article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
the vaidya metric@xcite , which in radiation coordinates @xmath1 has the form [ vaidya ] ds^2 = -(1-)dw^2 + 2cdrdw + r^2d^2 , where @xmath2 , @xmath3 , is a solution of einstein s equations with spherical symmetry in the eikonal approximation to a radial flow of unpolarized radiation . for the case of an ingoing radial flow , @xmath4 and @xmath5 is a monotone increasing mass function in the advanced time @xmath6 , while @xmath7 corresponds to an outgoing radial flow , with @xmath5 being in this case a monotone decreasing mass function in the retarded time @xmath6 . for many years , it has been used in the analysis of spherical collapse and the formation of naked singularities ( for references , see the extensive list of @xcite and also @xcite ) .
it is known that vaidya metric can be obtained , by taking appropriate limits in the self - similar case , from the tolman metric representing spherically symmetric dust distribution@xcite .
this result has shed some light on the nature of the so - called shell - focusing singularities@xcite , discussed in details in @xcite .
the vaidya metric has also proved to be useful in the study of hawking radiation and the process of black - hole evaporation@xcite , and more recently , in the stochastic gravity program@xcite . motivated by the longly known
fact that the radiation coordinates are defective at the horizon@xcite , implying that the vaidya metric ( [ vaidya ] ) is not geodesically complete ( see @xcite for considerations about possible analytical extensions ) , waugh and lake@xcite considered the problem of casting the vaidya metric in double - null coordinates @xmath8 . as
all previous attempts to construct a general transformation from radiation to double - null coordinates have failed , they followed synge@xcite and considered einstein s equation with spherical symmetry in double - null coordinates _
ab initio_. the spherically symmetric line element in double - null coordinates is [ uv ] ds^2 = -2f(u , v)dudv + r^2(u , v)d^2 , where @xmath9 and @xmath10 are non vanishing functions .
the energy - momentum tensor of a unidirectional radial flow of unpolarized radiation in the eikonal approximation is given by [ t ] t_ab = h(u , v)k_a k_b , where @xmath11 is a radial null vector . as waugh and lake
, we will consider , without loss of generality , the case of the flow along @xmath12-direction : @xmath13 .
the case of simultaneous ingoing and outgoing flows was considered in @xcite .
einstein s equations are less constrained in such case , allowing the construction of some similarity solutions . in our case , the metric ( [ uv ] ) with the energy - momentum tensor ( [ t ] )
reduce to the following set of equations@xcite : @xmath14 where @xmath15 and @xmath16 are arbitrary functions obeying , according to the weak energy condition , [ we ] b(v)a(v ) 0 .
note that from ( [ ef ] ) , the regularity of @xmath9 requires @xmath17 .
schwarzschild solution in the kruskal - szekeres form corresponds to @xmath18 . for @xmath19 , the choice 2b(v )
= - allows one to interpret @xmath16 as the mass of the solution and @xmath12 as the proper time as measured in the rest frame at infinity for the asymptotically flat case@xcite .
we assume here @xmath20 .
the radial flow is ingoing if @xmath21 , and outgoing if @xmath22 .
note that if the weak energy condition ( [ we ] ) holds , the function @xmath16 is monotone , implying that the radial flow must be ingoing or outgoing for all @xmath12 .
it is not possible , for instance , to have `` oscillating '' mass functions @xmath16 in double null coordinates .
the problem may be stated in the following way : given the functions @xmath16 and @xmath15 , eq . ( [ er ] ) shall be integrated and @xmath10 is obtained .
then , @xmath9 and @xmath23 are calculated from ( [ ef ] ) and ( [ eh ] ) .
the arbitrary function of @xmath24 appearing in the integration of ( [ er ] ) must be chosen properly to have a non - vanishing @xmath9 function from ( [ ef ] ) .
unfortunately , this procedure is not analytically solvable in general .
waugh and lake , nevertheless , were able to find regular solutions for eqs .
( [ ef])-([eh ] ) for linear ( @xmath25 ) and a certain exponential ( @xmath26 ) mass functions ( @xmath27 , and @xmath28 are positive constants , @xmath3 , corresponding to ingoing / outgoing flow , respectively . ) . to the best of our knowledge ,
these are the only varying mass analytical solutions obtained in double - null coordinates so far .
we notice , however , that kuroda was able to construct a transformation from radiation to double - null coordinates for some particular mass functions@xcite . in the following section , we will present a semi - analytical approach to attack the problem of solving eqs .
( [ ef])-([eh ] ) for general mass functions obeying the weak energy condition ( [ we ] ) .
the approach allows us to construct qualitatively conformal diagrams , identifying horizons and singularities , and also to evaluate specific geometric quantities . before , however
, we notice that one can solve analytically eqs .
( [ ef])-([eh ] ) also for the case of [ inv ] a(v ) = , being @xmath29 a massive parameter , and @xmath30 . with the mass function ( [ inv ] ) , eq .
( [ er ] ) can be integrated as [ l ] l(r , u , v ) = p(u)v+2e^^2- v_0^e^s^2 ds=0 , where =( + 2 ) and @xmath31 is an arbitrary function .
it is quite easy to show that @xmath32 is an invariant surface of ( [ er ] ) , _
, one has = _ v r + = l along the solutions of ( [ er ] ) . from ( [ ef ] ) , we have f = -p(u ) v the function @xmath31 must the chosen to preclude zeros of @xmath9 . with @xmath33
, we have from ( [ l ] ) [ r ] = 2e^^2- _ 0^e^s^2 ds . equation ( [ r ] ) defines @xmath10 .
its graphics and the corresponding conformal diagram is presented in fig .
[ fig1 ] .
the only spacetime singularity for @xmath34 is @xmath35 .
we recall that he kretschmann scalar for metrics of the form ( [ uv ] ) obeying ( [ ef])-([eh ] ) is given by@xcite k = r_abcd r^abcd = 48 .
the resulting spacetime corresponds to a naked singularity that vanishes smoothly , giving rise to an empty spacetime .
its causal structure is identical to the first linear case ( @xmath36 ) of waugh and lake , with time reversed ( see fig .
[ fig3 ] below ) .
we finish this section noticing that algebraic manipulation programs ( maple , for instance ) are able to find solutions of ( [ er ] ) for the case @xmath37 in terms of airy functions , but they seem too cumbersome to be useful . furthermore ,
their causal structure is always similar to the ( @xmath36 ) linear case of waugh and lake .
also , kuroda has considered before the case of @xmath38 for small @xmath6 and @xmath39 in radiation coordinates , obtaining some local properties of the singularity @xmath40@xcite .
( [ er ] ) along @xmath24 constant is a first order ordinary differential equation in @xmath12 .
one can evaluate the function @xmath10 in any point by solving the @xmath12-initial value problem knowing @xmath41 .
the trivial example of minkowski spacetime ( @xmath42 ) , for instance , can be obtained by choosing @xmath43 .
the curve @xmath44 is the frontier of two regions of the @xmath45 plane where the solutions of ( [ er ] ) have qualitative distinct behaviors .
suppose , for instance , @xmath46 ( and @xmath47 .
the case of an outgoing radiation flow follows in a straightforward manner ) . for all points of the plane @xmath45 below this curve , @xmath48 ( see fig .
[ fig2 ] ) .
hence , any solution entering in this region will , unavoidably , reach the singularity at @xmath35 , with finite @xmath12 .
suppose a given solution @xmath49 with initial condition @xmath50 enters into this region .
as the uniqueness of solutions for ( [ er ] ) is guaranteed for any point with @xmath51 , solutions never cross each other in the plane @xmath45 for @xmath51 .
hence , any solution starting at @xmath52 is confined the the region below @xmath49 , and it will also reach the singularity at @xmath35 with finite @xmath12 . on the other hand ,
suppose that a given solution @xmath53 with initial condition @xmath54 never enters into region bellow the curve @xmath44 .
any solution starting at @xmath55 is , therefore , confined to the region above @xmath53 and will escape from the singularity at @xmath35 . keeping in mind that these solutions correspond to null trajectories with @xmath24 constant , it is possible to infer the causal structure of the spacetime from their qualitative behavior and an appropriate choice for the initial conditions @xmath41 .
for instance , in the case of fig .
[ fig2 ] , supposing that the mass function @xmath16 has an asymptotic value such that @xmath56 , an event horizon shall be located somewhere between @xmath49 and @xmath53 .
event horizons are global structures .
one needs to know the total ( @xmath57 ) spacetime evolution to define them . on the other hand ,
apparent horizons are defined locally@xcite .
indeed , they are the real relevant ones in our semi - analytical approach .
the curve @xmath44 is an apparent horizon . for a given @xmath12 , all solutions in fig .
[ fig2 ] such that @xmath58 will be captured by the singularity .
solutions for which @xmath59 are temporally free , but they may find themselves trapped later if @xmath16 increases . the event horizon is the last of these solutions trapped by the singularity . we remind that eq .
( [ ef ] ) requires that @xmath60 , implying that , for @xmath12 constant , @xmath61 for all @xmath24 .
this conditions holds everywhere we have uniqueness of the solutions of ( [ er ] ) , provided that the initial conditions are such that @xmath62 . moreover , if @xmath63 , then @xmath64 , implying from ( [ ef ] ) that the sign of @xmath9 will be determined by @xmath15 . for the ingoing radiation case , one has @xmath65 , and from ( [ uv ] ) one sees that the vector field @xmath66 is timelike .
our conformal diagram are oriented such that this vector field points upward .
if the condition @xmath67 holds , the conformal diagrams obtained for a given @xmath16 from different initial conditions are always equivalent .
some explicit examples will clarify the proposed approach .
the linear mass function solution discovered by waugh and lake corresponds to ( the ingoing radiation case , the outgoing one also follows in a straightforward way ) @xmath68 and @xmath69 . note that with these choices , eq .
( [ er ] ) does not satisfy the lipschitz condition in @xmath35 , implying that uniqueness is not guaranteed for solutions passing there .
this will be a crucial point to clarify the nature of the three qualitative different cases identified by waugh and lake , corresponding to @xmath70 , @xmath71 , and @xmath72 .
we notice that the linear mass case was also considered in @xcite in great detail and in a more general situation ( the case of a charged radial null fluid ) . the frontier of the region where all the solutions reach the singularity at @xmath35 is , in this case , the straight line @xmath73 .
taking the @xmath12-derivative of ( [ er ] ) , one gets r
= ( ( 1 -)-1 ) .
the regions in the plane @xmath45 where the solutions obeys @xmath74 are the straight lines [ rr ] r = ( 1 ) .
one can reproduce the analysis of waugh of lake for the three qualitative different cases according to the value of @xmath75 by considering the possible solutions of ( [ rr ] ) .
for this purpose , we will consider the solutions of ( [ er ] ) with the initial condition @xmath76 . for @xmath36
, the case showed in fig .
( [ fig3 ] ) , @xmath77 for all points with @xmath34 , and the only relevant frontier is the @xmath78 straight line ( the apparent horizon ) . all solutions of ( [ er ] ) are concave functions and cross the @xmath79 line , reaching the singularity with a finite @xmath12 which increases monotonically with @xmath24 .
the causal structure of the corresponding spacetime is very simple .
there is no horizon , and the future cone of all points ends in the singularity at @xmath35 .
each curve in fig [ fig3 ] corresponds to a constant @xmath24 slice of the the full solution @xmath10 .
it is possible , in principle , to reconstruct the 2-dimensional surface @xmath10 and to plot its lines of constant @xmath80 on the @xmath81 plane , as done in the fig .
( [ fig1 ] ) , reproducing all original results of waugh and lake@xcite . however , these lines are not necessary to construct the conformal diagrams . for @xmath82 , @xmath74 on the line @xmath83 .
this straight line itself is a solution .
all other solutions are concave functions .
we have two distinct qualitative behavior for the null trajectories along @xmath24 constant .
all solutions starting at @xmath84 are confined to the region above @xmath83 .
they never reach the singularity , all trajectories reach @xmath85 .
however , in the region below @xmath83 , we have infinitely many concave trajectories starting and ending in the ( shell - focusing ) singularity .
they start at @xmath86 , increase in the region between @xmath83 and @xmath87 , cross the last line and reach unavoidably @xmath35 again , with finite @xmath12 . the trajectory @xmath83 plays the role of an event horizon , separating two regions with distinct qualitative behavior : one where constant @xmath24 null trajectories reach @xmath85 and another where they start and end in the singularity
this situation is showed in the fig .
this behavior is only possible , of course , because the solutions of ( [ er ] ) fail to be unique at @xmath35 . for @xmath88 ( fig .
[ fig5 ] ) , we have three distinct regions according to the concavity of the solutions .
they are limited by the two straight lines ( [ rr ] ) , that are also solutions of ( [ er ] ) . between them , solutions are convex . above @xmath89 and below @xmath90 , solutions are concave .
the last line is the inner horizon .
inside , the null trajectories with constant @xmath24 start and end in the singularity @xmath35 .
the line @xmath89 is an outer horizon . between them ,
the constant @xmath24 null trajectories starts in the naked singularity and reach @xmath85 . beyond the outer horizon ,
all constant @xmath24 null trajectories escape the singularity . as an example of the proposed approach applied to new mass functions ,
let us consider first the case of a collapse of incoming radial null fluid , starting with the empty space .
this could be described , for instance , by the mass function [ collap ] a(v ) = ( 1+v ) .
the corresponding spacetime is empty for @xmath91 , it receives smoothly a radial flow of null fluid , and finishes as a black hole of mass @xmath92 for @xmath93 .
the solutions of ( [ er ] ) are presented in the fig .
[ fig6 ] . with the mass function ( [ collap ] )
, we always have a singularity at @xmath35 for @xmath94 .
the constant @xmath24 null trajectories approach @xmath35 asymptotically for @xmath91 .
some of them , confined by the horizon , are captured by the collapse and reach the singularity , and others scape to @xmath95 . any initial condition obeying @xmath96 leads to the same causal structure .
the proposed semi - analytical procedure can be used also to derive some qualitative results .
one could , for instance , locate accurately the event horizon in this case .
since we do not expect deviations from the minkowski spacetime for @xmath91 , we could start with initial conditions such that @xmath97 .
then , we identity in the solutions of ( [ fig6 ] ) that one ( the horizon @xmath98 ) approaching asymptotically the curve @xmath44 for large @xmath12 . for large values of @xmath99 , the horizon
would be located at @xmath100 . as a last example , let us consider the physically relevant case of black hole evaporation .
we remind that much work has been devoted on vaidya metric to describe black hole evaporation@xcite .
our approach allow us to analyze this question in the more convenient double - null coordinates .
the process of black hole evaporations supposes @xmath22 and an outgoing radiation flow .
hence , one needs to chose @xmath101 in our procedure . with this choice
, we see from ( [ ef ] ) that @xmath102 , implying that the field @xmath103 is not timelike anymore . it is namely its orthogonal vector field @xmath104 that will define the time evolution in this case . in order to keep the future direction pointing upward in the conformal diagrams
, one needs to re - orient the coordinates @xmath24 and @xmath12 .
the appropriate transformation here is [ tr ] vu , u - v . with this choice and
@xmath101 , equation ( [ er ] ) reads _
u r(u , v ) = ( 1- ) , which is formally identical to the equation for the ingoing case characterized by the mass function @xmath105 with radiation flow along the @xmath24 direction .
hence , the description of an outgoing situation with decreasing mass function @xmath106 can be obtained by applying the transformation ( [ tr ] ) for the ingoing case corresponding the mass function @xmath107 . with the help of ( [ tr ] )
, one could construct the conformal diagram corresponding to the mass function @xmath108 starting from the last example .
such a case , however , does not correspond really to an evaporating black hole since , according to this mass function , the black hole never disappears , the singularity at @xmath35 is ever present for @xmath109 . in a real process of black hole evaporation due to hawking radiation ,
the black hole loses mass with a rate @xmath110 , disappearing completely in a finite time and giving rise to an empty spacetime .
the mass function [ evap ] a(u ) = \ { cc -mu , & u<0 , + 0 , & u0 , . can be used to simulate the vanishing of a black hole in a finite time . for the ingoing radiation case @xmath107 , the situation is similar to the last example , but with the crucial difference that for @xmath111 the spacetime is empty , and the solutions can reach ( and cross ) @xmath35 .
we notice that the form of @xmath106 is not important . in order to construct the conformal diagram of a evaporating black hole
, one only needs that @xmath112 after some @xmath109 . in this case , the full conformal diagram ( see fig . [ fig7 ] ) is constructed adding to that one obtained from the ingoing collapse case by using the transformation ( [ tr ] ) , a patch of minkowski spacetime , just after the instant @xmath113 corresponding to the vanishing of the black hole .
the resulting conformal diagram represents a black hole that evaporates leaving behind a naked singularity .
we presented a semi - analytical approach to construct the vaidya metric in double - null coordinates for general monotonic mass functions .
we applied it to elucidate the nature of the three distinct qualitative causal structures identified by waugh and lake@xcite for linear mass functions , to construct a solution corresponding to a smooth radial collapse of a null fluid , starting from empty space and finishing with a black - hole , and to construct the conformal diagram of a evaporating black hole .
the approach involves an arbitrary function @xmath41 . however , any choice for which @xmath62 will result in causally equivalent regular ( @xmath114 ) space - times .
the analytical solutions of this problem also involve an arbitrary function , as @xmath31 in ( [ l ] ) , that must be chosen properly in order to get a regular spacetime .
the double - null form of the vaidya metric is suitable to the study of quasi - normal modes of time depending solutions . in double - null coordinates ,
the equations for scalar perturbations are separable and we obtain an effective schrdinger equation with a time depending potential for the perturbations .
this situation has been recently considered in the heuristic analysis of @xcite .
the full problem is now under investigation .
99 h. stephani _
_ , _ exact solutions of einstein s field equations _ , cambridge university press , second edition , 2002 .
k. lake , phys .
lett . * 68 * , 3129 ( 1992 ) .
joshi , _ global aspects in gravitation and cosmology _ , oxford university press , 1993 .
lemos , phys .
lett . * 68 * , 1447 ( 1992 ) ; c. hellaby , phys . rev . *
d49 * , 6484 ( 1994 ) .
d.m . eardley and l. smarr , phys .
* d19 * , 2239 ( 1979 ) .
y. kuroda , progr .
* 72 * , 63 ( 1984 ) .
k. lake and t. zannias , phys . rev . *
d41 * , 3866 ( 1990 ) .
k. lake and t. zannias , phys .
* d43 * , 1798 ( 1991 ) .
hiscock , phys . rev . * d23 * , 2813 ( 1981 ) . y. kuroda , progr . theor . phys . * 71 * , 100 ( 1984 ) . y. kuroda , progr . theor . phys . *
71 * , 1422 ( 1984 ) .
w. biernacki , phys . rev . *
d41 * , 1356 ( 1990 ) .
r. parentani , phys . rev .
* d63 * , 041503 ( 2001 ) .
hu and e. verdaguer , _ stochastic gravity : theory and applications _ , gr - qc/0307032 , living rev .
relativity 7 , 3 ( 2004 ) , http://www.livingreviews.org/lrr-2004-3 r. lindquist , r. schwartz , and c. misner , phys .
rev . * 137 * , 1364 ( 1965 ) .
f. fayos , m.m .
martin - prats , and j.m.m .
senovilla , class .
quantum grav .
* 12 * 2565 ( 1993 ) .
b. waugh and k. lake , phys . rev .
* d34 * , 2978 ( 1986 ) .
j. synge , ann . mat .
* 98 * , 239 ( 1974 ) .
holvorcem , p.s .
letelier , and a. wang , j. math .
phys . * 36 * , 3663 ( 1995 ) .
s.w . hawking and g.f .
ellis , _ the large scale structure of space - time _
, cambridge university press , 1973 .
s. hod , phys . rev . *
d66 * , 024001 ( 2002 ) . | we reexamine here a problem considered in detail before by waugh and lake : the solution of spherically symmetric einstein s equations with a radial flow of unpolarized radiation ( the vaidya metric ) in double - null coordinates .
this problem is known to be not analytically solvable , the only known explicit solutions correspond to the constant mass case ( schwarzschild solution in kruskal - szekeres form ) and the linear and exponential mass functions originally discovered by waugh and lake .
we present here a semi - analytical approach that can be used to discuss some qualitative and quantitative aspects of the vaidya metric in double - null coordinates for generic mass functions .
we present also a new analytical solution corresponding to @xmath0-mass function and discuss some physical examples . |
many physicists consider neutrino oscillations to be the first clear signature for physics beyond the standard model . with the most recent data from sno and super - kamiokande ,
there remains virtually no doubt about the existence of neutrino flavor oscillations .
indeed , the parameter space for neutrino mass differences and mixing angles has already become significantly constrained
. there remains , however , the paramount issue concerning how these experimentally determined parameters can be accommodated within a theoretically motivated model .
while there have been numerous avenues that have been explored within recent years , models with large extra dimensions have received considerable attention ever since it was realized @xcite that extra spacetime dimensions have the potential to provide new , intrinsically geometric perspectives on the hierarchies between various energy scales present within and beyond the standard model ( sm ) .
these include the planck scale @xcite , the gut scale @xcite , and the string scale @xcite , as well as a possible scale for susy - breaking @xcite .
neutrino phenomenology has also been investigated within this higher - dimensional context , with the fundamental idea @xcite being that since the right - handed neutrino is a standard - model singlet , it need not necessarily be bound to the brane to which the other standard - model particles are restricted .
the right - handed neutrino can therefore propagate into the bulk of extra spacetime dimensions and accrue an infinite tower of kaluza - klein ( kk ) excitations , all of which will then take part in neutrino oscillations @xcite .
this idea , along with numerous variations , has spawned a relatively large literature investigating various aspects of higher - dimensional neutrino phenomenology ( see , _ e.g. _ , refs .
@xcite ) .
the challenge for these bulk - neutrino models , however , is to accommodate a suitable flavor structure which can trigger the observed neutrino oscillations . to date
, two different directions have been explored in the literature .
one direction involves generalizing the original idea of refs .
@xcite by introducing a separate right - handed bulk neutrino for each flavor of neutrino on the brane .
this process thereby extends flavor into the bulk .
however , in such cases , the brane / bulk couplings become arbitrary @xmath0 mixing matrices whose parameters are in principle undetermined . moreover , the three bulk neutrinos can in principle correspond to different extra spacetime dimensions with different radii .
thus , one obtains a scenario with many undetermined parameters governing neutrino masses and mixing angles .
indeed , to a large extent , the role of the extra dimensions in such scenarios is reduced to providing a set of kaluza - klein states which function ( from a four - dimensional perspective ) as additional sterile neutrinos .
in contrast , flavor neutrino oscillations continue to be triggered through mixing angles which are ultimately introduced by hand , just as in four dimensions .
this therefore fails to yield any further insight regarding the structure or origins of the neutrino mixings . a second ,
more `` minimal '' idea first advanced in ref .
@xcite is to introduce only one bulk neutrino , and to provide this single bulk neutrino with flavor - universal couplings to all three brane neutrinos .
thus , in so doing , one is essentially considering flavor to be a feature internal to the standard model , a feature which is restricted to the brane and which therefore does not extend into the bulk .
it is even possible to take the brane theory to be flavor - diagonal , thereby avoiding mixing angles completely ; indeed , in the model of ref .
@xcite , the only flavor - sensitive feature that distinguishes these brane neutrinos is their differing bare majorana masses @xmath1 .
nevertheless , one finds @xcite that sizable neutrino flavor oscillations between the different neutrinos on the brane arise as a result of their indirect mixings with the kaluza - klein modes of the bulk right - handed neutrino .
thus , in such a model , the flavor oscillations are entirely `` bulk - mediated '' and no mixing angles are needed at all : it is the presence of the higher - dimensional bulk which is completely responsible for inducing the flavor oscillations on the brane .
although it is possible to consider cases in which the couplings between the brane neutrinos and the bulk neutrino are large , it is convenient ( and perhaps also phenomenologically necessary ) to consider a so - called `` perturbative '' limit in which these couplings are relatively small . in such cases ,
it is possible to integrate out the effects of the kaluza - klein bulk neutrinos in order to obtain an effective mixing matrix involving only the light modes in the theory . typically , these light modes consist of the three brane neutrinos as well as the zero mode(s ) of the bulk neutrino(s ) .
because this effective mixing matrix completely encapsulates the resulting flavor - dependence of the model , it inevitably lies at the center of any comparison between a given theoretical neutrino model and experimental data .
indeed , experimentalists will eventually provide a unique numerical mixing matrix which incorporates the entire observed phenomenology of neutrino oscillations .
it will then be the job of the theorist to solve the `` inverse '' problem of deducing the set of viable underlying neutrino models which can lead to this matrix .
the non - minimal models with many bulk neutrinos contain many mixing angles and can therefore give rise to relatively diverse effective mixing matrices . however , one difficulty with the minimal model of ref .
@xcite is that it leads to a rather rigid form for this effective mixing matrix .
since the minimal model involves only one bulk neutrino , it yields an effective @xmath2 mixing matrix which , as we shall see , has a texture of the form = [ dsmassmatrix ] where the first three rows / columns correspond to the brane neutrinos and the final row / column corresponds to the bulk zero mode . here
@xmath3 and @xmath4 are parameters associated with the bulk of the higher - dimensional theory and its coupling to the brane , while @xmath1 are the bare majorana masses of the neutrinos on the brane . while we see that the bulk physics ( through the term @xmath3 ) is responsible for yielding a non - diagonal @xmath0 brane mixing submatrix , this matrix is relatively rigid , with all off - diagonal entries forced to be exactly equal .
indeed , a mixing texture of this form may have difficulty accommodating observed neutrino oscillations of the correct sizes and magnitudes .
it is the purpose of this paper to propose an alternative `` minimal '' model which leads to a richer effective mixing matrix without introducing additional parameters or sacrificing any of the minimality of the single bulk - neutrino scheme . rather than incorporate a flavor structure through differing bare majorana masses @xmath1 on the brane as in ref .
@xcite , we shall instead incorporate a non - trivial flavor structure within the context of the so - called `` split - fermion '' scenario @xcite .
as is well known , the split - fermion scenario represents an intrinsically higher - dimensional method of explaining flavor hierarchies for the charged matter content of the standard model .
however , as we shall discuss below , this method faces certain unique difficulties when attempting to address the flavor structure of the neutrino sector .
thus , by combining the bulk - neutrino scenario with the split - fermion scenario , we are able to obtain a `` hybrid '' minimal model which has both strong theoretical motivations as well as rich prospects for neutrino phenomenology .
in the split - fermion scenario @xcite , the fermions of the standard model are located within a `` fat '' brane but centered around different positions within the brane .
their spatial extent within the fat brane is modeled by a gaussian distribution with a typical width @xmath5 , where @xmath5 is approximately one order of magnitude smaller than the width of the brane .
the gauge fields and the higgs are assumed to be equally distributed over the width of the brane . to obtain the effective four - dimensional couplings between different particles ,
the extra dimensions have to be integrated out .
since the overlap of the gaussian wavefunctions of the particles can be very small , the resulting effective couplings can be extremely suppressed .
indeed , it is easy to obtain a suppression by thirty orders of magnitude by delocalizing particles at locations which are separated by about ten times their gaussian widths .
this setup yields a solution to several problems .
it can be used to suppress the proton decay , to suppress flavor - changing operators , or to explain the observed mass hierarchy in the lepton sector .
many possible configurations of particles inside the brane are possible ; some of these are discussed in ref .
@xcite .
relative shifts between left - handed standard - model doublets and right - handed singlets have been successfully used to address the mass hierarchy of quarks and charged leptons .
but for the same reason the split - fermion scenario is easy to implement for charged fermions , it leads to difficulties when one tries to incorporate neutrinos @xcite .
experiments suggest with an increasing degree of certainty that neutrino mixing is maximal in the case of @xmath6 and almost maximal in the case of @xmath7 .
the common textures of the zeroth - order neutrino mass matrix that are compatible with such almost - bimaximal mixing can be classified by the hierarchy type into different categories normal , inverted , or degenerate .
however , all of these situations require several entries to be of the same order of magnitude .
unfortunately , in the split - fermion scenario , the coupling between any two particles is extremely sensitive to their relative distance within the brane .
the observed neutrino mixing can therefore be achieved only by carefully choosing the central locations of the neutrinos inside the brane .
in addition , suitable positions for the charged leptons must be found in order to avoid excessive flavor - changing processes .
together , these requirements put severe constraints on the allowed fermion locations @xcite .
while all experimental constraints can ultimately be accommodated , the resulting particle map is extremely sensitive to small perturbations .
thus , even though these models reduce the amount of fine - tuning for the yukawa couplings , a considerable degree of fine - tuning continues to be necessary .
there are , of course , various options for ameliorating this situation .
for example , introducing majorana neutrinos rather than dirac neutrinos has been proposed @xcite as a method of achieving reasonable neutrino masses within the split - fermion scenario .
the neutrino sector of the split - fermion scenario has also recently been examined @xcite within the context of the randall - sundrum model @xcite . in this paper , we will consider a different approach towards the neutrino sector by asking what consequences arise if the two scenarios split fermions and bulk neutrinos are examined in a joined framework . in this way
, we shall be combining the strengths of each individual scenario : the split - fermion mechanism will trigger the flavor structure of the neutrino sector , while the bulk sector will ameliorate the fine - tuning issues and naturally lead to large mixing angles .
indeed , it was already noted in ref .
@xcite that the combination of having brane neutrinos at different locations in the extra dimension would trigger a higher - dimensional seesaw mechanism , although this idea was not pursued and no consistent framework was provided . in this paper , we shall see that split fermions provide a natural context for this phenomenon , leading to flavor - dependent mixing properties which can be used to distinguish between the brane neutrinos .
there are also other benefits to combining these scenarios .
for example , as we have mentioned , one difficulty of the split - fermion scenario applied to neutrinos has been that the mass matrix is exponentially sensitive to neutrino displacement within the brane .
however , in our hybrid scenario , we shall see that the interactions between the brane and bulk naturally convert this dependence from exponential to linear .
thus , the previous exponential sensitivity is entirely eliminated .
another important feature is that our hybrid scenario provides its own natural regulator for kaluza - klein sums .
thus , in our scenario , all such kaluza - klein summations are manifestly finite , even in cases with multiple extra dimensions . as we shall see
, this arises because the heavy kaluza - klein modes with wavelengths that are small compared with the widths of the brane gaussians will average out when folded with the gaussian distribution .
this thereby eliminates the necessity of introducing a cutoff by hand . but most importantly
, we shall find that our mechanism completely decouples the effective neutrino flavor mixing angles from the magnitudes of the neutrino overlaps within the brane .
thus , we are able to obtain large neutrino mixing angles between neutrinos , even when these neutrinos have rather large spatial separations within the brane .
we now describe the framework for our hybrid model .
clearly , parts of this model will be similar to the model considered in ref .
@xcite . for concreteness
, we will work in this section with one bulk neutrino and one compactified extra dimension of radius @xmath8 , although the model can be trivially extended to more bulk fields and to more additional dimensions .
the primary features of this model can be summarized as follows .
we imagine a single extra dimension of radius @xmath8 , as well as a brane of width @xmath9 . * on the brane , we introduce @xmath10 left - handed neutrinos @xmath11 ; these are our flavor eigenstates .
we assume that these left - handed neutrinos have no corresponding majorana masses and are restricted to lie within the brane . * in the bulk , we introduce a single dirac fermion @xmath12 which does not carry any flavor indices and is therefore completely flavor - neutral .
* for the sake of minimality , we assume that the left - handed neutrinos do not mix or couple to each other directly .
thus , all mixing angles on the brane are set to zero .
* we introduce a single flavor - blind brane / bulk coupling @xmath13 between each of the brane neutrinos @xmath11 and the bulk neutrino @xmath12 . * finally ,
each of the active brane neutrinos @xmath11 is assumed to be centered around a different transverse location within the brane .
it is this feature , and only this feature , which is introduces a flavor - dependence into our model .
we shall take our spacetime coordinates as @xmath14 where @xmath15 are the four uncompactified coordinates on the brane and @xmath16 is the coordinate of the fifth dimension perpendicular to the brane . given the above assumptions
, the action for our model then takes the form @xmath17 where @xmath18 describes the physics on the brane , @xmath19 describes the physics in the bulk , and @xmath20 describes the brane / bulk couplings .
these individual contributions take the form _ & = & d^4 x dy _
= 1^n_f ^_(x , y ) i s_b & = & d^4 x dy |(x , y ) i^a _ a ( x , y ) + s_c & = & d^4 x dy _
= 1^n_f m _ * ^_(x , y ) + [ klag ] where @xmath21 represent the five - dimensional dirac matrices . here
the index @xmath22 runs over all @xmath10 flavors on the brane . in the above
, we have chosen to work in the weyl basis in which the dirac fermion @xmath12 can be decomposed into two two - component spinors : @xmath23 .
the bar on @xmath24 indicates that @xmath24 transforms as a different lorentz representation than @xmath25 .
we refer to @xmath25 and @xmath24 as left - handed and right - handed components , respectively .
we have also written @xmath26 where @xmath27 are the pauli matrices . given the action in eq .
( [ klag ] ) , the next step is to compactify the fifth dimension in order to obtain an effective four - dimensional theory . to do this
, we shall make the following assumptions .
first , we shall identify @xmath28 and impose the orbifold relations @xmath29 and @xmath30 , where @xmath16 is the coordinate of the fifth dimension .
thus , @xmath25 and @xmath24 have the kaluza - klein mode expansions _ + ( x , y ) & = & 1 _
+ ^(0)(x ) + & & + 1 _ n
> 0 _ + ^(n)(x ) ( nyr ) + of our orbifold boundary conditions , @xmath31 does not have a zero mode
. we shall also take our brane neutrinos to have wavefunctions of the form _ ( x , y ) = g(y - y _ , ) _ ( x ) [ nudecomp ] where @xmath32 denotes the usual four - dimensional part of the fermion wavefunction on our brane and where @xmath33 denotes a normalized gaussian wavefunction centered around location @xmath34 with width @xmath5 : g(y , ) = ( - ) .
note that for simplicity , we are assuming that each brane neutrino has the same width @xmath35 . inserting the wavefunctions in eqs .
( [ kkdecomp ] ) and ( [ nudecomp ] ) into the action ( [ klag ] ) and integrating over the fifth dimension , we obtain the effective four - dimensional actions : _ & = & d^4 x _
= 1^n_f ^_^ _ _ + s_b & = & d^4 x
^(0 ) i^_|_-^(0 ) + & & + _
n > 0 + & & + _
n > 0 ( nr ) _
+ ^(n ) |_-^(n ) + |_-^(n ) _ + ^(n )
+ s_c & = & d^4 x _ = 1 ^n_f ^_m |_-^(0 ) + & & + _
n > 0 ( m^_n,+ _ + ^(n ) + m^_n,- |_-^(n ) ) .
[ tglag3 ] note that the fields are now functions of @xmath15 only , and @xmath4 and @xmath36 are volume - suppressed @xmath37-dependent brane / bulk couplings resulting from the rescaling of the individual @xmath38 , @xmath39 kaluza - klein modes : m & & m _ * + m_n,+^ & & m ( - ) ( y _ ) + m_n,-^ & & m ( - ) ( y _ ) . [ couplings3 ] as we see , the brane / bulk coupling is suppressed by a volume factor @xmath40 .
moreover , for non - shifted brane neutrinos with @xmath41 , we see that the couplings @xmath42 actually vanish .
thus , it is only the possible displacement of the antisymmetric fermion modes @xmath39 away from @xmath41 which permits them to couple to the bulk zero mode . given the lagrangian in eq .
( [ tglag3 ] ) , we immediately see that the sm flavor - eigenstate neutrinos @xmath11 on the brane will mix with the entire tower of kaluza - klein states of the higher - dimensional @xmath12 field , even though they do not mix directly with each other . defining ^t ( _ , _
+ ^(0 ) , _ + ^(1 ) , |_-^(1 ) , _ + ^(2 ) , |_-^(2 ) , ... ) , [ calndef ] we see that the mass term of the lagrangian ( [ tglag3 ] ) takes the form @xmath43 where the mass matrix @xmath44 is given by = .
[ oldmatrixp ] of course , in this notation , the first row / column represents @xmath10 independent rows / columns , one for each flavor neutrino on the brane .
the fact that this mass matrix is non - diagonal implies that each of the @xmath10 brane neutrinos will undergo oscillations with the bulk kk modes . through these mixings with the bulk kk neutrinos , the brane neutrinos will thereby undergo effective flavor oscillations with each other .
the next step is to determine the eigenvalues and eigenvectors of the mass matrix ( [ oldmatrixp ] ) .
we can begin this task as follows . defining the eigenvector @xmath45
, we see that the eigenvalue equation @xmath46 yields the simultaneous equations : m v^(0 ) + _ n>0 ( m_n,+^v_+^(n ) + m^_n,- v_-^(n ) ) & = & v^ + m _ v^&= & v^(0 ) + _ m^_n,+ v^+ v_-^(n ) & = & v_+^(n ) + _ m^_n,- v^+ v_+^(n ) & = & v_-^(n ) .
[ simulteqs ] combining the last two of these equations yields a partial solution for @xmath47 , v_^(n)= _ v^ ( m^_n , + m^_n , nr ) , [ interstep ] and inserting this into the top line of eq .
( [ simulteqs ] ) along with the definitions in eq .
( [ couplings3 ] ) yields the relation & & -0.30 truein v^ = m v^(0 ) + 2 m^2 _
n>0 + & & -0.3 truein _ v^ ( + ) .
[ 7ev ] thus , we see that the task of solving the remaining equations has now been reformulated as an equivalent @xmath48-dimensional eigenvalue problem of the form ( ) = [ 8ev ] where @xmath49 and where @xmath50 is a @xmath51-dependent `` reduced '' mass matrix of the form ( ) = . [ tildem ] here @xmath52 , where the dimensionless coefficients @xmath53 are given by & & -0.3 truein f _ ( ) = _ n>0 + & & ( r + n ) . [ fdef ] note that for numerical purposes , eq .
( [ 8ev ] ) provides a particularly useful simplification of the original infinite - dimensional eigenvalue problem . solving eq .
( [ 8ev ] ) for @xmath51 and @xmath54 and inserting these results into eq .
( [ interstep ] ) then enables one to obtain the full infinite - dimensional eigenvector @xmath55 .
thus far , we have not made any approximations .
however , in order to proceed analytically , we shall now take the so - called `` perturbative '' limit @xmath56 in which the brane / bulk coupling is small compared with the compactification scale .
introducing this limit is one way of ensuring that the brane sector of our theory ( or , as we shall see , a subset of that sector ) will experience little or no net loss of probability into the `` sterile '' tower of bulk excitations . in this limit ( see , _
e.g. _ , ref .
@xcite ) , the eigenvalues of the excited kk neutrino modes will be approximately @xmath57 , corresponding to the bulk / bulk sector of the mixing matrix , while the remaining @xmath58 eigenvalues will be much smaller than the masses of the excited modes . specifically , we see from the form of the reduced mass matrix in eq .
( [ tildem ] ) that the eigenvalues of these remaining light modes will scale either as @xmath59 or as @xmath60 .
however , the assumption that @xmath61 implies that @xmath62 in either case : if @xmath63 then clearly @xmath64 , while if @xmath65 then r ~ m_^2 r ~ ( m r)^2 1 .
thus in either case we can approximate f _ - _
n>0 1n ( - ) [ interm ] for these remaining light modes . as we shall demonstrate in appendix a
, this result may also be derived through a direct perturbative diagonalization of our original mass matrix .
note that the only case in which it is not justified to pass from eq .
( [ fdef ] ) to eq .
( [ interm ] ) occurs when @xmath66 is exactly equal to @xmath67 , for in this case the second term in eq .
( [ fdef ] ) vanishes exactly for all @xmath37 and is thus always smaller than the first .
however , since @xmath51 is significantly suppressed for the light modes we are considering , we see that the resulting value of @xmath68 in this case is significantly smaller than it is for other values of @xmath69 and @xmath66 . the approximation in eq .
( [ interm ] ) , which sets this value of @xmath68 to zero , therefore continues to be roughly accurate even in this case , indicating a strongly suppressed value for @xmath68 in the region near @xmath70 .
it is relatively straightforward to evaluate the kk sum in eq .
( [ interm ] ) , especially since the presence of the gaussian width for the brane neutrinos serves as a natural regulator for the kk sum which renders it manifestly finite . introducing the variable @xmath71 and noting that @xmath72
, we see that the sum can be approximated as an integral , yielding f _ & & - _ 0^ ( -k^2/ ) k ( y_+y_)/ + & = & - 2 erf .
[ effmassmatrix ] note that the error function is an odd function , @xmath73 , with approximate behavior ( x ) thus , defining the rescaled ( dimensionless ) displacements @xmath74 , we conclude that for the light modes in our theory , our reduced @xmath75-dimensional mass matrix in eq .
( [ tildem ] ) takes the form m [ tildem2 ] with m _
-m_^2 2 [ mresult ] we see from eq .
( [ mresult ] ) that when the neutrino displacements are of the same order as their gaussian widths , the entries of this mass matrix depend _ linearly _ rather than exponentially on the displacements !
this is an important observation , implying that the usual exponential sensitivity to the fermion displacements has been eliminated in our hybrid scenario , with the interaction with the bulk neutrino serving to convert this sensitivity from exponential to linear .
moreover , we also observe the interesting fact that the interactions with the bulk neutrino actually remove _ all _ sensitivity to the neutrino displacements when these displacements become significantly larger than their gaussian widths , with mass - matrix values saturating near the universal value @xmath76 in such cases .
thus , in such cases , all flavor dependence is lost .
we also notice from eq .
( [ mresult ] ) that our effective mixings on the brane are not translationally invariant within the brane , since they depend not on the difference @xmath77 , but rather on the _ sum _ @xmath78 .
indeed , our off - diagonal mixing terms cancel only for @xmath79 rather than @xmath80 .
moreover , since this dependence is on the sum rather than the difference , the net distance from the origin at @xmath34 becomes an important quantity , determining when the error function passes from its linear to constant regime .
once again , however , these features represent the effects of the mixings with the bulk neutrino modes , for it is the kk orbifold relations just above eq .
( [ kkdecomp ] ) which implicitly select the origin @xmath34 as a special point , thereby breaking translational invariance within the brane .
these features imply that it is possible to obtain large effective flavor mixing angles for the active neutrinos on the brane _ even when the overlaps between these neutrinos vanish_. in other words , large neutrino overlaps are no longer needed for large effective mixing angles . to illustrate this behavior more concretely , let us consider a simplified two - flavor example in which we have only one effective mixing angle .
given the result in eq .
( [ mresult ] ) , we may identify this effective mixing angle @xmath81 through the relation 2 = 2 erf(y_1+y_2 ) ( 2 y_2 ) - erf(2 y_1 ) .
[ effangle ] for a given fixed value of @xmath82 ( the position of the first neutrino ) , we can now examine how this effective mixing angle depends on @xmath83 ( the position of the second neutrino ) .
there are several cases to consider .
if @xmath84 ( so that the first neutrino is centered very close to the origin ) , we find that @xmath85 as @xmath86 .
thus , we find that @xmath87 as @xmath86 , which represents nearly maximal mixing .
note that this result applies even though the second neutrino is being pulled relatively far from the first .
indeed , one can verify the mixing remains nearly maximal in this case _
regardless _ of the value of @xmath83 .
by contrast , if @xmath88 , our asymptotic behavior changes .
for concreteness , let us take @xmath82 positive ( otherwise we can change all @xmath89 without altering the results ) .
in this case , erf@xmath90 and we see from eq .
( [ effangle ] ) that @xmath91 as @xmath92 while @xmath85 as @xmath93 .
thus , asymptotically , we have relatively large mixing angles in either case
. however , near @xmath94 , we find that our mixing angle passes through zero and changes sign .
thus , there is a region near @xmath94 for which smaller mixing angles can be realized without exponential fine - tuning . finally , for @xmath95 , the resulting behavior interpolates between the two behaviors described above .
again taking @xmath96 for simplicity , we find that @xmath97 asymptotes to values between @xmath98 and @xmath99 as @xmath93 , and between @xmath99 and @xmath100 as @xmath101 .
however , as in all cases , the effective mixing angle hits zero at @xmath102 . 3.7 truein this behavior is shown in fig .
[ figure ] for @xmath103 . in each case
, we observe that large mixing angles dominate , arising in most regions of the parameter space .
nevertheless , we see that small and intermediate mixing angles are easy to achieve without significant fine - tuning in the region near @xmath104 .
as indicated in the introduction , the flavor mixing in this model is significantly different from that of an earlier model @xcite in which all brane neutrinos are located at @xmath105 with zero width , distinguished only through their different `` bare '' majorana masses @xmath1 on the brane .
in such a model , the brane neutrinos couple to only the even bulk modes @xmath38 , whereupon we see ( following the same analysis as performed here ) that the effective @xmath75 `` reduced '' mixing matrix @xmath106 takes the form shown in eq .
( [ dsmassmatrix ] ) , where x = -m^2 + m^2 r ( r ) .
the important point here is that @xmath3 is completely flavor - independent .
thus all of the off - diagonal entries in the effective mass matrix are identical , leading to a relatively rigid flavor structure .
thus far , we have focused on the case with only a single extra spacetime dimension .
it is therefore worth understanding how the relevant mass scales in the mass matrix ( [ tildem2 ] ) are altered if we consider the analogous situation with @xmath107 extra dimensions . with @xmath108 extra dimensions ,
our three - brane has a `` width '' in @xmath108 directions , so that our brane neutrinos are @xmath108-dimensional gaussians .
tracing our previous steps , this implies that the brane / bulk coupling @xmath4 defined in eq .
( [ couplings3 ] ) will accrue a further volume factor of @xmath40 for each additional spacetime dimension , so that we can generally replace m m ( /r)^(-1)/2 ~m_(/r)^/2 .
[ replace ] on the other hand , it is shown in appendix b that the entries of @xmath109 continue to scale as @xmath110 regardless of the number or radii of extra dimensions .
thus , for general @xmath108 , we obtain an effective @xmath111-dimensional mass matrix whose entries scale as m ~ m _ [ tildemdelta ] where @xmath68 denotes a @xmath108-dependent set of flavor - dependent coefficients of order one .
we shall not pursue a detailed calculation of these coefficients for @xmath107 , and shall instead rely on our previous results for @xmath112 as a general illustration of the potential possibilities for the flavor dependence .
these observations are especially relevant for determining the neutrino mass eigenvalues resulting from our mass matrices .
indeed , one of the unique features of these higher - dimensional scenarios involving right - handed neutrinos in the bulk is that it is the _ same _ mass matrix that determines not only the neutrino flavor mixings but also the neutrino mass eigenvalues .
this is different from the traditional four - dimensional framework in which couplings to heavy right - handed neutrinos determine the neutrino masses ( through the traditional seesaw mechanism ) , whereas mixings amongst the light left - handed neutrinos are primarily responsible for determining the effective flavor mixing angles .
it turns out that the eigenvalue phenomenology is very different depending on whether @xmath112 or @xmath107 . for @xmath112 , we have the mass matrix given in eq .
