet - visible signatures .
apjs 181 , 351 - 359 , 2009 merline , w. j. , howell , s. b. a realistic model for point - sources imaged on array detectors : the model and initial results .
exa 6 , 163 - 210 , 1995 morgan , t. h. , killen , r. m. c. a non - stoichiometric model of the composition of the atmospheres of mercury and the moon .
p&ss 45 , 81 - 94 , 1997 morgan , t. h. , shemansky , d. e. limits to the lunar atmosphere .
jgr 96 , 1351 - 1367 , 1991 morrissey , p. , et al .
the on - orbit performance of the galaxy evolution explorer , apjl 619 , 7 - 10 , 2005 mortara , l. , fowler , a. evaluations of charge - coupled device ccd performance for astronomical use .
spie 290 , 28 - 33 , 1981 pieters , c.m .
, et al . character and spatial distribution of @xmath7 on the surface of the moon seen by m3 on chandrayaan-1 .
science 326 , 568 - 572 , 2009 schleicher , d. g. , ahearn , m. f. the fluorescence of cometary oh . apj 331 , 1058 - 1077 , 1988 sneep , m. , ubachs , w. direct measurement of the rayleigh scattering cross section in various gases .
jqsrt 92 , 293 - 310 , 2005 sridharan , r. , ahmed , s. m. , pratim .
d. , et al .
direct evidence for water ( @xmath8 ) in the sunlit lunar ambience from chace on mip of chandrayaan i. p&ss 58 , 947 - 950 , 2010 sridharan , r , , ahmed , s. m. , pratim .
d. , et al . the sunlit lunar atmosphere : a comprehensive study by chace on the moon impact probe of chandrayaan-1 .
p&ss 58 , 1567 - 1577 , 2010 starukhina , l.v . ,
shkuratov , y.g .
note : the lunar poles : water ice or chemically trapped hydrogen ?
icar 147 , 585 - 587 , 2000 starukhina , l.v . ,
shkuratov , y.g .
simulation of 3-m absorption band in lunar spectra : water or solar wind induced hydroxyl ?
lpi 41 , 1385 - 1386 , 2010 stern , s. a. the lunar atmosphere : history , status , current problems , and context .
rvgeo 37 , 453 - 492 , 1999 stern , s. a. , parker , j. w. , morgan , t. h. , et al . note : an hst search for magnesium in the lunar atmosphere .
icarus 127 , 523 - 526 , 1997 stubbs , t. j. , glenar , d. a. , colaprete , a. , et al .
optical scattering processes observed at the moon : predictions for the ladee ultraviolet spectrometer .
p&ss 58 , 830 - 837 , 2010 sunshine , j .
m. , farnham , t. l. , feaga , l. m. , et al .
temporal and spatial variability of lunar hydration as observed by the deep impact spacecraft .
science 326 , 565 - 568 , 2009 tarafdar , s. p. , vardya , m. s. the importance of molecular rayleigh scattering in the atmospheres of very late type stars .
mnras 163 , 261 - 277 , 1973 wurz , p. , rohner , u. , whitby , j. a. , et al .
the lunar exosphere : the sputtering contribution .
icarus 191 , 486 - 496 , 2007 | the potential effect of the lunar exosphere on the near - ultraviolet sky background emission is predicted for lunar - based ultraviolet telescope ( lut : a funded chinese scientific payload for the change - iii mission ) . using the upper limit on the oh concentration inferred from the recent mip chace results ,
our calculations show that the sky brightness due to the illuminated exosphere is @xmath0 within the wavelength range 245 - 340 nm . by evaluating the signal - to - noise ratios of observations of an ab=13 mag point source at a series of sky background levels
, our analysis indicates that the detection performance of lut can be moderately degraded by the lunar exosphere emission in most cases .
an ab=13 mag point source can still be detected by the telescope at a signal - to - noise ratio more than 8 when the oh concentration is less than @xmath1 .
however , the effect on the performance is considerable when the exosphere is as dense as suggested by chace .
lunar exosphere , sky background , rayleigh scattering , resonance fluorescence , emissive photodissociation , lunar - based ultraviolet telescope = 0.5 cm |
magnetic reconnection ( e.g. , * ? ? ? * and references therein ) is of interest in diverse areas of astrophysics , yet its mechanics remains incompletely understood .
a tearing instability is thought to be necessary to initiate reconnection in reversing magnetic field configurations , but many astrophysical plasmas have collisional resistivities that are insufficient to facilitate its growth .
interpretations of space plasma measurements ( e.g. , * ? ? ?
* ; * ? ? ?
* ) and astronomical observations , however , suggest that efficient collisionless reconnection is ubiquitous
. therefore , collisionless effects , which operate on plasma kinetic scales , seem to be required to provide the dissipation necessary for effecting a change of magnetic topology .
magnetohydrodynamic ( mhd ) models for the dynamics of systems undergoing magnetic reconnection have been available for a long time ( e.g. , * ? ? ? * and references therein ) .
unfortunately , these models do not describe the underlying nature of ( possibly multiscale ) plasma organization in the reconnection layer where magnetic energy is being dissipated and the assumptions of ideal mhd do not apply . understanding the detailed plasma organization on all length scales , from the likely relatively small , plasma kinetic scales , to the potentially much larger scales on which astrophysical dynamical systems `` prepare '' reconnection sites , and where ideal mhd may be valid , is paramount for completing the theories of a wide variety of astrophysical phenomena and for interpreting space plasma measurements and astronomical observations . in an effort to develop a picture of magnetic reconnection from first principles ,
recent particle - in - cell ( pic ) simulations have examined the dynamics of reconnection layers that start with a current sheet as thin as the plasma skin depth .
pic simulations of reconnection in pair plasmas have been carried out in two spatial dimensions @xcite and three dimensions @xcite . among these
, several @xcite have investigated the role of departure from the idealized , exactly antiparallel reconnection by introducing a perpendicular `` guide '' field .
these various simulations have revealed novel forms of small - scale plasma self - organization that are interesting in their own right , but that must ultimately be related to and embedded within the appropriate larger astrophysical contexts ( e.g. , * ? ? ?
while spacecraft measurements , which can be done in situ , can provide direct clues how to establish this embedding in space plasmas , in extrasolar contexts only an indirect relation can be established between the reconnection process and the observed emission ( e.g. , * ? ? ?
* ; * ? ? ?
common features seen in many pic simulations of magnetic reconnection include the formation of chains of magnetic flux ropes ( in three dimensions with a guide field ; otherwise , the common terms `` islands '' or `` plasmoids '' may still be more appropriate ) , the merging of smaller flux ropes into larger ones , and an energization of the plasma in the reconnection layer . in three dimensional simulations ,
kink - like and oblique modes , as well as secondary instabilities , can impart three dimensional structure to the reconnection layer .
typically , the simulations are initialized in the so - called harris sheet equilibrium describing a current sheet with a thickness similar to the plasma skin depth .
the tearing instability first sets in on scales of the initial current sheet thickness .
its nonlinear development produces a chain of skin - depth - scale flux ropes alternating with magnetic x - lines , the three dimensional generalization of two dimensional x - points . in x - lines ,
violation of flux freezing and magnetic line reconnection can be facilitated by a pressure tensor anisotropy ( * ? ? ?
* see also , e.g. , @xcite , and references therein ) .
smaller flux ropes tend to merge with each other to form larger ones ; this gives rise to magnetic organization on increasingly larger spatial scales .
three - dimensional pic simulations of guide field reconnection in electron - ion plasmas exhibit these same features , e.g. , @xcite found that oblique modes dominated over tearing modes when guide field was strong .
additional effects specific to plasmas with electron - ion mass disparity have also been identified , but are not relevant for the present work .
energization of particles in reconnection layers has been investigated in a number of pic simulations @xcite .
less attention has been given to particle energization in the general case of reconnection with a guide field in three dimensions @xcite .
in two dimensional simulations , particle acceleration producing a nonthermal energy spectrum , an apparent power - law , is often reported .
@xcite , however , instead detect a new ultrarelativistic thermal component energized by the reconnection . that the reconnection layers should energize particles
is in agreement with analytical considerations ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , which find that particles in the vicinity of the x - line in the reconnection layer are accelerated by the nearly - uniform electric field as they repeatedly cross , and are trapped within the converging plasma flows .
other mechanisms focusing on energetic particles that have moved from the x - line region into the flanking islands have also been suggested ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . in three dimensional simulations ,
evidence for a nonthermal spectrum is less solid .
it remains poorly understood which processes limit the energy to which particles can be accelerated in fully dynamical , three - dimensional reconnection layers , and how do the particle energy spectrum , the degree of accelerated particle beaming , and the temporal evolution of the accelerated population depend on the parameters of the reconnection layer . in this work
we employ three dimensional pic simulations to investigate the evolution of current sheets in relativistic pair plasmas undergoing magnetic reconnection .
our simulations add to the small but growing family of three dimensional pic simulation of relativistic reconnection with a guide field . with the intention to complement existing work , we initialize our simulations slightly differently than it is normally done , not assuming the usual harris sheet equilibrium .
also , we explore a parameter regime , involving magnetic to kinetic pressure ratios of the order of unity , that has thus far not received sufficient attention .
we observe an evolution of magnetic field geometry that constrains the viability of models for high - lundquist - number reconnection layers in which the diffusion region contains a hierarchy of interacting plasmoids ( e.g. , * ? ? ?
* ; * ? ? ?
the simulations also allow us to explore the character of particle energization in dynamical , fully three dimensional reconnection layers .
the paper is organized as follows .
section [ sec : simulations ] describes our methodology and simulation setup , while section [ sec : results ] presents the results .
section [ sec : discussion ] discusses our findings concerning development of kinetic instabilities in the current sheet , as well as our findings on particle energization , in view of the existing work on these topics .
finally , section [ sec : conclusions ] reviews our main conclusions .
the spatial domain is rectangular with @xmath1 , @xmath2 , and @xmath3 .
the boundary conditions are periodic in all directions .
the initial magnetic field is the same as that in the harris equilibrium @xmath4\hat{{\mathbf z}}\nonumber\\ & & + \kappa \,b_0\ , ( -\hat{\mathbf y } ) , \label{eq : harris_field}\end{aligned}\ ] ] where @xmath5 is the half - width of the initial current sheet and @xmath6 is a parameter defining the strength of the uniform guide field perpendicular to the opposing field , which in our simulations is oriented in the @xmath7 direction .
the current sheets are located at @xmath8 and @xmath9 and carry antiparallel currents . the current density profile that satisfies ampre
s law is @xmath10\hat{{\mathbf y } } .
\label{curprofile}\ ] ] to ensure that ampre s law is satisfied , if we set the particle density to be uniform , the particles have a spatially - dependent drift velocity @xmath11 in the @xmath7-direction .
the current density @xmath12 is related to the velocity @xmath13 by the relation @xmath14 where @xmath15 is magnitude of the unit charge carried by electron and ion macroparticles .
this results in the drift velocity profile @xmath16\hat{{\mathbf y } } , \ ] ] where @xmath17 . in this work ,
we initialize the simulation outside of pressure equilibrium with a uniform initial density @xmath18 and spatially varying drift velocity @xmath13 .
we can still attempt to relate the parameters of our simulation to those of preceding investigations .
in setting up initial conditions for a plasma with the magnetic field given in equation ( [ eq : harris_field ] ) , there are multiple ways in which pressure equilibrium can be satisfied depending on the spatial variation of the plasma density @xmath19 , temperature @xmath20 , charge drift velocity @xmath13 , reversing field strength @xmath21 , and guide field strength @xmath22 .
it is common to assume that the initial temperature is uniform and equal to @xmath23 , and only the density varies across the current sheet . in practice , harris sheets are often set up with a strong excess density in the current sheet for pressure balance .
it is common to split the particle population into two components , one a uniform background with density @xmath24 , and another spatially varying with maximum density @xmath18 at the center of the current sheet .
the latter component ensures pressure equilibrium and carries the current in the reconnection layer . in section [ sec : equilibrium ] below , we discuss how the initial configuration adjusts to approximate pressure equilibrium .
one can define the magnetic to kinetic pressure ratio via @xmath25 where here and henceforth we express the temperature in energy units . then in the isothermal harris sheet pressure equilibria used in preceding investigations
, the ratio simply equals the density contrast in the current sheet , @xmath26 . in our simulation
, @xmath27 is uniform outside the current sheet and equals @xmath28 note that @xmath27 is defined not taking into account the magnetic pressure of the guide field .
we do not introduce any initial perturbation to the initial field geometry described here .
the structure that develops is thus seeded by numerical fluctuations .
we initialize the simulation at temperature @xmath29 .
all the particles are drawn from the relativistic maxwellian distribution , implying that the average kinetic energy of the particles is @xmath30 .
we are interested in the dependence of reconnection mechanics and evolution of particle energy distribution on the dimensionless ratio of the magnetic pressure to the particle pressure @xmath27 , and the guide field amplitude parameter @xmath31 .
a nonvanishing @xmath31 indicates that the magnetic field is twisted in the current sheet and the plasma in the sheet center is magnetized .
we run a grid of nine simulations , with @xmath32 and @xmath33 , and carry out a detailed study of the run with @xmath34 and @xmath35 .
the chosen values of @xmath27 are low because we are interested in finding the lower limit of magnetization at which reconnection can produce significant particle energization .
lccccccccccc & 64&40&40&2&1 & 0 & 1.5&72&@xmath36 & 11 & 48.4 + s1k025 & 64&40&40&2&1 & @xmath37&1.0&28&@xmath36 & 23 & 34.3 + s1k1 & 64&40&40&2&1 & @xmath38&1.0&4.3&@xmath39 & 32 & 34.2 + s2k0 & 64&60&60&3&2 & 0&2.3&55&@xmath40 & 30 & 52.6 + s2k025 & 64&60&60&3&2 & @xmath37 & 2.3&44&@xmath41 & 29 & 54.4 + s2k025l & 128&120&120&3&2 & @xmath37&2.3&27 & @xmath42 & 38 & 58.9 + s2k025p & 64&60&60&3&2 & @xmath37 & 2.3&40&@xmath43 & 18 & 115 + s2k05 & 64&60&60&3&2 & @xmath44 & 2.3&25&@xmath45 & 27 & 44.9 + s2k1 & 64&60&60&3&2 & @xmath38&2.3&6.7&@xmath46 & 28 & 42.8 because of the low growth rate , the simulations with @xmath47 did not develop the tearing instability over the time period of the simulations and thus did not undergo reconnection , therefore , we do not show the results of these simulations in what follows .
the parameters of the simulations that did undergo reconnection are shown in table [ tab : simulations ] .
we use the relativistic particle - in - cell ( pic ) plasma code tristan - mp @xcite to simulate the evolution of a reconnection configuration in a pair plasma . in a pic simulation
, the number of macroparticles of each species located in each grid cell must be large enough to resolve variations in the current density and limit high - frequency particle noise .
tristan - mp uses a current filtering algorithm to reduce high - frequency particle noise , substantially reducing the required number of macroparticles per cell per species .
we initialize our simulations with 4 macroparticles per cell per species , which is fewer than found in other recent pic simulations @xcite . to verify that this low particle density does not lead to cell evacuation ,
we calculate the total particle density including both species in each cell in run s1k025 during the flux rope merging phase discussed in section [ sec : results ] below . we find that the total particle density in each cell calculated with a cubic cloud - in - cell kernel of size equal to the grid spacing is always larger than 1 , with a total density larger than @xmath48 in over @xmath49 of cells .
cells with low particle density occur in regions characterized by weak spatial field gradients that should not be sensitive to particle noise .
the cells with strong field gradients that would be sensitive to particle noise invariably have a higher than average particle density .
to ensure that the particle density is indeed sufficient to resolve the physics of reconnection , we have carried out two dimensional convergence tests , as well as a longer three dimensional test run with 80 particles per cell per species . in our simulations , the tearing instability leading to reconnection is seeded by numerical fluctuations that vary from simulation to simulation , so we do not expect all observables to be the same in each run at a given absolute time measured from the beginning of the simulation .
we therefore compare measurable quantities in each simulation at similar points in their evolution ; typically , we calculate these quantities at the end of the flux rope merging phase , which is discussed in section [ sec : results ] below .
the coarse time discretization employed in recording simulation output makes the determination of these evolution points imprecise .
this results in inaccuracies in the estimation of observables that vary quickly with time ; such quantities will therefore vary with resolution even if the resolution is sufficient to capture the physics of reconnection .
our two dimensional convergence tests use initial conditions matching those in the three dimensional simulations with @xmath34 and @xmath50 , but with a larger number , up to 16 , of macroparticles per species per cell .
we find that the only significant discrepancy , up to @xmath51 , is seen in the peak linear growth rate of the kinetic tearing instability ; this quantity is very sensitive to the coarse time discretization because time differencing is required for growth rate calculation . to test that we have sufficient macroparticle density to account for 3d effects in our simulations , we also carry out a three dimensional simulation with 80 macroparticles per cell per species , and the same parameters as in run s2k025 with @xmath34 and @xmath35 .
we again find that the measurable quantities do not deviate greatly from those found in run s2k025 , with the maximum deviation of @xmath52 again seen in the growth rate of the kinetic tearing instability .
we also find that the evolution of the total magnetic energy and the particle energy spectrum in this simulation with higher macroparticle density does not differ greatly from that found in run s2k025 in section [ sec : energization ] .
the main difference is that in the high - density simulation , the flux rope merging phase identified in section [ sec : results ] begins slightly later ; this may be understood as the result of slightly lower particle noise in the high density simulations , which results in later growth of instability . to ensure that our simulations resolve the plasma skin depth , we set @xmath53 , where the skin depth is given by @xmath54 here
, @xmath55 is the average lorentz factor of particles in the simulation and the additional factor of 2 in the denominator reflects the fact that the electrons and positrons oscillate together . to determine whether this value for @xmath56 is sufficient to resolve
the physics of reconnection , we have carried out further two dimensional convergence test simulations with up to 3 times larger number of grid cells per skin depth than in the three dimensional simulations , and again compared results at the end of the flux rope merging phase . here , the most resolution - dependent quantity was the maximum value of the reconnected magnetic field @xmath57 , which varied by @xmath58 with resolution .
this is a quantity that varies quickly with time near the end of the flux rope merging phase , so the apparent variation with resolution is again likely a result of coarse time discretization .
we focus on the regime in which the particle drift velocity in equation ( [ eq : beta_profile ] ) giving rise to the current density in equation ( [ curprofile ] ) is nonrelativistic .
this places a constraint on the initial current sheet width @xmath5 . combining equations ( [ curprofile ] ) and ( [ curdrift ] ) with the definition of @xmath27 in equation ( [ eq : sigma_uniform ] )
, it can be shown that @xmath59 where @xmath60 is the ratio of the particle pressure to the particle energy density ( which includes the particle rest energy ) ; in our simulations , @xmath61 . is related to the temperature via @xmath62 , \ ] ] where @xmath63 and @xmath64 is the modified bessel function of the second kind . ] with this estimate of the relation between @xmath5 and @xmath65 , in simulations with @xmath66 we choose @xmath67 , while in simulations with @xmath34 we choose @xmath68 .
this implies that @xmath69 , ensuring that the drift velocities are not relativistic .
the uniform initial density and temperature and nonuniform magnetic field in the initial configuration defined by equations ( [ eq : harris_field ] ) , ( [ curdrift ] ) , and ( [ eq : beta_profile ] ) are not in force balance , but they achieve approximate force balance following a brief readjustment . the initial adjustment to equilibrium results in a reduction of the current sheet thickness and an accompanying compression of the plasma by a factor @xmath70 where henceforth , @xmath71 denotes the current sheet half - thickness after the readjustment .
we further define @xmath72 to denote the light crossing times of the initial and the compressed current sheet .
the plasma takes @xmath73 to adjust to a state close to pressure equilibrium .
the magnitude of the compression is shown in table [ tab : simulations ] .
with readjustment , both the density of the particles and the strength of the guide field increase in the center of the current sheet in such a way as to balance the large pressure gradient that the reversing field exhibits from the center to the surface of the current sheet .
longer - term transients are present in the form of acoustic oscillations .
these oscillations could have the effect of enhancing the tearing instability , which relies on a velocity inflow to the current sheet
. the initial readjustment does not give rise to any significant magnetic dissipation or particle energization .
for a simulation to capture the physics of reconnection , its size must be sufficient to include the fastest growing modes of the important instabilities of relativistic harris current sheets in pair plasmas .
the known instabilities include the kinetic relativistic tearing instability ( kti ; hereafter the tearing mode ) , the drift - kink instability ( dki ; hereafter the kink mode ) , and the oblique mode which is similar to the tearing mode . to estimate their wavelengths , we follow @xcite .
the growth rate @xmath74 of the tearing mode with wave number @xmath75 in a harris current sheet of half - thickness @xmath71 and non - relativistic drift velocity @xmath76 is given by @xmath77\,\frac{1}{\tau_c } , \label{eq : tearing}\ ] ] where is a dimensionless function of the plasma temperature that in the limits of a cold and a relativistically hot plasma equals @xmath78 and @xmath79 .
the resulting maximum growth rate occurs for @xmath80 , corresponding to a wavelength of @xmath81 .
if we use the form of @xmath82 appropriate at ultrarelativistic temperatures , the growth rate for this mode is @xmath83 the situation is more complicated for the kink mode , because as @xcite find , the analytical maximum value of the growth rate @xmath84 occurs for @xmath85 , a wavenumber for which kinetic effects become important in a thin current sheet .
simulation results in their figure 20 indicate that the maximum growth rate occurs at @xmath86 corresponding to a wavelength of @xmath87 .
in addition to these two dimensional modes , it is also necessary to resolve oblique modes with @xmath88 that combine tearing and kink components ; these were identified in three dimensional simulations by @xcite and @xcite . the typical fastest - growing oblique mode in both of these simulations had @xmath89 ; this corresponds to a wavelength in both directions of @xmath90 . to resolve all three types of modes , in all simulations except for the larger size run s2k025l , we set @xmath91 , large enough to contain at least one wavelength of the tearing mode and two wavelengths of the kink mode . because the initial adjustment leads to a significant narrowing of the current sheet ,
several wavelengths of the fastest - growing tearing and drift - kink modes of the narrower current sheet are included in the simulations , and at least one wavelength of the oblique modes should also be resolved .
the resulting overall length scales are @xmath92 and @xmath93 or @xmath94 , depending on the simulation , as is shown in table [ tab : simulations ] . to ensure that little interaction takes place between the two current sheets in the periodic box , we set @xmath95 .
the current sheets in our larger size simulation s2k025l are as thick as in the other ones , but their separation and the dimensions of the box are twice as large , i.e. , @xmath96 and @xmath97 .
this simulation was carried out on a @xmath98 grid and contained a total of @xmath99 particles .
the smaller simulations with @xmath100 were carried out on a @xmath101 grid and contained @xmath102 particles , whereas the simulations with @xmath34 were carried out on a @xmath103 grid with @xmath104 particles . in our simulations
the growth times of the tearing and kink modes are on the same order as and smaller than the alfvn crossing time of the box . to capture the physics of reconnection in its nonlinear regime and on length scales similar to the box size , the duration of the simulation must be larger than these time scales .
the relativistic alfvn velocity is given by @xmath105 } } .\ ] ] the alfvn crossing time of the box @xmath106 is @xmath107 in the runs with @xmath108 .
we run all of our simulations for at least @xmath109 time steps , which amounts to @xmath110 in the large simulation s2k025l .
the evolution of the reconnection layer in all simulations exhibits the same common that have been observed in previous pic simulations of magnetic reconnection in pair plasmas .
the tearing instability grows and produces a chain of alternating x - lines and flux ropes .
here , we adopt `` flux ropes '' to denote a magnetic structure sometimes referred to as `` plasmoid '' ( or `` island '' in two dimensional treatments ) in which plasma is pinched by a helical magnetic field which can have a braided structure .
the flux ropes can have a finite lengths , with the field lines opening up and extending arbitrarily far from the rope axis .
the flux ropes can also split into sub - ropes , which opens the possibility of an organization of the ropes in a flux rope network . in all simulations except for s1k0 where the kink instability disrupts the current sheet , the flux ropes merge in quasi - hierarchical fashion until either only one flux rope is left in each current sheet or , in the large simulation s2k025l , the flux ropes have been disrupted by a transition to a disordered , three - dimensional state .
the period of flux rope merging is accompanied by fast magnetic - to - kinetic energy conversion .
this conversion produces a tail of energized particles . in simulations without guide field
, the kink instability also grows , resulting in some corrugation of the current sheet . for @xmath34 in simulation s2k0
, the kink instability does not disrupt the quasi - hierarchical merging process . for @xmath100 in simulation s1k0 , however , the kinking disrupts the merging and brings the two current sheets into contact where they can interact ; this is accompanied by rapid conversion of magnetic to thermal energy but the resulting spectrum does exhibit signatures of a secondary energized component . in agreement with @xcite , we find that that the presence of a guide field suppresses the kink instability .
however we find that in the large simulation s2k025l , a transition to three - dimensional evolution still occurs , likely due to a lack of large - scale phase coherence in the tearing instability .
we proceed to discuss our results in detail . in section [ sec : network ]
we discuss the evolution of the global current sheet and the formation of what will turn out to be a network of interconnected flux ropes in our largest simulation . in section [ sec : dissipation_flux_ropes ] we identify sites of magnetic reconnection within the network . in section [ sec : mode_analysis ] we carry out a fourier decomposition of perturbations in the largest simulation and discuss their growth rates . in section
[ sec : timescales ] we analyze the time scales associated with the evolution of the flux rope network . in section [ sec : reconnection_rate ] we measure the overall reconnection rate , while in section [ sec : ohms_law ] we analyze components of the nonideal electric field as reflected in the generalized ohm s law .
finally , in section [ sec : energization ] we discuss the overall rate of magnetic - to - thermal energy conversion as well as the efficiencies , mechanisms , and properties of particle energization in the simulations .
figure [ fig : slices ] shows a time sequence of slices at @xmath111 of the total current density @xmath112 and the plasma number density @xmath113 in simulation s2k025l . outside the two evolving current sheets , plane - parallel acoustic waves resulting from the initial pressure imbalance can be seen in the plasma density .
the waves traverse the computational box several times in the course of the simulation .
the early evolution of each current sheet , if observed at a single value of @xmath114 , is that correctly described by the familiar flux rope ( or island , plasmoid ) merging paradigm .
larger flux ropes produced by the merging of smaller ones contain substructure reflected in multiple , curved , embedded current sheets , each with a half - thickness @xmath115 .
very similar substructure also appears in the recent two - dimensional pic simulations of an electron - ion plasma by @xcite . to examine three dimensional aspects of the current sheet evolution , in figure [ fig : projections ] we show projections into the @xmath116 plane of the total current density @xmath117 , plasma number density @xmath113 , and the parallel electric field magnitude @xmath118 in each of the two current sheets in simulation s2k025l
the flux ropes that initially develop from the linear tearing instability are approximately parallel to the @xmath114 axis , which is the direction of the initial current flow .
however , each flux rope exhibits one or more discontinuities where the @xmath119 coordinate of the rope suddenly changes .
we interpret these discontinuities as arising from a lack of tearing mode phase coherence on scales larger than those in causal contact during the initial development of the instability .
the lack of phase coherence should occur on scales @xmath120 on which locations on the current sheet are out of causal contact during the first @xmath15-folding of the instability ; here , @xmath121 is the linear growth rate of a mode growing from numerical noise .
this sets a hard upper limit on the scale on which an instability can grow coherently ; other constraints may limit this scale further .
the initial departure from perfect translational invariance in the @xmath114 direction implies that the nonlinear flux rope merging will itself not occur coherently . in any three adjacent flux ropes ,
the middle rope can merge with one of the flanking ropes at one @xmath114 , and with the other rope at another @xmath114 ; at still other values of @xmath114 , the three ropes may remain separate for a time being .
this creates lateral linkage between the flux ropes ; already at time @xmath122 , all the flux ropes are mutually linked inside their current sheet .
the linkage implies each flux rope experiences a strongly @xmath114-dependent magnetic tension force .
this magnetic tension from neighboring flux ropes leads to rope tilting and kinking that is very distinct in origin from that arising from the well - known linear oblique and kink instabilities of a current sheet .
pairs of large flux ropes emerging from two generations of flux rope merging , but still oriented largely parallel to the @xmath114 axis , are often connected by minor flux ropes that are highly tilted toward the @xmath119 axis . @xcite
have discovered similar structures in a large three - dimensional simulation of electron - ion reconnection with a strong guide field .
it is clear in figure [ fig : projections ] that by @xmath123 , all flux rope segments have substantial tilts and any semblance of translational invariance in the direction of the original current flow is lost .
vertices of the flux rope network contain highly localized , intense current sheets and filaments .
by @xmath124 , the flux ropes are largely completely disrupted and the current sheet contains a disordered network of knot- and sheet - like structures . the force @xmath125 peaks at the primary x - line current sheets as well as flux rope perimeters , but is on average much smaller in flux rope interiors .
this suggests that the flux rope interiors are organized in a state of force free quasi - equilibrium constrained by a nonvanishing magnetic helicity ( i.e. , twisting , braiding of the field lines ) . to gain further insight into the structure of the flux rope network
, we examine the spatial variation of @xmath126 , the component of the electric field parallel to the magnetic field . the parallel field vanishes for the purely inductive electric field of ideal mhd and thus it can be obtained purely from the non - ideal , reconnection electric field @xmath127 defined as the difference between the actual and the induction electric field , @xmath128 in high magnetic reynolds number plasmas , the parallel field is present only in magnetic diffusion regions , and in particular , in locations where a change of magnetic connectivity is under way ( e.g. , * ? ? ?
exhibits noise in the entire domain due to small scale electrostatic fluctuations ; in our analysis , we filter these fluctuations on scales @xmath129 . ] the parallel electric field may also be instrumental in maintaining the pressure anisotropy , that is responsible for the breaking of flux freezing in the reconnection layer ( see , e.g. , * ? ? ?
the two rightmost columns of figure [ fig : projections ] show that the magnitude of the parallel electric field @xmath130 strongly peaks in the thin , long current sheets containing primary x - lines and connecting magnetic flux ropes .
however prior to flux rope disruption and the transition to a disordered state , the parallel field is normally very small inside magnetic flux ropes even when they contain embedded , substructure - level current sheets .
an exception are flux rope segments directly undergoing dynamical transformation , such as merging or stretching , where very isolated , spatially and temporally intermittent sites of significant @xmath118 occasionally appear .
the flux rope interiors are not undergoing pervasive , steady reconnection , in spite of their complex magnetic substructure .
the intermittent @xmath118 is also common after the flux rope network has become disordered , indicating that the disordered network remains active by developing transient , localized dissipation events in the braided network of disrupted flux ropes that effect further evolution of magnetic connectivity .
reconnection intermittency is also evident in figure [ fig : vapor ] , showing views of the current sheets @xmath131 , during the initial flux rope merging phase , and at @xmath132 , after the flux rope has become disordered .
the field lines shown are those passing through the sites where @xmath118 is significant , where the field line color is shown in green and field line geometry is that of an x - line .
the reconnected magnetic field snaps back to become incorporated into that of the flux ropes .
it is also noticeable that the flux ropes are not cylindrically symmetric but highly flattened ( ribbon - like ) and exhibit a clear longitudinal twist .
a common measure of magnetic connectivity is the concept of magnetic helicity , but its definition on the entire periodic domain presents unique challenges ( e.g. , * ? ? ? * and references therein ) . in spatially and temporally localized , simply - connected reconnection regions ,
a generalized helicity can be meaningfully defined using the finn - antonsen approach ( e.g. , * ? ? ?
* ) . in guide - field reconnection
, a change of the generalized helicity implies a change of global field line connectivity @xcite .
since the generalized helicity is locally generated by the source term @xmath133 that is closely related to the parallel electric field , the twisting and braiding of the field lines interior to a flux rope is a consequence of ( possibly @xmath114-coordinate dependent ) field connectivity change in the primary x - line region
. should be uniform in each individual x - line region . ] to identify modes present in the simulation , we carried out fourier decomposition of the electric and magnetic field for wave vectors @xmath134 parallel to the plane of the current sheet , @xmath135 , in simulation s2k025l .
we search for signatures of the tearing and kink instability , as well as for oblique modes , see , e.g. , @xcite and @xcite .
the tearing instability has @xmath136 and creates a perturbation in @xmath57 and @xmath137 .
the kink instability has @xmath138 and in the linear order creates a perturbation in @xmath139 if a guide field is present .
the oblique mode is similar to the tearing mode , except that the wave vector is tilted in the @xmath116 plane , that of the initial current sheet . in figure
[ fig : fft ] , we show fourier power spectra of the magnetic field component @xmath57 in the linear phase at @xmath140 and at the end of the fully nonlinear flux rope merging phase at @xmath122 .
we also show the corresponding growth rates evaluated over an interval of @xmath141 preceding each of the two times , @xmath142 in the figure , the growth rates are multiplied by the light crossing time of the compressed current sheet @xmath143 .
the modes with fastest growth rates are clustered around the @xmath144 tearing mode at @xmath140 , but the fastest growth then shifts to longer wavelengths and oblique directions by @xmath122 . at the end of the linear phase , the power spectrum for @xmath57 and @xmath137 peaks at @xmath145 , which is the tearing mode , consistent with the presence of @xmath146 flux ropes in each current sheet .
the peak , however , is broadened , likely by an initial lack of phase coherence on scales given by equation ( [ eq : coherence_length ] ) . for the tearing mode , @xmath147 . from this
we expect lack of phase coherence on scales @xmath148 , which can explain the observed broadening .
however the broadening also allows the possibility that an authentic oblique component is present , too . at the end of the flux rope merging phase , which we define as the time @xmath149 when the orderly flux rope merging gives way to a more disordered evolution ,
the peak has shifted to longer wavelengths @xmath150 with significant additional power occurring in the oblique direction , for @xmath151 .
we emphasize , again , that the oblique power may be an outcome of phase decoherence , as well as of secondary instabilities developing in the nonlinear regime , rather than of a genuine primary oblique instability mode .
turning to our search for the kink mode , we find that @xmath139 develops nonvanishing albeit weak @xmath152 power at @xmath153 with the peak wavelength corresponding to @xmath154 .
however , most of the power in @xmath139 is still along @xmath155 .
other components of @xmath156 and @xmath157 are devoid of power along @xmath152 .
this suggests that the simulation exhibits no evidence for a linear kink mode and that any variation with @xmath114 emerges in the nonlinear development of the tearing mode .
we compare these results to previous analyses of tearing , kink , and oblique modes in section [ sec : mode_discussion ] below .
once the tearing instability has produced a collection of nonlinear flux ropes at a time which we will denote with @xmath158 , the ropes begin to merge .
the merging time should scale with the island separation @xmath159 as @xmath160 where @xmath161 is a dimensionless coefficient encapsulating the dynamics of interaction between flux ropes .
then , the number of flux ropes per unit length along the current sheet , which we denote with @xmath162 , can be estimated as @xmath163 by visually counting the number of flux ropes in each current sheet in simulation s2k025l in figure [ fig : projections ] , we found that @xmath164 suggesting that @xmath165 in that simulation . comparing equations ( [ eq : island_density_model ] ) and ( [ eq : island_density_fit ] ) we estimate that @xmath166 .
given that the reconnection outflow velocities reach only about a half of the alfvn velocity in the simulation , it is not surprising that the characteristic velocity associated with flux rope merging @xmath167 is only a fraction of the alfvn velocity .
in reconnection configurations starting from a one - dimensional current sheet lacking any @xmath114 and @xmath119 dependence , the three - dimensionality manifested in an emerging @xmath114 dependence can arise from several effects which include an intrinsic obliqueness of the tearing modes , the development of kink modes , and a lack of coherence on scales on which the phase of the tearing mode is uncorrelated ( see section [ sec : unstable_modes ] ) . in reconnection regimes
in which oblique and kink modes are suppressed , flux rope dynamics acquires three dimensional character when the flux rope separation @xmath168 starts substantially exceeding the coherence length @xmath169 . from equations ( [ eq : kti_max ] ) , ( [ eq : coherence_length ] ) , and ( [ eq : island_density_model ] ) , this happens at times @xmath170 yielding an estimate @xmath171 for simulations s2k025 and s2k025l .
indeed , the flux rope geometry in figure [ fig : projections ] seems to acquire three - dimensional character at @xmath172 .
this is consistent with the rise of power at oblique wave vectors @xmath173 in figure [ fig : fft ] at similar times .
this analysis suggests that the reconnection layers acquire three dimensional structure on relatively short time scales regardless of the growth rates of the oblique and kink modes .
we now discuss the reconnection efficiency at primary x - lines in the early stages of the evolution of the current sheet . in this regime
, the reconnection site can be described using a typical two - dimensional model of reconnection in which two oppositely oriented fields are separated by a thin current layer in which the field reverses . in this layer , magnetic field lines diffuse across the plasma to reconnect at one or more x - lines .
magnetized plasma approaches the central plane of the layer , toward an x - line , with an asymptotic inflow velocity @xmath174 , which is also known as the reconnection velocity . after passing the x - line
, plasma is expelled from the vicinity of the x - line to either side at the outflow velocity @xmath175 , which is typically assumed to equal the alfv ' en velocity @xmath176 .
the orientation of the reconnected magnetic field is approximately perpendicular to that of the original field .
defining @xmath177 and @xmath178 to be the thickness and length of the current sheet , the conservation of mass from the inflow to the outflow requires @xmath179 the inflow velocity can be related to the electric field in the reconnection region @xmath180 which is approximately uniform and equal to the dynamical electric field the plasma inflow just outside the current sheet @xmath181 because the inflow velocity and the reversing magnetic field are perpendicular , the dimensionless reconnection rate @xmath182 , defined as the ratio of the inflow and the outflow velocity , and the electric field in the reconnection region , can be related via @xmath183 this calculation of the reconnection rate makes the common assumption that the outflow velocity is equal to the alfvn velocity in the inflow region . in our simulations , this assumption is not valid as the outflow velocity is smaller than the alfv ' en velocity . in simulation s2k025 , we find that indeed , the total electric field in the reconnection region is approximately uniform across the current sheet . given that the alfvn velocity is @xmath184 in this simulation , the dimensionless reconnection rate calculated from the electric field following equation ( [ rece ] ) is @xmath185 at @xmath122 , in the middle of the initial flux rope merging phase .
this is similar to the peak reconnection rates of about @xmath186 found in other three dimensional pic simulations with higher magnetizations ( e.g. , * ? ? ?
* ; * ? ? ?
this comparison assumes that reconnection at the largest x - line can be described accurately by a two dimensional reconnection model .
the inflow and outflow velocities at @xmath122 are @xmath187 and @xmath188 , respectively .
the peak plasma outflow velocity slightly later in the simulation reaches @xmath189 , just above half of the alfvn velocity .
since the alfvn velocity is not ultrarelativistic , it is not surprising that we are not finding ultrarelativistic flows in any of our simulations.^{-1/2}\gg 1 $ ] , but this may not be sufficient , e.g. , @xcite found maximum flow speeds of @xmath190 even in their two - dimensional simulations with high magnetization and no guide field . ]
if the dimensionless reconnection rate is calculated directly from the inflow and outflow velocities , it is @xmath191 .
the difference between the outflow velocity and the alfvn velocity , which are equal in idealized two dimensional reconnection models , is likely due to the dynamical , non - steady state nature of the plasma flow in our simulations ; similar discrepancies were found by @xcite in their nonrelativistic pic simulations , and @xcite in their relativistic mhd simulations . .
the center of the current sheet is at @xmath192 .
the lines represent the measured electric field @xmath137 ( solid , black ) , the induction component ( dotted , red ) , the pressure component ( dashed , green ) , the inertial component ( dot - dashed , blue ) , and the total ( pink , solid ) of the three calculated electric field terms .
the generalized ohm s law mandates that the latter should equal the actual electric field , which is the case apart from fluctuations arising from numerical discreteness .
the ohm s law components are smoothed with a gaussian kernel of standard deviation @xmath193 .
it is clear that at the x - line , the pressure term dominates over the inertial term.[fig : ohmslawy],scaledwidth=47.5% ] a central question in magnetic reconnection is the nature of the process that facilitates the violation of flux freezing in the diffusion region .
the total electric field @xmath156 in a collisionless plasma is related to the magnetic field @xmath157 and to moments of the particle distribution function of particle species @xmath194 by the generalized ohm s law ( e.g. , * ? ? ?
* ) @xmath195 where @xmath196 and @xmath197 are the average velocities and momenta of particles , @xmath198 is the pressure tensor , @xmath199 is the particle number density , and @xmath200 and @xmath201 are the particle charge and mass . averaging over the two species @xmath202 we recover the non - ideal , reconnection electric field @xmath203 on the left hand side of equation ( [ eq : g_ohms ] ) . in collisionless plasmas with long - lived magnetization ( counterexamples being the plasmas excited by kinetic instabilities , such as the filamentation instability ) , the pressure and the inertial terms on the right hand side of equation ( [ eq : g_ohms ] ) are small almost everywhere . both of these terms ,
however , can become important in the diffusion region of a magnetic reconnection layer . to determine which terms give rise to @xmath203 , we calculate the terms on both sides of equation ( [ eq : g_ohms ] ) as they vary across the x - line in simulation s2k025 .
we find no significant reconnection electric field in the @xmath204 or @xmath119 directions ; there is only a reconnection field in the @xmath114 direction , consistent with two dimensional models of reconnection .
meanwhile , during the linear growth phase of the tearing instability , spatial gradients are present only in the @xmath204 and @xmath119 directions , implying that the off - diagonal @xmath205 and @xmath206 components of the pressure tensor contribute to the first term on the right hand side of equation ( [ eq : g_ohms ] ) .
in particular , we find that most of the pressure term is provided by @xmath207 .
figure [ fig : ohmslawy ] shows the @xmath114 component of various terms contributing to the electric field .
the pressure term is dominant , and the total of the inductive , pressure , and inertial terms approximates the electric field , with departures arising from shot noise .
these results are consistent with previous investigations ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) that found that the spatial variation of off - diagonal terms of the pressure tensor was the main contributor to the reconnection electric field .
the results of our ohm s law calculation with a moderate guide field , combined with the similar results of other simulations with strong guide field ( * ? ? ?
* nonrelativistic simulations ) and no guide field ( * ? ? ?
* relativistic simulations ) , constitute evidence that for moderate magnetizations and all guide field magnitudes , the reconnection electric field is produced by the pressure term in three as well as in two dimensions . in an exception to this agreement
, @xcite found that the inertial term became important in highly magnetized relativistic current sheets with a guide field , which is a regime we do not probe .
we turn to the topic of magnetic to kinetic energy conversion and particle energization .
figure [ fig : energy_evolution ] shows the evolution of the total magnetic field , electric field , and particle kinetic energies in simulations s2k025 and s2k025l .
energy was conserved to within @xmath208 in all the simulations .
the initial kinetic and electromagnetic energy fraction differs in the two simulations because the initial current sheet occupies a fraction of the volume of the simulation box that is twice as large in simulation s2k025 as in s2k025l .
the trends , however , are consistent taking the volume difference into account . an initial reversible fluctuation of the magnetic - to - kinetic ratio during the first @xmath209 is associated with the readjustment to pressure equilibrium .
dissipative conversion of magnetic to kinetic energy seems to first occur in a couple of bursts at @xmath210 and @xmath211 .
then , the system enters a phase , lasting until the end of the simulation , in which dissipation is relatively steady .
the two bursts seem to coincide with initial tearing instability and the formation of @xmath212 flux ropes ( magnetic islands ) across the simulation box per current sheet , and the subsequent flux rope merging , yielding @xmath213 flux ropes per current sheet by @xmath214 , after which the overall geometry of the reconnection layer becomes fully three dimensional .
by @xmath215 , about @xmath51 and @xmath216 of the initial magnetic energy is converted into particle kinetic energy in the smaller and larger simulation , respectively . in table
[ tab : simulations ] , we show @xmath217 , the fraction of magnetic field energy that is converted to particle kinetic energy in each simulation . here
, we define @xmath218 to be the initial energy in the total magnetic field , which includes both the reversing field and the guide field , and @xmath219 is the change of the total magnetic energy from the beginning to the end of the simulation .
the energy in the reversing field only is a factor of @xmath220 smaller than @xmath218 if we ignore the reduction of the reversing field in the current sheets .
independent of this correction , @xmath217 exhibits a strong decreasing trend with increasing guide field strength @xmath31 .
it also exhibits a weaker increasing trend with @xmath27 ; note that the very high @xmath221 in simulation s1k0 is the result of destructive current sheet interaction . in the simulations with @xmath222 , and also in simulation s1k025 with @xmath35 and @xmath100 ,
the conversion of magnetic to kinetic energy is substantially weaker than in the other simulations .
we also note that the rate of development and evolution of the flux rope network exhibits similar behavior in the sense that flux ropes of a given size assemble later ( @xmath223 vs. @xmath224 ) in the @xmath222 and s1k025 simulations .
the variation of @xmath217 and of the overall evolution rate with @xmath31 is likely a consequence of the decrease of plasma compressibility with increasing guide field strength ( e.g. , * ? ? ?