( [ tildem2 ] ) .
however , in this case the effective brane / bulk coupling @xmath4 is necessarily larger than the effective brane / brane components @xmath113 by a factor @xmath114 .
thus , the mixing with the bulk zero mode is dominant and we must actually invert this ( @xmath58)-dimensional mass matrix in order to examine its mass eigenvalues .
we find that in general , @xmath115 eigenvalues will be of ( the smaller ) size @xmath116 , while the remaining two eigenvalues will be of ( the larger ) size @xmath4 ; likewise , the eigenstates corresponding to @xmath117 will have an extremely small component involving the right - handed bulk neutrino , while the eigenstates corresponding to @xmath63 will have a fairly significantly component involving the right - handed bulk neutrino , with probability @xmath118 .
thus , the @xmath115 lightest neutrino components are easily interpretable as our light flavor neutrino mass eigenstates with significant flavor mixings amongst them provided we choose @xmath10 to _ exceed _ our eventual desired number of light neutrinos by one .
given these results , we see that our scenario for @xmath112 actually requires an _ additional _ left - handed neutrino on the brane which participates in the flavor oscillations but is otherwise exceedingly heavy . in some sense , this neutrino may be regarded as an additional `` sterile '' neutrino , since its flavor index need not correspond to any of the flavors exhibited by the charged fermions of the standard model ; moreover , by choosing an appropriate location for this neutrino , its flavor mixings with the other left - handed neutrinos can be minimized and/or eliminated .
however , the unique prediction of this model ( as opposed to other higher - dimensional models ) is that this sterile neutrino is necessarily _ left - handed _ , living on the brane . of course
, its mass is of the same scale ( @xmath119 ) as that of the right - handed zero - mode neutrino living in the bulk .
note that the remaining kk bulk excitations have masses scaling as @xmath120 , and thus their masses are even heavier by an additional hierarchical factor of @xmath114 . for @xmath107 , by contrast ,
the brane / bulk coupling @xmath4 is suppressed by additional factors of @xmath121 relative to the brane / brane components @xmath113 .
thus , assuming that @xmath122 , we see that the brane / bulk coupling will be extremely _ small _ compared to the brane / brane components , leading to @xmath10 eigenvalues of mass @xmath123 and a single eigenvalue of mass @xmath124 .
although this single mass eigenstate will be much lighter than the @xmath10 active eigenstates , it will be overwhelmingly sterile and thus will not interact with the standard - model particles .
we see , then , that regardless of the number of extra spacetime dimensions , our active neutrino mass eigenstates will have masses which scale as @xmath125 in the current setup .
there are therefore a number of values for @xmath126 and @xmath5 which can yield neutrino masses in a phenomenologically acceptable range . one appealing possibility , for example , is to take @xmath126 ( the fundamental mass scale in our theory ) of order @xmath127 , while our three - branes have widths of order @xmath128 .
by contrast , the only solution for which @xmath126 and @xmath129 are of the same order of magnitude requires that we take @xmath126 in the sub - ev range .
likewise , if we demand @xmath130 , then we must take @xmath131@xmath132 .
however , there are other ways of adjusting the overall mass eigenvalues in this scenario without affecting the flavor mixing angles . for example , if there are two extra dimensions whose coordinates are @xmath16 and @xmath133 , we can easily imagine that the standard - model brane is located at @xmath134 while the right handed neutrino has bulk excitations in the @xmath16-direction but is localized around @xmath135 .
if both localizations have common width @xmath5 , we can retrace our previous steps to find that this location `` mismatch '' in the second extra dimension has the effect of inserting an extra suppression factor - 2 ( z_1-z_2)^2 ^ 2for each entry in the resulting effective mass matrix @xmath44 .
while such an overall rescaling of the mass matrix does not affect the magnitudes of the flavor mixing angles , it directly suppresses the magnitudes of the mass eigenvalues .
thus , in this way , we are free to adjust our mass eigenvalues as needed without disturbing the flavor structure of our model .
other similar scenarios are also possible .
in this paper , we have presented a `` hybrid '' model in which the split - fermion scenario and the bulk - neutrino scenario are joined together .
as we have seen , this joining has succeeded in yielding a rich spectrum of flavor mixing angles without significant fine - tuning , thereby ameliorating some of the difficulties inherent in either approach alone . needless to say ,
we have pursued only the most `` minimal '' approach towards constructing this hybrid model .
further possibilities that we have not exploited include , for example , possible variable widths of the gaussians @xcite , the possibility of several right - handed neutrinos , possible orbifold twists @xcite , or possible additional flavor - dependent yukawa couplings of order one .
we have also avoided primordial flavor mixings on the brane , and we have taken our brane neutrinos to have vanishing majorana masses .
while all of these assumptions can be relaxed , our main purpose in this paper has been to illustrate the range of theoretical possibilities that emerge solely from the joining of the split - fermion and bulk - neutrino models .
likewise , we have not performed a detailed fit of the parameters involved , nor have we tried to embed this neutrino sector of the standard model within a larger split - fermion framework that also involves the quarks and charged leptons .
however , we caution that in order to investigate whether such a hybrid model can accommodate all of the observed neutrino - oscillation properties , it is generally not sufficient merely to focus on reproducing the preferred mass differences and mixing angles often quoted in the neutrino literature . this is because such neutrino mixing parameters are usually quoted within the theoretical context of a simple two- or three - state neutrino mixing . in higher - dimensional neutrino models , by contrast , one usually faces very complex , multi - component neutrino oscillations with a variety of varying oscillation lengths and magnitudes ( see , _
e.g. _ , ref .
@xcite for examples of the complexities involved ) .
indeed , even in the small - coupling limit @xmath61 , one typically has a four - neutrino ( or multi - neutrino ) oscillation , depending on the number of bulk neutrinos which are present .
thus , in order to perform a proper comparison with experiment , one must go back to the original oscillation data , plotted in terms of appropriate baselines , and perform a direct fit between predicted and observed oscillation patterns ( and suitable time - averages thereof ) without regard for pre - existing theoretical prejudices .
this too represents a direction for further investigation .
there are also several important theoretical issues which we have not examined in this paper .
one pressing issue , for example , concerns how such hybrid settings can arise on a fundamental level .
although mechanisms of trapping fermions inside a fat brane have been examined closely , and although various ways to realize a split - fermion scenario seem to be possible , it would be interesting to have a more complete picture in which the required fermion locations can be explained from some deeper principle and/or generated dynamically .
nevertheless , we have seen in this paper that joining the split - fermion and bulk - neutrino scenarios has produced a rich theoretical model which naturally gives rise to large neutrino flavor mixing angles without significant fine - tuning . indeed , it is the coupling to the kk modes of the bulk neutrino which eliminates the usual fine - tuning of the split - fermion scenario , and which produces the relatively large mixing angles that we have observed .
thus , within the context of the split - fermion scenario , we have a natural explanation for the somewhat surprising fact that quark mixing angles are generally small while lepton mixing angles are apparently large : while the lepton sector contains a right - handed neutrino which is a gauge singlet and can therefore propagate in the bulk , all of the fields of the corresponding quark sector carry gauge charges and thus must live within the brane .
thus only the lepton sector is capable of receiving the enhancement to the flavor mixing angles that comes indirectly from the couplings to the kk modes of the right - handed neutrino .
we believe that this observation is critical , and perhaps provides one of the strongest motivations for considering such higher - dimensional models of neutrino phenomenology .
this work is supported in part by the national science foundation under grant phy/0301998 , by the department of energy under grant de - fg02 - 04er-41298 , and by a dfg grant from the german government .
we are happy to thank z. chacko , h .- s .
goh , and s. scherer for discussions .
the neutrino mass matrix @xmath44 shows a typical structure which is common for seesaw scenarios .
we can therefore analyze some of its features by similar means . following standard treatments ,
let us write @xmath44 in the block form = where the @xmath136 block corresponds to the brane neutrinos as well as the zero mode of the bulk neutrino , where the @xmath137 block corresponds to the excited kk modes of the bulk neutrino , and where @xmath138 corresponds to the couplings between these two groups of states .
assuming that these couplings are small ( _ i.e. _ , assuming @xmath61 , so that we are in the `` perturbative '' regime ) , we can then approximately block - diagonalize this matrix by finding a matrix @xmath139 such that u^t m u = . in general , the solution for @xmath139 is given by u = with m_d m_r^-1 , whereupon we find m_l m_l - m_d m_r^-1 m_d^t , m_r m_r . since our original mass matrix is real and symmetric , the eigenvalues are necessarily real .
the next - order corrections have the form @xmath140 , and an inspection shows that these are suppressed by a factor of @xmath141 relative to those considered above .
likewise , the mixing between the excited kk modes and the brane / zero - mode subsector is suppressed by a factor of @xmath142 .
note that in the above , we have chosen to group our blocks in such a way that the zero mode of the bulk neutrino is joined with the brane neutrinos rather than with the bulk - neutrino excited kk states .
this is because the kk zero mode is the only mode from the kk tower which is not heavy , and which therefore fails to decouple . by arranging our blocks in this manner
, we are therefore `` integrating out '' only the excited kk modes but retaining the zero mode in our low - energy reduced mass matrix @xmath143 .
given the parameters in our specific model , a straightforward calculation gives = rn , whereupon we find m_d^t & = & rn + & = & with m _
= -_n=1^ ( m^t_n,+ m^_n,- + m^t_n,- m^_n,+ ) . [ malphabeta ]
we thus find _
l , whereupon inserting the definitions in eq .
( [ couplings3 ] ) yields the result given in eq .
( [ interm ] ) .
we now consider the mass scales involved in our mass matrix @xmath144 .
we are particularly interested in the mass scales associated with the expression @xmath113 in eq .
( [ malphabeta ] ) .
thus , there are three components of this expression that we need to generalize to @xmath108 dimensions : we need to generalize the couplings @xmath36 ; we need to generalize expressions such as @xmath145 ; and we need to generalize the kk sum
. we shall therefore begin by addressing each of these in turn . for excitations of the bulk field in more than one extra spacetime dimension , the kaluza - klein expansion in eq .
( [ kkdecomp ] ) will now involve products of sines and cosines .
this in turn means that the couplings @xmath36 in eq .
( [ couplings3 ] ) will also be products of sines and cosines whose arguments display the positions @xmath146 of our fields .
likewise , the form of the higher - dimensional clifford algebra implies that a dirac spinor in @xmath147 spacetime dimensions will have @xmath148 components , and the inverse of the wave vector @xmath37 will be a linear combination of its components .
however , since our goal is merely to estimate relevant mass scales , we shall ignore these spinorial details and retain a notation whereby we treat quantities such as @xmath145 as a single number .
let us now put these ingredients together .
first , we notice that in the higher - dimensional case with @xmath107 , the couplings from eq .
( [ kkdecomp ] ) will enter into eq .
( [ fdef ] ) only for mixtures of even and odd fields .
this means that the factors that must be summed as part of the @xmath37-summation will consist of at least one sine function and one cosine function . since the contributions to this sum are naturally damped beyond @xmath151 , and since we assume the locations @xmath152 to be of order @xmath5 , we can approximate each occurrence of the sine function by the linear term in its taylor expansion ( and set each cosine factor to @xmath100 ) .
let @xmath153 denote the number of sine functions in the product , where @xmath154 , and likewise let @xmath155 denote those directions for which these sine functions appear .
each of these sine functions will then make a linear contribution to @xmath113 of the form @xmath156 , whereupon we find that eqs .
( [ 8ev ] ) and ( [ fdef ] ) will generalize to yield an expression of the form m _
~ m^2 r _ n ( - ) _ i=1^s ( y_i ) .
note that this @xmath37-summation continues to represent a sum over @xmath108-dimensional vectors @xmath37 .
defining @xmath157 , we see that this sum can be approximated by an integral m _ ~ m^2 r^^-1 _ i ( y_i/ ) d k k^+s-2 ( - k^2/^2 ) .
[ step ] however , the integral is now purely dimensionless and @xmath158 ; likewise , for @xmath159 , we see that each of the product factors are also @xmath158 .
thus the overall mass scale is set by the prefactor in eq .
( [ step ] ) . however , given the result in eq .
( [ replace ] ) , we see that we can write this prefactor in the form @xmath160 , obtaining the same overall mass scale as we found for @xmath112 .
thus , we conclude that the overall mass scale for @xmath113 is independent of both @xmath8 and @xmath108 . indeed , it is easy to verify that this result remains true even when all of the proper numerical and spinorial factors are included in the analysis .
n. arkani - hamed , s. dimopoulos and g. r. dvali , phys .
b * 429 * , 263 ( 1998 ) ; phys . rev .
d * 59 * , 086004 ( 1999 ) ; i. antoniadis , n. arkani - hamed , s. dimopoulos and g. r. dvali , phys .
b * 436 * , 257 ( 1998 ) .
e.g. _ , g. r. dvali and a. y. smirnov , nucl .
b * 563 * , 63 ( 1999 ) ; a. e. faraggi and m. pospelov , phys .
b * 458 * , 237 ( 1999 ) ; a. k. das and o. c. w. kong , phys .
b * 470 * , 149 ( 1999 ) ; r. n. mohapatra , s. nandi and a. perez - lorenzana , phys .
b * 466 * , 115 ( 1999 ) ; a. ioannisian and a. pilaftsis , phys .
d * 62 * , 066001 ( 2000 ) ; r. n. mohapatra and a. perez - lorenzana , nucl .
b * 576 * , 466 ( 2000 ) ; nucl
. phys .
b * 593 * , 451 ( 2001 ) ; phys .
d * 67 * , 075015 ( 2003 ) ; a. ioannisian and j. w. f. valle , phys .
d * 63 * , 073002 ( 2001 ) ; r. barbieri , p. creminelli and a. strumia , nucl .
b * 585 * , 28 ( 2000 ) ; e. ma , m. raidal and u. sarkar , phys .
* 85 * , 3769 ( 2000 ) ; e. ma , g. rajasekaran and u. sarkar , phys . lett .
b * 495 * , 363 ( 2000 ) ; g. c. mclaughlin and j. n. ng , phys .
b * 493 * , 88 ( 2000 ) ; a. lukas , p. ramond , a. romanino and g. g. ross , phys .
b * 495 * , 136 ( 2000 ) ; jhep * 0104 * , 010 ( 2001 ) ; d. o. caldwell , r. n. mohapatra and s. j. yellin , phys .
lett . * 87 * , 041601 ( 2001 ) ; phys .
d * 64 * , 073001 ( 2001 ) ; n. cosme , j. m. frere , y. gouverneur , f. s. ling , d. monderen and v. van elewyck , phys .
d * 63 * , 113018 ( 2001 ) ; k. agashe and g. h. wu , phys .
b * 498 * , 230 ( 2001 ) ; c. s. lam and j. n. ng , phys .
d * 64 * , 113006 ( 2001 ) ; a. de gouvea , g. f. giudice , a. strumia and k. tobe , nucl . phys .
b * 623 * , 395 ( 2002 ) ; j. m. frere , m. v. libanov and s. v. troitsky , jhep * 0111 * , 025 ( 2001 ) ; c. s. lam , phys .
d * 65 * , 053009 ( 2002 ) ; h. davoudiasl , p. langacker and m. perelstein , phys .
d * 65 * , 105015 ( 2002 ) ; q. h. cao , s. gopalakrishna and c. p. yuan , phys .
d * 69 * , 115003 ( 2004 ) ; phys . rev .
d * 70 * , 075020 ( 2004 ) ; j. l. hewett , p. roy and s. roy , phys .
d * 70 * , 051903 ( 2004 ) ; e. dudas , c. grojean and s. k. vempati , arxiv : hep - ph/0511001 ; m. t. eisele and n. haba , arxiv : hep - ph/0603158 .
n. arkani - hamed and m. schmaltz , phys .
d * 61 * , 033005 ( 2000 ) ; n. arkani - hamed , y. grossman and m. schmaltz , phys .
d * 61 * , 115004 ( 2000 ) ; e. a. mirabelli and m. schmaltz , phys .
d * 61 * , 113011 ( 2000 ) .
e.g. _ , g. r. dvali and m. a. shifman , phys .
b * 475 * , 295 ( 2000 ) ; y. grossman , int .
j. mod .
a * 15 * , 2419 ( 2000 ) ; h. georgi , a. k. grant and g. hailu , phys .
d * 63 * , 064027 ( 2001 ) ; t. g. rizzo , phys . rev .
d * 64 * , 015003 ( 2001 ) ; s. nussinov and r. shrock , phys .
b * 526 * , 137 ( 2002 ) ; h. c. cheng , c. t. hill , s. pokorski and j. wang , phys .
d * 64 * , 065007 ( 2001 ) ; n. maru , phys .
b * 522 * , 117 ( 2001 ) ; d. e. kaplan and t. m. p. tait , jhep * 0111 * , 051 ( 2001 ) ; w. f. chang , i. l. ho and j. n. ng , phys .
d * 66 * , 076004 ( 2002 ) ; n. haba and n. maru , phys . rev .
d * 66 * , 055005 ( 2002 ) ; y. grossman and g. perez , phys . rev .
d * 67 * , 015011 ( 2003 ) ; w. f. chang and j. n. ng , jhep * 0212 * , 077 ( 2002 ) ; c. biggio , f. feruglio , i. masina and m. perez - victoria , nucl .
b * 677 * , 451 ( 2004 ) ; b. lillie and j. l. hewett , phys .
d * 68 * , 116002 ( 2003 ) ; k. w. choi , i. w. kim and w. y. song , nucl .
b * 687 * , 101 ( 2004 ) ; b. lillie , jhep * 0312 * , 030 ( 2003 ) ; y. nagatani and g. perez , jhep * 0502 * , 068 ( 2005 ) .
y. grossman , r. harnik , g. perez , m. d. schwartz and z. surujon , phys .
d * 71 * , 056007 ( 2005 ) ; r. harnik , g. perez , m. d. schwartz and y. shirman , jhep * 0503 * , 068 ( 2005 ) . | higher - dimensional models of neutrino physics with one or more right - handed neutrinos in the bulk have attracted considerable attention in recent years . however ,
a critical issue for such models is to find a way of introducing the required flavor dependence needed for generating neutrino oscillations . in this paper
, we point out that a natural `` minimal '' framework that accomplishes this can be constructed by combining the bulk - neutrino hypothesis for right - handed neutrinos with the split - fermion scenario for left - handed neutrinos .
this combination leads to a unique flavor signature for neutrino phenomenology which easily incorporates large flavor mixing angles .
this hybrid scenario also has a number of additional important features .
for example , one previous difficulty of the split - fermion scenario applied to neutrinos has been that the mass matrix is exponentially sensitive to neutrino displacements within the brane .
however , in our hybrid scenario , the interactions between the brane and bulk naturally convert this dependence from exponential to linear .
another important feature is that our hybrid scenario provides its own natural regulator for kaluza - klein sums .
thus , in our scenario , all kaluza - klein summations are manifestly finite , even in cases with multiple extra dimensions .
but most importantly , our mechanism completely decouples the effective neutrino flavor mixing angles from the sizes of the overlaps between the neutrino wavefunctions within the brane .
thus , we are able to obtain large neutrino mixing angles even when these neutrinos have significant spatial separations and their overlaps vanish . .7ex
.7ex
= cmss10 = cmss10 at 7pt |
state - space models ( ssm ) are a very popular class of non - linear and non - gaussian time series models in statistics , econometrics and information engineering ; see for example @xcite , @xcite , @xcite .
an ssm is comprised of a pair of discrete - time stochastic processes , @xmath0 and @xmath1 , where the former is an @xmath2-valued unobserved process and the latter is a @xmath3-valued process which is observed .
the hidden process @xmath4 is a markov process with initial density @xmath5 and markov transition density @xmath6 , i.e.@xmath7 it is assumed that the observations @xmath1 conditioned upon @xmath0 are statistically independent and have marginal density @xmath8 , i.e.@xmath9 we also assume that @xmath10 , @xmath11 and @xmath8 are densities with respect to ( w.r.t . )
suitable dominating measures denoted generically as @xmath12 and @xmath13 . for example , if @xmath14 and @xmath15 then the dominating measures could be the lebesgue measures .
the variable @xmath16 in the densities of these random variables are the particular parameters of the model .
the set of possible values for @xmath16 is denoted @xmath17 .
the model ( [ eq : evol])-([eq : obs ] ) is also referred to as a hidden markov model ( hmm ) in the literature . for any sequence @xmath18 and integers @xmath19 ,
let @xmath20 denote the set @xmath21 .
( when @xmath22 this is to be understood as the empty set . ) equations ( [ eq : evol ] ) and ( [ eq : obs ] ) define the joint density of @xmath23,@xmath24 which yields the marginal likelihood , @xmath25 let @xmath26 , @xmath27 , be a sequence of functions and @xmath28 , @xmath29 , be the corresponding sequence of additive functionals constructed from @xmath30 as follows on @xmath31 , i.e. @xmath32 is the sums of term of the form @xmath33 , is merely a matter of redefining @xmath30 in the computations to follow.]@xmath34 there are many instances where it is necessary to be able to compute the following expectations recursively in time,@xmath35 .
\label{eq : representationadditivefunctional}\ ] ] the conditioning implies the expectation should be computed w.r.t .
the density of @xmath36 given @xmath37 , i.e. @xmath38 and for this reason @xmath39 is referred to as a _ smoothed additive functional_. as the first example of the need to perform such computations , consider the problem of computing the score vector , @xmath40 .
the score is a vector in @xmath41 and its @xmath42 component is@xmath43 _ { i}=\frac{\partial\log\text { } p_{\theta}\left ( y_{0:n}\right ) } { \partial \theta^{i}}. \label{eq : score}\ ] ] using fisher s identity , the problem of computing the score becomes an instance of ( [ eq : representationadditivefunctional ] ) , i.e.@xmath44 + \sum \limits_{k=0}^{n}\mathbb{e}_{\theta}\left [ \left .
\nabla\log g_{\theta } \left ( \left .
y_{k}\right\vert x_{k}\right ) \right\vert y_{0:n}\right ] \nonumber\\ & + \mathbb{e}_{\theta}\left [ \left . \nabla\log\mu_{\theta}\left ( x_{0}\right ) \right\vert y_{0:n}\right ] .
\label{eq : scoreadditivefunctional}\ ] ] an alternative representation of the score as a smoothed additive functional based on infinitesimal perturbation analysis is given in @xcite .
the score has applications to maximum likelihood ( ml ) parameter estimation @xcite , @xcite .
the second example is ml parameter estimation using the expectation - maximization ( em ) algorithm .
let @xmath45 be a batch of data and the aim is to maximise @xmath46 w.r.t .
@xmath16 . given a current estimate @xmath47 , a new estimate @xmath48 is obtained by maximizing the function@xmath49 + \sum \limits_{k=0}^{n}\mathbb{e}_{\theta^{\prime}}\left [ \left .
\log g_{\theta } \left ( \left .
y_{k}\right\vert x_{k}\right ) \right\vert y_{0:n}\right ] \\ & + \mathbb{e}_{\theta^{\prime}}\left [ \left . \log\mu_{\theta}\left ( x_{0}\right ) \right\vert y_{0:n}\right]\end{aligned}\ ] ] w.r.t . @xmath16 and setting @xmath50 to the maximising argument .
a fundamental property of the em algorithm is @xmath51 . for linear gaussian models and finite state - space hmm ,
it is possible to perform the computations in the definition of @xmath52 . for general non - linear non - gaussian state - space models of the form ( [ eq : evol])-([eq : obs ] ) , we need to rely on numerical approximation schemes .
smc methods are a class of algorithms that sequentially approximate the sequence of posterior distributions @xmath53 using a set of @xmath54 weighted random samples called particles . specifically , the smc approximation of @xmath55 , for @xmath56 , is @xmath57 where @xmath58 is the importance weight associated to particle @xmath59 and @xmath60 is the dirac measure with an atom at @xmath61 .
the particles are propagated forward in time using a combination of importance sampling and resampling steps and there are several variants of both these steps ; see @xcite , @xcite for details .
smc methods are parallelisable and flexible , the latter in the sense that smc approximations of the posterior densities for a variety of ssms can be constructed quite easily .
smc methods were popularized by the many successful applications to ssm .
a smc approximation of @xmath39 may be constructed by replacing @xmath62 in eq .
( [ eq : representationadditivefunctional ] ) with its smc approximation in eq .
( [ eq : filteringdistribution ] ) - we call this the _ path space _
method since the smc approximation of @xmath62 , which is a probability distribution on @xmath63 , is used .
fortunately there is no need to store the entire ancestry of each particle , i.e. @xmath64 , which would require a growing memory .
also , this estimate can be computed recursively . however , the reliance on the approximation of the joint distribution @xmath65 is bad .
it is well - known in the smc literature that the approximation of @xmath65 becomes progressively impoverished as @xmath66 increases because of the successive resampling steps @xcite , @xcite , @xcite .
that is , the number of distinct samples representing @xmath67 for any fixed @xmath68 diminishes as @xmath66 increases this is known as the _ particle path degeneracy _ problem . hence ,
whatever being the number of particles , @xmath67 will eventually be approximated by a single unique particle for all ( sufficiently large ) @xmath66 .
this has severe consequences for the smc estimate @xmath39 . in @xcite , under favourable mixing assumptions , the authors established an upper bound on the @xmath69 error which is proportional to @xmath70 . under similar assumptions ,
it was shown in @xcite that the asymptotic variance of this estimate increases at least quadratically with @xmath66 . to reduce the variance ,
alternative methods have been proposed .
the technique proposed in @xcite relies on the fact that for a ssm with good forgetting properties , @xmath71 when the horizon @xmath72 is large enough ; that is observations collected after times @xmath73 bring little additional information about @xmath74 .
( see ( * ? ? ?
* corollary 2.9 ) for exponential error bounds . )
this suggests that a very simple scheme to curb particle degeneracy is to stop updating the smc estimate beyond time @xmath73 .
this algorithm is trivial to implement but the main practical problem is that of determining an appropriate value for @xmath72 such that the two densities in eq .
( [ eq : forgetting ] ) are close enough and particle degeneracy is low .
these are conflicting requirements .
a too small value for the horizon will result in @xmath75 being a poor approximation of @xmath76 but the particle degeneracy will be low .
on the other hand , a larger horizon improves the approximation in eq .
( [ eq : forgetting ] ) but particle degeneracy will creep in .
automating the selection of @xmath72 is difficult .
additionally , for any finite @xmath72 the smc estimate of @xmath39 will suffer from a non vanishing bias even as @xmath77 . in @xcite , for an optimized value of @xmath72 which is dependent on @xmath66 and the typically unknown mixing properties of the model , the smc estimates of @xmath39 based on the approximation in eq .
( [ eq : forgetting ] ) were shown to have an @xmath69 error and bias upper bounded by quantities proportional to @xmath78 and @xmath79 under regularity assumptions .
the computational cost of the smc approximation of @xmath39 computed using either the path space method or the truncated horizon method of @xcite is @xmath80 .
a standard alternative to computing @xmath39 is to use smc approximations of fixed - interval smoothing techniques such as the forward filtering backward smoothing ( ffbs ) algorithm @xcite , @xcite .
theoretical results on the smc approximations of the ffbs algorithm have been recently established in @xcite ; this includes a central limit theorem and exponential deviation inequalities . in particular , under appropriate mixing assumptions ,
the authors have obtained time - uniform deviation inequalities for the smc - ffbs approximations of the marginals @xmath81 ( * ? ? ? * section 5 ) ; see @xcite for alternative proofs and complementary results .
let @xmath82 denote the smc - ffbs estimate of @xmath39 .
in this work it is established that the asymptotic variance of @xmath83 only grows linearly with time @xmath66 ; a fact which was also alluded to in @xcite .
the main advantage of the smc implementation of the ffbs algorithm is that it does not have any tuning parameter other than the number of particles @xmath54 .
however , the improved theoretical properties comes at a computational price ; this algorithm has a computational complexity of @xmath84 compared to @xmath85 for the methods previously discussed .
( it is possible to use fast computational methods to reduce the computational cost to @xmath86 @xcite . )
another restriction is that the smc implementation of the ffbs algorithm does not yield an online algorithm .
the contributions of this article are as follows .
* we propose an original online implementation of the smc - ffbs estimate of @xmath39 .
a particular case of this new algorithm was presented in @xcite , @xcite to compute the score vector ( [ eq : score ] ) . however , because it was catered to estimating the score , the authors failed to realise its full generality . * an upper bound for the _ non - asymptotic _ mean square error of the smc - ffbs estimate @xmath82 of @xmath39 is derived under regularity assumptions .
it follows from this bound that the asymptotic variance of @xmath87 is bounded by a quantity proportional to @xmath66 .
this complements results recently obtained in @xcite , @xcite .
* we demonstrate how the online implementation of the smc - ffbs estimate of @xmath39 can be applied to the problem of recursively estimating the parameters of a ssm from data .
we present original smc implementations of recursive maximum likelihood ( rml ) @xcite , @xcite , @xcite and of the online em algorithm @xcite , @xcite , @xcite , @xcite , @xcite , ( * ? ? ?
* section 3.2 . ) .
these smc implementations do not suffer from the particle path degeneracy problem .
the remainder of this paper is organized as follows . in section [ sec : smoothedadditivefunctionals ] the standard ffbs recursion and its smc implementation is presented .
it is then shown how this recursion and its smc implementation can be implemented exactly with only a forward pass .
a non - asymptotic variance bound is presented in section [ sec : theory ] .
recursive parameter estimation procedures are presented in section [ sec : experiments ] and numerical results are given in section [ sec : simulations ] .
we conclude in section [ sec : discussion ] and the proof of the main theoretical result is given in the appendix .
we first review the standard ffbs recursion and its smc approximation @xcite , @xcite .
this is then followed by a derivation of a forward - only version of the ffbs recursion and its corresponding smc implementation .
the algorithms presented in this section do not depend on any specific smc implementation to approximate @xmath88 .
recall the definition of @xmath39 in eq .
( [ eq : representationadditivefunctional ] ) . the standard ffbs procedure to compute @xmath39 proceeds in two steps . in the first step , which is the forward pass ,
the filtering densities @xmath89 are computed using bayes formula:@xmath90 the second step is the backward pass that computes the following marginal smoothed densities which are needed to evaluate each term in the sum that defines @xmath39:@xmath91 where @xmath92 we compute eq .
( [ eq : standardforwardbackward ] ) commencing at @xmath93 and then , decrementing @xmath94 each time , until @xmath95 .
( integrating eq .
( [ eq : standardforwardbackward ] ) w.r.t .
@xmath96 will yield @xmath97 which is needed for the next computation . ) to compute @xmath39 , @xmath66 backward steps must be executed and then @xmath66 expectations computed .
this must then be repeated at time @xmath98 to incorporate the effect of the new observation @xmath99 on these calculations .
clearly this is not an online procedure for computing @xmath100 .
the smc implementation of the ffbs recursion is straightforward @xcite . in the forward pass ,
we compute and store the smc approximation @xmath101 of @xmath102 for @xmath103 . in the backward pass
, we simply substitute this smc approximation in the place of @xmath102 in eq .
( [ eq : standardforwardbackward ] ) .
let @xmath104 be the smc approximation of @xmath105 , @xmath106 , initialised at @xmath93 by setting @xmath107 . by substituting @xmath108 for @xmath109 in eq .
( [ eq : conditionaldensity ] ) , we obtain @xmath110 this approximation is combined with @xmath111 ( see eq .
( [ eq : standardforwardbackward ] ) ) to obtain @xmath112 marginalising this approximation will give the approximation to @xmath113 , that is @xmath114 , where @xmath115 the smc estimate @xmath82 of @xmath39 is then given by @xmath116 the backward recursion for the weights , given in eq .
( [ eq : backwardweights ] ) , makes this an off - line algorithm for computing @xmath82 . to circumvent the need for the backward pass in the computation of @xmath39 , the following auxiliary function ( on @xmath2 )
is introduced , @xmath117 it is apparent that @xmath118 the following proposition establishes a forward recursion to compute @xmath119 , which is henceforth referred to as the _ forward smoothing recursion_. for sake of completeness , the proof of this proposition is given .
[ propdn ] for @xmath120 , we have@xmath121 p_{\theta}\left ( \left . x_{n-1}\right\vert y_{0:n-1},x_{n}\right ) dx_{n-1 } , \label{eq : recursionadditivefunctional}\ ] ] where @xmath122 . * proof . *
the proof is straightforward @xmath123 \text { } p_{\theta}\left ( \left .
x_{0:n-1}\right\vert y_{0:n-1},x_{n}\right ) dx_{0:n-1}\\ & = \int\left [ \int s_{n-1}\left ( x_{0:n-1}\right ) p_{\theta}\left ( \left .
x_{0:n-2}\right\vert y_{0:n-2},x_{n-1}\right ) dx_{0:n-2}\right ] p_{\theta}\left ( \left .
x_{n-1}\right\vert y_{0:n-1},x_{n}\right ) dx_{n-1}\\ & + \int s_{n}\left ( x_{n-1},x_{n}\right ) \text { } p_{\theta}\left ( \left .
x_{n-1}\right\vert y_{0:n-1},x_{n}\right ) dx_{n-1}.\end{aligned}\ ] ] the integrand in the first equality is @xmath124 while the integrand in the first integral of the second equality is @xmath125 .
@xmath126 this recursion is not new and is actually a special instance of dynamic programming for markov processes ; see for example @xcite . for a fully observed markov process with transition density @xmath127 , the dynamic programming recursion to compute the expectation of @xmath128 with respect to the law of the markov process
is usually implemented using a backward recursion going from time @xmath66 to time @xmath129 . in the partially observed scenario considered here ,
@xmath130 conditional on @xmath45 is a backward markov process with non - homogeneous transition densities @xmath131 .
thus ( [ eq : recursionadditivefunctional ] ) is the corresponding dynamic programming recursion to compute @xmath39 with respect to @xmath132 for this backward markov chain .
this recursion is the foundation of the online em algorithm and is described at length in @xcite ( pioneered in @xcite ) where the density @xmath133 appearing in @xmath134 is usually written as @xmath135 or as in eq .
( [ eq : conditionaldensity ] ) in @xcite , @xcite , @xcite .
the forward smoothing recursion has been rediscovered independently several times ; see @xcite , @xcite for example .
a simple smc scheme to approximate @xmath39 can be devised by exploiting equations ( [ eq : additivesmoothfunctionalsasfunctionoft ] ) and ( [ eq : recursionadditivefunctional ] ) .
this is summarised as algorithm smc - fs below .
* algorithm smc - fs : forward - only smc computation of the ffbs estimate * @xmath136 @xmath137@xmath138@xmath139@xmath140@xmath141 @xmath136 @xmath66@xmath142@xmath143@xmath144 } { \sum_{j=1}^{n}w_{n-1}^{\left ( j\right ) } f_{\theta}\left ( x_{n}^{(i)}|x_{n-1}^{(j)}\right ) } , \quad1\leq i\leq n,\label{eq : tapproximation}\\ \widehat{\mathcal{s}}_{n}^{\theta } & = \sum_{i=1}^{n}w_{n}^{\left ( i\right ) } \text { } \widehat{t}_{n}^{\theta}\left ( x_{n}^{\left ( i\right ) } \right ) .
\label{eq : smcapproxadditivefunctionals}\ ] ] this algorithm is initialized by setting @xmath145 for @xmath146 it has a computational complexity of @xmath84 which can be reduced by using fast computational methods @xcite .
the rationale for this algorithm is as follows . by using @xmath147
defined in eq .
( [ eq : mcconditionaldensity ] ) in place of @xmath148 in eq .
( [ eq : recursionadditivefunctional ] ) , we obtain an approximation @xmath149 of @xmath134 which is computed at the particle locations @xmath150 . the approximation of @xmath39 in eq .
( [ eq : smcapproxadditivefunctionals ] ) now follows from eq .
( [ eq : additivesmoothfunctionalsasfunctionoft ] ) by using @xmath151 in place of @xmath152 .
it is valid to use the same notation for the estimates in eq .
( [ eq : batchsmcestimate ] ) and in eq .
( [ eq : smcapproxadditivefunctionals ] ) as they are indeed the same . the verification of this assertion may be accomplished by unfolding the recursion in eq .
( [ eq : tapproximation ] ) .
in this section , we present a bound on the non - asymptotic mean square error of the estimate @xmath82 of @xmath153 . for sake of simplicity , the result is established for additive functionals of the type @xmath154 where @xmath155 , and when algorithm smc - fs is implemented using the bootstrap particle filter ; see @xcite , @xcite for a definition of this vanilla particle filter .
the result can be generalised to accommodate an auxiliary implementation of the particle filter @xcite , @xcite , @xcite . likewise , the conclusion is also valid for additive functionals of the type in ( [ eq : additivefunctional ] ) ; the proof uses similar arguments but is more complicated
. the following regularity condition will be assumed . * ( a ) * there exist constants @xmath156 such that for all @xmath157 , @xmath158 and @xmath159 , @xmath160 admittedly , this assumption is restrictive and typically holds when @xmath2 and @xmath3 are finite or are compact spaces . in general , quantifying the errors of smc approximations under weaker assumptions is possible @xcite .
( more precise but complicated error bounds for the particle estimate of @xmath39 are also presented in @xcite under weaker assumptions .
) however , when ( a ) holds , the bounds can be greatly simplified to the extent that they can usually be expressed as linear or quadratic functions of the time horizon @xmath66 .
these simple rates of growth are meaningful as they have also been observed in numerical studies even in scenarios where assumption a is not satisfied @xcite . for a function @xmath161 ,
let @xmath162 .
the oscillation of @xmath163 , denoted @xmath164 , is defined to be @xmath165 .
the main result in this section is the following non - asymptotic bound for the mean square error of the estimate @xmath82 of @xmath39 given in eq .
( [ eq : smcapproxadditivefunctionals ] ) .
[ nonasymptheo ] assume ( a ) .
consider the additive functional @xmath32 in ( [ eq : additivefunctionalsimple ] ) with @xmath166 and @xmath167 for @xmath168 .
then , for any @xmath56 and @xmath159 , @xmath169 where @xmath170 is a finite constant that is independent of time @xmath66 , @xmath16 and the particular choice of functions @xmath171 . the proof is given in the appendix .
it follows that the asymptotic variance of@xmath87 , i.e. as the number of particles @xmath54 goes to infinity , is upper bounded by a quantity proportional to @xmath172 as the bias of the estimate is @xmath173 ( * ? ? ?
* corollary 5.3 ) .
let @xmath174 denote the smc estimate of @xmath39 obtained with the standard path space method .
this estimate can have a much larger asymptotic variance as is illustrated with the following very simple model .
let @xmath175 , i.e. @xmath176 is an i.i.d .
sequence , and let @xmath177 and @xmath178 for all @xmath179 where @xmath163 is some real valued function on @xmath2 , and @xmath180 .
it can be easily established that the formula for the asymptotic variance of @xmath181 given in @xcite , ( * ? ? ?
( 9.13 ) , page 304 ) simplifies to @xmath182 ^{2}}{\mu_{\theta}\left ( x\right ) } dx+\frac{n\left ( n-1\right ) } { 2}\int\frac{\pi_{\theta}\left ( \left .
x\right\vert y\right ) ^{2}}{\mu_{\theta}\left ( x\right ) } dx\text { } \int\widetilde{s}_{\theta}\left ( x\right ) ^{2}\pi_{\theta}\left ( \left .
x\right\vert y\right ) dx \label{eq : asymptoticvariancepathbased}\ ] ] where @xmath183 thus the asymptotic variance increases quadratically with time @xmath66 .
note though that the asymptotic variance of @xmath184 converges as @xmath66 tends to infinity to a positive constant .
thus path space method can provide stable estimates of @xmath185 $ ] , i.e. when the additive functionals are time - averaged .
let @xmath186 where @xmath187 is a positive non - increasing sequence that satisfies the following constraints : @xmath188 and @xmath189 .
when @xmath190 then @xmath191 .
one important choice for recursive parameter estimation ( see section [ sec : experiments ] ) is @xmath192 it is also of interest to quantify the stability of the path space method when applied to estimate @xmath193 $ ] in this more general time - averaging setting .
once again let @xmath194 denote the smc estimate of @xmath195 $ ] obtained with the standard path space method .
using the formula for the asymptotic variance of @xmath196 given in @xcite , ( * ? ? ?
( 9.13 ) , page 304 ) it can be verified that this asymptotic variance is @xmath197 ^{2}}{\mu_{\theta}\left ( x\right ) } dx\sum\limits_{k=1}^{n}\gamma_{k}^{2}\prod\limits_{i = k+1}^{n}(1-\gamma_{i})^{2}\\ & + \int\frac{\pi_{\theta}\left ( \left .
x\right\vert y\right ) ^{2}}{\mu_{\theta}\left ( x\right ) } dx\text { } \int\widetilde{s}_{\theta}\left ( x\right ) ^{2}\pi_{\theta}\left ( \left .
x\right\vert y\right ) dx\sum\limits_{k=2}^{n}\sum\limits_{i=1}^{k-1}\gamma_{i}^{2}(1-\gamma _ { i+1})^{2}\cdots(1-\gamma_{n})^{2}\ ] ] it follows from lemma [ lem : stepdiscounting ] in appendix that any accumulation point of this sequence ( in @xmath66 ) has to be positive .
in contrast , the asymptotic variance of @xmath198 , i.e. when @xmath199 $ ] is computed using algorithm smc - fs , will converge to zero as @xmath66 tends to infinity .
an important application of the forward smoothing recursion is to parameter estimation for non - linear non - gaussian ssms .
we will assume that observations are generated from an unknown ` true ' model with parameter value @xmath200 , i.e. @xmath201 . the static parameter estimation problem has generated a lot of interest over the past decade and many smc techniques have been proposed to solve it ; see @xcite for a recent review . in a bayesian approach to the problem
, a prior distribution is assigned to @xmath16 and the sequence of posterior densities @xmath202 is estimated recursively using smc algorithms combined with markov chain monte carlo ( mcmc ) steps @xcite , @xcite , @xcite .
unfortunately these methods suffer from the particle path degeneracy problem and will result in unreliable estimates of the model parameters ; see @xcite , @xcite for a discussion of this issue . given a fixed observation record @xmath45 , an alternative offline mcmc approach to estimate @xmath203 has been recently proposed which relies on proposals built using the smc approximation of @xmath204 @xcite . in a ml approach ,
the estimate of @xmath205 is the maximising argument of the likelihood of the observed data .
the ml estimate can be calculated using a gradient ascent algorithm either offline for a fixed batch of data or online @xcite ; see section [ subsec : rml ] .
likewise , the em algorithm can also be implemented offline or online .
the online em algorithm , assuming all calculations can be performed exactly , is presented in @xcite , @xcite , @xcite , @xcite and @xcite . for a general ssm for which the quantities required by the online em can not be calculated exactly , an smc implementation is possible @xcite , ( * ? ? ?
* section 3.2 . ) ; see section [ sec : parameterestimation ] . to maximise the likelihood @xmath206 w.r.t .
@xmath16 , we can use a simple gradient algorithm .
let @xmath207 be the sequence of parameter estimates of the gradient algorithm .
we update the parameter at iteration @xmath208 using@xmath209 where @xmath210 is the score vector computed at @xmath211 and @xmath212 is a sequence of positive non - increasing step - sizes defined in ( [ eq : stepchoice ] ) . for a general ssm
, we need to approximate @xmath213 .
as mentioned in the introduction , the score vector admits several smoothed additive functional representations ; see eq .
( [ eq : score ] ) and @xcite .
( [ eq : score ] ) , it is possible to approximate the score with algorithm smc - fs . in the online implementation , the parameter estimate at time @xmath98 is updated according to @xcite , @xcite @xmath214 upon receiving @xmath215 , @xmath216 is updated in the direction of ascent of the predictive density of this new observation .
a necessary requirement for an online implementation is that the previous values of the model parameter estimates ( other than @xmath216 ) are also used in the evaluation of @xmath217 at @xmath218 .
this is indicated in the notation @xmath219 .
( not doing so would require browsing through the entire history of observations . )
this approach was suggested by @xcite for the finite state - space case and is named rml .
the asymptotic properties of this algorithm ( i.e. the behavior of @xmath216 in the limit as @xmath66 goes to infinity ) have been studied in the case of an i.i.d . hidden process by @xcite and for an hmm with a finite state - space by @xcite . under suitable regularity assumptions ,
convergence to @xmath205 and a central limit theorem for the estimation error has been established . for a general ssm
, we can compute a smc estimate of @xmath220 using algorithm smc - fs upon noting that @xmath219 is equal to @xmath221 in particular , at time @xmath66 , a particle approximation @xmath222 of @xmath223 is computed using the particle approximation at time @xmath137 and parameter value @xmath218 . similarly , the computation of eq .
( [ eq : tapproximation ] ) is performed using @xmath218 and with @xmath224 the estimate of @xmath219 is now the difference of the estimate in eq .
( [ eq : smcapproxadditivefunctionals ] ) with the same estimate computed at time @xmath137 . under the regularity assumptions given in section [ sec : theory ] , it follows from the results in the appendix that the asymptotic variance ( i.e. as @xmath77 ) of the smc estimate of @xmath219 computed using algorithm smc - fs is uniformly ( in time ) bounded . on the contrary , the standard path - based smc estimate of @xmath220 has an asymptotic variance that increases linearly with @xmath66 .
gradient ascent algorithms are more generally applicable than the em algorithm . however , their main drawback in practice is that it is difficult to properly scale the components of the computed gradient vector . for this reason
the em algorithm is usually favoured by practitioners whenever it is applicable .
let @xmath207 be the sequence of parameter estimates of the em algorithm . in the offline approach , at iteration @xmath208 ,
the function@xmath225 is computed and then maximized .
the maximizing argument is the new estimate @xmath226 .
if @xmath227 belongs to the exponential family , then the maximization step is usually straightforward .
we now give an example of this .
let @xmath228 , @xmath229 , be a collection of functions with corresponding additive functionals@xmath230 and let @xmath231 the collection @xmath232 is also referred to as the _ summary statistics _ in the literature .
typically , the maximising argument of @xmath233 can be characterised explicitly through a suitable function @xmath234 , i.e. @xmath235 where @xmath236_{l}=\mathcal{s}_{l , n}^{\theta}$ ] .
as an example of this , consider the following stochastic volatility model @xcite .
[ ex : stochvol]the stochastic volatility model is a ssm defined by the following equations : @xmath237 where @xmath238 and @xmath239 are independent and identically distributed standard normal noise sequences , which are also independent of each other and of the initial state @xmath240 .
the model parameters @xmath241 are to be estimated .
to apply the em algorithm to this model , let@xmath242 for large @xmath66 , we can safely ignore the terms associated to the initial density @xmath10 and the solution to the maximisation step is characterised by the function @xmath243 the smc implementation of the forward smoothing recursion has advantages even for the batch em algorithm . as there is no backward pass ,
there is no need to store the particle approximations of @xmath244 , which can result in a significant memory saving for large data sets . in the online implementation ,
running averages of the sufficient statistics are computed instead @xcite , @xcite , @xcite , @xcite , ( * ? ? ?
* section 3.2 . ) , @xcite .
let @xmath245 be the sequence of parameter estimates of the online em algorithm computed sequentially based on @xmath45 .
when @xmath99 is received , for each @xmath229 , compute @xmath246{l}$\mathcal{s}_{l , n+1}=\gamma_{n+1}\text { } \int s^{l}\left ( x_{n},x_{n+1},y_{n+1}\right ) p_{\theta_{0:n}}(x_{n},x_{n+1}|y_{0:n+1})dx_{n : n+1}$\\ $ \ \ \ \ \ \ \ \ \ \ \ + \left ( 1-\gamma_{n+1}\right ) \int\sum_{k=1}^{n}\left ( \prod\limits_{i = k+1}^{n}\left ( 1-\gamma_{i}\right ) \right ) \gamma_{k}s^{l}\left ( x_{k-1},x_{k},y_{k}\right ) p_{\theta_{0:n}}(x_{0:n}|y_{0:n+1})dx_{0:n},$\end{tabular } \ \ \ \ \ \ \label{eq : suffstatonline}\ ] ] and then set@xmath247 where @xmath248_{l}=\mathcal{s}_{l , n+1}$ ] . here
@xmath249 is a step - size sequence satisfying the same conditions stipulated for the rml in section [ subsec : rml ] .
( the recursive implementation of @xmath250 is standard @xcite . )
the subscript @xmath251on @xmath252 indicates that the posterior density is being computed sequentially using the parameter @xmath253 at time @xmath94 ( and @xmath254 at time @xmath129 . )
references @xcite , @xcite , @xcite and @xcite have proposed an online em algorithm , implemented as above , for finite state hmms . in the finite state setting all computations involved can be done exactly in contrast to general ssms where numerical procedures are called for .
it is also possible to do all the calculations exactly for linear gaussian models @xcite .
define the vector valued function @xmath255 as follows : @xmath256^{\text{t}}$ ] .
computing @xmath257 sequentially using smc - fs is straightforward and detailed in the following algorithm .
* smc - fs implementation of online em * @xmath258@xmath254 @xmath258@xmath259@xmath260 @xmath258@xmath261@xmath262 @xmath258@xmath263@xmath264 .
@xmath258@xmath260 , @xmath265 } { \sum_{j=1}^{n}w_{n-1}^{\left ( j\right ) } f_{\theta_{n-1}}\left ( x_{n}^{(i)}|x_{n-1}^{(j)}\right ) } .\ ] ] @xmath258@xmath266 @xmath267 it was suggested in ( * ? ? ?
* section 3.2 . ) that the two other smc methods discussed in section [ sec : litreview ] could be used to approximate @xmath257 ; the path space approach to implement the online em was also independently proposed in @xcite .
doing so would yield a cheaper alternative to algorithm smc - em above with computational cost @xmath80 , but not without its drawbacks .
the fixed - lag approximation of @xcite would introduce a bias which might be difficult to control and the path space approach suffers from the usual particle path degeneracy problem .
consider the step - size sequence in ( [ eq : stepchoice ] ) .
if the path space method is used to estimate @xmath257 then the theory in section [ sec : theory ] tells us that , even under strong mixing assumptions , the asymptotic variance of the estimate of @xmath257 will not converge to zero for @xmath268 .
thus it will not yield a theoretically convergent algorithm .
numerical experiments in @xcite appear to provide stable results which we attribute to the fact that this variance might be very small in the scenarios considered is assigned a prior distribution and we estimate @xmath269 @xcite , @xcite , @xcite , the path degeneracy problem has much more severe consequences than in the ml framework considered here as illustrated in @xcite .
indeed in the ml framework , the filter @xmath270 will have , under regularity assumptions , exponential forgetting properties for any @xmath159 whereas this will never be the case for @xmath271 .
in contrast , the asymptotic variance of the @xmath84 estimate converges to zero in time @xmath66 for the entire range @xmath272 under the same mixing conditions . the original @xmath84 implementation proposed here
has been recently successfully adopted in @xcite to solve a complex parameter estimation problem arising in robotics .
we commence with a study of a scalar linear gaussian ssm for which we may calculate smoothed functionals analytically .
we use these exact values as benchmarks for the smc approximations .
the model is @xmath273 where @xmath274 and @xmath275 .
we compared the exact values of the following smoothed functionals@xmath276 , \quad\mathcal{s}_{2,n}^{\theta } = \mathbb{e}_{\theta}\left [ \left .
\sum_{k=1}^{n}x_{k-1}\right\vert y_{0:n}\right ] , \quad\mathcal{s}_{3,n}^{\theta}=\mathbb{e}_{\theta}\left [ \left
. \sum_{k=1}^{n}x_{k-1}x_{k}\right\vert y_{0:n}\right ] , \label{eq : benchmarkfunctionals}\ ] ] computed at @xmath277 with the bootstrap filter implementation of algorithm smc - fs and the path space method .
comparisons were made after 2500 , 5000 , 7500 and 10,000 observations to monitor the increase in variance and the experiment was replicated 50 times to generate the box - plots in figure [ combinednandn2boxplots ] .
( all replications used the same data record . )
both estimates were computed using @xmath278 particles . ) for a linear gaussian ssm .
estimates were computed with path space method ( left column ) and algorithm smc - fs ( right column ) .
the long horizontal line intersecting the box indicates the true value.,scaledwidth=100.0% ] from figure [ combinednandn2boxplots ] it is evident that the smc estimates of algorithm smc - fs significantly outperforms the corresponding smc estimates of the path space method .
however one should bear in mind that the former algorithm has @xmath279 computational complexity while the latter is @xmath280 .
thus a comparison that takes this difference into consideration is important . from theorem [ nonasymptheo ] and the discussion after it
, we expect the variance of algorithm smc - fs s estimate to grow only linearly with the time index compared to a quadratic in time growth of variance for the path space method .
hence , for the same computational effort we argue that , for large observation records , the estimate of algorithm smc - fs is always going to outperform the path space estimates .
specifically , for a large enough @xmath66 , the variance of algorithm smc - fs s estimate with @xmath54 particles will be significantly less than the variance of the path space estimate with @xmath281 particles .
if the number of observations is small then , taking into account the computational complexity , it might be better to use the path space estimate as the variance benefit of using algorithm smc - fs may not be appreciable to justify the increased computational load .
figure [ fig : constantstep ] shows the parameter estimates obtained using the smc implementation of online em for the stochastic volatility model discussed in example [ ex : stochvol ] .
the true value of the parameters were @xmath282 and 500 particles were used .
smc - em was started at the initial guess @xmath283 .
for the first 100 observations , only the e - step was executed .
that is the step @xmath284 , which is the m - step was skipped .
smc - em was run in its entirety for observations 101 and onwards .
the step size used was @xmath285 for @xmath286 and @xmath287 for @xmath288 .
figure [ fig : constantstep ] shows the sequence of parameter estimates computed with a very long observation sequence . .
true and converged values ( average of the last 1000 iterations ) are indicated on the left and the right of the plot respectively.,scaledwidth=60.0% ]
we proposed a new smc algorithm to compute the expectation of additive functionals recursively in time . essentially , it is an online implementation of the ffbs smc algorithm proposed in @xcite .
this algorithm has an @xmath279 computational complexity where @xmath54 is the number of particles .
it was mentioned how a standard path space smc estimator to compute the same expectations recursively in time could be developed .
this would have an @xmath280 computational complexity .
however , as conjectured in @xcite , it was shown here that the asymptotic variance of the smc - ffbs estimator increased linearly with time whereas that of the @xmath280 method increased quadratically .
the online smc - ffbs estimator was then used to perform recursive parameter estimation .
while the convergence of rml and online em have been established when they can be implemented exactly , the convergence of the smc implementation of these algorithms have yet to be established and is currently under investigation .
the authors would like to thank olivier capp , thomas flury , sinan yildirim and ric moulines for comments and references that helped improve the first version of this paper .
the authors are also grateful to rong chen for pointing out the link between the forward smoothing recursion and dynamic programming
. finally , we are thankful to robert elliott to have pointed out to us references @xcite , @xcite and @xcite .
the proofs in this section hold for any fixed @xmath16 and therefore @xmath289 is omitted from the notation .
this section commences with some essential definitions .
consider the measurable space @xmath290 .
let @xmath291 denote the set of all finite signed measures and @xmath292 the set of all probability measures on @xmath293 .
let @xmath294 denote the banach space of all bounded and measurable functions @xmath295 equipped with the uniform norm @xmath296 .
let @xmath297 , i.e. @xmath298 is the lebesgue integral of the function @xmath299 w.r.t .
the measure @xmath300 . if @xmath301 is a density w.r.t .
some dominating measure @xmath12 on @xmath293 then , @xmath302 . we recall that a bounded integral kernel @xmath303 from a measurable space @xmath290 into an auxiliary measurable space @xmath304 is an operator @xmath305 from @xmath306 into @xmath294 such that the functions @xmath307 are @xmath308-measurable and bounded , for any @xmath309 . in the above displayed formulae
, @xmath310 stands for an infinitesimal neighborhood of a point @xmath311 in @xmath312 .
let @xmath313 denote the dobrushin coefficient of @xmath314 which defined by the following formula @xmath315 where @xmath316 stands the set of @xmath317-measurable functions @xmath295 with oscillation less than or equal to 1 .
the kernel @xmath314 also generates a dual operator @xmath318 from @xmath291 into @xmath319 defined by @xmath320 .
a markov kernel is a positive and bounded integral operator @xmath314 with @xmath321 . given a pair of bounded integral operators @xmath322 , we let @xmath323 the composition operator defined by @xmath324 . for time
homogenous state spaces , we denote by @xmath325 the @xmath326-th composition of a given bounded integral operator @xmath314 , with @xmath327 . given a positive function @xmath328 on @xmath293 ,
let @xmath329 be the bayes transformation defined by @xmath330 the definitions above also apply if @xmath301 is a density and @xmath314 is a transition density .
in this case all instances of @xmath331 should be replaced with @xmath332 and @xmath303 by @xmath333 where @xmath12 and @xmath310 are the dominating measures .
the proofs below will apply to any fixed sequence of observation @xmath334 and it is convenient to introduce the following transition kernels , @xmath335 with the convention that @xmath336 , the identity operator .
note that @xmath337 .
let the mapping @xmath338 , @xmath179 , be defined as follows @xmath339 several probability densities and their smc approximations are introduced to simplify the exposition .
the _ predicted filter _ is denoted by @xmath340 with the understanding that @xmath341 is the initial distribution of @xmath240 .
let @xmath342 denote its smc approximation with @xmath54 particles .
( this notation for the smc approximation is opted for , instead of the usual @xmath343 , to make the number of particles explicit . )
the bounded integral operator @xmath344 from @xmath2 into @xmath63 is defined as @xmath345 @xmath344 is defined for any pair of time indices @xmath346 satisfying @xmath347 with the convention that @xmath348 for @xmath349 and @xmath350 . the smc approximation , @xmath351 , is@xmath352 where @xmath353 is the smc approximation of @xmath354 obtained from the smc - ffbs approximation of section [ sec : ffbs ] , i.e.@xmath355 where the backward markov transition kernels @xmath356 are defined through @xmath357 it is easily established that the smc - ffbs approximation of @xmath358 , @xmath106 , is precisely the marginal of @xmath359 where @xmath360 was defined in ( [ eq : filteringdistribution ] ) . finally , we define @xmath361 the following estimates are a straightforward consequence of assumption ( a ) . for time indices
@xmath362 , @xmath363 and for @xmath364 , @xmath365 [ lem : kintchine ] let @xmath368 , @xmath56 , be the natural filtration associated with the @xmath54-particle approximation model and @xmath369 be the trivial sigma field . for any @xmath370 , there exist a finite ( non random ) constant @xmath371 such that the following inequality holds for all @xmath372 and @xmath373measurable functions @xmath374 s.t . @xmath375,@xmath376
[ lem : lperrorfilter]for any @xmath370 , there exists a constant @xmath371 such that the following inequality holds for all @xmath372 and @xmath378 s.t .
@xmath379 , @xmath380(\varphi ) \right\vert ^{r}\right ) ^{\frac{1}{r}}\leq a_{r}~\sum_{k=0}^{n}~b_{k , n}~\beta\left ( \frac{q_{k , n}}{q_{k , n}(1)}\right ) .
\label{eq : lperrorfilter}\ ] ] the following decomposition is central @xmath385with the convention that @xmath386 , for @xmath349 . lemma [ lem : dpn ] states that @xmath387and therefore the decomposition can be also written as @xmath388with the convention @xmath389 , for @xmath349 . let @xmath390then every term in the r.h.s .
of ( [ eq : decompsn ] ) takes the following form @xmath391where the integral operators @xmath392 are defined as follows , @xmath393finally , using ( [ eq : decompsn ] ) and ( [ eq : decompsnterm ] ) , @xmath394 is expressed as @xmath395where the first order term is @xmath396and the second order remainder term is @xmath397the non - asymptotic variance bound is based on the triangle inequality @xmath398and bounds are derived below for the individual expressions on the right - hand side of this equation.using the fact that @xmath399 is zero mean and uncorrelated,@xmath400the following results are needed to bound the right - hand side of ( eq : fisrtordererror ) .
first , observe that @xmath401 , and @xmath402 .
now using the decomposition , @xmath403 ~\psi _ { q_{k , n}(1)}(\phi _ { k}(\eta _ { k-1}^{n}))(dx_{k}^{\prime } ) , \end{array}\]]it follows that @xmath404for linear functionals of the form ( [ eq : additivefunctionalsimple ] ) , it is easily checked that @xmath405 ( s_{q})+\sum_{k < q\leq n}q_{k , q}(s_{q}~q_{q , n}(1))\]]with the convention @xmath406 , the identity operator , for @xmath407 . recalling that @xmath401 , we conclude that @xmath408 ( s_{q})+\sum_{k <
q\leq n}\frac{q_{k , q}(q_{q , n}(1)~s_{q})}{q_{k , q}(q_{q , n}(1))}\]]and therefore @xmath409 ( s_{q})+\sum_{k\leq q\leq n}\frac{q_{k , q}(q_{q , n}(1)~s_{q})}{q_{k , q}(q_{q , n}(1))}\]]thus , @xmath410using the estimates in ( [ eq : contractionest ] ) and ( eq : contractionest2 ) for the contraction coefficients , and the estimate in ( [ eq : contractionest ] ) for @xmath411 , it follows that there exists some finite ( non random ) constant @xmath412 such that the bound @xmath413holds for any pair of time indexes @xmath346 satisfying @xmath168 , particle number @xmath54 and choice of functions @xmath171 . the desired bound for ( [ eq : oscboundparticleppn ] ) is now obtained by combining this result with lemma [ lem : kintchine]:@xmath414where @xmath415 is a constant whose value does not depend on @xmath416.concerning the term @xmath417 in ( [ eq : nonasympvartriainequality]).@xmath418 ^{2}\right\ } ^{\frac{1}{2 } } \notag \\ & \leq \sum_{0\leq k\leq n}\frac{1}{\sqrt{n}}b_{k ,
n}\mathbb{e}\left\ { \left [ \sqrt{n}\left ( \eta _ { k}-\eta _ { k}^{n}\right ) \overline{d}_{k , n}(1)\times \sqrt{n}v_{k}^{n}\left ( \overline{d}_{k , n}^{n}(\widetilde{s}_{k , n}^{n})\right ) \right ] ^{2}\right\ } ^{\frac{1}{2 } } \notag \\ & \leq \sum_{0\leq k\leq n}\frac{1}{\sqrt{n}}b_{k , n}\mathbb{e}\left\ { \left [ \sqrt{n}\left ( \eta _ { k}-\eta _ { k}^{n}\right ) \overline{d}_{k , n}(1)\right ] ^{4}\right\ } ^{\frac{1}{4 } } \notag \\ & \times \mathbb{e}\left\ { \left [ \sqrt{n}v_{k}^{n}\left ( \overline{d}_{k , n}^{n}(\widetilde{s}_{k , n}^{n})\right ) \right ] ^{4}\right\ } ^{\frac{1}{4 } } \notag \\ & \leq \frac{1}{\sqrt{n}}e(n+1 ) \label{eq : secondorderrem_l2}\end{aligned}\]]where @xmath419 is a constant whose value does not depend on @xmath416 .
the second line follows from ( [ eq : contractionest ] ) and the third by the cauchy - schwartz inequality .
the final line was arrived at by the same reasoning used to derive bound ( [ eq : fisrtordererror_meansquare ] ) and lemma [ lem : lperrorfilter ]
. the assertion of the theorem may be verified by substituting bounds ( [ eq : fisrtordererror_meansquare ] ) and ( eq : secondorderrem_l2 ) into ( [ eq : nonasympvartriainequality ] ) .
let @xmath425 denote the largest integer less than or equal to @xmath170 .
since the result is obvious for @xmath426 , let @xmath427.@xmath428 where @xmath429 and @xmath430 it may be verified that @xmath431 and @xmath432 hence the result follows .
rou , f. , del moral , p. and guyader , a. ( 2008 ) . a non asymptotic variance theorem for unnormalized feynman - kac particle models , technical report inria-00337392 .
available at http://hal.inria.fr/inria-00337392_v1/ elliott , r.j .
, ford , j.j . and moore , j.b .
( 2002 ) on - line almost - sure parameter estimation for partially observed discrete - time linear systems with known noise characteristics . _
control sig .
_ , * 16 * , 435 - 453 .
kantas , n. , doucet , a. , singh , s.s . and maciejowski , j.m .
an overview of sequential monte carlo methods for parameter estimation in general state - space models . in _
proceedings ifac system identification _ ( sysid ) meeting .
poyiadjis , g. , doucet , a. and singh , s.s .
particle approximations of the score and observed information matrix in state - space models with application to parameter estimation . _
biometrika _ , to appear . | sequential monte carlo ( smc ) methods are a widely used set of computational tools for inference in non - linear non - gaussian state - space models .
we propose a new smc algorithm to compute the expectation of additive functionals recursively . essentially , it is an online or forward - only implementation of a forward filtering backward smoothing smc algorithm proposed in @xcite . compared to the standard path space smc estimator
whose asymptotic variance increases quadratically with time even under favourable mixing assumptions , the asymptotic variance of the proposed smc estimator only increases linearly with time .
this forward smoothing procedure allows us to implement on - line maximum likelihood parameter estimation algorithms which do not suffer from the particle path degeneracy problem . _ some key words _ : expectation - maximization , forward filtering backward smoothing , recursive maximum likelihood , sequential monte carlo , smoothing , state - space models . |
a differentiable lefschetz fibration is a differentiable surjection @xmath7 of a closed oriented manifold @xmath8 with only finitely many critical points of the form @xmath9 . here
@xmath10 and @xmath11 are complex coordinates on @xmath8 and @xmath3 respectively that are compatible with the orientations .
it is natural to ask how far differentiable lefschetz fibrations are from holomorphic ones .
this question becomes even more interesting in view of donaldson s result on the existence of symplectic lefschetz pencils on arbitrary symplectic manifolds @xcite .
conversely , by an observation of gompf total spaces of differentiable lefschetz fibrations have a symplectic structure that is unique up to isotopy .
the study of differentiable lefschetz fibrations is therefore essentially equivalent to the study of symplectic manifolds . in dimension @xmath12
apparent invariants of a lefschetz fibration are the genus of the non - singular fibers and the number and types of irreducible fibers . by the work of gromov and mcduff @xcite any genus-@xmath1 lefschetz fibration is in fact holomorphic .
likewise , for genus-@xmath13 the topological classification of elliptic fibrations by moishezon and livn @xcite implies holomorphicity in all cases .
our main theorem generalizes these results to genus @xmath14 . for the statement of our result recall that the mapping class group @xmath15 is a @xmath16-extension of the braid group @xmath17 .
we say that a genus-@xmath14 lefschetz fibration has _ transitive monodromy _ if the monodromy acts transitively on the set of strands .
[ genus-2 lefschetz fibrations ] let @xmath7 be a genus-@xmath14 differentiable lefschetz fibration with transitive monodromy .
if all singular fibers are irreducible then @xmath18 is isomorphic to a holomorphic lefschetz fibration .
note that the conclusion of the theorem becomes false if we allow reducible fibers , see e.g. @xcite .
the authors expect that a genus-@xmath14 lefschetz fibration with @xmath19 singular fibers , @xmath11 of which reducible , is holomorphic if @xmath20 for some universal constant @xmath21 .
in particular , any genus-@xmath14 lefschetz fibration should become holomorphic after fiber sum with sufficiently many copies of the rational genus-@xmath14 lefschetz fibration with @xmath22 irreducible singular fibers .
this problem should also be solvable by the method presented in this paper .
in @xcite we reduced theorem a to a holomorphicity statement for certain symplectic surfaces in @xmath3-bundles over @xmath3 .
the main theorem therefore follows essentially from the following isotopy result for symplectic surfaces in rational ruled symplectic @xmath12-manifolds .
[ isotopy in rational ruled ] let @xmath7 be an @xmath3-bundle and @xmath23 a connected surface symplectic with respect to a symplectic form that is isotopic to a khler form .
if @xmath24 then @xmath25 is symplectically isotopic to a holomorphic curve in @xmath8 , for some choice of complex structure on @xmath8 .
[ wlog sigma>0 ] by gromov - witten theory there exist surfaces @xmath26 , homologous to a section with self - intersection @xmath1 or @xmath13 and a fiber , respectively , with @xmath27 , @xmath28 .
it follows that @xmath29 unless @xmath25 is homologous to a negative section . in the latter case proposition [ fibered j ] produces an isotopy to a section with negative self - intersection number .
the result follows then by the classification of @xmath3-bundles with section .
we may therefore add the positivity assumption @xmath29 to the hypothesis of the theorem .
the complex structure on @xmath8 may then be taken generic , thus leading to @xmath5 or the first hirzebruch surface @xmath30 .
disconnectedness of the branch locus causes a difficulty to extend our theorem on genus-@xmath14 lefschetz fibrations to all cases .
one possibility may be to employ braid - theoretic arguments to reduce to the connected case .
we will discuss this in a future paper .
a similar result holds for surfaces of low degree in @xmath5 .
[ isotopy in p^2 ] any symplectic surface in @xmath5 of degree @xmath31 is symplectically isotopic to an algebraic curve . for @xmath32
this theorem is due to gromov @xcite , for @xmath33 to sikorav @xcite and for @xmath34 to shevchishin @xcite . together with the classification of symplectic structures on @xmath3-bundles over @xmath3 by mcduff , lalonde , a.k .
liu and t.j .
li ( see @xcite and references therein ) our results indeed imply a stronger classification of sympletic submanifolds up to hamiltonian symplectomorphism . here
we wish to add only the simple observation that a symplectic isotopy of symplectic submanifolds comes from a family of hamiltonian symplectomorphisms .
let @xmath35 be a symplectic @xmath12-manifold and assume that @xmath36 , @xmath37 $ ] is a family of symplectic submanifolds .
then there exists a family @xmath38 of hamiltonian symplectomorphisms of @xmath8 with @xmath39 and @xmath40 for every @xmath11 . at
a @xmath41 choose complex darboux coordinates @xmath42 , @xmath43 with @xmath44 describing @xmath45 .
in particular , @xmath46 . for @xmath11 close to @xmath47 let @xmath48 , @xmath49 be the functions describing @xmath50 as graph @xmath51 . define @xmath52 then for every fixed @xmath11 @xmath53 thus along @xmath50 @xmath54 the hamiltonian vector field belonging to @xmath55 thus induces the given deformation of @xmath50 . to globalize patch the functions @xmath55 constructed locally over @xmath45 by a partition of unity on @xmath45 . as @xmath55 vanishes along @xmath50 , at time @xmath11
the associated hamiltonian vector field along @xmath50 remains unchanged .
extend @xmath55 to all of @xmath8 arbitrarily .
finally extend the construction to all @xmath56 $ ] by a partition of unity argument in @xmath11 .
_ conventions : _
we endow complex manifolds such as @xmath57 or @xmath58 with their integrable complex structures , when viewed as almost complex manifolds .
a map @xmath59 of almost complex manifolds is _ pseudo - holomorphic _ if @xmath60 .
a _ pseudo - holomorphic curve _
@xmath61 in @xmath62 is the image of a pseudo - holomorphic map @xmath63 with @xmath25 a not necessarily connected riemann surface .
if @xmath25 may be chosen connected then @xmath61 is _ irreducible _ and its _ genus _
@xmath64 is the genus of @xmath25 for the generically injective @xmath65 . if @xmath66 then @xmath61 is _
rational_. a _ @xmath67-holomorphic @xmath14-cycle _ in an almost complex manifold @xmath62 is a locally finite formal linear combination @xmath68 where @xmath69 and @xmath70 is a @xmath67-holomorphic curve .
the support @xmath71 of @xmath61 will be denoted @xmath72 .
the subset of singular and regular points of @xmath72 are denoted @xmath73 and @xmath74 respectively . if all @xmath75 the cycle is _
reduced_. we identify such @xmath61 with their associated pseudo - holomorphic curve @xmath72 .
smoothing _ of a pseudo - holomorphic cycle @xmath61 is a sequence @xmath76 of smooth pseudo - holomorphic cycles with @xmath77 in the @xmath78-topology , see section [ cycle topology ] . by abuse of notation
we often just speak of a smoothing @xmath79 of @xmath61 meaning @xmath80 with @xmath81 as needed . for an almost complex manifold @xmath82
denotes the bundle of @xmath83-forms . _
complex coordinates _ on an even - dimensional , oriented manifold @xmath8 are the components of an oriented chart @xmath84 . throughout the paper we fix some @xmath85 .
almost complex structures will be of class @xmath86 for some sufficiently large integer @xmath87 unless otherwise mentioned . the unit disk in @xmath88
is denoted @xmath89 .
if @xmath90 is a finite set then @xmath91 is its cardinality .
we measure distances on @xmath8 with respect to any riemannian metric , chosen once and for all . the symbol @xmath92 denotes homological equivalence .
an _ exceptional sphere _ in an oriented manifold is an embedded @xmath14-sphere with self - intersection number @xmath93 .
in our proof of the isotopy theorems it will be crucial to reduce to a fibered situation . in sections
[ sect1 ] , [ sect2 ] and [ sect4 ] we introduce the notation and some of the tools that we have at disposal in this case .
let @xmath94 be a smooth @xmath3-fiber bundle .
if @xmath95 is a symplectic manifold and all fibers @xmath96 are symplectic we speak of a _
symplectic @xmath3-bundle_. if @xmath97 and @xmath98 are almost complex manifolds and @xmath18 is pseudo - holomorphic we speak of a _ pseudo - holomorphic _ @xmath3-bundle .
if both preceding instances apply and @xmath99 tames @xmath67 then @xmath100 is a _
symplectic pseudo - holomorphic @xmath3-bundle_. in the sequel we will only consider the case @xmath101 . then @xmath102 is differentiably isomorphic to one of the holomorphic @xmath103-bundles @xmath104 or @xmath105 .
any almost complex structure making a symplectic fiber bundle over a symplectic base pseudo - holomorphic is tamed by some symplectic form . to simplify computations
we restrict ourselves to dimension @xmath12 .
[ always tamed ] let @xmath35 be a closed symplectic @xmath12-manifold and @xmath106 a smooth fiber bundle with all fibers symplectic . then for any symplectic form @xmath107 on @xmath108 and any almost complex structure @xmath67 on @xmath8 making the fibers of @xmath18 pseudo - holomorphic
, @xmath109 tames @xmath67 for @xmath110 .
since tamedness is an open condition and @xmath8 is compact it suffices to verify the claim at one point @xmath111 .
write @xmath112 . choose a frame @xmath113 for @xmath114 with @xmath115
similarly let @xmath116 be a frame for the @xmath99-perpendicular plane @xmath117 with @xmath118 for some @xmath119 .
by rescaling @xmath107 we may also assume @xmath120 .
replacing @xmath116 by @xmath121 , @xmath122 , @xmath123 $ ] , the coefficients @xmath124 , @xmath125 vary in a compact set
. it therefore suffices to check that for @xmath126 @xmath127 is positive for all @xmath128 .
this term is minimal for @xmath129 where the value is @xmath130 .
this is positive for @xmath131 .
denote by @xmath132 the anti - holomorphic tangent bundle of an almost complex manifold @xmath62 .
consider a submersion @xmath133 of an almost complex @xmath12-manifold with all fibers pseudo - holomorphic curves .
let @xmath134 be complex coordinates on @xmath8 with @xmath135 fiberwise holomorphic .
then @xmath136 for some complex - valued functions @xmath137 .
clearly , @xmath138 vanishes precisely when @xmath18 is pseudo - holomorphic for some almost complex structure on @xmath108 .
the nijenhuis tensor @xmath139 , defined by @xmath140-[x , y]-j[x , jy]-j[jx , y],\ ] ] is antisymmetric and @xmath67-antilinear in each entry . in dimension
@xmath12 it is therefore completely determined by its value on a pair of vectors that do not belong to a proper @xmath67-invariant subspace .
for the complexified tensor it suffices to compute @xmath141 + \frac{i}{2}j[\partial_{\bar z } + a\partial_z + b \partial_w , \partial_{\bar w}]\\ & = & \frac{1}{2}(\partial_{\bar w } a)\big(\partial_z -i
j\partial_z\big ) + ( \partial_{\bar w } b)\partial_w.\end{aligned}\ ] ] since @xmath142 and @xmath143 are linearly independent we conclude : [ integrable cx str ] an almost complex structure @xmath67 on an open set @xmath144 with @xmath145 is integrable if and only if @xmath146 .
let @xmath147 .
then @xmath148 and @xmath149 are holomorphic coordinates on @xmath8 .
the lemma gives a convenient characterization of integrable complex structures in terms of the functions @xmath137 defining @xmath150 .
to globalize we need a connection for @xmath18
. the interesting case will be @xmath18 pseudo - holomorphic or @xmath151 , to which we restrict from now on .
[ connection and almost cx strs .
] let @xmath94 be a submersion endowed with a connection @xmath152 and let @xmath153 be an almost complex structure on @xmath108
. then the set of almost complex structures @xmath67 making @xmath154 pseudo - holomorphic is in one - to - one correspondence with pairs @xmath155 where 1 .
@xmath156 is an endomorphism of @xmath157 with @xmath158 .
2 . @xmath159 is a homomorphism @xmath160 that is complex anti - linear with respect to @xmath153 and @xmath156 : @xmath161 identifying @xmath162 via @xmath152 the correspondence is @xmath163 the only point that might not be immediately clear is the equivalence of @xmath164 with complex anti - linearity of @xmath159 .
this follows by computing @xmath165 [ adapted coordinates ] let @xmath166 be a pseudo - holomorphic submersion , @xmath167 , @xmath168 . then locally in @xmath8 there exists a differentiable map @xmath169 inducing a pseudo - holomorphic embedding @xmath170 for every @xmath171 .
moreover , to any such @xmath172 let @xmath173 be the homomorphism provided by lemma [ connection and almost cx strs .
] applied to the connection belonging to @xmath172 .
let @xmath135 be the pull - back by @xmath172 of the linear coordinate on @xmath88 and @xmath174 a holomorphic coordinate on @xmath108 .
then @xmath175 and @xmath135 are complex coordinates on @xmath8 , and @xmath176 for the @xmath88-valued function @xmath177 on @xmath8 with @xmath178 since @xmath18 is pseudo - holomorphic , @xmath67 induces a complex structure on the fibers @xmath179 , varying smoothly with @xmath171 . hence locally in @xmath8 there exists a @xmath88-valued function @xmath135 that fiberwise restricts to a holomorphic coordinate .
this defines the trivialization @xmath172 . in the coordinates @xmath180
define @xmath177 via @xmath181 .
then @xmath182 so the projection of @xmath183 onto @xmath150 is @xmath184 the two lemmas also say how to define an almost complex structure making a given @xmath94 pseudo - holomorphic , when starting from a complex structure on the base , a fiberwise conformal structure , and a connection for @xmath18 . for the symplectic isotopy problem we can reduce to a fibered situation by the following device .
[ fibered j ] let @xmath185 be a symplectic @xmath3-bundle .
let @xmath23 be a symplectic submanifold .
then there exists an @xmath99-tamed almost complex structure @xmath67 on @xmath8 and a map @xmath186 with the following properties . 1 .
@xmath187 is isotopic to @xmath18 .
2 . @xmath187 is pseudo - holomorphic .
@xmath25 is @xmath67-holomorphic .
moreover , if @xmath188 is a family of symplectic submanifolds there exist families @xmath189 and @xmath190 with the analogous properties for every @xmath11 .
we explained in @xcite , proposition 4.1 how to obtain a symplectic @xmath3-bundle @xmath191 , isotopic to @xmath18 , so that all critical points of the projection @xmath192 are simple and positive .
this means that near any critical point there exist complex coordinates @xmath180 on @xmath8 with @xmath193 for some holomorphic coordinate @xmath174 on @xmath103 and so that @xmath25 is the zero locus of @xmath194 .
we may take these coordinates in such a way that @xmath44 defines a symplectic submanifold .
this property will enter below when we discuss tamedness . since the fibers of @xmath187 are symplectic
the @xmath99-perpendicular complement to @xmath195 in @xmath196 defines a subbundle mapping isomorphically to @xmath197 .
this defines a connection @xmath152 for @xmath187 . by changing @xmath152 slightly near the critical points
we may assume that it agrees with the connection defined by the projections @xmath198 .
the coordinate @xmath135 defines an almost complex structure along the fibers of @xmath187 near any critical point .
since at @xmath199 the tangent space of @xmath25 agrees with @xmath195 , this almost complex structure is tamed at the critical points with respect to the restriction @xmath200 of @xmath99 to the fibers .
choose a complex structure @xmath201 on @xmath195 that is @xmath200-tamed and that restricts to this fiberwise almost complex structure near the critical points .
by lemma [ connection and almost cx strs . ]
it remains to define an appropriate endomorphism @xmath202 by construction of @xmath152 and the local form of @xmath25 we may put @xmath203 near the critical points . away from the critical points let @xmath193 and @xmath135 be complex coordinates as in lemma [ adapted coordinates ] .
then @xmath25 is locally a graph @xmath204
. this graph will be @xmath205-holomorphic if and only if @xmath206 here @xmath177 is related to @xmath159 via @xmath207 .
hence this defines @xmath159 along @xmath25 away from the critical points .
we want to extend @xmath159 to all of @xmath8 keeping an eye on tamedness . for non - zero @xmath208
the latter requires @xmath209 near the critical points we know that @xmath210 because @xmath44 defines a symplectic submanifold .
away from the critical points @xmath211 and @xmath212 lie in the @xmath99-perpendicular complement of a symplectic submanifold and therefore @xmath210 too .
possibly after shrinking the neighbourhoods of the critical points above we may therefore assume that tamedness holds for @xmath213 . by construction it also holds with the already defined @xmath159 along @xmath25 .
extend this @xmath159 differentiably to all of @xmath8 arbitrarily .
let @xmath214 $ ] be a function with @xmath215 and with support contained in an @xmath216-neighbourhood of @xmath25 .
then for @xmath216 sufficiently small , @xmath217 does the job .
if @xmath25 varies in a family argue analogously with an additional parameter @xmath11 . in the next section
we will see some implications of the fibered situation for the space of pseudo - holomorphic cycles .
one advantage of having @xmath8 fibered by pseudo - holomorphic curves is that it allows to describe @xmath67-holomorphic cycles by weierstrapolynomials , cf .
globally we are dealing with sections of a relative symmetric product .
this is the topic of the present section . throughout @xmath133
is a pseudo - holomorphic @xmath3-bundle . to study @xmath67-holomorphic curves @xmath218 of degree @xmath219 over @xmath108 we consider the @xmath219-fold relative symmetric product @xmath220}\to b$ ] of @xmath8 over @xmath108
this is the quotient of the @xmath219-fold fibered product @xmath221 by the permutation action of the symmetric group @xmath222 .
set - theoretically @xmath220}$ ] consists of @xmath1-cycles in the fibers of @xmath18 of length @xmath219 . [ m^[d ] ] there is a well - defined differentiable structure on @xmath220}$ ] , depending only on the fiberwise conformal structure on @xmath8 over @xmath108 .
let @xmath223 be a local trivialization of @xmath18 that is compatible with the fiberwise conformal structure , see the proof of lemma [ integrable complex structure ] for existence .
let @xmath174 be a complex coordinate on @xmath224 . to define a chart near a @xmath1-cycle @xmath225 choose @xmath226 and a biholomorphism @xmath227 .
the @xmath219-tuples with entries disjoint from @xmath228 give an open @xmath222-invariant subset @xmath229 now the ring of symmetric polynomials on @xmath230 is free .
a set of generators @xmath231 together with @xmath232 provide complex , fiberwise holomorphic coordinates on @xmath233}$ ] .
different choices lead to fiberwise biholomorphic transformations .
the differentiable structure is therefore well - defined .
we emphasize that different choices of the fiberwise conformal structure on @xmath8 over @xmath108 lead to different differentiable structures on @xmath220}$ ] .
note also that @xmath234}$ ] is a branched galois covering with covering group @xmath222 .
the branch locus is stratified according to partitions of @xmath219 , parametrizing cycles with the corresponding multiplicities .
the _ discriminant locus _ is the union of all lower - dimensional strata .
the stratum belonging to a partition @xmath235 with @xmath236 and @xmath237 occurring @xmath238-times is canonically isomorphic to the complement of the discriminant locus in @xmath239}\times_b\ldots\times_b m^{[d_s]}$ ] .
[ cx structure on m^[d ] ] there exists a unique continuous almost complex structure on @xmath220}$ ] making the covering @xmath240}\ ] ] pseudo - holomorphic .
it suffices to check the claim locally in @xmath220}$ ] .
let @xmath241 be a fiberwise holomorphic coordinate as in the previous lemma .
let @xmath232 and @xmath177 be as in lemma [ adapted coordinates ] , so @xmath242 let @xmath243 be the pull - back of @xmath135 by the @xmath244-th projection @xmath245 . by the definition of the differentiable structure on @xmath220}$ ] ,
the @xmath246-th elementary symmetric polynomial @xmath247 descends to a locally defined smooth function @xmath248 on @xmath220}$ ] .
the pull - back of @xmath174 to @xmath220}$ ] , also denoted by @xmath148 , and the @xmath248 provide local complex coordinates on @xmath220}$ ] .
the almost complex structure on @xmath249 is @xmath250 the horizontal anti - holomorphic vector field @xmath251 is @xmath222-invariant , hence descends to a continuous vector field @xmath252 on @xmath220}$ ] . together with the requirement that fiberwise the map @xmath253}$ ] be holomorphic , @xmath252 determines the almost complex structure on @xmath220}$ ] .
the horizontal vector field @xmath252 in the lemma , hence the almost complex structure on @xmath220}$ ] , will generally only be of hlder class @xmath254 in the fiber directions , for some @xmath255 , see @xcite .
however , at @xmath1-cycles @xmath256 with @xmath67 integrable near @xmath257 it will be smooth and integrable as well .
in fact , by lemma [ integrable cx str ] integrability is equivalent to holomorphicity of @xmath177 along the fibers .
this observation will be crucial below .
[ cycles - section ] there is an injective map from the space of @xmath67-holomorphic cycles on @xmath8 of degree @xmath219 over @xmath108 and without fiber components to the space of @xmath258}}$]-holomorphic sections of @xmath259}\to b$ ] .
a cycle @xmath260 maps to the section @xmath261 the intersection taken with multiplicities .
the image of the subset of reduced cycles are the sections with image not entirely lying in the discriminant locus .
we may reduce to the local problem of cycles in @xmath262 . in this case
the statement follows from @xcite , theorem i. by using the stratification of @xmath220}$ ] by fibered products @xmath239 } \times_b\ldots\times_b m^{[d_k]}$ ] with @xmath263 it is also possible to treat cycles with multiple components .
in fact , one can show that a pseudo - holomorphic section of @xmath220}$ ] has image in exactly one stratum except at finitely many points .
now the almost complex structure on a stratum agrees with the almost complex induced from the factors .
the claim thus follows from the proposition applied to each factor . because this result is inessential to what follows details are left to the reader . to study deformations of a @xmath67-holomorphic cycle it therefore suffices to look at deformations of the associated section of @xmath220}$ ] .
essentially this is what we will do ; but as we also have to treat fiber components we describe our cycles by certain polynomials with coefficients taking values in holomorphic line bundles over @xmath108 . we restrict ourselves to the case @xmath101 .
the description depends on the choice of an integrable complex structure on @xmath8 fiberwise agreeing with @xmath67 .
thus we assume now that @xmath264 is a holomorphic @xmath103-bundle .
there exist disjoint sections @xmath265 with @xmath266 .
then @xmath267 where @xmath268 is a fiber , and @xmath269 .
denote the holomorphic line bundles corresponding to @xmath270 by @xmath271 and @xmath272 .
let @xmath273 be holomorphic sections of @xmath274 respectively with zero loci @xmath90 and @xmath275 .
we also choose an isomorphism @xmath276 , where @xmath277 is the holomorphic line bundle on @xmath103 of degree @xmath13 .
note that if @xmath278}$ ] denotes the divisor of cycles @xmath279 with @xmath280 for some @xmath138 then @xmath281}\setminus h_d= s^d(l^{-e})=\bigoplus_{\nu=1}^d l^{-\nu e}.\ ] ] in fact , @xmath282 . [ cycles - coefficients ] let @xmath67 be an almost complex structure on @xmath8 making @xmath283 pseudo - holomorphic and assume @xmath284 near @xmath275 and along the fibers of @xmath18 .
1 ) let @xmath285 be a @xmath67-holomorphic @xmath14-cycle homologous to @xmath286 , @xmath287 , and assume @xmath288 .
let @xmath289 be a holomorphic section of @xmath290 with zero locus @xmath291 , with multiplicities
. then there are unique continuous sections @xmath292 of @xmath293 , @xmath294 , so that @xmath61 is the zero locus of @xmath295 as cycle .
2 ) there exist hlder continuous maps @xmath296 so that a local section @xmath297 of @xmath220}\setminus h_d$ ] is @xmath258}}$]-holomorphic if and only if @xmath298 3 ) let @xmath61 be a @xmath67-holomorphic @xmath14-cycle homologous to @xmath286 and with @xmath288 .
decompose @xmath299 with the second term containing all fiber components .
assume that @xmath284 also near @xmath300 let @xmath301 be sections of @xmath293 associated to @xmath61 according to ( 1 ) . then there exists a neighbourhood @xmath302 of the graph of @xmath303 and hlder continuous maps @xmath304 so that a section @xmath305 of @xmath306 with @xmath289 holomorphic defines a @xmath67-holomorphic cycle if and only if @xmath307 conversely , any solution of ( [ pde ] ) with @xmath308 corresponds to a @xmath67-holomorphic cycle without multiple components . here
@xmath309 is the discriminant .
moreover , if @xmath67 is integrable near @xmath72 then the @xmath310 are smooth near the corresponding points of @xmath311 .
1 ) assume first that @xmath312 and @xmath313 . then either @xmath61 is a fiber and @xmath314 is the defining polynomial ; or @xmath61 defines a section of @xmath220}$ ] as in proposition [ cycles - section ] .
any @xmath1-cycle of length @xmath219 on @xmath315 is the zero locus of a section of @xmath316 that is unique up to rescaling .
the restrictions of @xmath317 to any fiber form a basis for the space of global sections of @xmath316 .
hence , after choice of @xmath289 the @xmath292 are determined uniquely for @xmath318 away from the zero locus of @xmath289 . if @xmath319 choose a neighbourhood @xmath224 of @xmath320 so that @xmath321 with @xmath322 , @xmath323 . by the same argument as before we have unique weierstra polynomials of the form @xmath324 defining @xmath325 and @xmath326 respectively . multiplying produces a polynomial defining @xmath61 .
the first coefficient @xmath327 vanishes to the same order at @xmath320 as @xmath289 .
in fact , this order equals the intersection index of @xmath325 and @xmath61 with @xmath275 respectively .
this shows @xmath328 for some holomorphic function @xmath329 on @xmath224 with @xmath330 .
therefore @xmath331 extend continuously over @xmath320 .
in the general case let @xmath332 be the polynomial just obtained for @xmath333 .
put @xmath334 the coefficient of @xmath335 has the same zero locus as @xmath314 ; so after rescaling by a constant , @xmath336 has the desired form .
2 ) since @xmath67 and @xmath337 agree fiberwise and both make @xmath18 pseudo - holomorphic , the homomorphism @xmath338 factors over @xmath339 and takes values in @xmath195 .
let @xmath159 be the section of @xmath340 thus defined .
locally @xmath159 is nothing but the homomorphism obtained by applying lemma [ adapted coordinates ] .
to a local @xmath337-holomorphic trivialization of @xmath8 . because @xmath282 there is a canonical isomorphism @xmath341 this isomorphism understood we obtain an @xmath222-invariant map @xmath342 from @xmath343 to @xmath344 . define @xmath345 as the induced map from @xmath346}\setminus h_d$ ] .
the claim on pseudo - holomorphic sections of @xmath220}\setminus h_d$ ] is clear from the definition of @xmath258}}$ ] in proposition [ cx structure on m^[d ] ] and the description of @xmath159 in lemma [ adapted coordinates ] .
hlder continuity of the @xmath347 follows from the local consideration in @xcite .
3 ) let @xmath224 be a neighbourhood of @xmath348 with @xmath284 on @xmath349
. over @xmath350 define @xmath311 as those tuples @xmath305 with @xmath351 if @xmath352 for @xmath353 and @xmath354 define @xmath355 where @xmath356 is @xmath347 from ( 2 ) for @xmath357 .
we also put @xmath358 .
we claim that the @xmath310 are continuous . over @xmath224
this is clear as all terms vanish .
let @xmath135 be a complex coordinate on @xmath8 defining a local @xmath337-holomorphic trivialization of @xmath359 .
let @xmath360 be the induced coordinate functions on @xmath361 and @xmath177 the function encoding @xmath67 .
pulling back @xmath310 via @xmath362}$ ] gives @xmath363 it remains to show that if @xmath364 and @xmath365 are sequences with @xmath366 converging to @xmath367 with @xmath354 , @xmath368 , then ( [ b_r ] ) converges towards @xmath369 .
after reordering we may assume that @xmath370 correspond to the @xmath371 points converging to @xmath275 .
by hypothesis @xmath372 for @xmath373 .
moreover , since @xmath374 converges with non - zero limit and all @xmath375 are bounded , the @xmath376 stay uniformly bounded away from @xmath1 .
hence for any subset @xmath377 @xmath378 converges .
the limit is @xmath1 if @xmath379 .
evaluating expression ( [ b_r ] ) at @xmath380 and taking the limit gives @xmath381 as had to be shown .
the expression for @xmath310 also shows that the local equation for pseudo - holomorphicity of a section @xmath382 of @xmath220}\setminus h_d$ ] is @xmath383 this extends over the zeros of @xmath289 .
the converse follows from the local situation already discussed at length in @xcite .
finally we discuss regularity of the @xmath310 .
the partial derivatives of @xmath310 in the @xmath148-direction lead to expressions of the same form as @xmath310 with @xmath384 replaced by @xmath385 .
these are continuous by the argument in ( 2 ) . if @xmath67 is integrable near @xmath72 then @xmath177 is holomorphic there along the fibers of @xmath18 .
hence the @xmath310 and its derivatives in the @xmath148-direction are continuous and fiberwise holomorphic .
uniform boundedness thus implies the desired estimates on higher mixed derivatives .
it is instructive to compare the linearizations of the equations characterizing @xmath67-holomorphic cycles of the coordinate dependent description in this proposition and the intrinsic one in proposition [ cycles - section ] .
we have to assume that @xmath61 has no fiber components .
let @xmath386 be the section of @xmath387}\to { \mathbb{c}}{\mathbb{p}}^1 $ ] associated to @xmath61 by proposition [ cycles - section ] .
there is a pde acting on sections of @xmath388}/{\mathbb{c}}{\mathbb{p}}^1})$ ] governing ( pseudo- ) holomorphic deformations of @xmath386 . for the integrable complex structure
this is simply the @xmath389-equation .
there is a well - known exact sequence @xmath390}/{\mathbb{c}}{\mathbb{p}}^1}\big ) { \longrightarrow}0,\ ] ] describing the pull - back of the relative tangent bundle .
the @xmath391-equation giving @xmath67-holomorphic deformations of @xmath386 acts on the latter bundle . on the other hand ,
the middle term exhibits variations of the coefficients @xmath392 .
the constant bundle on the left deals with rescalings .
the final result of this section characterizes certain smooth cycles .