* and references therein ) .
since most of the magnetic to particle kinetic energy conversion takes place during the relaxation of reconnection field lines to their final equilibrium in flux ropes , the less compressible case of a strong guide field leads to larger flux ropes and smaller energy conversion .
figure [ fig : energy_spectrum ] , top panel , shows the energy spectrum at a reference time @xmath215 in six of the smaller size simulations .
each of the spectra exhibits a peak in agreement with the initial thermal distribution as well as a new tail resulting from particle energization extending to maximum lorentz factors in the range @xmath225 ( for the precise values at the end of the simulations , see table [ tab : simulations ] ) .
the simulations with the strongest particle energization are those with higher magnetization @xmath34 and zero or moderate guide field @xmath226 .
rather weak energization is seen in all simulations with strong guide field @xmath227 and in the simulation s1k025 with weak magnetization and moderate guide field .
we observe an intermediate level of energization in simulation s1k0 with weak magnetization and no guide field ; this simulation is unique in that at the reference time , the two current sheets have already interacted with each other .
it is important to note that all of these spectra are calculated at the same reference time ; therefore , these differences in energization efficiency might result from the differing energy conversion rates discussed in the previous section .
however , we will find below that the spectrum exhibits the strongest evolution during the flux rope merging phase ; because the merging phase is complete by the reference time in all the simulations , it seems that the differences in the level of energization between the runs can not be attributed solely to the differences in the rate of reconnection . in order to quantify the degree of particle energization in the simulations , following @xcite ,
we compute the particle kinetic energy @xmath228 contained in the difference between the measured spectrum and a thermal spectrum containing the same number of particles and the same energy at particle energies at which the measured spectrum is in excess of the thermal spectrum .
we then calculate the ratio of @xmath228 to the total particle kinetic energy @xmath64 , both computed at the end of the simulation .
it is worth remarking that we do _ not _ go as far as @xcite to identify the particles contributing to @xmath228 as a nonthermal population ; the following section we show this tail may be better described with a combination of several thermal populations .
consistent with the trend observed in the shape of the spectrum , we find that @xmath229 , which is shown in table [ tab : simulations ] , is the largest in the simulations with the higher magnetization @xmath34 and at most moderate guide field @xmath226 and is the smallest in simulation s1k1 with lower magnetization @xmath100 and strong guide field @xmath227 . thus , in contrast with @xcite , we find that @xmath229 decreases with increasing guide field .
the reason for this dissimilarity seems to be that the kink instability does not compromise the development of reconnection in any of our simulations other than s1k0 ; even in that run , a minor flux rope merger occurs before the current sheet is disrupted . without the kink instability , more particle energization can take place .
another possibly more interesting measure of particle energization is the ratio of the kinetic energy in energized particles to the magnetic energy converted to kinetic energy . combining equations ( [ eq : harris_field ] ) and
( [ eq : sigma_uniform ] ) with the definition of @xmath230 , and taking account of the conservation of energy @xmath231 , this ratio can be written as @xmath232 here , the dimensionless coefficient @xmath233 is the ratio of the initial particle kinetic energy density to the magnetic energy density including the guide field , given by @xmath234 where @xmath235 and all the quantities entering the definition of @xmath233 are evaluated at the beginning of the simulation .
the ratio , which ranges between @xmath236 and @xmath237 , is shown in table [ tab : simulations ] .
the trend seen at low magnetization @xmath100 is an increase of @xmath238 with guide field strength , approximately the opposite of that seen in @xmath229 .
the trend can be understood by noting that , as we will find in section [ sec : energization_mechanism ] , most of the particle energization contributing to @xmath228 occurs in x - line regions .
meanwhile , @xmath219 measures the change in magnetic energy both during reconnection in x - line regions and , more importantly , during the subsequent contraction of reconnected magnetic field lines into flux ropes .
therefore the ratio @xmath239 should indeed increase with stronger guide field because the guide field reduces plasma compressibility and moderates field line contraction .
figure [ fig : energy_spectrum ] , middle panel , shows the evolution of the particle energy spectrum in the large simulation s2k025l .
the spectrum evolves little during the linear tearing phase and the first round of flux rope merging at @xmath240 , but then it quickly hardens by @xmath241 as the flux rope merging concludes and the reconnection layer transitions to a disordered state .
further hardening takes place until the end of the simulation at @xmath215 . for comparison
, we also show the particle energy spectrum at this time from the smaller , equivalent simulation s2k025 .
the two spectra are consistent at lorentz factors @xmath242 indicating convergence with increasing box size at relatively low energies .
however , the larger simulation has progressively more particles at still higher energies , with lorentz factors reaching @xmath243 . we proceed to model the entire spectrum in simulation s2k025l as is , while keeping in mind that in a still larger simulation , further evolution of the spectrum at the highest energies is likely to be expected .
we experimented with composite populations containing thermal as well as power law components , and found that at lorentz factors @xmath244 , a model containing three thermal populations , each described by a relativistic maxwellian , seems to work best .
the model and the three thermal components are shown the bottom panel of figure [ fig : energy_spectrum ] .
we fix the temperature of the first population to be equal to the temperature of the initial plasma , @xmath245 .
a least - squares fit yielded temperatures @xmath246 and @xmath247 for the second and third population , respectively .
the first component contains @xmath248 of the particles and @xmath249 of particle kinetic energy , the second component contains @xmath250 of particles and @xmath251 of particle kinetic energy , and the third component contains only @xmath252 of the particles and @xmath253 of particle kinetic energy .
we show in section [ sec : energization_sites ] that the energized populations and especially the particles with @xmath254 are located close to the primary x - lines ( within about one larmor radius ) and in flux ropes . to gain insight in the nature of the particle energization process , in simulation s2k025 we select a number , @xmath255 , of particles reaching the highest energies , corresponding to lorentz factors @xmath256 , at the end of the flux rope merging phase at @xmath257 .
we trace the orbits of these particles throughout the merging phase over the time interval @xmath258 .
this is the period during which the particles experience coherent energization .
the particles generally begin in the current sheet and have momenta that , on average , are aligned with the direction of the electric force @xmath259 in the current sheet , with a median inclination of @xmath260 from the direction of the force . during flux rope merging , the traced particle energies increase approximately linearly .
however at the instance of merging for the two largest flux ropes , which occurs at @xmath261 , most of the particles incur an energy increment accounting for @xmath262 of the total energy gain . following the flux rope merging phase , some traced particles gain additional energy , but others lose energy , and both the gain and the loss could be considered manifestations of a thermalization process . indeed ,
the energization @xmath229 does not further increase after the flux rope merging is complete .
[ sec : energization_sites ] to pin down the geometric location of particle energization , for the traced particles , we compute the total energy gain per particle per unit volume averaged over @xmath114 via @xmath263\nonumber\\ & & \times \delta[{\mathbf x}-{\mathbf x}_j(t)]\ , dt \,\frac{dy}{l_y } , \end{aligned}\ ] ] where @xmath264 and @xmath265 are the position and velocity of traced particle @xmath266 at time @xmath267 , @xmath177 is the dirac delta , and the time integral covers the flux rope merging phase . in figure
[ fig : acceleration ] , we compare the energy gain @xmath268 to the corresponding absolute value of the @xmath114- and @xmath267-averaged current density @xmath269 . the highest energy gain is clearly associated with the largest , best defined x - line on the lower left .
substantial energization is also detected in the outflow regions flanking this x - line .
the total energy gain in the flanking regions , seen vertically above and below the x - line in the figure , is larger than that in the narrow vicinity of the x - line .
it is puzzling that the overall energization efficiency is weaker in the second current sheet shown on the right in the figure . in both sheets
, the energization seems to be associated with some current density maxima but not with others .
energization seems to prefer long , continuous x - line regions with thin current sheets , perhaps because these are the sites of plasma inflow . in figure
[ fig : trajplot ] we show the orbits of six representative traced particles belong to the current sheet showing on the left of figure [ fig : acceleration ] .
they all start in the current sheet within @xmath270 from the x - line and have initial lorentz factors in the range @xmath271 . while in the x - line region
, the particles oscillate across the current sheet on speiser - like orbits .
the particles reach lorentz factors of @xmath272 ( this is how they were selected ; see figure [ fig : trajplot ] , right panel ) after drifting from the x - line region into the neighboring islands . at the final lorentz factor ,
the particles have larmor radii evaluated using the reversing magnetic field @xmath273 , larger than the width of the flux ropes in the simulation . because the magnetic field in flux ropes is only somewhat enhanced compared to the reversing field , the accelerated particles are not easily trapped within the flux ropes . in the simulation s2k025 ( left panel ) , and the evolution of the lorentz factor as a function of coordinate @xmath119 ( right panel ) .
the dashed line indicates the center of the current sheet .
note that many of the particles start at @xmath274 and close to the center of the current sheet ; this is the location of the large x - line on the lower left in figure [ fig : acceleration ] .
the crosses indicate the particle position at the beginning of the acceleration phase .
[ fig : trajplot ] , scaledwidth=43.0% ] the electric field is nearly uniform across the reconnection region , hence the electric force acting on a particle trapped in a speiser orbit is approximately constant in time provided that the the inclination of the particle s velocity relative to the direction of the electric force is relatively small .
most of the traced particles happen to fulfill this condition .
then , the work done by the electric force on the particle is independent of the initial particle energy .
we have also checked that , we expect for the speiser orbit acceleration , it is the @xmath114-component of the electric field that contributes the most to the energy gain . the average total energy gain per particle obtained by integrating @xmath268 over the volume of the simulation is @xmath275 . because the rate of particle acceleration in the @xmath114-direction is constant in time
, the total energy gain is limited by the time the particle spends in the x - line region ( see , e.g. , * ? ? ?
particles displaced from the very center of the x - line experience a lorentz force due to the reconnected magnetic field @xmath57 that deflects them away from the x - line and toward the flux rope . once a particle has left the x - line region and entered a flux rope , it no longer experiences coherent acceleration .
the lack of acceleration internal to flux ropes of the type proposed by @xcite was attributed in previous works to the presence of the guide field @xcite . here , in addition
the presence of the guide field , we note that unlike the particles traced in @xcite , our energized particles already have very large larmor radii by the time they transition from the x - line region into a flux rope and the magnetic geometry and dynamics also seem markedly different .
if the motion of high energy particles accelerated in a reconnection region is anisotropic , then the synchrotron radiation emitted by these particles carries angular dependence .
radiation emitted by beams of accelerated particles can come in and out of view of an observer ; at any given time , the observer detects only the radiation emitted by particles with momenta making angles @xmath276 from the line of sight .
such beaming could explain the temporal variability often seen in astrophysical sources , such as grbs ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , blazars ( e.g. , * ? ? ?
* ) , that may be powered by reconnection .
it could also influence the characteristics of the gamma - ray emission from other candidate reconnection - powered astrophysical sources , such as magnetars @xcite and pulsar wind nebulae ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
we investigate the angular distribution of the momenta of the highest energy particles in our simulation , and compare with the results of @xcite , who recently reported a high degree of beaming in a two - dimensional pic simulation of pair plasma reconnection initialized with @xmath277 .
figure [ fig : directions ] shows the orientations of the momenta of all of the electrons and ions with @xmath278 in simulation s2k025l .
we select the larger simulation for this analysis to include a larger range of three - dimensional effects to which the particle anisotropy may be sensitive .
the median inclination of electrons ( ions ) to the @xmath114-axis ( negative @xmath114-axis ) is @xmath279 , hence a half of particles in each charge species occupies a fraction @xmath280 of the full solid angle .
this moderate degree of beaming is a natural consequence of x - line acceleration by the electric field in the reconnection region , since the electric field is uniform and accelerates the particles preferentially along the @xmath114-axis .
the degree of beaming , however , is much smaller than in @xcite , where the highest energy particles , with lorentz factors @xmath281 , were strongly beamed , occupying a solid angle fraction as small as @xmath282 .
the dependence of the beaming on the parameters of the reconnection layer can crudely be understood as follows .
assuming for simplicity that the plasma is ultrarelativistic and highly magnetized , and that the energy of the accelerated particle increases by a large factor , the components of its momentum following acceleration are approximately @xmath283 and @xmath284 , where @xmath285 is the initial lorentz factor of an accelerated particle , and @xmath286 is the lorentz factor gain during acceleration , which increases the particle momentum in the @xmath114 direction .
therefore , the particle s degree of beaming is proportional to @xmath287 . using @xmath288 , where @xmath289 is the accelerating electric field ( equation [ rece ] ) and @xmath290 is the duration of acceleration ( equation [ eq : tau_merge ] ) , and expressing the magnetic field in terms of the magnetization parameter @xmath291 , where @xmath235 ( equations [ eq : sigma_uniform ] and [ plasma ] ) , we obtain @xmath292 this result is accurate in the regime @xmath293 whereas cerutti et al . set up their simulation with @xmath294 . ignoring this concern
, we estimate that the factors in parentheses on the right hand side of equation ( [ eq : beaming_scaling ] ) are similar in our simulation and in cerutti et al .
we therefore expect that the beaming solid angle @xmath295 in their simulation should be smaller by the inverse ratio of the 3s in the two simulations , which is @xmath296 .
thus , the crude expectation is that the particles accelerated in cerutti et al.should more beamed within a solid angle over an order of magnitude smaller than in our simulation , consistent with the observed beaming ratio between the two simulations . in conclusion
, this analysis suggests that high magnetization and large physical size of the x - line region can both give rise to a high degree of beaming .
a number of studies have examined the three - dimensional evolution of a current sheet without a guide field ; here , we focus on results relevant to pair plasmas .
@xcite found that the kink mode dominated the tearing mode in antiparallel reconnection .
in contrast , @xcite found that the initial evolution was dominated by the tearing mode , leading to an initial merging phase similar to that in two dimensions .
after the completion of the merging phase , however , they found that a secondary kink instability set in and rendered the reconnection region turbulent and filled with plasmoids covering a range of dimensions . the reduced influence of the kink mode in the simulations of @xcite may be explained by initially relativistic drift velocities of @xmath297 .
analytical calculations suggest that while the kink mode dominates at low drift velocities , the tearing mode becomes dominant for drift velocities exceeding @xmath298 @xcite .
however , the growth rate of the kink mode , unlike the tearing mode , depends strongly on the initial structure of the current sheet @xcite . in guide field reconnection
, previous three dimensional studies have found that the kink mode is stabilized , but , under certain conditions , oblique modes dominate the evolution .
@xcite found that with a guide field present , the tearing mode and a sausage - like mode are combined into an oblique `` relativistic - drift - sausage - tearing '' mode .
@xcite found that in electron - ion plasmas , oblique modes give rise to a network of interconnected flux ropes . in both cases ,
the nonlinear development of the oblique modes led to turbulence in the current sheet . in nonrelativistic electron - ion simulations with a strong guide field
, @xcite found that at low temperatures , the current sheet filamented and that the turbulent transport played a key role in the diffusion of the field . at higher ,
albeit still nonrelativistic temperatures , their filamentation instability became weaker and the current sheet did not become turbulent .
one must be aware of the possibility that unlike the tearing mode , the other instabilities identified in simulations starting with harris sheet equilibria are idiosyncrasies of the initial configuration and have little to do with astrophysical magnetic reconnection .
our simulation , starting from a non - harris like configuration lacking a density enhancement in the initial current sheet , should help discern artifacts of the initial configurations from genuine properties of reconnection dynamics .
we have found that in our simulations , in which we considered guide fields at most equal to the reversing field , the tearing mode dominates the evolution at all guide field strengths . to understand the difference between our simulations without guide field and the studies that detected a dominant kink mode , we have carried out two dimensional simulations with @xmath34 and @xmath50 in both the @xmath205 and @xmath299 planes .
we find that the kink mode grows faster in a harris current sheet , while the tearing mode grows faster starting with our initial conditions .
our initial configuration seems to enhance the growth rate of the tearing mode and inhibit the growth of the kink mode .
the plasma inflow occurring during readjustment to equilibrium may enhance growth of the tearing mode above the value expected in pressure equilibrium .
furthermore , our initial configuration may be endowed with a somewhat stronger drift velocity shear than the harris sheet and this could stabilize the kink mode @xcite while having little effect on the tearing mode @xcite . in our simulations with guide field , we do not find any oblique modes similar to those observed by @xcite or @xcite .
therefore , the only linear mode present is the tearing mode ; the lack of phase coherence is the mechanism responsible for breaking translational symmetry .
because this mechanism is a general result of causal constraints , it is likely that this mechanism is present in harris current sheets as well as in our simulations , but is masked by the more rapid effects of kink and oblique modes .
lccccc s1k0 & 0.365&1.5&3&0.12&0.06 + s1k025 & 0.365&1.0&2&0.08&0.04 + s1k1 & 0.365&1.0&2&0.08&0.03 + s2k0&0.344&2.3&4&0.16&0.14 + s2k025 & 0.344&2.3&4&0.16&0.13 + s2k025l&0.344&2.3&9&0.16&0.11 + s2k1 & 0.344&2.3&2&0.09&0.07 we proceed to calculate the theoretical and actual tearing mode growth rates in each simulation . in table
[ tab : tearing ] we show the two rates which we label @xmath300 and @xmath301 , along with the initial particle drift velocity @xmath65 , the initial current sheet compression factor @xmath302 , and the number of flux ropes per current sheet determined by visual inspection @xmath303 .
because the fastest - growing tearing wavelength computing from the theoretical result in equation ( [ eq : tearing ] ) is @xmath304 and the box size in all by simulation s2k025l is @xmath305 , the number of flux ropes per current sheet produced by the tearing mode should be @xmath306 ; the number should be twice as large in s2k025l .
the actual number of flux ropes formed in the simulation is agreement with this prediction , with the exception of the simulation s2k1 with the strongest guide field .
this run produced only 2 flux ropes per current sheet even though one expects 4 .
as discussed in @xcite , the strong guide field may prefer oblique rather than pure tearing modes , but the former grow the fastest at long wavelengths @xmath307 , about half the box size of the simulation , and the finite box size interferes with ability of arbitrary oblique modes to grow .
the values of @xmath300 for the various runs are shown in table [ tab : tearing ] ; it should be noted that these are upper limits to the possible tearing mode growth rates , because any measured growth rates are found from estimates in the nonlinear regime .
we also compute the simulated growth rates @xmath301 of the tearing mode in our simulations by calculating the growth rate of the average reconnected magnetic field perturbation @xmath57 in the current sheet .
growth rates of the kti calculated with the latter method should be higher than those obtained by examining a single fourier mode , as we did in section [ sec : mode_analysis ] , because multiple fourier modes contribute to the amplitude of the perturbation .
we find that the theoretical and simulated growth rates were relatively similar in all simulations , with theoretical rates being at most double the simulated rates .
@xcite identified a similar discrepancy between the theoretical and simulated growth rates .
the growth rates also show that decreased guide field @xmath31 , and especially higher magnetization @xmath27 , lead to higher tearing mode growth rates both in theoretical calculations and in the simulations .
this reinforces the effects of the variation in @xmath303 , which has a similar dependence on @xmath27 and @xmath31 .
we have shown that the tearing mode is responsible for producing magnetic flux ropes in most of our simulations .
the tearing mode can also explain many of the differences between the simulations .
the runs with higher @xmath27 and smaller @xmath31 have larger values of @xmath302 , which leads to higher tearing mode growth rates and the formation of a larger number of flux ropes @xmath303 .
the faster evolution of runs with higher @xmath27 and smaller @xmath31 can be explained by the faster growth of the tearing mode in such runs .
because runs with higher @xmath27 and smaller @xmath31 are also associated with a larger @xmath303 , more flux rope merging can take place in such runs .
flux rope merging is strongly associated with energy transfer and particle acceleration ( as we show in the following sections ) , so the larger energy transfer in runs with higher @xmath27 and smaller @xmath31 can be explained by the larger @xmath303 in those runs .
this suggests that a large portion of the variation between runs is a result of the effect of @xmath27 and @xmath31 on the tearing mode .
various particle energization channels have been identified in pic simulations of plasma reconnection ( see , e.g. , * ? ? ? * and references therein ) .
three specific loci where particle energization was detected include : inside or near the diffusion region containing reconnection x - lines , in the magnetic islands ( or plasmoids , flux ropes ) where the reconnected magnetic flux accumulates , and between an x - line and the edge of a flanking island , where the plasma flowing out of the x - line first encounters a strong magnetic field gradient .
a number of studies point to the conclusion that significant energization occurs near the primary ( i.e. , largest , or highest rank ) x - lines . there ,
the electric field , which is perpendicular to the reversing field and is aligned with the current density vector in the middle of the current sheet , can accelerate particles oscillating within or across the current sheet ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
a variant of this mechanism involves the trapping of particles in secondary magnetic islands appearing within the diffusion region of an x - line @xcite ; we will not discuss this as we do not observe secondary flux rope formation in our simulations .
particles can also be accelerated outside of the x - lines , such as in primary magnetic islands .
if an island contracts in the course of its relaxation to mhd equilibrium , the particles trapped inside it can be accelerated by a fermi - type process ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
yet another location for particle acceleration is in the pileup region between the x - line and a flanking island , which is where the reconnected magnetic flux accumulates .
there , relativistic speiser motion can combine with curvature drift along the magnetic field gradient to create significant acceleration ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
these mechanisms can operate in reconnection sites where the plasma configuration , involving inflow into the diffusion region and outflow toward the flanking islands , is relatively stationary , or inside a single , autonomously evolving magnetic island . there may be other mechanisms that operate in course of spatial rearrangement and merging of magnetic islands .
converging islands can give rise to fermi - type acceleration as particles bounce between them ( e.g. , * ? ? ?
* ; * ? ? ?
a dynamically active reconnection region containing many interacting islands can also allow fermi - type stochastic particle acceleration @xcite .
it is worth noting that even if particle acceleration is not directly driven by the merging , it is normally most active during island merging episodes in the course of a reconnection event @xcite .
the presence of a guide field has a significant effect on particle acceleration in reconnection @xcite . without a guide field ,
the direct x - line acceleration was typically found to be less efficient than the other mechanisms discussed . however , with a guide field , the effectiveness of acceleration in the pileup region @xcite and inside magnetic islands @xcite was diminished , thus leaving x - line acceleration of particles on speiser orbits as the dominant mechanism .
our results are consistent with this conclusion . to permit an interpretation of the nonthermal radiation spectra in observed high - energy astrophysical sources in terms of synchrotron and inverse - compton radiation , a relatively hard power law particle energy spectrum , which places a significant portion of the total energy in high - energy particles , is required . in most pic simulations including the present work , reconnection produced energized particle populations , but whether the populations represented genuine power laws tails has remained a matter of interpretation .
alternatively , an energized population can be interpreted as a composite of one or more thermal sub - populations , each at a different temperature , for example .
two dimensional simulations typically provide the dynamic range to make a tentative distinction between a thermal or a power law spectrum in an energized population . in most three - dimensional simulations ,
however , such a determination is dubious . in what follows
, we discuss the characteristics of the particle energy spectra in the literature and provide a comparison with our results .
then , we briefly reflect on the expected nature of particle energy spectra produced by systems experiencing magnetic reconnection .
most investigations involving pic simulations of plasma reconnection have interpreted a section of the high energy tail of the particle energy spectrum as a power law @xmath308 . in the x - line region of two dimensional simulations , spectra with power law indices as hard as @xmath309 have been reported @xcite .
the spectrum in the whole simulation box also contains a power law component , but with a softer index of @xmath310 . in their two- and three - dimensional simulations of shock - induced reconnection ,
@xcite find that as long as the region containing a reversing magnetic field that can undergo reconnection is reasonably large compared to the thickness of the reconnection layer , the particle energy spectrum at high energies is a power law with an index @xmath311 over a decade in energy .
other three dimensional investigations have identified relatively soft power law spectra with indices @xmath312 both in the x - line region and in the entire simulation box @xcite .
other investigations have interpreted particle energy spectra in terms of multiple thermal and other exponentially truncated populations ( e.g. , * ? ? ?
* ) . in their two - dimensional simulation of reconnection in an initially nonrelativistic plasma at temperature @xmath313 ,
@xcite find a new thermal population at temperature @xmath314 .
in a three - dimensional simulation beginning with a relativistic plasma at temperature @xmath315 , @xcite find that a thermal population with @xmath316 is produced .
in a recent work presenting two - dimensional simulations without a guide field , @xcite detect a spectrum of the form @xmath317 , where @xmath318 is a constant of the order of unity . among the cited descriptions of particle populations energized by reconnection , the spectra observed in our simulations bear resemblance with those involving multiple thermal sub - populations .
for example , the spectrum at the end of simulation s2k025l can be modeled with three thermal components at temperatures similar to those found in @xcite ; specifically , the two energized sub - populations have temperatures @xmath319 and @xmath320 .
neither a power law nor the @xcite spectral form present a good fit to the spectrum in this simulation .
it is important to note that the continued evolution of the particle energy spectrum we observe after the current sheet has become disordered , and the likely additional evolution that would be taking place in am even larger simulation , mean that the spectral form has not converged .
for example , it is possible that in addition to the two thermal components that we have detected , additional such populations at still higher temperatures would appear in larger simulations , and that the combination of those would constitute a hard spectral tail , e.g. , a power law . in pic simulations
, distinguishing between the various forms of energized particle spectra is difficult due to dynamic range limitations .
another complication is how the spectrum of the non - energized , yet possibly adiabatically heated background plasma is to be subtracted to isolate the genuine nonthermal component .
insight can separately be gained by examining test particle trajectories in magnetic field geometries modeling reconnection layers .
particle ballistics in x - line as well as magnetic island geometries introduced at various levels of approximation has been investigated in numerous studies ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
often , the models entail transport terms describing particle acceleration while confined in a single x - line region or an island , and other terms describing particle escape and the termination of acceleration .
then , the terminal energy spectrum of the escaping particles is obtained by taking the input energy spectrum and deterministically transforming it by the transport terms .
the crudest such models have suggested that the terminal spectrum could be a power law , e.g. , of the form @xmath321 with @xmath322 , where @xmath323 is the reconnected magnetic field and @xmath324 is the electric field in the reconnection region discussed in section [ sec : reconnection_rate ] @xcite .
models taking a more detailed accounting of the transport of particle phase space coordinates and the kinematics of escape , however , can instead imply distinctly non - power - law , softer spectra ( e.g. , * ? ? ?
it is also important to note that the breaking of translational invariance in the direction of the current , e.g. , by three dimensional phase decoherence ( see section [ sec : network ] ) , may limit particle acceleration and prevent the production of the same energized particle tail produced in two dimensional geometries .
power - law spectra are generically expected in acceleration processes entailing stochasticity .
a prime example is the linear diffusive shock acceleration ( dsa ) in which the random outcome of particle scattering in the shock downstream determines whether the particle will return to the shock upstream and be subjected to another acceleration cycle @xcite .
in contrast , the acceleration in an idealized time - independent and two - dimensional x - line region that lacks substructure in the form of secondary and embedded islands , should be deterministic
. therefore the models of @xcite , @xcite , and @xcite , can be thought of as producing a power - law - like spectrum only `` by coincidence . '' realistic reconnection regions should plausibly allow particles to be accelerated in stochastic fashion as they contain time dependence and structure on multiple scales ; they may also have a mechanism for returning particles that have escaped an acceleration site into another such site ( see , e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
since in this work we have found that the reconnection layer transitions into a disordered network of interacting flux ropes , it will be particularly interesting to investigate , in subsequent study , if this network allows for stochastic acceleration of particles in intermittent , secondary reconnection sites that may appear in the course of the flux rope network evolution .
in this paper , we carried out three dimensional pic simulations of magnetic reconnection in a relativistic pair plasma with varying guide field strength .
plasma magnetizations , expressed in terms of the magnetic to kinetic pressure ratio , were of the order of unity .
the initial conditions differed from the usual harris sheet configuration by not having a large density contrast between the center of the current sheet and the background plasma .
we investigated the growth of unstable kinetic modes in the current sheet , as well as the nonlinear development of a three dimensional flux rope network .
we also investigated the character and efficiency of particle energization .
our main results can be summarized as follows : the current sheets in all simulations develop significant magnetic reconnection accompanied with conversion of magnetic to particle kinetic energy . with the aid of fourier decomposition , we ascertained that the linear tearing mode is dominant in the early evolution of the current sheet .
we find that no significant growth occurs in the linear kink and oblique modes .
the nonlinear development of the tearing mode produces a chain of flux ropes separated by primary x - lines .
the flux ropes merge in hierarchical fashion whereby the merging time scale is proportional to the flux rope separation . during this phase magnetic
reconnection takes place at the x - lines .
we find that the dimensionless reconnection rates @xmath325 and the maximum outflow speeds @xmath326 in our simulations are similar to those detected in other three dimensional simulations of reconnection in pair plasmas .
we also find that spatial variation of an off - diagonal component of the pressure tensor is responsible for the breaking of flux freezing at the x - lines , consistent with existing results . while the hierarchical flux rope merging process initially appears similar to that found in two dimensional simulations , in fact it is three - dimensional from the outset .
this is because a lack of initial phase coherence in the linear tearing mode on scales larger than those allowed by causality breaks translational invariance in the direction of the initial current flow .
the flux ropes form a topologically interconnected , dynamically evolving network .
dynamical interaction between neighboring flux ropes is provided by magnetic tension forces . with time
, the flux ropes break up into segments with more isotropic orientations .
the strongly three - dimensional character of the reconnection layer seems to suggest that global reconnection models invoking quasi - two - dimensional plasmoid hierarchies ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) require revision to account for the inter - plasmoid magnetic linkage and isotropization of plasmoid orientations .
the larger flux ropes produced during flux rope merging contain substructure down to plasma skin depth scales which is reflected in embedded , twisted and braided current filaments and sheets .
overall , this substructure is force - free and evolves relatively slowly .
however , isolated sites within the evolved flux rope network contain spatially and temporally intermittent sites characterized by strong nonideal conditions @xmath327 where a change of magnetic connectivity continues to take place even after flux rope merging has saturated on length scales equal to the size of the computational box .
this intermittency may produce the observed variability of nonthermal emission in systems in which the emitting particles are energized by magnetic reconnection . during the early ,
ordered flux rope merging phase , particles are accelerated to high lorentz factors by the electric field in primary x - lines ; the trajectories of these particles are well described by speiser orbits .
particles continue to be energized in the later , disordered phase we identify in our largest simulation , but we leave the analysis of energization in the disordered regime to a subsequent investigation . simulations with higher magnetization and lower guide field strength exhibit greater and faster energy conversion and particle energization .
the efficiency of particle energization measured in terms of the energy in the accelerated particles per unit magnetic energy dissipated in the simulation is an increasing function of the guide field strength , which can be interpreted as resulting from a decreasing plasma compressibility with increasing guide field .
the final particle energy spectrum in the largest simulation is best fit by the inclusion of new thermal components at temperatures @xmath328 , and @xmath329 , in addition to the initial thermal component with temperature @xmath330 .
we , however , acknowledge that a larger size or longer duration simulation is likely to produce a still more pronounced energized component , possibly even a population described with a power law spectrum . energetic positrons ( electrons ) with lorentz factors @xmath254 are moderately beamed in ( opposite to ) the direction of the initial current flow with median inclinations of @xmath279 .
the degree of beaming is determined by a particle s energy gain during acceleration .
we speculate that more highly magnetized plasmas and reconnection sites with larger size x - line regions should give rise to stronger beaming . in this work , we have investigated a narrow range of magnetizations with @xmath331 , but astrophysical reconnection sites can also have high magnetizations @xmath332 .
we can speculate about the applicability of our results in the latter limit .
the linear tearing mode responsible for the initiation of reconnection is insensitive to the degree of magnetization far from the current sheet .
the phase decoherence that produces the initial breakdown of translational invariance is determined by the tearing mode growth time and should thus also persist at high magnetizations .
therefore , we expect the qualitative structure of the reconnection region at higher values of @xmath27 to be similar to that found in our simulations .
the primary effect of high magnetization is that the alfvn velocity approaches the speed of light , which could give rise to ultrarelativistic outflows from the x - line region . in such outflows
the inertial term of the generalized ohm s law becomes important in the breaking of flux freezing @xcite .
this in turn may increase the dimensionless reconnection rate @xmath182 relative to the value found in our simulations .
it remains to be seen whether the associated reconnection process is more or less intermittent .
an increased magnetization is likely to increase the efficiency and the degree of beaming in particle energization .
we consider these results and the immediate questions they raise an incremental step in the development of a multiscale view of collisionless plasma self - organization during magnetic reconnection .
further work is clearly required to place our key finding , the evolution of the simulated , periodic reconnection layer into a disordered network of interacting magnetic flux ropes , in the macroscopic context of a realistic reconnection site characterized by outflow boundary conditions and altogether different field line asymptotics at large distances from the x - line .
it will be particularly interesting to see if the reconnected - flux - carrying outflow from the macroscopic x - line will possess the disordered , interlinked magnetic field topology we observe and investigate what will be the character of magnetic fluctuations in the outflow .
this work was initially inspired by professor pawan kumar s reflections on the physics of prompt emission in gamma - ray burst sources and we would like to thank professor kumar for many stimulating conversations .
d. k. would also like to thank professors richard fitzpatrick and wendell horton for useful discussions .
a. s. is supported by nsf grant ast-0807381 and nasa grants nnx09at95 g and nnx10a039 g . the authors acknowledge the texas advanced computing center ( tacc ) at the university of texas at austin for providing hpc resources that have contributed to the research results reported within this paper .
m. m. and a. s. also gratefully acknowledge support and hospitality from the kavli institute for theoretical physics during their workshop particle acceleration in astrophysical plasmas supported in part by the national science foundation under grant no.phy05-51164 .
we thank the computational information systems laboratory at the national center for atmospheric research for providing the vapor analysis tool @xcite .
chen , l. , bhattacharjee , a. , puhl - quinn , p. a. , yang , h. , bessho , n. , imada , s. , mhlbachler , s. , daly , p. w. , lefebvre , b. , khotyaintsev , y. , vaivads , a. , fazakerley , a. , & georgescu , e. 2008 , nature phys . , 4 , 19 | we investigate magnetic reconnection and particle acceleration in relativistic pair plasmas with three - dimensional particle - in - cell ( pic ) simulations of a kinetic - scale current sheet in a periodic geometry .
we include a guide field that introduces an inclination between the reconnecting field lines and explore outside - of - the - current sheet magnetizations that are significantly below those considered by other authors carrying out similar calculations .
thus our simulations probe the transitional regime in which the magnetic and plasma pressures are of the same order of magnitude .
the tearing instability is the dominant mode in the current sheet for all guide field strengths , while the linear kink mode is less important even without guide field .
oblique modes seem to be suppressed entirely . in its nonlinear evolution ,
the reconnection layer develops a network of interconnected and interacting magnetic flux ropes .
as smaller flux ropes merge into larger ones , the reconnection layer evolves toward a three - dimensional , disordered state in which the resulting flux rope segments contain magnetic substructure on plasma skin depth scales .
embedded in the flux ropes , we detect spatially and temporally intermittent sites of dissipation reflected in peaks in the parallel electric field . magnetic dissipation and particle acceleration
persist until the end of the simulations , with simulations with higher magnetization and lower guide field strength exhibiting greater and faster energy conversion and particle energization . at the end of our largest simulation ,
the particle energy spectrum attains a tail extending to high lorentz factors that is best modeled with a combination of two additional thermal components .
we confirm that the primary energization mechanism is acceleration by the electric field in the x - line region .
the highest energy positrons ( electrons ) are moderately beamed with median angles @xmath0 relative to ( the opposite of ) the direction of the initial current density , but we speculate that reconnection in more highly magnetized plasmas would give rise to stronger beaming .
lastly , we discuss the implications of our results for macroscopic reconnection sites , and which of our results may be expected to hold in systems with higher magnetizations . |
_ computable set theory _ is a research field studying the decidability of the satisfiability problem for collections of set - theoretic formulae ( also called _ syllogistics _ ) .
the main results in computable set theory up to 2001 have been collected in @xcite .
we also mention that the most efficient decision procedures have been implemented in the proof verifier @xcite and form its inferential core .
most of the decidability results established in computable set theory regard one - sorted multi - level syllogistics , namely collections of formulae involving variables of one type only , ranging over the von neumann universe of sets . on the other hand ,
few decidability results have been proved for multi - sorted stratified syllogistics , admitting variables of several types .
this , despite of the fact that in many fields of computer science and mathematics often one deals with multi - sorted languages . an efficient decision procedure for the satisfiability of the two - level syllogistic language ( @xmath3 ) , a fragment admitting variables of two sorts for individuals and sets of individuals , basic set - theoretic operators such as @xmath4 , @xmath5 , @xmath6 , the relators @xmath7 , @xmath8 , @xmath9 , and propositional connectives , has been presented in @xcite .
subsequently , in @xcite , the extension of @xmath3with the singleton operator and the cartesian product operator has been proved decidable .
tarski s and presburger s arithmetics extended with sets have been studied in @xcite .
the three - sorted language @xmath10(three - level syllogistic with singleton , powerset , and general union ) , allowing three types of variables , and the singleton , powerset , and general union operators , in addition to the operators and predicates already contained in @xmath3 , has been proved decidable in @xcite .
more recently , in @xcite , the three - level quantified syllogistic @xmath11 , involving variables of three sorts has been shown to have a decidable satisfiability problem .
later , in @xcite , the satisfiability problem for @xmath12 , a four - level quantified syllogistic admitting variables of four sorts has been proved to be decidable .
the latter result has been exploited in @xcite to prove that @xmath13 , an expressive description logic , has the consistency problem for its knowledge bases decidable . in this paper
we present a decidability result for the satisfiability problem of the set - theoretic language @xmath0 ( three - level quantified syllogistic with finite enumerations and restricted quantifiers ) , which is a three - sorted quantified syllogistic involving _ individual variables _ , _ set variables _ , and _ collection variables _ , ranging respectively over the elements of a given nonempty universe @xmath14 , over the subsets of @xmath14 , and over the collections of subsets of @xmath14 . the language of @xmath0 admits the predicate symbols @xmath7 and @xmath8 and a restricted form of quantification over individual and set variables .
@xmath0 extends the fragment @xmath11presented in @xcite since it admits the finite enumeration operator @xmath1 over individual variables . in spite of its simplicity
, @xmath0 allows one to express several constructs of set theory . among them ,
the most comprehensive one is the set former , which in turn allows one to express other set - theoretic operators like several variants of the powerset and the unordered cartesian product .
we will present two different @xmath0 representations of the latter construct : the first , more straightforward one involves finite enumerations and has linear length in the size of the unordered cartesian product , the second one does not involve finite enumerations , is exponentially longer than the first representation , and is expressible also in @xmath11 .
we will prove that @xmath0 enjoys a small model property by showing how to extract , out of a given model satisfying a @xmath0-formula @xmath2 , another model of @xmath2 but of bounded finite cardinality .
the paper is organized as follows . in section [ genericlanguage ] , we describe the syntax and semantics of a more general language , denoted @xmath15 , which contains @xmath0 as a proper fragment .
subsequently , in section [ machinery ] the machinery needed to prove our main decidability result is provided . in section [ satisfiability ] , the small model property for @xmath0
is established , thus solving the satisfiability problem for @xmath0 .
then , in section [ expressiveness ] , we show how @xmath0 can be used to express several set theoretical operators . finally , in section [ conclusions ] , we draw our conclusions .
we begin by defining the syntax and the semantics of the more general three - level quantified language @xmath15 .
then , in section [ restrictionquant ] , we show how to characterize @xmath0-formulae by suitable restrictions on the usage of quantifiers in formulae of @xmath15 .
the three - level quantified language @xmath15 involves * a collection @xmath16 of _ individual _ or _ sort @xmath17 variables _ , denoted by @xmath18 ; * a collection @xmath19 of _ set _ or _ sort @xmath20 variables _ , denoted by @xmath21 ; * a collection @xmath22 of _ collection _ or _
sort @xmath23 variables _ , denoted by @xmath24 .
in addition to variables @xmath15 involves also _ finite enumerations _ of type @xmath25 , with @xmath26 , @xmath27 .
@xmath15-_quantifier - free atomic formulae _ are classified as : * _ level @xmath17 _ : @xmath28 , @xmath29 , @xmath30 , @xmath31 , where @xmath32 , @xmath33 , @xmath34 , and @xmath35 ; * _ level @xmath20 _ : @xmath36 , @xmath37 , where @xmath38 and @xmath35 .
@xmath15 _ purely universal formulae _ are classified as : * _ level @xmath17 _ : @xmath39 , with @xmath40 a propositional combination of level @xmath17 quantifier - free atoms and @xmath41 variables of sort @xmath17 , with @xmath42 ; , conjunction @xmath43 , disjunction @xmath44 , implication @xmath45 , and biimplication @xmath46 . ]
* _ level @xmath20 _ : @xmath47 , where @xmath48 is a propositional combination of quantifier - free atomic formulae of any level and of purely universal formulae of level @xmath17 , and @xmath49 are variables of sort @xmath20 , with @xmath50 . finally , the _ formulae of @xmath15 _ are all the propositional combinations of quantifier - free atomic formulae and of purely universal formulae of levels @xmath17 and @xmath20 . a _
@xmath15-interpretation _ is a pair @xmath51 , where @xmath14 is any nonempty collection of objects , called the _
domain _ or _ universe _ of @xmath52 , and @xmath53 is an assignment over the variables of @xmath15 such that * @xmath54 , for each individual variable @xmath55 ; * @xmath56 , for each set variable @xmath57 ; * @xmath58 , for all collection variables @xmath35 .
+ ( we recall that @xmath59 denotes the powerset of @xmath60 ) next , let - @xmath61 , @xmath62 , - @xmath63 , @xmath64 . by @xmath65 $ ]
we denote the @xmath15-interpretation @xmath66 such that @xmath67 , for @xmath68 , @xmath69 , for @xmath70 , and which otherwise coincides with @xmath53 on all remaining variables . throughout the paper
we will use the abbreviations : @xmath71[0 ex]{\tiny def}}}}{\ensuremath{\mbox{\boldmath $ \mathcal{m}$}}\xspace}[z_{1}/u_{1},\ldots , z_{n}/u_{n}]$ ] , @xmath72[0 ex]{\tiny def}}}}{\ensuremath{\mbox{\boldmath $ \mathcal{m}$}}\xspace}[z_{1}/u_{1},\ldots , z_{m}/u_{m}]\,,$ ] where the variables @xmath73 and @xmath74 , the individuals @xmath75 , and the subsets @xmath76 are understood from the context .
let @xmath2 be a @xmath15-formula and let @xmath77 be a @xmath15-interpretation .
the notion of _ satisfiability _ for @xmath2 with respect to @xmath52 ( denoted by @xmath78 ) is defined recursively over the structure of @xmath79 .
the evaluation of quantifier - free atomic formulae is carried out as usual according to the standard meaning of the predicates ` @xmath8 ' and ` @xmath7 ' .
purely universal formulae are interpreted as follows : * @xmath80 iff @xmath81 \models \varphi_0 $ ] , + for all @xmath82 ; * @xmath83 iff @xmath84 \models \varphi_1 $ ] , + for all @xmath85 .
finally , compound formulae are evaluated according to the standard rules of propositional logic .
let @xmath2 be a @xmath15-formula . if @xmath78 ( i.e. , @xmath52 _ satisfies _
@xmath2 ) , then @xmath52 is said to be a _
@xmath15-model for @xmath2_. a @xmath15-formula is said to be _
satisfiable _ if it has a @xmath15-model .
a @xmath15-formula is _ valid _ if it is satisfied by all @xmath15-interpretations .
@xmath0 is the collection of the @xmath15-formulae @xmath2 such that , for _ every _ purely universal formula @xmath86 of level 1 occurring in @xmath2 and _ every _ purely universal formula @xmath39 of level 0 occurring in @xmath48 , the condition @xmath87 is a valid @xmath15-formula ( in this case we say that the purely universal formula @xmath39 is _ linked to the variables @xmath88 _ ) .
condition ( [ condition ] ) guarantees that , if a given interpretation assigns to @xmath41 elements of the domain that make @xmath40 false , then all such values must be contained as elements in the intersection of the sets assigned to @xmath49 .
this fact is used in the proof of lemma [ quantifiedformtwo ] to make sure that satisfiability is preserved in the finite model .
as the examples in section [ expressiveness ] will illustrate , condition ( [ condition ] ) is not particularly restrictive .
the following question arises : how one can establish whether a given @xmath15-formula is a @xmath0-formula ?
observe that neither quantification nor collection variables are involved in condition ( [ condition ] ) .
indeed , it turns out that ( [ condition ] ) is a @xmath3-formula and therefore one could use the decision procedures in @xcite to test its validity , as @xmath15 is a conservative extension of @xmath3 .
we mention also that in most cases of interest , as will be shown in detail in section [ expressiveness ] , condition ( [ condition ] ) is just an instance of the simple propositional tautology @xmath89 , and therefore its validity can follow just by inspection .
small models of satisfiable @xmath0-formulae will be expressed in terms of _ relativized interpretations _ with respect to a suitable domain .
[ relintrp ] let @xmath51 be a @xmath15-interpretation and let @xmath90 , @xmath91 , @xmath92 , @xmath93 , and @xmath94 .
the _ relativized interpretation _ @xmath95 of @xmath52 with respect to @xmath96 , @xmath97 , @xmath98 , @xmath99 , and @xmath100 is the interpretation @xmath101 such that @xmath102 for ease of notation , we will often omit the reference to the element @xmath91 and write simply @xmath103 in place of @xmath95 .
our goal is to show that any given satisfiable @xmath0-formula @xmath2 is satisfied by a small model of the form @xmath103 , where @xmath51 is a model of @xmath2 and @xmath96 is a suitable subset of @xmath14 of bounded finite size . at first ,
we state a slightly stronger result for @xmath0-formulae which are propositional combinations of quantifier - free atomic formulae of levels 0 and 1 .
[ wasle_basic ] let @xmath51 and @xmath104 be , respectively , a @xmath15-interpretation and the relativized interpretation of @xmath52 with respect to @xmath105 , @xmath91 , @xmath106 , @xmath107 , and @xmath94 .
furthermore , let @xmath108 be a level @xmath17 quantifier - free atomic formula of the form @xmath28 or @xmath29 , with @xmath109 and @xmath34 , let @xmath110 be a level 0 quantifier - free atomic formula of the form @xmath111 or @xmath112 , with @xmath26 , @xmath34 , @xmath35 , @xmath113 , and let @xmath114 be a level @xmath20 quantifier - free atomic formula of the form @xmath36 or @xmath37 , with @xmath115 , and @xmath116
. then we have : * if @xmath117 , for every @xmath118 in @xmath108 , then @xmath119 iff @xmath120 ; * if ( b1 ) @xmath117 , for every @xmath118 in @xmath108 , ( b2 ) @xmath121 , if @xmath122 and @xmath123 otherwise , for every @xmath124 , and ( b3 ) @xmath121 , for every @xmath125 occurring in @xmath126 such that @xmath127 , then @xmath128 iff @xmath129 ; * if ( c1 ) @xmath121 , if @xmath122 and @xmath123 otherwise , for every @xmath124 , and ( c2 ) @xmath130 , denotes the symmetric difference operator defined by @xmath131 . ] for all @xmath132 such that @xmath133 , then @xmath134 iff @xmath135 .
let us prove case ( a ) first .
assume @xmath136 .
@xmath137 if and only if @xmath138 .
since @xmath117 , by definition [ relintrp ] , @xmath139 and thus @xmath138 if and only if @xmath140 .
since @xmath141 , @xmath140 if and only if @xmath142 .
thus , by definition [ relintrp ] , @xmath142 if and only if @xmath143 , and finally @xmath143 if and only if @xmath144 , as we wished to prove .
next , let @xmath145 .
@xmath146 if and only if @xmath147 . since @xmath148 , by definition [ relintrp ] , @xmath139 and @xmath149 and thus @xmath147 if and only if @xmath150 . finally @xmath150 if and only if @xmath151 , and the thesis follows . for what concerns case ( b ) ,
let us assume first that @xmath152 , with @xmath153 . if @xmath154 , then @xmath155 and , since @xmath113 , @xmath122 and therefore @xmath156 .
moreover @xmath157 and thus , by definition [ relintrp ] , @xmath158 , for @xmath159 .
thus , if @xmath155 , it holds that @xmath160 , and finally that @xmath161 , as we wished to prove .
conversely , assume that @xmath162
. then @xmath163 .
if @xmath122 , @xmath156 , moreover , reasoning as above , @xmath158 , for @xmath159 . thus , if @xmath163 , it holds that @xmath164 , hence @xmath165 and the thesis follows . finally , if @xmath166 , @xmath167 and thus @xmath168 . as a consequence
, it follows that @xmath164 and thus @xmath165 , as we wished to prove .
next , assume that @xmath152 , with @xmath169 . since @xmath156 and @xmath158 , for @xmath159 , @xmath154 if and only if @xmath155 if and only if @xmath160 if and only if @xmath161 .
hence , even in this case the thesis holds .
finally , let @xmath170 . if @xmath171 , then @xmath172 .
in order to show that @xmath173 , we have to prove that @xmath174 . since @xmath157 , by definition [ relintrp ] , @xmath158 , for @xmath159 . thus @xmath175 and @xmath176
. we may have that @xmath177 only in one of the following two cases .
the first case is : @xmath178 and @xmath179 .
this can not occur because in fact @xmath176 .
the other case to be considered is @xmath160 with @xmath180 , for some @xmath153 . if @xmath160 , for some @xmath181 , then @xmath182 and , therefore , @xmath156 .
since @xmath183 , the assumption @xmath180 contradicts the hypothesis that @xmath176 .
hence , we must admit that if @xmath172 , then @xmath174 . on the other hand , if @xmath184 , then @xmath185 .
assume , by contradiction , that @xmath174 .
since @xmath186 , for @xmath159 , @xmath175 and thus , @xmath174 only in the case @xmath187 , for some @xmath188 such that @xmath189 . since @xmath190
, it holds that @xmath191 and thus @xmath192 , and since @xmath193 , we have @xmath172 , absurd .
finally , let us prove case ( c ) .
let @xmath194 . if @xmath195 , then @xmath196 .
thus @xmath197 and , by definition [ relintrp ] , @xmath198 .
since @xmath198 , it immediately follows that @xmath199 . on the other hand , if @xmath200 , then @xmath133 .
thus @xmath130 and , consequently , @xmath201 .
if @xmath201 , by definition [ relintrp ] , @xmath202 and finally @xmath203 .
next , let us assume that @xmath204 . if @xmath205 , then @xmath206 holds trivially . on the other hand , if @xmath180 , but @xmath206 , then either @xmath207 or @xmath208 , for some @xmath209 such that @xmath189 .
in the first case , since @xmath182 , by ( c1 ) it holds that @xmath210 and thus @xmath205 , absurd . in the other case , since @xmath189 it holds that @xmath211 , and thus , by ( c2 ) , @xmath212 .
the latter implies @xmath213 , a contradiction . by propositional logic ,
lemma [ wasle_basic ] implies at once the following corollary .
[ cor_basic ] let @xmath51 and @xmath104 be , respectively , a @xmath15-interpretation and the relativized interpretation of @xmath52 with respect to @xmath105 , @xmath91 , @xmath106 , @xmath107 , and @xmath94 .
furthermore , let @xmath2 be a propositional combination of quantifier - free atomic formulae of the types @xmath214 such that * @xmath117 , for every level @xmath17 variable @xmath215 in @xmath2 ; * @xmath113 ; * @xmath153 , for every level @xmath20 variable @xmath125 in quantifier - free atomic formulae of level @xmath20 ( namely of the form @xmath36 or @xmath37 ) occurring in @xmath2 ; * @xmath156 if @xmath122 and @xmath167 , otherwise , for every level @xmath20 variable @xmath216 ; * @xmath130 , for all @xmath115 such that @xmath133 . * @xmath156 , for every level @xmath20 variable @xmath217 occurring in @xmath2
. then @xmath78 if and only if @xmath218 . the preceding corollary yields at once a small model property for the collection @xmath219of propositional combinations of quantifier - free atomic formulae of the types @xmath214 indeed , let @xmath2 be a satisfiable @xmath219-formula and let @xmath51 be a model for it and let @xmath100 be the maximal length of finite enumerations @xmath25 occurring in @xmath2 .
let @xmath220 and @xmath221 be respectively the collections of variables of sort 0 and of sort 1 occurring in @xmath2 .
* for each pair of variables @xmath222 such that @xmath133 , let us select an element @xmath223 ; * construct a set @xmath224 such that @xmath225 , for every @xmath226 .
then put @xmath227 also , let @xmath97 be an arbitrarily chosen element of @xmath96 .
then , from corollary [ cor_basic ] it follows that the relativized interpretation @xmath228 is a _ small _ model for @xmath2 , as @xmath229 in the symmetric differences @xmath230 can be selected in such a way that @xmath231 holds ( see @xcite ) .
summing up , the following result holds : let @xmath2 be a @xmath219-formula , i.e. , a propositional combination of quantifier - free atomic formulae of the following forms @xmath214 and let @xmath220 and @xmath221 be the collections of variables of sort 0 and of sort 1 occurring in @xmath2 , respectively
. then @xmath2 is satisfiable if and only if is satisfied by a @xmath15-interpretation @xmath51 such that @xmath231 .
since the @xmath15-interpretations over a bounded domain are finitely many and can be effectively generated , the decidability of the satisfiability problem for @xmath219-formulae follows . to state the main results on quantified formulae , namely that the relativized interpretation @xmath232 of a model @xmath51 for a purely universal @xmath0-formula @xmath2 of level @xmath17 or @xmath20 also satisfies @xmath2 under suitable conditions on @xmath96 , @xmath233 , @xmath234 , and @xmath100 ( lemmas [ quantifiedformone ] and [ quantifiedformtwo ] below )
, it is convenient to introduce the following abbreviations : @xmath235[0 ex]{\tiny def } } } } & { \mbox{\rm rel}}({{\ensuremath{\mbox{\boldmath $ \mathcal{m}$}}\xspace}^{\vec{z}}},d^{*},d^*,\mathcal{v}'_{0},\mathcal{v}'_{1},l ) \\ { { \ensuremath{\mbox{\boldmath $ \mathcal{m}$}}\xspace}^{*,\vec{z } } } & { \mathrel{\makebox{\:= \hspace{-.2 cm } \raisebox{-0.5 ex}[0 ex][0 ex]{\tiny def } } } } & { { \ensuremath{\mbox{\boldmath $ \mathcal{m}$}}\xspace}^{*}}[z_{1}/u_{1},\ldots , z_{n}/u_{n } ] \\ { { \ensuremath{\mbox{\boldmath $ \mathcal{m}$}}\xspace}^{\vec{z } , * } } & { \mathrel{\makebox{\:= \hspace{-.2 cm } \raisebox{-0.5 ex}[0 ex][0 ex]{\tiny def } } } } & { \mbox{\rm rel}}({{\ensuremath{\mbox{\boldmath $ \mathcal{m}$}}\xspace}^{\vec{z}}},d^{*},d^*,\mathcal{v}'_{0},\mathcal{v}'_{1 } , l ) \\ { { \ensuremath{\mbox{\boldmath $ \mathcal{m}$}}\xspace}^{*,\vec{z } } } & { \mathrel{\makebox{\:= \hspace{-.2 cm } \raisebox{-0.5 ex}[0 ex][0 ex]{\tiny def } } } } & { { \ensuremath{\mbox{\boldmath $ \mathcal{m}$}}\xspace}^{*}}[z_{1}/u_{1},\ldots , z_{m}/u_{m}]\ , , \ ] ] with @xmath236 , @xmath237 , @xmath238 , @xmath239 . when @xmath240 , the @xmath15-interpretations @xmath241 and @xmath242 coincide , as stated in the following lemma , whose proof is routine and is omitted for brevity .
[ le_m*zmz * ] let @xmath51 be a @xmath15-interpretation , @xmath105 , @xmath240 , and @xmath243 .
then the @xmath15-interpretations @xmath241 and @xmath242 coincide . likewise , under some conditions
, the @xmath15-interpretations @xmath244 and @xmath245 coincide too , as stated in the following lemma . [ le_eqm*zmz * ] let @xmath51 be a @xmath15-interpretation , @xmath105 , @xmath93 , @xmath237 , and @xmath246 .
then the @xmath15-interpretations @xmath244 and @xmath245 coincide .
we are now ready to prove the main result of the present section , namely that if @xmath51 satisfies a purely universal @xmath0-formula @xmath2 of level @xmath17 or @xmath20 , then , under suitable conditions , the relativized interpretation @xmath232 of @xmath52 satisfies @xmath2 too .
this will be done in the following two lemmas .
[ quantifiedformone ] let @xmath51 be a @xmath15-interpretation , @xmath247 , @xmath91 , @xmath92 , @xmath93 , @xmath94 , and let @xmath248 be such that @xmath121 , if @xmath122 and @xmath123 otherwise , for every @xmath124 .
furthermore , let @xmath249 be a purely universal @xmath0-formula of level @xmath17 such that * @xmath250 , for every @xmath251 occurring free in it ; * each occurrence of finite enumeration @xmath25 in @xmath2 , with @xmath252 , for every @xmath253 , is such that @xmath113 ; * @xmath254 ; * @xmath156 , for every variable @xmath125 of level @xmath20 in @xmath2 such that @xmath255 .
then @xmath256 let @xmath52 and @xmath257 be as in the lemma , and assume that @xmath80 whereas @xmath258 .
then there must exist @xmath259 such that @xmath260 \not\models \varphi_0 $ ] , i.e. , @xmath261 . since , by ( iii ) , @xmath254 , by lemma [ le_m*zmz * ] , @xmath262 . by ( i ) and by the definition of @xmath263 , it is easy to see that @xmath264 , for every @xmath265 occurring in @xmath266 .
moreover , by ( ii ) each occurrence of finite enumeration @xmath25 in @xmath266 , with @xmath252 , for every @xmath253 , is such that @xmath113 . finally ,
since @xmath267 and @xmath268 , for every variable @xmath34 , it can be checked that * @xmath269 , for every variable @xmath125 of level @xmath20 occurring in @xmath2 such that @xmath255 ( by ( iv ) ) , and * @xmath270 , if @xmath271 and @xmath272 otherwise , for every @xmath124 ( because @xmath121 , if @xmath122 and @xmath123 otherwise , for every @xmath124 ) .
thus , by lemma [ wasle_basic ] ( a ) and ( b ) we have @xmath273 , which yields @xmath274 , a contradiction . [ quantifiedformtwo ]
let @xmath51 be a @xmath15-interpretation , @xmath105 , @xmath91 , @xmath92 , @xmath93 , @xmath94 , @xmath275 , and let @xmath47 be a purely universal @xmath0-formula of level @xmath20 such that 1 .
[ new_i ] @xmath276 ; 2 .
[ old_ii ] @xmath277 , for every variable @xmath278 occurring free in @xmath279 ; 3 .
[ old_i ] @xmath250 , for every @xmath251 occurring free in @xmath47 ; 4 .
[ newtup_i ] @xmath121 , if @xmath122 and @xmath123 otherwise , for every @xmath124 ; 5 .
[ old_iii ] @xmath130 , for all @xmath280 such that @xmath133 ; 6 .
[ newtup_ii ] each occurrence of finite enumeration @xmath25 in @xmath48 , with @xmath252 , for every @xmath253 , is such that @xmath113 ; 7 .
[ old_iv ] for every purely universal formula @xmath281 of level @xmath17 occurring in @xmath282 and variables @xmath283 such that @xmath284 , there are @xmath259 such that @xmath285\not\models ( \varphi_0)_{x_1,\ldots , x_m}^{z_1,\ldots , z_m}$ ] ; and variables @xmath286 , by @xmath287 we mean the formula obtained by simultaneously substituting @xmath49 with @xmath288 in @xmath2 . ]
[ newtup_iii ] for every purely universal formula @xmath281 of level @xmath17 occurring in @xmath282 , @xmath254 .
then @xmath289 let @xmath52 , @xmath257 , and @xmath290 be as in the lemma , and assume that @xmath291 whereas @xmath292 .
then there must exist @xmath293 such that @xmath294 \not\models \varphi_1 $ ] , i.e. , @xmath295 without loss of generality , we may assume that there exists @xmath296 such that * @xmath297 , for @xmath298 , for some variables @xmath299 in @xmath300 , and * @xmath301 , for all @xmath302 .
let @xmath303[0 ex]{\tiny def}}}}(\varphi_{1})^{z_{1}\ldots z_{h}}_{x_{1}\ldots x_{h}}$ ] ( i.e. , @xmath304 is the formula obtained by simultaneously substituting @xmath305 with @xmath299 in @xmath48 ) and let @xmath306[0 ex]{\tiny def}}}}{\ensuremath{\mbox{\boldmath $ \mathcal{m}$}}\xspace}[z_{h+1}/u_{h+1},\ldots , z_{m}/u_m]\,.\ ] ] our plan is to show that @xmath307 holds .
then , since ( [ modelzm ] ) readily implies @xmath308 where @xmath309[0 ex]{\tiny def}}}}{\ensuremath{\mbox{\boldmath $ \mathcal{m}$}}\xspace}[z_{1}/mx_{1},\ldots , z_{h}/mx_{h } , z_{h+1}/u_{h+1},\ldots , z_{m}/u_{m}]\,,\ ] ] and ( [ eq_five ] ) in its turn yields @xmath310 , a contradiction would be derived , proving that @xmath311 implies @xmath312 ( and hence completing the proof of the lemma ) .
thus , in what follows we will just show that ( [ eq_mstarz ] ) implies ( [ modelzm ] ) .
to begin with , let @xmath313[0 ex]{\tiny def}}}}{{\ensuremath{\mbox{\boldmath $ \mathcal{m}$}}\xspace}^{*}}[z_{h+1}/u_{h+1},\ldots , z_{m}/u_m]$ ] .
plainly , ( [ eq_mstarz ] ) implies at once @xmath314 . since , by hypothesis [ new_i ] of the lemma and by lemma [ le_eqm*zmz * ] , @xmath315 and @xmath316 coincide , so that @xmath317 holds , to prove ( [ modelzm ] ) it will be enough , by propositional logic , to show that @xmath316 and @xmath318 coincide on all propositional componentsby definition , a formula @xmath2 of @xmath15 is a propositional combination of certain atomic formulae of level @xmath17 , @xmath20 , and @xmath23 .
these are the propositional components of @xmath2 .
] of @xmath304 , which is what we do next . by hypotheses
[ old_ii ] , [ old_i ] , [ newtup_i ] , [ old_iii ] , and [ newtup_ii ] of the lemma and by lemma [ wasle_basic ] , @xmath316 and @xmath318 coincide on all propositional components of @xmath304 of any of the following types : * @xmath28 , @xmath29 ( with @xmath109 and @xmath319 ) , * @xmath111 , @xmath112 ( with @xmath26 , @xmath34 , and @xmath35 ) , and * @xmath36 , @xmath37 ( with @xmath115 and @xmath35 ) . thus , to complete the proof ,
we are only left with showing that @xmath316 and @xmath318 coincide also on the propositional components of @xmath304 of the remaining types , namely those of the form : * @xmath320 , @xmath321 , @xmath322 + ( with @xmath323 , @xmath35 , and @xmath302 ) , and * level @xmath17 purely universal formulae . for propositional components of @xmath304 of type @xmath320 ( with @xmath324 and @xmath302 ) , we have : @xmath325 analogously , for propositional components of @xmath304 of type @xmath321 , with @xmath326 and @xmath302 . for propositional components of @xmath304 of type @xmath327 ( with @xmath35 and @xmath328 )
, we have : @xmath329 finally , let @xmath39 be a propositional component of @xmath282 and let @xmath330[0 ex]{\tiny def}}}}(\varphi_0)_{x_1,\ldots , x_h}^{z_1,\ldots , z_h}$ ] .
we show that @xmath331 let us first assume that @xmath332 but , by way of contradiction , that @xmath333 we will distinguish two cases , according to whether @xmath334 ( i.e. , @xmath335 ) or @xmath336 ( i.e. , @xmath337 ) .
* case @xmath338 : * from @xmath339 , it follows that there exist @xmath340 such that @xmath341 \not\models \bar\varphi_0 $ ] .
let us put @xmath342[0 ex]{\tiny def}}}}{{\ensuremath{\mbox{\boldmath $ \mathcal{m}$}}\xspace}^{\vec{z}^{-}}}[z_{1}/u_{1},\ldots , z_{n}/u_{n}]$ ] .
then we have @xmath343 recalling that by definition of @xmath0-formulae ( cf.section [ restrictionquant ] ) the formula @xmath344 must be linked to the variables @xmath345 , then we have @xmath346 ( cf . condition ( [ condition ] ) ) , so that @xmath347{0pt}{8mm}_{x_{1},\ldots , x_{h}}^{z_{1},\ldots , z_{h}}\,,\ ] ] i.e. , @xmath348 thus , by ( [ temp1 ] ) , @xmath349 so that , for @xmath350 , @xmath351 therefore , for @xmath350 , @xmath352 in view of ( [ eq_uiindstar ] ) and by conditions [ old_ii ] , [ old_i ] , [ newtup_i ] , [ old_iii ] , [ newtup_ii ] , and [ newtup_iii ] of this lemma , we can apply corollary [ cor_basic ] and lemma [ le_m*zmz * ] in the deductions which follow : @xmath353 & & \hfill\text{{\small present lemma , and corollary~\ref{cor_basic } ) } } \\[4pt ] & \longrightarrow & ( { { \ensuremath{\mbox{\boldmath $ \mathcal{m}$}}\xspace}^{\vec{z}^{-}}})^{*,\vec{z } } \models \neg \bar\varphi_0 \hfill\text{{\small ( from ( \ref{eq_uiindstar } ) , condition \ref{newtup_iii } , } } \\[4pt ] & & \hfill\text{{\small and lemma \ref{le_m*zmz * } ) } } \\[4pt ] & \longrightarrow & { { \ensuremath{\mbox{\boldmath $ \mathcal{m}$}}\xspace}^{\vec{z}^{-},*}}\not\models ( \forall z_{1})\ldots(\forall z_{n } ) \bar\varphi_0 \ , .
\end{array}\ ] ] hence @xmath354 holds , contradicting our initial assumption ( [ initialassumption ] ) and therefore proving that the case @xmath355 can not arise .
* case @xmath356 : * when @xmath336 , the interpretations @xmath318 and @xmath316 are just @xmath52 and @xmath257 , respectively .
thus , our contradictory assumption ( [ contradictoryassumption ] ) becomes @xmath357 , which , by condition [ old_iv ] of the lemma , implies the existence of elements @xmath358 such that @xmath359 \not\models \bar\varphi_0 $ ] , i.e. , @xmath360 .
but , @xmath361 & \longrightarrow&{{\ensuremath{\mbox{\boldmath $ \mathcal{m}$}}\xspace}^{*,\vec{z}}}\not\models \bar\varphi_0 \hfill \text{{\small ( from ( \ref{uindstar } ) , condition \ref{newtup_iii } , and lemma \ref{le_m*zmz * } ) } } \\[4pt ] & \longrightarrow&{{\ensuremath{\mbox{\boldmath $ \mathcal{m}$}}\xspace}^{*}}\not\models ( \forall z_{1})\ldots(\forall z_{n } ) \bar\varphi_0 \ , .
\end{array}\ ] ] therefore @xmath354 , which contradicts our assumption ( [ initialassumption ] ) .
thus , even the current case @xmath336 can not arise . since in any case
we get a contradiction , we have the following implication : @xmath362 to complete the proof of ( [ eqequivalence ] ) , we need to establish also the converse implication .
but this follows at once , by observing that if @xmath363 , then by conditions [ old_i ] , [ newtup_ii ] , and [ newtup_iii ] of the lemma and by lemma [ quantifiedformone ] we have @xmath364 .
this concludes the proof of the lemma .
we will solve the satisfiability problem for @xmath0 , i.e. , the problem of establishing for any given formula of @xmath0 whether it is satisfiable or not , as follows : * firstly , we will reduce effectively the satisfiability problem for @xmath0-formulae to the same problem for normalized @xmath0-conjunctions ( these will be defined precisely below ) ; * secondly , we will prove that the collection of normalized @xmath0-conjunctions enjoys a small model property . from ( a ) and ( b ) , the solvability of the satisfiability problem for @xmath0 will follow immediately .
in fact , by further elaborating on point ( a ) , it could easily be shown that the whole collection of @xmath0-formulae enjoys a small model property .
let @xmath2 be a formula of @xmath0 and let @xmath365 be a disjunctive normal form of @xmath2 .
we observe that the disjuncts of @xmath365 are conjunctions of @xmath0-literals , namely quantifier - free atomic formulae of levels @xmath17 and @xmath20 , or their negations , and of purely universal formulae of levels @xmath17 and @xmath20 , or their negations , satisfying the linkedness condition ( [ condition ] ) . by a suitable renaming of variables
, we can assume that no bound variable can occur in more than one quantifier in the same disjunct of @xmath365 and that no variable can have both bound and free occurrences in the same disjunct . without disrupting satisfiability
, we replace negative literals of the form @xmath366 and @xmath367 occurring in @xmath365 by their negated matrices @xmath368 and @xmath369 , respectively , since for any given @xmath370-interpretation @xmath77 one has @xmath371 if and only if @xmath81 \models \neg \varphi_0 $ ] , for some @xmath82 , and , likewise , @xmath372 if and only if @xmath84 \models \neg \varphi_1 $ ] , for some @xmath373 .
then , if needed , we bring back the resulting formula into disjunctive normal form , eliminate as above the residual negative literals of the form @xmath374 which might have been introduced by the previous elimination of negative literals of the form @xmath375 from @xmath365 , and transform again the resulting formula in disjunctive normal form .
let @xmath376 be the formula so obtained .
observe that all the above steps preserve satisfiability , so that our initial formula @xmath2 is satisfiable if so is @xmath376 . in addition , the formula @xmath376 is satisfiable if and only if so is at least one of its disjuncts .
it is an easy matter to check the each disjunct of @xmath376 is a conjunction of @xmath0-literals of the following types : @xmath377{cccc } x = y\ , , & \quad x \in x\ , , & \quad \{x_1,\ldots , x_k\ } = x\ , , & \quad \{x_1,\ldots , x_k\ } \in a\,,\\ \neg ( x = y)\ , , & \quad \neg ( x \in x)\ , , & \quad \neg(\{x_1,\ldots , x_k\ } = x)\ , , & \quad \neg(\{x_1,\ldots , x_k\ } \in a)\,,\\ x = y\ , , & \quad x \in a\ , , & \quad\neg(x = y)\ , , & \quad \neg(x \in a)\ , , \end{array } \label{was1}\tag{i}\ ] ] where @xmath32 , @xmath38 , and @xmath378 ; @xmath379 where @xmath380 and @xmath40 is a propositional combination of quantifier - free level @xmath17 atoms ; and @xmath381 where @xmath382 and @xmath48 is a propositional combination of quantifier - free atomic formulae of any level and of purely universal formulae of level @xmath17 , where the propositional components in @xmath282 of type @xmath39 are linked to the bound variables @xmath88 .
we call such formulae _
normalized @xmath0-conjunctions_. the above discussion can then be summarized in the following lemma .
the satisfiability problem for @xmath0-formulae can be effectively reduced to the satisfiability problem for @xmath0-conjunctions .
let @xmath2 be a normalized @xmath0-conjunction and assume that @xmath77 is a model for @xmath2 .
we show how to construct , out of @xmath52 , a finite `` small '' @xmath15-interpretation @xmath383 which is a model of @xmath2 .
we proceed as follows .
first we outline a procedure to build a nonempty finite universe @xmath384 whose size depends solely on @xmath2 and can be computed _ a priori_. then , following definition [ relintrp ] , we construct a relativized @xmath15-interpretation @xmath383 with respect to suitable collections @xmath385 and @xmath386 of variables , and to a positive number @xmath100 , and show that @xmath257 satisfies @xmath2 .
let @xmath388 , @xmath389 , and @xmath390 be the collections of the variables of sort @xmath17 , @xmath20 , and @xmath23 occurring in @xmath2 , respectivelyand , an let @xmath391 be smallest number such that @xmath392 , for every finite enumeration @xmath25 occurring in @xmath2 .
we compute @xmath387 by means of the procedure below .
let @xmath393 be the conjuncts of @xmath2 of the form ( [ was3 ] ) . to each such conjunct @xmath394 , we associate the collection @xmath395 of the propositional components of its matrix @xmath396 and call the variables @xmath397 the _ arguments of _ @xmath395 .
then we put @xmath398[0 ex]{\tiny def}}}}\ { \varphi_{ij } : 1 { \leqslant}i { \leqslant}h \hbox { and } 1 { \leqslant}j { \leqslant}\ell_i \}.\ ] ] by applying the procedure _ distinguish _ described in @xcite to the collection @xmath399 , it is possible to construct a set @xmath400 such that * @xmath401 , for all @xmath402 such that @xmath133 , and * @xmath403 .
next , we construct a set @xmath224 satisfying that @xmath404 , for every @xmath405 .
plainly , we can assume that @xmath406 .
then , after initializing @xmath387 with the set @xmath407 , for each @xmath408 of the form @xmath39 having @xmath49 as arguments and for each ordered @xmath409-tuple @xmath410 of variables in @xmath389 such that @xmath411 , we insert in @xmath387 elements @xmath412 such that @xmath285 \not\models ( { \varphi_0})_{x_{i_1},\ldots , x_{i_m}}^{z_1\;,\ldots,\:z_m}$ ] . from the previous construction it follows easily that @xmath413 where @xmath53 and @xmath414 are , respectauto.ively , the maximal number of quantifiers in purely universal formulae of level @xmath20 occurring in @xmath415 and the maximal number of quantifiers in purely universal formulae of level @xmath17 occurring in purely universal formulae of level @xmath20 in @xmath415 .
thus , in general , the domain of the small model @xmath387 is exponential in the size of the input formula @xmath2 .
let us put @xmath416 .
we have to show that , if @xmath78 , then @xmath417 .
[ correctness ] let @xmath52 be a @xmath15-interpretation satisfying a normalized @xmath0-conjunction @xmath2 .
further , let @xmath418 be the @xmath15-interpretation defined according to definition [ relintrp ] , where @xmath387 is constructed as above , @xmath388 and @xmath389 are the collections of variables of levels @xmath17 and @xmath20 occurring in @xmath2 , respectively , and @xmath391 is the smallest number such that @xmath392 , for every finite enumeration @xmath25 of level @xmath17 variables occurring in @xmath2 . then @xmath218 .
we have to prove that @xmath419 , for every conjunct @xmath420 in @xmath2 .
each conjunct @xmath420 is of one of the three types ( [ was1 ] ) , ( [ was2 ] ) , and ( [ was3 ] ) introduced in section [ normal3lqs ] . by applying lemmas [ wasle_basic ] , [ quantifiedformone ] , or [ quantifiedformtwo ] to every @xmath420 in @xmath2 ( according to the type of @xmath420 )
we obtain the thesis .
notice that the hypotheses of lemmas [ wasle_basic ] , [ quantifiedformone ] , and [ quantifiedformtwo ] are fulfilled by the construction of @xmath387 outlined above . indeed , 1 .
@xmath421 ; 2 .
@xmath422 , for every variable @xmath278 occurring free in @xmath2 ; 3 .
@xmath250 , for every @xmath251 occurring free in @xmath2 ; 4 .
@xmath121 , if @xmath423 and @xmath424 otherwise , for every @xmath425 ; 5 .
@xmath130 , for all @xmath426 such that @xmath133 ; 6 .
each occurrence of finite enumeration @xmath25 in @xmath2 , with @xmath252 , for every @xmath253 , is such that @xmath392 ; 7 . for every purely universal formula @xmath281 of level @xmath17 occurring in a purely universal formula of level @xmath20 , and variables
@xmath427 such that @xmath284 , there are @xmath259 such that @xmath285\not\models ( \varphi_0)_{x_1,\ldots , x_m}^{z_1,\ldots , z_m}$ ] ; 8 . for every purely universal formula @xmath281 of level @xmath17 occurring in @xmath282 , @xmath428 . from the above reduction and relativization steps , it is not hard to derive the following result : the fragment @xmath0 enjoys a small model property ( and therefore its satisfiability problem is solvable ) .
reasoning as in @xcite , it is possible to define a class of subtheories @xmath429 of @xmath0 , whose formulae have quantifier prefixes of length bounded by the constant @xmath430 and satisfy certain syntactic constraints , having an -complete satisfiability problem .
such subtheories are quite expressive , in fact several set - theoretic constructs treated in section [ expressiveness ] such as , for instance , some variants of the powerset operator can be represented in them .
moreover , it can be shown that the modal logic can be represented in @xmath431 .
several constructs of elementary set theory are easily expressible within the language @xmath0 . in particular , it is possible to express with @xmath0-formulae a restricted variant of the set former , which in turn allows one to express other significant set operators such as binary union , intersection , set difference , set complementation , the powerset operator and some of its variants , etc .
more specifically , a set former of the form @xmath432 can be expressed in @xmath0 by the formula @xmath433 ( in which case it is called an _ admissible set former of level @xmath17 for _ @xmath434 ) provided that after transforming it into prenex normal form one obtains a formula satisfying the syntactic constraints of @xmath0 .
this , in particular , is always the case whenever @xmath435 is a quantifier - free formula of @xmath0 . in [ tablesetformers1 ] some examples of formulae expressible by admissible set formers of level @xmath17 for @xmath434are reported , where @xmath436 and @xmath437 stand respectively for the empty set and for the domain of the discourse , and @xmath438 is the complementation operator with respect to the domain of the discourse .
the formulae in the first column of [ tablesetformers1 ] are the allowed atoms in the fragment @xmath3(two - level syllogistic ) which has been proved decidable in @xcite .
since @xmath111 is a level @xmath17 quantifier - free atomic formula in @xmath434 , @xmath3with finite enumerations turns out to be expressible by @xmath434-formulae .
c|c & _ admissible set formers for @xmath434of level @xmath17 _ + + @xmath439 & @xmath440 + @xmath441 & @xmath442 + @xmath443 & @xmath444 + @xmath445 & @xmath446 + @xmath447 & @xmath448 + @xmath449 & @xmath450 + in addition to the formulae in [ tablesetformers1 ] the following literals latexmath:[\[\label{furtherliterals } z \subseteq x\ , , \quad |z| { \leqslant}h\ , , \quad |z| < h + 1\ , , \quad @xmath434-formulae of level @xmath17 , where @xmath452 denotes the cardinality operator and @xmath453 stands for a nonnegative integer constant ( cf . [ tablesetformers2 ] ) .
in fact , it turns out that all literals ( [ furtherliterals ] ) can be expressed by level @xmath17 purely universal @xmath434-formulae which are linked to the variable @xmath454 , so that they can freely be used in the matrix @xmath455 of a level @xmath20 universal formula of the form @xmath456 .
let us consider , for instance , the formula @xmath457 which expresses the literal @xmath458 .
the linkedness condition for it relative to the variable @xmath454 is @xmath459 which is plainly a valid @xmath434-formula since it is an instance of the propositional tautology @xmath460 , showing that ( [ zkey ] ) is linked to the variable @xmath454 . similarly , one can show that the remaining formulae in ( [ furtherliterals ] ) can also be expressed by level @xmath17 purely universal @xmath434-formulae which are linked to the variable @xmath454 .
c|c & @xmath0-formulae + + @xmath461 & @xmath462 + @xmath458 & @xmath463 + @xmath464 & @xmath458 + @xmath465 & @xmath466 + @xmath467 & @xmath468 + @xmath469 & @xmath470 + similar remarks apply also to the set former of the form @xmath471 .
this can be expressed by the @xmath0-formula @xmath472 ( in which case it is called an _ admissible set former of level @xmath20 for _ @xmath434 ) provided that @xmath455 does not contain any quantifier over variables of sort @xmath20 , and all quantified variables of sort @xmath17 in @xmath455 are linked to the variable @xmath454 as specified in condition ( [ condition ] ) .
c|c & _ admissible set formers of level @xmath20 for @xmath434 _ + + @xmath473 & @xmath474 + @xmath475 & @xmath476 + @xmath477 & @xmath478 + @xmath479 & @xmath480 + @xmath481 & @xmath482 + @xmath483 & @xmath484 + @xmath485 & @xmath486 + @xmath487 & @xmath488 + @xmath489 & @xmath490 + @xmath491 & @xmath492 + @xmath493 & @xmath494 + @xmath495 & @xmath490 + @xmath496 & @xmath496 + some examples of formulae expressible by admissible set formers of level @xmath20 for @xmath434are reported in [ tablesetformers3 ] . in this case
the symbol @xmath437 stands for the powerset of the domain of the discourse .
the meaning of the overloaded symbol @xmath437 can always be correctly disambiguated from the context . in view of the fact
that , as already remarked , the literals ( [ furtherliterals ] ) can be expressed by level @xmath17 purely universal @xmath434-formulae which are linked to the variable @xmath454 , it follows that all set formers in [ tablesetformers3 ] are indeed admissible .
the propositional combination of the following literals @xmath497 present in the first column of [ tablesetformers3 ] form the proper fragment @xmath498of the theory @xmath10(three - level syllogistic with singleton , powerset , and unionset ) whose decision problem has been solved in @xcite .
we recall that in addition to the formulae in ( [ 3lssp ] ) , @xmath10involves also unionset clauses of the form @xmath499 , with @xmath125 a variable of sort @xmath20 and @xmath500 a variable of sort @xmath23 , which , however , are not expressible by @xmath434-formulae . besides the ordinary powerset operator ,
@xmath434-formulae allow one also to express the variants @xmath501 , @xmath502 , and @xmath503 reported in [ tablesetformers3 ] , which denote respectively the collection of all the subsets of @xmath125 with at most @xmath453 distinct elements , with exactly @xmath453 elements , and with at least @xmath453 distinct elements .
it is interesting to observe that the satisfiability problem for the propositional combination of literals of the forms @xmath504 , @xmath505 , @xmath506 , @xmath507 , involving also one occurrence of literals of the form @xmath508 , has been proved to be decidable in @xcite , when sets are interpreted in the standard von neumann hierarchy ( cf .
@xcite ) . another interesting variant of the powerset operator is the @xmath509 operator introduced in @xcite in the solution to the satisfiability problem for the extension of @xmath510with the powerset and singleton operators .
we recall that given sets @xmath511 , @xmath512 denotes the collection of all subsets of @xmath513 which have nonempty intersection with each set @xmath514 , for @xmath515 . in symbols , @xmath516[0 ex]{\tiny def } } } } & \displaystyle\left\{z : z \subseteq \bigcup_{i=1}^k x_i \wedge \bigwedge_{i=1}^{k } z \cap x_i \neq \emptyset \right\ } \\ & \!\!\!\!\!= & \displaystyle\left\{z : z \subseteq \bigcup_{i=1}^k x_i \wedge \bigwedge_{i=1}^{k } \neg ( z \subseteq \overline x_i ) \right\}\ , .
\end{array}\ ] ] from the latter expression , it readily follows that the literal @xmath517 can be expressed by a @xmath434-formula .
@xmath518 , the unordered cartesian product @xmath519 is the set @xmath520[0 ex]{\tiny def}}}}\big\{\{x_{1},\ldots , x_{n}\ } : x_{1 } \in x_{1 } , \ldots , x_{n } \in x_{n}\big\}\,.\ ] ] then , the literal @xmath521 where @xmath500 stands for a variable of level 2 and @xmath518 here stand for variables of level 1 , can be expressed by the @xmath0-formula @xmath522 in what follows , we show that ( [ not3lqsr ] ) can be expressed without making use of the finite enumeration operator .
when the sets @xmath518 are pairwise disjoint or , on the opposite side , when they all coincide , we can readily express the literal ( [ ucp ] ) by a @xmath11-formula .
for instance , if the sets @xmath518 are pairwise disjoint , then @xmath523 if and only if 1 . [ first ] @xmath524 , and 2 . [ second ] there exist @xmath525 such that @xmath526 , , @xmath527 . the above conditions can be used to express the literal ( [ ucp ] ) by the following @xmath11-formula @xmath528 as is easy to check , where @xmath529[0 ex]{\tiny def } } } } & |z| { \leqslant}n \wedge |z| { \geqslant}n \\ |z| { \leqslant}n & { \mathrel{\makebox{\ : $ \equiv$ \hspace{-.2 cm } \raisebox{-0.5 ex}[0 ex][0 ex]{\tiny def } } } } & ( \forall x_{1})\ldots(\forall x_{n+1})\bigg(\displaystyle\bigwedge_{i=1}^{n+1 } x_{i } \in z
\rightarrow \bigvee_{1 { \leqslant}i < j { \leqslant}n+1 } x_{i } = x_{j}\bigg ) \\
|z| { \geqslant}n & { \mathrel{\makebox{\ : $ \equiv$ \hspace{-.2 cm } \raisebox{-0.5 ex}[0 ex][0 ex]{\tiny def } } } } & \neg(|z| { \leqslant}n-1 ) \end{array } $ ] ( notice that @xmath530 is linked to the variable @xmath454 ) . when @xmath531 , then @xmath532 if and only if @xmath533 thus , in this particular case , the literal ( [ ucp ] ) can be expressed by the @xmath11-formula @xmath534 however , if we make no assumption on the sets @xmath518 , in order to characterize the sets @xmath454 belonging to @xmath535 by a @xmath11-formula , we have to consider separately the cases in which @xmath524 , @xmath536 , etc . , listing explicitly , for each of them , all the allowed membership configurations of the members of @xmath454 .
for instance , if @xmath537 , we have @xmath538 if and only if * @xmath539 and there exist distinct @xmath540 and @xmath541 s. t. @xmath542 ; or * @xmath543 and the intersection @xmath544 is nonempty .
thus the following @xmath11-formula expresses the literal @xmath545 : @xmath546 likewise , in the case @xmath547 , we have @xmath548 if and only if * @xmath549 and there exist pairwise distinct @xmath540 , @xmath541 , and @xmath550 such that @xmath551 ; or * @xmath539 and there exist distinct @xmath552 and @xmath553 such that either * * @xmath554 and @xmath555 , or * * @xmath556 and @xmath541 , or * * @xmath557 and @xmath558 , + and such that @xmath542 ; or * @xmath543 and the intersection @xmath559 is nonempty .
[ characterization ] let @xmath518 be given sets
. then @xmath523 if and only there exists a partition @xmath560 of the set @xmath561 and a bijection @xmath562 such that @xmath563 let @xmath523
. then @xmath564 , for some @xmath565 . for @xmath566
, let us put @xmath567[0 ex]{\tiny def}}}}\{i : x = x_{i}\}\,.\ ] ] then it is an easy matter to check that @xmath568[0 ex]{\tiny def}}}}\{\sigma(x ) : x \in x\}$ ] is a partition of @xmath561 and @xmath569 is a bijection from @xmath454 into @xmath560 which satisfies ( [ condition1 ] ) .
conversely , assume that @xmath562 is a bijection satisfying ( [ condition1 ] ) , for a partition @xmath560 of @xmath561 and a set @xmath454 , and put @xmath570[0 ex]{\tiny def}}}}\sigma^{-1}(p_{i})\,,\ ] ] where @xmath571 is the block of @xmath560 containing @xmath572 .
then it plainly follows that @xmath573 , for @xmath350 and that @xmath574 , proving that @xmath575 .
let @xmath576 be the collection of all partitions of the set @xmath561 .
for any partition @xmath577 , we will assume that the blocks @xmath578 of @xmath560 are ordered by a total order @xmath579 in such a way that @xmath580 then , based on lemma [ characterization ] , the literal @xmath581 is expressed by the following @xmath11-formula @xmath582 let @xmath583 be the length of the formula ( [ mostro ] ) . then the following bounds on @xmath583 hold : @xmath584 where @xmath585 is the @xmath586th bell s number .
using the bounds on @xmath587 by berend and tassa ( cf .
@xcite ) @xmath588 the bounds ( [ bounds1 ] ) yield @xmath589
we have presented a three - sorted stratified set - theoretic fragment , @xmath0 , and given a decision procedure for its satisfiability problem .
the fragment turns out to be quite expressive since it allows to represent several set - theoretic construct such as variants of the powerset operator and the unordered cartesian product .
thanks to the presence of the finite enumeration operator , @xmath0 allows to represent the unordered cartesian product by means of a formula which is linear in the size of the product .
another representation of the latter construct is possible without resorting to the finite enumeration operator , but is this case the formula turns out to be exponentially longer .
proceeding as in @xcite it is possible to single out a family @xmath590 of sublanguages of @xmath0 , characterized by imposing further constraints in the construction of the formulae , such that each language in the family has the satisfiability problem -complete , and to show that the modal logic @xmath591 can be formalized in @xmath592 .
we further intend to study the possibility of formulating non - classical logics in the context of well - founded set theory constructing suitable extensions of the @xmath0 fragment .
d. cantone .
decision procedures for elementary sublanguages of set theory : x. multilevel syllogistic extended by the singleton and powerset operators . ,
volume 7 , number 2 , pages 193230 , kluwer academic publishers , hingham , ma , usa , 1991 .
d. cantone and v. cutello . a decidable fragment of the elementary theory of relations and some applications . in
issac 90 : proceedings of the international symposium on symbolic and algebraic computation _ , pages 2429 , new york , ny , usa , 1990 . acm press .
d. cantone and v. cutello .
ecision procedures for stratified set - theoretic syllogistics . in manuel bronstein , editor , _ proceedings of the 1993 international symposium on symbolic and algebraic computation , issac93 ( kiev , ukraine , july 6 - 8 , 1993 ) _ , pages 105110 , new york , 1993 .
acm press .
d. cantone , v. cutello , and j.t . schwartz .
decision problems for tarski and presburger arithmetics extended with sets . in _ csl 90 : proceedings of the 4th workshop on computer science logic _ , pages 95109 , london , uk , 1991 .
springer - verlag . | we solve the satisfiability problem for a three - sorted fragment of set theory ( denoted @xmath0 ) , which admits a restricted form of quantification over individual and set variables and the finite enumeration operator @xmath1 over individual variables , by showing that it enjoys a small model property , i.e. , any satisfiable formula @xmath2 of @xmath0 has a finite model whose size depends solely on the length of @xmath2 itself .
several set - theoretic constructs are expressible by @xmath0-formulae , such as some variants of the power set operator and the unordered cartesian product .
in particular , concerning the unordered cartesian product , we show that when finite enumerations are used to represent the construct , the resulting formula is exponentially shorter than the one that can be constructed without resorting to such terms
. please list keywords from your paper here , separated by commas . |
the authors of the last variant @xcite of known @xcite method for determination of level density @xmath0 and radiative strength functions @xmath1 from total gamma - spectra of reactions @xmath2he,@xmath3 ( and @xmath4 respectively ) made in @xcite the statement on practical impossibility to determine systematical errors for obtained by them values of enumerated parameters .
one can assume that this statement concerns not only axel - brink hypothesis @xcite , but also rather ordinary systematical errors in determination of absolute intensities of the total gamma - spectra .
i d est , reliability of all the data obtained by them is not completely determined .
consequently , all conclusions made by authors @xcite on parameters of cascade gamma - decay can be wholly mistaken due to rather significant coefficients @xmath5 of transport of experimental errors @xmath6 of measured of the total gamma - spectra intensity onto the required functions @xmath0 and @xmath7 .
increase in @xmath8 is strongly caused by the fact that any first generation spectra " in the region of low gamma - quanta energies are small difference of two large values ( moreover , one of them can not be determined experimentally even in principle ) .
the errors @xmath9 of function @xmath10 and its parameters @xmath11 , @xmath12 in vicinities of their most probable values are connected by approximate matrix equation : @xmath13 where @xmath14 - the matrix of derivatives of nonlinear function @xmath10 on its parameters @xmath0 and @xmath7 . matrix equation ( 1 ) can be solved on any modern computer .
the program for calculation of jacobi matrix @xmath14 ( of derivatives @xmath15 ) in analytic form for the system of nonlinear equations solved in @xcite was prepared and tested in @xcite .
the sole problem by work with this program is caused by the only circumstance
level density in diapason of neutron binding energy changes by 4 - 5 orders of magnitude .
that is why , specific of computer arithmetic noticeably influences on results of calculation .
the presence of very significant non - diagonal elements in characteristic matrix @xmath16 of the likelihood function for the system of equations solved in @xcite , causes their degeneration for any possible weight matrix @xmath17 even by the use of additional data on density of neutron resonances , low - lying levels and by fixation of values of the total radiative widths of neutron resonances .
therefore , the desired @xmath0 and @xmath7 parameters and their errors have a multitude of values with equal probability .
this circumstance was not pointed out in the all papers performed with the use of method @xcite .
authors of @xcite postulated that the error in level density derived from the spectra of evaporated nucleons does not depend on error in level density obtained with the help of @xcite .
, they did not take into account a possibility of strong correlation between systematical errors in determination of @xmath0 .
this correlation is stipulated , for example , by the use of identical hypothesis axel - brink @xcite for prediction of gamma - quanta emission partial widths and bohr - mottelson @xcite for nucleon products of nuclear reactions .
consequently , a multitude of equal - in - probability vectors @xmath18 must be found from solution of equation ( 1 ) for matrixes @xmath14 corresponding to different equal - in - probability vectors @xmath19 . the width of interval of the possible values of the vector elements @xmath18 can be considered as the measure of experimental errors of @xmath0 and @xmath7 .
the necessary for this operation vector @xmath6 ( or its upper estimates ) in any experiment is determined only by analysis of factors carrying systematical distortions in the measured spectra .
its correct estimation can be done only by the authors of method oslo " @xcite by means of corresponding calculations and additional experiments . for the present , this circumstance does not allow one to use equation ( 1 ) for estimation of region of the possible @xmath18 vector values . therefore , qualitative estimation of the expected systematical errors of @xmath20 and @xmath21 can be made only in the other way by comparison of the difference of the total gamma - spectra intensities at decay of levels from any spin window @xmath22 for arbitrary excitation energy @xmath23 , with difference of functional dependencies of @xmath0 and @xmath7 for pairs of their different model or experimental data sets used in the calculation .
naturally , this brings to loss of dependence on all the distortions carried in @xcite at intermediate steps .
the last concerns not only the procedure for determination of the first generation spectra " , but and the use of mistaken ( this is pointed out in first time in @xcite ) axel - brink hypothesis @xcite and , to a very high extent , to the method used for normalization of the total spectra for arbitrary energy @xmath23 to the same number of decays . therefore , all the conclusions obtained below give notions only of the biggest permissible value ( by using @xcite ) of systematical errors in @xmath10 .
the considerations presented above permit one to formulate main conditions and criterions for solution of the problem of partial evaluation of the error transfer coefficients of the measured total gamma - spectra onto the values of level density and radiative strength functions derived from them .
first of all , there must be whose parameters @xmath0 and @xmath7 were determined by two independent and really in principle different methods in the excitation energy interval being maximally close to neutron binding energy .
it is also desirable to have the @xmath0 and @xmath7 values in one of analyzed sets with minimum possible uncertainty estimated by independent and traditional methods of error determination .
they must be extracted from experiment without the use of untested hypotheses ( like @xcite ) .
it is also desirable to have the maximum different types of experimental data for functional dependencies @xmath0 and @xmath7 on excitation and gamma - quantum energies .
the mentioned requirements are satisfied , for example , in compound nuclei @xmath24mo and @xmath25yb .
the total gamma - spectrum at maximum excitation energy @xmath23 is ( to a precision of different population of initial levels with different @xmath26 ) a superposition of the first generation spectra " for this and lower excitation energies .
that is why , one can expect that the estimated below effect of influence of the total gamma - spectra systematic errors on uncertainties of desired @xmath0 and @xmath7 values has no principle differences relative the case @xcite .
the total gamma - spectrum following decay of levels from narrow interval @xmath27 for any excitation energy bin of a nucleus under study can be normalized to both the given decay number of initial levels or to the total cascade energy @xmath23 . in the first case , systematical error @xmath28 is an algebraic sum of errors in determination of form of measured spectra at different energies of gamma - quanta and inevitable error of absolute normalization . in the second case : @xmath29 all the systematical errors are minimal and have sign - changeable dependence on @xmath30 and in sum are equal to error @xmath31 .
the use in calculation of different functional dependences for @xmath0 and @xmath7 brings ( at the of normalization ( 2 ) ) to analogous effect .
its value can be characterized by the parameter @xmath32 for any pairs of the standard " and tested " total gamma - spectra @xmath33 and @xmath34 , respectively .
it is postulated below that any sets of @xmath0 and @xmath7 can be obtained with equal probability from analysis like that in @xcite , if only mean - square difference @xmath35 of the total gamma - spectra calculated with the use of them is approximately the same and equals square root mean of errors of the experimental spectrum .
correspondingly , reliable determination of experimental values of desirable parameters @xmath0 and @xmath7 requires that mean - square error of experimental spectra must be much less than mean - square difference @xmath35 of two corresponding spectra .
the existing differences in the @xmath0 and @xmath7 values give notion on real magnitudes of their systematical errors . in practice ,
extraction of the @xmath0 and @xmath7 values from @xmath10 is performed in several steps @xcite .
therefore , each step brings to additional increase in systematical errors of the determined values .
this circumstance must be taken into account at estimation of required precision at determination of @xmath10 .
comparison between the data on @xmath0 and @xmath7 obtained in oslo within the method @xcite and dubna data @xcite allows us to reveal their characteristic peculiarities and to determine the minimal permissible systematical error @xmath6 which guaranties reliable identification of minimal difference of the tested parameters .
it is determined , first all , by the presence of the step - like structure and correlated with it in position peak in the radiative strength functions in the dubna data .
the oslo data on level densities are very close to the existing primitive models ( for example , @xcite ) . unlike the dubna data they do not show abrupt change in nuclear properties below neutron binding energy .
practically , there were used in calculations the models of radiative strength functions@xcite for @xmath36-transitions , @xcite for level density , results of approximation of dubna parameters of cascade gamma - decay @xcite and the experimental data on @xmath0 and @xmath7 for @xmath24mo @xmath25yb , obtained in oslo @xcite for reactions induced by @xmath2he .
comparison between results of calculation for combinations : + 1 .
@xcite , + 2 .
@xcite , + 3 .
@xcite , + 4 .
@xcite , + 5 .
@xcite , + 6 .
@xcite , + 7 .