[ smooth cycles ] in the situation of proposition [ cycles - section ] let @xmath386 be a differentiable section of @xmath220}\to s^2 $ ] intersecting the discriminant divisor transversally . then the 2-cycle @xmath61 belonging to @xmath386 is a submanifold and the projection @xmath393 is a branched cover with only simple branch points .
moreover , @xmath61 varies differentiably under @xmath394-small variations of @xmath386 .
away from points of intersection with the discriminant divisor the symmetrization map @xmath253}$ ] is locally a diffeomorphism , and the result is clear .
moreover , the discriminant divisor is smooth only at points @xmath395 with @xmath396 ; this is the locus where exactly two points come together .
we may hence assume @xmath397 and @xmath75 for @xmath398 .
at @xmath395 the @xmath399 coordinates @xmath400 at @xmath401 , @xmath402 , and @xmath403 , @xmath404 descend to complex coordinates on @xmath405 . similarly , the variation of the @xmath401 for @xmath402 lead only to multiplication of @xmath406 by a smooth function without zeros .
it therefore suffices to discuss the case @xmath407 .
then @xmath61 is the zero locus @xmath408 the assumption says that , say , @xmath409 is a simple zero of @xmath410 by assumption there exists a function @xmath411 with @xmath412 and @xmath413 .
replacing @xmath135 by @xmath414 brings @xmath61 into standard from @xmath415 . hence @xmath61 is smooth and the projection to @xmath148 has a simple branch point over @xmath409 .
the same argument is valid for small deformations of @xmath386 .
this section contains a discussion of the topology on the space of pseudo - holomorphic cycles , which we denote @xmath416 throughout .
let @xmath417 be the space of pseudo - holomorphic stable maps .
an element of @xmath417 is an isomorphism class of pseudo - holomorphic maps @xmath418 where @xmath25 is a nodal riemann surface , with the property that there are no infinitesimal biholomorphisms of @xmath25 compatible with @xmath65 .
the @xmath78-topology on @xmath417 is generated by open sets @xmath419 defined for @xmath420 and @xmath421 a neighbourhood of @xmath422 as follows .
to compare @xmath423 with @xmath65 consider maps @xmath424 that are a diffeomorphism away from @xmath422 and that over a branch of @xmath25 at a node have the form @xmath425 for some @xmath426 .
then @xmath423 belongs to @xmath419 if such a @xmath427 exists with maximal dilation over @xmath428 less than @xmath216 and with @xmath429 recall that the dilation measures the failure of a map between riemann surfaces to be holomorphic .
note also that an intrinsic measure for the size of the neighbourhood @xmath421 of the singular set on non - contracted components is the diameter of @xmath430 in @xmath8 ; on contracted components one may take the smallest @xmath216 with @xmath421 contained in the _ @xmath216-thin part_. the latter consists of endpoints of loops around the singular points of length @xmath431 in the poincar metric . for a fixed almost complex structure of class @xmath432 ,
@xmath78-convergence of pseudo - holomorphic stable maps implies @xmath433-convergence away from the singular points of the limit . if one wants convergence of derivatives away from the singularities for varying @xmath67 one needs @xmath434-convergence of @xmath67 for some @xmath435 .
we will impose this condition separately each time we need it .
the @xmath78-topology on @xmath417 induces a topology on @xmath436 via the map @xmath437 here @xmath438 is the covering degree of @xmath439 . from this point of view
the compactness theorem for @xmath440 follows immediately from the version for stable maps .
we call this topology on @xmath440 the
_ @xmath78-topology_. alternatively , one may view @xmath436 as a closed subset of the space of currents on @xmath8 , or of the space of measures on @xmath8 .
we will not use this point of view here .
next we turn to a semi - continuity property of pseudo - holomorphic cycles in the @xmath78-topology . for a pseudo - holomorphic curve singularity @xmath441 in a @xmath12-dimensional
almost complex manifold @xmath8 define @xmath442 as the virtual number of double points .
this is the number of nodes of the image of a small , general , @xmath67-holomorphic deformation of the parametrization map from a union of unit disks to @xmath8 belonging to @xmath441 .
this number occurs in the genus formula .
if @xmath443 is the decomposition of a pseudo - holomorphic curve into irreducible components , the genus formula says @xmath444 we emphasize that in this formula @xmath61 has no multiple components . for a proof perform a small , general pseudo - holomorphic deformation @xmath445 of the pseudo - holomorphic maps with image @xmath446
this is possible by changing @xmath67 slightly away from @xmath447 .
the result is a @xmath448-holomorphic nodal curve for some small perturbation @xmath448 of @xmath67 .
the degree of the complex line bundle @xmath449 equals @xmath450 minus the number of double points of @xmath446 .
this expresses the genus of @xmath451 in terms of @xmath450 , @xmath452 and @xmath453 .
sum over @xmath138 and adjust by the intersections of @xmath446 with @xmath454 for @xmath455 to deduce ( [ genus formula ] ) . as a measure for
how far a pseudo - holomorphic cycle differs from a cycle with only ordinary double points we introduce @xmath456 similarly , as a measure for non - reducedness of a pseudo - holomorphic cycle @xmath285 put @xmath457 so @xmath458 if and only if @xmath72 has only ordinary double points and @xmath459 if and only if @xmath61 has no multiple components .
the definition of the @xmath78-topology on the space of pseudo - holomorphic cycles implies semi - continuity of the pair @xmath460 .
[ semicontinuity of m , delta ] let @xmath62 be a @xmath12-dimensional almost complex manifold with @xmath67 tamed by some symplectic form .
let @xmath461 be @xmath462-holomorphic cycles with @xmath463 in @xmath78 and in @xmath434 away from a set not containing any closed pseudo - holomorphic curves , and assume @xmath464 in the @xmath78-topology .
then @xmath465 for @xmath81 , and if @xmath466 for all @xmath467 then @xmath468 .
moreover , if also @xmath469 for all @xmath467 then for @xmath470 there is a bijection between the irreducible components of @xmath471 and of @xmath472 respecting the genera . by the definition of the cycle topology , for @xmath81 each component of @xmath473 deforms to parts of some component of @xmath474 .
this sets up a surjective multi - valued map @xmath89 from the set of irreducible components of @xmath473 to the set of irreducible components of @xmath474 .
the claim on semi - continuity of @xmath371 follows once we show that the sum of the multiplicities of the components @xmath475 does not exceed the multiplicity of @xmath476 . by the compactness theorem
we may assume that the @xmath474 lift to a converging sequence of stable maps @xmath477 .
let @xmath478 be the limit .
this is a stable map lifting @xmath473 . for a component @xmath476 of @xmath473 of multiplicity @xmath438 choose a point @xmath479 in the part of @xmath434-convergence of the @xmath462 and away from the critical values of @xmath480 .
let @xmath481 be a local submanifold of real codimension @xmath14 intersecting @xmath72 transversally and positively in @xmath482 .
as @xmath434-convergence of almost complex structures implies convergence of tangent spaces away from the critical values , @xmath275 is transverse to @xmath474 for @xmath81 with all intersections positive .
now any component of @xmath474 with a part degenerating to @xmath476 intersects @xmath275 , and @xmath483 gives the multiplicity of @xmath476 in @xmath473 .
the claimed semi - continuity of multiplicities thus follows from the deformation invariance of intersection numbers .
the argument also shows that equality @xmath484 can only hold if @xmath89 induces a bijection between the non - reduced irreducible components of @xmath474 and @xmath473 respecting the multiplicities .
this implies convergence @xmath485 , so we may henceforth assume @xmath474 and @xmath473 to be reduced , and @xmath486 to be injective .
note that @xmath480 may contract some irreducible components of @xmath487 . in any case , @xmath488 is the sum of the genera of the non - contracted irreducible components of @xmath487 , and it is not larger than the respective sum for @xmath489 . the latter equals @xmath490 if @xmath491 . by the genus formula ( [ genus formula ] )
we conclude @xmath492 for @xmath493 in homology and because the number of irreducible components is semicontinuous by surjectivity of @xmath89
. moreover , the number of singular points can at most decrease in the limit .
this shows @xmath468 .
if equality holds , there is a bijection between the singular points of @xmath471 and @xmath472 respecting the number of virtual double points .
the genus formula then shows that @xmath89 respects the genera of the irreducible components
. in the fibered situation of proposition [ cycles - section ] convergence in @xmath416 in the @xmath78-topology implies convergence of the section of @xmath220}$ ] : [ section convergence ] let @xmath94 be an @xmath3-bundle . for every @xmath467 let @xmath474 be a pseudo - holomorphic curve of degree @xmath219 over @xmath108 for some almost complex structure making @xmath18 pseudo - holomorphic .
assume that @xmath77 in the @xmath78-topology and that @xmath61 contains no fiber components .
let @xmath494 and @xmath386 be the sections of @xmath220}\to b$ ] corresponding to @xmath474 and @xmath61 , respectively , according to proposition [ cycles - section ] .
then @xmath495 we have to show the following .
let @xmath496 be a local trivialization with @xmath497 a closed ball , and let @xmath498 be an open set so that @xmath499 is proper .
let @xmath500 be the degree of @xmath501 over @xmath502 , counted with multiplicities .
then for @xmath467 sufficiently large @xmath503 shall be a ( branched ) covering of the same degree @xmath500 .
in fact , any @xmath504 has neighbourhoods of this form with @xmath421 arbitrarily small .
away from the critical points of the projection to @xmath502 both @xmath61 and @xmath474 would then have exactly @xmath500 branches on @xmath505 , counted with multiplicities . in the coordinates on @xmath220}$ ] exhibited in proposition [ m^[d ] ] the components of @xmath494 are elementary symmetric functions in these branches .
as @xmath421 can be chosen arbitrarily small this implies @xmath78-convergence of @xmath494 towards @xmath386 . by the definition of the topology on @xmath417
the @xmath474 lie in arbitrarily small neighbourhoods of @xmath72 .
properness of @xmath506 implies @xmath507 by compactness of @xmath508 we may replace @xmath72 in this inclusion by a neighbourhood .
therefore ( [ boundary inclusion ] ) holds with @xmath474 replacing @xmath72 , for @xmath81 .
we conclude that @xmath509 is proper for @xmath470 too , hence a branched covering .
let @xmath510 be its covering degree .
convergence of the @xmath474 in the @xmath78-topology implies that for every @xmath467 there exist stable maps @xmath477 , @xmath511 lifting @xmath474 and @xmath61 respectively and a @xmath512 as above with @xmath513 uniformly .
let @xmath514 be the union of the critical values of @xmath515 and of @xmath516 for all @xmath467 .
this is a countable set , hence has dense complement .
choose @xmath517 and put @xmath518 .
by hypothesis @xmath268 is @xmath462-holomorphic and transverse to @xmath486 and @xmath519 for every @xmath467 .
therefore , for each @xmath467 the cardinality of @xmath520 and of @xmath521 are @xmath510 and @xmath500 respectively .
since @xmath482 is a regular value of @xmath516 , for @xmath81 the image @xmath522 lies entirely in the regular part of non - contracted components of @xmath523 . on this part
the pull - back of the riemannian metric on @xmath8 allows uniform measurements of distances . in this
metric the distance of @xmath522 from @xmath521 , viewed as @xmath1-cycle , tends to zero for @xmath524 .
therefore @xmath525 for @xmath81 .
in a situation where the description of proposition [ cycles - coefficients],3 applies we obtain convergence of coefficients , even under the presence of fiber components in the limit .
[ coefficient convergence ] given the data @xmath526 , @xmath337 , @xmath527 , @xmath528 , @xmath61 , @xmath292 of proposition [ cycles - coefficients],3 assume that @xmath462 is a sequence of almost complex structures making @xmath18 pseudo - holomorphic and so that @xmath529 on a neighbourhood of @xmath530 that is independent of @xmath467 .
let @xmath531 be a sequence of @xmath462-holomorphic curves converging to @xmath532 in the @xmath533-topology .
let @xmath534 be holomorphic sections of @xmath290 with zero locus @xmath535 converging uniformly to @xmath289 .
then the sections @xmath536 of @xmath293 corresponding to @xmath474 converge uniformly to @xmath292 for all @xmath246 . from proposition
[ cycles - coefficients],3 the @xmath536 fulfill equations @xmath537 with uniformly bounded right - hand side .
cover @xmath103 with 2 disks intersecting in an annulus @xmath538 whose closure does not contain any zeros of @xmath289 . then @xmath539 . thus over @xmath538 the branches of @xmath474 stay uniformly bounded away from @xmath275 , hence the @xmath536 are uniformly bounded over @xmath538 .
the cauchy integral formula on each of the two disks implies a uniform estimate @xmath540 therefore , in view of boundedness of @xmath541 everywhere and of @xmath536 on @xmath538 we deduce a uniform estimate on the hlder norm @xmath542 thus it suffices to prove pointwise convergence of the @xmath536 on a dense set .
away from the zeros of @xmath289 union @xmath543 convergence follows from proposition [ section convergence ] .
in fact , the quotients @xmath544 occur as coefficients of the local section @xmath545 of @xmath220}\setminus h_d$ ] , see proposition [ cycles - coefficients],2 .
these sections correspond to a sequence of pseudo - holomorphic curves converging to a pseudo - holomorphic cycle without fiber components as considered in proposition [ section convergence ] .
note that since @xmath533-convergence @xmath546 implies convergence of the @xmath1-cycles @xmath547 , any sequence of holomorphic sections @xmath534 with zero locus @xmath535 converges after rescaling .
we are interested in finding _ unobstructed _ deformations of a pseudo - holomorphic cycle @xmath61 in a pseudo - holomorphic @xmath3-bundle . in the relevant situations this is possible after changing the almost complex structure . in this section
we give sufficient conditions for unobstructedness , while the construction of an appropriate almost complex structure occupies the next section .
the describing pde follows from proposition [ cycles - coefficients],3 .
recall the assumptions of loc.cit .
: @xmath67 integrable near @xmath72 , standard fiberwise and near @xmath275 union all fiber components of @xmath72 union @xmath548 . in the notation of loc.cit . , to set up the operator choose @xmath549 , @xmath550 with @xmath551 take @xmath552 and write @xmath553 for the open set of sobolev sections taking values in @xmath554 .
view pde ( [ pde ] ) in proposition [ cycles - coefficients ] as family of differentiable maps @xmath555 ( a_r)_{r=1,\ldots , d}&\longmapsto & \big(\bar\partial a_r- b_r(a_0,a_1,\ldots , a_d)\big)_{r=1,\ldots , d } , \end{array}\end{aligned}\ ] ] parametrized by @xmath556 . because the @xmath310 depend holomorphically on @xmath292 the linearization of this map takes the form @xmath557 here @xmath108 is a @xmath558-matrix with entries in @xmath559 .
[ linearization surjective ] assume that there exists a @xmath67-holomorphic section @xmath560 representing @xmath561 and that @xmath562 .
then @xmath563 is surjective .
moreover , for any @xmath350 the restriction map @xmath564 is surjective . unlike the case of rank @xmath13 surjectivity of @xmath563
does not follow from topological considerations .
instead we are going to identify the kernel of this operator with sections of the holomorphic normal sheaf @xmath565 of @xmath61 in @xmath8 , with zeros along @xmath275 . here
@xmath566 is the ideal sheaf of the possibly non - reduced subspace @xmath61 of a neighbourhood of @xmath72 in @xmath8 where @xmath67 is integrable .
the vanishing of @xmath567 then follows essentially topologically .
the proof proceeds in three lemmas .
local solutions of @xmath568 form a locally free @xmath569-module @xmath570 of rank @xmath219 , that is , the sheaf of holomorphic sections of a holomorphic vector bundle over @xmath103 of rank @xmath219 .
similarly , because @xmath18 is holomorphic in a neighbourhood of @xmath61 the push - forward sheaf @xmath571 is a locally free @xmath572-module of the same rank @xmath219 .
[ shk ] @xmath573 .
this correspondence holds for any base @xmath108 , so let us write @xmath108 instead of @xmath103 for brevity .
consider first the case @xmath61 smooth and transverse to @xmath275 and @xmath574 having only simple branch points .
then @xmath575 is the sheaf of holomorphic sections of @xmath576 .
let @xmath577 be the section of @xmath220}$ ] associated to @xmath61 according to proposition [ cycles - section ] .
away from the critical points of @xmath578 the inclusion @xmath579 induces an isomorphism @xmath580 .
let @xmath581 be a holomorphic section of @xmath582 along @xmath583 , @xmath584 open .
since @xmath585 , @xmath586 the @xmath244-th projection , @xmath581 induces an @xmath222-invariant holomorphic section @xmath587 of @xmath588 over @xmath589 .
recall the symmetrization map @xmath590}$ ] from proposition [ cx structure on m^[d ] ] . by the definition of the almost complex structure on @xmath220}$ ] this map is holomorphic near @xmath591 .
thus @xmath592 is a holomorphic section of @xmath593}/b})$ ] .
the vanishing of @xmath581 on @xmath594 translates into the vanishing of the normal component of @xmath592 along the divisor @xmath278}$ ] introduced before proposition [ cycles - coefficients ] . in other words
, @xmath592 is a section of @xmath595}/b } ( -\log h_d ) \big)$ ] .
it is clear that this sets up an isomorphism between @xmath596 and @xmath595}/b } ( -\log h_d ) \big)$ ] .
we claim that the module of sections over @xmath224 of the latter sheaf is canonically isomorphic to @xmath570
. then define @xmath597 away from the critical points of @xmath578 by sending @xmath581 to @xmath598 . to prove the claim we have to characterize holomorphic sections of @xmath599}/b}$ ] along the image of @xmath577 in the coordinates @xmath600 .
let @xmath601 be the fiberwise holomorphic coordinates on @xmath220}$ ] induced by @xmath135 , see proposition [ cx structure on m^[d ] ] .
a section @xmath602 of @xmath599}/b}$ ] is holomorphic iff the @xmath603 depend holomorphically on the @xmath604 and @xmath605 = 0 $ ] , where @xmath606 is the antiholomorphic horizontal vector field defining the complex structure .
the vanishing of the bracket expresses the fact that @xmath607 should imply @xmath608 .
in particular , if @xmath609 is a holomorphic function on @xmath220}$ ] then so will be @xmath610 . pulling back by @xmath611 and observing @xmath612 ( proposition [ cycles - coefficients],2 ) gives @xmath613 as condition for @xmath614 to be a holomorphic vector field along @xmath615 .
rescaling the coordinates @xmath248 by @xmath289 leads to @xmath616 .
this corresponds to a holomorphic section of @xmath617}/b } \big)$ ] iff @xmath618 .
the map @xmath619 is an isomorphism wherever defined so far .
to finish the case @xmath61 smooth , transverse to @xmath275 and with only simple branch points it remains to extend over the critical points of @xmath578 .
as we are in a purely holomorphic situation now we are free to work in actual holomorphic coordinates .
note that if @xmath61 splits into several connected components then @xmath220}$ ] is naturally a fibered product with one factor for each component of @xmath61 .
since our isomorphism @xmath619 respects this decomposition the problem is local in @xmath61 .
we may therefore assume @xmath61 to be defined by @xmath415 with @xmath620 holomorphic coordinates and @xmath148 descending to @xmath108 .
in this case @xmath621 and @xmath622 are a frame for @xmath623 . away from @xmath409
take linear combinations with @xmath624 to lift to @xmath625 .
thus @xmath626 in @xmath575 .
compute @xmath627 \psi(-u\partial_z)&= & \psi \big(\frac{1}{2}\partial_u\big ) & = & \big ( \partial_{\sigma_1 } + \frac{\sigma_1}{2 } \partial_{\sigma_2}\big ) \big| _ { \genfrac{}{}{0pt}{}{\sigma_1=0}{\sigma_2=0 } } & = & \partial_{\sigma_1}. \end{array}\ ] ] therefore @xmath619 extends to an isomorphism over all of @xmath108 . for the general case we know by @xcite that locally in @xmath108 there exists a holomorphic deformation @xmath628 of @xmath61 over the @xmath14-polydisk @xmath629 , say , with @xmath630 smooth and with @xmath631 only simply branched for all @xmath632 .
the previous reasoning gives an isomorphism of holomorphic vector bundles over @xmath633 .
since the vector bundles extend over @xmath634 this morphism extends uniquely .
define @xmath619 as the restriction of this extension to @xmath634 .
then @xmath619 is an isomorphism because the determinant of the extension does not vanish in codimension @xmath13 ; and it is unique because any two deformations of @xmath61 fit into a joint deformation with the locus of non - simply branched curves having higher codimension .
[ h^1=0 ] if @xmath562 then @xmath635 and @xmath596 is globally generated . by serre duality on @xmath61 @xmath636
now the first part of the statement follows by lemma [ l has no sections ] below with @xmath637 . to verify the hypotheses write @xmath638 with @xmath639 at most for @xmath640 .
if @xmath641 then for @xmath642 @xmath643 and @xmath644 only if @xmath645 .
hence @xmath646 and @xmath647 if @xmath648 then @xmath649 , @xmath650 and for @xmath651 @xmath652 hence in any case the inequalities required in lemma [ l has no sections ] hold .
global generatedness follows by a standard dimension argument with the riemann - roch formula provided @xmath653 .
this is also true here because @xmath654 and @xmath655 . summarizing , lemma [ shk ] identified the sheaf of local solutions of @xmath568 with @xmath596 .
because @xmath563 is locally surjective standard arguments of cohomology theory then give an identification @xmath656 because @xmath578 is a finite morphism the latter sheaf equals @xmath657 .
the latter vanishes by lemma [ h^1=0 ] .
moreover , since by the same lemma @xmath596 is globally generated , so is @xmath570 .
this gives the claimed surjectivity of the restriction [ l has no sections ] let @xmath658 , @xmath659 , be a compact holomorphic @xmath13-cycle on a complex manifold @xmath211 of dimension @xmath14 and let @xmath277 be a holomorphic line bundle over @xmath61 .
assume that for all @xmath660 , @xmath661 : @xmath662 then @xmath663 .
by induction over @xmath664 .
we identify effective @xmath13-cycles with compact complex subspaces without further notice . if @xmath665 we are dealing with a holomorphic line bundle of negative degree over a reduced space @xmath666 , which has no non - zero global sections . in the general case
let @xmath667 be the sheaf of holomorphic sections of @xmath277 and @xmath668 .
the effective cycle @xmath669 fulfills the induction hypothesis . in view of the exact sequence @xmath670 the section @xmath671 lifts to @xmath672 . because @xmath673 the second factor is @xmath674 thus @xmath672 is the sheaf of holomorphic sections of a line bundle over @xmath675 of degree @xmath676 which is @xmath677 by assumption .
hence @xmath671 vanishes identically . in place of proposition
[ cycles - coefficients ] and proposition [ linearization surjective ] one can use the fact from complex analytic geometry that the moduli space of compact complex hypersurfaces in a complex manifold @xmath211 is smooth at points @xmath61 with @xmath678 . since this result is not trivial we preferred to give the elementary if somewhat cumbersome explicit method described here .
surjectivity of @xmath679 implies a parametrization of pseudo - holomorphic deformations of @xmath61 by a finite - dimensional manifold .
our unobstructedness result , proposition [ unobstructedness of cycle space ] below , states this in a form appropriate for the isotopy problem . in the proof
we need the following version of the sard - smale theorem .
[ transverse ] let @xmath680 be banach manifolds , @xmath681 a smooth map with @xmath682 fredholm for all @xmath683 .
if @xmath684 is a direct submanifold ( differential of the inclusion map splittable ) that is transverse to @xmath685 then the set @xmath686 is of second category in @xmath90 .
apply the sard - smale theorem to the projection @xmath687 .
[ unobstructedness of cycle space ] let @xmath688 be a holomorphic @xmath103-bundle with @xmath270 disjoint holomorphic sections , @xmath689 .
for @xmath690 open sets with @xmath691 consider the space @xmath692 of almost complex structures @xmath67 on @xmath8 with @xmath284 fiberwise and on @xmath421 , integrable on @xmath224 and making @xmath90 holomorphic .
write @xmath693 with @xmath694 the space of @xmath67-holomorphic cycles in @xmath8 .
assume that @xmath299 is a @xmath67-holomorphic cycle homologous to @xmath286 with @xmath695 , for @xmath696 with @xmath697 , @xmath288 , @xmath421 containing @xmath698 .
here @xmath699 contains all non - fiber components of @xmath61
. then 1 .
@xmath700 and @xmath694 are banach manifolds at @xmath61 .
the map @xmath701 is locally around @xmath61 a projection .
the subset of singular cycles in @xmath694 is nowhere dense and does not locally disconnect @xmath694 at @xmath61 . similarly for @xmath700 . in proposition [ linearization surjective ]
we established surjectivity of the linearization of the map ( [ pde - map ] ) .
an application of the implicit function theorem with @xmath702 and @xmath289 as parameters thus establishes ( 1 ) and ( 2 ) .
we show the density claim in ( 3 ) for @xmath703 , the case of @xmath700 works analogously . for the time being assume that @xmath61 has no fiber components .
apply proposition [ transverse ] with @xmath704 an open neighbourhood of @xmath705 and @xmath706},\quad ( c',q)\longmapsto c'\cap p^{-1}(q).\ ] ] for the definition of @xmath220}$ ] see section [ sect2 ] .
this map is well - defined for @xmath325 close to @xmath61 by the absence of fiber components . in local coordinates provided by proposition [ cycles
- coefficients],1 it is evaluation of @xmath305 at points of @xmath103 , hence smooth . for @xmath707}$ ] choose the strata @xmath708}$ ] of the discriminant locus parametrizing @xmath1-cycles @xmath709 with fixed partition @xmath710 indexed by @xmath244 .
the top - dimensional stratum @xmath711 parametrizes @xmath1-cycles with exactly one point of multiplicity @xmath14 ; it is a locally closed submanifold of @xmath220}$ ] of codimension @xmath14 .
all other strata @xmath712 , @xmath713 , have codimension at least @xmath12 . since @xmath211 and @xmath714 are finite - dimensional here , the fredholm condition is vacuous .
transversality of @xmath685 to @xmath715 follows for all @xmath244 from the following lemma .
[ ev is transverse ] for any @xmath350 the map @xmath716 is a submersion at @xmath61 . on solutions of the linearized equation @xmath717 the differential of the map in question
is evaluation at @xmath320 .
the claim thus follows from the generatedness statement in proposition [ linearization surjective ] . now since @xmath718 for @xmath713 transversality of @xmath719 means that @xmath720 has no point of multiplicity larger than @xmath14 , for all @xmath350 .
on the other hand , by proposition [ smooth cycles ] transversal intersections with @xmath711 translate into smooth points @xmath482 of @xmath325 with the projection @xmath721 being simply branched at @xmath482 . for the remaining part of claim ( 3 ) we apply proposition [ transverse ] with @xmath90 the space of paths @xmath722\to \mathscr{m}_j\ ] ] connecting two smooth curves @xmath723 sufficiently close to @xmath61 .
the map is @xmath724\times{\mathbb{c}}{\mathbb{p}}^1){\longrightarrow}m^{[d]},\quad ( \gamma , t , q)\longmapsto \gamma(t)\cap p^{-1}(q).\ ] ] and @xmath252 runs over the @xmath712 as before .
again , transversality follows by lemma [ ev is transverse ] . for dimension reasons we still obtain @xmath725 for @xmath713 .
it remains to argue that not only @xmath726 is transverse to @xmath711 but even @xmath727 for every @xmath37 $ ] .
let @xmath728\times { \mathbb{c}}{\mathbb{p}}^1 $ ] be the one - dimensional submanifold of @xmath729 with @xmath727 having a point of multiplicity @xmath14 over @xmath320 .
let @xmath730 be in @xmath731 where @xmath732 is the projection @xmath733.\ ] ] then since @xmath734}$ ] is an analytic divisor and the differential of @xmath685 along @xmath735 is complex linear , it follows that @xmath736 is also in @xmath731 .
but @xmath737 is one - dimensional , so @xmath738 as had to be shown .
this finishes the proof , provided @xmath61 does not have fiber components . in the general case
let @xmath739 be the subset of cycles with fiber components .
a cycle @xmath740 has a unique decomposition @xmath741 for certain fibers @xmath332 and with @xmath742 .
the space of such configurations is of real codimension @xmath743 .
therefore @xmath252 is a union of submanifolds of real codimension at least @xmath14 , and these may be avoided in any path by small perturbations .
apply the previously established density result to this perturbed path to obtain a path of smooth cycles .
our objective is now to construct an almost complex structure @xmath67 as required in proposition [ unobstructedness of cycle space ] , making an arbitrary pseudo - holomorphic curve in a pseudo - holomorphic @xmath3-bundle @xmath67-holomorphic .
the next result constructs an appropriate integrable complex structure @xmath337 .
[ integrable complex structure ] let @xmath7 be an @xmath3-bundle and let @xmath156 be a complex structure on the fibers of @xmath18 .
let @xmath744 be disjoint sections .
then there exists an integrable complex structure @xmath337 on @xmath8 with @xmath745 and making @xmath18 a holomorphic map and @xmath270 holomorphic divisors .
moreover , if @xmath746 is an open subset and @xmath747 is a trivialization mapping @xmath270 to constant sections then @xmath337 may be chosen to make this trivialization holomorphic .
since @xmath748 we may assume @xmath749 . put @xmath750 , and let @xmath268 be a fiber with @xmath751 if @xmath752 .
it suffices to produce a map @xmath753 with the following properties . 1 .
@xmath754 is a biholomorphism for every @xmath755 .
2 . @xmath756 , @xmath757 .
3 . there exists a complex coordinate @xmath174 on an open set @xmath758 containing @xmath759 so that @xmath760 extends differentiably to a map @xmath761 inducing a biholomorphism @xmath762 .
in fact , away from @xmath268 this map may be used to define a holomorphic trivialization of @xmath18 , while near @xmath268 one may take @xmath763 . since @xmath764 a tubular neighbourhood of @xmath275 is diffeomorphic to a neighbourhood of the zero section in the complex line bundle of degree @xmath329 over @xmath103 .
let @xmath180 be complex coordinates near @xmath765 with @xmath232 , @xmath135 fiberwise holomorphic and @xmath275 given by @xmath44 .
consider the zero locus of @xmath766 .
this is a local section of @xmath18 intersecting @xmath275 at @xmath482 of multiplicity @xmath767 .
hence this zero locus extends to a section @xmath768 isotopic to @xmath275 and with @xmath769 , @xmath770 . under the presence of a trivialization over @xmath746 mapping @xmath270 to constant sections
choose @xmath768 holomorphic over @xmath224 .
away from @xmath268 we now have an @xmath3-bundle with @xmath771 disjoint sections and fiberwise complex structure .
the uniformization theorem thus provides a unique map @xmath772 that is fiberwise biholomorphic and maps @xmath773 to @xmath774 respectively .
it remains to verify ( 3 ) .
take for @xmath174 the function with @xmath232 as before .
multiplying @xmath609 by a constant @xmath775 on sections has the effect of keeping @xmath275 and @xmath90 fixed , but scaling @xmath776 by @xmath777 .
thus @xmath763 corresponds to the family with @xmath768 replaced by the graph of @xmath778 .
this family extends over @xmath779 .
we are now in position to construct an almost complex structure so that a given pseudo - holomorphic cycle has unobstructed deformations .
[ tilde j ] let @xmath526 be a pseudo - holomorphic @xmath3-bundle and @xmath270 disjoint @xmath67-holomorphic sections .
let @xmath218 be a @xmath67-holomorphic curve .
then for every @xmath780 there exist a @xmath781-diffeomorphism @xmath782 and an almost complex structure @xmath783 on @xmath8 with the following properties . 1
. @xmath685 is smooth away from a finite subset @xmath784 , and @xmath785 , @xmath786 .
2 . @xmath787 , @xmath788 .
3 . @xmath789 and @xmath270 are @xmath783-holomorphic .
4 . @xmath790 is a pseudo - holomorphic @xmath3-bundle .
@xmath783 is integrable in a neighbourhood of @xmath72 .
there exists an integrable complex structure @xmath337 on @xmath8 and an open set @xmath791 containing @xmath275 and all fiber components of @xmath61 , so that @xmath792 fiberwise and on @xmath421 .
define @xmath793 . without restriction
we may assume @xmath794 and @xmath795 .
decompose @xmath796 with the second term containing the fiber components . to avoid discussions of special cases replace @xmath61 by the closure of @xmath797 . for the construction of @xmath337 we would like to apply lemma [ integrable complex structure ] .
however , since we want @xmath792 near the fiber components of @xmath61 and @xmath785 , it is not in general possible to achieve @xmath798 . for
each @xmath799 take a local @xmath67-holomorphic section @xmath712 of @xmath18 through @xmath320 .
for each @xmath138 there exists a local trivialization @xmath800 restricting to a biholomorphism @xmath801 and sending @xmath712 and @xmath270 to constant sections . define the fiberwise complex structure @xmath201 near the @xmath332 by pulling back the complex structure on @xmath103 via these trivializations . extend to the rest of @xmath8 arbitrarily .
now define the reference complex structure @xmath337 by applying lemma [ integrable complex structure ] with the data @xmath802 and the chosen trivialization near @xmath803 .
next we construct the diffeomorphism @xmath685 .
put @xmath804 where @xmath805 denotes the set of critical points of a map .
this is a finite set .
for each @xmath806 let @xmath180 be @xmath337-holomorphic coordinates near @xmath482 with @xmath232 , @xmath807 .
if @xmath808 we also require @xmath809 . let @xmath810 be the function defining @xmath67 near @xmath482 via @xmath811 . then @xmath812 is a chart for @xmath8 mapping @xmath813 to the standard complex structure on @xmath814 .
note that in this chart @xmath18 is the projection onto the first coordinate of @xmath815 .
now @xmath816 is pseudo - holomorphic with respect to an almost complex structure agreeing at @xmath817 with the standard complex structure @xmath818 on @xmath815 .
thus theorem 6.2 of @xcite applies .
it gives a diffeomorphism @xmath819 of a neighbourhood of the origin in @xmath815 of class @xmath394 mapping @xmath816 to a holomorphic curve , and with @xmath820 .
moreover , by our choices @xmath821 is already holomorphic and hence , by the construction in @xcite remains pointwise fixed under @xmath819 .
therefore there exists , for any sufficiently small @xmath780 , a diffeomorphism @xmath685 of @xmath8 with @xmath822 and with @xmath823 , @xmath824 .
this is the desired diffeomorphism of @xmath8 . for the definition of @xmath783 observe that on @xmath825 , for @xmath309 sufficiently small and @xmath806 ,
the transformed curve @xmath789 is pseudo - holomorphic with respect to @xmath826 . on this part of @xmath8
define @xmath827 .
moreover , for @xmath828 it holds @xmath829 , hence @xmath830 .
this is true at @xmath831 by the definition of @xmath337 , and for @xmath832 because @xmath18 and @xmath270 are pseudo - holomorphic for both @xmath67 and @xmath337 .
therefore we may put @xmath792 on @xmath833 for @xmath309 sufficiently small .
so far we have defined @xmath783 on @xmath834 . to extend to the rest of @xmath8
let @xmath835 be the restriction of a meromorphic function on @xmath836 inducing a biholomorphism on each fiber as in the proof of lemma [ integrable complex structure ] .
let @xmath837 and put @xmath232 . to define @xmath783 agreeing with @xmath838 fiberwise is equivalent to giving a complex valued function @xmath177 via @xmath839
see lemmas [ connection and almost cx strs .
] , [ adapted coordinates ] .
the condition that @xmath789 be pseudo - holomorphic prescribes @xmath177 along @xmath840 .
moreover , @xmath783 coincides with @xmath838 near @xmath841 iff @xmath177 has compact support , and the already made definition of @xmath783 on @xmath421 forces @xmath177 to also vanish there .
this fixes @xmath177 on @xmath842 .
there exists an extension of @xmath177 to @xmath815 with compact support and so that @xmath843 in a neighbourhood of @xmath789 .
let @xmath844 be a non - critical point of @xmath845 . in a neighbourhood @xmath846 of @xmath482
write @xmath789 as graph @xmath847 .
we define @xmath848 on @xmath849 .
cover a neighbourhood of @xmath850 with finitely many such @xmath849 .
let @xmath851 be a partition of unity subordinate to the cover @xmath852 of @xmath850 .
for any @xmath482 the projection @xmath853 is an open embedding .
hence there exists a function @xmath854 on @xmath855 with @xmath856 .
then @xmath857 is a partition of unity for @xmath858 in a neighbourhood @xmath224 of @xmath850 in @xmath8 .
put @xmath859 for well - definedness it is crucial that @xmath135 is globally defined on @xmath860 .
now @xmath810 is a smooth function on a neighbourhood of @xmath861 in @xmath8 with the desired properties .
extend arbitrarily to @xmath862 with compact support .
to finish the proof of lemma [ tilde j ] it remains to remark that @xmath783 keeps @xmath18 pseudo - holomorphic by construction . the results of this and the last section will be useful for the isotopy problem in combination with the following lemma ( cf .
also @xcite , lemma 6.2.5 ) .
[ tilde j_n ] in the situation of lemma [ tilde j ] let @xmath863 be a sequence of almost complex structures making @xmath18 pseudo - holomorphic and converging towards @xmath67 in the @xmath533-topology on all of @xmath8 and in the @xmath864-topology on @xmath865 . for every @xmath467 let @xmath474 be a smooth @xmath462-holomorphic curve with @xmath866 and so that @xmath546 in the @xmath533-topology .
let @xmath685 , @xmath783 be a diffeomorphism and almost complex structure from the conclusion of lemma [ tilde j ] .
then , possibly after going over to a subsequence , there exists a finite set @xmath867 containing @xmath868 and almost complex structures @xmath869 with the following properties . 1
. @xmath18 is @xmath869-holomorphic .
@xmath870 is @xmath869-holomorphic .
3 . @xmath871 in @xmath533 on @xmath8 and in @xmath872 on @xmath873 .
an analogous statement holds for sequences of paths @xmath874 , @xmath875 uniformly converging to @xmath61 and @xmath67 respectively . by the gromov compactness theorem a subsequence of the @xmath474 converges as stable maps . if @xmath876 is the limit then @xmath877 .
define @xmath878 as the union of @xmath868 and of @xmath879 of the set of critical points of @xmath880 .
note that by the definition of convergence of stable maps , away from @xmath878 the convergence @xmath881 is as tuples of sections .
in other words , for @xmath882 , say of multiplicity @xmath371 in @xmath61 , there exists a neighbourhood @xmath849 of @xmath482 and disjoint @xmath462-holomorphic sections @xmath883 of @xmath18 over @xmath884 so that @xmath885 and @xmath886 has image @xmath887 . by elliptic regularity and @xmath888-convergence of the @xmath462 away from @xmath868
this convergence is even in @xmath889 . to save on notation
we now write @xmath61 , @xmath474 , @xmath462 instead of @xmath890 , @xmath870 , @xmath891 respectively .
the assumptions remain the same except that @xmath462 may only be continuous at points of @xmath868 .
a diagonal argument reduces the statement to convergence in @xmath892 for any fixed small @xmath420 in place of @xmath864-convergence on @xmath873 .
the construction of @xmath869 proceeds on @xmath771 types of regions , which are @xmath893 , @xmath894 , and @xmath895 . on @xmath893
take @xmath896 .
for the definition near @xmath878 observe that @xmath462 fulfills all requirements except that it is possibly only continuous at @xmath868 .
however , since the distance @xmath897 from @xmath474 to @xmath868 is positive , there exist smooth @xmath869 agreeing with @xmath462 away from @xmath898 and still converging to @xmath783 in the @xmath533-topology .
the interesting region is @xmath899 .
let @xmath900 be local complex coordinates on @xmath8 near some @xmath901 with @xmath174 holomorphic and @xmath135 fiberwise @xmath902-holomorphic .
assume that @xmath903 describe the branches of @xmath474 near @xmath482 as above .
since @xmath18 is pseudo - holomorphic but @xmath135 need not be fiberwise pseudo - holomorphic the almost complex structure @xmath462 is now equivalent to @xmath14 functions @xmath904 via @xmath905 pseudo - holomorphicity of @xmath906 means @xmath907 to define @xmath869 fiberwise agreeing with @xmath783 requires a function @xmath908 with @xmath909 for all @xmath244 .
the intersection of the fibers of @xmath18 with @xmath910 defines a family of closed disks @xmath911 near @xmath482 , @xmath912 .
take a triangulation of @xmath913 with vertices @xmath914 in the interior of @xmath913 as in the following figure .
+ _ triangulation of @xmath914 .
_ the lines in the interior are straight in the @xmath135-coordinate .
take @xmath309 so small that the triangulation deforms to @xmath911 for all @xmath915 .
let @xmath916 be the fiberwise piecewise linear function on @xmath917 restricting to @xmath918 along the @xmath244-th branch of @xmath474 and vanishing at the vertices on the boundary .
@xmath919 the difference @xmath920 of two branches fulfills the elliptic equation @xmath921 elliptic regularity gives an estimate @xmath922 with @xmath21 not depending on @xmath467 .
this implies the desired convergence .
the lemma implies that the hlder norm of @xmath916 tends to @xmath1 for @xmath524 .
let @xmath923 be a smoothing of @xmath916 agreeing with @xmath916 at the vertices of the triangulation and so that @xmath924 then near @xmath482 the desired almost complex structure @xmath869 will be defined by @xmath925 . to glue keeping @xmath474 and @xmath18 pseudo - holomorphic we observe that our local candidates for @xmath869 fiberwise all agree with @xmath902 .
it therefore suffices to glue the corresponding sections @xmath926 of @xmath927 ( lemma [ adapted coordinates ] ) using a partition of unity . since lemma
[ adapted coordinates ] requires a fiberwise coordinate @xmath928 do this in three steps : first on a neighbourhood of a general fiber , minus a general section , then in a neighbourhood of @xmath275 , and finally on @xmath862 .
the statement for paths @xmath929 , @xmath930 follows locally in @xmath11 by the same reasoning with an additional parameter @xmath11 ; extend to all @xmath11 with a partition of unity argument .
in this section we discuss the existence of certain generic paths of almost complex structures .
let @xmath931 be a symplectic pseudo - holomorphic @xmath3-bundle
. for the purpose of this section @xmath932 denotes the space of tamed almost complex structures on @xmath8 _ making @xmath18 pseudo - holomorphic_. endowed with the @xmath86-topology @xmath932 is a separable banach manifold .
we will use the following notion of positivity .
monotone ] an almost complex manifold @xmath62 is _ monotone _ if for every @xmath67-holomorphic curve @xmath218 it holds @xmath933 [ genericity c_1 ] for any @xmath67 in a path connected baire subset @xmath934 there exist disjoint @xmath67-holomorphic sections @xmath265 with @xmath935 .
moreover , such @xmath67 enjoy the following properties . 1 .
any irreducible @xmath67-holomorphic curve @xmath218 not equal to @xmath90 is homologous to @xmath936 where @xmath268 is the class of a fiber .
2 . @xmath62 is monotone ( definition [ def .
monotone ] ) .
in particular , there are no @xmath67-holomorphic exceptional spheres on @xmath8 except possibly @xmath90 . for @xmath937
the ( generic ) integrable complex structure @xmath90 is a holomorphic section of minimal self - intersection number
. then @xmath938 is poincar dual to @xmath939 we first consider the case @xmath940 .
the expected complex dimension of the space of smooth @xmath67-holomorphic spheres representing @xmath90 is @xmath941 this is no surprise as @xmath90 is an exceptional sphere . for @xmath942 any reducible @xmath67-holomorphic curve representing @xmath90
is the union of one section representing @xmath943 and @xmath944 fibers .
the expected complex dimension of such configurations is @xmath945 standard transversality arguments show that such reducible @xmath67-holomorphic representatives of @xmath90 do not occur for @xmath67 in a path connected baire subset of @xmath932 .
note that for any curve @xmath218 variations of @xmath942 span the cokernel of the @xmath946-operator , so transversality indeed applies .
( if @xmath947 is a fiber the cokernel is trivial since deformations of @xmath61 as fiber span the normal bundle . )
a similar reasoning gives the existence of a @xmath67-holomorphic section @xmath948 . here
the expected complex dimension is @xmath14 , so we impose @xmath14 incidence conditions to reduce to dimension @xmath1 .
now if @xmath949 is a @xmath67-holomorphic curve different from @xmath90 then @xmath950 shows that @xmath951 , @xmath952 unless @xmath953 . in any case
@xmath954 for @xmath955 we have @xmath956 , @xmath957 , the expected complex dimension of spheres representing @xmath90 is @xmath13 , and the expected complex dimension of singular configurations splitting off @xmath87 fibers is @xmath958 .
impose one incidence condition to reduce the expected complex dimensions by @xmath13 .
proceed as before to deduce the existence of disjoint @xmath67-holomorphic sections representing @xmath959 for generic @xmath67 . this time @xmath960 shows that @xmath951 with at least one inequality strict .
thus again @xmath961 for any non - trivial @xmath218 .
the other , much deeper genericity result that we will use , is due to shevchishin .
for the readers convenience we state it here adapted to our situation , and give a sketch of the proof .
[ genericity seva ] ( @xcite , theorems 4.5.1 and 4.5.3 . ) let @xmath8 be a symplectic @xmath12-manifold and @xmath560 a finite subset .
there is a baire subset @xmath962 of the space of tamed almost complex structures on @xmath8 with the following properties . 1 .
@xmath962 is path connected .
2 . for a path @xmath963}$ ] in @xmath962 let @xmath964 be the disjoint union over @xmath56 $ ] of the moduli spaces of non - multiple pseudo - holomorphic maps @xmath965 with @xmath966 , for any closed riemann surface @xmath25 . then there exists a baire subset of paths @xmath190 such that @xmath964 is a manifold , at @xmath65 of dimension @xmath967 + 2g(\sigma ) -1 - 2\#s,\ ] ] and the projection @xmath968\ ] ] is open at all @xmath65 except possibly if @xmath969-\sharp s\le 0 $ ] , @xmath970 and @xmath65 is an immersion .
( sketch ) we assume @xmath971 , the general case is similar .
let @xmath972 be the disjoint union of the moduli spaces of @xmath67-holomorphic maps to @xmath8 for every @xmath973 .
define @xmath974 to be the subset of pairs @xmath975 with cokernel of the linearization of the @xmath391-operator at @xmath65 having dimension at most @xmath13 .
let @xmath962 be the complement of the image of @xmath976 in @xmath932 .
a standard transversality argument shows that a generic path of almost complex structures lies entirely in @xmath962 .
then one estimates the codimension of subsets of @xmath974 where the cokernel is @xmath13-dimensional .
this subset is further stratified according to the so - called _ order _ and _ secondary cusp index _ of the critical points of @xmath65 .
it turns out that a generic path misses all strata except possibly for the case of order @xmath14 and secondary cusp index @xmath13 . for this the crucial point
is the existence of explicit second order perturbations of @xmath65 showing that the map @xmath977 is surjective , see @xcite , lemma 4.4.1 . here
@xmath978 is a @xmath389-operator on the torsion free part of @xmath979 . in the immersive case we saw this operator already in the sketch of the proof of theorem [ smoothing nodal curves ] .
the implicit function theorem then allows to compute the dimension of strata with @xmath980 of specified dimension .
the remaining singularities of order @xmath14 and secondary cusp index @xmath13 have local expressions @xmath981 so are topologically ordinary cusps .
moreover , there must be at least @xmath982 $ ] cusps present .
a further ingredient of the proof is that the presence of a sufficiently generic cusp contributes complex directions in the second variation of the @xmath946-equation .
this implies the following .
let @xmath65 be a critical point of the projection @xmath983 $ ] and assume @xmath982>0 $ ] .
then there is a 2-dimensional submanifold @xmath984 at @xmath65 with real coordinates @xmath985 so that @xmath986,\quad ( x , y)\longmapsto x^2-y^2.\ ] ] therefore @xmath987 $ ] is open at @xmath65 .
[ generic j ] 1 .
the proof of the theorem shows that we may restrict to any subspace @xmath932 in the space of tamed almost complex structures on @xmath8 having the following properties : for any @xmath942 and any @xmath67-holomorphic map @xmath988 there exist variations @xmath989 of @xmath67 in @xmath932 and @xmath990 of @xmath153 so that terms of the form @xmath991 span the cokernel of the linearization of the @xmath946-operator at @xmath65 .
moreover , we need enough freedom in varying @xmath992 in the normal direction to find solutions of equation ( 4.4.6 ) in @xcite
. variations of @xmath67 and @xmath153 enter in the form @xmath993 into this linearization .
therefore , both conditions are fulfilled if on some open set of smooth points of @xmath994 variations inside @xmath995 can be prescribed arbitrarily in the normal direction to @xmath994 in @xmath8 .
similarly , if variations of @xmath67 inside @xmath932 fulfill the two conditions only on an open subset @xmath996 then the analogous conclusions of the theorem for @xmath997 hold true .
the theorem also holds if we replace @xmath90 by a path @xmath998}$ ] .
[ very generic j ] for @xmath8 the total space of a symplectic @xmath3-bundle , we may restrict in the statement of theorem [ genericity seva ] to almost complex structures making @xmath18 pseudo - holomorphic ; moreover , we may assume that for any @xmath999 also the conclusions of lemma [ genericity c_1 ] hold . by the remark , for @xmath942 the only curves that might cause problems are fibers of @xmath18 .
these have always unobstructed deformations . by the same token , in the definition of @xmath962 we are free to remove the set of bad almost complex structures from lemma [ genericity c_1 ] .
the main application of this is the existence of smoothings of @xmath67-holomorphic cycles occurring along generic paths of almost complex structures .
our proof uses the unobstructedness of deformations of nodal curves in monotone manifolds , due to sikorav .
it generalizes the well - known unobstructedness lemma for smooth pseudo - holomorphic curves @xmath61 with @xmath1000 @xcite 2.1c1 , @xcite . in the case where all components are rational
the result is due to barraud @xcite . for the readers convenience
we include a sketch of the proof .