@xcite + was performed for the energy of initial level @xmath37 for two its possible parities and spins excited at the thermal neutron capture . below @xmath38 mev for @xmath24mo and 1.85 mev for @xmath25yb in calculation
was used experimental information @xcite on decay modes and parameters of known levels of these nuclei .
parameters of the model @xcite were chosen from those obtained in @xcite in such a way that to have smallest discrepancy with @xcite .
information on energy resolution of scintillation detectors was included in calculation as well .
it was supposed in all calculations that the level densities of different parity above the energy @xmath39 are equal ; only dipole transitions were taken into account .
the sum of @xmath40 in calculation of parameter @xmath41 for different strength functions was re - normalized so that the total radiative width of decaying level was equal in all variants of calculation .
all the total gamma - spectra were normalized to the energy @xmath42 ( in correspondence with ( 2 ) ) .
the total area of calculated spectrum at normalization ( 2 ) to the cascade energy is obviously equal for any set of @xmath0 and @xmath7 .
therefore , all the calculated variants of spectra differ each from other only by shape of dependence @xmath43 .
just this circumstance decreases sensibility of the method like @xcite as compared with extraction of @xmath0 and @xmath7 from the two - step cascade intensities .
this conclusion is true only for the data for @xmath0 and @xmath7 , obtained in dubna .
the results of the cascade intensity analysis performed in prague ( and by other groups in the same manner ) are not reliable due to the presence in corresponding method of three principle mistakes .
this was shown in @xcite .
it is naturally that the forms of calculated spectra depend also on ratio between excitation probability of initial levels with different @xmath26 , on difference in level densities with different parity , ratio between strength functions of dipole gamma - transitions of different type and so on .
but , judging by the data listed below , limitation in extent of variation of calculation parameters by the mentioned upper variants can not bring to radical change of the made below conclusions on required precision in determination of @xmath10 . at calculation of concrete values of parameters @xmath35
there were introduced the low and high thresholds for @xmath30 .
they were equal to 0.5 and 7.0 mev for molybdenum and 0.5 and 6.0 mev for ytterbium .
this was done for reduction of contribution in value @xmath35 of parts of spectra with low intensity .
seven obtained variants of the calculated total gamma - spectra for these isotopes are shown in fig .
their parts corresponding to spectra of only primary gamma - transitions are presented in fig . 2 in the same scale .
it is natural that the ratio between intensities of the spectra presented in figures 1 and 2 practically does not depend on the used method of normalization .
the used in the calculation level densities ( excited by the primary dipole gamma - transitions following thermal neutron capture in nuclei under study ) are shown in fig .
3 , and strength functions in fig .
4 . the ratios @xmath44 intensity of the calculated total gamma - spectra to intensity of only their primary transitions
is presented in fig .
modern nuclear models take into account influence of structure of levels connected by gamma - transition on its probability .
that is why a necessity to use the axel - brink hypothesis @xcite is absent in modern theory .
but , a precision of any theoretical calculation is determined by obviously insufficient accuracy of the experimental data which were the basis for development and parameterization of the model like qpnm @xcite .
unambiguous conclusion on existence of strong dependence @xmath45 was firstly obtained from the experiment @xcite .
unfortunately , due to methodical reasons information on this function is limited by the region @xmath46 .
the degree of mistakenness of the axel - brink hypothesis at high excitation energy can be estimated only by experimental study of cascades with three and more successively emitted gamma - quanta .
corresponding experiment was realized by authors of @xcite . in spite of limited character of the data @xcite ,
they allow one to expect a possibility of very strong or even complete compensation of decrease of level density at any excitation energy @xmath47 with respect to general trend by increasing of the gamma - transition widths directly follows to corresponding levels .
it was shown in @xcite that the analogous effect should be taken into account and in analysis in the spectra of evaporated nucleons also .
there are no principle difficulties to calculate the total gamma - spectrum under assumption of complete compensation of decrease in level density by simultaneous increase in strength functions . in the limit of the possible complete @xmath48 compensation of deviations of parameters of gamma - decay from general trend
the calculation of the total gamma - spectrum with or without accounting for hypothesis @xcite must give the same total gamma - spectrum . in the other words ,
the method @xcite is weakly sensitive or completely insensitive to shape of function @xmath45 . but , cascade population of any level at any excitation energy depends on the cross - section of gamma - quantum interaction with excited nucleus and is not compensated by change in level density
. therefore , analysis of cascade intensities @xcite allows one to study real dependence of strength functions on both gamma - quantum energy and structure of nucleus . the method like @xcite has not such possibilities .
the results of comparison of mean - square relative differences of the functions under consideration are listed in table .
the parameters @xmath35 for all combinations were determined at choice of the @xmath0 and @xmath7 values as the base " first variant .
this combination is more often used in different calculations . _ table .
the square mean differences of intensities of calculated total gamma - spectra @xmath49 , intensities of only primary gamma - transitions @xmath50 , the used level densities @xmath51 and radiative strength functions @xmath41 .
@xmath52 , @xmath53 , @xmath54 and @xmath55 _ the modules of their maximal relative differences . [ cols="<,>,>,>,>,>,^,>",options="header " , ] _ @xmath56experimental level density of @xmath24mo , given in @xcite , completely corresponds to model @xcite .
_ these data allow one to get some notion on extent of increase in spectrum error of the primary gamma - transitions used in @xcite for determination of the @xmath0 and @xmath7 values . to a precision of coefficient @xmath57 ( or some more ) , the data in fig .
5 gives an extent of increase of error of total gamma - spectrum intensity at its transformation in the spectrum of only primary gamma - transitions .
coefficients of increase in systematical errors of @xmath0 and @xmath7 owing to the use of axel - brink hypothesis instead of the radiative strength function depending on energy of excited level ( structure of its wave function ) can not be less than 2 ( obtained from comparison of @xmath0 for the same nuclei in @xcite and @xcite ) .
it can have arbitrary value .
the way for its observation variation of forms of the initial @xmath0 and @xmath7 functions as in the method @xcite and using only library programs of multidimensional fitting .
maximal value of @xmath49 for the parameters used by the testing does not exceed 0.25 0.28 .
correspondingly , maximal error in measurement of the total spectra must be 3 - 5 times less .
the procedure of determination of the fist generation spectra " brings to additional increase of experimental errors in the region above threshold of the experiment in 3 - 10 times and more ( fig .
5 ) . in practice , infringement of the axel - brink hypothesis appears itself in simultaneous increase in @xmath7 at decrease in @xmath0 values in the region of the step - like structure .
this effect was revealed for both primary and secondary gamma - transitions .
but , observation of compensation of decrease in @xmath0 by increase in @xmath7 , as it was obtained in @xcite , requires of both rise of precision at determination of intensity of total spectrum and to decrease of the @xmath49 value .
corresponding coefficient in our estimation can be adopted to equal to that presented above . as a result
, one can assume in the first approach that certain and unambiguous identification of different values of level densities and strength functions requires that the total relative error of determination of absolute intensity of the total gamma - spectra in method @xcite must be , most probably , less than 0.01 at least for low energy bins of the measured spectra .
this conclusion quite unambiguously from the made in @xcite attempts to reproduce intensity of two - step cascades in @xmath58fe , @xmath24mo and @xmath25yb by means of level density and strength functions derived by
oslo method " .
there are made in publications the attempts to reproduce the sum of unknown two - step cascade intensities with close values of energies of the primary and secondary gamma - transitions by calculations with corresponding pairs of @xmath0 and @xmath7 .
impossibility to prove unambiguous reproduction of each of these items is obvious even if it was achieved for the sum and for all energies of the cascade gamma - transitions . in the other case the data
@xcite moreover can not correspond to the @xmath0 and @xmath7 values which reproduce intensity of cascades .
analysis of these data @xcite performed in correspondence with requirements of mathematical statistics and mathematics completely excludes reliability of determination of @xmath0 and @xmath7 by method @xcite , at least , for the gamma - decay of compound states excited at thermal neutron capture .
moreover , this question was stated in details in @xcite .
so , the method @xcite for determination of the @xmath0 and @xmath7 values can give reliable information on nuclear parameters only in the experiment which provides rather high precision of measurement of the total gamma - spectra .
it is probable that the spectrometer used for this aim in oslo can not provide for required precision even in principle .
99 a. schiller et al . , nucl .
. res . * a447 * ( 2000 ) 498 .
bartholomew et al . , advances in nuclear physics , * 7 * ( 1973 ) 229 .
voinov et al . , phys .
c * 77 * , ( 2008 ) 034613 .
p. axel , phys .
* 126(2 ) * , ( 1962 ) 671 .
d. m. brink , ph .
d. thesis , oxford university ( 1955 ) .
v. a. khitrov et al . , in : _ proceedings of the xi international seminar on interactions of neutrons with nuclei , dubna , may 2003 _ , e3 - 2004 - 9 ( dubna , 2004 ) , p. 107
; nucl - ex/0305006 . o. bohr , b.r . mottelson,_nuclear structure _ , vol .
1 ( benjamin , ny , amsterdam , 1969 ) .
a.m. sukhovoj , v.a .
khitrov , phys . particl . and nuclei , * 36(4 ) * ( 2005 ) 359 .
+ http://www1.jinr.ru/pepan/pepan-index.html ( in russian ) w. dilg , w. schantl , h. vonach , m. uhl , nucl
* a217 * ( 1973 ) 269 .
kadmenskij , v.p .
markushev , v.i .
furman , sov .
* 37 * ( 1983 ) 165 .
strutinsky , in proc . of int .
( paris , 1958 ) , p. 617 .
a.m. sukhovoj , v.a .
khitrov , physics of paricl . and nuclei , * 37(6 ) * ( 2006 ) 899 .
http://www1.jinr.ru/pepan/pepan-index.html ( in russian ) .
a.m. sukhovoj , w.i .
furman , v.a .
khitrov , physics of atomic nuclei , * 71(6 ) * ( 2008 ) 982 .
r. chankova et al .
, phys.rev .
c * 73 * ( 2006 ) 034311 , + m. guttormsenet et al . ,
c * 71 * ( 2005 ) 044307 .
u. agvaanluvsan et al .
, phys.rev .
c * 70 * ( 2004 ) 054611 . http://www.nndc.bnl.gov/nndc/ensdf .
a.m. sukhovoj , v.a .
khitrov , preprint no .
e3 - 2008 - 134 , jinr ( dubna , 2008 ) .
sukhovoj a. m. , physics of atomic nuclei , * 71 * ( 2008 ) 1907 .
soloviev , theory of atomic nuclei : quasiparticles and phonons , institute of physics publishing , bristol and philadelphia , 1992 .
k. furutaka et al . , in international conference on nuclear data for science and technology 2007 , nice , 2007 , p. 517 , + m. oshima et all , 13-th international symposium on capture gamma - ray spectroscopy , book of apstracts , 2008 , p. 83 .
a.m. sukhovoj , v.a .
khitrov , in : xvii international seminar on interaction of neutrons with nuclei , dubna , may 2009 , e3 - 2010 - 36 , dubna , 2010 , p. 282 .
vasilieva , a.m. sukhovoj , v.a .
khitrov , phys . at
* 64(2 ) * ( 2001 ) 153 , ( nucl - ex/0110017 ) a.v
. voinov et al . , phys.rev.lett .
* 93 * ( 2004 ) 142504 .
m. krticka , f. becvar , i. tomandl et al .
, phys.rev .
c * 77 * , ( 2008 ) 054319 .
a. schiller et al . , phys.lett .
b * 633 * ( 2006 ) 225 .
a.m. sukhovoj , v.a .
khitrov , li chol , pham dinh khang , vuong huu tan , + nguyen xuan hai , in : xiii international seminar on interaction of neutrons with nuclei , dubna , may 2006 , e3 - 2006 - 7 , dubna , 2006 , p. 72 .
khitrov , li chol , a.m. sukhovoj , xii international seminar on interaction of neutrons with nuclei , dubna , may 2004 , e3 - 2004 - 169 , dubna , 2004 , p. 438 .
+ http://arxiv.org/abs/nucl-ex/0409016 .
the accepted in calculation level density excited by the dipole primary gamma - transitions at the thermal neutron capture .
data @xcite , lines - @xcite , histogram level density from reaction @xmath59 ) , obtained by analogy with @xcite . | from a comparison of the total gamma - spectra calculated for different functional dependencies of level density and radiative strength functions , there were obtained both their square root relative differences and analogous data for the used parameters .
the analysis of these data showed that the total uncertainty in determination of gamma - spectra intensities which is necessary to obtain reliable values of parameters of cascade gamma - decay , most probably , must not exceed one percent .
* estimation of maximum permissible errors in the total gamma - spectra intensities at determination from them of level density and radiative strength functions * * a.m. sukhovoj , v.a .
khitrov * + _ joint institute for nuclear research , 141980 , dubna , russia _ + |
the mass of the recently - discovered top quark @xcite has been measured with impressive accuracy , @xmath2 gev @xcite , by the cdf and d0 experiments at the fermilab tevatron .
the uncertainty will be reduced even further , to perhaps 1 - 2 gev , with additional running at the tevatron @xcite , or at the cern large hadron collider @xcite .
high - energy @xmath3 @xcite or @xmath4 @xcite colliders operating at the @xmath5 threshold hold the promise of yet more precise measurements of @xmath6 , to 200 mev or even better . with such increasingly - precise measurements on the horizon , it is important to have a firm grasp of exactly what is meant by the top - quark mass . thus far
the top - quark mass has been experimentally defined by the position of the peak in the invariant - mass distribution of the top - quark s decay products , a @xmath7 boson and a @xmath8-quark jet @xcite .
this closely corresponds to the pole mass of the top quark , defined as the real part of the pole in the top - quark propagator .
the propagator of a top quark with four - momentum @xmath9 has a pole at the complex position @xmath10 , and yields a peak in the @xmath11 invariant - mass distribution ( for experimentally - accessible real values of @xmath9 ) when @xmath12 .
the pole mass of a stable quark is well - defined in the context of finite - order perturbation theory @xcite .
however , the all - orders resummation of a certain class of diagrams , associated with `` infrared renormalons '' , indicates that the pole mass of a stable heavy .
] quark is ambiguous by an amount proportional to @xmath1 , as a result of nonperturbative qcd @xcite .
physically , this is a satisfying result , because we believe that quarks are permanently confined within hadrons , precluding the unambiguous definition of a quark pole mass @xcite .
the top quark decays very quickly , having a width @xmath13 , approximately an order of magnitude greater than the strong - interaction energy scale @xmath14 .
such a short lifetime means that the top quark decays before it has time to hadronize @xcite .
the large top - quark width can act as an infrared cutoff , potentially insulating the top quark from the effects of nonperturbative qcd @xcite .
given this information , one might expect the top - quark pole mass to be free of the ambiguities associated with nonperturbative qcd .
the purpose of this article is to demonstrate that this is not the case .
the top - quark pole mass , like the mass of a stable heavy quark , is unavoidably ambiguous by an amount proportional to @xmath1 .
we demonstrate this in two ways , first by a general argument using @xmath15-matrix theory , second by a consideration of infrared renormalons .
the ambiguity in the pole mass in the specific context of the @xmath11 invariant - mass distribution is discussed at the end of the next section .
consider a scattering process with asymptotic states consisting of stable particles .
we first ask if it is possible for the scattering amplitude to have a pole at the mass of a stable quark .
this would correspond to a quark propagator connecting two subamplitudes , as depicted in fig . 1 ; the pole in the quark propagator would correspond to the pole in the amplitude .
such a configuration is impossible , however : the subamplitudes which the quark propagator connects have external states which are color singlets ( due to confinement ) , while the quark is a color triplet , so color is not conserved .
thus there can not be a pole in the amplitude at the quark mass .
this argument applies equally well to an unstable quark , such as the top quark .
the fact that the quark is unstable evidently plays no role in the argument ; it only shifts the imagined pole in the propagator into the complex plane . as in the case of a stable quark
, there can not be a pole in the amplitude , regardless of how short - lived the quark .
in particular , the fact that the top - quark lifetime is much less than @xmath0 is irrelevant . there is another way to understand why the short top - quark lifetime is irrelevant
let us return to fig . 1 , and again consider first the case of a stable quark .
imagine that there is a pole in the amplitude at the quark mass .
near the pole , the scattering amplitude would factorize into the production of the stable quark by scattering subprocess a , followed by its propagation over a large proper time , and concluding with its participation in scattering subprocess b. thus the quark could be considered as an asymptotic state .
this demonstrates that the poles in the scattering amplitude of a theory correspond to its asymptotic states @xcite .
since quarks are not asymptotic states , due to confinement , there can not be a pole at the quark mass .
a similar argument applies to an unstable quark , such as the top quark .
the imagined pole position is now located at a complex value . because the scattering amplitude is an analytic function , the analytic continuation to complex momentum is well - defined . near the pole
, the scattering amplitude would factorize as before , although this would no longer correspond to a true physical process since the top - quark would have complex momentum @xcite .
the top quark would propagate over a large proper time , and could not escape confinement
. there would be no asymptotic top - quark state , and hence no pole .
we are left with the following physical picture .
a state with momentum near its pole corresponds to a long - lived particle , regardless of whether the pole is real or complex .
if the particle is colored , it will be confined , preventing an unambiguous definition of the pole mass of the particle .
these arguments imply that the nonperturbative aspect of the strong interaction will stand in the way of any attempt to unambiguously extract the top - quark pole mass from experiment .
for example , consider the extraction of the pole mass from the peak in the @xmath11 invariant - mass distribution . in perturbation theory ,
the final state is a @xmath7 and a @xmath8 quark , as depicted in fig .
however , the @xmath8 quark manifests itself experimentally as a jet of colorless hadrons , due to confinement .
at least one of the quarks which resides in these hadrons comes from elsewhere in the diagram , and can not be considered as a decay product of the top quark , as depicted in fig .
this leads to an irreducible uncertainty in the @xmath11 invariant mass of @xmath16 , and hence an ambiguity of this amount in the extracted top - quark pole mass .
we now turn to an investigation of the top - quark pole mass from the perspective of infrared renormalons .
we first review the argument which demonstrates the existence of a renormalon ambiguity in the pole mass of a stable heavy quark @xcite .
we then extend the argument to take into account the finite width of the top quark .
finally , we investigate the existence of a renormalon ambiguity in the top - quark width itself .
the pole mass of a quark is defined by the position of the pole in the quark propagator .
the propagator of a quark of four - momentum @xmath9 is @xmath17 where @xmath18 is a renormalized short - distance mass , mass ) evaluated at a scale much greater than @xmath1 . ] and @xmath19 is the renormalized one - particle - irreducible quark self - energy .
the equation for the position of the pole is @xmath20 this is an implicit equation for @xmath21 that can be solved perturbatively .
we first work to leading order in @xmath22 , which gives @xmath23 where @xmath24 is the one - loop quark self - energy shown fig .
this quantity is real , so the pole position is real .
renormalons arise from the class of diagrams generated by the insertion of @xmath25 vacuum - polarization subdiagrams into the gluon propagator in the one - loop self - energy diagram , as shown in fig .
3(a@xmath26 ) .
one can express this as @xmath27 where @xmath28 and @xmath29 is the one - loop qcd beta - function coefficient , @xmath30 .
formally , these are the dominant qcd corrections in the `` large-@xmath29 '' limit
. thus @xmath31 in eq . (
[ sigma ] ) is calculated at leading order in @xmath22 , but to all orders in @xmath32 . for large @xmath25 the coefficients @xmath33 grow factorially , and
are given by @xcite @xmath34 where @xmath35 is a finite renormalization - scheme - dependent constant .
scheme , @xmath36 . ] the series in eq .
( [ sigma ] ) is therefore divergent .
one can attempt to sum the series using the technique of borel resummation @xcite .
the borel transform ( with respect to @xmath32 ) of the self - energy is obtained from the series coefficients , eq .
( [ coeff ] ) , via @xmath37 where @xmath38 is the borel parameter . because the coefficients @xmath33 are divided by @xmath39 in the above expression , the series has a finite radius of convergence in @xmath38 , and can be analytically continued into the entire @xmath38 plane .
the self - energy is then reconstructed via the inverse borel transform , given formally by @xmath40 the integral in eq .
( [ inverse ] ) is only formal , because the borel transform of the quark self - energy possesses singularities on the real @xmath38-axis , which impede the evaluation of the integral .
these singularities are referred to as infrared renormalons because they arise from the region of soft gluon momentum in fig .
3(a@xmath26 ) .
the series for the self - energy in eq .
( [ sigma ] ) is therefore not borel summable .
the divergence of the series for the self - energy is governed by the infrared renormalon closest to the origin , which lies at @xmath41 .
this renormalon is not associated with the condensate of a local operator , so it can not be absorbed into a nonperturbative redefinition of the pole mass @xcite .
instead , one can choose some _ ad hoc _ prescription to circumvent the singularity in the integral .
the difference between various prescriptions is a measure of the ambiguity in the pole mass . estimating the ambiguity as half the difference between deforming the integration contour above and below the singularity gives @xcite @xmath42
so the pole mass is ambiguous by an amount proportional to @xmath1 .
we now include the @xmath43 contribution to the top - quark self - energy shown in fig .
the pole position is still given by eq .
( [ pole1 ] ) , but where @xmath24 includes both figs .
3(a ) and ( b ) .
since fig .
3(b ) has an imaginary part , the pole moves off the real axis .
the imaginary part of the one - loop pole position defines the tree - level top - quark width via as before , to extend the calculation to all orders in @xmath32 , we replace fig .
3(a ) by fig .
3(a@xmath26 ) .
this contribution to the pole mass remains the same as for a stable quark , and has the same renormalon ambiguity . at leading order in @xmath44
, the infrared renormalons do not know about the top - quark width .
the @xmath45 contribution to the top - quark self - energy learns about the top - quark width if one works to all orders in @xmath44 , via a schwinger - dyson representation @xcite , as shown in fig .
the circles on the internal propagators and the vertex in figs .
4(a ) and ( b ) represent the weak corrections to all orders in @xmath44 .. ] we wish to solve for the pole position as given by eq .
( [ pole ] ) .
we denote the pole position at zeroth order in @xmath22 , but to all orders in @xmath44 , by the complex value @xmath46 , with @xmath47 , where @xmath48 is the top - quark width to all orders in @xmath44 . at leading order in @xmath22 , the pole position is then given by @xmath49 where @xmath50 is given by figs .
4(a ) and ( b ) .
again , we extend this calculation to all orders in @xmath32 by making @xmath25 vacuum - polarization insertions in the gluon propagator , as depicted in fig 4(a@xmath26 ) .
this yields a series in @xmath32 , which we denote by @xmath51 in analogy with eq .
( [ sigma ] ) . to investigate whether the width might cut off the infrared renormalons generated by these diagrams ,
we need only consider the contribution of soft gluons .
in the limit of vanishing gluon momentum , the internal propagator reduces to @xmath52 , where @xmath53 is the wavefunction - renormalization factor .
the ward identity tells us that , in this same limit , the dressed vertex is simply @xmath54 .
thus , in the infrared limit , @xmath51 is formally identical to @xmath31 with @xmath18 replaced by @xmath46 everywhere .
the infrared renormalons , which are associated with the borel transform with respect to @xmath32 , are unaffected .
the width does not act as an cutoff for infrared renormalons , despite the fact that it is much greater than @xmath1 .
we conclude that the pole mass of the top quark is ambiguous by an amount proportional to @xmath1 , just as for the case of a stable quark .
we next ask whether the top - quark width suffers from a similar renormalon ambiguity . because the first - order calculation yields the top - quark width at tree level only , it is insufficient to address this question .
the solution to eq .
( [ pole ] ) at @xmath55 is @xmath56 where the superscripts on @xmath57 indicate the order at which it is to be evaluated .
the imaginary part of this equation ( times @xmath58 ) defines the top - quark width at @xmath55 .
one may calculate the imaginary part of eq .
( [ pole2 ] ) using the cutkosky rules .
this reduces to the calculation of the qcd correction to the process @xmath59 .
corresponds to the wavefunction renormalization of the top quark . ]
the presence of renormalons in this process was investigated in refs .
if the width is expressed in terms of the pole mass , then it has an infrared renormalon at @xmath41 , corresponding to an ambiguity proportional to @xmath1 .
however , if the width is expressed in terms of a short - distance mass , such as the @xmath60 mass , there is no renormalon at @xmath41 , and hence no ambiguity proportional to @xmath1 .
although the top - quark lifetime is much less than the strong - interaction time scale , @xmath0 , there are nonperturbative contributions to the top - quark pole mass , just as in the case of a stable heavy quark .
these nonperturbative contributions are signaled by the divergent behavior at large orders of an expansion in @xmath61 .
this leads to an unavoidable ambiguity of @xmath62 in the pole mass of the top quark . a short - distance mass , such as the @xmath60 mass ,
can in principle be measured with arbitrary accuracy .
this may require nonperturbative information , depending on the measurement .
it is sensible to adopt the @xmath60 mass as the standard definition of the top - quark mass , as is the convention for the lighter quarks @xcite .
the relation between the top - quark pole mass and the @xmath60 mass evaluated at the pole mass , @xmath63 , is known to two loops @xcite : @xmath64 where the last term reminds us that the pole mass has an unavoidable ambiguity of @xmath62 .
given that the pole mass is ambiguous , we suggest as the standard the @xmath60 mass evaluated at the @xmath60 mass , which is related to the pole mass by @xmath65 the difference in the coefficients of the two @xmath66 terms above is exactly 8/3 . for a top - quark pole mass of @xmath67 gev , @xmath68 gev . gev . ]
the considerations of this paper apply to any colored particle , stable or unstable .
thus , if nature is supersymmetric , the pole masses of squarks and gluinos will necessarily be ambiguous by an amount proportional to @xmath1 .
we are grateful for conversations with e. braaten , a. el - khadra , r. leigh , t. liss , and t. stelzer .
this work was supported in part by department of energy grant de - fg02 - 91er40677 .
we gratefully acknowledge the support of a gaann fellowship , under grant number de - p200a10532 , from the u. s. department of education for m. s. v. fadin and v. khoze , pisma zh .
. fiz . * 46 * , 417 ( 1987 ) [ jetp lett . * 46 * , 525 ( 1987 ) ] ; yad
. fiz . * 48 * , 487 ( 1988 ) [ sov .
phys . , * 48 * , 309 ( 1988 ) ] ; v. fadin , v. khoze , and t. sjstrand , 48 613 1990 .
g. t hooft , in _ the why s of subnuclear physics _ , proceedings of the international school of subnuclear physics , erice , 1977 , ed .
a. zichichi ( plenum , new york , 1979 ) , p. 943 ; a. mueller , in _ qcd - 20 years later _ ,
proceedings of the workshop , aachen , germany , 1992 , eds .
p. zerwas and h. kastrup ( world scientific , singapore , 1993 ) , vol . 1 , | the top quark decays more quickly than the strong - interaction time scale , @xmath0 , and might be expected to escape the effects of nonperturbative qcd . nevertheless , the top - quark pole mass , like the mass of a stable heavy quark , is ambiguous by an amount proportional to @xmath1 .
-.5 in 6.5 in # 1 # 2 # 3 phys .
rev .
* # 1 * , # 2 ( # 3 ) # 1 # 2 # 3 phys .
rev .
lett . * # 1 * , # 2 ( # 3 ) # 1 # 2 # 3 phys .
rev .
d * # 1 * , # 2 ( # 3 ) # 1 # 2 # 3 phys .
lett .
* b#1 * , # 2 ( # 3 ) # 1 # 2 # 3 nucl .
phys .
* b#1 * , # 2 ( # 3 ) # 1 # 2 # 3 rev . mod .
phys . * # 1 * , # 2 ( # 3 ) # 1 # 2 # 3 z. phys .
c * # 1 * , # 2 ( # 3 ) * top - quark pole mass * + + department of physics + university of illinois + 1110 west green street + urbana , il 61801 + |
quantum fields with impurities ( defects ) are recently subject of intensive studies in the context of condensed matter physics @xcite-@xcite , conformal field theory @xcite and integrable systems @xcite-@xcite .
most of the papers concern the case of one point - like defect . for more realistic applications
however , the generalization to @xmath4 defects would be essential . in the present paper
we face the multi - defect problem by extending the algebraic technique developed in @xcite-@xcite for @xmath5
. we will show below that this technique perfectly applies to the case @xmath4 as well , the key point being the reformulation of the boundary value problem at hand in algebraic terms .
an immediate application of this formalism is the derivation of the casimir force acting on a test particle . since
this issue is of some experimental interest , we treat it in detail , deriving the general expression of force .
we consider also a complex scalar field and determine the relative charge density . as an explicit example we focus on two delta - type defects .
it appears that the intensity and direction of the force depend on the position .
the force is discontinuous at the defects and actually diverges when one is approaching any of them from the left ( right ) .
a compensation occurs however in the symmetric limit , which is finite .
the paper is organized as follows . introducing the notation , in the next section we briefly summarize the algebraic approach to one defect .
this approach is generalized in section [ sec:2defect ] , where the solution for two defects is established .
the latter is extended to @xmath6 defects in section [ sec : ndef ] . in section
[ sec : fieldtheo ] we establish the two - point correlation function both at zero and finite temperature . using this results
we compute the casimir force ( section [ sec : cas ] ) and the charge density of the complex scalar field ( section [ sec : charged ] ) . in section
[ sec : delta ] we consider in details two delta - defects and discover a kind of resonance effect in the behavior of the density between the two defects .
some remarks and our conclusions are collected in section [ sec : conclu ] .
it is useful to recall first the algebraic treatment @xcite of one defect on the real line .
we take a defect with arbitrary position @xmath7 .
it divides the real line in two domains @xmath8 and @xmath9 . for simplicity
we focus on the real massive scalar field @xmath10 , satisfying @xmath11 \varphi ( t , x ) = 0\ , , \qquad x\not= y \ , .
\label{eqm}\ ] ] it is well known @xcite that all possible dissipationless interactions of @xmath10 with the defect @xmath12 are described by the boundary condition @xmath13 where @xmath14 following @xcite , one can solve equations ( [ eqm],[bc ] ) by introducing the so called reflection - transmission algebra with the following structure . with each domain
@xmath15 one associates a creation " and annihilation " operator @xmath16 which satisfy @xmath17 } & = & 0 = { [ a^*_i(k)\,,\,a^*_j ( p)]}\ , , \nonumber \\ { [ a_i(k)\ , , \ , a^*_j ( p ) ] } & = & \left [ \delta_{ij}+t_{ij}(k)\right ] 2\pi \delta(k - p)+r_{ij}(k ) 2\pi \delta(k+p)\ , , \label{ccr}\end{aligned}\ ] ] where @xmath18 represent the _ reflection _ and _ transmission coefficients _ from the impurity . in terms of the parameters
@xmath19 one has @xmath20 let us stress that the reflection coefficients ( but not the transmission ones ) depend on the position @xmath12 of the impurity : in the following , we will also use the ` bare ' coefficients @xmath21 which are just the part independent from the position @xmath12 : @xmath22 the associated defect _ scattering matrix _ @xmath23 satisfies unitarity and hermitian analyticity @xmath24 now , the solution of ( [ eqm ] ) can be written in the form @xcite @xmath25 \ , .
\label{f1}\ ] ] where @xmath26 . the boundary condition ( [ bc ] ) then implies that @xmath27 and @xmath28 are not independent , but satisfy the constraints @xmath29 which are consistent with ( [ ccr ] ) because of ( [ sm2 ] ) .
the algebra @xmath30 defined by equations ( [ ccr ] , [ c1 ] , [ c2 ] ) is known as reflection - transmission ( rt ) algebra @xcite . introducing the matrices @xmath31 equations ( [ c1 ] , [ c2 ] )
can be conveniently written in the form @xmath32 by construction the matrix @xmath33 satisfies @xmath34 where the apex @xmath35 denotes transposition .
we end this summary with two observations . in the domain of @xmath19
where bound states are absent@xcite , the field @xmath36 satisfies in addition to ( [ eqm ] , [ bc ] ) the equal - time canonical commutation relations @xmath37 = 0\ , , \qquad [ ( { \partial}_t\varphi ) ( t , x_1)\ , , \ , \varphi ( t , x_2 ) ] = -i\delta ( x_1-x_2 ) \ , .
\label{initial}\ ] ] we will not consider in this paper the issue of bound states , referring for the details about their impact to @xcite .
we stress in conclusion that the above construction is purely algebraic and applies for any representation of the rt algebra @xmath30 .
this fact allows to investigate the physical effects associated with nonequivalent representations and in particular , to introduce @xcite the concept of temperature , which is relevant for the applications in section 5 .
we show now how the rt algebraic framework allows to compute in a very systematic and straightforward way the global reflection and transmission coefficients when there are two defects .
we confirm the results of @xcite , where these coefficients where computed through the calculation of green functions .
we consider two defects located at @xmath38 . at each of them
@xmath39 let @xmath40}(k)$ ] and @xmath41}(k)$ ] be the reflection and transition matrices defined by ( [ coef ] ) in terms of the parameters @xmath42 . according to ( [ def1 ] ) we set @xmath43}(k )
\equiv \left(\begin{array}{cc } r^{[1]}_{11}(k ) & t^{[1]}_{12}(k ) \\
t^{[1]}_{21}(-k ) & r^{[1]}_{22}(-k ) \end{array}\right)\ , , \qquad c^{[2]}(k ) \equiv \left(\begin{array}{cc }
r^{[2]}_{22}(k ) & t^{[2]}_{23}(k ) \\
t^{[2]}_{32}(-k ) & r^{[2]}_{33}(-k ) \end{array}\right)\ , , \label{def2}\ ] ] which satisfy by construction @xmath44}(k)\ , c^{[\alpha ] } ( -k)={\mbox{${\mathbb i}$}}\ , \qquad c^{[\alpha ] } ( k)^\dag = c^{[\alpha ] } ( -k)^t \ , , \qquad \alpha = 1,2\ , . \label{cunit}\ ] ] the two defects separate now the real line in three domains @xmath45 following the previous section , we associate with each of them the operators @xmath46 and define the field @xmath36 by ( [ f1 ] ) , where now @xmath47 . since @xmath36 must satisfy ( [ bc2 ] ) , we impose @xmath48}(k ) = c^{[\alpha ] } ( k)\ , a^{[\alpha ] } ( -k)\ , , \qquad a^{[\alpha ] } ( k)^\dag = a^{[\alpha ] } ( -k)^\dag\ , c^{[\alpha ] } ( k)^\dag\ , , \qquad \alpha = 1,2\ , , \label{cond2}\ ] ] where @xmath48}(k ) \equiv \left(\begin{array}{c } a_{\alpha } ( k ) \\ a_{\alpha + 1}(-k ) \end{array}\right)\ , , \qquad \alpha = 1,2\ , .\ ] ] the operators @xmath49 describe the propagation of @xmath36 in @xmath15 . only @xmath27 and @xmath50
give origin to asymptotic states .
therefore it is natural to expect that @xmath51 can be expressed in terms of @xmath27 and @xmath50 .
indeed , from ( [ cond2 ] ) it follows that @xmath52}_{11}(k ) a_2(-k)&=&a_2(k)- c^{[2]}_{12}(k ) a_3(k)\ , , { \nonumber \\}c^{[1]}_{22}(-k ) a_2(-k)&=&a_2(k)- c^{[1]}_{21}(-k ) a_1(k)\ , .
\end{aligned}\ ] ] eliminating @xmath53 from these two equations , one gets @xmath54}_{11}(k)-c^{[1]}_{22}(-k)\right ] a_{2}(k ) = c^{[2]}_{11}(k)\,c^{[1]}_{21}(-k)\,a_{1}(k)-c^{[1]}_{22}(-k)\,c^{[2]}_{12}(k)\ , a_{3}(k)\ , .
\label{eq : a2-a13}\ ] ] remark that because of the factors @xmath55 and @xmath56 appearing in @xmath57}_{22}(k)$ ] and @xmath58}_{22}(k)$ ] respectively , the coefficient @xmath59}_{11}(k)-c^{[1]}_{22}(-k)$ ] never vanishes for generic @xmath60 and ( [ eq : a2-a13 ] ) uniquely determines @xmath61 in terms of @xmath62 and @xmath63 .
one has @xmath64}_{22}(k )
t^{[1]}_{21}(-k ) a_1(k ) - r^{[1]}_{22}(k ) t^{[2]}_{23}(k ) a_3(k ) } { r^{[2]}_{22}(k ) - r^{[1]}_{22}(k)}\ , , \label{a2}\ ] ] analogously , @xmath65 is given in terms of @xmath66 and @xmath67 , @xmath68}_{22}(-k ) t^{[1]}_{12}(-k ) a^*_1(k ) - r^{[1]}_{22}(-k ) t^{[2]}_{32}(k ) a^*_3(k ) } { r^{[2]}_{22}(-k ) - r^{[1]}_{22}(-k)}\ , , \label{a*2}\ ] ] we conclude therefore that in order to define the rt algebra for two impurities @xmath69 , it is enough to fix the commutation relations between @xmath27 and @xmath50 .
for this purpose we have to determine the global reflection and transmission matrices @xmath70 in terms of the local ones @xmath71}$ ] and @xmath72}$ ] . according to ( [ c1 ] ) ,
@xmath73 and @xmath74 can be read from the relations between @xmath62 and @xmath63 , which can be derived as follows . plugging the expression ( [ eq : a2-a13 ] ) into ( [ cond2 ] )
, we get @xmath75}_{22}(-k)^{-1}\,c^{[1]}_{11}(k)\ , \frac{c^{[1]}_{22}(-k)-c^{[2]}_{11}(k ) } { 1-c^{[1]}_{22}(k)\,c^{[2]}_{11}(k)}\,a_{1}(-k ) + \frac{c^{[1]}_{12}(k)\,c^{[2]}_{12}(k ) } { 1-c^{[1]}_{22}(k)\,c^{[2]}_{11}(k)}\,a_{3}(k ) \ , . \label{r1}\ ] ] by means of the identities @xmath43}_{22}(-k)^{-1 } = c^{[1]}_{22}(k ) + c^{[1]}_{22}(-k)^{-1}\,c^{[1]}_{12}(k)\,c^{[1]}_{21}(-k)\ , , \ ] ] @xmath43}_{12}(-k)\,c^{[1]}_{22}(-k)^{-1 } = -c^{[1]}_{11}(k)^{-1}\,c^{[1]}_{12}(k)\ , , \ ] ] following from ( [ cunit ] ) , equation ( [ r1 ] ) can be rewritten in the form @xmath76}_{11}(k ) + \frac{c^{[1]}_{12}(k)\,c^{[1]}_{21}(k)\,c^{[2]}_{11}(k ) } { 1-c^{[1]}_{22}(k)\,c^{[2]}_{11}(k ) } \right ] a_{1}(-k ) + \frac{c^{[1]}_{12}(k)\,c^{[2]}_{12}(k ) } { 1-c^{[1]}_{22}(k)\,c^{[2]}_{11}(k ) } \,a_{3}(k ) \ , .
\label{r2}\ ] ] the same type of calculations provides @xmath77}_{22}(-k ) + \frac{c^{[2]}_{21}(-k)\,c^{[2]}_{12}(-k)\,c^{[1]}_{22}(-k ) } { 1-c^{[1]}_{22}(-k)\,c^{[2]}_{11}(-k)}\right ] a_{3}(-k ) + \frac{c^{[2]}_{21}(-k)\,c^{[1]}_{21}(-k ) } { 1-c^{[1]}_{22}(-k)\,c^{[2]}_{11}(-k ) } \,a_{1}(k ) \ , .
\label{r3}\ ] ] comparing ( [ r2 ] , [ r3 ] ) with ( [ c1 ] ) , one infers @xmath78}_{11}(k ) + \frac{c^{[1]}_{12}(k)\,c^{[1]}_{21}(k)\,c^{[2]}_{11}(k ) } { 1-c^{[1]}_{22}(k)\,c^{[2]}_{11}(k ) } \ : = \ { \widetilde{\cr}}_{11}(k)\,\e^{-2\ri ky_{1 } } \label{r11}\\ { \widetilde{\cr}}_{11}(k ) & = & { \widetilde{r}}^{[1]}_{11}(k ) + \frac{t^{[1]}_{12}(k)\,t^{[1]}_{21}(-k)\,{\widetilde{r}}^{[2]}_{22}(k)\,\e^{2\ri ky_{12 } } } { 1-{\widetilde{r}}^{[1]}_{22}(-k)\,{\widetilde{r}}^{[2]}_{22}(k)\,\e^{2\ri ky_{12 } } } \ , , \label{eq : r11tilde}\end{aligned}\ ] ] @xmath79}_{22}(-k ) + \frac{c^{[2]}_{21}(-k)\,c^{[2]}_{12}(-k)\,c^{[1]}_{22}(-k ) } { 1-c^{[1]}_{22}(-k)\,c^{[2]}_{11}(-k)}\ : = \ { \widetilde{\cr}}_{33}(k)\,\e^{-2\ri ky_{2 } } \label{r33}\\ { \widetilde{\cr}}_{33}(k ) & = & { \widetilde{r}}^{[2]}_{33}(k ) + \frac{t^{[2]}_{32}(k)\,t^{[2]}_{23}(-k)\ , { \widetilde{r}}^{[1]}_{22}(k)\,\e^{-2\ri ky_{12 } } } { 1-{\widetilde{r}}^{[1]}_{22}(k)\,{\widetilde{r}}^{[2]}_{22}(-k)\,\e^{-2\ri ky_{12}}}\ , , \label{eq : r33tilde}\end{aligned}\ ] ] @xmath80}_{12}(k)\,c^{[2]}_{12}(k ) } { 1-c^{[1]}_{22}(k)\,c^{[2]}_{11}(k ) } = \frac{t^{[1]}_{12}(k)\,t^{[2]}_{23}(k ) } { 1-{\widetilde{r}}^{[1]}_{22}(-k)\,{\widetilde{r}}^{[2]}_{22}(k)\,\e^{2\ri ky_{12 } } } \ , , \label{t13}\end{aligned}\ ] ] @xmath81}_{21}(-k)\,c^{[1]}_{21}(-k ) } { 1-c^{[1]}_{22}(-k)\,c^{[2]}_{11}(-k ) } = \frac{t^{[2]}_{32}(k)\,t^{[1]}_{21}(k ) } { 1-{\widetilde{r}}^{[1]}_{22}(k)\,{\widetilde{r}}^{[2]}_{22}(-k)\,\e^{-2\ri ky_{12}}}\ , , \label{t31}\end{aligned}\ ] ] which express the global two - impurity reflection and transmission coefficients in terms of the local ones . as for the one - impurity case ,
we have introduced ` bare ' reflection coefficients which , as the transmission coefficients , depend only on the distance between the two impurities , @xmath82 .
remark that when one of the two defect is ` trivial ' , i.e. @xmath83 and @xmath84 , the global coefficients @xmath85 and @xmath86 become identical to the ones of the remaining defect .
the relative @xmath87-matrix is @xmath88 using the expressions ( [ r33]-[t31 ] ) and the properties of @xmath71}(k)$ ] and @xmath72}(k)$ ] one can verify that ( [ sm3 ] ) satisfies both unitarity and hermitian analyticity ( [ sm2 ] ) , which are fundamental for the construction of the rt algebra @xmath69 below . equations ( [ r33]-[t31 ] ) describe in a compact way all ( possible infinite ) processes of reflection and transmission from the two defects , relating an incoming to an outgoing wave .
accordingly , we impose on the @xmath69-generators @xmath89 } & = & 0 = { [ a^*_i(k)\,,\,a^*_j(p)]}\ , , \nonumber \\ { [ a_i(k)\,,\,a^*_j(p ) ] } & = & \left [ \delta_{ij}+\ct_{ij}(k)\right ] 2\pi \delta(k - p)+\cr_{ij}(k)2\pi \delta(k+p)\ , , \label{ccr1}\end{aligned}\ ] ] where @xmath90 take _ only _ the values 1 and 3 . the exchange relations involving @xmath51 are determined by ( [ ccr1 ] ) and the relations ( [ a2 ] , [ a*2 ] ) .
summarizing , the rt algebra @xmath69 for two defects has the following simple structure .
the elements @xmath51 , associated with the field evolution between the defects , are linear combinations of @xmath27 and @xmath50 , which generate the asymptotic states and satisfy quadratic exchange relations ( [ ccr1 ] ) .
the latter have the same form as in the one - impurity case ( [ ccr ] ) , but involve the global two - impurity reflection and transmission matrices @xmath91 and @xmath92 , which are uniquely determined by the single defects via ( [ r11]-[t31 ] ) .
we will show in the next section that this result has a straightforward generalization to an arbitrary number @xmath6 of defects .
we consider here @xmath93 defects on @xmath94 with coordinates @xmath95 .
they define the @xmath96 domains @xmath97 the relative rt algebra @xmath98 is therefore generated by @xmath99 . extrapolating from the previous section , we expect that all @xmath100 can be expressed in terms of @xmath101 and @xmath102 , which generate in turn the asymptotic states .
let us now verify this scenario and determine the algebra @xmath98 . setting @xmath44}(k )
\equiv \left(\begin{array}{cc } r^{[\alpha ] } _ { \alpha \alpha}(k ) & t^{[\alpha ] } _ { \alpha ( \alpha+1)}(k ) \\ t^{[\alpha ] } _ { ( \alpha+1 ) \alpha } ( -k ) & r^{[\alpha]}_{(\alpha+1)(\alpha + 1)}(-k ) \end{array}\right)\ , , \qquad \alpha = 1, ...
,{n}\ , , \label{defn}\ ] ] we infer from the boundary condition ( [ bc2 ] ) that ( [ cond2 ] , [ a2 ] ) hold now for any @xmath103 . indeed , applying recursively the technics used in the case of two defects , one deduces that @xmath104}(k)=\left(\begin{array}{c } a_{j}(k ) \\
a_{l}(-k ) \end{array}\right)\ , , \qquad j , k=1,\ldots,{n}+1\ , , \ ] ] obeys the relations @xmath105}(k ) & = & c^{[j\ldots l]}(k)\,a^{[j , l]}(-k)\ , , \\ c^{[j\ldots l]}(k)^\dag & = & c^{[j\ldots l]}(-k)^t { { \hspace{4 mm}}\mbox{and}{\hspace{4 mm}}}c^{[j\ldots l]}(k)\,c^{[j\ldots l]}(-k)={\mbox{${\mathbb i}$}}\ , , \quad 1\leq j < l\leq { n}+1 \ , , \end{aligned}\ ] ] where @xmath106}(k)$ ] is some matrix built on the @xmath107}(k)$ ] matrices , @xmath108 . starting from @xmath109}_{21}(k)\,a_{1}(k ) + c^{[1\ldots j]}_{22}(k)\,a_{j}(-k)\ , , \\
a_{j}(k ) & = & c^{[j\ldots { n}+1]}_{11}(k)\,a_{j}(-k ) + c^{[j\ldots { n}+1]}_{12}(k)\,a_{{n}+1}(k)\ , , \end{aligned}\ ] ] one gets @xmath110}(k)-c_{22}^{[1\ldots j]}(k)\big)\,a_{j}(k ) = c_{21}^{[1\ldots j]}(k)\,c_{11}^{[j\ldots { n}+1]}(k)\,a_{1}(k ) -c_{22}^{[1\ldots j]}(k)\,c_{12}^{[j\ldots { n}+1]}(k)\,a_{{n}+1}(k).\ ] ] therefore , the expressions found in the case of two defects also apply here , with the rules @xmath111}(k)\,\to\,c^{[1\ldots j]}(k ) \ ; \ c^{[2]}(k)\,\to\,c^{[j\ldots { n}+1]}(k)\ , , \\ a_{1}(k)\,\to\,a_{1}(k)\ ; \
a_{2}(k)\,\to\,a_{j}(k ) \ ; \ a_{3}(k)\,\to\,a_{{n}+1}(k)\ , .\end{aligned}\ ] ] hence , it remains to compute the matrices @xmath106}(k)$ ] from the matrices @xmath107}(k)$ ] .
we give here a recursion formula for them .
we consider the relations @xmath112}(k ) & = & c^{[1]}(k)\,a^{[1]}(-k)\ , , \\
c^{[1]}(k)^\dag & = & c^{[1]}(-k)^t { { \hspace{4 mm}}\mbox{and}{\hspace{4 mm}}}c^{[1]}(k)\,c^{[1]}(-k)={\mbox{${\mathbb i}$}}\ , \\
a^{[2,{n}+1]}(k ) & = & c^{[2\ldots { n}+1]}(k)\,a^{[2,{n}+1]}(-k)\ , , \\ c^{[2\ldots { n}+1]}(k)^\dag & = & c^{[2\ldots { n}+1]}(-k)^t { { \hspace{4 mm}}\mbox{and}{\hspace{4 mm}}}c^{[2,{n}+1]}(k)\,c^{[2\ldots { n}+1]}(-k)={\mbox{${\mathbb i}$}}\ , , \nonumber\end{aligned}\ ] ] which amounts to take @xmath113 in the previous calculations .
this leads to @xmath114}(k ) & = & c^{[1\ldots { n}+1]}(k)\,a^{[1,{n}+1]}(-k)\ , , \end{aligned}\ ] ] or in components @xmath115}_{11}(k)\,a_{1}(-k ) + c^{[1\ldots { n}+1]}_{12}(k)\,a_{{n}+1}(-k)\ , , \\
a_{{n}+1}(k ) & = & c^{[1\ldots { n}+1]}_{21}(k)\,a_{1}(-k ) + c^{[1\ldots { n}+1]}_{22}(k)\,a_{{n}+1}(-k)\ , , \end{aligned}\ ] ] where @xmath116}_{11}(k ) & = & c^{[1]}_{11}(k ) + \frac{c^{[1]}_{12}(k)\,c^{[1]}_{21}(k)\,c^{[2\ldots { n}+1]}_{22}(k ) } { 1-c^{[1]}_{22}(k)\,c^{[2\ldots { n}+1]}_{11}(k ) } \ , , \\ c^{[1\ldots { n}+1]}_{12}(k ) & = & \frac{c^{[1]}_{12}(k)\,c^{[2\ldots { n}+1]}_{12}(k ) } { 1-c^{[1]}_{22}(k)\,c^{[2\ldots { n}+1]}_{11}(k ) } \ , , \\ c^{[1\ldots { n}+1]}_{21}(k ) & = & \frac{c^{[2\ldots { n}+1]}_{21}(k)\,c^{[1]}_{21}(k ) } { 1-c^{[1]}_{22}(k)\,c^{[2\ldots { n}+1]}_{11}(k)}\ , , \\ c^{[1\ldots { n}+1]}_{22}(k ) & = & c^{[2\ldots { n}+1]}_{22}(k ) + \frac{c^{[2\ldots { n}+1]}_{21}(k)\,c^{[2\ldots { n}+1]}_{12}(k)\,c^{[1]}_{22}(k ) } { 1-c^{[1]}_{22}(k)\,c^{[2\ldots { n}+1]}_{11}(-k)}\ , .\end{aligned}\ ] ] hence , an iterative use of these formulae , starting from the ` local ' coefficients @xmath107}(k)$ ] , allows to compute all the @xmath106}(k)$ ] .
alternative recursion formulas can be obtained particularizing the last defect instead of the first one .
one gets @xmath116}_{11}(k ) & = & c^{[1\ldots { n}]}_{11}(k ) + \frac{c^{[1\ldots { n}]}_{12}(k)\,c^{[1\ldots { n}]}_{21}(k)\,c^{[{n}+1]}_{22}(k ) } { 1-c^{[1\ldots { n}]}_{22}(k)\,c^{[{n}+1]}_{11}(k ) } \ , , \\
c^{[1\ldots { n}+1]}_{12}(k ) & = & \frac{c^{[1\ldots { n}]}_{12}(k)\,c^{[{n}+1]}_{12}(k ) } { 1-c^{[1\ldots { n}]}_{22}(k)\,c^{[{n}+1]}_{11}(k ) } \ , , \\ c^{[1\ldots { n}+1]}_{21}(k ) & = & \frac{c^{[{n}+1]}_{21}(k)\,c^{[1\ldots { n}]}_{21}(k ) } { 1-c^{[1\ldots { n}]}_{22}(k)\,c^{[{n}+1]}_{11}(k)}\ , , \\ c^{[1\ldots { n}+1]}_{22}(k ) & = & c^{[{n}+1]}_{22}(k ) + \frac{c^{[{n}+1]}_{21}(k)\,c^{[{n}+1]}_{12}(k)\,c^{[1\ldots { n}]}_{22}(k ) } { 1-c^{[1\ldots { n}]}_{22}(k)\,c^{[{n}+1]}_{11}(-k)}\ , .\end{aligned}\ ] ] note also that one can deduce physical properties directly from the recursion formulae .
in particular , from unitarity and hermiticity relations of @xmath117}(k)$ ] and @xmath118}(k)$ ] , it is easy to check that these relations are also obeyed by @xmath119}(k)$ ] : @xmath116}(k)\,c^{[1\ldots { n}+1]}(-k)={\mbox{${\mathbb i}$}}{{\hspace{4 mm}}\mbox{and}{\hspace{4 mm}}}c^{[1\ldots { n}+1]}(k)^\dag = c^{[1\ldots { n}+1]}(-k)^t \ , .\end{aligned}\ ] ] in the same way , one proves recursively that the transmission coefficients @xmath119}_{12}(k)$ ] and @xmath119}_{21}(k)$ ] depend only on the distances between the different defects , while the reflection coefficients obey @xmath120}_{11}(k)={\widetilde{c}}^{[1\ldots { n}+1]}_{11}(k)\,\e^{-\ri k\,y_{1 } } { { \hspace{4 mm}}\mbox{and}{\hspace{4 mm } } } c^{[1\ldots { n}+1]}_{22}(k)={\widetilde{c}}^{[1\ldots { n}+1]}_{22}(k)\,\e^{\ri k\,y_{{n}+1}}\ , , \ ] ] where the ` bare ' coefficients @xmath121}_{jj}(k)$ ] depend only on the distances between the different defects .
this section is devoted to some aspects of quantum field theory with defects on the line . in order to keep the discussion as simple as possible we consider one scalar field and @xmath122 defects .
the results however can be easily extended to more general cases using the results of section [ sec : ndef ] . in the fock representation @xmath123 of @xmath69
the basic correlation functions are @xcite @xmath124 \delta ( k - p ) + \cr_{ij}(k ) \delta ( k+p ) \right \}\ , , \quad \langle a^*_i(k ) a_j(p)\rangle = 0\ , , \label{fock1}\ ] ] where @xmath125 and @xmath85 and @xmath86 are given by ( [ r11]-[t31 ] ) , together with @xmath126 because of ( [ a2 ] , [ a*2 ] ) , the correlators involving @xmath65 and @xmath61 can be expressed in terms of ( [ fock1 ] ) .
now , using ( [ f1 ] ) one has @xmath127 which can be computed for any @xmath128 and have the following general form @xmath129\,\e^{\ri kx_{12 } } + \cr_{ij}(k)\ , \e^{\ri k\tx_{12}}\right \}\ , , \label{cf0}\ ] ] where @xmath130 , @xmath131 and @xmath132 .
we recall that the dispersion relation is @xmath133 for the correlators involving @xmath134 or @xmath113 , we use the expressions ( [ a2 ] , [ a*2 ] ) and after some algebra we find @xmath135}_{22}(-k){\widetilde{r}}^{[2]}_{22}(k)\,\e^{2\ri ky_{12 } } } { 1-{\widetilde{r}}^{[1]}_{22}(-k){\widetilde{r}}^{[2]}_{22}(k)\,\e^{2\ri ky_{12 } } } + \frac{{\widetilde{r}}^{[1]}_{22}(k){\widetilde{r}}^{[2]}_{22}(-k)\,\e^{-2\ri ky_{12 } } } { 1-{\widetilde{r}}^{[1]}_{22}(k){\widetilde{r}}^{[2]}_{22}(-k)\,\e^{-2\ri ky_{12 } } } \ , , \\
\cr_{22}(k)&= & \frac{{\widetilde{r}}^{[2]}_{22}(k)\,\e^{-2\ri ky_{2 } } } { 1-{\widetilde{r}}^{[1]}_{22}(-k){\widetilde{r}}^{[2]}_{22}(k)\,\e^{2\ri ky_{12 } } } + \frac{{\widetilde{r}}^{[1]}_{22}(k)\,\e^{-2\ri ky_{1 } } } { 1-{\widetilde{r}}^{[1]}_{22}(k){\widetilde{r}}^{[2]}_{22}(-k)\,\e^{-2\ri ky_{12 } } } \ , , \label{cf2}\end{aligned}\ ] ] @xmath136}_{12}(k)}{1-{\widetilde{r}}^{[1]}_{22}(-k){\widetilde{r}}^{[2]}_{22}(k)\,\e^{2\ri ky_{12 } } } \ , , \qquad \cr_{12}(k)= \frac{t^{[1]}_{12}(k){\widetilde{r}}^{[2]}_{22}(k)\,\e^{-2\ri ky_{2 } } } { 1-{\widetilde{r}}^{[1]}_{22}(-k){\widetilde{r}}^{[2]}_{22}(k)\,\e^{2\ri ky_{12}}}\ , , \label{cf4}\end{aligned}\ ] ] @xmath137}_{23}(k)}{1- { \widetilde{r}}^{[1]}_{22}(-k ) { \widetilde{r}}^{[2]}_{22}(k)\,\e^{2\ri ky_{12}}}\ , , \qquad \cr_{23}(k)=\frac{t^{[2]}_{23}(-k ) { \widetilde{r}}^{[1]}_{22}(k)\,\e^{-2\ri ky_{1 } } } { 1- { \widetilde{r}}^{[1]}_{22}(k){\widetilde{r}}^{[2]}_{22}(-k)\,\e^{-2\ri ky_{12}}}\ , , \label{cf6}\end{aligned}\ ] ] where we have used ` bare ' coefficients to make the @xmath138 dependence explicit .
the remaining coefficients are computed thanks to ( [ fock2 ] ) , that is to say @xmath139 the energy density of the field @xmath36 reads @xmath140 + m^2 \varphi^2 \right \}(t , x)\ , .
\label{cas1}\ ] ] one can compute the expectation value @xmath141 via point - splitting from the two - point function ( [ cf0 ] ) .
like in the standard casimir effect , this expectation value diverges at coinciding points .
this fact should not be surprising because it is the force , which is actually the physical observable in this context .
it is given by @xmath142 which determines a well - defined _ distribution _ in @xmath143 .
in general , @xmath144 may be singular at @xmath145 .
we will illustrate this feature below on the example of two delta defects . in order to investigate the casimir force at finite ( inverse ) temperature @xmath146
, we introduce the so called gibbs representation @xmath147 of @xmath69 . in contrast to the fock vacuum , the fundamental state of @xmath147 is not annihilated by @xmath148 and represents an appropriate idealization of a thermal bath which keeps the system in equilibrium . in @xmath147 one has @xcite @xmath149 \delta ( k - p ) + \cr_{ij}(k ) \delta ( k+p ) \right \}\ , , \label{gibbs1}\ ] ] @xmath150 \delta ( k - p ) + \cr_{ji}(-k ) \delta ( k+p ) \right \ } \ , , \label{gibbs2}\ ] ] where @xmath151 .
it is easy to show at this point that the thermal correlation functions @xmath152 are obtained by substitution @xmath153 in the integrand of the respective zero - temperature expressions ( [ cf0 ] ) . in this way one
now gets for the force @xmath154 showing an explicit temperature dependence . in this section
we consider a complex scalar field @xmath155 \ , , \ ] ] where both @xmath156 and @xmath157 are two copies of the scalar field with the same two defects constructed in section 3 above .
the theory admits now a @xmath158 symmetry leading to the conservation of the current @xmath159 \
, , \label{current}\ ] ] where @xmath160 stands for the normal product in the rt algebra @xmath69 .
we are interested in the charge density @xmath161 of the system in the gibbs state , defined in the previous subsection . using the thermal two - point function of @xmath162 ,
one obtains @xmath163 the density ( [ cd2 ] ) vanishes in the limit of zero temperature ( @xmath164 ) : the first term in the integrand gives the density at finite temperature , whereas the second one generates the correction due to the defects .
we consider the case of two defects defined by the boundary conditions @xmath165 { \partial}_x \varphi ( t,+y_{\alpha } ) - { \partial}_x \varphi ( t ,- y_{\alpha } ) = \mu_{\alpha}\,\varphi ( t ,- y_{\alpha})\ , , \end{array } \quad \alpha=1,2\ , , \quad \forall t\in { \mbox{${\mathbb r}$}}\ , .\ ] ] the local reflection and transmission coefficients have the form @xmath166}_{11}(k ) = { \widetilde{r}}^{[\alpha]}_{22}(-k ) = \frac{-i\mu_{\alpha}}{k+i\mu_{\alpha } } { { \hspace{4 mm}}\mbox{and}{\hspace{4 mm } } }
t^{[\alpha]}_{12}(k ) = t^{[\alpha]}_{21}(-k ) = \frac{k}{k+i\mu_{\alpha } } \,,\quad \alpha=1,2\,.\quad\ ] ] to avoid the presence of bound states we will assume that the product @xmath167 is strictly positive . in the computation of the casimir force we will also suppose for simplicity that @xmath168 . using eq . ( [ for - cas ] ) , one can compute the force on a test particle when @xmath169 , that is to say @xmath170 .
we get @xmath171 { \nonumber \\}&&\qquad\qquad\qquad\qquad\quad + \mu_1 ^ 2\,\sin(2k(x - y_{1}))+\mu_2 ^ 2\,\sin(2k(x - y_{2}))\big\}\ , , { \nonumber \\}g(k ) & = & k^2\big [ ( \mu_1+\mu_2)^2 - 4\mu_1\mu_2\,\sin^2(k\,y_{12})\big ] + 2k\,\mu_1\mu_2(\mu_1+\mu_2)\,\sin(2k\,y_{12}){\nonumber \\ } & & + 4\mu_1 ^ 2\mu_2 ^ 2\,\sin^2(k\,y_{12})\ , .
\nonumber\end{aligned}\ ] ] one can check that the integrand has no real poles , so that the divergent parts are obtain in the limit @xmath172 .
they can be computed analytically and we obtain : @xmath173 where we have introduced @xmath174+\mu_2\,\cos[2k(x - y_{2})]\big\}\ , , { \nonumber \\}\fs_{2}(x , y_{1},y_{2 } ) & = & \int_{0}^{+\infty}dk\,\frac{g(k)}{k^4+g(k)}\big\ { \mu_1 ^ 2\,\sin[2k(x - y_{1})]+\mu_2 ^ 2\,\sin[2k(x - y_{2})]\big\}\ , .
\nonumber\end{aligned}\ ] ] the integral @xmath175 has logarithmic singularities at @xmath176 , @xmath177 and @xmath178 , which can be extracted by means of the meijer s function @xmath179 we get @xmath180 the integrals @xmath181 and @xmath182 are absolutely convergent and can be computed numerically .
we represent the force for different values of @xmath183 and @xmath184 in figures [ fig : f2xmu2 ] and [ fig : f2xmu1 ] .
we find a point between the defects where the force changes sign . in this point
the test particle is in stable equilibrium .
we perform the same kind of calculation when @xmath185 , that is to say @xmath186 . extracting the poles ,
we get @xmath187 & & -\fc_{1}(x , y_{1},y_{2})-\fs_{1}(x , y_{1},y_{2})+ \int_{0}^{+\infty}dk\,\frac{k\,\fa_{1}(k)+\fa_{0}(k)}{k^4+g(k)}\ , , \end{aligned}\ ] ] where we have introduced : @xmath188+\mu_2\,\cos[2k(x - y_{2})]\big\}\ , , { \nonumber \\}\fs_{1}(x , y_{1},y_{2 } ) & = & \int_{0}^{+\infty}dk\,\frac{g(k)}{k^4+g(k)}\big\ { \mu_1 ^ 2\,\sin[2k(x - y_{1})]+(\mu_2 ^ 2 + 2\mu_1\mu_2)\,\sin[2k(x - y_{2})]\big\}\ , , \nonumber\end{aligned}\ ] ] @xmath189-\mu_2\cos[2k(x - y_{2 } ) ] + \mu_1\sin[2k(x - y_{1})]\,\sin(2k\,y_{12})\big\}\ , , { \nonumber \\}\fa_{0}(k ) & = & 2\mu_1 ^ 2\mu_2 ^ 2\,\sin[2k(x - y_{1})]\ , \big[1-\cos(2k\,y_{12})\big]\ , .\nonumber\end{aligned}\ ] ] again , one can extract from @xmath190 the meijer s contribution @xmath191 the remaining integrals are absolutely convergent and can be computed numerically .
we represent the force in figures [ fig : f1xmu2 ] and [ fig : f1xmu1 ] , which show that it is repulsive close to the defect at @xmath192 , but for @xmath193 there are also domains of attraction .
as already mentioned , the force is a distribution with singularities at the positions @xmath176 of the defects . to compute the force acting for instance at the defect in @xmath194
, one needs to sum the two contributions computed in sections [ sec : casmid ] and [ sec : caslft ] and take the limit @xmath195 , getting rid of the divergences .
this is achieved by taking the symmetric limit : @xmath196 one obtains the force as a function of the distance between the two defects : @xmath197{\nonumber \\}&&-2\mu_2\,\int_{0}^{+\infty}dk\ , \frac{k^{3}}{k^4+g(k)}\big[\mu_1\mu_2\cos(4ky_{12 } ) + ( \mu_1 ^ 2+\mu_2 ^ 2+\mu_1\mu_2)\cos(2ky_{12})\big]\ , .
\nonumber\end{aligned}\ ] ] using the meijer s function , one finds for the last integral @xmath198 where @xmath199 is absolutely convergent . computing the integrals numerically
, one gets the figures [ fig : f12xmu2 ] and [ fig : f12xmu1 ] .
the charged density for a complex field and two delta - defects is obtained from ( [ cd2 ] ) .
the integrals can be computed numerically , the mass @xmath200 and the temperature @xmath201 playing the role of regulators at @xmath202 and @xmath203 respectively .
we get @xmath204 with @xmath205 -k \mu_2 \sin [ 2 k ( x - y_2)]\big\}\,k^{3}{\nonumber \\ } & & + 2\,\big\{\mu_1 ^ 2\cos [ 2 k ( x - y_1)]+\mu_2 ^ 2 \cos [ 2 k ( x - y_2)]\big\}\,k^2\ , , { \nonumber \\}\fr_{1}(x ) & = & 2\,\big\{\mu_1 \sin [ 2 k ( x - y_1)]+\mu_2 \sin [ 2 k ( x - y_2)]\big\ } k^3{\nonumber \\}&&-2\,\big\{\mu_1 ^ 2\cos [ 2 k ( x - y_1 ) ] + \mu_2 ( 2 \mu_1+\mu_2 ) \cos [ 2 k ( x - y_2)]\big\ } k^2{\nonumber \\}&&-4\,\mu_1\mu_2\sin ( 2 k y_{12 } ) \,\big\ { \mu_1\cos [ 2 k ( x- y_1 ) ] + \mu_2 \cos [ k ( 2x - y_1-y_2)]\big\ } k{\nonumber \\}&&-4\,\mu_1 ^ 2\mu_2 ^ 2\,\big[1- \cos ( 2 k y_{12 } ) \big]\cos [ 2 k ( x - y_1)]\ , .
\end{aligned}\ ] ] and a similar expression for @xmath206 .
we remind that @xmath207 . when @xmath168 , there is a divergence at @xmath202 reflecting the possibility of bose condensation .
when @xmath208 ( and @xmath209 ) , the integral defining @xmath210 , @xmath211 or 3 , converges .
the same is true for @xmath212 and generic values of the parameters @xmath183 , @xmath184 , @xmath194 and @xmath213 .
however , when the relation @xmath214 is fulfilled , @xmath212 diverges ( see figure [ fig : d2 - 13 ] ) .
this relation corresponds to a change in the behavior of @xmath215 around @xmath202 : @xmath216 and could be interpreted as a kind of resonance effect .
one can check that , at this point in the parameter space , the integrals entering in @xmath210 , @xmath211 or 3 , as well as in the force studied in the previous sections , still converge .
the plots of @xmath217 are presented in figures [ fig : d1mul]-[fig : d2xmu021 ] .
+ @xmath218 @xmath219 + + + @xmath220 @xmath221 + + + @xmath222 @xmath223 +
in the present paper we generalized the rt algebra framework @xcite-@xcite for dealing with one defect to any number of defects .
more precisely , we considered a scalar field , which interacts with multiple defects on the line and freely propagates away of them .
we studied the most general point - like defects , parametrizing the interaction by local reflection and transmission coefficients .
as expected , the algebraic formulation considerably simplifies the boundary value problem at hand and is very efficient for deriving the correlation functions both at zero and finite temperature .
applied to the casimir effect , the algebraic technique leads to a nice and compact expression ( in terms of the reflection coefficients ) for the force . as an explicit illustration
, we derived the casimir force for two delta - type defects .
it turns out that both the intensity and the direction of the force depend on the position .
we established also the charge density distribution of a complex scalar field interacting with the defects .
our framework can be generalized in several directions
. one can apply it to the case where the field @xmath224 has @xmath225 internal degrees of freedom .
then , each of the operators @xmath226 , @xmath227 , becomes a vector of length @xmath228 , while the ` coefficients ' @xmath107}_{ij}$ ] , @xmath229 are now @xmath230 matrices .
the above results remain valid if these matrices commute among them .
another very attractive application concerns quantum graphs @xcite , which are networks of wires connected at nodes .
each node is characterized by a scattering matrix @xcite and can be viewed as a defect @xcite .
the rt algebra approach has been already applied @xcite to the study of quantum fields on graphs with one vertex ( star graphs ) .
using the results of this paper , one can formulate quantum field theory on a generic quantum graph and address in this framework the interesting problems of conductance and vacuum energy .
warmly thanks e. pilon and jp . guillet for useful advice on the analytical and numerical ( resp . )
estimates of the integrals computed in section [ sec : delta ] .
m. oshikawa , c. chamon and i. affleck , j. stat .
mech . * 0602 * ( 2006 ) p008 [ arxiv : cond - mat/0509675 ] .
e. fradkin and j. e. moore , phys . rev
. lett .
* 97 * ( 2006 ) 050404 [ arxiv : cond - mat/0605683 ] .
h. saleur , lectures on non - perturbative field theory and quantum impurity problems " , arxiv : cond - mat/9812110
. h. saleur , `` lectures on non - perturbative field theory and quantum impurity problems ii '' , arxiv : cond - mat/0007309 .
i. cherednik , int .
j. mod .
phys . a * 7 * ( 1992 ) 109 .
g. delfino , g. mussardo and p. simonetti , nucl .
b * 432 * ( 1994 ) 518 [ arxiv : hep - th/9409076 ] .
r. konik and a. leclair , nucl .
b * 538 * ( 1999 ) 587 [ arxiv : hep - th/9703085 ] .
o. castro - alvaredo and a. fring , nucl .
b * 649 * ( 2003 ) 449 [ arxiv : hep - th/0205076 ] .
p. bowcock , e. corrigan and c. zambon , int .
j. mod .
a * 19s2 * ( 2004 ) 82 [ arxiv : hep - th/0305022 ] .
p. bowcock , e. corrigan and c. zambon , jhep * 0401 * ( 2004 ) 056 [ arxiv : hep - th/0401020 ] .
m. mintchev , e. ragoucy and p. sorba , phys . lett .
b * 547 * ( 2002 ) 313 .
[ arxiv : hep - th/0209052 ] .
m. mintchev , e. ragoucy and p. sorba , j. phys .
a * 36 * ( 2003 ) 10407 [ arxiv : hep - th/0303187 ] . v. caudrelier , m. mintchev , e. ragoucy and p. sorba , j. phys . a * 38 * ( 2005 ) 3431 [ arxiv : hep - th/0412159 ] .
m. mintchev and p. sorba , jstat * 0407 * ( 2004 ) p001 [ arxiv : hep - th/0405264 ] .
m. mintchev and p. sorba , annales henri poincare * 7 * ( 2006 ) 1375 [ arxiv : hep - th/0511162 ] .
s. albeverio , l. dabrowski and p. kurasov , lett .
* 45 * ( 1998 ) 33 .
v. kostrykin and r. schrader , j. phys .
a * 32 * ( 1999 ) 595 .
b. bellazzini and m. mintchev , j. phys .
a * 39 * ( 2006 ) 11101 [ arxiv : hep - th/0605036 ] .
b. bellazzini , m. mintchev and p. sorba , j. phys . a * 40 * ( 2007 ) 2485 [ arxiv : hep - th/0611090 ] . | an algebraic framework for quantization in presence of arbitrary number of point - like defects on the line is developed .
we consider a scalar field which interacts with the defects and freely propagates away of them . as an application we compute the casimir force both at zero and finite temperature .
we derive also the charge density in the gibbs state of a complex scalar field with defects .
the example of two delta - defects is treated in detail . *
algebraic approach to multiple defects on the line + and application to casimir force * + m. mintchev@xmath0 and e. ragoucy@xmath1 + _ @xmath2 infn and dipartimento di fisica , universit di pisa , largo pontecorvo 3 , 56127 pisa , italy + @xmath3 lapth , 9 , chemin de bellevue , bp 110 , f-74941 annecy - le - vieux cedex , france _ |
let @xmath1 be a projective algebraic variety over the field of complex numbers .
given two coherent sheaves @xmath2 , on @xmath1 the algebraic group @xmath3 acts naturally on the affine space @xmath4 by @xmath5
. if two morphisms are in the same @xmath6orbit then they have isomorphic cokernels and kernels .
therefore it is natural to ask for good quotients of such actions in the sense of geometric invariant theory .
morphisms of type @xmath7 .
+ in general @xmath8 and @xmath9 will be decomposable such that @xmath6 is not reductive .
more specifically let @xmath8 and @xmath9 be direct sums @xmath10 where @xmath11 and @xmath12 are finite dimensional vector spaces and @xmath13 , @xmath14 are simple sheaves , i.e. their only endomorphisms are the homotheties , and such that @xmath15 for @xmath16 and @xmath17 for @xmath18 . in this case
we call homomorphisms @xmath19 of type @xmath7
. then the groups @xmath20 and @xmath21 can be viewed as groups of matrices of the following type .
the group @xmath22 , say , is the group of matrices @xmath23 where @xmath24 and @xmath25 . in the literature on moduli of vector bundles and coherent sheaves many quotients of spaces @xmath26 of type @xmath27 by the reductive group @xmath28
have been investigated , see for example @xcite , @xcite , @xcite , @xcite , @xcite .
the moduli spaces described in this way are the simplest ones , and this allows to test in these cases some conjectures that are expected to be true on more general moduli spaces of sheaves ( cf .
@xcite , @xcite ) .
we think that the moduli spaces of morphisms of type @xmath7 will be as useful to treat other less simple moduli problems of sheaves .
in fact , if one wants to use the spaces @xmath29 as parameter spaces for moduli spaces of sheaves , which are as close as possible to the moduli spaces , the higher types @xmath7 are unavoidable . the homomorphisms in a beilinson complex of a bundle on projective @xmath30space , for example , have in general arbitrary type @xmath7 depending on the dimensions of the cohomology spaces of the bundle . in several papers , see @xcite , @xcite for example , semi
stable sheaves or ideal sheaves of subschemes of projective spaces , are represented as quotients of injective morphisms of type @xmath7 , and one should expect that the moduli spaces of such sheaves are isomorphic to a good quotient of an open subset of the corresponding space of homomorphisms . in some cases of type @xmath31
this has been verified for semi
stable sheaves on @xmath32 in @xcite . in case of type @xmath7
there are good and projective quotients if one restricts the action to the reductive subgroup @xmath33 this has been shown recently by a. king in @xcite .
the quotient problem for @xmath29 of type @xmath7 with respect to the full group @xmath28 is however the generic one and indispensable .
unfortunately the by now standard geometric invariant theory ( git ) does nt provide a direct answer for these quotient problems in case @xmath34 is not reductive .
there are several papers dealing with the action of an arbitrary algebraic group like @xcite , @xcite , @xcite , @xcite and older ones , but their results are insufficient for the above problem .
the conditions of @xcite are close to what we need , but they do nt allow a concrete description of the set of semi
stable points in our case and they do nt guarantee good or projective quotients , see remark [ fauntleroy ] . the main idea +
our procedure is very close to standard git and we finally reduce the problem of the quotient to the one of a reductive group action . we introduce polarizations @xmath35 of tuples of rational numbers for the action of @xmath6 on the affine space in analogy to the ones of a. king in @xcite , which are refinements of the polarizations by ample line bundles on the projective space @xmath36 , and then introduce open sets @xmath37 of stable and semi stable points depending on @xmath38 and study the quotient problem for these open subsets .
there are chambers in @xmath39 such that the polarizations in one chamber define the same open set , in accordance with the chamber structure in neron
severi spaces of polarizations in the reductive case , see f.e .
@xcite i. dolgachev - y. hu and @xcite m. thaddeus .
however , in contrast to the reductive case , good quotients @xmath40 do nt exist for all polarizations , see [ prem_ex ] . as a main achievement we are providing numerical conditions on the polarizations , depending on the dimensions of the spaces @xmath11 and @xmath12 , under which such quotients exist .
the main step for that is to embed the group actions @xmath41 into an action @xmath42 of a reductive group @xmath43 and to compare the open sets @xmath44 and @xmath45 , where @xmath46 is a polarization for the @xmath43action associated to @xmath38 .
construction of quotients by non reductive groups .
+ to be more precise , a polarization @xmath38 is a tuple @xmath47 of positive rational numbers , called weights of the factors @xmath48 and @xmath49 respectively , which satisfy @xmath50 , where @xmath51 denote the dimensions of the spaces of the same name .
we use then the numerical criterion of a. king , @xcite , as definition for semi stability with respect to the reductive group @xmath52 .
an element @xmath53 is @xmath54stable if for any proper choice of subspaces @xmath55 such that @xmath56 maps @xmath57 into @xmath58 , we have @xmath59 , or semi stable if equality is allowed .
let @xmath60 denote the set of stable and semi stable points so defined .
if @xmath61 is the unipotent radical of @xmath6 , which is generated by the homomorphisms @xmath62 and @xmath63 for @xmath64 and @xmath65 , we say that @xmath56 is @xmath66stable if @xmath67 is @xmath54(semi)stable for any @xmath68 , see [ defstab ] .
we thus have open subsets @xmath69 and @xmath70 . _
the main result of our paper is that there are sufficient numerical and effective bounds for the polarizations @xmath38 such that @xmath44 admits a good and even projective quotient @xmath40 and that in addition @xmath71 admits a geometric quotient , which is smooth and quasi projective , see proposition [ goodqu ] and the results [ sone ] , [ equality ] , and section 8 .
_ the definitions of good and geometric quotients are recalled in [ constquot ] . by using correspondences between spaces of morphisms , called _ mutations _ , it is possible to deduce from our results other polarizations such that there exists a good projective quotient ( see @xcite , @xcite ) .
all this is achieved by embedding the action @xmath72 into an action @xmath73 of a reductive group and then imposing conditions for the equality @xmath74 , where @xmath46 is the associated polarization .
the quotient is then the quotient of the saturated subvariety @xmath75
. the quotient will be projective if @xmath76 does nt meet @xmath45 . also for this
, numerical conditions can be found in section 8 .
the idea of embedding the non reductive action @xmath72 into the action @xmath77 is simply to replace the @xmath13 by @xmath78 using the evaluation maps @xmath79 .
it is explained in [ motiv ] and [ remmotiv ] that this is the outcome when we start to replace the sheaves @xmath13 step by step and similarly for the sheaves @xmath14 .
since we have to deal everywhere with the dimensions of the vector spaces @xmath80 and @xmath81 which form the components of the unipotent group @xmath82 , we have translated the whole setup into an abstract multilinear setting and related actions by technical reasons .
this gives more general results although we have only applications in the theory of sheaves .
the reader should always keep in mind the motivation in [ motiv ] .
the results obtained in the simplest case ( morphisms of type ( 2,1 ) or ( 1,2 ) ) are stated in [ result-2 - 1 ] .
they are characteristic for the general case in which only the conditions are more complicated .
remark on finite generatedness + one would expect that the quotients of @xmath83 could be obtained by first forming the quotient @xmath84 with respect to the unipotent radical @xmath82 and then in a second step a quotient of @xmath84 by @xmath85 .
however , the actions of unipotent groups behave generally very badly , @xcite , and we are not able to prove that the algebra @xmath86 $ ] is finitely generated .
this would be an essential step in a direct construction of the quotient .
of course , the main difficulty also in this paper arises from the presence of the group @xmath82 .
the counterexample of m.nagata , @xcite , also shows that the finite generatedness depends on the dimensions of the problem .
so from a philosophical point of view we are determining bounds for the dimensions involved under which we can expect local affine @xmath6-invariant coordinate rings which are finitely generated , and thus to obtain good quasi projective quotients , even so the bounds might not be the best .
the simple examples [ prem_ex ] , [ remgeoqu ] show that a good quotient @xmath40 might not exist if the conditions are not fulfilled .
morphisms of type ( 2,1)[result-2 - 1 ] + in this case the homomorphisms of sheaves are of the type @xmath87 where we use the notation @xmath88 for @xmath89 . for this type
a polarization is given by a pair @xmath90 of positive rational numbers such that @xmath91 .
it is determined by the rational number @xmath92 which lies in @xmath93 $ ] .
writing @xmath94 for @xmath95 and @xmath96 for @xmath97 for the moment , our results depend on constants @xmath98 defined as follows : let @xmath99 be the linear map induced by the composition map @xmath100 , and @xmath101 let @xmath102 be the set of proper linear subspaces @xmath103 such that for every proper linear subspace @xmath104 , @xmath105 is not contained in @xmath106 .
let @xmath107 [ theo_main]*theorem : * there exists a good projective quotient @xmath108 and a geometric quotient @xmath109 if @xmath110 in the case of morphisms @xmath111 on projective spaces the constants have been computed in @xcite and we obtain the more explicit result : [ theo_main2]*theorem : * let @xmath112 be an integer .
there exists a good projective quotient @xmath108 and a geometric quotient @xmath109 in the case of morphisms @xmath113 on the projective space @xmath114 if @xmath115 t & > & \dsp\frac{(n+1)m_2 ^ 2(m_2 - 1)}{2n_1(m_2(n+1)-1)}\txte{if}2\leq m_2\leq n+1,\\[2.5ex ] t & > & \dsp\frac{(n+1)^2m_2}{2(n+2)n_1}\txte{if}m_2>n+1 .
\end{array}\ ] ] construction of fine moduli spaces of torsion free sheaves[int_fine ] + in section [ fine_mod ] we construct smooth projective fine moduli spaces of torsion free coherent sheaves on @xmath114 using morphisms @xmath116 ( for @xmath117 ) . more precisely we prove that for all polarizations , semi - stable morphisms are injective outside a closed subvariety of codimension @xmath118 , hence their cokernels are torsion free sheaves .
a generic morphism is injective and its cokernel is locally free . in this case
we can construct @xmath119\ ] ] distinct smooth projective moduli spaces @xmath120 of such morphisms , of dimension .
moreover , all the @xmath121 are birational to each other .
for @xmath122 , we construct a coherent sheaf @xmath13 on @xmath123 , flat over @xmath121 , such that for every closed point @xmath124 , @xmath125 is isomorphic to the cokernel of the morphism @xmath126 corresponding to @xmath127 .
we prove that @xmath121 is a fine moduli space of torsion free sheaves with universal sheaf @xmath13 .
in particular , this means that for every closed point @xmath128 , the kodara - spencer map @xmath129 is bijective , and for any two distinct closed points @xmath130 , the sheaves @xmath131 , @xmath132 are not isomorphic .
open problems[quest ] + even in the simplest case of morphisms of type ( 2,1 ) we do not know what all the polarizations are for which a good quotient @xmath133 exists .
more generally it would be interesting to find all the saturated open subsets @xmath134 of @xmath83 such that a good quotient ( quasiprojective or not ) exists , or all the open subsets @xmath134 such that a geometric quotient @xmath135 exists .
the corresponding problem for reductive groups has been studied in @xcite , 1.12 , 1.13 , and in @xcite , @xcite .
organization of the paper[org ] + in section 2 we describe our problem in terms of multilinear algebra .
in section 3 we recall results of a. king , @xcite .
the reductive group actions considered in this paper , the action of @xmath52 on @xmath83 and that of @xmath136 on @xmath137 , are particular cases of @xcite .
we also discuss the relation of @xmath38(semi)stability in @xmath83 with that in the projective space @xmath36 .
but we can not work solely on the projective niveau , because the embedding @xmath138 is not linear . after defining @xmath6(semi)stability for the non reductive group in section 4
we describe the embedding in section 5 and introduce the associated polarizations .
section 6 contains the step of constructing the quotient @xmath40 using the git
quotient @xmath139 of a. king .
sections 7 and 8 are the hard parts of the paper . here
the conditions of the weights which define good polarizations are derived .
it seems that the constants appearing in these estimates had not been considered before . in section 9
we are investigating a few examples in order to test the strength of the bounds . here
we restrict ourselves to small type @xmath140 in order to avoid long computations of the constants which give the bounds for the polarizations .
what we discover in varying the polarizations are flips between the moduli spaces , as one has to expect from the general results on the variation of linearizations of group actions , cf .
@xcite , @xcite , @xcite .
in example [ ex1 ] we have a very simple effect of a flip , but in example [ ex2gen ] the chambers of the polarizations look already very complicated . in section 10
we define new fine moduli spaces of torsion free sheaves using our moduli spaces of morphisms . acknowledgement .
the work on this paper was supported by dfg .
the first author wishes to thank the university of kaiserslautern , where the work was started , for its hospitality .
let @xmath141 and @xmath142 be semi
simple sheaves as in the introduction . in order to describe
the action of @xmath3 on @xmath4 in greater detail we use the abbreviations @xmath143 such that we are given the natural pairings @xmath144 the group @xmath6 consists now of pairs @xmath145 of matrices @xmath146 with diagonal elements @xmath147 and @xmath148 .
similarly a homomorphism @xmath149 is represented by a matrix @xmath150 of homomorphisms @xmath151 . using the natural pairings ,
the left action @xmath152 of @xmath6 on @xmath83 is described by the matrix product @xmath153 \vdots & & \vdots\\[1.5ex ] \varphi_{s1 } & \cdots & \varphi_{sr}\end{array}\right)\ \circ\ \left ( \begin{array}{cccc } g_1 & 0 & \cdots & 0\\ u_{21 } & g_2 & & \vdots\\ \vdots & \ddots & \ddots & 0\\ u_{r1 } & \cdots & u_{r , r-1 } & g_r \end{array } \right)^{-1}\raisebox{-5ex}{,}\ ] ] where the compositions @xmath154 and @xmath155 are compositions as sheaf homomorphisms but can also be interpreted as compositions induced by the pairings of the vector spaces above .
thus the group @xmath6 , the space @xmath83 and the action are already determined by the vector spaces @xmath156 and the pairings between them .
therefore , in the following we define @xmath157 and the actions @xmath72 by abstractly given vector spaces and pairings .
the resulting statements can then be applied to systems of sheaves by specifying the spaces as spaces of homomorphisms as above .
[ abstr]the abstract setting let @xmath158 be positive integers and let for @xmath159 finite dimensional vector spaces @xmath156 be given , where we assume that @xmath160 and @xmath161 .
moreover we suppose that we are given linear maps , called _ compositions _
, @xmath162 we assume that all these maps and the induced maps @xmath163 are * surjective*. this is the case when all the spaces are spaces of sheaf homomorphisms as above for which the sheaves @xmath13 and @xmath14 are line bundles on a projective space or each of them is a bundle @xmath164 .
we may and do assume that these pairings are the identities if @xmath165 , @xmath166 etc . . finally , we suppose that these maps verify the natural associative properties of compositions .
this means that the induced diagrams @xmath167\ar[d ] & a_{ki}\otimes a_{ih}\ar[d]\\ a_{kj}\ot a_{jh}\ar[r ] & a_{kh } } \ \ \ \ \ \xymatrix { b_{on}\ot b_{nm}\ot b_{ml}\ar[r]\ar[d ] & b_{om}\otimes b_{nl}\ar[d]\\ b_{on}\ot b_{nl}\ar[r ] & b_{ol } } \ ] ] @xmath168\ar[d ] & h_{lj}\ot a_{ji}\ar[d]\\ h_{lk}\ot a_{ki}\ar[r ] & h_{li } } \ \ \ \ \
\xymatrix { b_{nm}\ot b_{ml}\ot h_{li}\ar[r]\ar[d ] & b_{nl}\ot h_{li}\ar[d]\\ b_{nm}\ot h_{mi}\ar[r ] & h_{ni } } \ ] ] @xmath169\ar[d ] & h_{mj}\ot a_{ji}\ar[d]\\ b_{ml}\ot h_{li}\ar[r ] & h_{mi } } \ ] ] are commutative for all possible combinations of indices . in our setup
we also let finite dimensional vector spaces @xmath11 for @xmath170 and @xmath12 for @xmath171 be given and we consider finally the vector space @xmath172 where summation is over @xmath170 and @xmath171 .
this is the space of homomorphisms in the abstract setting . the group @xmath6 and its action on @xmath83 are now also given in the abstract setting as follows .
[ group]the group @xmath6 we define @xmath6 as a product @xmath173 of two groups where the left group @xmath174 replaces @xmath22 and the right group @xmath175 replaces @xmath21 in our motivation .
let @xmath174 be the set of matrices @xmath23 with @xmath176 and @xmath177 .
the group law in @xmath174 is now defined as matrix multiplication where we define the compositions @xmath178 naturally according to the given pairings as the composition @xmath179 explicitly , if @xmath180 has the entries @xmath181 and @xmath182 has the entries @xmath183 then the product @xmath184 in @xmath174 is defined as the matrix with the entries @xmath185 in the diagonal and @xmath186 for @xmath187 . the verification
that this defines a group structure on @xmath174 is now straightforward . as a set @xmath174 is the product of all the @xmath188 and all @xmath189 for @xmath64 and thus has the structure of an affine variety .
since multiplication is composed by a system of bilinear maps it is a morphism of affine varieties .
hence @xmath174 is naturally endowed with the structure of an algebraic group .
the group @xmath175 is now defined in the same way by replacing the spaces @xmath11 and @xmath190 by @xmath12 and @xmath191 . finally @xmath192 is defined as an algebraic group .
[ action]the action of @xmath6 and @xmath83 we will define a left action of @xmath175 and a right action of @xmath174 on @xmath83 such that the action of @xmath6 on @xmath83 can be defined by @xmath193 .
both actions are defined as matrix products as described above in the case of sheaf homomorphisms using the abstract compositions as in the definition of the group law .
if @xmath56 has the entries @xmath194 and @xmath195 has the entries @xmath196 and @xmath197 then @xmath198 is defined as the matrix product @xmath199 \vdots & & \vdots\\[1.6ex ] \varphi'_{s1 } & \cdots & \varphi_{sr } \end{array } \right)= \left ( \begin{array}{ccc } \varphi_{11 } & \cdots & \varphi_{1r}\\[1.6ex ] \vdots & & \vdots\\[1.6ex ] \varphi_{s1 } & \cdots & \varphi_{sr } \end{array } \right ) \
\left ( \begin{array}{cccc } g_1 & 0 & \cdots & 0\\ u_{21 } & g_2 & & \vdots\\ \vdots & \ddots & \ddots & 0\\ u_{r1 } & \cdots & u_{r , r-1 } & g_r \end{array}\right)\ ] ] with @xmath200 where @xmath201 is the composition @xmath202 or dually the composition @xmath203 the left action of @xmath175 is defined in the same way . in the next two sections we give an analysis of stability and semi - stability for the action of @xmath6 and its natural reductive subgroup @xmath52 . in the reductive case
this is due to a. king .
[ subgrp]canonical subgroups of @xmath6 we let @xmath204 and @xmath205 be the maximal normal unipotent subgroups of @xmath174 and @xmath175 defined by the condition that all @xmath206 and all @xmath207 .
then @xmath208 is a maximal normal unipotent subgroup of @xmath6 .
similarly we consider the reductive subgroups @xmath209 and @xmath210 of @xmath174 and @xmath175 defined by the conditions @xmath211 and @xmath212 for all indices .
then @xmath213 is a reductive subgroup of @xmath6 and it is easy to see that @xmath85 . the restricted action of @xmath52 is much simpler and reduces to the natural actions of @xmath188 on @xmath11 and @xmath214 on @xmath12
[ resking]results of a. king let @xmath215 be a finite set , @xmath216 a subset such that the union of the images of the two projections of @xmath217 is @xmath215 . for each @xmath218 ,
let @xmath219 be a positive integer , @xmath220 a vector space of dimension @xmath219 and for each @xmath221 , let @xmath222 be a finite dimensional nonzero vector space .
let @xmath223 on @xmath224 we have the following action of the reductive group @xmath225 arising naturally in this situation .
if @xmath226 and @xmath227 , then @xmath228 let @xmath229 be a sequence of integers such that @xmath230 to this sequence is associated the character @xmath231 of @xmath232 defined by @xmath233 this character is trivial on the canonical subgroup of @xmath232 isomorphic to @xmath234 ( for every @xmath235 , the element @xmath236 of @xmath232 corresponding to @xmath237 is such that @xmath238 for each @xmath239 ) .
this subgroup acts trivially on @xmath224 .
a point @xmath240 is called _ @xmath231-semi - stable _ if there exists an integer @xmath241 and a polynomial @xmath242 which is @xmath243-invariant and such that @xmath244 ( @xmath245 is called @xmath243-invariant if for every @xmath246 and @xmath247 we have @xmath248 ) . the point @xmath249 is called _ @xmath231-stable _ if moreover @xmath250 and if the action of @xmath232 on @xmath251 is closed .
a. king proves in @xcite the following results : 1 .
a point @xmath252 is @xmath231-semi - stable ( resp .
@xmath231-stable ) if and only if for each family @xmath253 of subspaces @xmath254 which is neither the trivial family ( 0 ) nor the given family @xmath255 and which satisfies @xmath256 for each @xmath221 , we have + @xmath257 2 .
let @xmath258 ( resp .
@xmath259 ) be the open subset of @xmath224 consisting of semi - stable ( resp stable ) points .
then there exists a good quotient @xmath260 by @xmath261 which is a projective variety .
the restriction of this quotient @xmath262 is a geometric quotient and @xmath263 is smooth .
[ polariz]polarizations the ( semi-)stable points of @xmath224 remain the same if we replace @xmath264 by @xmath265 , @xmath266 being a positive integer .
so the notion of ( semi-)stability is fully described by the reduced parameters @xmath267 , where @xmath268 so we can define the _ polarization _ of the action of @xmath232 on @xmath224 by any sequence @xmath269 of nonzero rational numbers such that @xmath270 by multiplying this sequence by the smallest common denominator of the @xmath271 we obtain a sequence @xmath264 of integers and the corresponding character of @xmath232 .
therefore the loci of stable and semi stable points of @xmath224 with respect to @xmath232 and a polarization @xmath272 are well defined and denoted by @xmath273 [ condstab]conditions imposed by the non - emptiness of the quotient if @xmath259 is not empty , the @xmath274 must satisfy some conditions .
we will derive this only in the three situations which occur in this paper .
polarizations satisfying these necessary conditions will be called _
proper_. the first is that of the action of @xmath52 in [ subgrp ] and the second is that of @xmath136 and @xmath137 in section 5 , and the third is the case in between occurring in [ equal1 ] .