[ smoothing nodal curves ] ( @xcite , corollary 2 ) let @xmath62 be an almost complex manifold and @xmath218 a @xmath67-holomorphic curve with at most nodes and @xmath1001 a finite set .
assume that for each irreducible component @xmath1002 it holds @xmath1003 then a neighbourhood of @xmath61 in the space of @xmath67-holomorphic cycles is parametrized by an open set in @xmath230 with @xmath1004 .
the subset parametrizing nodal curves is a union of complex coordinate hyperplanes .
each such hyperplane parametrizes deformations of @xmath61 with one of the nodes unsmoothed .
in particular , for @xmath287 a @xmath67-holomorphic smoothing of @xmath61 exists .
( sketch ) we indicate the proof for @xmath1005 , the general case is similar .
let @xmath1006 be the injective @xmath67-holomorphic stable map with image @xmath61 .
standard gluing techniques for @xmath67-holomorphic curves give a parametrization of deformations of @xmath65 as @xmath67-holomorphic stable map by the finite - dimensional solution space of a non - linear equation on @xmath1007 ; here @xmath1008 parametrizes a certain universal holomorphic deformation of @xmath25 together with some marked points and @xmath421 is a linear subspace of finite codimension in a space of sections of @xmath1009 .
the differential of the equation in the @xmath421-direction is fredholm and varies continuously with @xmath683 .
the precise setup differs from approach to approach , see for example @xcite , @xcite .
the following discussion holds for either of these .
the statement of the theorem follows from this by two observations .
first , if @xmath25 has @xmath246 nodes then @xmath90 is naturally a product @xmath1010 , where @xmath1011 parametrizes deformations of the nodes of @xmath25 , while @xmath1012 takes care of changes of the complex structure of the ( normalization of the ) irreducible components of @xmath25 together with points marking the position of the singular points of @xmath25 .
in particular , the @xmath244-th coordinate hyperplane in @xmath1013 corresponds to deformations of @xmath25 with the @xmath244-th node unsmoothed .
the observation is that the @xmath946-equation is not only differentiable relative @xmath90 but even relative @xmath1013 .
in fact , variations along the @xmath1012-direction merely change the complex structure of @xmath25 away from the nodes .
this variation is manifestly differentiable in all of the gluing constructions . for the second observation
let @xmath1014 be the normalization of @xmath25 ; this is the unique generically injective proper holomorphic map with @xmath1015 smooth .
write @xmath1016 for the composition with @xmath65 .
let @xmath1017 be the linearization of the @xmath946-operator acting on sections of @xmath1009 , and @xmath1018 the analogous operator with variations in the @xmath1012-directions included .
there is a similar operator @xmath978 acting on sections of @xmath1019 , see e.g. @xcite , 1.5 for details .
both @xmath1017 and @xmath978 are @xmath1-order perturbations of the @xmath389-operator on @xmath1009 and on @xmath1020 respectively , for the natural holomorphic structures on these bundles .
the adjoint of @xmath978 thus has an interpretation as @xmath1-order perturbation of the @xmath389-operator acting on sections of @xmath1021 .
thus @xmath1022 restricted to @xmath1023 consists of pseudo - analytic sections of a holomorphic line bundle of degree @xmath1024 $ ] , which is @xmath677 by hypothesis . on the other hand ,
every zero of a pseudo - analytic section contributes positively , see e.g. @xcite .
thus @xmath978 is surjective .
the point is that this well - known surjectivity implies surjectivity of @xmath1018 . partially descending a similar diagram of section spaces on @xmath1025 gives the following commutative diagram with exact rows : @xmath1026 here smoothness of a section at a node means smoothness on each branch plus continuity .
note that the surjectivity of @xmath1027 holds only in dimension @xmath12 and because @xmath61 has at most nodes .
the diagram implies @xmath1028 this is the same as @xmath1029 .
in fact , @xmath1030 applied to variations of the complex structure of @xmath25 spans @xmath1031 modulo @xmath1032 , and @xmath1033 putting together we have shown that @xmath1034 , and @xmath1018 is the differential relative @xmath1013 of the @xmath389-operator acting on sections of @xmath1009 .
we are now ready for the main result of this section .
[ nodal smoothing ] let @xmath35 be a symplectic @xmath12-manifold and @xmath67 a tamed almost complex structure .
assume that @xmath1035 for some generic path @xmath190 as in theorem [ genericity seva],2 .
let @xmath285 be a @xmath67-holomorphic cycle with @xmath1036 for all @xmath138 .
suppose that @xmath454 is an exceptional sphere precisely for @xmath1037 and that @xmath1038 holds for every @xmath1037 .
then there exists a @xmath67-holomorphic smoothing @xmath79 of @xmath61 .
moreover , if @xmath1039 is open and @xmath1040 a decomposition without common irreducible components , then there exists a deformation of @xmath61 into a nodal curve @xmath79 with all nodes contained in @xmath224 and @xmath1041 the union of separate smoothings of @xmath1042 and @xmath1043 .
assume first that no @xmath446 is an exceptional sphere .
let @xmath445 be the generically injective @xmath67-holomorphic map with image @xmath446 .
consider the moduli space @xmath1044 of tuples of @xmath989-holomorphic maps @xmath1045 together with @xmath56 $ ] . here
@xmath1046 denotes the closed surface underlying @xmath451 with some complex structure depending on @xmath244 . by theorem [ genericity seva ]
, @xmath1044 is a smooth manifold of expected dimension and the projection @xmath1047\ ] ] is open at any @xmath1048 .
the subset of tuples @xmath1048 with one of the maps @xmath1049 having a critical point is of real codimension at least @xmath14 . similarly , since we have excluded the possibility of exceptional spheres , the condition that @xmath1049 and @xmath1050 for @xmath1051 have a common tangent line is also of real codimension @xmath14 . thus taking @xmath190 generic
we may assume these configurations to correspond to a locally finite union of locally closed submanifolds of real codimension at least @xmath14 .
we may thus choose a @xmath67-holomorphic perturbation @xmath1048 of the tuple with entries @xmath1052 avoiding the subset of singular configurations .
the image of @xmath1048 is a @xmath67-holomorphic curve @xmath61 with at most nodes and such that each component evaluates positively on @xmath1053 .
an application of theorem [ smoothing nodal curves ] now shows that a smoothing of @xmath61 exists .
this finishes the proof under the absence of exceptional spheres . in the general case
write @xmath1054 with the second term containing the exceptional spheres .
the previous construction applied to @xmath699 yields a smooth @xmath67-holomorphic curve @xmath1055 . because @xmath25 contains no exceptional sphere and @xmath1035 ,
the space of smooth @xmath67-holomorphic curves with a common tangent with @xmath1056 and homologous to @xmath25 is nowhere dense in the space of all such @xmath67-holomorphic maps .
this follows by another application of theorem [ genericity seva ] .
deforming @xmath25 slightly we may therefore assume the intersection with @xmath1056 to be transverse .
now apply theorem [ smoothing nodal curves ] to @xmath1057 . because @xmath1058 the result is a @xmath67-holomorphic smoothing @xmath1059 of @xmath1060 not containing @xmath1056 .
an induction over @xmath1061 finishes the construction .
the assumption on the intersection numbers of exceptional components with the rest of the curve guarantees that in the induction process the intersection of any exceptional component with @xmath1062 remains non - empty .
the statement on the existence of partial smoothings is clear from the proof .
[ j - hol spheres with prescribed sing s ] let @xmath526 be a pseudo - holomorphic @xmath3-bundle .
let @xmath1063 be open and @xmath1064 a proper , injective , @xmath67-holomorphic map .
assume that there exists an integrable complex structure @xmath337 on @xmath8 making @xmath18 a holomorphic map , and with @xmath1065 , @xmath1066 .
then for any @xmath1067 there exists a @xmath67-holomorphic sphere @xmath1068 approximating @xmath65 to @xmath1069-th order at @xmath1 : @xmath1070 let @xmath1071 be disjoint holomorphic sections of @xmath18 with @xmath1072 , and @xmath268 a fiber not containing @xmath1073
. there exists a holomorphic map @xmath1074 with @xmath1075 , @xmath1076 and inducing an isomorphism on each fiber , cf.lemma [ integrable complex structure ] .
let @xmath1077 be a holomophic coordinate .
put @xmath232 and @xmath1078 .
thus @xmath180 are holomorphic coordinates on @xmath860 with @xmath1079 and so that @xmath1080 see lemma [ adapted coordinates ] .
we may assume that the image of @xmath65 is not contained in a fiber , for otherwise the proposition is trivial . by changing the domain of @xmath65
slightly we may then assume @xmath1081 for the standard coordinate @xmath1082 on the domain @xmath89 of @xmath65 . the taylor expansion with respect to the coordinates @xmath180 of @xmath65 at @xmath1083 up to order @xmath1069 defines a polynomial map @xmath1084 . for @xmath1085
this approximation of @xmath65 takes the form @xmath1086 for some polynomial @xmath1087 of degree at most @xmath1069 .
since @xmath8 is a projective algebraic variety , for any @xmath1088 the map @xmath1089 extends to a holomorphic map @xmath1090 osculating to @xmath65 at @xmath1083 to order at least @xmath1069 . by the choice of @xmath135 it holds @xmath1091 .
hence @xmath1092\cdot s = k',\ ] ] and @xmath1093\sim mh+k ' f$ ] .
let @xmath195 be the relative tangent bundle . because @xmath18 is pseudo - holomorphic this is a complex line bundle .
put @xmath1094 .
now @xmath1095 is poincar - dual to @xmath1096 , hence @xmath1097 to show the existence of @xmath1098 in the statement of the proposition we consider a modified gromov - witten invariant .
let @xmath932 be the space of almost complex structures @xmath448 on @xmath8 of class @xmath86 with @xmath1099 , @xmath1100 and making @xmath18 pseudo - holomorphic .
note that by lemma [ always tamed ] tamedness on any compact subset in @xmath932 is implicit in this definition .
let @xmath972 be the disjoint union over @xmath1101 of the moduli spaces of @xmath448-holomorphic maps @xmath1102 with 1 .
@xmath1103=(\psi_0)_*[{\mathbb{c}}{\mathbb{p}}^1]$ ] in @xmath1104 .
2 . in the chosen holomorphic coordinates @xmath1082 on @xmath103 and @xmath1105 on @xmath1106 : @xmath1107 in particular , @xmath1108 .
the forgetful map @xmath1109 is a fredholm map of banach manifolds of index @xmath1 .
let @xmath1110 be @xmath448-holomorphic .
let @xmath1111 be the function describing @xmath448 on @xmath1112 .
then @xmath1113 , and @xmath448-holomorphicity of @xmath1114 is equivalent to @xmath1115 therefore a deformation @xmath1116 of @xmath1114 inside @xmath972 is @xmath448-holomorphic iff @xmath1117 linearizing with fixed @xmath1118 gives the equation @xmath1119 as @xmath1120 globalizes as a section of @xmath1121 , the intrinsic meaning of @xmath1122 is as a section of @xmath1123 .
so globally our pde is an equation of cr - type acting on sections of a complex line bundle over @xmath103 of degree @xmath93 .
the index is computed by the riemann - roch formula to be @xmath1 . moreover , as @xmath1114 is generically injective , the cokernel can be spanned by variations of @xmath1118 .
an application of the implicit function theorem finishes the proof of the lemma .
since @xmath932 is connected there exists a path @xmath190 in @xmath932 connecting @xmath337 with @xmath1124 . by a standard application of the sard - smale theorem the restriction of @xmath972 to a _
path @xmath190 is a one dimensional manifold @xmath1044 over @xmath1125 $ ] .
we claim that @xmath1044 is compact .
let @xmath1126 be @xmath1127-holomorphic , and @xmath1128 .
the gromov compactness theorem gives a @xmath1129-holomorphic cycle @xmath1130 to which a subsequence of @xmath1131 converges . by the homological condition and the local form of the elements of @xmath972 near @xmath1073
there is exactly one component of @xmath473 that projects onto @xmath103 .
all other components are fibers .
so the expected dimension drops by @xmath14 for each such bubbling off . hence bubbling off
can be avoided in a generic path as claimed .
we have thus shown that @xmath1132 $ ] is a cobordism .
it remains to observe the following fact .
the fiber of @xmath1047\ ] ] over @xmath1 consists of one element .
let @xmath1133 be @xmath337-holomorphic . if @xmath1134 then the intersection index of @xmath1135 with @xmath1136 at @xmath1137 is at least @xmath1138 .
but @xmath1139 which is absurd .
hence @xmath1140 is the only holomorphic map in @xmath1044 .
we are now ready to finish the proof of proposition [ j - hol spheres with prescribed sing s ] .
the parity of the cardinality of the fiber stays constant in a one - dimensional cobordism .
hence , by the lemma , the fiber of @xmath1132 $ ] over @xmath13 must be non - empty .
we may take for @xmath1098 any element in this fiber .
( end of proof of proposition . )
our main technical result runs as follows .
[ isotopy lemma ] let @xmath526 be a pseudo - holomorphic @xmath3-bundle with disjoint @xmath67-holomorphic sections @xmath270 .
let @xmath1141 be a sequence of almost complex structures making @xmath18 pseudo - holomorphic .
suppose that @xmath461 , @xmath1142 , are smooth @xmath462-holomorphic curves and that @xmath1143 in the @xmath78-topology , with @xmath1144 for every @xmath138 and @xmath463 in @xmath78 and in @xmath434 away from a finite set @xmath784 .
we also assume : 1 .
if @xmath1145 is a non - zero @xmath448-holomorphic cycle @xmath533-close to a subcycle of @xmath1146 , with @xmath1147 as in proposition [ very generic j ] , then @xmath1148 then any @xmath67-holomorphic smoothing @xmath1149 of @xmath473 is symplectically isotopic to some @xmath474 .
the isotopy from @xmath474 to @xmath1149 can be chosen to stay arbitrarily close to @xmath473 in the @xmath78-topology , and to be pseudo - holomorphic for a path of almost complex structures that stays arbitrarily close to @xmath67 in @xmath533 and in @xmath872 away from a finite set .
[ smoothing exists ] a @xmath67-holomorphic smoothing of @xmath473 exists by proposition [ nodal smoothing ] provided @xmath1035 for some generic path @xmath1150 as in proposition [ genericity seva ] . for the condition on exceptional spheres observe that for homological reasons @xmath8 contains at most one of them , say @xmath90 .
then either @xmath1151 and there is nothing to prove , or @xmath1152 and @xmath1153 _ proof of the lemma . _ by lemma [ always tamed ] there exists a symplectic structure taming @xmath67 .
we may therefore asume @xmath67 tamed whenever needed .
recall the functions @xmath1154 defined in section [ cycle topology ] .
order the pairs @xmath460 lexicographically : @xmath1155 we do an induction over @xmath1156 .
the induction process is inspired by the proof of @xcite , theorem 6.2.3 , but the logic is different under the presence of multiple components .
the starting point is @xmath1157 .
then @xmath473 is a smooth pseudo - holomorphic curve and the statement is trivial .
so let us assume the theorem has been established for all @xmath460 strictly less than @xmath1158 .
the proof of the induction step proceeds in 8 paragraphs , referred to as steps 18 .
1 ) _ enhancing @xmath67 . _
+ one of the two sections @xmath270 deform @xmath67-holomorphically , say @xmath275 .
we may therefore assume @xmath1159 .
apply lemma [ tilde j ] to @xmath1160 and @xmath67 .
the result is a @xmath781-diffeomorphism @xmath685 of @xmath8 , smooth away from a finite set @xmath1161 , and an almost complex structure @xmath783 making @xmath18 and @xmath270 pseudo - holomorphic . by perturbing @xmath783 slightly away from @xmath472
we may assume that @xmath1162 is monotone .
moreover , @xmath1163 is now a @xmath783-holomorphic cycle with unobstructed deformation theory in the sense made precise in proposition [ unobstructedness of cycle space ] .
the homological condition on @xmath473 required there follows because @xmath1164 .
we claim that it suffices to prove the theorem under the assumption that the limit almost complex structure @xmath67 has this special form and _ for any particular smoothing @xmath1149_. by taking the union with the finite set in the hypothesis we may assume @xmath1165 .
change @xmath474 and @xmath462 slightly to achieve @xmath1166 .
for example , apply a diffeomorphism that is a small translation near @xmath868 and the identity away from this set .
after going over to a subsequence lemma [ tilde j_n ] now provides almost complex structures @xmath869 on @xmath8 making @xmath18 pseudo - holomorphic with @xmath871 in @xmath78 and in @xmath434 away from @xmath868 , and so that @xmath870 is @xmath869-holomorphic .
the sequence @xmath1167 fulfills the hypothesis of the lemma . assuming that we can prove the lemma for @xmath1168 , pick for every @xmath81 an isotopy @xmath1169}$ ] between @xmath870 and a smoothing of @xmath1163 ; here @xmath1170 is pseudo - holomorphic for a path @xmath1171 with @xmath1172 and @xmath1173 uniformly in the respective @xmath78-topologies for @xmath524 .
changing @xmath1170 and @xmath1174 slightly near @xmath868 by an appropriate translation we may also assume @xmath1175 for all @xmath11 .
then @xmath1176 is an isotopy connecting @xmath474 with a smoothing of @xmath473 .
another application of lemma [ tilde j_n ] , now to @xmath1177 , allows to find paths @xmath930 of smooth almost complex structures , converging to @xmath67 uniformly in @xmath78 for @xmath524 , so that @xmath1178 is @xmath1179-holomorphic . to get the statement of the lemma ,
apply this reasoning both to the original sequence @xmath76 and to the given sequence of @xmath67-holomorphic smoothings of @xmath473 .
this gives an isotopy of @xmath474 and of some smoothing @xmath1149 with the @xmath685-preimage of some @xmath783-holomorphic smoothings of @xmath1163 .
finally use the fact that according to proposition [ unobstructedness of cycle space ] , @xmath902-holomorphic smoothings of @xmath1163 are unique up to isotopy .
this step is crucial to make the connection between @xmath902-holomorphic smoothings and @xmath67-holomorphic smoothing .
here we need also proposition [ coefficient convergence ] to the effect that @xmath533-convergence of cycles implies convergence of coefficients in the description of proposition [ cycles - coefficients],3 .
we can henceforth add the following hypotheses to the lemma . _
_ 1 . for some integrable complex structure @xmath337 on @xmath8
the pair @xmath1180 fulfills the hypotheses of proposition [ unobstructedness of cycle space ] by proposition [ unobstructedness of cycle space ] it now even suffices to produce an isotopy of @xmath474 with some @xmath448-holomorphic smoothing for @xmath448 sufficiently close to @xmath67 and of the form as required in loc.cit .
in step 5 we will ask @xmath67 to have a certain genericity property , which can be achieved by perturbing @xmath783 in the normal direction near some smooth points of @xmath472 .
2 ) _ replacement of non - multiple components of @xmath473 by @xmath67-holomorphic spheres . _
+ let @xmath1181 and write @xmath1182 for any pseudo - holomorphic parametrization of a branch of @xmath472 that belongs to a non - multiple component of @xmath473 at @xmath482
. denote by @xmath1183 the sum of the multiple components of @xmath473 , as cycle . for the singularities at @xmath482 of the curves underlying @xmath473 and @xmath1183 we have @xmath1184 since @xmath284 near @xmath482 , by hypothesis ( @xmath1185 ) from step 1
, proposition [ j - hol spheres with prescribed sing s ] applies .
we obtain a sequence @xmath1186 of @xmath67-holomorphic spheres approximating @xmath65 at @xmath1187 to order @xmath1069 .
note that the approximation of fiber components is exact .
we do not indicate the dependence of @xmath1098 on @xmath65 in the notation .
for @xmath110 the topological types of the curve singularities @xmath1188 and of @xmath1189 agree .
fix such a @xmath1069 . for
@xmath780 define the annulus @xmath1190 . for @xmath780
sufficiently small the intersections @xmath1191 are isotopic by an isotopy in @xmath1192 that is the identity near @xmath1193 .
therefore , for each non - multiple branch @xmath65 there exists a map @xmath1102 which agrees with @xmath1098 over @xmath1194 and with @xmath65 over @xmath1195 , and which is isotopic to @xmath65 on @xmath1192 .
let @xmath448 be a small perturbation of @xmath67 agreeing with @xmath67 near @xmath1193 and making @xmath18 and @xmath1114 pseudo - holomorphic .
at the expense of further changes of @xmath67 and @xmath1114 away from @xmath1196 , we obtain an almost complex structure @xmath448 and a @xmath448-holomorphic map @xmath1197 with * @xmath448 agrees with @xmath67 in a neighbourhood of @xmath1193 .
* @xmath1198 is isomorphic to @xmath65 over a neighbourhood of @xmath482 . * except possibly at @xmath482 the map @xmath1198 is an immersion intersecting @xmath1193 transversely .
do this inductively for all reduced branches of @xmath473 at all singular points of @xmath472 .
the result is an almost complex structure @xmath448 agreeing with @xmath67 near @xmath1193 and making @xmath18 pseudo - holomorphic , and a @xmath448-holomorphic cycle @xmath1199 all of whose non - multiple components are rational .
we refer to the part of the added components away from the singularities of @xmath472 as _ parasitic_. let @xmath1039 be a neighbourhood of @xmath1196 so that @xmath1199 agrees with @xmath473 on @xmath224 except for smooth branches of parasitic components _ away _ from the singularities of @xmath472 .
we can take the intersection of @xmath224 with the parasitic part of @xmath1200 to be a union of disks .
similarly adjust the sequence @xmath474 . by @xmath1201-convergence of the almost complex structures the tangent spaces of @xmath474
converge to the tangent spaces of @xmath472 away from the multiple components of @xmath473 .
hence , for all @xmath81 there exist an almost complex structure @xmath1202 making @xmath18 pseudo - holomorphic , a @xmath1202-holomorphic immersion @xmath1203 and an open set @xmath1204 with the following properties .
* @xmath1205 on @xmath224 , @xmath1206 in @xmath78 and in @xmath434 away from @xmath1207 .
* @xmath1208 .
* @xmath1209 is isomorphic to the inclusion @xmath1210 , which is a part of the original curve , as pseudo - holomorphic map .
define @xmath1211 .
we call @xmath1212 and @xmath1213 the _ parasitic _ part of @xmath1214 and of @xmath1215 respectively . a note on notation : because for the most part of the proof we work with curves having parasitic part we henceforth drop the primes on all symbols .
to refer to the original curves and almost complex structures we place an upper index @xmath1 , so the original sequences now read @xmath1216 and @xmath1217 .
3 ) _ description of further strategy . _
+ if @xmath473 has multiple components the further strategy is to deform @xmath486 as pseudo - holomorphic map _ away _ from @xmath473 , uniformly in @xmath467 .
in the deformation process we fix enough points so that any occurring degeneration has better singularities , measured by @xmath460 .
thus induction applies and we can continue with any smoothing of the degeneration . therefore , the deformation process is always successful .
the result is a sequence of @xmath462-holomorphic curves @xmath1218 with @xmath1215 containing the set of chosen points and symplectically isotopic to @xmath474 in the required way , but with a uniform distance from @xmath472 . by the same reason as before
@xmath1219 is a @xmath67-holomorphic cycle with better singularities .
next we replace the parasitic part with the removed part of @xmath1220 and apply the induction hypothesis , if necessary .
this will eventually lead to a @xmath448-holomorphic smoothing @xmath1221 of the original cycle @xmath1220 that is isotopic to @xmath1222 , with @xmath448 arbitrarily close to @xmath67 and fulfilling the other conditions requested in proposition [ unobstructedness of cycle space ] .
this was left to be shown in step 1 .
if @xmath473 is reduced this argument does not work because we can not prescribe enough points in the deformation to move away .
instead the pseudo - holomorphic deformation @xmath1223 of @xmath486 will now be with respect to a generic path @xmath875 of almost complex structures connecting @xmath462 with @xmath67 .
the deformation process is successful as long as @xmath1223 stays close to @xmath486 .
otherwise a diagonal argument gives a sequence @xmath1224 converging to a @xmath67-holomorphic curve with improved singularities as above .
alternatively , the reduced case follows from the local isotopy theorem due to shevchishin @xcite , theorem 6.2.3 .
in fact , in this case our proof comes down to a global version of the proof given there . for the sake of completeness we nevertheless provide full details here .
4 ) _ replacements of parasitic parts . _ + the rest of the proof will repeatedly require replacements of the parasitic part by the removed part of the original curves @xmath1225 or @xmath1226 . the purpose of this paragraph is to make this process precise .
a point of caution concerns the nodes that have been generated by the introduction of parasitic components . to simplify the proof we use the fibered structure here ,
although this is not strictly necessary . denote by @xmath1227 the union of the non - multiple irreducible components of @xmath473 _ that are not fibers of @xmath18_. away from @xmath1228 the projection @xmath1229 is a holomorphic ( branched ) covering . for small @xmath1230 @xmath1231 is a union of annuli containing no branch points of this covering .
take @xmath1230 also so small that @xmath1232 , @xmath224 the neighbourhood of @xmath1233 from step 2 .
then @xmath1234 is also a union of annuli , one for each branch of @xmath1227 near a singular point of @xmath1235 .
choose @xmath1236 and less than @xmath1237 of @xmath878 with image a union of connected components of @xmath1238 and so that @xmath1239 , @xmath1240 , are @xmath67-holomorphic disks contained in the fibers of @xmath18 . by construction
it holds @xmath1241 for the intersection number , hence also @xmath1242 for @xmath81 .
therefore , @xmath1243 is the graph of a function @xmath1244 .
the same reasoning applies to any small pseudo - holomorphic deformation of @xmath486 with respect to almost complex structures keeping @xmath18 pseudo - holomorphic .
recall from step 2 that the gluing took place outside of @xmath224 .
we will reglue on @xmath1245 , that is , away from a smaller neighbourhood @xmath1246 chosen disjoint from @xmath1245 now .
denote by @xmath1247 a metric on @xmath1248 inducing the @xmath533-topology .
normalize @xmath1249 in such a way that @xmath1250 implies @xmath1251 .
[ replace parasitic ] for @xmath1252 and @xmath81 the following holds .
let @xmath1253 , be a continuous family of almost complex structures with @xmath1254 pseudo - holomorphic , @xmath1255 and @xmath1256 for all @xmath1257 .
assume that @xmath1258 is a continuous family of @xmath1179-holomorphic immersions with @xmath1259 , injective over @xmath1260 and such that @xmath1261 then there exist a continuous family @xmath1262 of almost complex structures making @xmath18 pseudo - holomorphic and a continuous family of smooth @xmath1263-holomorphic curves @xmath1264 with the following properties .
1 . @xmath1265 , @xmath1266 .
@xmath1267 , @xmath1268 for all @xmath1257 .
3 . @xmath1269 .
@xmath1270 is a union of irreducible components of @xmath1271 containing @xmath1272 . for @xmath1252 and @xmath470
the preimage of @xmath1273 under @xmath1223 is a union of smooth circles @xmath1274 , moving continuously with @xmath11 . since @xmath1275 ,
the parasitic part of @xmath486 lies in the complement of @xmath1276 . by the construction of the parasitic part
any connected component of @xmath1277 containing some parasitic part is a disk . at time
@xmath11 remove all connected components from @xmath1278 that are deformations of such a parasitic component .
call the resulting map @xmath1279 .
the assumptions imply that @xmath1279 is an embedding , provided @xmath420 is sufficiently small and @xmath467 sufficiently large .
now for @xmath1252 and @xmath470 , on @xmath1245 the @xmath1179 get arbitrarily close to @xmath67 in the @xmath434-topology , and the image of @xmath1223 is a graph over @xmath878 .
elliptic regularity thus allows to bound the slope of @xmath1280 in terms of @xmath1281 .
hence we can glue @xmath1282 with the reduced part of @xmath1222 at only a slight change of slopes to obtain @xmath1283 .
therefore , for @xmath1252 and @xmath81 there exist @xmath1263 , @xmath1283 enjoying the requested properties .
5 ) _ the incidence conditions @xmath1284 ( note that by our convention from the end of step 2 the @xmath438 and @xmath476 in the hypothesis of the lemma are now @xmath1285 . ) for each @xmath138 choose points @xmath1286 in general position with @xmath1287 since @xmath1144 this number is non - negative .
note that hypothesis ( @xmath1288 ) applied to @xmath1289 together with rationality of the reduced part of @xmath473 implies @xmath1290 the inequality is strict if @xmath473 has multiple components .
in the reduced case all components of @xmath473 are rational and then equality holds .
write @xmath1291 and @xmath1292 for the concatenation of the @xmath1293 .
let @xmath1294 be the generically injective @xmath67-holomorphic map with image @xmath476 .
consider the moduli space @xmath1295 of pseudo - holomorphic maps @xmath1296 representing @xmath1297 $ ] in homology and with @xmath1298 . here @xmath1299 may vary . by remark [ generic j],2
we may assume @xmath67 to be regular for all small , non - trivial deformations of @xmath1300 keeping the incidence conditions @xmath1293 ; it suffices to change the almost complex structure in the normal direction near a general point of @xmath476 . if @xmath1301 this leads to a change of @xmath783 in step 1 ; if @xmath75 change @xmath448 away from the neighbourhood @xmath224 of @xmath1193 in its construction in step 2 .
now the dimension of @xmath1295 is @xmath1302 hence @xmath1300 is an isolated point of @xmath1295 for every @xmath138 .
this implies that the singularities of @xmath473 under any non - trivial deformation as @xmath67-holomorphic cycle containing @xmath1303 improve .
we measure this improvement through the pair @xmath460 .
let @xmath1304 be the number of virtual double points of @xmath472 restricted to a neighbourhood of the parasitic part . because the parasitic part intersects @xmath1305 transversally it holds @xmath1306 [ no equisingular deformations of c_infty ] let @xmath1307 be a @xmath1308-holomorphic cycle sufficiently close to @xmath473 with @xmath1309 .
assume that one of the following applies .
1 . @xmath1310 for a general path @xmath190 of almost complex structures and @xmath1311 is not a nodal curve .
2 . @xmath1312 , @xmath1313
. then @xmath1314 lemma [ semicontinuity of m , delta ] implies @xmath1315 for @xmath1307 sufficiently close to @xmath473 in the @xmath78-topology .
moreover , if equality holds for some @xmath1307 containing @xmath1316 then every component of @xmath1307 is the image of some @xmath1308-holomorphic deformation @xmath1317 of @xmath1300 with @xmath1318 . inside the parametrizing moduli space of generically injective pseudo - holomorphic maps with incidence conditions the subspace of non - immersions is of real codimension at least @xmath14 .
hence such maps do not occur over a general path of almost complex structures . in the second case
@xmath1312 we saw before that @xmath1300 is an isolated point in @xmath1295 .
so equality can only occur if @xmath1319 for all @xmath138 , that is @xmath1320 .
6 ) _ the non - reduced case : deforming @xmath474 uniformly away from @xmath473 . _
+ deform @xmath474 and @xmath462 slightly to achieve @xmath1321 for all @xmath467 .
for example , near @xmath1322 apply a locally supported diffeomorphism of @xmath8 moving one branch of @xmath474 to @xmath1322 . as before let @xmath1323 be the generically injective @xmath462-holomorphic map with image @xmath474 .
we first treat the case when @xmath473 has multiple components .
the purpose of this section is to establish the following result .
[ c_n ] for every sufficiently small @xmath420 there exists , for every @xmath81 , a @xmath462-holomorphic immersion @xmath1324 with 1 .
@xmath1325 is homotopic to @xmath486 through a family of immersions , pseudo - holomorphic with respect to a path of almost complex structures staying arbitrarily close to @xmath462 in @xmath872 .
2 . @xmath1326 .
3 . @xmath1327 .
the proof is by descending induction over the degree @xmath1328 . for a first try take paths @xmath1329 on @xmath8 with @xmath1330 and @xmath1331 .
such paths exist for all @xmath467 provided that @xmath216 is sufficiently small .
write @xmath1332 for the concatenation @xmath1333 of @xmath1303 and @xmath1329 . choose a @xmath872-small deformation @xmath1334}$ ] of @xmath462 so that * @xmath930 fulfills the properties of theorem [ genericity seva],2 with @xmath1335 .
note that the expected ( real ) dimension of deformations of @xmath473 not decreasing @xmath460 and containing @xmath1336 is @xmath1337 , so these do not occur for generic @xmath930 , cf.proposition [ no equisingular deformations of c_infty ] .
now try to deform @xmath486 as @xmath1179-holomorphic map with @xmath1338 contained in the image . by theorem [ genericity seva ]
the set of @xmath56 $ ] where this is possible , is open . for closedness let @xmath1339 be the image of a deformation of @xmath486 at times @xmath1340 . by the gromov compactness theorem we may assume cycle - theoretic convergence @xmath1341 if @xmath1342 is reduced and irreducible lemma [ nodal smoothing ] gives a @xmath1343-holomorphic partial smoothing of @xmath1342 containing @xmath1338 , see also remark [ smoothing exists ] .
the smoothing has to be partial because we want to keep deformations of the parasitic components @xmath1344 .
it remains to verify that this partial smoothing is symplectically isotopic to @xmath1339 for @xmath1345 . replace the parasitic part by the original curve , both for the sequence @xmath1339 and for the partial smoothing .
this is possible by lemma [ replace parasitic ] .
since @xmath1342 is reduced the induction hypothesis provides a symplectic isotopy between the curves with parasitic parts replaced .
glue back the parasitic parts throughout the isotopy to obtain the requested homotopy of immersions .
if @xmath1342 is non - reduced or reducible write @xmath1346 here the @xmath1347 may not be distinct .
split @xmath1303 arbitrarily into @xmath289 tuples @xmath1348 with @xmath1349 let @xmath1350 be the number of entries of @xmath1293 . since by ( [ number of constraints ] ) the number of entries @xmath1351 of @xmath1303 is less than @xmath1352 there exists an @xmath138 with @xmath1353 let @xmath1354 be the generically injective map with image @xmath1347 . by lemma [ nodal smoothing ] there exists a @xmath1343-holomorphic deformation of @xmath1354 to an immersion @xmath1355 containing @xmath1348 in the image . in the further deformation process
we want to only deform @xmath1354 and keep the other components . to this end change the path @xmath1179 for @xmath1356 so that it stays constant on a neighbourhood of @xmath1357 .
this is possible by remark [ generic j],2 .
we also choose a new path @xmath1358 so that @xmath1359
. now apply the previous reasoning with @xmath486 replaced by @xmath1355 and @xmath1332 by the concatenation of @xmath1303 and @xmath1358 .
because the degree @xmath1360 decreases by an integral amount each time we split the curve , after finitely many changes of @xmath930 the deformation will be successful for all @xmath11 .
the result is a family of cycles @xmath1361 that is pseudo - holomorphic for a family of @xmath872-small perturbations of @xmath462 and containing a point @xmath1329 with @xmath1362 finally , the inclusion @xmath1363 for all @xmath11 implies that @xmath1364 is always smaller than @xmath1365 .
note that a potential drop of @xmath309 happens on @xmath1366 .
hence upon replacement of the parasitic part @xmath309 decreases by @xmath1304 , so the induction hypothesis applies to the reglued curves .
as above we thus deduce that for any @xmath11 there exists a partial smoothing of @xmath1361 that is the image of a @xmath1179-holomorphic map @xmath1223 homotopic to @xmath486 through a family of pseudo - holomorphic immersions as demanded .
hence there exists @xmath1367 with @xmath1368 fulfilling the desired properties 13 .
7 ) _ the reduced case : pseudo - holomorphic deformation of @xmath474 over a path @xmath1369}$ ] connecting @xmath462 with @xmath67 . _
+ if @xmath473 is reduced there is no room for imposing one more point constraint without spoiling unobstructedness of the deformation . instead choose general paths @xmath1369}$ ] of tamed almost complex structures connecting @xmath462 with @xmath67 and such that @xmath1370 uniformly in the @xmath533-topology everywhere and in the @xmath872-topology on @xmath865 . by remark
[ generic j],2 we can also arrange @xmath1371 for every @xmath1372 . for @xmath470 and @xmath1373
we seek to find a @xmath1179-holomorphic immersion @xmath1374 with * @xmath1375 .
* @xmath1223 is homotopic to @xmath486 through a family of immersions , pseudo - holomorphic with respect to a path of almost complex structures staying @xmath216-close to @xmath462 in @xmath1376 and in @xmath872 on @xmath1377 .
* @xmath1378 .
[ dichotomy ] for every @xmath420 one of the following two cases occurs . 1 .
there exists an @xmath1379 so that @xmath1223 with properties ( i)(iii ) exists for all @xmath1380 .
2 . for every @xmath470
there exists @xmath1224 with properties ( i)(iii ) and so that @xmath1381 for @xmath1382 we may take @xmath1383 .
let @xmath1384 be the maximal number so that for each @xmath1385 a map @xmath1223 obeying ( i)(iii ) exists .
assume that @xmath1386 .
let @xmath1387 be a sequence of maps @xmath1388 obeying ( i)(iii ) and with @xmath1389 for @xmath1390 . by the gromov compactness theorem ,
after going over to a subsequence , we may assume cycle - theoretic convergence @xmath1391 since @xmath930 is generic at @xmath1386 and @xmath1392 , proposition [ no equisingular deformations of c_infty ] implies @xmath1393 unless @xmath473 is nodal .
if @xmath460 drops argue as in the non - reduced case : regluing of the non - parasitic part and application of the induction hypothesis shows the existence of @xmath1223 for @xmath1394 with the requested properties , provided @xmath1395 .
otherwise there exists @xmath1224 as in ( 2 ) ( with slightly smaller @xmath216 ) . in the nodal case
the existence of @xmath1223 for @xmath1394 follows from the deformation theory of nodal curves theorem [ smoothing nodal curves ] .
note that this line of reasoning fails in the non - reduced case , because in the proof of closedness @xmath1316 may end up unevenly distributed on @xmath1396 if @xmath1397 is reducible near a multiple component of @xmath473 .
a smoothing of @xmath1396 preserving the incidence conditions may then not exist .
8) _ taking the limit . _
+ assume first that either @xmath473 is non - reduced or that lemma [ dichotomy],2 applies .
let @xmath1215 be the sequence of deformations of @xmath474 constructed in this lemma or in lemma [ c_n ] in step 6 respectively . by the gromov compactness theorem we may assume convergence @xmath1398 by construction @xmath1399 , hence @xmath1400 .
because also @xmath1401 , proposition [ no equisingular deformations of c_infty ] implies @xmath1402 at the expense of going over to a deformation @xmath1403 of @xmath838 as allowed by proposition [ unobstructedness of cycle space ] replace the parasitic parts by the removed part of @xmath1220 , cf.step 4 .
take a @xmath1403-holomorphic smoothing @xmath1404 of this curve , existent by remark [ smoothing exists ] .
the induction hypothesis shows that @xmath1404 is symplectically isotopic to @xmath1215 with the parasitic part replaced , hence to @xmath1225 . in the remaining case
lemma [ dichotomy],1 there is a family of reglued curves @xmath1283 that is pseudo - holomorphic for a family of almost complex structures that are integrable in a fixed neighbourhood of @xmath472 and fiberwise agreeing with @xmath67 for @xmath1372 . in this case
the isotopy statement follows for @xmath467 sufficiently large and @xmath11 close to @xmath13 from the unobstructedness result proposition [ unobstructedness of cycle space ] . in the notation of the statement of the lemma this produces the desired isotopy between @xmath474 and a smoothing of @xmath473 under the additional hypothesis ( @xmath1185 )
this was left to be shown in step 1 .
the proof of lemma [ isotopy lemma ] is finished .
this section provides the proofs of the three main theorems stated in the introduction .
we shall use one more lemma .
[ degree lemma ] let @xmath62 be an almost complex manifold diffeomorphic to @xmath5 or to an @xmath3-bundle over @xmath3 .
assume that @xmath218 is a non - trivial irreducible pseudo - holomorphic curve with @xmath1405 .
then for all @xmath1406 @xmath1407 in either of the following cases .
1 . @xmath1408 and @xmath1409 with @xmath1410 , @xmath1411 a hyperplane .
2 . @xmath955 or @xmath58 and @xmath1412 with @xmath1413 , @xmath1414 , where @xmath268 and @xmath275 are a fiber and a section of the symplectic @xmath3-bundle @xmath1415 respectively with @xmath1416 .
@xmath940 and @xmath949 with @xmath1417 , notations as in ( ii ) . since @xmath1000 it suffices to establish ( [ degree estimate ] ) for @xmath1418 . for ( i )
the genus formula ( [ genus formula ] ) yields @xmath1419 inequality ( [ degree estimate ] ) thus follows from @xmath1420 this is true for @xmath1421 . for @xmath955
it holds @xmath1422 and the genus formula reads @xmath1423 so ( [ degree estimate ] ) will follow from @xmath1424 which is equivalent to @xmath1425 this is true for any @xmath1426 as long as @xmath1413 and for @xmath1427 , @xmath1428 . finally the case @xmath940
. then @xmath1429 , @xmath1430 , @xmath1431 and ( [ degree estimate ] ) follows from @xmath1432 or @xmath1433 for @xmath1414 and @xmath1428 the first term is non - positive and @xmath1434 so the result follows . for @xmath1435 , @xmath1436 we obtain @xmath1437 if @xmath1438 but @xmath1439 inequality ( [ inequality f_1 ] ) implies as sufficient condition @xmath1440 this inequality holds true for all @xmath219 .
the proof is finished .
we gave ( i ) in the lemma only to illustrate the origin of the degree restriction @xmath31 for @xmath5 .
for technical reasons we have to do the computation on the @xmath103-bundle @xmath58 , with some additional properties covered by case ( iii ) in the lemma .
_ proof of theorem b. _ let @xmath99 be the symplectic form on @xmath8 . denote by @xmath23 the symplectic submanifold . by proposition [ fibered j ] there
exists a map @xmath1441 and an @xmath99-tamed almost complex structure @xmath67 on @xmath8 making @xmath18 a pseudo - holomorphic map and @xmath25 a @xmath67-holomorphic curve .
let @xmath1442 be two disjoint sections and a fiber of @xmath18 respectively , with @xmath1416 .
according to remark [ wlog sigma>0 ] we may assume @xmath1443 .
then deformations of @xmath25 as pseudo - holomorphic curve are unobstructed by the smooth case of theorem [ smoothing nodal curves ] .
possibly after deforming @xmath25 slightly we may therefore assume @xmath1444 for the baire subset @xmath934 introduced in section [ generic paths ] .
since @xmath99 tames also the integrable complex structure @xmath818 on @xmath8 , lemma [ genericity c_1 ] , theorem [ genericity seva ] and proposition [ very generic j ] imply the existence of a path @xmath1445}$ ] of @xmath99-tamed almost complex structures having the following properties . 1 .
@xmath1446 , @xmath1447 and @xmath1448 is a symplectic pseudo - holomorphic @xmath3-bundle for every @xmath11 .
2 . @xmath1449 is monotone for every @xmath11 .
the path @xmath1445}$ ] is generic in the sense of theorem [ genericity seva],2 .
any @xmath989-holomorphic curve @xmath218 , @xmath1450 is homologous to @xmath286 with @xmath1451 , @xmath562 .
consider the set @xmath538 of @xmath1452 $ ] so that for every @xmath1453 a smooth @xmath989-holomorphic curve @xmath1454 exists that is isotopic to @xmath25 through an isotopy of @xmath99-symplectic submanifolds . by definition @xmath538
is an interval .
it is open as subset of @xmath1125 $ ] by the smooth case of theorem [ smoothing nodal curves ] .
it remains to show it is closed .
let @xmath1455 , @xmath1456 . for each @xmath467 choose a @xmath1457-holomorphic curve @xmath474 symplectically isotopic to @xmath25 .
by the gromov compactness theorem we may assume that the @xmath474 converge to a @xmath1458-holomorphic cycle @xmath1459 .
now @xmath1460 , @xmath464 fulfill the hypotheses of lemma [ isotopy lemma ] .
condition ( @xmath1288 ) follows from lemma [ degree lemma],ii .
the conclusion of lemma [ isotopy lemma ] in conjunction with remark [ smoothing exists ] now shows that @xmath1461 . hence @xmath1462 $ ] .
( end of proof of theorem b. ) _ _ proof of theorem c.__the proof is essentially the same as that of theorem b. as we are not in a fibered situation we proceed as follows .
let @xmath1463 be the blowing up in a point @xmath1464 .
choose a khler form @xmath1465 on @xmath58 that agrees with @xmath1466 away from @xmath1467 for @xmath1468 .
now run the program in the proof of theorem b to the @xmath1465-symplectic surface @xmath1469 . by remark [ generic j ]
we are free to take the @xmath989 and all intermediately occurring almost complex structures to agree with the standard integrable complex structure on @xmath1467 .
then @xmath989 descends to an @xmath99-tamed almost complex structure @xmath1470 on @xmath5 .
if @xmath1471 is a smooth @xmath989-holomorphic curve homologous to @xmath1469 then @xmath1471 is disjoint from @xmath1472 for homological reasons .
thus @xmath1473 is a smooth @xmath1470-holomorphic curve , hence symplectic with respect to @xmath99 .
it remains to verify assumption ( @xmath1288 ) in lemma [ isotopy lemma ] .
let @xmath1474 be a @xmath989-holomorphic cycle homologous to @xmath1475 , where @xmath275 is a section with @xmath1429 .
by lemma [ genericity c_1],1 there are no @xmath989-holomorphic curves homologous to @xmath1476 with @xmath1477 except @xmath1478 . therefore , if @xmath1479 with @xmath1480 for @xmath1481 , and @xmath1482 then @xmath1483 hence @xmath1484 .
now if @xmath31 and @xmath325 has a multiple component then @xmath1485 for all @xmath138 .
assumption ( @xmath1288 ) therefore follows from lemma [ degree lemma],iii .
( end of proof of theorem c. ) _ proof of theorem a. _ the main theorem of @xcite implies that @xmath1486 factors over a degree 2 cover of an @xmath3-bundle @xmath1487 , branched along a symplectic surface @xmath1488 with @xmath1489 a simply branched cover of @xmath3 .
the assumption on the monodromy means that @xmath108 is connected . by theorem
b there exists a symplectic isotopy @xmath1490 connecting @xmath1491 to a holomorphic curve @xmath1492 with respect to the integrable complex structure
. apply proposition [ fibered j ] with @xmath1493 .
the result is a family @xmath989 of almost complex structures and a family of @xmath3-bundles @xmath1494 with perturbing @xmath1490 slightly using the smooth case of theorem [ smoothing nodal curves ] we can achieve that @xmath1496 has only simple branch points for every @xmath11 .
let @xmath1497 be the two - fold cover of @xmath482 branched along @xmath1490 .
then @xmath1498 is a genus-@xmath14 symplectic lefschetz fibration for every @xmath11 .
this shows that @xmath1486 is even isotopic to a holomorphic lefschetz fibration .
( end of proof of theorem a. ) xxxxxx j .- f .
barraud : _ nodal symplectic spheres in @xmath5 with positive self - intersection _ , intern .
res . not .
* 9 * ( 1999 ) 495508 s.k .
donaldson : _ lefschetz pencils on symplectic manifolds _ , preprint 1999 m. gromov : _ pseudo - holomorphic curves in symplectic manifolds _ , inv . math .
* 82 * ( 1985 ) , 307347 h. hofer , v. lizan , j .- c .
sikorav : _ on genericity of holomorphic curves in 4-dimensional almost - complex manifolds _ , j. geom .
7 * ( 1998 ) , 149159 f. lalonde , d. mcduff : _ @xmath67-curves and the classification of rational and ruled symplectic @xmath12-manifolds _ , in : contact and symplectic geometry , 342 , publ .
newton inst . 8 , cambridge univ . press 1996 j. li , g. tian : _ virtual moduli cycles and gromov - witten invariants of general symplectic manifolds _ , in : _ proceedings of the first ip conference at uc , irvine _ , r. stern ( ed . ) d. mcduff : _ the structure of rational and ruled symplectic four - manifolds _ , journal amer
. math .
soc . * 3 * ( 1990 ) , 679712 m. micallef , b. white : _ the structure of branch points in minimal surfaces and in pseudoholomorphic curves _ , ann . of math . * 141 * ( 1995 ) , 3585 b. moishezon : _ complex surfaces and connected sums of complex projective planes _ , lecture notes in math .
* 603 * , springer 1977 b. ozbai , a. stipsicz : _ noncomplex smooth 4-manifolds with genus-2 lefschetz fibrations _ ,
preprint math.gt/9810007 v. shevchishin : _ pseudoholomorphic curves and the symplectic isotopy problem _ , preprint math.sg/0010262 , 95 pp .
b. siebert : _ gromov - witten invariants for general symplectic manifolds _ , preprint dg - ga 9608005 b. siebert , g. tian : _ on hyperelliptic symplectic lefschetz fibrations of four manifolds _ , communications cont .
* 1 * ( 1999 ) , 255280 b. siebert , g. tian : _ weierstra polynomials and plane pseudo - holomorphic curves _ , preprint 2000 , 9 pp ; to appear in : chinese ann . of math .
b * 23 * ( 2002 ) j .- c .
sikorav : _ the gluing construction for normally generic @xmath67-holomorphic curves _ , preprint ens lyon 2000 , 24 pp . c. h. taubes : _
sw @xmath1499 gr : from the seiberg - witten equation to pseudo - holomorphic curves _ , j. amer .
* 9 * ( 1996 ) , 854918 i. n. vekua : _ generalized analytic functions _ , pergamon 1962 | we prove that any genus-2 lefschetz fibration without reducible fibers and with `` transitive monodromy '' is holomorphic .
the latter condition comprises all cases where the number of singular fibers @xmath0 is not congruent to @xmath1 modulo @xmath2 .
an auxiliary statement of independent interest is the holomorphicity of symplectic surfaces in @xmath3-bundles over @xmath3 , of relative degree @xmath4 over the base , and of symplectic surfaces in @xmath5 of degree @xmath6 . supported by the heisenberg program of the dfg ] |
as a part of our continuing studies of the magnetic , electronic and transport properties of ternary rare - earth ( r ) intermetallic compounds of the type r@xmath0t@xmath1x@xmath2 where t is a transition metal and x is an s - p element , we have recently focused our attention to the search for new yb based compounds of the 235 type with interesting physical properties .
in fact there are very few yb based compounds of the type r@xmath0t@xmath1x@xmath2 known due to the difficulty in preparation and characterization of these compounds .
we had recently @xcite synthesized high quality single phase samples of the compound yb@xmath0fe@xmath1si@xmath2 and established it to be a kondo - lattice compound with heavy fermion behavior ( among the highest @xmath7 values = .5 j / yb @xmath8 , for yb compounds ) and antiferromagnetic ordering at low temperatures .
pressure studies are in progress to look for superconductivity in this compound given that both the tm and lu compounds of the same series show superconductivity at low temperatures @xcite .
hence it is of interest to search for new yb compounds with a similar structure because one can expect to see similar interesting physical properties .
there have been some investigations on the compound ce@xmath0ir@xmath1ge@xmath2 @xcite which forms in u@xmath0co@xmath1si@xmath2 structure , a structure closely related to the yb@xmath0fe@xmath1si@xmath2 structure .
it has been shown to be a kondo - lattice system with antiferromagnetic ordering and moderate heavy fermion behavior .
the electron - hole analogy between ce and yb makes the compound yb@xmath0ir@xmath1ge@xmath2 an interesting candidate to investigate . in this paper
we report a comprehensive study of the structure , electrical resistivity , magnetic susceptibility and heat capacity of the new compound yb@xmath0ir@xmath1ge@xmath2 .
the susceptibility measurements indicate a mixed valent nature for the yb ions while the low temperature resistivity follows a typical kondo lattice behavior and the heat capacity data shows an upturn below 5 k giving a hint that the compound may order magnetically below the lowest temperature of measurement of 0.4 k. such unusual properties have been reported earlier @xcite only for certain ce compounds of the form ce@xmath0t@xmath1x@xmath9 where t = rh , ir with x = al , ga . we find that yb@xmath0ir@xmath1ge@xmath2 compound also shows such unusual properties where the magnetic response shows a typical mixed valent behavior while the low temperature resistivity and heat capacity data indicate heavy fermion or kondo - lattice and magnetic ordering type of response .
yb@xmath0ir@xmath1ge@xmath2 forms in the pmmn structure which allows the possibility of 2 sites for yb ion . in this compound
it is possible for yb to have different valency in different sites .
the paper is organised as follows .
we first describe the details of sample preparation and characterisation and the measurement techniques in the section on experimental details .
we then describe one by one the results and analysis of the magnetic susceptibility data , the resistivity data and the heat capacity data in subsequent sections .
we then discuss the combined picture provided by all the measurements in the section on discussion . in the final section on conclusion
we summarise all the results for the sample .
a polycrystalline sample of yb@xmath0ir@xmath1ge@xmath2 was prepared by arc melting .
first the binary ir@xmath1ge@xmath2 was prepared by arc melting the constituent elements taken in stoichiometric proportion on a water - cooled hearth in high purity argon atmosphere .
the resultant button was flipped over and remelted 6 times to ensure homogenous mixing .
the weight loss in this stage was negligible .
the binary was crushed into a fine powder and small pieces of yb ( with 5% extra yb ) were mixed with it .
the whole material was pelletized by a press into a closely packed tablet .
this tablet was then arc melted and remelted only four times to make a compromise between yb loss and homogenous mixing .
we found that by pressing the yb along with already prepared binary powder we could reduce the yb evaporation considerably .
no attempts to anneal the sample were made .
a small piece of the as cast sample was used for powder x - ray diffraction .
the crystal structure of the given sample was refined using * full prof * @xcite and it is shown in fig . 1 .
the slight difference in the intensity profile is probably associated with preferred orientation effects , which have not been taken into account in the present analysis .
the refinement yielded the following values for the lattice constants : a = 18.9474 , b = 15.4697 and c = 4.5960 the temperature dependence of the dc magnetic susceptibility ( @xmath10 ) was measured using the commercial squid magnetometer ( mpms5 , quantum design , usa ) in a field of 1 koe in the temperature range from 1.8 to 300 k. the resistivity was measured using a four - probe dc technique with contacts made using silver paint on a bar shaped sample of 1 mm thick , 10 mm length and 2 mm width cut from the arc melted ingot .
the temperature was measured using a calibrated si diode ( lake shore inc .
, usa ) sensor .
the heat - capacity in zero field between 0.4 to 35 k was measured using an automated adiabatic heat pulse method . a calibrated germanium resistance thermometer ( lake shore inc , usa )
was used as the temperature sensor in this range .
the temperature dependence of the inverse dc magnetic susceptibility ( 1/@xmath11 ) of polycrystalline yb@xmath0ir@xmath1ge@xmath2 in a field of 1 koe from 1.8 to 300 k is shown in the fig . 2 .
with the inset showing the low temperature @xmath10(t ) data .
the high temperature susceptibility ( 100 k@xmath12t@xmath12300 k ) was fitted to a modified curie - weiss expression which is given by , @xmath13 here , @xmath14 is the temperature independent susceptibility , including the core - electron diamagnetism and the pauli paramagnetism and the van - vleck terms .
c is the curie constant and @xmath15 is the curie - weiss temperature .
the effective moment can be calculated from the value of the curie constant from the relation , @xmath16 where @xmath17 is the concentration of yb ions ( x = 2 for yb@xmath0ir@xmath1ge@xmath2 ) . the value which we get for @xmath10@xmath18 is ( 1.65@xmath1910@xmath20 emu / mol ) which is quite small .
the value of the effective magnetic moment ( @xmath21 ) was found to be 3.66 @xmath22 and the value for the curie - weiss temperature @xmath23 was found to be @xmath2415.19 k. the estimated effective moment of only 3.66@xmath22 is much smaller than the value 4.54 @xmath22 for magnetic yb@xmath25 ions .
this indicates that yb may possibly be of mixed valent nature in this compound
. however , usually for mixed valent compounds , one finds large and positive values of @xmath23 whereas we find a small and negative value = -15.19 k for @xmath23 which is typical of compounds which undergo antiferromagnetic ordering at low temperatures .
it is possible that the susceptibility would follow a curie - weiss law with an effective moment closer to the theoretical yb@xmath25 value only at much higher temperatures above 300 k and crystalline electric fields are strong enough even at temperatures of the order of 100 - 300 k. we have also analyzed the data after subtracting the susceptibility of the pauli paramagnetic lu@xmath0ir@xmath1ge@xmath2 sample .
the value of either @xmath26 or @xmath23 did not change significantly .
it is also noteworthy that we do not observe any broad hump in the susceptibility typical of mixed valent systems .
there is a strong deviation of the data from the curie - weiss law below 100 k. this could be due to the combined result of the kondo effect and crystalline electric fields .
such a conjecture is substantiated from the resistivity and heat - capacity studies described below .
we do not observe any indication of magnetic ordering down to 1.8 k from the magnetization measurements .
the resistivity ( @xmath27 ) vs temperature data of yb@xmath0ir@xmath1ge@xmath2 from 0.4 k to 300 k is shown in fig .
the room temperature resistivity value is about 200@xmath3@xmath28-cm which is quite common among ternary rare earth silicides and germanides .
the compound shows metallic conductivity with some negative curvature of @xmath27(t ) below about 150 k , which may result from crystal field interactions and/or s - d interband scattering of conduction electrons . at about 10
k the resistivity goes through a broad shallow minimum and then slightly increases towards lower temperatures ( see the inset of fig .
however , below 2 k , the resistivity starts to decrease again as in the case of typical kondo lattice compounds on the onset of coherent scattering .
the latter hypothesis seems supported by the logarithmic character of the resistivity rise below 8 k ( inset of fig .
4 ) . as is apparent from fig .
4 , the temperature variation of the resistivity of yb@xmath0ir@xmath1ge@xmath2 in the temperature range 2 k to 300 k can be well described by the formula , @xmath29 which comprises the residual scattering of conduction electrons on defects , impurities and spin - disorder ( @xmath30+@xmath31 ) , the bloch - grneisen scattering on phonons ( third term ) , the mott s - d scattering ( fourth term ) and the kondo scattering ( fifth term ) .
the least squares fitting procedure yields the following values of the parameters : @xmath30+@xmath31 = 126.9 @xmath3@xmath28 cm , r = 0.29 @xmath3@xmath28 cm / k , @xmath32 = 110.5 k , k = 4.8@xmath1910@xmath33@xmath3@xmath28 cm / k@xmath34 and c@xmath35 = 0.32 @xmath3@xmath28 cm .
the non - negligible value of c@xmath35 suggests that a weak kondo effect is operative at low temperature .
the data could be fitted only up to 2 k below which the coherence sets in .
the heat capacity c@xmath36 data of yb@xmath0ir@xmath1ge@xmath2 in the temperature range 0.4 k to 35 k is shown in the top panel a of fig .
4 . in this figure
we have plotted the total heat capacity @xmath37 , the heat capacity of iso - structural lu@xmath0ir@xmath1ge@xmath2 to estimate the lattice contribution @xmath38 to the total heat capacity and the magnetic heat capacity @xmath39 calculated by subtracting the heat capacity of lu@xmath0ir@xmath1ge@xmath2 from the total heat capacity .
we have taken care of the mass difference between yb and lu while estimating the lattice contribution for the yb sample .
the low temperature heat capacity shows an upturn below 6 k and starts increasing as we go down in temperature .
panel b shows the data plotted as @xmath37/ t vs @xmath40 .
this plot emphasizes the sharp upturn in the data at low temperatures .
the magnitude and sharpness of the upturn as seen in the inset panel i of panel b strongly indicates a possible magnetic ordering on going further down in temperature .
the inset ii of the panel b shows the temperature dependence of the estimated magnetic entropy s. the entropy reaches a value of about .27 rln(8 ) at 35 k which is much lower than the value @xmath41ln(2@xmath42 + 1 ) with @xmath42 = 7/2 for yb .
thus the total entropy has not been released even upto a temperature of 35 k. this indicates a significant contribution from crystal field effects . the magnetic contribution to the heat capacity @xmath39 plotted in the panel
a shows a broad hump peaked around 27 k indicating the presence of a schottky anomaly .
the schottky type anomaly at higher temperatures and the upturn in heat capacity at low temperatures makes it impossible to fit the data in any temperature range to the equation @xmath7 t + @xmath43@xmath44 to get a measure of the sommerfeld constant @xmath7 .
let us now try to explain why the combined data of all the measurements described above is unusual before we give arguments , which attempt to understand such a behavior .
the nature of the ground state ( gs ) in such systems where an array of 4f moments is embedded in a metallic environment results from a competition between two energy scales .
the onsite kondo effect ( favoring a non - magnetic gs ) with a characteristic energy given by the kondo temperature t@xmath35 and the intersite rkky ( characterized by t@xmath45)interaction favoring a magnetic gs .
the gs physics depends on the final balance between these two mechanisms which is driven by the strength of the hybridization between the localized f electrons and the itinerant conduction electrons . in general t@xmath35 values
are smaller for heavy fermion compounds which display kondo lattice behavior with a large @xmath7 and a fall in @xmath27 due to coherent scattering .
the effective moment of the 4f ions ( estimated from the high temperature @xmath10 data ) will be nearly equal to its theoretical value .
however , one can see from our data that the curie - weiss fit to the high temperature inverse susceptibility gives an effective moment considerably reduced from the yb@xmath25 value suggesting that yb may be in a mixed valent state in this compound whereas the low temperature transport and heat capacity measurements show a kondo lattice type of behavior with strong indications of a possible magnetic ordering just below 0.4 k. although the calculated effective paramagnetic moment is quite low , the overall susceptibility behavior looks more like that of a system which is largely trivalent rather than one which is mixed - valent [ e.g. , compare the magnetic susceptibility of cenige@xmath0 or cenisn@xmath0 ( fig.6 of @xcite ) in which ce is in the 3@xmath46 state with that of cenisi@xmath0 ( fig.4 of @xcite ) in which ce is mixed - valent ] .
one can try to explain the behavior of yb@xmath0ir@xmath1ge@xmath2 if we realize that it forms in the pmmn structure which allows yb to have two inequivalent sites .
hence it is possible for yb ions to have a different valency at the two different sites .
we believe that this is infact the case and yb on one of the sites has a valency close to the 2@xmath47 while at the other site it has a valency of 3@xmath47 .
in such a case the non - magnetic yb@xmath48 ions would exhibit pauli paramagnetism while the magnetic yb@xmath25 ions would follow a curie - weiss kind of behavior . thus one would see a reduced effective moment depending on the relative occupancies of the two different sites .
the yb@xmath25 ions would also be the ones participating in the kondo effect and may undergo magnetic ordering at low temperature .
one needs to have single crystal x - ray data to calculate the exact positions and occupation factors of yb and other elements in the unit cell to get a clearer understanding of the underlying physics .
to conclude , we have synthesized a new yb based compound namely yb@xmath0ir@xmath1ge@xmath2 . which forms in the orthorhombic structure with space group pmmn unlike its ce analogue which forms in the tetragonal u@xmath0co@xmath1si@xmath2 type structure .
iso - structural lu@xmath0ir@xmath1ge@xmath2 was also synthesized to estimate the lattice contribution to the heat capacity .
it is also found to take the same structure as the yb compound .
a curie - weiss fit to the high temperature ( 100 - 300 k ) inverse magnetic susceptibility data gives an effective moment of 3.66@xmath22 which is far less than the value 4.54@xmath22 for magnetic yb@xmath25 ions indicating that yb may be in a mixed valent state in this compound .
however , the structure pmmn can have two sites for yb ions and the valency of yb may be different from 3 in one of the sites thus giving a net reduced effective paramagnetic moment .
such behavior has been observed in some cerium based compounds with multiple ce sites @xcite .
this is the first report to our knowledge on an yb based compound showing such behavior .
our resistivity measurements show that yb@xmath0ir@xmath1ge@xmath2 behaves like a kondo - lattice below 2 k. the heat capacity data shows signs of a strong cef influence with a schottky type anomaly around 27 k. finally we could not detect any magnetic ordering in the compound down to 0.4 k though the low temperature heat capacity indicates that the compound may possibly undergo magnetic ordering below 0.4 k. yogesh singh , s. ramakrishnan , z. hossain and c. geibel , phys .
b * 66 * , 014415 ( 2002 ) .
braun , phys . letters * 75a * , 386 ( 1980 ) .
braun , c.u .
segre , f. acker , m. rosenberg , s. dey , u. deppe , j. magn .
magn . mater .
* 25 * , 117 ( 1981 ) .
moodenbaugh , d.e .
cox , and h.f .
braun , phys .
b * 25 * , 4702 ( 1981 ) .
vining and r.n .
shelton , phys .
b * 28 * , 2732 ( 1983 ) .
g. venturini , m.meot-meyer , j.f .
marcche , b. malaman and b. roques , mater .
bull . * 21*,33 ( 1986 ) . c. godart , c. v. tomy , l. c. gupta and r. vijayaraghavan , solid state commun . *
67 * , 677 ( 1988 ) .
z. hossain , h. ohmoto , k. umeo , f. iga , t. suzuki , t. takabatake , n. takamoto and k. kindo , phys .
b , * 60 * , 10383 ( 1999 )
. b. buschinger , c.geibel , m. weiden , c. dietrich , g. cordier , g. olesch , j. koehler and f. steglich , j. alloys compd .
* 260 * , 44 ( 1997 ) .
b. buschinger , o.trovarelli , m.weiden , c.geibel and f. steglich , j. alloys compd * 275 - 277 * , 633 ( 1998 ) .
x - ray powder diffraction program available at collaborative computational project number 14 ( ccp14 ) ( www.ccp14.ac.uk ) .
pecharsky , k. a. gschneidner , jr . , and l. l. miller , phys .
b*43 * , 10906 ( 1991 ) .
k. ghosh , s. ramakrishnan , s. k. dhar , s. k. malik , girish chandra , k. a. gschneidner , jr .
, v. k. pecharsky , z. hu and w. b. yelon phys . rev . b*52 * , 7267 ( 1995 ) and references cited therein . | we report sample preparation , structure , electrical resistivity , magnetic susceptibility and heat capacity studies of a new compound yb@xmath0ir@xmath1ge@xmath2 .
we find that this compound crystallizes in an orthorhombic structure with a space group pmmn unlike the compound ce@xmath0ir@xmath1ge@xmath2 which crystallizes in the tetragonal ibam ( u@xmath0co@xmath1si@xmath2 type ) structure .
our resistivity measurements indicate that the compound yb@xmath0ir@xmath1ge@xmath2 behaves like a typical kondo lattice system with no ordering down to 0.4 k. however , a curie - weiss fit of the inverse magnetic susceptibility above 100 k gives an effective moment of only 3.66 @xmath3@xmath4 which is considerably less than the theoretical value of 4.54 @xmath3@xmath4 for magnetic yb@xmath5 ions .
the value of @xmath6 = -15.19 k is also considerably higher indicating the presence of strong hybridization .
an upturn in the low temperature heat capacity gives an indication that the system may order magnetically just below the lowest temperature of our heat capacity measurements ( 0.4 k ) .
the structure contains two sites for yb ions and the present investigation suggests that yb may be trivalent in one site while it may be significantly lower ( close to divalent ) in the other .
1truecm ms number pacs number : 72.10.fk , 72.15.qm , 75.20.hr , 75.30.mb + |
the arguments of this paper apply to the einstein - maxwell - scalar field system under spherical symmetry . for an introduction to this system
, the author can consult @xcite .
the einstein - scalar field system , whose rigorous study was initiated in @xcite , is of course a special case , where the maxwell part vanishes .
the results of this paper also apply in the case where the wave equation is decoupled from the einstein equations . for more on the relationship between these two problems ,
see @xcite .
recall that the quotient manifold @xmath7 inherits a @xmath8-dimensional lorentzian metric , which in null coordinates takes the form @xmath9 .
the metric of @xmath10 can then be written @xmath11 , where @xmath12 is the standard metric on @xmath13 , and @xmath14 is the area radius function @xmath15 defined by @xmath16 .
we recall the hawking mass @xmath17 , and the mass aspect function @xmath18 . under spherical symmetry , the maxwell equations decouple , contributing a @xmath19 term to the energy momentum tensor , where @xmath20 is a constant . defining the `` renormalized '' hawking mass @xmath21 , the evolution of the metric and scalar field
are then completely determined by the following sytem of equations : @xmath22 @xmath23 @xmath24 @xmath25 @xmath26 @xmath27 @xmath28 @xmath29 we also easily obtain from the above the equations @xmath30 and @xmath31
we will assume the existence of a spacetime bounded by an ingoing null curve , a timelike curve , and an outgoing null curve , which we will call the _ event horizon_. at @xmath32 , even though @xmath33 is part of the spacetime . ] certain conditions will be imposed on the initial ingoing null curve and the timelike boundary . in this paper , we will not concern ourselves with the explicit _ construction _ of solutions satisfying the conditions described here , although this can be accomplished easily enough using the methods of @xcite .
we consider a spacetime @xmath34 with penrose diagram depicted below , where @xmath35 is a timelike boundary : @xmath36 here , @xmath37 , but @xmath38 .
the point of intersection of the initial ingoing ray and @xmath35 will be @xmath39
. the ingoing ray will then be the curve @xmath40 .
the equations @xmath41@xmath42 are assumed to hold pointwise . on @xmath35 , @xmath43 , @xmath6 for some constant @xmath4 , and @xmath44 will be non - increasing as a function of @xmath1 , such that @xmath45 .
moreover , we assume that @xmath1 ranges in all of @xmath46 . on the ingoing null segment @xmath40 ,
we assume @xmath47 , @xmath48 bounded , @xmath49 .
this makes @xmath50 for some @xmath51 .
we will assume @xmath52 at @xmath53 .
we first show under the above assumptions , @xmath54 in @xmath34 , and @xmath55 in @xmath56 .
_ let @xmath57 be the region @xmath58 we first note that @xmath59 is continuous .
for @xmath59 is assumed to be differentiable in the @xmath1 direction , while differentiating @xmath60 in the @xmath61 direction one obtains an evolution equation for @xmath62 , which initially is @xmath63 . by continuity of @xmath59
, @xmath57 is an open subset of @xmath34 . in @xmath64 , we clearly have @xmath65 , and @xmath66 in @xmath67 .
thus , it follows that @xmath55 in @xmath68 . for
, if @xmath69 , and @xmath70 , then @xmath71 for all @xmath72 , and thus the curve @xmath73 can not intersect @xmath35 .. see @xcite ] thus , by @xmath74 , we have @xmath75 in @xmath76 , and so , denoting @xmath77}\varpi(u,1)$ ] we have @xmath78 in @xmath79 . integrating @xmath80 from @xmath40 ,
it is clear that @xmath81 . from @xmath82
, we have @xmath83 integrating @xmath60 , it follows that @xmath84 on @xmath79 .
thus @xmath85 .
this completes the proof .
we note now that from the above proposition and the results of @xcite , it follows that @xmath87 , @xmath88 in @xmath34 , and thus , since @xmath89 , we have @xmath52 throughout @xmath34 .
this immediately yields @xmath90 where we use the fact that @xmath91 and @xmath92 have the same sign to obtain the lower bound .
we will show the following the above two propositions and the archimedean property of the real numbers immediately yield the theorem .
_ proof of proposition 2_. we note that proposition 1 gives all the information necessary to apply section 5 of @xcite .
the propositions of that section immediately yield the result of this proposition . in the process , one obtains a lower bound @xmath102 which we will require below .
_ proof of proposition 3_. consider a dyadic decomposition of the event horizon , with respect to the @xmath1 coordinate , as in @xcite .
@xmath103 in each dyadic interval , we can select , again as in @xcite , a subinterval @xmath104 , split into two pieces as shown below , each of length @xmath105 , @xmath106 such that @xmath107 here @xmath1 denotes the @xmath1-value of any point in the dyadic interval .
( recall that these values are all comparable to the length of the dyadic interval . ) we note that the assumptions of propositions of section 6 of @xcite hold .
we thus obtain the bound @xmath108 on the interval @xmath109 .
it follows immediately , in view of the bounds @xmath110 , and @xmath111 that , @xmath112 now , by the inequalities @xmath75 , @xmath113 , the fact that @xmath44 is nonincreasing to the future along @xmath35 , and the equation @xmath74 , this implies immediately @xmath114 along the event horizon . by the dyadic nature of the decomposition
, @xmath115 holds for all values of @xmath1 , and not just for the @xmath1-values of the future endpoints of @xmath104 .
finally , by @xmath116 , this proves immediately @xmath117 . applying again the results of section 6 of @xcite ,
now with _ any _ subinterval of the event horizon of length @xmath118 , in view of @xmath117 , we obtain that the inequality @xmath119 holds throughout @xmath34 , in particular , on @xmath35 .
the boundary conditions , namely @xmath43 , @xmath120 , @xmath44 is non - increasing on @xmath35 , yield that @xmath121 on @xmath35 , and thus by @xmath122 we have @xmath123 on @xmath35 .
integrating now the equation @xmath124 from @xmath35 , in view of the bounds @xmath111 , @xmath116 , and @xmath122 , we obtain @xmath125 throughout @xmath34 .
this concludes the proof . | in recent work of allen et .
al . , heuristic and numerical arguments were put forth to suggest that boundary value problems for black hole evolution , where an appropriate sommerfeld radiation condition is imposed , would probably fail to produce price law tails .
the interest in this issue lies in its possible implications for numerical relativity , where black hole evolution is typically studied in terms of such boundary formulations . in this note
, it is shown rigorously that indeed , price law tails do not arise in this case , i.e. that sommerfeld ( and more general ) radiation conditions lead to decay faster than any polynomial power .
our setting is the collapse of a spherically symmetric self - gravitating scalar field .
we allow an additional gravitationally coupled maxwell field .
the proof also applies to the easier problem of a spherically symmetric solution of the wave equation on a schwarzschild or reissner - nordstrm background .
the method relies on previous work of the authors .
one of the characteristic features of black holes forming in the context of gravitational collapse is the appearance of so - called _ price law tails_. these describe the decay of radiation on timelike curves in the exterior region of the black hole , as well as the decay of the radiation flux along the event horizon . in the context of radiation described by a scalar field , price @xcite
put forth heuristic arguments indicating that this decay should be @xmath0 , where @xmath1 is an eddington - finkelstein - like advanced time coordinate . in @xcite , it was rigorously proven in the context of the coupled einstein - scalar field equations under spherical symmetry ( and more generally of the einstein - maxwell - scalar field equations ) that @xmath2 , for any @xmath3 and sufficiently large @xmath4 .
the decay rates @xmath5 also hold for spherically symmetric solutions of the wave equation on a fixed schwarzschild or reissner - nordstrm background . in the present paper
, it is proven that if appropriate boundary conditions are imposed at @xmath6 , for fixed @xmath4 , then decay can be proven at rates faster than any given polynomial power .
the conditions include standard sommerfeld boundary conditions .
this paper was motivated by @xcite , where the absense of price law tails was suggested on the basis of heuristic arguments and numerical calculations .
in particular , this paper confirms the validity of the heuristics of @xcite .
in fact , the argument of this paper is a rather straightforward application of the method of @xcite . in that paper ,
decay rates were achieved by an inductive argument , at each step of which , the rate was improved by an amount constrained by information coming from infinity . as we shall see below , with appropriate boundary conditions
, there is no such constraint , and the decay rate can be improved at each step by a fixed finite polynomial power .
thus , the inductive argument yields that any polynomial power can be achieved . |
let @xmath7 be a measurable space .
we equip it with an assumption that will be explained when required .
let @xmath8 be any transition probability on @xmath7 .
an homogeneous markov process @xmath9 is naturally associated to @xmath10 . in the target problem , we are interested in the probabilities of reaching the target class @xmath4 within @xmath11 steps , namely in @xmath12 the set @xmath4 is a priori given and does not change through the computations .
let @xmath7 be a measurable space and let @xmath13 .
let @xmath14 be a sub @xmath3-algebra of @xmath15 such that @xmath16 .
a function @xmath17 $ ] is a _ transition probability on @xmath18 _ if * @xmath19 is a probability measure on @xmath20 , for any @xmath21 ; * @xmath22 is @xmath20-measurable , for any @xmath23 .
given a transition probability @xmath8 on @xmath18 , we denote by @xmath24 the transition probability on @xmath18 given inductively by @xmath25 we denote by @xmath26 the set of the transition probabilities on @xmath18 .
we denote by @xmath27 the set of all transition probabilities on @xmath28 , that we equip with a suitable pseudometric @xmath29 @xmath30 it is such that @xmath31 this pseudo - metric is obviously compatible with the target @xmath4 and appears as a loss function of using @xmath32 instead of @xmath33 , whatever is the initial point @xmath34 ( @xmath35 is a discount rate ) .
the point will be , with no surprise , to let this distance go to @xmath36 .
a target problem is defined through a transition probability @xmath37 .
a _ target problem _ is a quadruple @xmath38 , where @xmath39 and @xmath40 .
a _ simple target problem _ is a target problem where @xmath20 is generated by an at most countable partition of @xmath28 .
the main purpose of this paper is to approximate any target problem with a sequence of simple target problems in the spirit of the construction the lebesgue integral , where the integral of a function @xmath41 is approximated by the integral of simple functions @xmath42 .
the lebesgue approximation requires at each step @xmath43 two choices : the choice of the subdivision @xmath44 and the choice of the function values @xmath45 on each subdivision .
we call _ strategy _ @xmath46 a sequence of maps @xmath47 from the set of the target problems to the set of the simple target problems .
in the `` lebesgue example '' given above , the strategy is related to the `` objective '' of the problem ( the integral ) and the pseudometric @xmath48 is required to go to @xmath36 as @xmath11 goes to infinity . here
also , a strategy is meaningful if @xmath49 tends to @xmath36 as @xmath11 goes to infinity .
moreover , for what concerns applications , given a target problem @xmath50 a good strategy should not need the computation of @xmath24 , @xmath51 .
the first main result of this paper is the existence of a class of good strategies , called _
target algorithms_. [ thm : mainta ] for any target problem @xmath52 and any target algorithm @xmath46 , @xmath53 where @xmath54 .
two questions immediately arise : does the sequence @xmath55 have a limit ( and in which sense ) ? moreover , since @xmath29 is defined as a pseudometric , does this limit depend on the choice of @xmath46 ?
the extension of the concept of compatible projection given in @xcite to our framework will enable us to understand better the answer to these questions .
a measurable set @xmath56 of a measurable space @xmath57 is an @xmath15-atom if it has no non - empty measurable proper subset .
no two distinct atoms intersect .
if the @xmath3-field is countably generated , say by the sequence @xmath58 then the atoms of @xmath15 are of the form @xmath59 where each @xmath60 is either @xmath61 or @xmath62 .
an equivalence relationship @xmath63 on a measurable set @xmath7 is _ measurable _ ( _ discrete _ ) if there exists a ( discrete ) random variable @xmath64 ( @xmath65 denotes the borel @xmath3-algebra ) , such that @xmath66 and we denote it by @xmath67 .
let @xmath38 be a target problem .
a _ compatible projection _ is a measurable equivalency @xmath68 such that @xmath69 a compatible projection @xmath63 is said to be _ optimal _ if @xmath70 , for any other compatible projection @xmath71 .
this definition is well posed if @xmath72 assumption ensures that the definition of measurable equivalency is indeed well posed .
this assumption will be stated and discussed in section [ sec : a0 ] .
[ thm : cp+tp ] if @xmath67 is a compatible projection for the target problem @xmath38 , then there exists a target problem @xmath73 .
such that @xmath74 for any @xmath75 .
it is not said `` a priori '' that an optimal compatible projection must exist . if it is the case , then this equivalence is obviously unique .
[ thm : maintb ] for any target problem @xmath52 , there exists a ( unique ) optimal compatible projection @xmath63 . to conclude the main results ,
let us first come back to the lebesgue example .
the simple function @xmath42 are chosen so that @xmath76 increases to @xmath77 and @xmath78 .
the following theorem guarantees these two facts by showing the `` convergence '' of any strategy to the optimal problem .
[ thm : maintc ] let @xmath79 , with @xmath46 target algorithm and let @xmath63 be the optimal compatible projection associated to the target problem @xmath38 .
then * @xmath80 for any @xmath11 , and @xmath81 ; * @xmath82 , for any @xmath83 .
in theorems [ thm : mainta][thm : maintc ] , we have proved the convergence of @xmath84 to @xmath85 with respect to the pseudometric @xmath29 .
the pseudometric topology @xmath86 is the topology induced by the open balls @xmath87 , which form a basis for the topology .
accordingly , the previous theorems may be reread in terms of convergence of @xmath88 to @xmath8 on the topological space @xmath89 . given a strategy @xmath90 , if we want to show a sort of weak convergence of @xmath91 to @xmath19 , for any @xmath34 , we face the two following problems * each @xmath91 is defined on a different domain ( namely , on @xmath92 ) ; * we did not have required a topology on @xmath28 . first , we want to introduce a new definition of probability convergence which takes into account the first restriction .
the idea is given in the following example .
[ exa : dyadic ] let @xmath93,i=0,\ldots,2^n-1\})$ ] be the @xmath3-algebra on @xmath94 $ ] generated by the dyadic subdivision .
suppose we know that @xmath95 $ ] is the unique probability on @xmath92 s.t . for any @xmath96 , @xmath97)=2^{-n}$ ] .
even if @xmath98 is not defined on the borel sets of @xmath94 $ ] , it is clear that in `` some '' sense , it must happen that @xmath99 , where @xmath100 is the lebesgue measure on the borel sets of @xmath94 $ ] .
note that the cumulative function of @xmath101 is not defined , and therefore a standard weak convergence can not be verified .
in fact , we know that @xmath102 \big ) = \nu_n \big ( \big ( 0 , \frac{i}{2^n } \big ] \big ) = \frac{i}{2^n},\ ] ] i.e. , in this case , as @xmath103 , we can determine the cumulative function in a dense subset .
this fact allows to hope that @xmath99 in a particular sense .
let @xmath104 be a filtered space , and set @xmath105 .
let @xmath106 $ ] , @xmath107 and @xmath108 $ ] be probability measures .
we say that @xmath98 _ converges totally to @xmath109 on the topological space @xmath110 _ as @xmath11 tends to infinity , if @xmath111 ( converges in weak sense on @xmath110 ) , for any @xmath112 $ ] , such that @xmath113 .
we write @xmath114 .
going back to the example , it is simple to check that @xmath115 } \nu_*$ ] , where @xmath116 are given in example [ exa : dyadic ] and @xmath117 $ ] is the standard topology on @xmath94 $ ] .
in fact , let @xmath118 be any extension of @xmath119 to the borel sets of @xmath94 $ ] . for any @xmath120 , we have by that @xmath121 where @xmath122 is the cumulative function of @xmath123 , which implies the weak convergence of @xmath123 to @xmath100 and , therefore , @xmath124 } \nu_*$ ] . for what concerns the topology on @xmath28 , we will define the topological space @xmath125 induced by the pseudometric @xmath126 associated to the target problem @xmath38 , and the pseudometric @xmath29 . in this way
@xmath127 is defined only with the data of the problem .
one may ask : is this topology too poor ?
the answer is no , since it is defined by the interesting pseudometric @xmath126 .
in fact , @xmath128 means that @xmath34 and @xmath129 play `` almost the same role '' with respect to @xmath4 . a direct algorithm which takes @xmath126 into account needs the computation of @xmath24 at each step . in any case , even if @xmath126 may not be computable , it defines a nontrivial interesting topology @xmath127 on @xmath28 .
as expected , we have the following theorem .
[ prop : top ] let @xmath79 , with @xmath46 target algorithm
. then @xmath130
in this section , we introduce the core of the approximating target problem , namely a set of strategies @xmath46 which solves the target problem . given a measurable space @xmath7 and a target problem @xmath52 , the target algorithm is built in the spirit of the exact one given in @xcite , which starts from the largest classes @xmath4 and @xmath131 and then reaches the optimal classes according to a backward construction .
the target algorithm defines a strategy @xmath132 , where @xmath133 and it consists of three steps : 1 .
the choice of a sequence @xmath134 of equivalences on the simplex on the unit ball of @xmath135 with @xmath136 ; 2 . the definition of a filtration @xmath137 based on @xmath134 where each @xmath92 is generated by a countable partition of @xmath28 ; 3 . the choice of a suitable measure @xmath138 and the definition of @xmath139 .
associated to each countably generated sub @xmath3-algebra @xmath140 , we define the equivalence relationship @xmath141 induced by the atoms of @xmath142 : @xmath143_{\mathcal a } : = \cap \{a\in\mathcal a\colon x\in a\}= \cap \{a\in\mathcal a\colon y\in a\}=:[y]_{\mathcal a}.\ ] ] thus , if @xmath144 is a sequence of countably generated @xmath3-algebras , then @xmath145 now , the atoms of the @xmath3-algebra @xmath146 of each simple target problem @xmath147 are at most countable , by definition .
then @xmath148 may be represented as a transition matrix on the state set @xmath149 .
each row of @xmath148 is a distribution probability on @xmath150 ( i.e. a sequence @xmath151 in the simplex @xmath152 of @xmath135 ) .
the first step of the target algorithm is to equip @xmath152 with the @xmath135norm and then to define an @xmath1-equivalence on @xmath152 .
we will alternatively use both the discrete equivalencies and the countable measurable partitions , as a consequence of the following result , whose proof is left to appendix . [
lem : discrete_measur ] given a measurable space @xmath7 , there exists a natural bijection between the set of discrete equivalencies on @xmath28 and the set of the countable measurable partitions of it .
let @xmath153 be the unit ball in @xmath135 and @xmath154 be the simplex on @xmath135 .
let @xmath155 $ ] , for any @xmath11 , and @xmath156 be the standard topology on @xmath157 $ ] .
denote by @xmath158}$ ] the borel @xmath3-algebra on @xmath157 $ ] generated by @xmath156 .
we look at @xmath152 as a subset of @xmath159 so that the borel @xmath3-algebra @xmath160 induced on @xmath152 is @xmath161 } \cap s$ ] .
@xmath162 is an _
@xmath1-equivalence on @xmath152 _ if it is a discrete equivalence on @xmath163 and @xmath164 whenever @xmath165 .
the choice of @xmath135norm on @xmath152 is linked to the total variation distance between probability measures .
the total variation distance between two probability measures @xmath8 and @xmath148 is defined by @xmath166 . now
the total variation of a measure @xmath138 is @xmath167 , where the supremum is taken over all the possible partitions of @xmath168 . as @xmath169
, we have that @xmath170 , see @xcite . to each @xmath171
corresponds the probability measure @xmath8 on @xmath150 with @xmath172 ( and viceversa ) .
therefore , since @xmath173 , we have @xmath174 [ exa : e - cut ] define the @xmath1-cut as follows .
@xmath175 , where @xmath176 denotes the entire part of @xmath34 .
then @xmath162 is an @xmath1-equivalence on @xmath152 .
indeed , * @xmath177 is at most countable ( since we divide each @xmath157 $ ] into classes of length @xmath178 ) .
* for any @xmath179 @xmath180 = \{q \in s\colon \pi_{\sim_{\epsilon}}(q ) = \pi_{\sim_{\epsilon}}(p ) \ } = \prod_{n } \bigg [ \frac{\big\lfloor \frac{2^np_n}{\epsilon } \big\rfloor\epsilon}{2^{n } } , \frac{\big(\big\lfloor \frac{2^np_n}{\epsilon } \big\rfloor+1\big)\epsilon}{2^{n } } \bigg ) \ , { { \bigcap s}}\ ] ] is measurable with respect to @xmath181 * @xmath182 , @xmath183 given a sequence @xmath184 of @xmath1-equivalences on @xmath152 , we define the choice of @xmath137 inductively .
this algorithm is a good candidate to be a strategy for the approximating problem we are facing and it is based on this idea : consider the equivalence classes given by @xmath185 and divide them again according to the following rule .
starting from any two points in the same class , we check whether the probabilities to attain any other @xmath185-classes are @xmath1-the same .
mathematically speaking : step @xmath36 : @xmath186 step @xmath11 : @xmath187 is based on the equivalence @xmath185 and on @xmath188 , inductively
. @xmath185 is generated by a countable partition of @xmath28 , say @xmath189 .
we define , for any couple @xmath190 , @xmath191 the following lemma [ lem : contmeas ] shows that @xmath192 is a discrete equivalency on @xmath7 , and therefore it defines @xmath193 as generated by a countable partitions of @xmath28 . in applications
, @xmath137 must be finitely generated .
this is not a big restriction .
in fact one can prove inductively that this is always the case if the projection @xmath188 divides each component of @xmath152 into a finite number of subsets , as in example [ exa : e - cut ] .
the choice of the `` optimal '' sequence @xmath134 is not the scope of this work .
we only note that the definition of @xmath162 can be relaxed and the choice of the sequence @xmath134 may be done interactively , obtaining a fewer number of classes @xmath194 at each step .
[ lem : contmeas ] @xmath137 is a filtration on @xmath18 .
moreover , for any @xmath43 , @xmath192 is a discrete equivalency on @xmath7 .
the monotonicity of @xmath137 is a simple consequence of .
the statement is true for @xmath195 , since @xmath196 . for the induction step ,
let @xmath197 be the measurable countable partition of @xmath28 given by @xmath198 .
the map @xmath199 given by @xmath200 is therefore measurable .
as @xmath188 is a discrete equivalency on @xmath163 , the map @xmath201 is also measurable , where @xmath202 is the natural projection associated with @xmath188 .
thus , two points @xmath203 are such that @xmath204 if and only if their image by @xmath205 is the same point of @xmath206 .
the new partition of @xmath28 built by @xmath192 is thus obtained as an intersection of the sets @xmath207 which formed the @xmath208-partition with the counter - images of @xmath206 by @xmath205 .
intersections between two measurable countable partitions of @xmath28 being a measurable countable partition of @xmath28 , we are done . before defining @xmath139 , we need the following result , which will be proved in section [ sec : proofs ] .
[ thm : pi_infty ] let @xmath209 be defined as in the previous section and let @xmath210 . then @xmath211 is a compatible projection . as a consequence of theorem [ thm : cp+tp ] and of theorem [ thm : pi_infty ] , a target problem @xmath212 is well defined .
we intend to define @xmath88 as the @xmath138weighted mean average of @xmath213 given the information carried by @xmath187 .
more precisely , let @xmath138 be a probability measure on @xmath214 such that @xmath215 , for any @xmath216 ( the existence of such a measure is shown in example [ exa : exist_meas ] ) .
for any @xmath217 , let @xmath218 be the @xmath219-random variable such that @xmath220 .
define @xmath221(x ) , \qquad \forall x\in x , \forall f\in { \mathcal f}_n.\ ] ] @xmath222 is uniquely defined on @xmath223 , the only @xmath138-null set of @xmath187 being the empty set .
then we can ensure that @xmath224 is a probability measure , for any @xmath21 .
we give in the following an example of the measure @xmath138 that has been used in equation which justifies its existence .
[ exa : exist_meas ] let @xmath225 be a sequence of independent and identically distributed geometric random variables , with @xmath226 .
let @xmath227 and set @xmath228 .
there exists a probability measure @xmath229 on @xmath230 such that @xmath231 and thus , @xmath232 , @xmath233 .
moreover , it follows that for any @xmath11 , @xmath234 we check by induction that we can embed @xmath235 into @xmath236 , for any @xmath237 .
the searched measure @xmath138 will be the trace of @xmath238 on the embedded @xmath3-field @xmath219 . for @xmath195 , define @xmath239 , @xmath240 .
the embedding forms a nontrivial partition , and therefore the restriction of @xmath238 to the embedding of @xmath241 defines a probability measure on @xmath242 with @xmath243 if @xmath244 . for the induction step , suppose it is true for @xmath11 .
given @xmath245 , we then have @xmath246 , where @xmath247 is a nontrivial partition in @xmath248 and therefore the restriction of @xmath238 to the embedding of @xmath249 defines a probability measure @xmath250 on @xmath187 with @xmath251 if @xmath244 . given @xmath252 , let @xmath253 .
the monotonicity of @xmath192 ensures that each @xmath254 will belong to one and only one @xmath255 .
moreover , by definition of @xmath254 , we have that @xmath256 since @xmath257 is at most countable , we may order @xmath255 for any @xmath96 .
we have accordingly defined an injective map @xmath258 , where @xmath259 according to the cardinality of @xmath255 , define the @xmath260-embedding @xmath261 by definition of @xmath262 and , it follows that we have mapped @xmath263 into a partition in @xmath264 .
moreover , @xmath265 as a consequence of .
the restriction of @xmath238 to the embedding of @xmath266 defines a probability measure on @xmath263 with @xmath267 if @xmath244 .
note that @xmath268 is by construction an extension of @xmath250 to @xmath263 .
finally , the extension theorem ensures the existence of the required @xmath138 , which is just mapped to the trace of @xmath238 on the embedded @xmath269 .
let @xmath11 objects @xmath270 be picked repeatedly with probability @xmath271 that object @xmath272 is picked on a given try , with @xmath273 .
find the earliest time at which all @xmath11 objects have been picked at least once .
it is not difficult to show that the general coupon collector s problem may be embedded into a markow network of @xmath274nodes ( see , @xcite ) .
thus , let @xmath8 be a @xmath275-transition matrix on the state set @xmath276 . for any @xmath277 ( @xmath278 )
, we define the @xmath279-matrix @xmath280 then @xmath281 is a nonnegative @xmath279-matrix .
it is a transition probability matrix from @xmath28 to @xmath282 ( we called it @xmath283 ) .
each row @xmath96 represents the restriction of @xmath284 to @xmath285 . as noted in the proof of theorem [ thm : mainta ] , we should choose a probability measure @xmath138 on @xmath28 and then define a new matrix @xmath286 on @xmath287 with .
a `` neutral '' choice for @xmath138 is @xmath288 .
accordingly , by , for any @xmath289 , @xmath290 a simple computation gives @xmath291 where @xmath292 .
we have tested two target algorithms on a coupon collector problem with @xmath293 objects . in this case
@xmath8 is given by a @xmath294-sparse matrix .
the computation of @xmath24 is not practicable .
the number of components of each class of @xmath295 is plotted for @xmath296 in the following figure [ fig : coupon1 ] and figure [ fig : coupon2 ] .
the sequence @xmath297 varies according to the target algorithm .
we remark that both strategies converge to the same exact solution .
[ cm][ct ] [ cm][ct ] [ cm][ct ] [ cm][ct ] ( log scale ) , with @xmath298 , @xmath299 , @xmath300 .
the choice of @xmath134 is made as in example [ exa : e - cut].,title="fig:",scaledwidth=95.0% ] + [ cm][ct ] [ cm][ct ] [ cm][ct ] [ cm][ct ] ( log scale ) with @xmath298 , @xmath299 , @xmath300 , with a different choice of the target algorithm.,title="fig:",scaledwidth=95.0% ] +
the problem of approximation is mathematically different if we start from a markov process with a countable set of states or with an uncountable one .
let us consider , for the moment , the countable case : @xmath28 is the at most countable set of the states and @xmath301 is the power set .
each function on @xmath28 is measurable .
if we take any equivalence relation on @xmath28 , it is both measurable and identified by the @xmath3-algebra it induces ( see theorem [ thm : why_a0 ] below ) .
this is not in general the case when we deal with a measurable space @xmath7 , with @xmath28 uncountable . in this section ,
we want to connect the process of approximation with the upgrading information .
more precisely , a measurable equivalence @xmath67 defines both the partition @xmath282 and the sigma algebra @xmath77 .
one wishes these two objects to be related , in the sense that ordering should be preserved .
example [ exa : why_a0 ] below shows a paradox concerning @xmath68 and @xmath77 when @xmath28 is uncountable .
in fact , [ lem : ordering_1 ] let @xmath302 be countably generated sub @xmath3-algebras of a measurable space @xmath7 .
then @xmath303_{\mathcal a_1}\supseteq [ x]_{\mathcal a_2}$ ] . in particular ,
let @xmath304 be random variables .
if @xmath305 , then @xmath306 . see appendix [ app : measuring ] .
the problem is that even if a partition is more informative than another one , it is not true that it generates a finer @xmath3-algebra , _ i.e. _ , the following implication is not always true for any couple of random variables @xmath41 and @xmath307 @xmath308 then lemma [ lem : ordering_1 ] is not invertible , if we do not require the further assumption on the measurable space @xmath7 .
this last fact connects the space @xmath7 with the theory of blackwell spaces ( see lemma [ lem : a0_a1 ] ) . we will assume the sole assumption . we give here a counterexample to assumption , where two random variables @xmath304 generate two different sigma algebras @xmath309 with the same set of atoms
obviously , assumption does not hold .
let @xmath310 be a polish space and suppose @xmath311 .
let @xmath312 and consider the sequence @xmath313 that determines @xmath314 , i.e. @xmath315 .
let @xmath316 .