[ first]first case let @xmath275 be positive integers .
we take @xmath276 this is the case of morphisms of type @xmath7 . for @xmath170 ,
let @xmath277 if @xmath278 , and @xmath279 otherwise , and for @xmath171 , let @xmath280 .
then if one @xmath281 is not positive , we have @xmath282 and @xmath283 , so in this case no point of @xmath224 is stable .
so we obtain , if @xmath259 is non - empty , the conditions @xmath284 a proper polarization is in this case a sequence @xmath285 of rational numbers such that the @xmath286 and the @xmath287 are positive and satisfy @xmath288 [ second]second case this case appears when we use a bigger reductive group to define the quotient ( this is the case of @xmath289 later on )
. let @xmath275 be positive integers .
here we take @xmath290 then the necessary conditions for @xmath259 to be non - empty are : @xmath291 to derive the first set of conditions we consider for any @xmath292 the family @xmath293 for which @xmath294 if @xmath295 and @xmath296 for all other @xmath297 .
then @xmath298 for any @xmath299 and any @xmath300 .
if @xmath245 is stable we obtain @xmath301 moreover , if the family @xmath293 is defined by @xmath294 for @xmath302 if @xmath303 and @xmath296 else , we obtain directly @xmath304 a proper polarization in this case is then a sequence @xmath305 of rational numbers satisfying @xmath306 and @xmath307 we could also drop the normalization condition .
[ third]third case this case is a combination of the first and second case .
it appears in the proof of the equivalence of semi stability in [ conv2 ] . here
@xmath215 is the same as in the previous cases and @xmath308 now the necessary conditions for @xmath259 to be non empty are : @xmath309 the first condition follows as in the second case when we consider the family @xmath293 with @xmath294 for @xmath310 and @xmath296 for all other @xmath297 .
the second condition follows when all @xmath311 are zero except @xmath280 for one @xmath312 .
again a proper polarization in this case is a sequence @xmath313 with @xmath314 and @xmath315 [ projact]the action of @xmath232 on @xmath316 we suppose that we are in one of the first two preceding cases and that there exist stable points in @xmath224 .
let @xmath317 be a nonzero homogeneous polynomial , @xmath243-invariant for some positive integer @xmath30 .
the @xmath243invariance implies that @xmath317 has degree @xmath318 where in case 1 ( action of @xmath52 on @xmath83 ) @xmath319 and in case 2 ( action of @xmath320 on @xmath289 ) @xmath321 to see this let @xmath322 and let @xmath180 be given by @xmath323 and @xmath324 in the first case and by @xmath325 and @xmath326 in the second case
. then @xmath327 and @xmath328 in both cases , such that @xmath329 .
now we will see that there exists a @xmath232-line bundle @xmath330 on @xmath316 such that the set @xmath258 of semi - stable points is exactly the set of points over @xmath331 , which is the set of semi - stable points in the sense of geometric invariant theory corresponding to @xmath332 cf .
@xcite , @xcite , @xcite . here
the action of @xmath232 on @xmath333 is the natural action multiplied by @xmath231 .
more precisely , the action of @xmath232 on @xmath224 induces an action of this group on @xmath334 and on @xmath335 by : @xmath336 for all @xmath247 , @xmath246 and @xmath337 , viewed as an homogeneous polynomial of degree @xmath338 on @xmath224 .
the line bundle space @xmath339 of @xmath330 is acted on by @xmath232 in the same way : if @xmath340 then @xmath341 is the form on @xmath342 given by @xmath343 .
we modify now the action of @xmath232 on @xmath339 ( resp .
@xmath344 ) by multiplying with @xmath345 : @xmath346
now @xmath347 is an invariant section if and only if @xmath317 is a homogeneous polynomial of degree @xmath348 which satisfies @xmath349 from the definition of semi - stable points in @xmath224 and @xmath316 with respect to the modified @xmath232-structure on @xmath350 , we get immediately [ sslocus]*lemma : * assume that @xmath351 and let @xmath338 be defined as above in the two cases of @xmath224 . then the set @xmath352 is the cone of the set @xmath353 as defined in g.i.t .
there are two definitions of stable points in @xmath316 , the classical one , given in @xcite , @xcite , and a more recent one , given in @xcite .
if we take d. mumford s definition , the cone of the set of stable points in @xmath316 does not coincide with @xmath259 because every point of @xmath316 has a stabilizer of positive dimension .
in fact there is a subgroup of @xmath354 of positive dimension which acts trivially on @xmath316 . in the first case for example such a group
is given by @xmath355 and @xmath356 with @xmath357 .
if we want to keep the coincidence between the sets of stable points for one and the same group , we would have to consider the action of a smaller reductive group in order to eliminate additional stabilizers .
we will do this in [ detgrp ] only in the first case . if we take the definition of v.l .
popov and e.g. vinberg , then we obtain that the set @xmath358 is exactly the cone of the set @xmath359 [ detgrp]the group @xmath360 let @xmath6 and @xmath83 be as in section 2 and let @xmath361 be a proper polarization as in [ first ] for the action of @xmath52 on @xmath83 .
it is then convenient to use the subgroup @xmath362 of @xmath52 consisting of elements @xmath363 satisfying @xmath364 we consider the action of @xmath362 on @xmath330 induced by the modified @xmath231-action of @xmath52 . now the set @xmath365 of @xmath231-stable points of @xmath83 is exactly the cone over the locus @xmath366 of stable points of @xmath367 in the sense of geometric invariant theory .
let @xmath6 and @xmath83 be as in section 2 .
a character @xmath231 on @xmath52 as in king s setup can be extended to a character of @xmath6 .
also the modified action of @xmath52 on @xmath330 can be extended to an action of @xmath6 .
let @xmath360 be the subgroup of @xmath6 defined by the same equations as for @xmath362 .
it contains @xmath82 and @xmath362 , and we have @xmath368 . in the case of the action of @xmath52 on @xmath83 a proper polarization
is given by a sequence @xmath369 of positive rational numbers such that @xmath370 more precisely , the polarization is exactly the sequence @xmath371 .
the parameter @xmath286 ( resp .
@xmath287 ) will be called the weight of the vector space @xmath11 ( resp .
@xmath12 ) .
we see that the dimension of the set of possible proper polarizations is @xmath372 .
let @xmath338 denote the smallest common denominator of the numbers @xmath286 and @xmath287 and @xmath231 the character of @xmath52 defined by the sequence of integers @xmath373 .
let @xmath374 as we have seen , if we consider the modified action of @xmath52 on @xmath330 , then the @xmath231-semi - stable points of @xmath83 are exactly those over the semi - stable points of @xmath367 in the sense of geometric invariant theory with respect to the action of @xmath375 on @xmath330 .
the @xmath376-invariant polynomials are the @xmath52-invariant sections of @xmath377 .
we are now going to define a notion of ( semi)stability for the points of @xmath83 with respect to the given action of the non reductive group @xmath6 .
let @xmath61 be the above unipotent group , see also [ subgrp ] .
[ defstab]*definition : * a point @xmath53 is called @xmath6semi stable ( resp .
@xmath6stable ) with respect to the ( proper ) polarization @xmath378 if every point of @xmath379 is @xmath52semi stable ( resp .
@xmath52stable ) with respect to this polarization .
we denote these sets by @xmath380 resp .
@xmath381 . for many of the quotient problems for the spaces of homomorphisms between @xmath382 and @xmath383 and their cokernel sheaves
this is a fruitful notion . in [ prem_ex ] we investigate an example with an explicit description of the open sets @xmath384 .
this example also shows that the existence of a good quotient depends on the choice of the polarization .
[ defexam]_situation for type ( 2,1 ) : _ + in the case of morphisms of type @xmath31 we have @xmath385 and the polarization is completely described by the single parameter @xmath92 .
we must have @xmath386 . a polarization such that there exists integers @xmath387 , @xmath388 , @xmath389 , with @xmath390 , @xmath391 , such that @xmath392 , @xmath393 are not both 0 , and that @xmath394 is called _
singular_. there are only finitely many singular polarizations , corresponding to the values @xmath395 of @xmath338 .
let @xmath396 , @xmath397 .
if @xmath38 , @xmath398 are polarizations corresponding to parameters @xmath338 , @xmath399 such that for some @xmath400 we have @xmath401 , then @xmath402 hence there are exactly @xmath403 notions of @xmath6-(semi-)stability in this case . moreover , if @xmath404 , @xmath405 and @xmath406 are relatively prime , and @xmath38 is a non singular polarization , we have @xmath407 . in the general case of morphisms of type @xmath7 , it is not difficult to see that there are only finitely many notions of @xmath6-(semi-)stability .
[ fauntleroy]*remark : * in @xcite semi - stability is defined as follows : a point @xmath53 is semi - stable if there exists a positive integer @xmath408 and a @xmath360-invariant section @xmath409 of @xmath410 such that @xmath411 ( there is also a condition on the action of @xmath82 ) .
it is clear that a semi - stable point in the sense of fauntleroy is also @xmath6-semi - stable with respect to @xmath371 .
it is proved in @xcite that there exists a _
categorical quotient _ of the open subset of semi - stable points in the sense of @xcite , but it is not clear that all @xmath6-semi - stable points are semi - stable .
moreover , in the general situation of @xcite there is no way to impose conditions which would imply that the categorical quotient is a good quotient or even projective . using definition [ defstab ] we are able to derive a criterion for the existence of a good and projective quotient of @xmath83 under the action of @xmath6 .
[ prem_ex]existence and non existence of good quotients , an example we show here that we can not expect that a good quotient @xmath412 will exist for any polarization @xmath38 .
we consider morphisms @xmath413 on @xmath32 .
there are 3 notions of stability in this case , two corresponding to non singular polarizations .
for one of the non singular polarizations the quotient @xmath414 exists and for the other we prove the inexistence of a good quotient @xmath415 .
let @xmath416 be a complex vector space of dimension 3 , and @xmath417 .
let on @xmath32 .
a polarization for the action of @xmath6 on @xmath83 is a triple @xmath418 of positive rational numbers such that @xmath419 . as in [ defexam ]
a polarization depends only on @xmath420 .
there is only one singular polarization , corresponding to @xmath421 .
hence if we consider only non singular polarizations there are only two notions of @xmath6-(semi-)stability , the first one corresponding to polarizations such that @xmath422 and the second to polarizations such that @xmath423 . in both cases semi
stable points are already stable .
we are going to show that in the first case @xmath71 has a geometric quotient which is projective and smooth and that in the second case @xmath71 does nt even admit a good quotient .
the elements @xmath424 and @xmath425 are written as matrices @xmath426 where @xmath427 , @xmath428 and @xmath429 .
the case @xmath430[case1 ] + in this case @xmath71 has a geometric quotient which is the universal cubic @xmath431 of the hilbert scheme of plane cubic curves in @xmath417 .
the quotient map is given by @xmath432 _ remark _ : if @xmath433 , then @xmath434 and the conditions of [ theo_main ] ( in the dual case @xmath435)for a good and projective quotient to exist in this case are satisfied .
the proof is done in several steps .
\(1 ) * claim 1 * : let @xmath424 be as above .
then 1 .
@xmath436 if and only if @xmath437 in @xmath438 and @xmath439 are not both zero . 2 .
@xmath440 if and only if @xmath437 and @xmath441 in @xmath442 .
_ proof of claim 1_. ( i ) follows easily from the criterion ( 1 ) in [ resking ] . as for ( ii )
let @xmath436 with @xmath443 .
then @xmath444 for any @xmath445 which implies that also @xmath446 .
let conversely @xmath447 .
then @xmath443 because otherwise there is a linear form @xmath448 with @xmath449 and @xmath450 and with @xmath451 the element @xmath452 is the matrix @xmath453 which is not in @xmath454 .
\(2 ) by the result of a. king in [ resking ] , ( 3 ) , there is a geometric quotient @xmath455 which is smooth and projective .
* claim 2 * : @xmath456 . here
@xmath457(1 ) is the dual of the tautological quotient bundle over @xmath458 .
( the dimension of this quotient variety is 13 while @xmath459 and @xmath460 . to verify claim 2 we consider the map @xmath461 from @xmath454 to @xmath462 where we identify @xmath463 with @xmath458 via @xmath464
. then each @xmath465 because @xmath466 is the subspace of forms vanishing in @xmath467 .
it follows immediately that @xmath239 is a morphism @xmath468 which is surjective and @xmath52equivariant .
it induces a morphism of the geometric quotient to @xmath469 which is even bijective . since both ,
the quotient and the target are smooth , this is an isomorphism .
\(3 ) since @xmath470 we have an induced homomorphism @xmath471 .
it is the middle part of the canonical exact sequence @xmath472 of vector bundles on @xmath458 .
let @xmath473 be the kernel of @xmath474 .
from the left part of the sequence we obtain the affine bundle @xmath475 here @xmath476 is nothing but the universal cubic and the fibres of @xmath477 are isomorphic to @xmath478 .
* claim 3 * : @xmath479 is the inverse image of @xmath480 under @xmath239 and @xmath481 is a geometric quotient with respect to @xmath52 .
_ proof of claim 3_. @xmath482 belongs to @xmath483 if and only if @xmath484 , see ( ii ) of claim 1 .
\(4 ) let now @xmath485 be the morphism @xmath486 , given by @xmath487 , where @xmath488 .
it is obviously @xmath6equivariant and its fibres coincide with the @xmath6orbits . since @xmath239 is a geometric quotient and @xmath477
is an affine bundle , then @xmath489 is also a geometric quotient . * remark * : the variety @xmath490 is isomorphic to the moduli space @xmath491 of stable coherent sheaves on @xmath32 with hilbert polynomial @xmath492 .
this had been verified by j. le potier in @xcite .
the space @xmath71 is a natural parametrization of @xmath493 because any @xmath494 can be presented in an extension sequence @xmath495 where @xmath496 is the cubic curve supporting @xmath9 and @xmath497 , and then @xmath9 has a resolution @xmath498 this resolution is the beilinson resolution as can easily be verified .
moreover , @xmath249 is @xmath66stable if and only if @xmath9 is stable .
( if @xmath499 is a smooth point of @xmath496 , then @xmath9 is the line bundle @xmath500 and if @xmath499 is a singular point of @xmath496 , then @xmath9 is the unique cohen
macaulay module on @xmath496 with the given polynomial ) .
there is an obvious universal family @xmath501 on @xmath502 which defines a @xmath6equivariant morphism @xmath503 and then a bijective morphism @xmath504 , which by smoothness , is an isomorphism .
one knows that @xmath493 carries a universal family @xmath8 .
this family can be obtained as the non trivial extension @xmath505 where @xmath506 and @xmath507 or can be obtained as the descent of the family @xmath501 .
more details can be found in @xcite .
the case @xmath508[case2 ] + we suppose now that the polarization @xmath38 is such that @xmath423 . in this case
an element @xmath249 of @xmath83 is @xmath6-stable if and only if @xmath509 , @xmath510 are not both zero , and if for every @xmath511 , and are linearly independent . * proposition:*[nonex ] for this polarization there does not exist a good quotient @xmath512 let @xmath509 be a non - zero element of @xmath513 , let , and let @xmath424 be the matrix @xmath514 then @xmath249 is stable .
* claim :* _ the orbit @xmath515 is closed and if @xmath516 is such that @xmath517 meets @xmath515 , then @xmath518 . _ before proving the claim , we will show that it implies proposition [ nonex ] .
the stabilizer of a generic point in @xmath381 is isomorphic to @xmath234 : it consists of pairs of homotheties @xmath519 .
it follows that if exists , then all the fibers of the quotient morphism are of dimension at least .
now suppose that the claim is true .
then this implies that .
but the stabilizer @xmath520 of @xmath249 has dimension 2 : it consists of pairs @xmath521 with @xmath522 , @xmath523 , and hence has dimension 2 .
it follows that , a contradiction . _ proof of the claim .
_ let @xmath516 such that @xmath524 .
let @xmath525 then @xmath509 is contained in the vector space spanned by @xmath127 and @xmath510 .
hence by replacing @xmath526 with an element of @xmath527 we can assume that @xmath528 and that @xmath529 if @xmath510 is a multiple of @xmath509 . according to lemma [ util_lemm ] there exists a smooth irreducible curve @xmath496 , @xmath530 , and a morphism @xmath531 such that @xmath532 & w \\ { \ \ \ \ \ \ \ \ \ } t \fmaps[r ] & \theta(t)y \\ } \ ] ] can be extended to , with .
we can write , for , @xmath533 with @xmath534 @xmath535 where @xmath237 , @xmath536 , @xmath537 , @xmath266 , @xmath538 are morphisms and .
the morphisms @xmath237 , @xmath536 , @xmath537 , @xmath266 , @xmath538 can be extended to morphisms , denoted by @xmath539 , @xmath540 , @xmath541 , @xmath542 , @xmath543 respectively , and @xmath544 extends to .
now we use the fact that @xmath545 is defined at @xmath546 .
the first consequence is that , , and if @xmath547 then @xmath541 and @xmath543 also vanish at @xmath546 .
the second is that the morphisms can be extended to , and we have , .
we will now consider three cases : , , .
suppose that .
if @xmath547 , then @xmath548 implies that @xmath549 is a multiple of @xmath509 , but this is not true . if @xmath529 then @xmath550 implies that @xmath551 is a multiple of @xmath552 and @xmath548 implies then that @xmath553 is also a multiple of @xmath509 , which is not true
hence we can not have .
suppose that .
if @xmath547 , then @xmath548 implies that @xmath554 & s^2v^ * \\ { \ \ \ \ \ \ \ \ \ } t \fmaps[r ] & a(t)q_1+b(t)q_2+u(t)z_1 \\ } \ ] ] and @xmath555 & s^2v^ * \\ { \ \ \ \ \ \ \ \ \ } t \fmaps[r ] & c(t)q_1+d(t)q_2+u(t)z_1 \\ } \ ] ] extend to morphisms @xmath556 which vanish at @xmath546 .
it follows from the fact that that , and that . since ( by @xmath6-stability of @xmath526 )
, we have .
but since , this contradicts the fact that .
if @xmath529 then we deduce from the fact that that , which contradicts the @xmath6-stability of @xmath526 .
it follows that we have . if @xmath547 , using the fact that and we see that and that , which contradicts the fact that .
hence we have @xmath529 .
it follows from @xmath550 that and that .
by @xmath548 we see that @xmath557 & s^2v^ * \\ { \ \ \ \ \ \ \ \ \ } t \fmaps[r ] & b(t)q_2+u(t)z_1 \\ } \ ] ] extends to @xmath496 and that @xmath558 we have , if @xmath559 @xmath560 it follows that @xmath561 is a multiple of @xmath509 : . we have then @xmath562 and @xmath563 as claimed .
it remains to show that @xmath515 is closed .
this can be proved easily by computing the stabilizers of all the points in @xmath381 .
we see then that @xmath520 has the maximal possible dimension , hence @xmath515 is closed .
we now give a proof of the lemma used in the preceeding proposition : [ util_lemm]*lemma :* let @xmath83 be a finite dimensional vector space , @xmath6 a linear algebraic group acting algebraically on @xmath83 , @xmath564 and .
then there exists a smooth curve @xmath496 , and a morphism @xmath531 such that the morphism @xmath532 & w \\ { \ \ \ \ \ \ \ \ \ } t \fmaps[r ] & \theta(t)y \\ } \ ] ] extends to and that .
let @xmath565 , @xmath566 .
the generic @xmath567-dimensional affine subspace through @xmath249 meets on a curve , and meets in a finite number of points .
hence we can find a curve that meets only at @xmath249 .
taking the normalization of @xmath1 and substracting a finite number of points or unnecessary components if needed , we obtain a morphism ( where @xmath490 is a smooth curve ) and a point such that and .
consider now the restriction of @xmath239 @xmath568 there exists a smooth curve @xmath569 and an etale surjective morphism such that the principal @xmath570-bundle on @xmath569 is locally trivial . by considering completions @xmath571 , @xmath572 of @xmath569 , @xmath490 and an extension of @xmath573 to a morphism
we obtain a smooth curve @xmath574 , @xmath575 and a morphism such that and that the principal @xmath570-bundle is defined on and locally trivial .
let @xmath576 be a nonempty open subset such that we have a @xmath570-isomorphism @xmath577 then we can take , , and for , we have @xmath578 where @xmath579 is the canonical morphism @xmath580 .
more general counterexamples of inexistence of geometric quotients[remgeoqu ] let @xmath83 be the space of homomorphisms @xmath581 over @xmath114 and let the homomorphism @xmath582 be given by the matrix @xmath583 where the @xmath584 are homogeneous coordinates .
the stabilizer of @xmath585 contains @xmath586 and the pairs @xmath587 in @xmath588 and thus has dimension @xmath118 . if @xmath589 is a polarization with @xmath590 , then it is easy to see that @xmath585 is @xmath38stable in the sense of [ defstab ] .
for example @xmath591 is the dimension vector of a @xmath585invariant choice of subspaces with @xmath592 .
there are however stable homomorphisms @xmath593 with stabilizer @xmath586 .
therefore @xmath594 can never admit the structure of a geometric quotient .
we will see in [ sone ] that a sufficient condition for that in the case of this @xmath83 is @xmath595 or @xmath596 because @xmath597 .
we will construct an algebraic reductive group @xmath320 , a finite dimensional vector space @xmath289 on which @xmath320 acts algebraically , and an injective morphism @xmath598 compatible with a morphism of groups @xmath599 the traces of @xmath320-orbits on @xmath600 will be exactly the @xmath6-orbits .
the space @xmath289 is of the same type as those studied in [ resking ] .
we will associate naturally to any polarization of the action of @xmath6 on @xmath83 a character @xmath231 of @xmath601 , i.e. a polarization of the action of @xmath320 on @xmath289 .
we will prove that in certain cases a point @xmath56 of @xmath83 is @xmath602(semi-)stable with respect to the given polarization if and only if @xmath603 is @xmath231-(semi-)stable with respect to the associated polarization .
the existence of a good and projective quotient of the open set of @xmath6-semi - stable points will follow from this .
[ motiv]motivation in terms of sheaves the idea for the embedding of @xmath83 into a space @xmath604 with a reductive group action is to replace the sheaves @xmath13 in @xmath605 by @xmath606 and dually the sheaves @xmath14 in @xmath607 by @xmath608 and then to consider the induced composed homomorphisms @xmath609 for @xmath610 @xmath611 in the bigger space @xmath612 of all homomorphisms between @xmath613 and @xmath614 .
this space is naturally acted on by the reductive group @xmath615 however it is not suitable enough for our purpose by two reasons .
it does not allow enough polarizations as in section 3 for direct sums in order to have consistency of ( semi-)stability and , secondly the group actions @xmath72 and @xmath616 do nt have consistent orbits .
both insufficiencies are however eliminated when we consider the following enlargement of @xmath612 .
we set @xmath617 and introduce the auxiliary spaces @xmath618 and define @xmath619 there are distinguished elements @xmath620 whose components are the natural composition maps .
the embedding of @xmath83 into @xmath621 will be defined as the affine map @xmath622 where @xmath609 is the above composition for a given @xmath623 the components of @xmath624 and @xmath625 will guarantee a compatible action of a reductive group and at the same time the possibility of choosing enough polarizations for this action .
[ remmotiv]*remark * : one might hope to be able to do induction on @xmath626 and/or @xmath409 by simply replacing @xmath627 by @xmath628 and keeping the other @xmath13 for @xmath629 .
but then we drop the information about the homomorphisms @xmath630 .
therefore we are lead to replace all @xmath631 , by @xmath78 at a time , i.e. by @xmath632 where @xmath633 .
moreover , in order to keep the information of the homomorphisms @xmath634 for @xmath635 we consider also the spaces @xmath636 together with the maps @xmath637 in the following . the reader may convince himself that only because of this the actions of the original group is compatible with the action of the bigger reductive group .
it is a beautiful outcome that then we are able to compare the semi stability with respect to related polarizations in section 7 .
[ defbigw]the abstract definition of @xmath621 the above motivating definition of the space @xmath621 can immediately be turned into the following final definition using the spaces @xmath638 and the pairings between them . for any possible @xmath292 and @xmath312
we introduce the spaces @xmath639 and we denote by @xmath640 and @xmath641 their dimensions . for @xmath642 and @xmath643
we let @xmath644 be the canonical morphisms , defined as follows . on the component @xmath645 of @xmath646 ,
the map @xmath647 is the map @xmath648 induced by the composition map of the spaces @xmath649 .
the map @xmath650 is defined in the same way . as in [ motiv ]
we set @xmath651 and @xmath652 in order to define the embedding @xmath653 we define the operator @xmath654 as follows . given @xmath655 ,
we let @xmath656 be the linear map defined by the matrix @xmath657 , for which each @xmath658 is the composed linear map @xmath659 where the first map is induced by @xmath660 , the second by the composition @xmath661 and the third by the dual composition @xmath662 the map @xmath653 can now be defined by @xmath663 [ inject]*lemma : * the linear map @xmath654 is injective and hence the morphism @xmath653 is a closed embedding of affine schemes . from the surjectivity assumptions in [ abstr ] we find that dually the composition @xmath664 is injective .
now it follows from the definition of @xmath658 that @xmath660 can be recovered from @xmath658 , by shifting @xmath665 to its dual .
[ bigg]the new group @xmath43 we consider now the natural action on @xmath621 as described in [ resking ] in the general situation , where the group is @xmath666 to be precise , this action is described in components by @xmath667 with @xmath668 and with @xmath669 the first and third expression describe the natural actions of @xmath670 on @xmath671 and of @xmath672 on @xmath673 there are also natural embeddings of @xmath674 into @xmath675 respectively . for
that it is enough to describe the embedding of @xmath676 in @xmath670 . given an element @xmath677 @xmath678 with @xmath679 we define @xmath680 as the matrix @xmath681 with respect to the decomposition of @xmath646 with the following components : @xmath682 on @xmath683 and for @xmath684 the map @xmath685 is the composition @xmath686 where the second arrow is induced by the given pairing . in case
@xmath687 we have @xmath688 and @xmath689 .
now we define the map @xmath690 it is then easy to verify that @xmath691 is an injective group homomorphism and defines a closed embedding of algebraic groups . with this embedding
we consider @xmath676 as a closed subgroup of @xmath670 . in the same way we obtain a closed embedding @xmath692 of @xmath693 .
finally we obtain the closed embedding @xmath694 of @xmath695 .
[ stabilizer]*lemma : * the subgroup @xmath696 ( respectively @xmath697 ) is the stabilizer of the distinguished element @xmath698 ( respectively @xmath699 ) it is enough to prove the statement only for @xmath676 because of duality . the fact that @xmath174 stabilizes @xmath700 is an easy consequence of the properties of the composition maps .
the converse can be proved by induction on @xmath626 .
it is trivial for @xmath701 .
suppose that @xmath702 and that the statement is true for @xmath703 .
let @xmath704 be an element of the stabilizer of @xmath700 .
when we replace the space @xmath83 by @xmath705 , corresponding to the spaces @xmath706 and the same spaces @xmath707 and similarly @xmath671 by @xmath708 , then @xmath709 is an element of the stabilizer of @xmath710 , so by the induction hypothesis it belongs to @xmath711 and there exists an element @xmath712 such that @xmath713 .
let now @xmath714 have the components @xmath715 the identity @xmath716 then shows that @xmath717 for @xmath718 for @xmath719 and @xmath720 for @xmath721 .
now let @xmath722 , @xmath723 , for @xmath724 , which are linear mappings @xmath725 .
then @xmath726 is an element of @xmath174 and we have @xmath727 * remark * : since the action of @xmath728 on @xmath624 is linear , it is clear that we have an isomorphism @xmath729 we will use this fact in section 8 .
using the associativity of the composition maps it is again easy to verify that the actions of @xmath6 on @xmath83 and @xmath43 on @xmath621 are compatible , i.e. that the diagram @xmath730 is commutative , in which the horizontal maps are the actions .
in addition we have the [ orbits]*corollary : * let @xmath731 .
then @xmath56 and @xmath732 are in the same @xmath6-orbit in @xmath83 if and only if @xmath603 and @xmath733 are in the same @xmath320-orbit in @xmath289 .
it follows from the compatibility of the actions that if @xmath734 in @xmath83 then also @xmath735 in @xmath621 by the last diagram .
conversely , if @xmath736 and @xmath737 then g stabilizes @xmath738 by the definition of @xmath653 in [ defbigw ] . by lemma [ stabilizer ] @xmath739 .
[ asspolar]the associated polarization in [ first ] and [ second ] we had introduced polarizations for the different types of actions of @xmath52 on @xmath83 and of @xmath43 on @xmath621 . in the following
we will describe polarizations on @xmath83 and @xmath621 which are compatible with the morphism @xmath740 .
their weight vectors are related by the following matrix equations and determine each other .
the entries of the matrices are just the dimensions of the spaces @xmath190 and @xmath191 . in the sequel
we will use the following * notation * : the dimension of a vector space will be the small version of its name .
so @xmath741 etc .
a proper polarization of the action of @xmath6 on @xmath83 is a tuple @xmath361 , where @xmath286 and @xmath287 are positive rational numbers such that @xmath742 we define the new sequence of rational numbers @xmath743 by the conditions @xmath744 then we have @xmath745 in particular the tuple @xmath746 is a polarization on @xmath621 such that @xmath747 is the weight of @xmath748 and @xmath749 the weight of @xmath750 .
it is called the _ associated polarization _ on @xmath621 .
it is compatible with @xmath653 in the following sense : if @xmath751 , and @xmath752 are linear subspaces , and if the subspaces of @xmath646 and @xmath753 are defined by @xmath754 respectively then we have @xmath755 if the set of stable points in @xmath289 with respect to the associated polarization is non - empty then by [ second ] the weights satisfy the conditions @xmath756 equivalently the conditions may also be written as @xmath757 substituting the weights of the original polarization on @xmath83 , we can reformulate these conditions . in the cases treated in the examples they reduce to the following [ weightcond]weight conditions .
let @xmath83 be of type @xmath7 and let @xmath361 be a proper polarization of @xmath83 with positive @xmath286 and @xmath287 .
if the set @xmath758 of stable points of @xmath621 with respect to the associated polarization @xmath759 is non - empty , then in case of type @xmath31:@xmath760 , type @xmath761 : @xmath762 , type @xmath763:@xmath764 .
[ invpoly]comparison of invariant polynomials in the following we assume that @xmath765 is the polarization on @xmath621 associated to the polarization @xmath361 . the semi stable locus @xmath45 with respect to this polarization is more precisely defined by the character @xmath766 associated to it as in [ resking ] .
if @xmath553 is lowest common denominator of @xmath743 , we have @xmath767 for an element @xmath736 with components @xmath768 and @xmath769 . by the matrix relations between the polarizations
@xmath553 is also a common denominator of @xmath369 , such that , if @xmath499 denotes the lowest , we have @xmath770 for some @xmath544 . the character @xmath231 with respect to the given polarization
can be defined by @xmath771 where the @xmath196 resp .
@xmath772 are the diagonal components of @xmath180 resp .
@xmath773 , see [ group ] .
now the relations between the polarizations imply by a straightforward calculation that @xmath774 if f is a @xmath775-invariant polynomial on @xmath621 it follows that @xmath776 i.e. that @xmath777 is a @xmath778-invariant polynomial on @xmath83 . as a consequence
we obtain the [ ssrelat]*lemma : * @xmath779 , i.e. if @xmath53 and @xmath603 is @xmath43-semi - stable in @xmath621 with respect to the polarization @xmath765 then @xmath56 is @xmath6-semi - stable in @xmath83 with respect to the polarization @xmath361 ( in the sense of [ defstab ] ) .
there exists a @xmath775-invariant polynomial f on @xmath621 such that @xmath780 .
then @xmath781 for any element @xmath782 in the unipotent subgroup @xmath61 .
this means that w is @xmath6-semi - stable .
[ remdet]*remark : * when we consider the subgroup @xmath783 defined by the condition @xmath784 we have @xmath785 as follows from the definition of @xmath360 in [ detgrp ] . with respect to these groups
the semi - stable points are those over the semi - stable loci in @xmath367 resp .
@xmath786 , with respect to the line bundles @xmath787 where @xmath338 and @xmath788 is defined as in [ projact ] in the different cases endowed with the modified action defined by the characters
. however , we can not compare @xmath367 and @xmath786 directly because the morphism @xmath653 does not descend .
we need the analogous statement of lemma [ ssrelat ] also in the case of stable points . for
that is is more convenient to use the subspace criterion ( 1 ) of a. king in the case of @xmath52 and @xmath43 .
this gives also another proof in the semi
stable case .
[ srelat ] * lemma : * with the same notation as in the previous lemma @xmath789
let @xmath790 be a point of @xmath83 with maps @xmath791 and suppose that @xmath56 is not @xmath6stable with respect to the polarization @xmath38 .
we can assume that it is not @xmath52stable , too .
then there are linear subspaces @xmath792 and @xmath752 for all @xmath292 and @xmath312 such that the family @xmath793 is proper and such that @xmath794 with these subspaces we can introduce the subspaces @xmath795 and @xmath796 as @xmath797 they form a proper family of subspaces and satisfy @xmath798 for the possible values of @xmath292 and @xmath312 .
but by the definition of the spaces and because @xmath46 is the associated polarization , the formulas of [ asspolar ] imply the dimension formula @xmath799 this states that also @xmath603 is not @xmath43stable . in section 7 we will derive sufficient conditions for the equality @xmath800 in the following section
we show how this equality implies the existence of a good and projective quotient @xmath40 using the result for @xmath801 from geometric invariant theory .
we keep the notation of the previous sections and let @xmath46 be the polarization on @xmath621 associated to the polarization @xmath38 on @xmath83 .
we do not require that they are proper here , but we will do that later for the examples .
in addition we introduce the saturation @xmath802 of the image of @xmath83 with respect to the action of @xmath43 .
construction of the quotient[constquot ] [ goodqu ] * proposition * : let @xmath83 and @xmath621 together with their @xmath6 and @xmath43structure be as in section 2 and 5 , let @xmath38 be a polarization for @xmath803 and @xmath46 be the associated polarization for @xmath804 .
\(1 ) if@xmath805 , then there exists a geometric quotient @xmath806 of @xmath807 by @xmath6 , which is a quasi projective nonsingular variety .
\(2 ) if in addition @xmath808and @xmath809 , then there exists a good quotient @xmath810 , such that @xmath493 is a normal projective variety , @xmath263 is an open subset of @xmath493 , and @xmath811 is the restriction of @xmath489 .
we recall here the definition of a good and a geometric quotient of c.s .
seshadri , see @xcite , @xcite .
let an algebraic group @xmath6 act on an algebraic variety or algebraic scheme @xmath1 .
then a pair @xmath812 of a variety and a morphism @xmath813 is called a good quotient if 1 .
@xmath814 is @xmath6equivariant ( for the trivial action of @xmath6 on @xmath574 ) , 2 .
@xmath814 is affine and surjective , 3 .
if @xmath134 is an open subset of @xmath574 then @xmath815 is an isomorphism @xmath816 , where the latter denotes the ring of @xmath6invariant functions , 4 .
if @xmath817 are disjoint closed and @xmath6invariant subvarieties of @xmath1 then @xmath818 are closed and disjoint . if in addition the fibres of @xmath814 are the orbits of the action and all have the same dimension , the quotient @xmath812 is called a geometric quotient . as usual
we write @xmath819 for a good quotient space and @xmath820 for a geometric quotient space .
we will prove the second statement first , assuming that the conditions of ( 1 ) and ( 2 ) are satisfied .
we use the abbreviations @xmath821 and similarly @xmath822 for the subsets of the stable points . by the result of a. king , [ resking ]
, there exists a good projective quotient of @xmath823 by the reductive group @xmath43 .
so there exists also a good and projective quotient of the closed invariant subvariety @xmath824 which we denote by @xmath825 by assumption ( 2 ) @xmath826 .
we let @xmath489 be the composition @xmath827 we know already that @xmath493 is projective .
we will then verify that @xmath828 is the good quotient of the proposition .
we consider first the commutative diagram @xmath829 ^ -{\mu}\ar[d]^p & \bg\zeta(w^{ss})\ar[d]^{\pi_0}\\ w^{ss}\ar[r]^-{\pi } & m } \ ] ] in which @xmath499 is the projection and @xmath830 is defined by @xmath831 .
there is an action of @xmath6 on @xmath832 by @xmath833 and it follows that @xmath830 is @xmath6equivariant . * claim * : _ the morphism @xmath830 is a geometric quotient of @xmath834 by @xmath6 . _ _ proof of the claim : _ we show first that the fibres of @xmath830 are the @xmath6orbits .
so let @xmath835 be two elements in @xmath836 such that @xmath837 .
then @xmath838 . by lemma [ stabilizer ]
@xmath839 and @xmath840 . the claim will be proved if we show that @xmath830 has local sections . for this
it suffices to use the remark following lemma [ stabilizer ] and a local section of the quotient map @xmath841 now we are going to verify the 4 properties of a good quotient for @xmath489 .
clearly ( i ) is satisfied by the definition of @xmath489 .
proof of ( ii ) .
it is clear that @xmath489 is surjective .
the morphism @xmath489 is affine because @xmath842 and @xmath843 and @xmath653 are affine .
proof of ( iii ) .
let @xmath844 be an open subset .
then @xmath845 since @xmath489 is @xmath6invariant .
conversely let @xmath846 .
the @xmath847 , and since @xmath830 is a geometric quotient , @xmath848 descends to an @xmath849 , which is @xmath43invariant .
now again @xmath850 descends because @xmath843 is a good quotient .
this proves equality @xmath851 .
proof of ( iv ) .
let @xmath817 be disjoint , closed , @xmath6invariant subvarieties of @xmath95 . then @xmath852 are disjoint , closed and @xmath853invariant subvarieties of @xmath832 . since @xmath830 is a good quotient , @xmath854 are disjoint , closed and @xmath43invariant in @xmath855 . finally , since @xmath843 is a good quotient , @xmath856 are disjoint and closed subvarieties of @xmath493 . but @xmath857 , which proves ( iv ) .
the normality of @xmath493 follows from the fact that @xmath855 is smooth and @xmath843 is a good quotient , @xcite , with respect to the reductive group @xmath43 . that @xmath489 becomes a geometric quotient on the open set @xmath97 of stable points follows from the fact that the @xmath43orbits in @xmath858 intersect @xmath97 in @xmath6orbits .
in particular the stabilizers of @xmath56 in @xmath6 and of @xmath603 in @xmath136 are isomorphic , such that all orbits have the same dimension .
the proof of ( 1 ) is a modification of the above . in any case @xmath843 induces the geometric quotient @xmath859 with @xmath860 open in @xmath493 .
now @xmath858 is a @xmath843-saturated open subset of @xmath861 , such that we obtain a geometric quotient @xmath862 with @xmath863 open . by the same arguments as
above applied to the diagram related to @xmath864 we conclude that @xmath865 is a geometric quotient .
_ remarks _ : 1 ) the idea of this proof comes from @xcite , and has already been used in @xcite and @xcite .
\2 ) if the second condition of ( 2 ) is not satisfied , we can not even prove that @xmath44 admits a good quasi projective quotient , because @xmath866 might not be saturated .
of course the projectivity of the quotient depends on this condition .
[ s - equi]s equivalence we suppose that the hypotheses of proposition [ goodqu ] are satisfied , with polarization @xmath38 for @xmath803 and associated polarization @xmath867 for @xmath804 .
it is easy to define the _ jordan - hlder filtration _ of @xmath43-semi - stable elements of @xmath621 with respect to @xmath867 ( cf.@xcite for a more general situation ) . using the preceding results we can also define a _
jordan - hlder filtration _ of a @xmath6-semi - stable element of @xmath83 with respect to @xmath38 .
let @xmath868 then there exist a positive integer @xmath499 , an element @xmath68 and filtrations @xmath869 with @xmath870 for each @xmath871 , such that @xmath872 satisfies @xmath873 and that if @xmath874 is the induced morphism , then @xmath875 is @xmath6-stable with respect to @xmath38 for any @xmath871 .
this filtration and @xmath773 need not be unique , but @xmath499 is unique and the @xmath876 , too , up to the order and isomorphisms .
conversely , an element of @xmath83 having such a filtration is @xmath6-semi - stable with respect to @xmath38 .
we say that two elements @xmath877 and @xmath878 of @xmath44 are _ s - equivalent _ if they have jordan - hlder decompositions @xmath879 respectively of the same length , and if there exists a permutation @xmath880 of @xmath881 such that @xmath882 is isomorphic to @xmath883 for any @xmath871 .
the following result is also easily deduced from [ goodqu ] .
[ equiv]*proposition * : let @xmath884 .
then @xmath885 if and only if @xmath56 and @xmath732 are s - equivalent .
it follows that the set of closed points of @xmath493 is exactly the set of s - equivalence classes of elements of @xmath95 .
we are going to investigate conditions for the weights of the polarizations under which a + ( semi)stable point @xmath53 is mapped to a ( semi)stable point @xmath886 . for the estimates we need the following constants which depend on the dimensions @xmath887 and the composition maps @xmath888 .
[ constants ] constants let @xmath102 be the family of proper linear subspaces @xmath889 such that @xmath105 is not contained in @xmath890 for any family @xmath891 of subspaces . for any @xmath312 we let the map @xmath892 be induced by the maps @xmath893 associated to the composition maps , which are supposed to be surjective , see [ abstr ] . we introduce the constant @xmath894
similarly we define the constants @xmath895 in the dual situation .
let @xmath896 be induced by the maps @xmath897 and let @xmath333 be the family of proper subspaces @xmath898 which are not contained in @xmath899 for any family @xmath900 of subspaces .
then we define @xmath901 [ monoton ] * lemma : * if @xmath902 for all @xmath903 , then @xmath904
. it will be sufficient to assume that @xmath905 for all @xmath292 except one , @xmath906 say .
then let @xmath907 be vector spaces of dimensions @xmath908 and suppose that @xmath909 for any @xmath910 we consider the subspace @xmath911 then @xmath912 and also @xmath913 because @xmath914 is a direct sum of the surjective operator @xmath915 such that @xmath916 equals @xmath917 and @xmath918 .
therefore @xmath919 .
once we have shown that also @xmath920 belongs to the analogous family @xmath921 , the lemma is proved . to see this
let @xmath922 and @xmath923 for @xmath924 be subspaces such that @xmath925 then in particular @xmath926 and thus @xmath927 .
but then @xmath928 with @xmath929 and it follows that @xmath930 since @xmath931 we obtain @xmath932 for all @xmath292 and then also @xmath933 .
[ conv1]study of the converse
i let @xmath934 be a polarization on @xmath83 and let @xmath935 be the associated polarization on @xmath621 ( the associated polarization has been defined in [ asspolar ] ) .
we had shown in [ ssrelat ] and [ srelat ] that if @xmath53 and @xmath603 is ( semi)stable in @xmath621 with respect to @xmath43 and @xmath759 , then so is @xmath56 with respect to @xmath6 and @xmath38 .
we are going to derive sufficient conditions for the converse , i.e. whether @xmath603 is ( semi)stable if @xmath56 is ( semi)stable . in the sequel we are going to use the following * notation * : given a family @xmath936 of subspaces @xmath751 we set @xmath937 and call a subspace @xmath795 saturated if there is such a family with @xmath938 .
note that in this case @xmath939 . similarly we introduce the spaces @xmath940 for a subfamily @xmath941 of @xmath942 and call them saturated .
let @xmath790 be given and assume that @xmath603 is not semi stable with respect to @xmath46 .
then there exist linear subspaces @xmath943 and @xmath796 such that @xmath944 and such that @xmath945 where as before the small characters denote the dimension of the spaces .
if there were subspaces @xmath751 and @xmath752 with @xmath938 and @xmath946 as in [ srelat ] , then @xmath947 would imply that @xmath948 and we would have @xmath949 and @xmath56 would not be semi stable . in the following we are going to construct families @xmath950 of subspaces @xmath951 and @xmath952 such that @xmath953 and @xmath954 are as close to @xmath955 as possible and such that there is a useful estimate for @xmath956 _ step 1 _ : we can assume that @xmath957 has a decomposition @xmath958 and such that @xmath959 . to derive this
, we remark that for a subspace @xmath960 of a direct sum @xmath961 of vector spaces there exists a linear map @xmath962 such that the isomorphism @xmath963 of @xmath961 transforms @xmath960 into @xmath964 , where @xmath965 is the projection of @xmath960 in @xmath966 and @xmath967 . using this and descending induction on @xmath292 we can find an element @xmath968 , see [ subgrp ] , such that the truncations @xmath969 , see [ bigg ] , map @xmath957 onto a direct sum @xmath970 for any @xmath292 . since @xmath971
we easily derive that @xmath972 for all possible @xmath292 .
we put @xmath973 note that @xmath974 .
_ step 2 _ : let @xmath975 be subspaces of @xmath976 respectively such that @xmath977 is minimal over @xmath957 for any @xmath292 . then @xmath978 since these spaces are the first components of @xmath979 respectively and we have @xmath980 .
we let @xmath981 _ step 3 _ : we are going to define the subspaces @xmath982 as images .
let @xmath983 be the map which is the sum of the composed maps @xmath984 then we define @xmath985 and @xmath986 it follows @xmath987 for any @xmath312 .
_ step 4 _ : if the weights @xmath988 are supposed to be positive , we may assume that @xmath989 for @xmath643 .
otherwise we could choose subspaces @xmath990 by descending induction as images .
then @xmath991 would improve the assumption on the choice of the spaces @xmath957 and @xmath992 .
now it follows that for any @xmath312 @xmath993 because @xmath994 is mapped to @xmath995 and the maps @xmath650 are the identity on the spaces @xmath996 .
note that we even have @xmath997 since @xmath998 factorises through @xmath999 as follows from the definition of @xmath1000 .
[ estimate ] * lemma * : suppose that all @xmath1001 , and let @xmath1002 then @xmath1003 let @xmath1004 since @xmath1005 is not contained in a direct sum with spaces smaller than @xmath1006 we get @xmath1007 by lemma [ monoton ] and above definitions we get @xmath1008 the map @xmath1009 sends @xmath1010 onto @xmath1011 by definition of @xmath1011 and also maps @xmath1012 onto @xmath1000 .
therefore , since @xmath980 , we have a surjection @xmath1013 and the dimension estimate @xmath1014 now we can derive the estimate of the lemma .
if there is no summation condition it is understood that the sum has to be taken over all indices of the given interval .
we have @xmath1015 substituting for @xmath1016 in the third sum and replacing the first by @xmath1017 and using the definition of @xmath1018 we get @xmath1019 now using the assumed estimate for the first sum and the derived estimate for @xmath1020 we get @xmath1021 [ sone]*corollary * : suppose that @xmath1022 , let @xmath1023 and let @xmath759 be the associated polarization @xmath1024 if all @xmath1025 and if @xmath1026 then @xmath1027 _ remarks _ : ( 1 ) note that by the normalization of the polarizations we must have @xmath1028 such that @xmath1029 is the only possible value for @xmath1030 .