@xmath317 . as a consequence of lemma [ lem : count_gen_new ] ,
there exist two random variables @xmath304 such that @xmath318 and @xmath319 .
the atoms of @xmath314 are the points of @xmath28 , and then the atoms of @xmath230 are also the points of @xmath28 , since @xmath320 .
we recall here the definition of blackwell spaces . a measurable space @xmath7 is said _ blackwell _ if @xmath15 is a countably generated @xmath3-algebra of @xmath28 and @xmath321 whenever @xmath142 is another countably generated @xmath3-algebra of @xmath28 such that @xmath322 , and @xmath142 has the same atoms as @xmath15 . a metric space @xmath28 is blackwell if , when endowed with its borel @xmath3-algebra , it is blackwell .
the measurable space @xmath7 is said to be a _ strongly blackwell space _
if @xmath15 is a countably generated @xmath3-algebra of @xmath28 and * @xmath323 if and only if the sets of their atoms coincide , where @xmath324 and @xmath325 are countably generated @xmath3-algebras with @xmath326 @xmath327 . for what concerns blackwell spaces ,
the literature is quite extensive .
d. blackwell proved that every analytic subset of a polish space is , with respect to its relative borel @xmath3-field , a strongly blackwell space ( see @xcite ) . therefore ,
if @xmath328 is ( an analytic subset of ) a polish space and @xmath329 , then @xmath7 can not be a weakly blackwell space . to see this ,
take @xmath330 a base of @xmath331 and @xmath332 .
then @xmath333 and @xmath334 have the same set of atoms ( the points of @xmath28 ) but @xmath335 ( or , equivalently , the identity function @xmath336 is not measurable ) .
moreover , as any ( at most ) countable set equipped with any @xmath3-algebra may be seen as an analytic subset of a polish space , then it is a strongly blackwell space .
a. maitra exhibited coanalytic sets that are not blackwell spaces ( see @xcite ) .
m. orkin constructed a nonanalytic ( in fact nonmeasurable ) set in a polish space that is a blackwell space ( see @xcite ) .
jasiski showed ( see @xcite ) that continuum hypothesis ( ch ) implies that there exist uncountable sierpiski and luzin subsets of @xmath337 which are blackwell spaces ( implying in a strong way that blackwell spaces do not have to be lebesgue measurable or have the baire property ) .
jasiski also showed that ch implies that there exist uncountable sierpiski and luzin subsets of @xmath337 which are not blackwell spaces ( implying in a strong way that lebesgue measurable sets and sets with the baire property do not have to be blackwell spaces ) .
this latter result is strengthened by r.m .
shortt in @xcite by showing that ch implies the existence of uncountable sierpiski and luzin subsets of @xmath337 which are highly non - blackwell in the sense that all blackwell subspaces of the two sets are countable .
note that assumption and assumption ( a1 ) coincide , as the following lemma states .
[ lem : a0_a1 ] let @xmath7 be a measurable space . then holds if and only if holds .
lemma [ lem : count_gen_new ] in appendix states that @xmath338 is countably generated if and only if there exists a random variable @xmath41 such that @xmath339 .
in addition , as a consequence of lemma [ lem : ordering_1 ] , we have only to prove that ( a1 ) implies . by contradiction ,
assume ( a1 ) , @xmath340 , but @xmath341 .
we have @xmath342 , and then @xmath343 by ( a1 ) and lemma [ lem : count_gen_new ] . on the other hand , as a consequence of eq . , we have that @xmath344 .
we call _ weakly blackwell space _ a measurable space @xmath7 such that assumption holds . if @xmath7 is a weakly blackwell space , then @xmath18 is a weakly blackwell space , for any @xmath345 .
moreover , every strong blackwell space is both a blackwell space and a weakly blackwell space whilst the other inclusions are not generally true . in @xcite , examples are provided of blackwell spaces which may be shown not to be weakly blackwell .
the following example shows that a weakly blackwell space need not be blackwell .
let @xmath28 be an uncountable set and @xmath15 be the countable
cocountable @xmath3-algebra on @xmath28 .
@xmath15 is easily shown to be not countably generated , and therefore @xmath7 is not a blackwell space .
take any countably generated @xmath3-field @xmath322 , i.e.@xmath346 .
* since each set ( or its complementary ) of @xmath15 is countable , then , without loss of generality , we can assume the cardinality of @xmath347 to be countable .
* each atom @xmath348 of @xmath349 is of the form @xmath350 note that the cardinality of the set @xmath351 is countable , as it is a countable union of countable sets . as a consequence of
, we face two types of atoms : 1 . for any @xmath96 , @xmath352 .
this is the atom made by the intersections of all the uncountable generators .
this is an uncountable atom , as it is equal to @xmath353 .
2 . exists @xmath96 such that @xmath354 .
this implies that this atom is a subset of the countable set @xmath355 .
therefore , all the atoms ( except @xmath353 ) are disjoint subsets of the countable set @xmath355 and hence they are countable .
it follows that the number of atoms of @xmath356 is at most countable .
thus , @xmath357 is a strongly blackwell space , i.e. @xmath7 is a weakly blackwell space .
[ exa : why_a0 ] suppose @xmath358 $ ] , @xmath359 where @xmath360 is the countable cocountable @xmath3-algebra on @xmath28 and @xmath361 .
consider a decisionmaker who chooses action 1 if @xmath362 and action 2 if @xmath363 .
suppose now that the information is modeled either as the partition of all elements of @xmath28 , @xmath364 and in this case the decisionmaker is perfectly informed , or as the partition @xmath365 if we deal with @xmath3-algebras as a model of information then @xmath366 and @xmath367 .
the partition @xmath156 is more informative than @xmath368 , whereas @xmath369 is not finer than @xmath370 .
in fact @xmath371 and therefore if the decisionmaker uses @xmath369 as its structure of information , believing it more detailed than @xmath370 , he will never know whether or not the event @xmath355 has occurred and can be led to take the wrong decision . in this case , @xmath3-algebras do not preserve information because they are not closed under arbitrary unions . however , if we deal with blackwell spaces , any countable @xmath3-algebra is identified by its atoms and therefore will possess an informational content ( see @xcite , for example ) . the following theorem , whose proof is in appendix [ app : a0 ] , links the measurability of any relation with the cardinality of the space and assumption .
it shows the main difference between the uncountable case and the countable one .
[ thm : why_a0 ] assume .
let @xmath7 be a measurable space .
the following properties are equivalent : 1 .
[ co:3 ] any equivalence relation @xmath63 on @xmath28 is measurable and assumption holds ; 2 .
[ co:2 ] @xmath372 is a weakly blackwell space ; 3 .
[ co:1 ] @xmath28 is countable and @xmath373 .
the following theorem mathematically motivates our approximation problem : any limit of a monotone sequence of discrete equivalence relationships is a measurable equivalence .
[ limit_eq ] for all @xmath43 , let @xmath192 be a discrete equivalency . then @xmath210 is a measurable equivalency .
conversely , for any measurable equivalency @xmath63 , there exists a sequence @xmath209 of discrete equivalencies such that @xmath210
. see appendix [ app : measuring ] .
let @xmath67 be a compatible projection .
we define @xmath374 what remains to prove is that @xmath375
. more precisely , we have to show that @xmath376 is @xmath77-measurable , @xmath377 . by contradiction
, there exists @xmath378 such that the random variable @xmath379 is not @xmath77-measurable .
then @xmath380 , and hence @xmath381 by assumption , which contradicts equation . as a consequence of theorem [ limit_eq ] , @xmath382 , where @xmath383 .
define @xmath384 we will prove that , for any @xmath75 , @xmath385 is @xmath77-measurable and consequently @xmath386 will be a compatible projection .
this implies that there exists a measurable function @xmath387 so that @xmath388 . therefore ,
if @xmath389 , then @xmath390 , which is the thesis . thus , we show that for any @xmath391 and @xmath392 , we have @xmath393 to prove equation , we first show that it is true when @xmath394 by proving that @xmath395 which implies that @xmath396 .
the inclusion @xmath397 is always true . for the other inclusion ,
let @xmath398 .
let @xmath399 ; there exists @xmath400 such that @xmath401 . therefore , equation and the definition of @xmath188 imply @xmath402 , for any @xmath403 . as @xmath404
, we obtain that @xmath405 .
then equation is true on the algebra @xmath406 .
+ actually , let @xmath407 such that @xmath408 .
we prove that equation holds for @xmath409 by showing that @xmath410 again , since @xmath411 , then @xmath412 and therefore @xmath413 .
conversely , the set @xmath414 is empty since the sequence of @xmath415-measurable maps @xmath416 converges to @xmath36 : @xmath417 then equation is true on the monotone class generated by the algebra @xmath418 , i.e. , equation is true on @xmath77 . given a target algorithm @xmath419 , let @xmath420 be defined as in theorem [ thm : pi_infty ] .
we show that @xmath211 is optimal .
let @xmath421 be another compatible projection and let @xmath422 be the target problem given by theorem [ thm : cp+tp ] .
we are going to prove by induction on @xmath11 that @xmath423 in fact , for @xmath195 it is sufficient to note that @xmath424 .
equation ( [ eq : projectionsta ] ) states that @xmath425 , where @xmath426 is the discrete random variable , given by lemma [ lem : discrete_measur ] , s.t.@xmath427 let @xmath428 $ ] be defined as @xmath429 . then @xmath430{e , r}{\epsilon_n } \node[3]{s/\epsilon_n } \end{diagram}\ ] ] obviously , @xmath431 . for the induction step , as @xmath432
, we have that @xmath433 is @xmath434-measurable , and therefore @xmath435
. then @xmath436 .
therefore @xmath437 , which implies @xmath438 by lemma [ lem : ordering_1 ] , and hence @xmath211 is optimal .
[ cor : uniqueness ] @xmath211 does not depend on the choice of @xmath46 .
@xmath439 is optimal , @xmath440 .
the optimal projection being unique , we are done .
let @xmath420 be defined as in theorem [ thm : pi_infty ] and @xmath441 be given by theorem [ thm : cp+tp ] so that @xmath442 for any @xmath75
. then each @xmath139 of definition [ eq : def_of_p_n ] can be rewritten as @xmath443_n } p_\infty(x , f)\mu(dz)}{\mu([x]_n ) } , \qquad \forall x\in x , \forall f\in { \mathcal f}_n,\ ] ] where @xmath303_n$ ] is the @xmath192-class of equivalence of @xmath34 and @xmath444_n)>0 $ ] since @xmath303_n\neq\varnothing$ ] .
note that @xmath445 .
then , for any @xmath446 , there exists an @xmath447 so that @xmath448 .
therefore we are going to prove by induction on @xmath11 that @xmath449 which completes the proof . if @xmath450 , then by definition of @xmath451 , since @xmath452 , we have that @xmath453_n}|p_\infty(z , t ) -p(x , t)|\mu(dz)}{\mu([x]_n ) } \\ & = \frac{\int_{[x]_n}|p(z , t ) -p(x , t)|\mu(dz)}{\mu([x]_n ) } \\ & \leq \epsilon_m \frac{\int_{[x]_n}{\mu(dz)}}{\mu([x]_n ) } = \epsilon_m .
\end{aligned}\ ] ] for the induction step , we note that @xmath454 where @xmath455 is the partition of @xmath28 given by @xmath456 . by induction hypothesis ,
@xmath457 for @xmath458 large enough .
since @xmath459_m = a_i^{(m)}$ ] if @xmath460 , it follows that @xmath461 equation becomes @xmath462 on the other hand , by equation , @xmath463_m}\frac{p_\infty(z , a^{(m)}_i)-p(x , a^{(m)}_i)}{\mu([x]_m)}\mu(dz).\ ] ] the definition of @xmath464 states that @xmath465 whenever @xmath466_m$ ] and therefore @xmath467_m}\sum_i \big\vert p_\infty(z , a^{(m)}_i)-p(x , a^{(m)}_i)\big\vert \frac{\mu(dz)}{\mu([x]_m ) } \leq \epsilon + \epsilon_{m+1}. \end{gathered}\ ] ] since @xmath468 as @xmath469 tends to infinity , we get the result . by definition [ eq : def_of_p_n ] and lemma [ lem : contmeas ] , @xmath470 is a martingale with respect to the filtration @xmath471 , for any @xmath472 . then , if @xmath473 as in definition [ eq : def_of_p_n ] , we have that @xmath474(x ) = y^f(x ) , \qquad \text{for $ \mu$-a.e . }
x\in x , \forall f\in \cup_m { \mathcal f}_m.\ ] ] let @xmath420 be defined as in theorem [ thm : pi_infty ] and @xmath441 given by theorem [ thm : cp+tp ] so that @xmath442 for any @xmath75
. then @xmath475 for any @xmath21 and @xmath83 , the only @xmath138-null set in @xmath476 being the empty set .
let the target problem @xmath52 be given and let @xmath477 , where @xmath478 be a target algorithm . in order to prove theorem [ prop : top ] , which states the total convergence of the probability measure @xmath224 towards @xmath19 , we proceed as follows : * first
, we define the topology @xmath127 on @xmath28 ; * then , we define a `` natural '' topology @xmath479 on @xmath28 associated to any target algorithm @xmath480 .
we prove in theorem [ thm : tot_conv ] the total convergence of @xmath139 to @xmath481 , under this topology ; * then , we define the topology @xmath482 on @xmath28 as the intersection of all the topologies @xmath479 ; * finally , we show theorem [ prop : top ] by proving that @xmath483 . the non - triviality of @xmath127 will imply that of @xmath482 .
we introduce the pseudometric @xmath126 on @xmath28 as follows : @xmath484 now , let @xmath479 be the topology generated by @xmath485 .
@xmath486 is a closed set if and only if @xmath487 .
in fact , if @xmath488 , for a given @xmath11 , then @xmath489 , for any @xmath490 and therefore @xmath486 is closed .
@xmath491 is a topological space .
let us go back to example [ exa : dyadic ] .
the topology defined by asking that the sets in each @xmath92 are closed is strictly finer than the standard topology . on the other hand , the same example
may be explained with left closed
right opened dyadic subdivisions , which leads to a different topology that also contains the natural one .
any other `` reasonable '' choice of subdivision will show the same : the topologies are different , and all contain the standard one . in the same manner
, we are going to show that all the topologies @xmath479 contain the standard one , @xmath127 .
[ thm : tot_conv ] let the target problem @xmath52 and the target algorithm @xmath419 be given .
for any target algorithm @xmath46 , @xmath492 let @xmath493 be a closed set of @xmath46 and let @xmath494 be any extension of @xmath222 to @xmath495 . we have to check that @xmath496 , for any given @xmath34 ( see , e.g. , @xcite )
note that , since @xmath497 , we have @xmath498 .
but , @xmath499 . actually , @xmath500 @xmath501 as @xmath11 tends to infinity , from the target algorithm , since as @xmath502
. for any @xmath503 , there exists @xmath504 , such that for any @xmath505 , @xmath506 .
+ @xmath507 as @xmath11 tends to infinity , from the continuity of the measure . for any @xmath503
, there exists @xmath508 and for any @xmath509 , @xmath510 an example of a natural extension of @xmath88 to @xmath494 is given by @xmath511(x ) , \qquad \forall x\in x , \forall f\in \vee_n\mathcal f_n,\ ] ] where , for any @xmath512 , @xmath218 is the @xmath219-random variable such that @xmath513 . as mentioned for @xmath222 ,
@xmath514 is a probability measure , for any @xmath21 . for any fixed strategy @xmath515 ,
let @xmath88 be as in theorem [ thm : mainta ] .
we have @xmath516 for any given @xmath34 . in order to describe the topology @xmath482
, we will denote by @xmath517_*$ ] the closure of a set @xmath518 in a given topology @xmath519 .
note that the monotonicity of @xmath192 implies @xmath520_{\tau_{\mathrm{str } } } = \cap_n [ [ f]]_{\tau_{\mathrm{str}_n}}\ ] ] where @xmath521 is the ( discrete ) topology on @xmath28 generated by @xmath92 . since @xmath482 is the intersection of all the topologies @xmath479
, we have @xmath520_{\tau_p } \supseteq [ [ f]]_{\tau_{\mathrm{str } } } = \cap_n [ [ f]]_{\tau_{\mathrm{str}_n } } , \qquad \forall f\in 2^x,\forall \mathrm{str}.\ ] ] let @xmath409 be the closed set in @xmath127 so defined @xmath522 i.e. , @xmath409 is the complementary of an open ball in @xmath523 with center @xmath34 and radius @xmath524 . if we show that @xmath525 , then we are done , since the arbitrary choice of @xmath34 and @xmath524 spans a base for the topology @xmath127 . we are going to prove @xmath526_{\tau_{\mathrm{str } } } = \cap_m [ [ f]]_{\tau_{\mathrm{str}_m } } , \qquad \forall \mathrm{str},\ ] ] which implies @xmath527_{\tau_p } = f $ ] .
it is always true that @xmath528_*$ ] ; we prove the nontrivial inclusion @xmath529_{\tau_{\mathrm{str}_m}}$ ] .
assume that @xmath530_{\tau_{\mathrm{str}}}$ ] .
now , @xmath530_{\tau_{\mathrm{str}_m}}$ ] , for any @xmath469 , and then there exists a sequence @xmath531 with @xmath532 such that @xmath533 , for any @xmath469 .
thus , @xmath534_m$ ] , where @xmath303_m$ ] is the @xmath456-class of equivalence of @xmath34 .
thus @xmath535 since @xmath536 is @xmath537-measurable .
by theorem [ thm : mainta ] , for any @xmath43 , @xmath538 now , let @xmath447 be such that @xmath539 and take @xmath540 sufficiently large s.t .
@xmath541 we have @xmath542 and therefore @xmath543 the arbitrary choice of @xmath1 implies @xmath544 , which is the thesis .
in this appendix we give the proof of auxiliary results that connect equivalency with measurability .
let @xmath67 be a discrete equivalency on @xmath28
. then @xmath282 defines a countable measurable partition of @xmath28 .
conversely , let @xmath545 be a countable measurable partition on @xmath28 . define @xmath546 s.t .
therefore @xmath41 is measurable and @xmath67 is a discrete equivalency on @xmath28 .
[ lem : mon_fun_alg_1 ] let @xmath304 be two random variables such that @xmath548
. then @xmath549 .
let @xmath550 be fixed .
we must prove that @xmath551 .
if @xmath552 or @xmath553 are empty , then we are done .
assume then that @xmath554 .
we have two cases - @xmath555 : @xmath556 such that @xmath557 .
+ by definition of @xmath558 , @xmath559 .
conversely , let @xmath560 .
since @xmath561 , then @xmath562 , _
i.e. _ @xmath563
. then @xmath564
. - @xmath565 : @xmath566 we have that @xmath567 .
+ then @xmath568 .
conversely , let @xmath560 .
since @xmath566 @xmath569 , then @xmath570 , which implies @xmath571 ,
_ i.e. _ @xmath572
. then @xmath573 .
the next lemma plays a central rle . its proof is common in set theory . [
lem : dens_count ] for all @xmath43 , let @xmath192 be a discrete measurable equivalency .
then there exists a random variable @xmath41 such that @xmath574 . before proving the core of the lemma
, we build a sequence @xmath575 of functions @xmath576 that will be used to define the function @xmath41 .
take @xmath577 to be the increasing function @xmath578 and let @xmath575 the sequence of function @xmath576 so defined : @xmath579 where , for all @xmath11 , @xmath580 and @xmath581 as a first consequence of the definition , note that for any choice of @xmath11 and @xmath582 , it holds that @xmath583 since @xmath584 .
we now prove by induction on @xmath585 that for any choice of @xmath586 , @xmath587 and @xmath588 , we have @xmath589 eq . is clearly true for @xmath590 , since @xmath591 is strictly monotone .
the same argument shows that eq .
is always true for @xmath592 and therefore we check it only for @xmath593 .
we assume by induction that eq .
is true for @xmath594 and we prove it for @xmath595 . by using twice the induction hypothesis , as @xmath596 , we obtain @xmath597 eq .
is now a consequence of eq . .
now , we come back to the proof of the lemma .
first note that , without loss of generality , we can ( and we do ) require the sequence @xmath598 to be monotone , by taking the sequence @xmath599 instead of @xmath192 .
@xmath600 is again a countable measurable equivalency on @xmath28 .
in fact , by lemma [ lem : discrete_measur ] we can read this statement in trivial terms of partitions : since at most countable intersection of families of countable measurable partition is a countable measurable partition .
moreover , by definition , @xmath601 .
let @xmath602 be the increasing sequence of countable measurable dissections of @xmath28 .
we are going to give a consistent inductive method of numbering the set of atoms of @xmath603 to build the functions @xmath604 .
let @xmath605 be any ordering of @xmath606 .
by induction , let @xmath607 be the partition of the atom @xmath608 given by @xmath609 .
define , for any @xmath43 , @xmath610 to complete the proof , we first show that @xmath611 , and then we prove @xmath612 by proving that @xmath613 pointwise . to prove that @xmath614 we show that @xmath615 .
one implication is a consequence of the fact that @xmath604 is defined on the partition of @xmath28 given by @xmath616 . for the converse , assume that @xmath617 and consider @xmath618 .
thus @xmath619 and , without loss of generalities , @xmath620 . by eq . , we have @xmath621 @xmath622
. + @xmath623 .
the sequence @xmath624 is monotone by definition and bounded by eq . .
then @xmath625 and thus @xmath626 .
let @xmath11 be fixed , and take @xmath203 with @xmath628 .
then , for any @xmath629 , @xmath630 implies @xmath631 as above , consider @xmath618 . as @xmath628
, we have @xmath619 and @xmath620 .
again , by eq . ,
for @xmath632 , @xmath633 _ i.e. _ , @xmath634 , @xmath635 . as @xmath636 , @xmath637
. apply lemma [ lem : mon_fun_alg_1 ] with @xmath638 to conclude that @xmath639 . as a consequence of lemma [ lem : dens_count ] ,
any countably generated sub @xmath3-algebra is generated by a measurable equivalence @xmath63 , as the following lemma states .
[ lem : count_gen_new ] @xmath338 is countably generated if and only if there exists a random variable @xmath41 such that @xmath339 .
@xmath640 let @xmath641 .
apply lemma [ lem : dens_count ] with @xmath642 .
@xmath643 take a countable base @xmath644 of @xmath645 and simply note that @xmath646 .
let @xmath21 be fixed . by hypothesis ,
if @xmath648 then @xmath325 will be of the form @xmath649 . without loss of generality ( if needed , by choosing @xmath650 instead of @xmath651 ) we can require @xmath652 , for any @xmath43 and @xmath653
. then @xmath303_{\mathcal a_2}=\cap_n ( a^1_n\cap a^2_n)\subseteq \cap_n a^1_n = [ x]_{\mathcal a_1}$ ] .
the last part of the proof is a consequence of lemma [ lem : count_gen_new ] and of the first point , since @xmath654_{\pi_f}\subseteq [ x]_{\pi_g}=g^{-1}(\{g(x)\}),\ ] ] or , equivalently , @xmath655 which is the thesis . note that @xmath656 is countable , generated by @xmath657 . then @xmath211 is a measurable equivalency by lemma [ lem : count_gen_new ] .
conversely , we can use the standard approximation technique : if @xmath67 is measurable , let @xmath658 for any @xmath11 .
since @xmath604 are discrete random variables , @xmath192 are defined through lemma [ lem : discrete_measur ] . by lemma [ lem : ordering_1 ] and eq . , the thesis @xmath659 will be a consequence of the fact that @xmath660 .
@xmath661 by definition , which implies @xmath662 . finally , as @xmath663
, we have @xmath626 , which completes the proof .
before proving the theorem , we state the following lemma . @xmath665 .
let @xmath666 be the identity relation : @xmath667 .
by hypothesis , there exists @xmath41 such that @xmath668 , and thus @xmath41 is injective . then @xmath669 .
now , take @xmath670 and let @xmath671 be the relation so defined : @xmath672 since any equivalency is measurable , then there exists @xmath673 such that @xmath674 . but
@xmath675 , which shows that @xmath676 , i.e.@xmath373 . @xmath677 .
since @xmath669 , there exists an injective function @xmath678 .
let @xmath63 be a equivalence relationship on @xmath28 , and define the following equivalence on @xmath679 : @xmath680 by definition of @xmath681 , if we denote by @xmath682 the canonical projection of @xmath679 on @xmath683 , then @xmath684 is such that @xmath685 the axiom of choice ensures the existence of a injective map @xmath686 .
then @xmath687 is such that @xmath67 .
@xmath41 is measurable since @xmath688 .
assume @xmath28 is uncountable .
by ch , exists @xmath691 s.t .
@xmath692 ( i.e. @xmath693 is in bijection with @xmath679 via @xmath694 ) . take a bijection @xmath695 .
then the map @xmath696 is a bijective map from @xmath697 to @xmath679 .
equip @xmath679 with the borel @xmath3-algebra @xmath645 and let @xmath698 .
@xmath324 is countably generated and its atoms are all the points in @xmath693 and the set @xmath699 .
now , take a non - borel set @xmath447 of the real line .
@xmath700 is also countably generated , @xmath701 and its atoms are all the points in @xmath693 and the set @xmath699 , too . since @xmath702 and @xmath703 , @xmath372 is not a weakly blackwell space by lemma [ lem : a0_a1 ] . @xmath704 . since @xmath28 is countable ,
then @xmath282 is .
therefore , lemma [ lem : discrete_measur ] ensures any equivalence @xmath63 is measurable , since @xmath688 .
finally , just note that each countable set is strongly blackwell . and
thus lemma [ lem : a0_a1 ] concludes the proof .
patrick billingsley , _ probability and measure _ , third ed .
, wiley series in probability and mathematical statistics , john wiley & sons inc . , new york , 1995 , a wiley - interscience publication .
mr mr1324786 ( 95k:60001 ) david blackwell , _ on a class of probability spaces _ , proceedings of the third berkeley symposium on mathematical statistics and probability , 19541955 , vol .
ii ( berkeley and los angeles ) , university of california press , 1956 , pp . 16 .
mr mr0084882 ( 18,940d ) | on a weakly blackwell space we show how to define a markov chain approximating problem , for the target problem .
the approximating problem is proved to converge to the optimal reduced problem under different pseudometrics .
a computational example of compression of information is discussed .
let @xmath0 be an homogeneous markov chain .
suppose the process stops once it reaches an absorbing class , called the target , according to a given stopping rule : the resulting problem is called target problem ( tp ) .
the idea is to reduce the available information in order to only use the necessary information which is relevant with respect to the target .
a new markov chain , associated with a new equivalent but reduced matrix is defined . in the ( large ) finite case
, the problem has been solved for tps : in @xcite , it has been proved that any tp on a finite set of states has its `` best target '' equivalent markov chain . moreover , this chain is unique and there exists a polynomial time algorithm to reach this optimum .
the question is now to find , in generality , an @xmath1approximation of the markov problem when the state space is measurable .
the idea is to merge into one group the points that @xmath1-behave the same with respect to the objective , but also in order to keep an almost equivalent markov chain , with respect to the other `` groups '' .
the construction of these groups is done through equivalence relations and hence each group corresponds to a class of equivalence .
in fact , there are many other mathematical fields where approximation problems are faced by equivalences . for instance , in integration theory , we use simple functions , in functional analysis , we use the density of countable generated subspaces and in numerical analysis , we use the finite elements method . in this paper ,
the approximation is made by means of discrete equivalences , which will be defined in the following .
the purpose of any approximation is to reach the exact solution when @xmath2 .
we prove that the sequence of approximations tends to the optimal exact equivalence relation defined in @xcite , when we refine the groups .
finer equivalence will imply better approximation , and accordingly the limit will be defined as a countably generated equivalence . under a very general blackwell type hypothesis on the measurable space
, we show that it is equivalent to speak on countably generated equivalence relationships or on measurable real functions on the measurable space of states .
if we do not work under this framework of blackwell spaces , we can be faced to paradoxes , as it is explained by @xcite , of enlarging @xmath3algebras , while decreasing the information available to a decision - maker .
the @xmath1approximation of the markov chain depends always upon the kind of objective . in @xcite ,
jerrum deals with ergodic markov chains .
his objective is to approximate the stationary distribution by means of a discrete approximating markov chain , whose limit distribution is close in a certain sense to the original one .
however , unlike our following work , his purpose is not the explicit and unified construction of the approximating process . in this paper , we focus on the target problem .
we solve extensively the tp , where the objective is connected with the conditional probability of reaching the target @xmath4 , namely @xmath5 , for any @xmath6 .
this part extends the work in @xcite , since tps approximation may help to understand the behavior of those tps where the best equivalent markov chain is also very large .
the setting of an approximating problem can be extended to a general form , but we will not develop it in this paper . |
universal scaling of elliptic flow ( @xmath0 ) has been recently observed at rhic @xcite .
that is , for a broad range of particle species , @xmath2 vs. @xmath3 scales to a single function ; here , @xmath4 and @xmath5 are the number of valence quarks ( @xmath6 for mesons and baryons respectively ) and the transverse kinetic energy of the particle .
this observation has been interpreted as evidence that transverse expansion of the matter produced in energetic rhic collisions , occurs during a phase dominated by partonic collectivity . in this contribution we further demonstrate ; ( i ) the partonic origin of @xmath0 via scaling tests for the @xmath7 meson ; ( ii ) essentially full thermalization for the charm quark via scaling tests for d meson @xmath0 ; and ( iii ) validate universal scaling of @xmath0 in au+au and cu+cu collisions at @xmath1 gev .
differential @xmath0 measurements for multi - strange hadrons ( @xmath7 , @xmath8 , @xmath9 ) allow us to address the important question of how the existence of a hadronic phase affects @xmath0 .
these particles are to have a small hadronic cross section for interaction with non - strange hadrons . consequently ,
if elliptic flow was established in a phase involving hadrons interacting with their standard hadronic cross sections ( post - hadronization ) , one would expect @xmath0 for the multi - strange hadrons to be significantly smaller than that for other hadrons ( e.g. @xmath10 , k , @xmath11 ) .
the left panel of fig . 1 compares phenix @xmath0 results for the @xmath7 meson , @xmath12 , @xmath13 and @xmath14 from 20 - 60% central au+au collisions ; they show that , despite its mass which is comparable to that for the @xmath10 and @xmath15 , its @xmath16 values are similar to those for the lighter mesons ( @xmath11 and @xmath17 ) whose hadronic re - scattering cross sections are expected to be large . the right panel of fig . 1 shows the positive results from a validation test for universal scaling of the @xmath7 meson @xmath0 .
the recent high - statistics preliminary star results for @xmath9 and @xmath8 produced in au+au collisions at @xmath18 gev @xcite show a similar universal scaling @xcite .
these results suggest that the @xmath0 values for hadrons comprised of u , d and s quarks develop in the partonic phase and the existence of a hadronic phase does not lead to their modification .
further insight into properties of the produced medium at rhic can be obtained from the study of elliptic flow of particles comprised of heavy quarks ( charm or bottom ) . if charm quarks flow like light quarks
, it would indicate an unexpectedly strong interaction of charm with the medium .
one method of accessing charm @xmath0 is to measure the @xmath0 for non - photonic " electrons @xcite . a recent high statistics phenix measurements show rather larger elliptic flow for non - photonic " electrons from au+au collsions at @xmath1 gev @xcite .
the non - photonic " electrons arise mainly from d meson decay for @xmath19 2gev / c .
therefore , the large heavy flavor electron @xmath0 indicates that the d meson and consequently the charm quark , do flow with a large @xmath0 .
this suggests that the in - medium interactions are strong and perhaps frequent enough to thermalize the charm quark .
if this is indeed the case , then @xmath16 for the d meson should show the same universal scaling presented earlier for other hadrons .
the @xmath0 values for d mesons can been obtained from the values for non - photonic electrons via detailed simulations , see @xcite for details . for these simulations , it is necessary to assume the shape for @xmath20 for the d mesons .
the left and right panels of fig . 2
compare the unscaled and scaled results ( respectively ) for @xmath0 vs. @xmath5 for @xmath21 and @xmath22 mesons measured in minimum bias au+au collisions at @xmath1 gev . here
, the assumed shape for the @xmath23 dependence of the d meson @xmath0 is that for the proton ( see @xcite ) .
this assumption is compatible with a large array of differential @xmath0 measurements for different heavy " particle species @xcite .
both panels of fig .
2 show robust scaling ; they confirm that the d mesons ( for @xmath24 gev / c ) and hence , their associated charm quarks , flow with a fluid velocity field similar to that for the other quarks .
this suggests that the medium responds as a thermalized fluid and the transport mean free path is small .
if @xmath0 develops in a phase dominated by partonic collectivity and @xmath25 , @xmath26 , @xmath27 and @xmath28 quarks flow with a common velocity field , one expects universal scaling for all particle species measured at rhic .
figure 3(a ) shows differential flow measurements @xmath20 , for several particle species produced at mid - rapidity in central and semi - central au+au collisions at @xmath29=200 gev ; they span essentially the full range of measurements ( several hundred data points ) at rhic .
the @xmath0 values for pions ( @xmath30 ) , kaons @xmath13 , ( anti-)protons ( @xmath31,p ) , ( anti-)deuterons ( @xmath32)d and the @xmath7 meson comprise results from the phenix @xcite .
the values for neutral kaons ( @xmath33 ) , lambdas ( @xmath34 ) , cascades ( @xmath35 ) and omegas ( @xmath36 ) represent results from the star @xcite .
3 ( b ) shows the scaled results ( @xmath37 vs @xmath3 ) obtained from the same data ; here @xmath38 is the integral @xmath0 of charged hadrons for each of the indicated centrality selections , multiplied by a constant factor @xmath39 ( i.e. @xmath40 ) @xcite .
recent measurements @xcite indicate that @xmath41 is independent of centrality and the size of the colliding system as would be expected from a hydrodynamic system .
3(b ) indicates that the relatively complicated dependence of @xmath0 on centrality , transverse momentum , particle type , etc .
, for particles produced at mid - rapidity can be scaled to a single function .
1(c ) demonstrates that the same scaling also holds for @xmath12,(@xmath13 ) and ( @xmath31,p ) produced at mid - rapidity in cu+cu collisions at @xmath29=200 gev . vs. @xmath23 and ( b ) @xmath42 ) vs. @xmath3 for several centralities and particle species as indicated .
( c ) @xmath42 ) vs. @xmath3 for @xmath12 , @xmath13 , and ( @xmath31,p ) from cu+cu and au+au collisions at @xmath18 gev , scaledwidth=85.0% ] vs. @xmath23 and ( b ) @xmath42 ) vs. @xmath3 for several centralities and particle species as indicated .
( c ) @xmath42 ) vs. @xmath3 for @xmath12 , @xmath13 , and ( @xmath31,p ) from cu+cu and au+au collisions at @xmath18 gev , scaledwidth=85.0% ] vs. @xmath23 and ( b ) @xmath42 ) vs. @xmath3 for several centralities and particle species as indicated .
( c ) @xmath42 ) vs. @xmath3 for @xmath12 , @xmath13 , and ( @xmath31,p ) from cu+cu and au+au collisions at @xmath18 gev , scaledwidth=85.0% ] vs. @xmath23 and ( b ) @xmath42 ) vs. @xmath3 for several centralities and particle species as indicated .
( c ) @xmath42 ) vs. @xmath3 for @xmath12 , @xmath13 , and ( @xmath31,p ) from cu+cu and au+au collisions at @xmath18 gev , scaledwidth=85.0% ]
validation tests for the universal scaling of @xmath0 at rhic , suggest that the transverse expansion dynamics leading to elliptic flow are not controlled by ordinary hadrons interacting with their standard hadronic cross sections . instead , they suggest a pre - hardonization state exhibiting partonic collectivity leading to the flow of light , strange and heavy quarks with a common expansion velocity field .
10 phenix collaboration , a. adare _ et.al . _ nucl - ex/0608033 phenix collaboration , m. issah _ et.al .
_ nucl - ex/0604011 r. a. lacey , a. taranenko , nucl - ex/0610029 star collaboration , m. oldenburg _ et.al._ , _ _ j .
phys . g__**32 * * , s563 ( 2006 ) phenix collaboration , s. sakai _ et.al._ , _ _ j .
phys . g__**32 * * , s551 ( 2006 ) ; phenix collaboration , a. adare _ et.al .
_ nucl - ex/0611018 ; star collaboration , j. adams _ et.al .
lett . _ * 92 * , 052302 ( 2004 ) ; j. adams _ et.al . _ _ phys .
rev . _ * c72 * , 014904 ( 2005 ) , j. adams _ et.al .
lett _ * 95 * , 122301 ( 2005 ) phenix collaboration , s. s. adler _ et.al . _
lett . _ * 96 * , 032302 ( 2006 ) | recent phenix elliptic flow ( @xmath0 ) measurements for identified particles produced in au+au and cu+cu collisions at @xmath1 gev are presented and compared to other rhic measurements .
they indicate universal scaling of @xmath0 compatible with partonic collectivity leading to the flow of light , strange and heavy quarks with a common expansion velocity field . |
many models of physics beyond the standard model ( sm ) have an extra abelian gauge group @xmath1 @xcite .
there are many options for this @xmath1 gauge symmetry and the corresponding @xmath0 from the broken symmetry can enable its identification .
the drell - yan process , wherein the @xmath0 is produced from quark - antiquark fusion and decays to a lepton - antilepton pair , can give a particularly clear signal at a hadron collider @xcite . however , the lepton pair search for a @xmath0 is nullified if the @xmath0 does not couple to the sm leptons .
searches can still be made for dijet decay products of a @xmath0 , but the qcd dijet backgrounds are huge and fog such a signal @xcite ; hence , a @xmath0 resonance may not be discovered in dijets @xcite , especially if its coupling strength to quarks is not large , although a signal in the top pair channel could be easier to recognize @xcite .
our interest here is in a 4-lepton signal from a leptophobic @xmath0 that can be produced at the lhc ( and the tevatron ) with a large cross section and give a 4-lepton signal comparable to that of the lepton pair signals of generic @xmath0 models .
specifically , we consider a @xmath0 resonance in which the 4-leptons final state is bridged by pair production of a new scalar boson ( @xmath2 ) .
the @xmath0 couples to quark pairs and @xmath2 , but not to lepton pairs , and the new scalar @xmath2 decays into a lepton pair ( see fig .
[ fig:4leptonresonance ] ) .
lhc experiments , and possibly tevatron experiments , can find or reject this distinctive 4-lepton signal .
a leptophobic @xmath0 may also appear as a resonance in a 6-lepton final state ; a future search for this signal at the lhc requires @xmath3 integrated luminosity at @xmath4 center - of - mass ( cm ) energy @xcite .
resonance diagram at a hadron collider.,scaledwidth=30.0% ]
we begin by introducing a specific model in which a 4-lepton @xmath0 resonance can be realized without having a corresponding lepton pair signal .
we consider a generic supersymmetry ( susy ) framework where scalar fields are abundant .
the baryon number ( @xmath5 ) is not preserved in the susy framework and in general the proton is unstable .
thus a gauged @xmath5 has been sought , but then additional fermions are required to cancel the anomaly @xcite .
one natural way of anomaly cancellation is to add a fourth - generation ( 4 g ) of fermions .
then , by requiring all quarks carry @xmath6 , the 4 g lepton charge is uniquely determined to be @xmath7 by the anomaly free condition : rlcrc & @xmath8 , & & & @xmath9 , + & @xmath8 , & & & -@xmath10 .
+ this is effectively @xmath11 for the sm fermions : every sm quark has @xmath5 as a charge , and every sm lepton has @xmath9 charge can be a possible source of the @xmath12 anomaly recently reported in the tevatron cdf experiment @xcite . ] .
although proton stability would not have been guaranteed once the @xmath11 is broken spontaneously , it turned out there exists a residual @xmath13 discrete symmetry , called baryon tetrality ( @xmath14 ) , that forbids proton decay @xcite . under @xmath14 , lepton number
violating operators can exist ( such as @xmath15 and @xmath16 ) , but not baryon number violating operators ( such as @xmath17 ) . in order to have the @xmath14 residual discrete symmetry , the higgs boson that spontaneously breaks the @xmath11 gauge symmetry ( typically , a new higgs singlet ) should have a @xmath11 charge of @xmath10 or @xmath7 @xcite .
since it coincides with the @xmath11 charge of the @xmath18 [ 4 g right - handed neutrino and sneutrino ( superpartner of neutrino ) ] , we can adopt the approach of ref .
@xcite in which the 4 g right - handed sneutrino ( let us call it @xmath19 ) with a vacuum expectation value ( vev ) is used to break the @xmath11 without the need for a separate singlet . in general , the 4 g sneutrino ( left - handed one ) can also have a vev through a mixing although we will assume the mixing is very small . because the 4 g majorana neutrino mass term is forbidden by the @xmath11 symmetry ,
the 4 g neutrino is a dirac particle on which the seesaw mechanism does not work , and thus can easily satisfy the lep @xmath20 width measurement that is compatible only with 3 light active neutrinos @xcite .
we take the 4 g sneutrino ( @xmath21 ) , the spin-0 companion of the 4 g neutrino , as a bridging scalar between the @xmath0 and the lepton final states .
it has a nonzero @xmath11 charge @xmath22 and can couple to the @xmath0 while the sneutrinos of the first 3 generations have vanishing @xmath11 charge .
we assume the @xmath21 is the lightest superpartner ( lsp ) ; it can decay into a lepton pair through the lepton number violating interaction @xmath23 .
( for instance , see ref .
@xcite . ) a nonzero @xmath11 charge for the 3-generation sneutrinos would inevitably have led to 2-lepton @xmath0 resonance @xcite .
in the remainder of this paper , we focus on the collider physics consequences of this scenario with the @xmath21 lsp .
our analysis does not necessitate this particular supersymmetric model , albeit well motivated .
rather , the model serves as an existence proof of a consistent theory of the 4-lepton @xmath0 resonance without a 2-lepton @xmath0 resonance , and also provides a specific realization of the phenomenology of a new scalar that couples to the @xmath0 and lepton pairs .
here , we discuss some characteristic features of the @xmath21 lsp decay exclusively through lepton number violating operators .
renormalizable operators @xmath24 and @xmath25 are forbidden by the @xmath11 gauge symmetry .
although operators with two 4 g fields such as @xmath26 and @xmath27 ( or @xmath28 ) are allowed at the renormalizable level , @xmath21 decays can not be mediated by these operators due to kinematics when @xmath21 is the lightest of the 4 g states .
thus , nonrenormalizable operators @xmath29 and @xmath30 with a heavy mass parameter @xmath31 allow @xmath21 decays since @xmath32 = 4 $ ] .
taking @xmath33 , @xmath34 and @xmath35 , for instance , effective coefficients @xmath36 , @xmath37 are obtained is expected in susy to keep the extra @xmath38-term contribution to the sfermion masses small , and @xmath39 satisfies bounds on the scale of new physics from various constraints such as neutral kaon mixing . ] . neglecting the light fermion masses ,
we obtain the partial widths & & ( _ 4 ^+_j ^-_k ) = ( ^eff_4jk)^2 m__4 , + & & ( _ 4 |d_j d_k ) = ( ^eff_4jk)^2 m__4 .
if we take all @xmath40 , the @xmath21 lsp would decay only through @xmath41 ( @xmath42 ) with a total decay width given by _ _ 4 = .
it is demanded that @xmath43 by the result of the lep @xmath20 decay experiment .
the @xmath44 can be constrained by various experiments such as @xmath45 , @xmath46 and similar @xmath47 decays .
the bounds depend on the final lepton flavor indices ( @xmath48 ) , and currently the most severe bound comes from @xmath49 , which translates into @xmath50 for @xmath51 . in a flavor - blind sense , it corresponds to @xmath52 for the @xmath21 lsp with the other @xmath53 s sufficiently heavy @xcite .
that is @xmath54 ( @xmath55 ) for @xmath56 ( @xmath57 ) , which falls into the ballpark of the aforementioned @xmath58 value for @xmath59 .
larger @xmath58 may be allowed for a specific choice of @xmath60 as the experimental bounds are flavor - dependent .
because of many free parameters in @xmath61 , a wide range of @xmath62 for a given @xmath63 can be accommodated . + ( a ) ( b ) ( c )
in this section we present quantitative cross section predictions of the 4-leptons channel for the lhc7 ( lhc with @xmath64 cm energy ) experiments . for the calculations we use comphep / calchep @xcite , with some modifications , and the parton distribution function of cteq6l @xcite . for definiteness ,
we take the @xmath0 gauge coupling constant to be @xmath65 ; the @xmath0 production cross section and @xmath0 width can be simply scaled by @xmath66 for other @xmath67 values .
we assume that the @xmath21 lsp is the lightest 4 g field , with @xmath56 , and that all new particles , except for the @xmath21 lsp , have masses larger than @xmath68 so that @xmath0 decays only into the sm fermions and the @xmath21 pair .
thus , the total @xmath0 width we take is the minimum value , which is @xmath69 for @xmath70 .
the 4-lepton @xmath0 resonance cross section is ( p p 4 ) ( p p z ) br(z _ 4 _ 4^ * ) br(_4 2)^2 .
the branching fraction is @xmath71 for @xmath70 .
the @xmath21 branching fractions to the light leptons ( @xmath72 , @xmath73 , @xmath74 ) are parameter dependent and flavor nonuniversality is expected .
we shall illustrate the case @xmath75 , which is indeed possible to arrange . figure [ fig : plots ] ( a ) shows the @xmath0 production cross section at the lhc ( solid curve ) and tevatron ( dashed curve ) , for the @xmath0 mass range @xmath76 .
the low mass region would have been excluded by the dilepton @xmath0 resonance searches at the lhc had a @xmath0 coupled to the light leptons .
for instance , the current bound on the sequential @xmath0 model is already @xmath77 @xcite though its couplings are larger than our benchmark coupling .
the ratio of the tevatron to lhc @xmath0 production cross sections is about @xmath78 for @xmath79 , and it drops rapidly at higher @xmath80 . though it might be possible to have an observable 4-lepton resonance at the tevatron , especially for the low @xmath0 mass region , we will focus on the lhc experiments in our analysis .
figure [ fig : plots ] ( b ) shows the 4-lepton @xmath0 resonance cross section at the lhc after the following typical acceptance cuts and @xmath0 invariant mass cut : + @xmath81 ( i ) @xmath82 ( each lepton ) , + @xmath81 ( ii ) @xmath83 ( each lepton ) , + @xmath81 ( iii ) @xmath84 ( 4-leptons ) .
the sm 4-lepton background to @xmath72 and @xmath74 pairs is principally from the @xmath85 subprocess .
as a recent atlas analysis shows , with nearly the same @xmath86 and @xmath87 cuts as ours , the sm background is negligible when the @xmath88 mass is outside the @xmath20 window of @xmath89 @xcite .
furthermore , some 4-lepton combinations ( such as @xmath90 , @xmath91 ) do not have any significant sm backgrounds .
thus , through all the @xmath0 mass range , we will require a small number of 4-lepton events ( 10 events ) after the acceptance cuts , in order to estimate the discovery reach .
figure [ fig : plots ] ( c ) shows the required luminosity at lhc7 to realize a signal of 10 events at a 4-lepton resonance as read from fig .