\(2 ) if all @xmath1031 , then the necessary conditions for @xmath1032 and @xmath1033 are both satisfied , see [ asspolar ] .
the condition of the corollary is an extra condition .
let us first assume that @xmath603 is not semi stable and let the spaces @xmath957 and @xmath1034 be as at the beginning of [ conv1 ] .
the only @xmath1035 is positive .
let the other spaces be chosen as in [ conv1 ] .
the difference @xmath1036 reduces to @xmath1037 , and since @xmath1038 , this difference is zero .
therefore @xmath1039 since all the @xmath747 are positive we have @xmath1040 moreover , @xmath1041 induces a surjection @xmath1042 because @xmath980 .
therefore we obtain the dimensions estimate @xmath1043 .
it follows that @xmath1044 since @xmath1045 the last expression is non negative .
this proves the case of semi stability .
for the case of stability we assume that @xmath56 is stable and that @xmath603 is already semi stable .
if @xmath603 were not stable , we would find subspaces @xmath957 and @xmath1046 as in [ conv1 ] such that @xmath1047 and such that at least one @xmath957 is different from @xmath748 .
now let the spaces @xmath1006 and @xmath1048 be constructed as above .
then we have @xmath1049 where @xmath1050 , and where we use that @xmath1051 . if the family @xmath1052 is different from @xmath493 , then @xmath1053 , and if it is equal , then @xmath1054 . in order to obtain a contradiction we have to show that @xmath1052 is different from @xmath493 .
assume that it is not .
then @xmath1055 and we must have @xmath1056 for @xmath924 and @xmath1057 .
if also @xmath1058 , then by the above estimate also @xmath1059 , contradicting the choice of the @xmath957 . therefore @xmath1060 and @xmath1061
but then @xmath1062 and we have @xmath1063 . from this
it is easy to see that @xmath1064 where @xmath1065 for @xmath1066 and @xmath1067 .
then we have @xmath1068 which contradicts the stability of @xmath56 . [ conv2 ]
study of the converse ii we keep the notation of [ conv1 ] and compare the ( semi)stability of points in @xmath83 and @xmath137 in two steps , each reducing to the case @xmath1022 .
we consider the intermediate space @xmath1069 and the maps @xmath1070 here @xmath1071 is defined by @xmath1072 where @xmath1073 is the map defined by @xmath790 as in [ conv1 ] .
the map @xmath1074 is defined by @xmath1075 where now @xmath1076 is induced by the tuple @xmath1077 as the sum of the compositions @xmath1078 which are induced by the @xmath1079 and the pairings @xmath1080 .
it is obvious that @xmath1081 note that both @xmath1071 and @xmath1074 are injective by the same reason as for @xmath653 . on @xmath1082
the group @xmath1083 acts naturally and we have the embedding @xmath1084 see [ bigg ] .
it follows as in section 5 that @xmath1071 is compatible with the group actions and that @xmath1085 are on the same @xmath6orbit if and only if @xmath1086 are on the same @xmath1087 orbit .
similarly we have the group embedding @xmath1088 and @xmath1074 is equivariant and satisfies the analogous statements for the orbits . given the polarization @xmath1089 for @xmath1090 we consider the polarization @xmath1091 for @xmath1092 where the @xmath747 are defined as in [ asspolar ] .
as in [ ssrelat ] , [ srelat ] it is easy to show that @xmath1093 and similarly that @xmath1094 note that as for @xmath1095 , we have unipotent sub - orbits in @xmath1096 and @xmath1097 , see [ defstab ] .
we are going to show that in all 4 cases equality holds under suitable conditions on the weights of the polarizations .
then the same is true for @xmath653 .
[ estim1]estimate for @xmath1098 let @xmath790 in @xmath83 be given and assume that @xmath1099 is not semi stable .
then there are linear subspaces @xmath795 and @xmath752 and a unipotent element @xmath1100 such that for @xmath1101 we have @xmath1102 for all @xmath903 and all @xmath312 , and such that @xmath1103 we may assume that @xmath1104 because @xmath1105 acts on @xmath83 in the same way and we can replace @xmath56 by @xmath67 .
moreover , we may assume that all @xmath1000 are equal to @xmath1106 since all @xmath1107 .
now we proceed as in [ conv1 ] replacing the spaces @xmath750 by @xmath12 .
therefore we find subspaces @xmath1108 such that @xmath980 and such that @xmath1109 and the family @xmath1052 is minimal with this property .
we denote @xmath1110 and let @xmath1111 as in [ estimate ] we consider the surjection @xmath1112 where @xmath1113 are the same , and we get the estimate @xmath1114 for any @xmath312 .
the estimation of the discriminant @xmath1115 is now simpler than in [ conv1 ] .
[ discrim1 ] * lemma * : with the above notation @xmath1116 where @xmath1117 . by replacing dimensions and inserting the estimate for @xmath1020 as in [ conv1 ]
we get @xmath1118 [ equal1 ] * corollary * : let @xmath1119 be a polarization for @xmath83 and let @xmath1120 be the associated polarization for @xmath1082 as in [ conv2 ] .
if all @xmath1121 and @xmath1122 then @xmath1123 the proof is the same as for [ sone ] , because the spaces @xmath957 and @xmath1124 are defined in the same way and we thus get the estimate @xmath1125 . [ estim2 ] estimate for @xmath1126 the analogous estimate for @xmath1074 follows by duality while we can assume that @xmath1022 or @xmath701 .
the proof could be done by formally transform it into a dual situation which is similar to that of [ estim1 ] , but it is better to keep direct track of the weights .
let @xmath1127 be given in @xmath1128 and assume that its image under @xmath1074 is not semi stable
. then there are subspaces @xmath795 and @xmath796 such that @xmath1129 where @xmath654 is defined as in [ conv2 ] , and such that @xmath1130 we assume that all @xmath1131 , and then we may assume that @xmath957 is maximal , i.e. the inverse image of @xmath1132 under @xmath1133 for @xmath903 , and similarly @xmath1134 in @xmath1135 under @xmath1136 as in [ estim1 ] we can find subspaces @xmath752 such that @xmath1137 we choose subspaces @xmath1138 which are maximal such that @xmath1139 we have @xmath1140 . we let @xmath1141 be the inverse image of @xmath1142 under @xmath1143 . then @xmath1144 .
furthermore we let inductively @xmath1145 be the inverse images for @xmath903 .
then we have injections @xmath1146 and induced by factorization the images @xmath1147 the induced injections @xmath1148 imply the dimension estimates @xmath1149 for @xmath903 .
next we consider the homomorphism @xmath1150 we have @xmath1151 and consider the subspace @xmath1152 by the definition of the constant @xmath1153 we get @xmath1154 where @xmath1155 further we have a surjective map @xmath1156 which is induced by the map @xmath1157 and the induced surjection @xmath1158 , since @xmath1159 .
so we get @xmath1160 now we can estimate the discriminant in [ discrim2 ] * lemma * : let all the @xmath1161 be non - negative and let @xmath1162 then @xmath1163 since @xmath1164 we also have @xmath1165 with the same steps as in the previous proofs we get @xmath1166 inserting the assumption on the first difference and the estimate for @xmath1167 we get the result .
as in the previous cases we obtain the [ equal2 ] * corollary * : in the above notation let all @xmath1168 , and all @xmath1169 , and let @xmath1170 then @xmath1171 : in the notation of [ estim2 ] there is a surjection @xmath1172 because @xmath1159
. therefore @xmath1173 . if the condition of the corollary is satisfied , then @xmath1174 follows , where we use @xmath1175 combining the results of [ equal1 ] and [ equal2 ] we get the [ equality ] * proposition * : let @xmath1176 be a polarization for @xmath803 and let @xmath1177 be the associated polarization for @xmath804 .
suppose that all @xmath1168 , and all @xmath1169 and that @xmath1178 then @xmath1179
the projectivity of the quotient in [ goodqu ] depends on the second condition in ( 2 ) , i.e. whether the boundary @xmath1180 of the saturated set contains no semi
stable points of @xmath1181 again this condition depends on the chosen polarization and conditions for the weights . in order to derive these conditions in some cases we describe the boundary in terms independent of the group action .
[ satbd]saturated boundary . the elements of @xmath1182 are tuples @xmath1183 of linear maps @xmath1184 if @xmath1185 , there are an element @xmath53 and automorphisms @xmath1186 such that @xmath1187 here @xmath1188 stands for the different identities of the spaces @xmath1189 we let @xmath1190 respectively @xmath1191 be the mapping @xmath1192 induced by @xmath1193 respectively @xmath647 for @xmath924 . from the relations between the @xmath1193 and @xmath647
it follows easily that for each @xmath924 the composition @xmath1194 has a factorization @xmath1195 \ar[r ] & p_1\\ p_i\otimes a_{i1}\ar[ur]_{x_{1i } } } \ ] ] where the vertical map is the surjection induced by the pairings .
this follows from the commutative diagrams induced by the automorphism @xmath1196 and because @xmath1197 admits such a factorization for each @xmath924 .
we put @xmath1198 . by the dual description for the maps
@xmath1199 we are given factorizations @xmath1200\\
q_s \ar[ur]^{y_{ls } } \ar[r ] & b^\ast_{s , s-1}\otimes \ldots\otimes b^\ast_{sl}\otimes q_l } \ ] ] of the maps @xmath1201 for @xmath1202 . by similar arguments
there are also factorizations @xmath1203
\ar[r]^-{\scriptscriptstyle { x_{1i}\otimes i d } } & p_1\otimes h^\ast_{s1}\ar[r]^{\gamma } & q_s\\ p_i\otimes h^\ast_{si}\ar[urr]_{\gamma_{si } } } \hspace{4cm}\raisebox{-7mm}{$(l_i)$}\ ] ] for all @xmath903 and dually factorizations @xmath1204\\
p_1\ar[urr]^{\gamma_{l1}}\ar[r]_-{\gamma } & q_s\otimes h_{s1}\ar[r]_-{\scriptscriptstyle{y_{ls}\otimes i d } } & b^\ast_{sl}\otimes q_l\otimes h_{s1 } } \hspace{4cm}\raisebox{-7mm}{$(r_l)$}\ ] ] for all @xmath312 .
moreover , there are further factorizations of the induced composed maps @xmath1205 \ar[r]^-{\scriptstyle{\gamma_{si}\otimes i d } } & q_s\otimes b_{sl } \ar[r]^-{\widetilde{y}_{ls } } & q_l\\ p_i\otimes h^\ast_{li}\ar[urr]_{\phi_{li } } } \hspace{4cm}\raisebox{-7mm}{$(l_{li})$}\ ] ] and dually @xmath1206\\ p_i \ar[urr]^{\psi_{li } } \ar[r]_-{\widetilde{x}_{1i } } & p_1\otimes a^\ast_{i1 } \ar[r]_-{\scriptstyle{\gamma_{l1}\otimes i d } } & q_l\otimes h_{l1}\otimes a^\ast_{i1 } } \hspace{4cm}\raisebox{-7mm}{$(r_{li})$}\ ] ] all these factorizations are based on mappings induced by the pairings .
all factorization conditions are independent of the chosen automorphisms .
one can rediscover the original components @xmath660 of @xmath56 from @xmath1207 or @xmath1208 if @xmath1209 and @xmath1210 for all @xmath871 and all @xmath312 .
in fact we have [ descrz ] * lemma * : let @xmath1211 .
then @xmath1212 if and only if 1 .
rank @xmath1213 for @xmath903 2 .
rank @xmath1214for @xmath1215 3 .
@xmath1194 has a factorization @xmath1216 for @xmath924 4 .
@xmath1217 has a factorization @xmath1218 5 .
@xmath1219 has factorizations @xmath1220 and @xmath1221 6 .
@xmath1222 has factorizations @xmath1223 and @xmath1224 .
if @xmath1185 , the three conditions are satisfied by the above , where @xmath1225 is the dimension of the image of @xmath647 and @xmath1226 is the @xmath1227 of @xmath650 as the map @xmath1228 .
let conversely @xmath1229 satisfy these conditions .
we proceed by descending induction to find automorphisms @xmath1196 by which the @xmath1193 can be identified with the @xmath647 .
note that the factorization conditions are maintained under automorphisms .
since @xmath1230 has maximal rank it is an injection @xmath1231 .
hence we can find an automorphism @xmath1232 of @xmath1233 such that @xmath1234 becomes @xmath1235 .
let us assume now that modulo some automorphisms @xmath1236 we have @xmath1209 for @xmath1237 .
we are going to find an automorphism @xmath1238 such that @xmath1239 .
because of the rank condition we can assume that @xmath1240 is the image of @xmath1193 in @xmath1241 . now using all the @xmath1242 we find that @xmath1193 has a factorization through the standard map @xmath1243 induced by the pairings .
now the rank condition implies that @xmath1244 induces an automorphism on @xmath1240 .
this can be used to make @xmath1244 the identity via an automorphism @xmath1245 .
now @xmath1246 . by the analogous dual procedure we can also find automorphism @xmath1247 such that we can assume that @xmath1210 . finally the factorizations @xmath1221 or @xmath1224
resulting from ( 3 ) and ( 3@xmath1248 ) yield mappings @xmath1207 or @xmath1208 from which we get @xmath660 as composition @xmath1249 it follows from the special type of the @xmath647 and @xmath650 that these are original components of an element @xmath790 inducing @xmath1250 .
[ descrbdz]*corollary * : with the same notation as in [ descrz ] , if @xmath1251 , then \(1 ) @xmath1252 and @xmath1253 with strict inequality for at least one @xmath292 or @xmath312 , and ( 2 ) , ( 2@xmath1248 ) , ( 3 ) , ( 3@xmath1248 ) of [ descrz ] are satisfied . all conditions are closed and thus hold for points in @xmath1254 . if @xmath1251 then by [ descrz ] equality in ( 1 ) can not hold for all @xmath292 and @xmath312 .
we are going to derive effective sufficient conditions for the projectivity of the quotient in the cases @xmath1255 .
[ proj1 ] * proposition * : let the polarizations @xmath38 and @xmath46 be as in proposition [ equality ] and let @xmath1256 .
then @xmath1180 contains no semi
stable point in the following cases 1 .
@xmath1257 2 .
@xmath1258 @xmath1259 we present only the case ( ii ) , case(i ) is an easier version of ( ii ) .
let @xmath1260 and let us assume that @xmath1261 is not maximal .
let @xmath105 be the kernel of @xmath1262 and let @xmath1263 be the smallest subspace such that @xmath105 is contained in @xmath1264 .
we put @xmath1265 @xmath1266 and consider @xmath1267 by definition @xmath1268 .
diagram @xmath1269 reduces in our case , with @xmath1270 replaced by @xmath1271 , to @xmath1272_{\delta_2}\ar@{->>}[r]^-{\scriptstyle{x_2\otimes i d } } & p'_1\otimes h^\ast_{21}\ar@{->>}[r]^{\gamma } & q'_2\\ m'_2\otimes h^\ast_{22}\ar@{->>}[urr]_{\gamma_{22 } } } \ ] ] and @xmath1273 vanishes on @xmath1274 because @xmath105 is the kernel of @xmath1275 .
therefore @xmath1276 in order to estimate @xmath1277 we consider diagram @xmath1278 enlarged by the commutative square of induced pairings @xmath1279\ar@{->>}[r ] & m'_2\otimes h^\ast_{22}\otimes b_{21 } \ar@{->>}[d]\ar@{->>}[r]^-{\scriptstyle{\gamma_{22}\otimes i d } } & q'_2\otimes b_{21}\ar@{->>}[r]^-{y_1 } & q'_1\\ m'_2\otimes a_{21}\otimes h^\ast_{11 } \ar@{->>}[r]^-{\delta_1 } & m'_2\otimes h^\ast_{12 } \ar@{->>}[urr]_{\phi_{12 } } } .\ ] ] again the map @xmath1280 vanishes on @xmath1281 and we get @xmath1282 now we have the estimate @xmath1283 therefore the condition @xmath1284 would be sufficient , because @xmath1285 .
we modify the last estimate as follows .
since the weights in case @xmath763 are related by @xmath1286 and since we have @xmath1287 we get the estimate @xmath1288 this shows that @xmath1174 if @xmath1275 is degenerate and the first condition of ( ii ) is satisfied . in case @xmath1289
is not maximal the second condition follows by the dual procedure .
[ proj2 ] the case ( 3,1 ) in order to derive a similar result in case @xmath761 we introduce the additional constant @xmath1290 analogous to @xmath1291 in [ constants ] .
let @xmath1292 be the linear map induced by the pairing and let @xmath102 be the family of all proper subspaces @xmath1293 which are not contained in @xmath1294 for any subspace @xmath1295 different from @xmath1296 .
we put @xmath1297 for brevity we write @xmath1298 and @xmath1299 [ projestim]*proposition * : let @xmath1300 , let @xmath1301 be a polarization for @xmath803 and @xmath1302 be the associated polarization for @xmath1303 , and assume that all @xmath1304 ( in this case @xmath1305 . )
if 1 .
@xmath1306 2 .
@xmath1307 3 .
@xmath1308 then @xmath1180 contains no semi stable point
. moreover , condition ( 1 ) may be replaced by any of the conditions 1 .
@xmath1309 2 .
@xmath1310 3 .
@xmath1311 * remark * : @xmath1180 contains no semi stable point also in each of the following cases 1 .
@xmath1312 2 .
@xmath1313 3 .
@xmath1314 .
this can be seen by a direct estimate of the discriminant @xmath1115 after substituting for @xmath1277 in the following proof .
let @xmath1315 .
we distinguish the following cases of degeneracy of @xmath1275 and @xmath1316 .
_ case 1 : @xmath1316 is injective _ : then by the proof of [ descrz ] we can assume that @xmath1317 is the canonical embedding and that @xmath1318 and @xmath1275 have a factorization @xmath1319 in the following diagram @xmath1320\ar@{>->}[r]^-{\widetilde{\xi}_3 } & ( m_2\oplus m_3\otimes a_{32})\otimes a_{21 } \ar[d]\ar[r]^-{x_2 } & p_1\\ m_3\otimes a_{31}\ar[urr]^{x_{13}}\ar@{>->}[r]_-{\xi'_3 } & m_2\otimes a_{21}\oplus m_3\otimes a_{31}\ar[ur]_{\bar{x}_2 } } .\ ] ] here also @xmath1321 is the canonical embedding .
moreover it is easy to verify that in this case also the composed map @xmath1322 admits a decomposition @xmath1323 here @xmath1324 since @xmath1319 can not be injective by the assumption on its rank .
we choose subspaces @xmath1325 such that @xmath1326 and such that these subspaces are minimal with this property .
now we consider the spaces @xmath1327 and their discriminant @xmath1328 by the definition of the constant @xmath1329 and the diagram @xmath1330\ar@{>>}[r ] & p'_1\otimes h^\ast_{11}\ar@{>>}[r ] & q'_1\\ m'_2\otimes h^\ast_{12 } \oplus m'_3\otimes h^\ast_{13}\ar@{>>}[urr ] } \ ] ] we obtain the estimate @xmath1331 where by the definition of @xmath1134 we have @xmath1332 inserting this we obtain @xmath1333 if @xmath1334 , conditions ( 2 ) and ( 3 ) imply that @xmath1174 .
if , however , @xmath1335 we have the direct estimate @xmath1336 this proves the proposition in the first case .
_ case 2 : @xmath1316 is not injective _
here we let @xmath105 denote the kernel of @xmath1316 and we choose a subspace @xmath1337 such that @xmath1338 and @xmath1339 is minimal with this property .
then we consider the subspaces @xmath1340 we have the exact sequences @xmath1341 where @xmath339 denotes the kernel of @xmath1318 . from the factorization properties restricted to the spaces @xmath957 and
@xmath1034 we extract the following commutative diagram of surjections @xmath1342^{\delta_1}\ar[r ] & p'_1\otimes h^\ast_{11}\ar[d]^{\gamma}\\ m'_3\otimes a_{32}\otimes a_{21}\otimes h^\ast_{11}\ar[ur ] \ar[dr ] & m'_3\otimes h^\ast_{13}\ar[r]^{\gamma_{13 } } & q'_1\\ & m'_3 \otimes a_{32}\otimes h^\ast_{12}\ar[r ] \ar[u]_{\tau } & p'_2\otimes h^\ast_{12}\ar[u]_{\gamma_{12 } } } \raisebox{-3cm}{.}\ ] ] from this we get again the estimates @xmath1343 where @xmath1344 and @xmath1345 .
let @xmath386 be a real number .
then we have @xmath1346 substituting this into the discriminant we get @xmath1347 now condition ( 1 ) enables us to find @xmath338 with @xmath1348 such that the first two terms of the estimate are non negative .
therefore @xmath1174 , and again @xmath1349 is not semi stable . in order to show that ( 1 ) can be replaced by one of ( i ) ,
( ii ) or ( iii ) we substitute @xmath747 and @xmath1350 and get after cancelation @xmath1351 if @xmath1352 , then by a direct estimate we get @xmath1174
. therefore we may assume that @xmath1353 .
since in addition @xmath1354 , we get @xmath1355 this shows that ( 1 ) can be replaced by ( i ) . in the same way one shows that ( 1 ) can be replaced by ( ii ) , using the other estimate of @xmath1277
that finally ( 1 ) can be replaced by ( iii ) can be shown by substituting first @xmath1356 and canceling @xmath1357 and then substituting @xmath1358 to get @xmath1359 proof of theorems [ theo_main ] and [ theo_main2 ] theorem [ theo_main ] is an immediate consequence of proposition [ goodqu ] , corollary [ sone ] and proposition [ proj1 ] .
theorem [ theo_main2 ] follows immediately from theorem [ theo_main ] and [ constant_calc ] .
[ constant_calc ] constants we give here some constants ( cf . [ constants ] ) used in the examples .
the following result is proved in @xcite , prop .
[ formula1]*proposition * : for homomorphisms of type @xmath1360 on a projective space of dimension @xmath30 we have @xmath1361 [ symbd2 ] * lemma * : for homomorphisms of type @xmath1362 on the projective space @xmath458 the constant @xmath1363 is @xmath1364 .
we put @xmath1365 .
the homomorphisms @xmath1366 of [ constants ] reduces here to the canonical map @xmath1367 if @xmath105 is a proper subspace of @xmath1368 which is not contained in one of the summands , it contains elements @xmath1369 with @xmath1370 or elements @xmath1369 with @xmath1371 .
but since @xmath1372 is surjective , the map @xmath1373 is surjective .
hence @xmath1374 .
if @xmath105 contains an element @xmath1375 with @xmath1371 , then @xmath1376 . for then @xmath1377 contains @xmath416 , and since @xmath1373 is surjective , if follows that @xmath1378 .
therefore , if @xmath1379 , there is a basis @xmath1380 of @xmath105 with @xmath1381 linearly independent , i.e. @xmath1382 or @xmath1383 .
therefore @xmath1384 .
but now we can find subspaces which realize this bound . for any @xmath1385 we let @xmath105 be the space of all @xmath1386 .
then @xmath1387 and it follows also that in this case @xmath1388
. then @xmath1389 .
[ ex1 ] first example of type @xmath31 we use the abbreviation @xmath1390 for @xmath1391 for a sheaf and a positive integer and consider here homomorphisms @xmath1392 over @xmath32 of type @xmath31 .
the polarization @xmath1393 is supposed to be proper for @xmath83 and @xmath137 , i.e. @xmath1394 and @xmath1031 for all @xmath292 .
the only constant involved here is @xmath1395 .
therefore the conditions of [ sone ] and [ proj1 ] are automatically satisfied by @xmath1396 .
hence all the quotients of @xmath44 will be good and projective under this condition .
since @xmath1397 and @xmath1398 , we can replace the polarization by the rational number @xmath1399 ( cf .
[ ex1gen ] ) . the numerical condition for ( semi)stability then becomes @xmath1400 where @xmath1401 is the dimension vector of a @xmath1402invariant sub - family of vector spaces , such that @xmath1403 .
one can easily check that @xmath1404 is the only value for which @xmath1115 might be zero , and this is the case for the values @xmath1405 and @xmath1406 . and indeed , the homomorphisms @xmath573 given by matrices @xmath1407 with generically chosen entries and linear forms @xmath1408 are semi stable and not stable for @xmath1404 .
[ ex1plus ] the case @xmath1409 it is easy to show that in this case @xmath1410 is @xmath338stable if and only if * @xmath1411 is nowhere zero * for any @xmath1412 with @xmath68 and any @xmath1413dimensional subspace @xmath1414 .
the first condition says that @xmath1415 is isomorphic to the universal quotient bundle @xmath215 on @xmath32 , and the second that @xmath1416 induces a @xmath1417dimensional subspace of @xmath1418 .
it follows that the sets @xmath96 of stable points are the same for @xmath1409 , which we denote by @xmath1419 .
moreover , from the above characterization of stable homomorphism we deduce that the geometric quotient @xmath1420 is isomorphic to the grassmannian @xmath1421 which is smooth of dimension @xmath1422 .
there is an interesting subvariety @xmath1423 which consists of the images of the homomorphisms @xmath1424 which belong to @xmath1419 .
these are those @xmath1410 for which the induced homomorphism @xmath1425 is not injective .
we will see next that @xmath490 is isomorphic to the non stable locus of @xmath860 below and is smooth of dimension @xmath1426 .
[ ex1zero ] the case @xmath1404 we write @xmath1427 for @xmath1428 . when considering the matrix representations we find that @xmath1429 and that the remaining part @xmath1430 consists of those homomorphisms for which @xmath1411 is zero in exactly one point .
such homomorphisms are equivalent to matrices @xmath1431 where @xmath1432 are independent linear and @xmath1433 are independent quadratic forms .
note , however , that @xmath1419 intersects the non stable locus of @xmath1427 in matrices equivalent to those of type ( 1 ) .
but the orbit closures in @xmath1427 of both types ( 1 ) and ( 2 ) of matrices contain the direct sums @xmath1434 of independent linear and quadratic forms . from that it
follows that the induced morphism @xmath1435 of the quotients is bijective and moreover an isomorphism by zariski s main theorem , because both spaces are normal .
the points of the non stable locus @xmath1436 are represented by matrices of type ( 3 ) .
it is again routine to deduce from this observation that @xmath1437 the subvariety @xmath1423 corresponds to this set under the isomorphism .
we can also identify the set @xmath1438 of stable points with @xmath1439 .
[ ex1minus ] the case @xmath1440 similarly to the case @xmath1419 we find that here @xmath1441 is independent of @xmath338 and that @xmath1442 . the remaining part consists now of all homomorphisms which are equivalent to a matrix of type ( 1 ) . note that now homomorphisms of type ( 2 ) are contained in @xmath1443 .
the induced morphism @xmath1444 is again surjective but not injective over @xmath1436 .
let @xmath574 be the inverse image of @xmath1436 .
then @xmath574 consists of the points which are represented by matrices of type ( 2 ) which are not equivalent to matrices of type ( 3 ) .
it is easy to check that the restricted morphism @xmath1445 is bijective and therefore also an isomorphism by zariski s main theorem .
we are going to verify that @xmath574 is a divisor in @xmath1446 .
there is a morphism @xmath1447 which assigns to the class of @xmath1410 the point @xmath249 at which @xmath1411 is degenerate . in this case @xmath1448 where @xmath1449 is the ideal sheaf of @xmath249 . for such @xmath1410
we are given an exact diagram @xmath1450 & & & { { \mathcal o}}\\ 0\ar[r ] & 2{{\mathcal o}}(-2)\oplus{{\mathcal o}}(-1)\ar[u]\ar[rr]^-{(\phi_1,\phi_2 ) } & & 3{{\mathcal o}}\ar[r ] & { { \mathcal f}}\ar[r]\ar[u ] & 0 \\ 0\ar[r ] & { { \mathcal o}}(-1)\ar[u]\ar[rr]^-{\phi_2 } & & 3{{\mathcal o}}\fleq[u]\ar[r ] & { { \mathcal o}}\oplus{{\mathcal i}}_x(1)\ar[u]\ar[r ] & 0\\ & 0\ar[u ] & & & 2{{\mathcal o}}(-2)\ar[u]\\ & & & & 0\ar[u ] } \ ] ] such that @xmath1410 corresponds to a @xmath1417dimensional subspace @xmath1451 .
the condition of defining a element of @xmath574 is that @xmath217 is neither contained in @xmath1452 nor in @xmath1453 for any section @xmath409 of @xmath1454 .
we let @xmath1455 denote the open subvariety of such @xmath217 . by assigning to @xmath217 the class of @xmath1456 where @xmath1416 is defined by a lifting in the above diagram
, we get a morphism @xmath1457 whose image is the fibre @xmath1458 .
the morphism @xmath1459 is nothing but the quotient of @xmath1460 by the algebraic group @xmath1461 .
it follows that @xmath1462 is a variety of dimension 23 . using the techniques of this paper for this quotient
, we can even prove that @xmath574 is smooth .
finally @xmath574 has dimension 25 and thus is a divisor in the irreducible and normal variety @xmath1446 .
_ remarks _ : ( 1 ) one would like to interpret the matrices of type ( 2 ) as representing extensions of the sheaves @xmath1463 and @xmath1464 .
indeed a matrix of type ( 2 ) defines such an extension , but this extension is isomorphic to the direct sum .
\(2 ) the above correspondence between @xmath1410 and @xmath217 indicates that the quotient spaces considered here are spaces of coherent systems as in @xcite .
[ flip1 ] the flip the diagram @xmath1465 can be interpreted as a flip .
it is induced by the inclusions @xmath1466 .
the orbits of stable points of type ( 2 ) in @xmath1443 and of type ( 1 ) in @xmath1419 do nt intersect in @xmath1427 but so do their closures in @xmath1427 .
thus the fibres of @xmath1467 and @xmath1468 correspond to the two different types of semi
stable orbits in @xmath1427 defining the same points in @xmath1436 .
[ ex1gen ] general homomorphisms of type @xmath31 in a more general situation of type @xmath31 we consider homomorphisms @xmath1469 over @xmath114 .
a polarization in this case is determined by the rational number @xmath92 with @xmath386 and @xmath1470 .
a @xmath38(semi)stable homomorphism is then called @xmath338(semi)stable .
we write @xmath94 and @xmath96 for @xmath44 and @xmath71 . in terms of @xmath338
the conditions are @xmath1471 the constant @xmath1472 is given in proposition [ formula1 ] .
such polarizations exist if and only if @xmath1473 in order to measure @xmath338stability we introduce the numbers @xmath1474 and call @xmath1475 @xmath573admissible if there are subspaces @xmath1476 of dimensions @xmath1477 such that @xmath573 maps @xmath1478 into @xmath1479
. then @xmath573 is @xmath338(semi)stable if and only if for any @xmath573admissible proper triple @xmath1475 , i.e. a triple which is neither @xmath1480 or @xmath1481 , we have @xmath1482 a polarization @xmath338 is called critical if there are proper triples with @xmath1483 .
thus the critical values of @xmath338 are the rational numbers @xmath1484 where we may assume @xmath1485 and thus @xmath1486 .
we let @xmath1487 be the maximal critical value if there are such with @xmath386 and put @xmath1488 otherwise .
if @xmath338 is not critical we have @xmath1489 . [ ex1stab ] * lemma * : suppose that @xmath405 and @xmath406 are relatively prime and that @xmath1490
. then @xmath1491 is @xmath338stable if and only if 1 .
@xmath1411 is stable with respect to the group @xmath1492 .
2 . for any @xmath1413dimensional subspace @xmath1493 , and any @xmath1494
the map @xmath1495 is not zero . by the characterization of stability in section 3 the homomorphism @xmath1411 is stable if and only if for any proper pair @xmath1496 of @xmath1411admissible subspaces @xmath1497 .
now let @xmath1410 be stable . if @xmath1411 were not stable there would be a proper @xmath1411admissible pair @xmath1498 with @xmath1499 .
but then @xmath1500 because @xmath1501 are supposed to be relatively prime .
then @xmath1502 because @xmath1503 is a critical value and thus @xmath1504 , contradicting the stability of @xmath1505 .
the condition ( 2 ) is trivially satisfied if @xmath1410 is @xmath338stable , because otherwise @xmath1506 would be admissible with @xmath1507 .
we have to show now that conversely ( 1 ) , ( 2 ) imply that @xmath1410 is @xmath338stable . for
this let @xmath1475 be a proper @xmath1410admissible triple .
if @xmath1508 and @xmath1509 , there is nothing to prove . if @xmath1510 then @xmath1497 by ( 1 ) and we have @xmath1511 and hence @xmath1512
if however @xmath1513 we have @xmath1512 in case @xmath1514 . since the case @xmath1515 is only possible
if @xmath1516 and then @xmath1517 by ( 2 ) , we can assume that @xmath1518 .
but then @xmath1519 because the fraction is a critical value , and last inequality is the inequality @xmath1512 .
now we are able to describe the space @xmath1520 for @xmath1521 which is independent of @xmath338 . according to the lemma @xmath96
can only be non empty if there are stable morphisms @xmath1411 .
this is the case if and only if @xmath1522 where @xmath1523 , see @xcite .
we restrict ourselves now to the case where in addition to the previous conditions on @xmath1524 we have @xmath1525 and @xmath1526
. then a stable @xmath1411 is injective and a subbundle ( except at finite number of points in case @xmath1527 , see @xcite , @xcite ) .
the quotient space of this space of stable homomorphisms by @xmath1528 is denoted by @xmath1529 .
it is a smooth projective variety and there is a universal sheaf @xmath8 on @xmath1530 .
for @xmath1531 let @xmath1532 denote the fibre sheaf representing @xmath249 . since it is the cokernel of the representing homomorphism @xmath1411 , we get @xmath1533 therefore @xmath1534 is locally free on @xmath1535 where @xmath499 denotes the first projection of @xmath1530 . now
@xmath1536 can be non empty only if @xmath1537 if conversely this is the case for any stable @xmath1411 and any subspace @xmath1538 where @xmath1539 $ ] , there is a lifting @xmath1540 of @xmath1541 , and @xmath1410 satisfies ( 1 ) , ( 2 ) of the lemma .
it follows now easily by considering corresponding families that @xmath1542 where @xmath1543 denotes the relative grassmannian .
it is more difficult to characterize the other moduli spaces @xmath1544 for the intervals between the critical values or for the critical values and to interpret the flips between them .
[ ex2 ] example of type @xmath763 we consider now a simple example of type @xmath763 on @xmath1545 of homomorphisms @xmath1546 again the polarizations @xmath1547 are supposed to be proper for @xmath83 and @xmath137 such that we have @xmath1548 and @xmath1549 all constants @xmath1550 and @xmath1551 are again zero , because @xmath1552 . then by the above conditions also the conditions for proposition [ equality ] and proposition [ proj1 ] are satisfied , such that there exists a good and projective quotient @xmath40 for any polarization satisfying the conditions .
since we have @xmath597 and @xmath1553 , the polarization @xmath38 is determined already by @xmath1554 and @xmath420 , for which the above conditions become @xmath1555 next we derive the conditions for the occurrence of true semi stable points . if @xmath1556 is the dimension vector of a @xmath573invariant sub - family we have to consider the equation @xmath1557 by inserting all possible dimension vectors we get the 6 conditions @xmath1558 for @xmath1559 .
if one of these is satisfied , there might be non stable points in @xmath44 . in the following figure 1
the lines with the equations ( 2 ) are shown together with the rectangle ( 1 ) ( lower right ) , for the points of which we get good and projective quotients .
( 960,500)(80,220 ) ( 310,560 ) ( 1 , 0)490 ( 800,560 ) ( 0,-1)490 ( 320,80)(1 , 0)600 ( 300,60)@xmath1560 ( 860 , 100)@xmath1554 ( 330,80)(-1 , 0 ) 10 ( 320,80 ) ( 0 , 1)600 ( 290,560)@xmath1413 ( 800,40)@xmath1413 ( 340,650)@xmath1561 ( 320,560 ) ( 1,-1)480 ( 800,80 ) ( 0 , 1 ) 5 ( 800,80)(-2 , 3)320 ( 800,80)(-1 , 3)160 ( 640,560)(-2,-3)320 ( 800,560)(-1,-1)480 ( 320,85 ) ( 0,-1 ) 5 ( 320,80 ) ( 1 ,
3)160 ( 800,285)(-10,0)49(-1,0)5 ( 705,560)(0,-10)49(0,-1)5 ( 270,285)@xmath1562 ( 705,40)@xmath1563 figure 1 the homomorphism @xmath573 defined by the matrix @xmath1564 where the @xmath1408 are homogeneous coordinates of @xmath1545 , is easily verified to be @xmath6stable for each polarization @xmath38 in the rectangle ( 1 ) .
therefore the moduli spaces are not empty .
on each of the 3 lines in the rectangle ( 1 ) each point defines one and the same open set @xmath44 and hence one and the same moduli space with semi stable and non stable points .
similarly , on each of the 4 open triangles we have one and the same moduli space , which is a smooth projective geometric quotient .
each of the 7 spaces has dimension @xmath1565 .
the reader may also verify that the moduli space for an open triangle admits a morphism to the moduli space of each of its edges , thereby defining a chain of flips .
[ ex2gen ] more general homomorphisms of type @xmath763 more general homomorphisms
for which we know the constants explicitly are homomorphisms of type @xmath1566 over @xmath1545 , say . by proposition [ constant_calc ] the constants are here @xmath1567 let @xmath83 be the space of those homomorphisms . a proper polarization + @xmath1568 for @xmath83 satisfies @xmath1569 with @xmath1570 positive .
we will also assume that @xmath1571 , i.e. @xmath1572 and @xmath1573 .
these four conditions can be replaced by @xmath1574 [ exclaim ] * claim * : there are polarizations @xmath38 such that @xmath44 admits a good and projective quotient in the following cases 1 .
@xmath1575 and @xmath1576 2 .
@xmath1577 and @xmath1578 3 .
@xmath1579 and @xmath1580 the conditions of [ sone ] for the equivalence of ( semi)stability become @xmath1581 and the conditions of [ proj1 ] for the projectivity of the quotient become @xmath1582 the first condition of ( 3 ) follows already from the first of ( 2 ) . after replacing @xmath1583 and @xmath1584 conditions ( 2 ) and ( 3 ) are equivalent to @xmath1585 \frac{7}{16}m_1\mu_1 & \geq & ( 4m_1 + 30)\lambda_2 - 15\\[1.0ex ] \mu_1 & \geq & \frac{4}{7}\lambda_2 \end{array } \eqno(4)\ ] ] using ( 1 ) for @xmath1554
, we find that ( 4 ) has a solution @xmath1586 if the system @xmath1587 \mu_1 & > & \frac{16}{7(8+m_1 ) } \end{array } \eqno(5)\ ] ] has a solution @xmath420 . for this
we distinguish the cases @xmath1588 . if @xmath1576 the first inequality of ( 5 ) has a solution @xmath1589 if @xmath1575 .
if @xmath1578 , then @xmath1577 , which is case ( i ) .
if @xmath1590 , the first inequality of ( 5 ) reduces to @xmath1591 then ( 5 ) has a solution @xmath420 if and only if @xmath1592 these inequalities reduce to @xmath1593 they are all satisfied if we suppose ( ii ) of the claim . in _ figure 2 _ the lines of the critical values of the polarizations i.e. of the pairs @xmath1594 are shown together with the small region of those pairs which satisfy the sufficient conditions ( 4 ) for the existence of a good and projective quotient , based on the values @xmath1595 and @xmath1596 . [ ex3 ] example of type @xmath761 as an example of type @xmath761 we consider only the space of homomorphisms @xmath1597 over @xmath1545
we assume again that all @xmath286 and all @xmath747 are positive
. then the conditions of [ sone ] together with the normalization of the polarization are @xmath1598 \lambda_2\geq \frac{4}{5 } c_1(1,1)\\ \phantom{xx } \end{array}\ ] ] as additional condition for the projectivity of the quotient we use condition ( a ) of the remark following proposition [ projestim ] . since in this case both the constants @xmath1599 and @xmath1600 are zero , this condition is just @xmath1601 and is already satisfied by our assumption . for homomorphisms of the above type
the condition @xmath1602 is necessary if @xmath1603 for if @xmath1604 is an element of @xmath83 then @xmath1605 has degree @xmath1413 and thus contains at most @xmath1606 independent components .
then @xmath1607 and @xmath1608 is a choice of dimensions of @xmath573invariant subspaces and the discriminant becomes @xmath1609 . by [ symbd2 ]
the value of @xmath1363 is @xmath1610 .
now it is easy to see that there exist polarizations @xmath38 which satisfy the above inequalities . that @xmath71 is then indeed non empty follows from the existence of generic matrices as in [ ex2 ] .
moreover there are again regions of polarizations for which the sets @xmath44 are the same and which are responsible for flips .
figure 2 + here the horizontal axis represents @xmath1611 and the vertical axis represents @xmath1612 + for @xmath1595 and @xmath1596 .
let @xmath30,@xmath408 be integers such that @xmath112 and @xmath1613 let @xmath416 be a vector space of dimension @xmath1614 , @xmath1615 .
we will study in this chapter morphisms of sheaves on @xmath114 of type @xmath1616 let @xmath1617 the linear map induced by @xmath1618 . for semistable morphisms
( with respect to a given polarization ) @xmath1619 is non zero .
so it is of rank 1 or 2 .
morphisms @xmath1620 such that @xmath1619 is of rank 2 are called _ generic _ , and those such that @xmath1619 is of rank 1 are called _ special_. suppose that @xmath1621 is a generic morphism .
let @xmath1622 and @xmath1623 be the linear subspace of zeroes of linear forms in @xmath317 .
then @xmath1618 is isomorphic to the canonical morphism @xmath1624 hence we have @xmath1625 , and @xmath1626 ( the ideal sheaf of @xmath1627 twisted by @xmath1628 ) .
let @xmath1629 be the restriction of @xmath1630 to @xmath1631 .
it vanishes on @xmath1627 and induces a linear map @xmath1632 let @xmath1634 . then .
the morphism @xmath1635 induced by @xmath1620 vanishes on @xmath1636 .
hence it induces a morphism @xmath1637 where @xmath1638 is the inclusion . since
, we can ( by replacing @xmath1620 by an element of its @xmath1639-orbit ) suppose that @xmath1640 .
it follows that , and since @xmath1620 is semi - stable , we have , i.e. @xmath1633 is injective .
suppose that @xmath1633 is injective .
then @xmath1642 is isomorphic to the canonical morphism @xmath1643 it is easy to see that @xmath317 and @xmath105 depend only on the @xmath6-orbit of @xmath1620 .
conversely , suppose @xmath317 and @xmath105 are given .
we can define an element @xmath1644 of @xmath83 associated to @xmath317 and @xmath105 as follows : let @xmath1645 be a basis of @xmath317 .
let be a basis of @xmath105 , with . using this basis we can identify @xmath105 and @xmath1646 with @xmath1647 .
we define @xmath1648 by @xmath1649 & v^*\\ ( \lambda,\mu)\ar@{|-_{>}}[r ] & \lambda z_1-\mu z_2 } \ ] ] and @xmath1650 over @xmath1651 by @xmath1652 let @xmath1651 . then @xmath1654 is non injective at @xmath249 if and only if all the elements of @xmath105 ( which are homogeneous polynomials of degree 3 ) vanish at @xmath249 .
suppose that @xmath1654 is non injective on an irreducible hypersurface @xmath960 .
then all the polynomials in @xmath105 vanish on @xmath960 .
let @xmath245 be an irreducible equation of @xmath960 .
then all the elements of @xmath105 are multiple of @xmath245 .
it follows that @xmath245 is of positive degree @xmath1655 , and @xmath1656 . but
this in impossible since @xmath1657 [ fine_lem3 ] * lemma * : let @xmath1658 be defined by @xmath1659 and .
suppose that there exists a polarization such that @xmath1620 is semi - stable .
then @xmath1620 is generically injective and @xmath1660 has no torsion .
moreover , if @xmath105 is generic , @xmath1620 is injective .
lemma [ fine_lem2 ] implies that @xmath1620 is injective outside a closed subvariety of codimension @xmath118 .
it follows that @xmath1620 is generically injective and that @xmath1660 has no torsion . to prove that @xmath1620 is injective for a generic @xmath105 , it suffices to find a @xmath105 such that @xmath1620 is injective
let be a basis of @xmath317 .
let , ( resp . )
be linearly independant elements of @xmath1661 that have no common zeroes in @xmath114 , with ( this is possible since ) .
let @xmath1662 it is easy to see that for such a @xmath105 , @xmath1620 is injective .
let @xmath1659 a plane , @xmath1623 the subspace defined by @xmath317 and a linear subspace of dimension @xmath408 .
let , where @xmath1620 is a morphism associated to @xmath317 and @xmath105 .
since the @xmath6-orbit of @xmath1620 is determined by @xmath317 and @xmath105 , @xmath1663 is well defined .
we will give another construction of @xmath1663 .
the exact sequence @xmath1667 implies @xmath1668 , @xmath1669 . using the exact sequence @xmath1670 we obtain the exact sequence @xmath1671 from @xmath126 we get , , where @xmath1672 is the inclusion . from the exact sequence @xmath1673 we deduce
the exect sequence @xmath1674 & \hom({{\mathcal i}}_{{{\mathbb p}}_{n-2}}(-1),{{\mathcal f}}_k)\ar[r ] & s^2v^*\ot p^{\bot}\ot k^ * \ar[rr]^-\theta & & s^3v^*\ot k^*/{{\mathbb c}}i_k } \ ] ] where @xmath1675 comes from the multiplication @xmath1676 the kernel of @xmath830 is canonically isomorphic to @xmath1677 and it is easy to see that @xmath1672 is contained in the image of @xmath1678 .
it follows that we have an exact sequence @xmath1679 and that @xmath1664 .
the last assertion follows from the commutative diagram with exact rows and columns : @xmath1680 & 0\ar[d ] & 0\ar[d ] \\ 0\ar[r ] & { { \mathcal o}}(-3)\ar[r]\ar[d ] & { { \mathcal o}}\otimes k^*\ar[r]\ar[d ] & { { \mathcal f}}_k\ar[r]\ar[d ] & 0 \\ 0\ar[r ] & { { \mathcal o}}(-2)\otimes{{\mathbb c}}^2\ar[r]\ar[d ] & { { \mathcal o}}(-1)\oplus({{\mathcal o}}\ot k^*)\ar[r]\ar[d ] & { { \mathcal e}}(p , k)\ar[r]\ar[d ] & 0 \\ 0\ar[r ] & { { \mathcal i}}_{{{\mathbb p}}_{n-2}}(-1)\ar[r]\ar[d ] & { { \mathcal o}}(-1)\ar[r]\ar[d ] & { { \mathcal o}}_{{{\mathbb p}}_{n-2}}(-1)\ar[r]\ar[d ] & 0 \\ & 0 & 0 & 0 } \ ] ] let @xmath1681 be the projective variety of pairs @xmath1682 , where @xmath317 is a plane of @xmath513 and @xmath1653 is a vector subspace of dimension @xmath408 ( @xmath1627 beeing the codimension 2 linear subspace of @xmath114 defined by @xmath317 )
. we can view @xmath1681 as a moduli space for generic morphisms .
we will give a construction of a _
universal sheaf _
@xmath1683 on @xmath1684 , i.e @xmath1683 is flat on @xmath1681 and for every @xmath1685 , @xmath1686 is isomorphic to the cokernel of a generic morphism associated to @xmath1682 .
it is also possible to define a _
universal morphism _ whose cokernel is isomorphic to @xmath1683 , but we will see this more generally in [ f_mod ] .
let @xmath1687 be the grassmannian of planes in @xmath513 and be the obvious projection .
let @xmath134 be the universal subsheaf of @xmath1688 on @xmath1687 .
let @xmath1689 be the projections .
then we have a canonical obvious morphism of vector bundles on @xmath1684 @xmath1690 let @xmath1691 be its cokernel .
it is a flat family of sheaves on @xmath114 . for every @xmath1685 we have .
let @xmath102 be the universal sheaf on @xmath1684 , such that . then we have a canonical obvious morphism of vector bundles on @xmath1684 @xmath1692 let @xmath9 be its cokernel .
then for every @xmath1693 , @xmath1694 is the sheaf that was noted @xmath1665 before . by lemma [ fine_lem4 ] ,
the sheaf @xmath1695 is a line bundle @xmath339 on @xmath1681 .
then we have a universal extension @xmath1696 on @xmath1684 .
then using lemma [ fine_lem4 ] it is easy to see that for every @xmath1693 , @xmath1686 is isomorphic to the cokernel of a generic morphism associated to @xmath1682 .
let @xmath1621 be a special morphism .
let the associated application of rank 1 .
let @xmath82 be the hyperplane of @xmath114 defined by @xmath1697 .
we have an exact sequence @xmath1674 & { { \mathcal o}}(-2)\ar[r ] & { { \mathcal o}}(-2)\ot{{\mathbb c}}^2\ar[r]^-{\phi_1 } & { { \mathcal o}}(-1)\ar[r ] & { { \mathcal o}}_h(-1)\ar[r ] & 0 . } \ ] ] let @xmath1698 be the linear map induced by @xmath1630 .
let @xmath1700 , @xmath1701 be the two factors @xmath1702 of .
we can suppose thet .
let @xmath1703 @xmath1704 , be the restrictions of @xmath1630 , defined by .
let @xmath1705 be a basis of @xmath513 , such that @xmath509 is an equation of @xmath82 . by using the action of on @xmath83 we can assume that .
now @xmath1699 is not zero on @xmath1700 : otherwise we would have @xmath1706 , and , with @xmath1707 and this would contradict the semi - stability of @xmath1620 .
hence , by considering the action of @xmath1708 , it suffices to prove that @xmath1699 does not vanish on @xmath1701 .
suppose it does .
then @xmath1630 vanishes on @xmath1709 , because , and again , with , which contradicts the semi - stability of @xmath1620 .
[ fine_lem6]*lemma * : suppose that @xmath1620 is semi - stable with respect to some polarization .
then it is injective outside of a closed subvariety of codimension @xmath118 , and @xmath1660 has no torsion .
it suffices to prove the first statement .
let @xmath1651 and @xmath1710 such that .
then we have either @xmath1711 or @xmath1712 and @xmath1713 .
suppose that @xmath1620 is not injective at all points of an irreducible hypersurface @xmath1714 .
then the same is true for @xmath1715 .
suppose that this morphism is defined by quadratic forms @xmath1716 .
these forms vanish on @xmath1717 , hence they are all multiple of an equation of @xmath1717 .
it follows as in the proof of [ fine_lem5 ] that , with @xmath1718 , which contradicts the semi - stability of @xmath1620 .
* dfinition :* [ fine_def]let @xmath960 be a smooth variety , @xmath9 a coherent sheaf on @xmath1719 , flat on @xmath960 .
we say that @xmath960 is a _ fine moduli space of sheaves _ with _ universal sheaf _
@xmath9 if the following properties are verified : 1 . for every closed point @xmath1720
the kodara - spencer map @xmath1721 is bijective .
2 . for every closed points
@xmath1722 with @xmath1723 , @xmath1724 and @xmath1725 are not isomorphic .
3 . for every flat family @xmath8 of coherent sheaves on @xmath114 parametrized by an algebraic variety @xmath1726 , and for any closed points @xmath1720 , @xmath1727 such that @xmath1728 , there exists an open neighbourhood @xmath134 of @xmath338 in @xmath1726 , and a morphism @xmath1729 such that @xmath1730 and @xmath1731 by theorem [ theo_main2 ] , there exists a projective good quotient of the open subset @xmath95 of semi - stable points as soon as @xmath1734 the critical polarizations in our range are given by @xmath1735 let @xmath1736 + 1\ ] ] ( where @xmath1737 $ ] denotes the integer part of @xmath249 ) . then we obtain exactly @xmath553 moduli spaces of morphisms corresponding to non critical values : @xmath120 , where for @xmath1738 @xmath1739 with @xmath1740 $ ] , @xmath1741 beeing a sufficiently small positive rational number .
we have ( cf .
the end of [ fine_obv ] ) .
suppose that we choose a polarization such that @xmath338 is not a critical value .
in this case we have @xmath1742 , and the stabilizer in @xmath6 of the points of @xmath97 is the canonical subgroup isomorphic to @xmath1702 .
let , and be the quotient map .
on @xmath1743 we have a universal morphism @xmath1744 ( where @xmath1745 is the projection @xmath1746 ) such that is a flat family of torsion free sheaves on @xmath114 parametrized by @xmath97 ( this is a consequence of lemmas [ fine_lem3 ] and [ fine_lem6 ] ) .
there is a canonical action of @xmath6 on @xmath9 such that @xmath1702 acts by multiplication .
[ fine_theo]*theorem :* there exists a @xmath6-line bundle @xmath333 on @xmath1747 such that @xmath1749 descends to @xmath1750 .
let @xmath8 be the corresponding sheaf on @xmath1747 .
then @xmath1750 is a fine moduli space of sheaves on @xmath114 with universal sheaf @xmath8 .
on @xmath97 we have a canonical action of @xmath6 on the bundles @xmath1751 , @xmath1752 and @xmath1753 . on these bundles @xmath1702 acts as ordinary multiplication by scalars .
let @xmath1754 , @xmath1755 be the @xmath6-bundles @xmath1756 ( where @xmath1757 is the projection @xmath1758 ) . on these bundles @xmath1702
acts trivially .
we can multiply the universal morphism with and we obtain a new universal morphism @xmath1759 now it is easy to see that the bundles @xmath1754 , @xmath1755 descend to @xmath1747 either directly from our construction of the quotient , or by using the more general results of @xcite , 2.3 .
the @xmath6-morphism @xmath1760 also descends and we get a universal morphism of vector bundles on @xmath1747 @xmath1761 we define now , and it is clear that .
now we prove that the kodara - spencer map of @xmath8 at @xmath1762 is bijective .
let @xmath1763 .
then we have a commutative diagram @xmath1764 ^ -{t\pi}\ar[d]^{\omega_w } & t_zm(t)\ar[d]^{\omega_z } \\ \ext^1({{\mathcal f}}_w,{{\mathcal f}}_w)\fleq[r ] & \ext^1({{\mathcal e}}_z,{{\mathcal e}}_z ) } \ ] ] the tangent map @xmath1765 is surjective because @xmath1750 is a geometric quotient .
so it suffices to prove that @xmath1766 is surjective and that .
consider the exact sequence @xmath1767 it is well known that ( up to a sign ) @xmath1768 is the composition @xmath1769 of maps induced by the preceeding exact sequence .
now the result follows easily from the exact sequence @xmath1770 we must now verify that if @xmath1771 are distinct closed points , then @xmath1772 and @xmath1773 are not isomorphic .
this follows from the more general following result : if two injective morphisms of vector bundles on @xmath114 @xmath1774 have isomorphic cokernels , then they are in the same orbit .
it follows that the @xmath553 moduli spaces of morphisms @xmath120 , with their corresponding universal sheaves , are also fine moduli spaces of torsion free sheaves on @xmath114 .
the moduli space @xmath1775 is the same as the _ obvious _ one * m * ( cf .
[ fine_obv ] ) , and the corresponding universal sheaf is the same ( up to an element of @xmath1776 ) as @xmath1683 .
these examples are generalizations of the case of @xmath32 ( with @xmath1777 ) that was treated in @xcite .
but in this case our results are not needed , because we get only two fine moduli spaces : one is the obvious moduli space and the other is the corresponding moduli space of stable sheaves on @xmath32 . 99 bialynicki - birula , a. , wicika , j. _ a recipe for finding open subsets of vector spaces with a good quotient .
77 , 1 ( 1998 ) , 97 - 114 bialynicki - birula , a. , wicika , j. _ open subsets of projective space with a good quotient by an action of a reductive group .
_ transform .
groups 1 , 3 ( 1996 ) , 153 - 185 dixmier , j. , _ quelques aspects de la thorie des invariants _ , gazette soc . math .
de france 1989 dixmier , j. , raynaud , m. , _ sur le quotient dune varit algbrique par un groupe algbrique _ , advances in mathematics , suppl .
studies , vol 7a , ( 1981 ) , 327 - 344 dolgachev , h. , hu , yi , _ variation of geometric invariant theory quotients _ , preprint alg - geom/94 02 008 drzet , j .-
_ fibrs exceptionnels et varits de modules de faisceaux semi - stables sur @xmath1779 .
reine angew .
( 1987 ) , 14 - 58 drzet , j .-
, _ cohomologie des varits de modules de hauteur nulle . _
ann . 281 ( 1988 ) , 43 - 85 drzet , j .-
m . , _ varits de modules extrmales de faisceaux semi - stables sur @xmath1779 ._math .
( 1991 ) , 727 - 770 drzet , j .-
m . , _ exceptional bundles and moduli spaces of stables sheaves on @xmath114 _ , in vector bundles in algebraic geometry , proceedings durham 1993 , london math . soc .
lecture note ser .
208 , cambridge 1995 drzet , j .-
_ quotients algbriques par des groupes non rductifs et varits de modules de complexes _ ,
journ . of math .
vol . 9 , 7 ( 1998 ) , 769 - 819 drzet , j
.- m . , _ varits de modules alternatives .
fourier 49 ( 1999 ) , 57 - 139 drzet , j .-
m . , _ espaces abstraits de morphismes et mutations .
reine angew .
518 ( 2000 ) , 41 - 93 drzet , j
.- m . , le potier , j. , _ fibrs stables et fibrs exceptionnels sur @xmath1779 _ ann . ec .
18 ( 1985 ) , 193 - 244 ellingsrud , g. , piene , r. , strmme , s.a .
_ on the variety of nets of quadrics defining twisted cubic curves .
_ in _ space curves _ , lect .
notes in math . 1266 , springer - verlag ( 1987 ) .
ellingsrud , g. , strmme , s.a . , _ on the chow ring of a geometric quotient _ , ann . of math . 130
( 1989 ) , 159 - 187 fauntleroy , a. , _ geometric invariant theory for general algebraic groups .
55 ( 1985 ) , 63 - 87 .
fauntleroy , a. , _ invariant theory for linear algebraic groups ii _ , comp .
68 ( 1983 ) , 23 - 29 freiermuth , h.g .
, _ on the moduli space @xmath1780 of semi
stable sheaves on @xmath1545 with hilbert polynomial @xmath1781 _ , diplomarbeit , kaiserslautlern 2000 .
greuel , g .-
pfister , g. , _ geometric quotients of unipotent group actions _
67 ( 1993 ) , 75 - 105 karpov , b.v . , _ semi - stable sheaves on a two - dimensional quadric and kronecker modules . _ math . izvestiya ams transl
. 40 ( 1993 ) , 33 - 66 .
king , a. , _ moduli of representations of finite dimensional algebras .
journ . of math .
oxford 45 ( 1994),515 - 530 .
kollr , j. , _ quotient spaces modulo algebraic groups _ , ann . of math .
145(1997 ) , 33 - 79 le potier , j. , _ systmes cohrents et structures de niveau _ , astrisque 214 , soc . math .
france , 1993 le potier , j. , _ faisceaux semi stables de dimension 1 sur le plan projectif _ , rev .
roumaine math .
pures et appl . 318 , 635 - 678 , 1993 mir - roig , r.m . , _ some moduli spaces for rank 2 stable reflexive sheaves on @xmath1782 .
299 ( 1987 ) , 699 - 717 .
mir - roig , r.m . , trautmann , g. , _ the moduli scheme @xmath1783 over @xmath1782_. math
. z. 216 ( 1994 ) , 283 - 315 .
mumford , d. , fogarty , j. , _ geometric invariant theory .
_ ergeb . math .
grenzgeb . bd .
berlin heidelberg new - york : springer ( 1982 ) nagata , m. , _ on the 14th problem of hilbert _ , proc .
intern . cong . of math .
1958 , edinburgh , 459 - 462 , cambridge university press 1960 newstead , p.e . , _ introduction to moduli problems and orbit spaces . _
berlin heidelberg new - york : springer ( 1978 ) okonek , c. , _ moduli extremer reflexiver garben auf @xmath1784 .
reine angew .
( 1983 ) , 183 - 194 .
popov , v.l .
, vinberg , e.g. _ invariant theory . _ in algebraic geometry iv : linear algebraic groups , invariant theory , encycl .
55 , 123 - 278 ( 1994 ) springer - verlag .
reid , m. , _ what is a flip _ , preprint 1992 schofield , a. , _ general representations of quivers .
65 ( 1992 ) , 46 - 64 .
seshadri , c.s . , _
mumford s conjecture for @xmath1708 and applications .
int . colloq . on algebraic geometry .
oxford univ .
press , 347 ( 1968 ) .
thaddeus , m. , _ geometric invariant theory and flips _ , journ .
( 1996 ) , 691 - 723 tjtta , e.n . _
rational curves on the space of determinantal nets of conics .
_ preprint math.ag/9802037 ( 1998 ) | we extend the methods of geometric invariant theory to actions of non reductive groups in the case of homomorphisms between decomposable sheaves whose automorphism groups are non reductive . given a linearization of the natural action of the group @xmath0 on hom(e , f ) , a homomorphism
is called stable if its orbit with respect to the unipotent radical is contained in the stable locus with respect to the natural reductive subgroup of the automorphism group .
we encounter effective numerical conditions for a linearization such that the corresponding open set of semi - stable homomorphisms admits a good and projective quotient in the sense of geometric invariant theory , and that this quotient is in addition a geometric quotient on the set of stable homomorphisms . |
the radiation environment in deep space or on the moon can be very harsh ( macfarlane et al . , 1991 ) . without the earth
s atmosphere and magnetic field to protect them , astronauts will be directly exposed to high - energy galactic cosmic rays ( gcr ) , to intense fluxes of protons and other particles from solar flares , and to radiation from on - board power plants or nuclear propulsion systems .
such radiation is also very dangerous to computers and electronics on spacecraft , as it may cause enough single - event upsets to lead to device failure .
the economic penalties of additional mass to provide shielding may be huge . to minimize this expense ,
we need to be able to reliably calculate the amount of shielding required . performing this task
requires nuclear data for reactions induced by different projectiles , on different targets , for a large range of incident energies .
similarly , addressing different astrophysical problems , such as investigation the origin and propagation of cosmic rays ( cr ) and the nuclide abundances in the solar system and in cr , again requires a large amount of different nuclear reaction data .
experiments to obtain such data are costly , there is a limited number of facilities available to make such measurements , and some reactions of interest can not yet be measured at current accelerators . therefore reliable models and codes are required to provide the necessary estimates . during recent years at the los alamos national laboratory
, we have developed an improved version of the cascade - exciton model ( cem ) , contained in the code cem2k , to describe nucleon - induced reactions at incident energies up to 5 gev ( mashnik and sierk , 2001 and 2002 ) and the los alamos version of the quark - gluon string model , realized in the high - energy code laqgsm ( gudima et al . , 2001 ) , able to describe both particle- and nucleus - induced reactions at energies up to 1000 gev / nucleon .
both codes have been tested against most of the available data and compared with predictions of other modern codes ( mashnik and sierk , 2001 and 2002 ; mashnik et al .
, 2002a-2002d ; titarenko et al . , 2002 ) .
our comparisons have shown that both codes describe a large variety of spallation , fission , and fragmentation reactions quite reliably and have one of the best predictive powers compared with other available monte - carlo codes . in the present paper ,
we outline our models and show several typical results that testify that both cem2k and laqgsm are reliable event - generators which can be used in many different applications . as an illustration of application of our evaluated cross sections in astrophysics ,
we show an estimate of the size of the cr halo obtained using the cr propagation code galprop ( strong and moskalenko , 1998 ; moskalenko et al . , 2002 ) .
a detailed description of the initial version of the cem may be found in gudima et al .
( 1983 ) , therefore we outline here only its basic assumptions .
the cem assumes that reactions occur in three stages .
the first stage is the intranuclear cascade ( inc ) in which primary particles can be re - scattered and produce secondary particles several times prior to absorption by or escape from the nucleus .
the excited residual nucleus remaining after the cascade determines the particle - hole configuration that is the starting point for the preequilibrium stage of the reaction .
the subsequent relaxation of the nuclear excitation is treated in terms of an improved modified exciton model ( mem ) of preequilibrium decay followed by the equilibrium evaporative final stage of the reaction .
generally , all three stages contribute to experimentally measured outcomes .
the improved cascade - exciton model in the code cem2k differs from the older cem95 version by incorporating new approximations for the elementary cross sections used in the cascade , using more precise values for nuclear masses and pairing energies , employing a corrected systematics for the level - density parameters , improving the approximation for the pion binding energy " , @xmath0 , adjusting the cross sections for pion absorption on quasi - deuteron pairs inside a nucleus , considering the effects of refractions and reflections from the nuclear potential on cascade particles , allowing for nuclear transparency of pions , including the pauli principle in the preequilibrium calculation , and improving the calculation of the fission widths .
implementation of significant refinements and improvements in the algorithms of many subroutines led to a decrease of the computing time by up to a factor of 6 for heavy nuclei , which is very important when performing simulations with transport codes .
essentially , cem2k has a longer cascade stage , less preequilibrium emission , and a longer evaporation stage with a higher excitation energy , as compared to its precursors cem97 and cem95 . besides the changes to cem97 and cem95 mentioned above
, we also made a number of other improvements and refinements , such as : ( i ) imposing momentum - energy conservation for each simulated event ( the monte carlo algorithm previously used in cem provides momentum - energy conservation only statistically , on the average , but not exactly for the cascade stage of each event ) , ( ii ) using real binding energies for nucleons at the cascade stage instead of the approximation of a constant separation energy of 7 mev used in previous versions of the cem , ( iii ) using reduced masses of particles in the calculation of their emission widths instead of using the approximation of no recoil used previously , and ( iv ) a better approximation of the total reaction cross sections . on the whole , this set of improvements led to a much better description of particle spectra and yields of residual nuclei and a better agreement with available data for a variety of reactions .
details , examples , and further references may be found in mashnik and sierk ( 2001 and 2002 ) and in titarenko et al .
( 2002 ) .
the los alamos version of the quark - gluon string model ( laqgsm ) by gudima et al . , ( 2001 ) is a next - generation of the quark - gluon string model ( qgsm ) by amelin et al .
( 1990 , and references therein ) and is intended to describe both particle- and nucleus - induced reactions at energies up to about 1 tev / nucleon .
the core of the qgsm is built on a time - dependent version of the intranuclear cascade model developed at dubna , often referred in the literature simply as the dubna intranuclear cascade model ( dcm ) ( see toneev and gudima , 1983 and references therein ) .
the dcm models interactions of fast cascade particles ( participants " ) with nucleon spectators of both the target and projectile nuclei and includes interactions of two participants ( cascade particles ) as well .
it uses experimental cross sections ( or those calculated by the quark - gluon string model for energies above 4.5 gev / nucleon ) for these elementary interactions to simulate angular and energy distributions of cascade particles , also considering the pauli exclusion principle .
when the cascade stage of a reaction is completed , qgsm uses the coalescence model described in toneev and gudima ( 1983 ) to create " high - energy d , t , @xmath1he , and @xmath2he by final state interactions among emitted cascade nucleons , already outside of the colliding nuclei . after calculating the coalescence stage of a reaction , qgsm moves to the description of the last slow stages of the interaction , namely to preequilibrium decay and evaporation , with a possible competition of fission using the standard version of the cem by gudima et al . , ( 1983 ) .
but if the residual nuclei have atomic numbers with @xmath3 , qgsm uses the fermi break - up model to calculate their further disintegration instead of using the preequilibrium and evaporation models .
laqgsm differs from qgsm by replacing the preequilibrium and evaporation parts of qgsm described according to the standard cem ( gudima et al . , 1983 ) with the new physics from cem2k ( mashnik and sierk , 2001 and 2002 ) and
has a number of improvements and refinements in the cascade and fermi break - up models ( in the current version of laqgsm , we use the fermi break - up model only for @xmath4 ) . a detailed description of laqgsm and further references may be found in gudima et al .
( 2001 ) . originally , both cem2k and laqgsm were not able to describe fission reactions and production of light fragments heavier than @xmath2he , as they had neither a high - energy - fission nor a fragmentation model .
recently , we addressed these problems ( mashnik et al
. , 2002a and 2002c ) by further improving our codes and by merging them with the generalized evaporation model code gem2 developed by furihata ( 2000 and 2001 ) .
we have benchmarked our codes on all reactions measured recently at gsi ( darmstadt , germany ) and on many other different reactions at lower and higher energies measured earlier at other laboratories .
we found that cem2k and laqgsm allow us to describe quite well a large variety of spallation , fission and fragmentation reactions at energies from 10 mev to 800 gev , without any free fitting parameters , though we still have to solve a number of problems for a better description of nuclides produced near the intersection of the spallation and fission regions .
our current versions of cem2k and laqgsm were incorporated recently into the mars ( mokhov , 1995 ) and lahet ( prael and lichtenstein , 1989 ) transport codes and are currently being incorporated into mcnpx ( waters , 1999 )
. this will allow others to use our codes as event - generators in these transport codes to simulate reactions with targets of practically arbitrary geometry and nuclide composition .
figure 1 shows examples of calculated by cem2k and laqgsm neutron spectra from interactions of protons with @xmath5pb at 0.8 gev , while figure 2 gives examples of p , d , t , @xmath1he , and @xmath2he spectra from p ( 61 mev ) + fe compared with experimental data by ishibashi et al .
( 1997 ) and bertrand and peelle ( 1973 ) .
we note that all reactions shown in figures 1 and 2 , and in all the following figures of this paper , were calculated within a single approach , without fitting any parameters of cem2k or laqgsm .
we see that both cem2k and laqgsm describe well spectra of secondary neutrons and protons .
similar results are obtained for other targets and energies of incident protons , as well as for reactions induced by neutrons , pions , and photons .
spectra of @xmath2he are also described by our codes quite well , while spectra of d , t , and @xmath1he are reproduced reasonably , but not as well as those of nucleons and @xmath2he .
this is partially due to the fact that we do not fit here the probability @xmath6 for @xmath7 excited nucleons ( excitons ) to condense into a complex particle which can be emitted during the preequilibrium stage of a reaction to get the best agreement with the data , as is often done in the literature by other authors ( see details in gudima et al . , 1983 and mashnik et al . , 2002c
) . in the present version of our models
we do not take into account direct production of complex particles like pick - up and knock - out , and this leads to some underestimation of emission of high - energy complex particles , and to under - prediction of the high - energy tails of their spectra seen in figure 2 .
we note that some of the evaporation models used often in the literature , like the gsi evaporation model by schmidt ( gaimard and schmidt , 1991 ; junghans et al . , 1998 ) , which is used in conjunction with the liege inc by cugnon ( boudard et al . ,
2002 ) in the code incl , do not consider preequilibrium particle emission ; they only evaporate n , p , and @xmath2he , and do not produce d , t , and @xmath1he at all .
recently , nakamura s group measured neutron double - differential cross sections from many reactions induced by light and medium nuclei on targets from @xmath8c to @xmath5pb , at several incident energies from 95 to 600 mev / nucleon ( iwata et al . , 2001 and references therein ) .
we have calculated all these cross sections using laqgsm . as an example ,
our results for interactions of 600 mev / nucleon @xmath9ne with cu are compared in figure 3 with experimental data and calculations with the qmd ( aichelin , 1991 ) and hic ( bertini et al . , 1974 ) models kindly provided to us by nakamura s group .
we see that laqgsm describes these data quite well and agrees with the measurements much better than do qmd and hic .
similar results are obtained for all the other reactions measured by this group .
recently , for the proton radiography ( prad ) project , we have performed ( mashnik et al . , 2002b ) a benchmark of qgsm ( and laqgsm ) , mars , and lahet3 ( prael , 2001 ) against all measured double - differential cross sections of proton - nucleus reactions at energies around 50 gev for all targets for which we were able to find experimental data .
we calculated , compared with experimental data , and analyzed more than 500 spectra of p , d , t , @xmath1he , @xmath2he , @xmath10 , @xmath11 , @xmath12 , @xmath13 , and @xmath14 emitted from targets from @xmath15be to @xmath5pb at angles from 0 to 159 deg at incident proton energies from 30 to 70 gev .
this study has shown that laqgsm describes well most of the measured spectra .
a detailed and comprehensive report on this work is now in preparation , while figure 4 shows just one example , namely spectra of deuterons emitted at 9.17 deg from p ( 70 gev ) + @xmath5pb .
we see that deuterons with momentum of up to about 15 gev / c are emitted and measured in this particular reaction .
utilizing the coalescence mechanism for complex particle emission , laqgsm is able to describe high - energy deuteron production , and agrees well with the measurement .
lahet3 does not consider the coalescence of complex particles and therefore describes emission of only evaporative and preequilibrium deuterons with momenta not higher than 1 gev / c .
even though the cross section for emission of deuterons with momentum of @xmath16 gev / c is more than six orders of magnitude lower than for evaporation of low energy deuterons , such high energy deuterons and other complex particles may be extremely dangerous to people and equipment in space .
therefore , we need to be able to calculate such reactions as well as possible , to accurately estimate the necessary shielding . recently at gsi in darmstadt , germany , a large amount of measurements have been performed using inverse kinematics for interactions of @xmath17fe , @xmath5pb and @xmath18u at 1 gev / nucleon and @xmath19au at 800 mev / nucleon with liquid @xmath20h .
these measurements provide a very rich set of cross sections for production of practically all possible isotopes from such reactions in a pure " form , _
i.e. _ , individual cross sections from a specific given bombarding isotope ( or target isotope , when considering reactions in the usual kinematics , p + a )
. such cross sections are much easier to compare to models than the camouflaged " data from @xmath21-spectrometry measurements .
these are often obtained only for a natural composition of isotopes in a target and are mainly for cumulative production , where measured cross sections contain contributions not only from the direct production of a given isotope , but also from all its decay - chain precursors . in addition , many reactions where a beam of light , medium , or heavy ions with energy near to or below 1 gev / nucleon interact with different nuclei , from the lightest , d , to the heaviest , @xmath5pb were measured recently at gsi .
references on these measurements and many tabulated experimental cross sections may be found on the web page of prof .
schmidt ( 2002 ) . we have analyzed with cem2k and laqgsm all measurements done at gsi
we are acquainted of , both for proton - nucleus and nucleus - nucleus interactions .
some examples of our cem2k results compared with the gsi data and calculations by other current models for proton - nucleus reactions may be found in mashnik and sierk ( 2001 and 2002 ) , mashnik et al .
( 2002a , 2002c , and 2002d ) , and titarenko et al .
figure 5 shows an example of laqgsm results for the reaction p(1 gev ) + @xmath5pb compared with gsi data and calculations by a version of lahet using the incl event - generator incorporated recently into lahet .
we have described incl above ; it is well known and often used in europe .
one can see that laqgsm merged with gem2 ( laqgsm+gem2 ) describes quite well the gsi data and agrees better with the measurement than incl does .
similar results are obtained for all other reactions measured at gsi for which we found data . figure 6 shows an example of the highest energy we calculated so far with laqgsm , namely results by laqgsm+gem2 for the reaction p(800 gev ) + a compared with experimental data ( sihver et al . , 1992 ) and calculations with the phenomenological code yieldx ( silberberg et al . , 1998 )
for the sake of brevity , we present in this figure only every third yield measured and tabulated in sihver et al .
( 1992 ) , though we calculated all possible products from this reaction and get similar results for nuclides not shown here .
one can see that laqgsm agrees reasonably with most of the measured yields and describes the data better than yieldx .
more than a half of the measured products are described by laqgsm+gem2 with an accuracy of a factor of two or better , though we find some large discrepances for several nuclides like @xmath22hf and @xmath23rh .
at last , figure 7 shows a heavy - ion induced reaction measured at gsi ( junghans , 1997 ; junghans et al . , 1998 ) , namely the yields of measured products ( black circles ) from the interaction of a 950 mev / nucleon @xmath18u beam with copper compared with our laqgsm+gem2 results ( open circles ) .
one can see that laqgsm+gem2 describes most of these data with an accuracy of a factor of two or better .
this section shows an example of using cross sections calculated by cem2k and laqgsm ( together with available data ) to put constraints on the galactic cr halo size . for this calculation
we use a state - of - the - art cr propagation code galprop .
the galprop models have been described in full detail elsewhere ( strong and moskalenko , 1998 ; moskalenko et al . , 2002 and references therein ) ; here we summarize their basic features .
the code solves a transport equation on a full 3d spacial grid @xmath24 or 2d grid .
the 2d models have cylindrical symmetry in the galaxy , and the basic coordinates are @xmath25 , where @xmath26 is galactocentric radius , @xmath27 is the distance from the galactic plane and @xmath28 is the total particle momentum .
the propagation region is bounded by @xmath29 ( taken @xmath30 kpc ) , @xmath31 beyond which free escape is assumed . for a given @xmath32 the diffusion coefficient as a function of momentum and the reacceleration parameters
is determined by the energy - dependence of the b / c ratio .
the spatial diffusion coefficient is taken as @xmath33 , assuming independence of position , where @xmath34 is rigidity .
for the case of reacceleration the momentum - space diffusion coefficient @xmath35 is related to the spatial coefficient .
the reacceleration is parameterized by @xmath36 where @xmath37 is the alfvn speed and @xmath38 the ratio of wave energy density to magnetic field energy density .
the source spectrum of nuclei is assumed to be a power law in momentum , @xmath39 for the injected particle density , if necessary with a break .
the interstellar hydrogen distribution uses h i and co surveys and information on the ionized component ; the helium fraction of the gas is taken as 0.11 by number .
energy losses of nuclei by ionization and coulomb interactions are included .
the distribution of cr sources is chosen to reproduce the cr distribution determined by analysis of egret gamma - ray data .
the primary source abundances are adjusted to give as good agreement as possible with the observed abundances after propagation , for a given set of cross - sections .
the heliospheric modulation is taken into account using the force - field approximation .
the nuclear reaction network is built using the nuclear data sheets .
the isotopic cross section database now includes the lanl t-16 compilation by mashnik et al .
( 1999 ) consisting of several tens of thousands of experimental points .
this includes a critical re - evaluation of some data and cross checks .
the isotopic cross sections are calculated in galprop using the authors fits to major beryllium and boron production cross sections c , n , o @xmath40 be , b , and to other major reactions ; all other cross sections at present are calculated using the webber et al .
( 1990 ) and/or silberberg et al .
( 1998 ) phenomenological approximations renormalized to the data where it exists .
the reaction network is solved starting at the heaviest nuclei ( i.e. @xmath41ni ) , solving the propagation equation , computing all the resulting secondary source functions , and proceeding to the nuclei with @xmath42 .
the procedure is repeated down to @xmath43 . in this way
all secondary , tertiary etc .
reactions are automatically accounted for . to be completely accurate for all isotopes , e.g. for some rare cases of @xmath44-decay
, the whole loop is repeated twice . for some astrophysically important reactions we produced our own evaluations of excitation functions ( e.g. , moskalenko et al . , 2001 ) instead of using only scarce experimental data or calculations by stand - alone phenomenological systematics or nuclear reaction models .
such evaluated data files ( mashnik et al . 1999 ) proved to be useful , e.g. , to study production of radioisotopes for medical and industrial applications using high power accelerators ( van riper et al . , 2001 ) . at present ,
neither available experimental data nor any of the current models or phenomenological systematics can be used alone to produce a reliable evaluated activation cross section library covering a wide range of target nuclides and incident energies .
instead , such an evaluated library should be created by constructing excitation functions using all available experimental data along with calculations employing some of the most reliable codes in the regions of targets and incident energies where they are most applicable .
when there are reliable experimental data , they , rather than model results , should be taken as the highest priority for the evaluation .
for this purpose we used all available to us experimental data from the lanl t-16 compilation ( mashnik et al .
, 1999 ) together with calculations by cem2k , and for several reactions , by laqgsm and the older versions of the cem code , cem97 and cem95 .
one example of such an evaluated excitation function , for the reaction @xmath45si(p , x)@xmath46al , is shown in figure 8 together with available data , cem2k results , and calculations using phenomenological systematics by webber et al .
( 1990 ) and silberberg et al . (
it is seen that cem2k has some problems in a correct description of this particular cross section near the threshold ; therefore we used abundant experimental data available for this reaction to produce our evaluated excitation function at these energies .
it is clear that neither the webber et al .
( 1990 ) systematics nor the silberberg et al .
( 1998 ) approximation describe correctly this excitation function ; using their results as an input to cr propagation codes may lead to errors in results and interpretation . the results of the calculation of galactic propagation of radioactive isotopes @xmath46al , @xmath47cl , and @xmath48mn are shown in figure 9 , where the evaluated excitation functions used were produced as described above .
the radioactive isotopes of these elements are the main astrophysical `` time clocks '' which together with stable secondary isotopes allow us to probe global galactic properties , such as the diffusion coefficient and the halo size . based on the cr data from spacecraft ( ace , ulysses , and voyager , for details
see moskalenko et al . , 2001 ) we were able to restrict the halo size as @xmath49 kpc . using the semiempirical systematics yields less consistent results ( see , e.g. , strong and moskalenko , 2001 ) rizing questions about the interpretation .
this result supports the conclusion that large uncertainties for the halo size obtained in previous works were mostly due to cross - section inaccuracies . in future
, we plan to develop evaluated data libraries for other astrophysical reactions of interest and to use them in future studies of galactic cr propagation .
from the results presented here and in the cited references , we conclude that cem2k and laqgsm describe well ( and without any fitting parameters ) a large variety of medium- and high - energy nuclear reactions and are suitable for evaluations of nuclear data for science and applications .
we continue our work on further improvements and development of both cem2k and laqgsm , but even in their present versions they are quite reliable and may be used as event - generators for astrophysical applications .
we thank prof . nakamura , drs .
iwata , and iwase for sending us numerical values of their measured neutron spectra and results of calculations with qmd and hic .
this study was supported by the u. s. department of energy and by the moldovan - u .
s. bilateral grants program , crdf project mp2 - 3025 .
s.g.m . and i.v.m
. acknowledge partial support from a nasa astrophysics theory program grant .
abramov , v. v. , b. yu .
baldin , a. f. buzulutskov , et al . , production of deuterons and antideuterons with large @xmath50 in @xmath51 and @xmath52 collisions at 70 gev , _ sov .
_ , * 45 * , 845 - 851 , 1987 .
aichelin , j. , quantum " molecular dynamics a dynamical microscopic @xmath53-body approach to investigate fragment formation and nuclear equation of state in heavy ion collisions , _ phys . rep .
_ , * 202 * 233 - 360 , 1991 .
amelin , n. s. , k. k. gudima , and v. d. toneev , further development of the model of quark - gluon strings for the description of high - energy collisions with a target nucleus , _ sov . j. nucl
_ , * 52 * , 172 - 178 , 1990 .
bertrand f. e. , and r. w. peelle , complete hydrogen and helium particle spectra from 30 to 60 mev proton bombardment on nuclei with a = 12 to 209 and comparison with the intranuclear cascade model , _ phys .
rev . _ * c8 * , 1045 - 1064 , 1973 .
ishibashi , k. , h. takada , t. nakomoto , et al .
, measurement of neutron - production double - differential cross sections for nuclear spallation reaction induced by 0.8 , 1.5 and 3.0 gev protons , _ j. nucl .
_ , * 34 * , 529 - 537 , 1997 .
mashnik , s. g. , a. j. sierk , k. a. van riper , and w. b. wilson , production and validation of isotope production cross section libraries for neutrons and protons to 1.7 gev , _ proc .
sare4 ( knoxville , usa ) _ , ed .
t. a. gabriel , pp .
151 - 162 , oak ridge , 1999 ; ( nucl - th/9812071 ) ; our compilation ( lanl t-16 lib ) is permanently updated as new experimental data become available to us .
mashnik s. g. , and a. j. sierk , recent developments of the cascade - exciton model of nuclear reactions , _ proc .
nd2001 ( tsukuba , japan ) _ , _ j. nucl .
_ , * supplement 2 * , 720 - 725 , 2002 ; ( nucl - th/0208074 ) .
mashnik , s. g. , k. k. gudima , and a. j. sierk , merging the cem2k and laqgsm codes with gem2 to describe fission and light - fragment production , _ proc .
satif-6 ( slac , usa ) _ ; lanl report la - ur-02 - 0608 , los alamos , 2002a .
mashnik , s. g. , k. k. gudima , n. v. mokhov , r. e. prael , and a. j. sierk , benchmarking three codes for proton radiography applications , _ proc .
satif-6 ( slac , usa ) _ ; lanl report la - ur-02 - 0304 , los alamos , 2002b .
mashnik , s. g. , a. j. sierk , and k. k. gudima , complex - particle and light - fragment emission in the cascade - exciton model of nuclear reactions , _ proc .
rpsd 2002 ( santa fe , nm ) _ ; lanl report la - ur-02 - 5185 , 2002c ; ( nucl - th/0208048 ) .
mashnik s. g. , r. e. prael , a. j. sierk , et al .
, benchmarking ten codes against the recent gsi measurements of the nuclide yields from @xmath5pb , @xmath19au , and @xmath18u + p reactions at 1 gev / nucleon , _ proc .
nd2001 ( tsukuba , japan ) _ , _ j. nucl .
sci . techn .
_ , * supplement 2 * , 785 - 788 , 2002d ; ( nucl - th/0208075 ) .
van riper , k. a. , s. g. mashnik , and w. b. wilson , a computer study of radionuclide production in high power accelerators for medical and industrial applications , _ nucl .
_ , * a463 * , 576 - 585 , 2001 .
webber , w. r. , j. c. kish , and d. a. schrier , formula for calculating partial cross sections for nuclear reactions of nuclei with @xmath58 mev / nucleon in hydrogen targets , _ phys .
_ , * c41 * , 566 - 571 , 1990 . | the cem2k and laqgsm codes have been recently developed at los alamos national laboratory to simulate nuclear reactions for a number of applications .
we have benchmarked our codes against most available data measured at incident particle energies from 10 mev to 800 gev and have compared our results with predictions of other current models used by the nuclear community . here , we present a brief description of our codes and show some illustrative results that testify that cem2k and laqgsm can be used as reliable event generators for space - radiation - shielding , cosmic - ray propagation , and other astrophysical applications . finally , we show an example of combining of our calculated cross sections with experimental data from our lanl t-16 compilation to produce evaluated files .
such evaluated files were successfully used in the model of particle propagation in the galaxy galprop to better constrain the size of the cosmic - ray halo . |
methanol masers are traditionally divided into two classes .
although masers of both classes often co - exist in the same star - forming region , they are usually seen apart from each other .
class ii masers are found in the vicinity of the massive ysos while class i masers are believed to trace distant parts of the outflows from these ysos .
class ii methanol masers display strong emission in transitions at 6.6 and 12.1 ghz ( @xcite ) .
more than 2 dozens of weaker class ii maser transitions are detected in several sources ( see , e.g. , @xcite ) . at present there
exist 3 unified catalogues of class ii methanol masers containing more than 500 objects ( @xcite , @xcite and @xcite ) and references to the initial data .
distinguishing feature of the @xcite catalogue is that it contains data on several maser transitions and cross - references to the data on molecular shock tracers .
most of the class ii maser sources were found toward positions of the iras sources while the blind surveys brought some detections in the places without other evidences of star formation . at present class ii masers
were found only in the high mass star forming regions ( @xcite ) .
class i methanol masers are less strong and widespread .
most of them were found in vicinities of class ii maser sources ( see , e.g. , @xcite and @xcite ) . there were extensive searches ( e.g. , @xcite and @xcite ) for class i sources but no unified catalogues of these objects do exist and no blind surveys were performed yet .
the transitions of the series at about 25 ghz in omc-1 were the first methanol masers detected in the interstellar medium .
existence of the interferometry data ( @xcite ) allowed construction of the models constraining characteristics of the turbulence in this source ( @xcite ) .
analysis of existing observational data on class i methanol masers allows to distinguish 4 regimes differing by the series of the brightest ( in terms of brightness temperature ) line .
most widespread maser regime is likely to have brightest lines which belong to the series .
the lines at 36.1 ghz and at 84.5 ghz are likely to be a weak masers under normal conditions of the massive star forming region .
usually maser nature of the lines in this regime is difficult to prove observationally .
anyhow , there are cases when the line profiles contain narrow spikes and the maser nature is proved interferometrically .
the sources sgr b2 , g30.8 - 0.1 , and g1.6 - 0.025 can be considered as representatives of this maser regime ( @xcite ) . in the other maser regime the lines of the series become prevalent .
numerous sources manifest definitely maser lines arising in the and transitions at 44.1 and 95.2 ghz , respectively .
masers in the sources dr21w , ngc2264 and omc-2 represent this regime ( @xcite ) .
preliminary theoretical analysis of the pumping shows that the lines of series become brightest in the models with rather high beaming ( @xmath0 ) and moderate column densities ( sobolev , ostrovskii & voronkov , in prep . ,
presented in the poster at iaus227 ) .
next maser regime is less widespread and is represented by the sources where the lines of the series at about 25 ghz are the brightest .
omc-1 is the famous example of such sources ( @xcite ) and some bright sources were found recently ( @xcite ) .
lines of this series become brightest in the models with the high specific column densities ( @xmath1 [ @xmath2c])@xmath3 ) and require relatively high temperatures ( @xmath4 k ) and densities ( @xmath5 [ @xmath6)=5 - 7 $ ] ) .
existence of the forth maser regime with the brightest line at 9.9 ghz from the series was previously uncertain .
recent atca observations have shown that this regime is actually realized in the sources w33a and g343.12 - 0.06 ( see figure [ fig : fg1 ] ) .
preliminary modelling shows that the line at 9.9 ghz becomes brightest in the models with specific column densities which are as high as required for the previous regime but prefers greater beaming and either lower densities or higher temperatures ( @xmath7 k )
. 25 ghz in g343.12 - 0.06 .
contours in the map are ( 1,2,3,4,5,6,7,8,9)x0.9 jy / beam.,width=508 ] so , class i methanol masers are potentially a good probe of such physical parameters as the temperature , density and column density . both observations and modelling of class ii masers show that the 6.6 ghz transition always manifests the highest brightness temperature ( @xcite , @xcite ) .
observations did not find significant differences in positions of the maser spots ( see , e.g. , @xcite , @xcite ) .
so , there is basically only one known pumping regime of class ii masers .
however , ratios of the brightnesses of different maser lines show considerable dependence on the values of the physical parameters .
this allows determination of the physical parameters in individual sources ( on the basis of multi - transitional observations , e.g. , @xcite , @xcite ) and construction of the model of the `` common '' class ii methanol maser source ( on the basis of extensive surveys , e.g. , @xcite and @xcite ) .
results of such studies are summarized in @xcite and @xcite and will not be considered here .
different response of the maser fluxes to the changes of the physical parameters provides possibility to elucidate nature of the time variability and estimate physical parameters in the source via modelling of variability in the different maser lines .
for example , interferometric studies of variability in the source g9.62 + 0.20 show that the changes in the different parts of the source are synchronized with the speed of light ( @xcite ) . in this case
the variability is most probably caused by the changes of the dust temperature which follow variations in the brightness of the central yso reflecting the character of the accretion process .
extensive surveys have shown that the radial velocities of class ii methanol masers are usually coincident with those of associated molecular cores to within several km / s ( see , e.g. , @xcite and @xcite ) .
this provides a possibility for making estimates of distances to these sources using models of rotation curves of the gas in the galaxy ( the kinematic distances ) . on the basis of such estimates
several authors discussed the distribution of methanol maser sources in the galaxy ( see , e.g.,@xcite and @xcite ) .
it should be noted that for the most of the sources kinematic estimates are ambiguous and allow 2 solutions : `` near '' and `` far '' ones .
anyhow , it was firmly established that the most of methanol masers are situated in the molecular ring of the galaxy .
we addressed the question of the kinematic distance ambiguity from the statistical point of view .
of course , sometimes the `` near '' estimate is closer to the real distance to the source and sometimes the `` far '' estimate appears to be more realistic .
we looked at the extremes .
this is reasonable because in the top view of the maser distribution there is a pronounced dip of the maser number density between the `` near '' and `` far '' distance domains .
firstly , we assumed that the `` far '' estimate of kinematic distance is correct .
it was found that in this case the dependence of the average peak maser flux on the distance from the sun is odd and shows general tendency to increase when the distances exceed 5 kpc .
such dependence is certainly unrealistic . secondly , we considered the dependence of the average peak maser flux on the distance from the sun for the `` near '' distance estimates . in this case
there is a pronounced tendency of the decrease with the distance from the sun .
taking into account that the weaker masers at greater distances are missed due to the limited sensitivity of the surveys we obtained the dependence which is quite close to the natural , i.e. , fall which is proportional to the square of the distance from the sun .
so , confinement to the `` near '' kinematic distances is likely to produce statistical results close to reality . guided by this hypothesis we constructed the radial dependence of the maser number density as a function of the distance from the galactic center which is shown in the figure [ fig : fg2 ] .
this dependence shows pronounced peaks at the distances which are close to those of the spiral arms in the current model of the galaxy .
maser statistics shows that the lifetime of the methanol masers is rather limited ( see , e.g.,@xcite and @xcite ) .
hence , maser sources are a good tracers of regions of the early massive star formation which are likely to be concentrated to the spiral arms .
so , the correlation which is found does look promising for the studies of the structure of our galaxy .
consideration of the selection effect related to the limited sensitivity of the surveys leaves the maser number density peaks at their places .
we also constructed the 6.6 ghz methanol maser luminosity function using `` near '' distances .
it was found that the population of masers is greatly dominated by the low luminosity sources and future sensitive surveys are likely to increase the number of the known sources substantially .
it should be noted that the surveys in the regions of the spiral arm tangents are the most promising in the sense of new detections but provide little information for the studies of the structure of the galaxy . using the unified catalogue we studied the spread of the ratios of the brightnesses in the different maser transitions as well as the correlation between the maser ranges and the breadths of such molecular shock tracers as cs and sio lines ( @xcite ) .
it was shown that : physical conditions within the usual maser source vary considerably ; maser brightness is determined by parameters of some distinguished part of the object - maser formation region ; class ii methanol masers are formed in the regions affected by the shock propagation .
we aimed on finding the relation between the bulk motions in the maser formation region with those of the dense gas of the ambient cloud .
observations show that the peak velocities of the different maser lines often do not coincide ( @xcite ) and the relative brightness of the peaks varies with time ( @xcite ) . in this situation
the mean velocity of the maser formation region is better represented by the center of the range of velocities of the maser features .
this should be true for the case when the source is strong enough . in order to get the data on the cs(2 - 1 ) velocities of the brightest masers we carried out observations of 31 source using the mopra antenna .
this brought us 18 new detections and a collection of the better quality spectra . as a result we obtained accurate cs(2 - 1 ) velocities for practically all masers with the 6.6 ghz flux density exceeding 150 jy .
we assumed that the maser has significantly shifted velocity if the center of the maser range was apart from the cs(2 - 1 ) line center by more than 0.75 km / s .
it was found that the difference in the numbers of the red - shifted and blue - shifted masers is statistically insignificant
. however , the distributions of maser sources with the different sense of the shift display considerable difference .
figure [ fig : fg3 ] shows that the blue - shifted maser sources with galactic longitudes from 320@xmath8 to 350@xmath8 form a distinguished group which is located in the region of the scutum - centaurus spiral arm .
this can reflect existence of a grand design , i.e. , grouping of the sources with similar peculiarity of morphology or of the evolutionary stage of the massive star forming regions . diagram with co emission in the background.,width=288 ]
we have shown that the methanol masers represent a powerful diagnostic tool for the studies of the nature of the massive star formation regions and their distribution in the galaxy . | methanol masers which are traditionally divided into two classes provide possibility to study important parts of the star forming regions : class ii masers trace vicinities of the massive ysos while class i masers are likely to trace more distant parts of the outflows where newer stars can form .
there are many methanol transitions which produce observed masers .
this allows to use pumping analysis for estimation of the physical parameters in the maser formation regions and its environment , for the study of their evolution .
extensive surveys in different masing transitions allow to conclude on the values of the temperatures , densities , dust properties , etc . in the bulk of masing regions .
variability of the brightest masers is monitored during several years . in some cases
it is probably caused by the changes of the dust temperature which follow variations in the brightness of the central yso reflecting the character of the accretion process .
a unified catalogue of the class ii methanol masers consisting of more than 500 objects is compiled .
analysis of the data shows that : physical conditions within the usual maser source vary considerably ; maser brightness is determined by parameters of some distinguished part of the object - maser formation region ; class ii methanol masers are formed not within the outflows but in the regions affected by their propagation .
it is shown that the `` near '' solutions for the kinematic distances to the sources can be used for statistical analysis . the luminosity function of the 6.7 ghz methanol masers is constructed .
it is shown that improvement of the sensitivity of surveys can increase number of detected maser sources considerably .
the distribution of class ii methanol masers in the galaxy is constructed on the basis of estimated kinematic distances .
it is shown that most of the sources are located in the molecular ring and that the dependence of the number of sources on the distance from the galactic center has significant peaks at the positions corresponding to the spiral arms .
a survey of cs(2 - 1 ) line emission tracing dense gas is performed at mopra toward the positions of the brightest class ii methanol masers .
velocity correlations between the maser and cs lines are analyzed .
it is shown that the sources with l from 320 to 350 deg in which the masers are relatively blue - shifted , form a group which is located in the region of the scutum - centaurus spiral arm
. this can reflect existence of a grand design , i.e. , grouping of the sources with similar peculiarity of morphology or evolutionary stage of the massive star forming regions . |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.