[ fig : plots ] ( b ) . for @xmath65 , an integrated luminosity at lhc7 of @xmath92 is needed for discovery ( 10 events ) of @xmath93 .
the existence of a 4-lepton @xmath0 resonance is already being probed at lhc7 in terms of the @xmath94 and @xmath67 ( with 3-events criteria and @xmath95 ) @xcite gives @xmath96 for @xmath65 . ] and an integrated luminosity of @xmath97 in each detector is expected before the end of 2011 .
the current lhc dijet search results ( with @xmath98 ) do not constrain the model for @xmath99 , as can be deduced from the estimates in ref . @xcite .
a 4-lepton signal could be confused initially with a possible higgs signal from @xmath100 with each @xmath20 decaying to lepton pairs .
there are several distinguishing characteristics of the signals : ( i ) @xmath20 decay includes neutrino decay modes that are absent in @xmath21 decay ; ( ii ) @xmath21 can decay into different lepton flavors which allows final states like @xmath90 and @xmath91 , although these could be switched off by @xmath101 ; ( iii ) the angular distribution of leptons in their rest frame is flat for the scalars ( higgs and sneutrino ) , but @xmath102-dependent for the vectors ( @xmath20 and @xmath0 ) ; ( iv ) if the @xmath21 mass differs from the @xmath20 boson mass , the lepton pair invariant mass distributions from the sneutrino decays would peak at a value different from @xmath103 , either lower or higher ; and ( v ) @xmath100 should be accompanied by @xmath104 , with a ratio of about 1 to 2 .
another exotic possibility for 4-lepton events is that a higgs - like boson is produced via gluon - gluon fusion and it decays to a pair of hidden sector fields ( vectors or scalars ) , each of which then decay to two leptons @xcite .
the production cross section for a higgs boson via gluon - gluon fusion would be much larger at lhc7 than at the tevatron .
though we have limited ourselves to only 4-lepton events , it is straightforward to extend the idea to other 4-fermion resonances depending on the values of @xmath105 and @xmath106 , such as @xmath107 , @xmath108 , @xmath109 , etc .
we have discussed a novel @xmath0 search channel in which a 4-lepton @xmath0 resonance can be produced at the lhc without an accompanying 2-lepton @xmath0 resonance signal .
we have shown that it is possible to construct a consistent supersymmetric model which has a @xmath0 particle with this property .
the @xmath1 symmetry of the model respects the baryon number for the first three generations .
the model is made anomaly free by the addition of a fourth generation of fermions .
then the @xmath0 can decay to the fourth generation sneutrino pair , which in turn decay into lepton pairs , thus giving the 4-lepton resonance signal .
the @xmath0 and the @xmath21 can be discovered or excluded in the near future by the lhc experiments .
we thank the organizers , p. fileviez - perez and y. kamyshkov , of the workshop on baryon & lepton number violation ( gatlinburg , tennessee , 2011
september ) , where some of this work was done .
hl thanks s .- c .
hsu for useful discussions .
this work was supported in part by the u.s .
department of energy under grant contract no .
de - fg02 - 95er40896 and no .
de - ac02 - 98ch10886 .
99 p. langacker , rev .
phys . * 81 * , 1199 ( 2009 ) [ arxiv:0801.1345 [ hep - ph ] ] .
g. aad _ et al .
_ [ atlas collaboration ] , phys .
lett . * 107 * , 272002 ( 2011 ) [ arxiv:1108.1582 [ hep - ex ] ] .
[ cms collaboration ] , cms - pas - exo-11 - 019 .
s. chatrchyan _ et al .
_ [ cms collaboration ] , phys .
b * 704 * , 123 ( 2011 ) [ arxiv:1107.4771 [ hep - ex ] ] .
g. aad _ et al .
_ [ atlas collaboration ] , phys .
b * 708 * , 37 ( 2012 ) [ arxiv:1108.6311 [ hep - ex ] ] .
[ cms collaboration ] , cms - pas - top-10 - 007 .
v. barger , p. langacker and h. s. lee , phys .
lett . * 103 * , 251802 ( 2009 ) [ arxiv:0909.2641 [ hep - ph ] ] . c. d. carone and h. murayama , phys .
lett . * 74 * , 3122 ( 1995 ) [ arxiv : hep - ph/9411256 ] .
d. c. bailey and s. davidson , phys .
b * 348 * , 185 ( 1995 ) [ arxiv : hep - ph/9411355 ] .
p. fileviez perez and m. b. wise , phys .
d * 82 * , 011901 ( 2010 ) [ erratum - ibid .
d * 82 * , 079901 ( 2010 ) ] [ arxiv:1002.1754 [ hep - ph ] ] .
p. ko and y. omura , phys .
b * 701 * , 363 ( 2011 ) [ arxiv:1012.4679 [ hep - ph ] ] .
p. fileviez perez and m. b. wise , phys .
d * 84 * , 055015 ( 2011 ) [ arxiv:1105.3190 [ hep - ph ] ] .
p. fileviez perez and m. b. wise ,
jhep * 1108 * , 068 ( 2011 ) [ arxiv:1106.0343 [ hep - ph ] ] .
r. f. lebed and v. e. mayes , phys .
d * 84 * , 075016 ( 2011 ) [ arxiv:1106.4347 [ hep - ph ] ] .
k. cheung and j. song , phys .
* 106 * , 211803 ( 2011 ) [ arxiv:1104.1375 [ hep - ph ] ] .
h. s. lee , phys .
b * 704 * , 316 ( 2011 ) [ arxiv:1007.1040 [ hep - ph ] ] .
v. barger , p. fileviez perez and s. spinner , phys .
rev . lett . * 102 * , 181802 ( 2009 ) [ arxiv:0812.3661 [ hep - ph ] ] .
k. nakamura _ et al .
_ [ particle data group ] , j. phys .
g * 37 * , 075021 ( 2010 ) . v. d. barger , g. f. giudice and t. han , phys .
d * 40 * , 2987 ( 1989 ) .
h. s. lee , phys .
b * 674 * , 87 ( 2009 ) [ arxiv:0812.1854 [ hep - ph ] ] .
r. barbier _ et al .
. rept .
* 420 * , 1 ( 2005 ) [ arxiv : hep - ph/0406039 ] .
a. pukhov _ et al .
_ , arxiv : hep - ph/9908288 .
a. pukhov , arxiv : hep - ph/0412191 .
j. pumplin , d. r. stump , j. huston , h. l. lai , p. m. nadolsky and w. k. tung , jhep * 0207 * , 012 ( 2002 ) [ arxiv : hep - ph/0201195 ] .
our model file is linked at + http://quark.phy.bnl.gov/~hlee/simulation/[http://quark.phy.bnl.gov/@xmath110hlee/simulation/ ] .
h. s. lee , z. liu and a. soni , phys .
b * 704 * , 30 ( 2011 ) [ arxiv:1105.3490 [ hep - ph ] ] .
s. gopalakrishna , s. jung and j. d. wells , phys .
d * 78 * , 055002 ( 2008 ) [ arxiv:0801.3456 [ hep - ph ] ] .
a. falkowski , j. t. ruderman , t. volansky and j. zupan , jhep * 1005 * , 077 ( 2010 ) [ arxiv:1002.2952 [ hep - ph ] ] . | a spin-1 weakly interacting vector boson , @xmath0 , is predicted by many new physics theories .
searches at colliders for such a @xmath0 resonance typically focus on lepton - antilepton or top - antitop events . here
we present a novel channel with a @xmath0 resonance that decays to 4 leptons , but not to 2 leptons , and discuss its possible discovery at the large hadron collider .
this baryonic gauge boson is well motivated in a supersymmetry framework . |
first studied in detail by r. peterson in the mid 1980s @xcite , the relativley rapid rotation of some horizontal branch stars remains an enigmatic feature of this phase of evolution .
rapid rotation was not seen in all horizontal branch ( hb ) stars , but the work by @xcite on m13 suggested that rapid rotation was more common in cooler hb stars .
later observations by @xcite of hb stars in m15 and in m13 @xcite further illuminate that there is a gap in the horizontal branches of m15 and m13 , with two distinct rotation rates on either side . at effective temperatures
less than about 11000 k , hb stars frequently show rapid rotation ( @xmath1 @xmath2 40 km / s ) while no rotation rates above 10 km / s appear in the hotter stars ( @xmath3 11000 k ) .
the work by @xcite agreed with a previous survey of m13 hb stars completed by @xcite .
further work by @xcite amplifies these results . in additon ,
a study of several other galactic globular clusters by @xcite and @xcite found similar results for the rotation of cool and hot hb stars .
rr lyra stars appear to show rotation with velocities below 10 km / s ( eg @xcite ) . while it is possible that the pulsations can drive a modest wind which in turn carries away angular momentum from these pulsating stars , this pause in the trend of rotation velocity with decreasing temperature is a mystery .
age and metallicity are the primary factors in determining a cluster s features and placement in the h
r diagram , and ( in principle ) the structure of the cluster s hb .
thus , if two clusters have similar metallicities and ages , one would expect them to show similar horizontal branch morphologies .
but that is not the case ; this is the infamous `` second parameter problem '' of globular cluster evolution .
the rotation anomaly may couple to the other questions raised by hb morphology through the second parameter problem @xcite so an understanding of the rotational break could be important in understanding the fundamentals of stellar evolution .
much of the work on horizontal branch stars naturally concentrates on stars in globular clusters .
however , field stars that correspond to the post helium core flash of low mass stars can perhaps play an important role in understanding the connection between rotation and hb morphology .
for population i stars , identification of post helium
flash objects is complicated by the paucity of stars in galactic clusters and distance determination for field stars .
however , one class of such stars , the subdwarf b stars , are very hot ( @xmath4 or more ) , and show high gravity ( @xmath5 ) .
thus they are readily identified as blue point sources in multicolor surveys , and spectroscopy reveals their stellar nature .
they are the field analog to the bluest hb stars seen in clusters , and have very thin surface hydrogen layers .
but these stars present their own enigmas - in particular the origin of sdb stars .
their origin as stars which have lost most but not all of their hydrogen envelope following rgb evolution is mysterious , as simple models of rgb evolution can not easily explain their miniscule surface hydrogen layers .
models proposed to explain their origin include interactions with a companion ( eg .
@xcite)and even mass stripping by planetary companions @xcite .
the high incidence of composite - spectrum binaries among the sdb stars is consistent with these kinds of models .
other scenarios involving single star evolution include variations of mass loss efficiency on the rgb @xcite , and subsequently experience the core helium flash while on the helium degenerate white dwarf cooling track ( eg .
@xcite ) . in some cases ,
the proposed origin scenario makes a clear prediction about the rotational properties of sdb stars .
for example , those that involve binary mergers or common envelope evolution should spin up the sdb product . while there is no direct spectroscopic evidence because of the signature of chemical diffusion , the field sdb stars appear to be population i objects .
their kinematics suggest that , in general , they derive from a younger population , and therefore we assume in our modeling that their progenitors have masses of 1 @xmath6 or greater . as far as their rotation , comparison with the results for pop .
ii stars suggests that they all should be slow rotators , with @xmath7 10km / s .
recently @xcite measured rotation velocities of a sample of field hb stars showed that a large majority are slow rotators , but that a sample of field hb stars do rotate faster .
the work by @xcite has shown that at least one star , pg 1605 , is a more rapid rotator as expected based on an asteroseismic analysis of the pulsations by @xcite .
attempts to understand the rotation observations , and perhaps link the rotation break to gaps in the horizontal branch or general hb morphology , illustrate the creativity of theorists .
the lack of a coherent answer to date is a testament to the difficulty of the problem .
following up on a previous study of differential rotation in rgb stars @xcite , a detailed investigation by @xcite ( hereafter sp2k ) explored possible solutions based on single star evolution with different assumptions about internal angular momentum transport . as we follow the same general procedure here in our study of sdb stars
, we describe the computations of sp2k in some detail .
sp2k modeled rotation on the red giant branch ( rgb ) and its effect on hb rotation .
starting with models that are rotating as solid bodies at the departure from the main sequence , sp2k followed evolution of the models through the helium flash and onto the hb .
they studied several different limiting cases of angular momentum transport , two of which we follow up on in our study of population i hot hb stars .
both cases conserve specific angular momentum ( angular momentum per unit mass , @xmath8 ) in convectively stable regions . in convective regions ,
convective mixing is assumed to produce either constant angular velocity in the convection zone
( case b ) or constant @xmath8 ( case d ) ; in both cases , the angular momentum contained within the convection zone is preserved .
these designations correspond to those used by sp2k . for low mass population ii stars , sp2k concluded that observations of the rapidly rotating cool hb stars are best understood with models with rapidly rotating cores that evolve from progenitors that have constant specific angular momentum in convection zones ( i.e. case d ) .
case d evolution produces models that can rotate at the rates seen in the cool hb stars .
the slow rotation in the hotter hb stars requires some choking of angular momentum transport by chemical composition gradients produced by diffusion - such effects are seen in hb stars ( and models ) at temperatures above 11,000k ( sp2k ) .
an alternative to sp2k s analysis is given by @xcite , who investigated interaction with a planetary companion as one suggestion for a `` second parameter . ''
@xcite argued that the progenitors of the rapidly rotating hb stars have been spun up by planets or brown dwarfs , masses approximately 5 times jupiter s mass and at separations of about 2 au .
while they state that these planetary interactions are the main second parameter , they also admit that planets do not completely solve the problem .
the models of @xcite rely on horizontal branch angular momentum loss , not ineffective angular momentum transport as in sp2k to explain the slower rotating blue horizontal branch stars .
a subset of sdb stars are known to be multiperiodic pulsating stars . as nonradial modes
must be present to explain the density of pulsation modes in a single star , we may be able to probe the interiors of these stars through asteroseismology .
such work on the ec 14026 stars ( or sdbv stars ) , has already begun .
the general properties of the sdbv stars are reviewed from an observational perspective by @xcite ; theoretical models of these stars are reviewed by @xcite .
models of rotation in single hb stars indicate that these stars should have rapidly rotating cores .
similarly , sdb stars that originate through binary mergers or common envelope evolution should show pathological rotation .
the pulsating sdbv stars , if they represent `` typical '' sdb stars and , by extension , typical hb stars , should allow asteroseismology to address the state of internal rotation of hb stars .
in fact , one sdbv star , pg 1605 was found to have rapid rotation based on the asteroseismic analysis by @xcite this was later partially verified by @xcite through measurement of rotational broadening of spectral lines .
armed with this initial success , it is likely that asteroseismology can be a useful tool to explore important questions of the second parameter problem in globular clusters and the origins of sdb stars in the field .
most single - star models of hb stars that include rotation were developed for pop ii stars .
the blue hb stars in globular clusters are quite old , and originate from stars that are less than 1@xmath6 . given the potential of sdbv stars for probing internal rotation , we therefore have computed new models of rgb stars and their hot hb decendents that are suited for seismic analysis appropriate to the sdbv stars .
the more recently formed single field hb stars such as the sdb stars probably originate from higher mass progenitors ( up to the maximum mass that undergoes the core helium flash ) .
this means that the stars that are now sdb stars could have had much higher initial angular momenta @xcite as well .
therefore , we have computed models of rotating rgb and sdb stars following the same strategy as @xcite , but using higher mass population i progenitors . also , we compute a dense grid of models to adequately sample the various structures that might be relevant for the pulsators .
we then look at the asteroseismic consequences of these rotation profiles , and begin to analyze the data on the pulsating sdb stars to see if they do show rapid core rotation .
this paper proceeds as follows . in the next section ,
we describe our stellar evolution code and our procedures for computing through the rgb and hb .
we also discuss the limiting cases of angular momentum transport .
section 3 outlines the results of these calculations in terms of the surface rotation velocities and internal angular velocity profiles of the models . using these profiles
, we then use section 4 to explore the influence of internal rotation on the observable pulsation frequencies in a preliminary asteroseismological analysis of the models . in section 5
we discuss the results of the asteroseismological analysis in terms of the observed properties of the pulsators .
we conclude in section 6 with a discussion of our results , and an outline for more detailed study of the pulsators .
the evolution of our stellar models was computed using isuevo , the iowa state university evolution code .
isuevo is a standard stellar evolution that is optimized for producing models for use in asteroseismology studies .
isuevo computes models ranging from the zams through the rgb and agb .
we also compute models of white dwarfs and planetary nebula central stars and of course the horizontal branch .
features of isuevo as applied to white dwarfs have been described in @xcite , @xcite , @xcite and @xcite , with further details in @xcite .
isuevo models have also been used for studies of hb stars @xcite and agb stars @xcite .
the constitutive physics used in this code are described in the above references , but we summarize relevant details here .
opacities are taken from the opal compilation for standard mixtures @xcite .
reaction rates utilize cross sections from @xcite .
we follow the standard mixing length treatment with @xmath9 . during
core helium burning on the hb we handle semiconvection in the core using a method similar to @xcite .
as we are modeling population i stars , we assume a metallicity z=0.02 and an initial mixture that is solar .
model sequences were computed with initial masses of 1.0 , 1.5 , and 1.8 @xmath6 , and evolved from the zams until degenerate ignition of helium in the core ( the helium core flash ) .
the core masses at the helium flash for these models are given in table 1 ( measured to the midpoint of the hydrogen
burning shell ) along with a sample of others from the literature . for this paper
, we did not compute the changes in the compositional structure caused by diffusion , and did not include mass loss while on the rgb . as dictated by observations ,
in the future we should probably consider models that deviate from this simple history ; for example , in the case of rgb mass loss and late core helium flash such as computed by @xcite , the core mass of hot horizontal branch stars could be significantly smaller than the standard values .
we did not follow the evolution of the model through the helium flash with our code , as the helium core flash may be a hydrodynamic event @xcite .
rather , we follow the conventional technique of computing a zero age horizontal branch ( zahb ) model that retains the properties of the rgb core at the flash @xcite . the hydrogen shell profile is fit to an analytic function during the construction of the zahb model . for envelopes thinner than 0.0003@xmath6 , we had to reduce the thickness of the hydrogen to helium transition zone to preserve the surface abundance of hydrogen at the pre - flash value .
for most models of interest for our study of sdb stars , the envelope thickness was too small to support an active hydrogen burning shell on the zahb , and the stellar luminosity is derived entirely by the convective helium burning core .
the transport of angular momentum within stars that are evolving up the giant branch is dominated by the effects of convection in the deep convective envelope .
starting from some initial state on or just after the main sequence , the distribution of specific angular momentum @xmath8 would stay relatively constant , affected only by slow diffusive processes ( and perhaps magnetic fields if present ) , in the absence of convection @xcite . in the radiative regions of our models , we assume that @xmath8 is conserved locally , which is roughly equivalent to the shellular rotation described by @xcite .
we note here that the angular velocity profiles in our models represent the extreme case of no diffusion of @xmath8 and that , in real stars , the angular velocity profiles will be similar to or shallower than what we report .
the total angular momentum within radiative zones is also an upper limit , as diffusive transport of @xmath8 usually drains angular momentum from the core into the more slowly spinning envelope . within convection zones
, we follow the procedure of sp2k and examine two limiting cases of @xmath8 transport . the first case
, corresponding to sp2k s case b , is that convective mixing produces constant angular velocity with radius in convection zones , with the total angular momentum of the convection zone being conserved .
there is some observational evidence for this , for example helioseismic observations of the rotation of the solar convection zone show constant differential rotation with depth ( but not necessarily with latitude ) @xcite .
case b models therefore transport @xmath8 outwards , as the outer layers of a convection zone have larger moments of inertia . for rapidly rotating convective regions ( i.e. those where the rotation velocities are comparable to the speed of convective eddys ) this seems like a reasonable approach .
the second case that we consider is comparable to sp2k , case d : the total angular momentum within the envelope convection zone is conserved , and @xmath8 is constant within . in case d evolution , then , the surface convection zone rotates faster in its inner regions where the radius ( and therefore moment of inertia ) is smaller .
in essence , case d treats @xmath8 as if it is a compositional element , mixing thoroughly throughout the convection zone . sp2k point out that such rotation in giant star envelopes is consistent with the much slower rotation velocities found in the convection zones of these stars .
even with our larger initial angular momenta ( see below ) the velocities of rotational motion are much smaller than typical convective velocities of a few km / s .
this suggests that the small amount of differential rotation resulting from the assumption of constant @xmath8 can be maintained by the turbulent mixing generated by convection . with the treatment of angular momentum described above , and
since we do not include the effects of rotation on the hydrostatic structure on the models , we can treat the rotational evolution of the models as a `` side calculation '' during model evolution .
in addition , we can compute the angular momentum distribution in a dimensionless way , by setting the initial model to constant angular velocity of 1 .
subsequent evolution results in changes in @xmath10 and @xmath8 that are proportional to the initial value , allowing us to scale our results with a simple multiplicative factor that corresponds to the initial angular velocity ( or , equivalently , the initial angular momentum ) . for comparison with observations
, we will generally assume that the initial rotation on the main sequence was as a solid body .
the initial rotation rates for our models are taken from @xcite for example , for the 1.8@xmath6 model , @xcite implies @xmath11 km / s , @xmath12 , and @xmath13 . for this preliminary investigation ,
the only angular momentum transport we consider is what is required , ad hoc , to produce the two limiting cases described above .
other mechanisms must act as well - hydrodynamic and thermal instabilities that arise would all serve to transport angular momentum down the angular velocity gradients , effectively reducing the angular velocity contrasts that develop .
these transport mechanisms will act on varying time scales . as long as these time scales are longer than or comparable to the evolutionary time scale on the rgb and hb
, the real angular velocity profiles will resemble the limiting cases above . for reference , the shortest eddington
sweet circulation time scale within these models is approximately @xmath14 to @xmath15 years .
however , given the steep gradients in @xmath16 that these models exhibit , other shear instabilities could act on much shorter time scales .
it is difficult to provide quantitative estimates of these time scales as they vary within the stellar model , and depend on unknown ( and perhaps unknowable ) properties such as the internal magnetic field strengths .
however , for the purposes of this preliminary investigation , we are content to present and discuss the limiting case scenarios for the angular velocity under the two assumptions outlined above , with full acknowledgement that the real stars will not show such extreme behavior .
generally , our results for rgb rotation profiles are very similar to those of sp2k . in all cases ,
evolution up the rgb means an increase in radius ( and moment of inertia ) meaning that the outer layers slow down . at the same time , the core is contracting and therefor spinning up .
thus evolution up the rgb steepens the angular velocity profile .
figure [ rgbrot ] shows the evolution of the angular velocity during this phase for the 1.8@xmath6 case b model ( constant @xmath10 in the surface convection zone ) .
since most of the moment of inertia is in the outermost layers , preserving the bulk angular momentum of the outer convection zone at a constant velocity means that the deepening ( and expanding ) star must spin down greatly .
as shown in this figure , the surface layers rotate nearly 4 orders of magnitude more slowly than the core , reflecting the growth of the star by a factor of nearly 100 in radius .
the case d model ( constant @xmath8 through the surface convection zone ) exhibits very similar behavior overall .
however , the inner parts of the envelope have greater @xmath8 than the previous case ( case b ) , and must rotate faster with their smaller moment of inertia .
figure [ rgbrotjm ] shows the evolution of the angular velocity on the early rgb for the 1.8@xmath6 case d model . in this case , as the convection zone deepens , the material that had a smaller @xmath8 is mixed with the envelope material with higher specific angular momentum , raising the angular velocity at the deepening base . for constant specific angular momentum ,
angular velocity decreases as moment of inertia increases , and the specific moment of inertia increases as radius squared . thus the rotation rate increases with decreasing radius in such a convection zone .
material below the base of the convection zone has slowed by expansion , and an angular velocity kink develops just below the base of the envelope convection zone .
as shown by sp2k , the dominant feature of the internal rotation profile is the change in angular velocity that occurs near @xmath17 from the center of the model .
the rotation profile drops precipitously at this point as the result of draining of angular momentum from the core by the deepening , and then retreating , convective envelope . during early rgb evolution ,
the convective envelope appears and deepens .
it reaches a maximum depth that is modestly dependent on the total stellar mass .
the maxiumum depth that the surface convection zone reaches on the rgb is given in table 1 . clearly , angular momentum transport in the convective envelope is critical in determining the eventual angular velocity profile of the model .
following the helium core flash , the envelope is ejected , and the remnant core settles onto the horizontal branch .
we have assumed that the specific angular momentum is preserved everywhere in the remnant core , and thus the angular velocity profile will reflect changes in the structure of the core as it converts from a degenerate configuration to a nondegenerate , helium burning region . as a degenerate core on the rgb ,
the structure resembles a polytrope with the same polytropic index as the convective helium burning core on the ehb ( i.e. @xcite ) .
therefore , the radius profile ( as a function of mass ) is a power law of the same power but different coefficient . the nondegenerate helium
burning core will have a larger radius , and therefore rotate more slowly ( by the same factor ) in the core than it did just prior to the helium core flash .
figure [ corergbhb ] shows this for the two cases of angular momentum transport in convection zones .
horizontal branch stars evolved from case b rgb models show much smaller surface rotation velocities than those with case d precursors .
these models preserve a rapidly rotating core with roughly constant angular velocity out to about 0.3@xmath6 .
material outside of that point had been drained of angular momentum by inclusion in the convective envelope early during rgb evolution , and so remain slowly rotating on the hb . with such small exterior rotation velocities ,
these stars contain a much larger contrast between the surface and core rotation velocities .
angular velocity drops by almost a factor of @xmath18 , as shown in figure [ omvsrhb ] .
the rotation profiles of models descended from case d rgb evolution contain relics of the high specific angular momentum of material in the outer layers of the rgb progenitor .
when the constant j / m envelope reached down to the 0.3@xmath6 zone , and then retreated , the higher specific angular momentum left behind produced a local maximum of angular velocity that persists into ehb evolution .
the inner core never had a chance to share in this high angular momentum pool , and was left rotating more slowly . on the ehb ,
the core becomes convective , but does not reach the maximum angular velocity material . therefore , the angular velocity profile changes little during ehb evolution - the core does spin down a bit as the angular momentum is mixed , but the inner core continues to spin about a factor of 5 slower than the fastest material .
the angular velocity drops from the maximum by a factor of about 10 or so to the surface .
the general features of the internal rotation profile remain nearly fixed through the stage of core helium burning .
as shown in figure [ omvsrhb ] , the rotation profile ( with radius ) shows the same general features at the start and end of hb evolution . for case b evolution
, a rapidly rotating core is preserved . for case d evolution , the outer decrease in angular velocity is preserved , while the inner , more slowly rotating core , corresponding to the fully convective helium burning zone , spins down at the start of ehb evolution as the core grows and mixes in lower angular momentum material .
further evolution results in little additional angular velocity change with radius .
given the uncertainty about the initial conditions ( e.g. initial angular momentum on the main sequence , and main sequence angular momentum evolution ) , this relatively small change in the angular velocity profile on the ehb could allow efficient parameterization for studies of their seismic influence . given that our plan is to use these models for asteroseismology , the structure of the internal rotation profile is the focus of our study rather than the surface rotation velocities .
still , it is useful to examine the surface rotation rates of these models for general comparison with observation . given the way we construct hb models , the surface rotation velocity of models of different mass reflect the value of @xmath8 at the surface which in turn results from the truncation of the envelope of the rgb progenitor .
for case b evolution , @xmath8 decreases with depth in the rgb progenitor .
hb models with thicker envelopes therefore have surface layers with larger @xmath8 , and faster rotation .
however , since we are interested here in the bluest hb stars , the envelope masses are all less than 0.005@xmath6 .
the change in @xmath8 in the seed model over that small range is less than a factor of 2 . in the case
d models , @xmath8 is constant outside of the core - reflecting the assumptions built into the rgb model - so all of our case d models have the same value of @xmath8 in their surface layers .
the surface angular velocity of the hb models depends on the initial value of @xmath8 at the surface ( since @xmath19 ) .
the surface rotation velocity @xmath20 scales as @xmath21 .
spectroscopic measurements of the sdb stars can determine @xmath22 to adequate precision to compare with our models .
therefore , for a fixed @xmath8 , the rotation velocity should scale with @xmath23 with the constant of proportionality depending on the models and the initial angular momentum .
the rotation velocities for sample models are shown in figure [ logvlogg ] for a variety of initial rotation velocities .
the pulsating sdb stars - known as the ec 14026 stars or simply sdbv stars , are found in the midst of the ( @xmath24,@xmath0 ) plane locus of the `` ordinary '' sdb stars .
they range in @xmath0 of slightly less than 30,000k to over 35,000k , and have @xmath24 between 6.1 and 5.3 .
most lie at the high - gravity end , clustering near the zaehb , with the cooler and lower gravity members numbering about 10% of the total .
figure [ hrdsdbv ] shows an h
r diagram that includes a representative sample of ec 14026 stars and sample evolutionary tracks from our sets of models . given the convergence of the tracks in this plane , it is clear that h r diagram position alone is insufficient to judge their evolutionary state , with stars that are close to the zaehb intermixed with stars that have depleted helium in their cores .
the lowest gravity ec 14026 stars fall sufficiently far from the zaehb that they can not be core helium burners for core masses that are consistent with single star evolution .
thus they are either on their way to the agb or will evolve directly to the white dwarf cooling track in a very short time .
another class of multiperiodic pulsating sdb stars exhibit longer periods ( between 0.3 and 2.5 hours ) .
these stars , colloquially known as the `` betsy stars '' after their discoverer @xcite , lie adjacent to the ec 14026 stars in the h
r diagram at the cool end .
fontaine et al . ( 2003 ) postulate that these are @xmath25-mode pulsators driven by the same mechanism that drives the pulsations in the ec 14026 stars .
the pulsators that lie in the main clump are shown in figure [ hrdsdbv ] surrounded by a box that represents the mean values of @xmath0 and @xmath24 ; the size of the box represents the standard deviation of these means . to explore the influence of a rapidly rotating core on the pulsations of sdbv stars , we employ representative models that fall within this box
many of the high gravity pulsating sdb stars show complex pulsations that require a large number of independent periodicities .
examples include pg 1047 @xcite and pg 0014 @xcite .
the rich pulsation spectra show a mode density that is too high to be explained using only radial pulsations .
nonradially pulsating stars have a large number of available pulsation modes manifested as a large number of possible oscillation frequencies .
even normal nonradial modes have a limited frequency distribution , meaning that some of the pulsators can not be understood without resorting to either high values of @xmath26 ( a large number of nodal lines on the stellar surface ) or to rapid rotation . generally , for nonradial pulsations , each mode samples a slightly different part of the stellar interior , and so we can in principle use the observed pulsations to probe the structure of the stellar interior .
of particular relevance to this paper , nonradial modes are sensitive to stellar rotation - rotation can significantly increase the number of distinct oscillation frequencies seen in pulsating stars by lifting the degenercy of modes with the same degree ( @xmath26 ) but different order ( @xmath27 ) as discussed below . in this section ,
we describe how nonradial pulsations in sdb stars may provide a window into their interiors , and test the hypothesis that horizontal branch stars have rapidly rotating interiors .
the discussion here is intended to be somewhat general so as to illustrate the process by which we can detect rapid rotation of the stellar interior using currently observed sdbv stars .
a future paper will describe our asteroseismic results in more detail and apply the analysis to individual pulsators .
nonradial pulsations in stars are generally expressed in terms of spheroidal modes .
the spectrum of available oscillations are characterised by the eigenfrequency @xmath28 corresponding to a spheroidal mode with quantum numbers @xmath29 , @xmath26 , and @xmath27 .
the values @xmath26 and @xmath27 refer to the spherical harmonic @xmath30 with @xmath26 nodal lines , @xmath27 of which pass through the axis of symmetry .
the sign of @xmath27 gives the direction of propogation of the corresponding surface running wave , and therefore there are @xmath31 possible values of @xmath27 for a given @xmath26 .
the radial order of the mode ( corresponding roughly to the number of nodes in the radial direction ) is @xmath29 . for perfect spherical symmetry ,
the oscillation frequencies are independent of @xmath27 and depend only on @xmath29 and @xmath26 . however , if rotation ( or another non spherical process ) is present , the value of the oscillation frequency will depend on @xmath27 . in the case of slow rotation as a solid body , the @xmath27 dependence of the oscillation frequency can be written as : @xmath32 where the quantity @xmath33 is a function of the spatial structure of the oscillations within the star ( the eigenfunctions ) . for the above case of slow rotation , @xmath34 \xi_h^2 ) \rho r^2 dr}}\ ] ] where the radial and horizontal perturbations at @xmath35 are given by @xmath36 and @xmath37 respectively . for a more complete description of rotational
splitting , see one of the standard texts of stellar pulsation ( i.e. @xcite ) .
equations ( 1 ) and ( 2 ) correspond to solid body rotation with a rotation frequency that is small compared to the oscillation frequency .
the constancy of @xmath10 means that it does not appear inside the integral in equation ( 2 ) , allowing simple determination of the effect of rotation in this special case .
clearly , rotation can split an @xmath38 oscillation into a triplet , spaced in frequency by the rotation frequency times @xmath39 .
sequences of such equally spaced triplets have been clearly identified in a number of pulsating white dwarf stars , for example , suggesting that they rotate nearly as solid bodies , at least in those portions of those stars that contribute to the integral ( 2 ) @xcite . if a star undergoes differential rotation with radius that is , if @xmath40 , then the simple form for calculating the value of @xmath28 above is not appropriate .
the splitting in this case ( where @xmath10 depends on radius , not on latitude ) is given by @xmath41 where @xmath42 \xi_h^2 - 2 \xi_r \xi_h ) \rho r^2 } { \int_0^r{(\xi_r^2 + \ell [ \ell + 1 ] \xi_h^2 ) \rho
r^2 dr}}\ ] ] is the rotation kernel corresponding to a mode with @xmath29 and @xmath26 .
thus for a star with internal differential rotation with radius , the observed splitting for a mode with a given @xmath29 and @xmath26 represents an average of @xmath16 weighted by the rotation kernel @xmath43 .
if the kernel function of a mode is nonzero where the star has very rapid rotation , then the resulting splitting can reveal the inner rapid rotation despite slow rotation at the surface layers .
figure [ casebmkern ] shows sample rotation kernels for an sdb model representative of the pulsating ec 14026 stars .
this particular model is nearing the end of core helium burning as a middle aged sdbv model .
the modes illustrated span the period range seen in the pulsating sdbv stars .
note that higher @xmath29 modes have more peaks , but that in all modes the kernel has nonzero value close to the stellar core . also indicated in figure [ casebmkern ] are lines representing @xmath44 for case b angular momentum transport in the rgb outer convection zone . note that even though the amplitude of @xmath45 is small in the inner regions , the large value of @xmath44 in the same region suggests that the value of the rotational splitting will be dominated by the core rotation . the case for @xmath46 modes
is more striking . for several modes in this model ,
the rotation kernel is quite large in the core .
this likely results from resonant mode trapping by the composition transition zones , in particular the he / c+o zone in the core . at this evolutionary stage
, the model shows some mixed character modes for @xmath46 which can have significant amplitude in the core @xcite .
tables 2 - 7 present the computed splittings for models of different evolutionary stages that lie within the box of figure [ hrdsdbv ] .
the first two tables give periods and splittings for a near - zahb model ; tables 4 and 5 show the same range of periods in a middle - aged sdb model , and tables 5 and 6 show results for a highly evolved sdb model .
tables show the splittings expected if the model is rotating as a solid body at the surface rate , and for models with differential rotation .
the splittings for differential rotation in models with case b are given in tables 2 , 4 , and 6 ; case d models are in tables 3 , 5 , and 7 .
as an illustration of the kind of effects that differential rotation can produce , consider tables 4 and 5 , which present the computed splittings for two models of middle - aged sdb stars . in
the case b model in table 4 , the surface rotation velocity , if representative of the entire star , would produce a splitting of less than @xmath47 for all modes ( surface rotation velocity of only 6 m / s ) , yet the rapid internal rotation results in splittings ranging from 1 to 80 @xmath48 depending on the value of @xmath29 and @xmath26 for the mode .
table 5 shows what happens for a case d model , which has an artificially scaled - down rotation rate of 10 km / s at the surface ( this model would have been rotating way too quickly if it had the initial angular momentum assigned from the main sequence progenitor ) . for this model
, the expected splitting would be about 15 @xmath48 given the surface rate and the assumption of solid body rotation .
the rapidly rotating core , however , produces splittings ranging from 24 to 52 @xmath48 . in both limits ,
the rotational splitting is much larger than expected from assuming solid body rotation at the surface rotation velocity .
even more interesting is the fact that the splitting changes ( sometimes dramatically ) from one mode to the next within a sequence of the same @xmath26 but different @xmath29 .
this reflects the fact that the low order modes in these stars sample portions of the core and envelope , and the exact parts sampled depend on the overtone number . with a steep drop in @xmath10 , the kernel functions ( see figure [ casebmkern ] as an example )
sweep across these fluctuations as @xmath29 increases .
in addition , mode trapping by the composition zones present in these models produces some strong core localization of the kernels for certain modes , and these therefore show larger rotational splitting .
the fact that the expected splittings can be quite large , coupled with the large differences in expected splittings from mode to mode , can make observational identification of rotational splitting quite difficult .
generally , the way this has been done in the past is to look for a sequence of equally spaced triplets in a fourier transform , and then identify them as @xmath38 for example . even if all members of such triplets are not seen , finding several pairs of modes split by the same amount ( or twice that amount , or half )
would be indicative of rotational splitting . for sdbv stars , though , with significant mode to mode differences in splittings expected , a series of rotationally split frequencies would be virtually indistinguishable from modes of different @xmath26 and @xmath29 .
for example , consider the model of an advanced - stage sdb star whose periods are listed in tables 6 and 7 .
the @xmath46 modes have periods that are close to @xmath49 ( radial ) modes . with splittings of 40 to 200 @xmath50hz for the lowest order modes
, the quintuplets @xmath51 and 3 could spread over very large frequency ranges that would encompass , or nearly reach , @xmath52 ; @xmath53 ; and @xmath54 . without appealing to differential rotation , such a rich mode spectrum would require appealing to other mechanisms to explain the periodicities .
we have shown that sdb stars , like horizontal branch stars in general , should retain rapidly rotating cores as a relic of their evolution on the rgb .
such rapid rotation could serve as a reservoir of angular momentum which , when tapped , can produce anomalously fast rotation at the surface on the hb . taken alone ,
our results are consistent with the models of sp2k in that without late angular momentum transport , standard models of single hb stars can not explain observed trends in the distribution of surface rotation velocities with effective temperature . however
, if angular momentum transport can tap the internal reservoir in the absence of composition gradients , the suggestion of sp2k that redder hb stars can spin up in the latter stages of core helium burning remains theoretically viable .
other mechanisms can be at work that would serve to reduce or eliminate the strong differential rotation present in these limiting case models .
also , other scenarios for the origin of sdb stars ( such as binary mergers ) and of sdb surface rotation velocities ( i.e. spin up by nearby planets ) make specific predictions about the internal angular velocity profiles . in any case , though , we can test the predictions of a wide variety of stellar models if we can measure the rotation rate of the subsurface layers of hb stars and , in particular , the hot subdwarfs
. pulsations can probe the interiors of these stars and thereby reveal the rotation velocity profile within .
rapidly rotating cores produce rotational splitting of nonradial modes , and the amount of the splittings is not what would be expected from spectroscopic measurements of @xmath1 .
evidence that we see such effects could be the rotation of pg 1605 as deduced from the pulsations .
the rotatonal splitting identified in that star by @xcite is about 3 times larger than the @xmath1 measured via spectroscopy @xcite . while this could simply be the result of a small inclination angle between the rotation axis of the star and our line of sight
, it could also result from differential rotation and a rapidly rotating core . clearly , it is important to consider the possibly large rotational splittings that might appear in the pulsation spectra of the ec 14026 stars .
identification of modes with large splittings , or of sequences of modes with varying splitting would point towards strong differential rotation . in either case ,
such a measurement would confirm the existence of a rapidly rotating core within these stars .
failure to find such effects would be convincing evidence that angular momentum transport mechanisms on the rgb or during the hb phase are very efficient at wiping out angular velocity gradients .
should that be the case , then the mystery of the rotation of horizontal branch stars will remain with us .
86.45 & 11567 & 0 & 4 & & + 87.30 & 11457 & 2 & 4 & 0.086 & 17.6 + 91.24 & 10963 & 1 & 4 & 0.087 & 7.8 + 97.25 & 10283 & 0 & 3 & & + 98.48 & 10155 & 2 & 3 & 0.084 & 19.8 + 105.5 & 9480 & 1 & 3 & 0.086 & 8.6 + 118.6 & 8434 & 0 & 2 & & + 121.3 & 8246 & 2 & 2 & 0.081 & 25.9 + 130.1 & 7685 & 1 & 2 & 0.087 & 2.8 + 136.9 & 7302 & 0 & 1 & & + 138.2 & 7237 & 2 & 1 & 0.084 & 11.2 + 144.3 & 6930 & 1 & 1 & 0.087 & 2.1 + 158.5 & 6309 & 0 & 0 & & + 168.6 & 5930 & 2 & 0 & 0.059 & 91.4 + 84.73 & 11800 & 0 & 4 & & + 85.50 & 11699 & 2 & 4 & 18.0 & 29.3 + 89.30 & 11198 & 1 & 4 & 18.1 & 32.2 + 95.88 & 10430 & 0 & 3 & & + 97.22 & 10287 & 2 & 3 & 17.6 & 32.0 + 104.5 & 9574 & 1 & 3 & 17.9 & 39.8 + 117.1 & 8540 & 0 & 2 & & + 119.5 & 8372 & 2 & 2 & 17.1 & 30.2 + 126.9 & 7880 & 1 & 2 & 18.1 & 28.9 + 133.3 & 7504 & 0 & 1 & & + 134.9 & 7414 & 2 & 1 & 17.4 & 25.0 + 142.4 & 7023 & 1 & 1 & 18.1 & 34.8 + 157.6 & 6344 & 0 & 0 & & + 167.9 & 5956 & 2 & 0 & 12.3 & 47.3 + 82.37 & 12140 & 0 & 6 & & + 84.20 & 11876 & 2 & 6 & 0.051 & 13.6 + 87.85 & 11384 & 1 & 6 & 0.052 & 7.1 + 91.14 & 10972 & 0 & 5 & & + 93.21 & 10728 & 2 & 5 & 0.051 & 12.6 + 98.65 & 10138 & 1 & 5 & 0.052 & 7.5 + 105.6 & 9467 & 0 & 4 & & + 108.3 & 9238 & 2 & 4 & 0.050 & 13.0 + 113.9 & 8784 & 1 & 4 & 0.052 & 3.2 + 118.8 & 8417 & 0 & 3 & & + 121.9 & 8206 & 2 & 3 & 0.050 & 12.2 + 130.3 & 7677 & 1 & 3 & 0.052 & 3.5 + 141.8 & 7052 & 0 & 2 & & + 147.6 & 6774 & 2 & 2 & 0.046 & 24.7 + 161.4 & 6196 & 1 & 2 & 0.052 & 1.6 + 167.2 & 5980 & 0 & 1 & & + 171.5 & 5832 & 2 & 1 & 0.049 & 16.9 + 179.4 & 5573 & 1 & 1 & 0.052 & 0.9 + 185.9 & 5379 & 0 & 0 & & + 191.1 & 5232 & 2 & 0 & 0.044 & 75.9 + 82.65 & 12099 & 0 & 6 & & + 83.33 & 11999 & 2 & 6 & 15.3 & 31.0 + 86.89 & 11508 & 1 & 6 & 15.5 & 31.5 + 91.45 & 12099 & 0 & 5 & & + 92.26 & 10839 & 2 & 5 & 15.3 & 30.2 + 97.68 & 10239 & 1 & 5 & 15.4 & 38.7 + 106.0 & 9435 & 0 & 4 & & + 107.2 & 9331 & 2 & 4 & 15.1 & 29.1 + 112.6 & 8881 & 1 & 4 & 15.5 & 29.5 + 119.1 & 8390 & 0 & 3 & & + 120.6 & 8287 & 2 & 3 & 14.9 & 27.3 + 129.0 & 7753 & 1 & 3 & 15.5 & 36.5 + 142.3 & 7026 & 0 & 2 & & + 146.5 & 6828 & 2 & 2 & 13.9 & 31.4 + 159.7 & 6261 & 1 & 2 & 15.5 & 30.4 + 167.8 & 5961 & 0 & 1 & & + 169.8 & 5890 & 2 & 1 & 14.6 & 23.6 + 177.5 & 5633 & 1 & 1 & 15.5 & 27.3 + 186.6 & 5360 & 0 & 0 & & + 190.7 & 5243 & 2 & 0 & 12.7 & 52.0 + 83.49 & 11977 & 0 & 7 & & + 83.75 & 11934 & 2 & 7 & 0.042 & 9.3 + 87.29 & 11453 & 1 & 7 & 0.043 & 5.7 + 93.48 & 10697 & 0 & 6 & & + 94.09 & 10626 & 2 & 6 & 0.042 & 11.6 + 97.43 & 10258 & 1 & 6 & 0.043 & 2.4 + 102.9 & 9716 & 0 & 5 & & + 104.8 & 9545 & 2 & 5 & 0.040 & 23.6 + 111.7 & 9418 & 1 & 5 & 0.043 & 2.0 + 115.7 & 8642 & 0 & 4 & & + 116.7 & 8567 & 2 & 4 & 0.041 & 17.8 + 125.9 & 7943 & 1 & 4 & 0.043 & 3.0 + 134.5 & 7438 & 0 & 3 & & + 135.8 & 7364 & 2 & 3 & 0.039 & 38.6 + 142.0 & 7041 & 1 & 3 & 0.043 & 0.9 + 147.0 & 6804 & 0 & 2 & & + 147.4 & 6784 & 2 & 2 & 0.038 & 103 + 173.5 & 5765 & 2 & 1 & 0.043 & 17.7 + 174.9 & 5716 & 1 & 2 & 0.043 & 0.2 + 176.3 & 5673 & 0 & 1 & & + 192.8 & 5188 & 2 & 0 & 0.034 & 210 + 85.23 & 11800 & 0 & 7 & & + 85.46 & 11701 & 2 & 7 & 13.8 & 28.4 + 88.48 & 11302 & 1 & 7 & 13.9 & 39.6 + 94.79 & 10257 & 0 & 6 & & + 95.73 & 10444 & 2 & 6 & 13.5 & 33.0 + 99.48 & 10052 & 1 & 6 & 14.1 & 30.7 + 104.0 & 9615 & 0 & 5 & & + 105.7 & 9463 & 2 & 5 & 12.9 & 38.6 + 113.8 & 8791 & 1 & 5 & 14.0 & 36.5 + 118.1 & 8467 & 0 & 4 & & + 118.9 & 8414 & 2 & 4 & 13.3 & 28.6 + 126.7 & 7895 & 1 & 4 & 13.9 & 46.1 + 135.8 & 7364 & 0 & 3 & & + 137.8 & 7257 & 2 & 3 & 12.4 & 49.7 + 144.9 & 6902 & 1 & 3 & 14.0 & 26.3 + 148.0 & 6758 & 2 & 2 & 12.9 & 64.5 + 148.2 & 6748 & 0 & 2 & & + 177.5 & 5634 & 2 & 1 & 14.1 & 30.3 + 179.3 & 5577 & 1 & 2 & 14.0 & 21.1 + 180.7 & 5534 & 0 & 1 & & + 194.0 & 5154 & 2 & 0 & 11.3 & 137.7 + | observations of the rotation rates of horizontal branch ( hb ) stars show puzzling systematics .
in particular , cooler hb stars often show rapid rotation ( with velocities in excess of 10 km / s ) , while hotter hb stars ( those with @xmath0 in excess of 11,000k ) typically show much smaller rotation velocities of 8 km / s or less .
simple models of angular momentum evolution of stars from the main sequence through the red giant branch fail to explain these effects .
assumption of solid body rotation throughout rotate much too slowly . on the other hand , assuming local conservation of angular momentum , but with solid body rotation in the convective regions , produces hb models also rotate too slowly at all @xmath0 but preserve a rapidly rotating core . in these cases , the observed angular velocities of hb stars require that some of the angular momentum stored in the core reaches the surface .
models that assume constant specific angular momentum in the surface convection zone have faster rotation in the envelope ; while the predictions of these models match the observed rotation rates of the cooler hb stars , hot hb stars rotate more slowly than the models .
thus , while there is not as yet a coherent explanation of the trends of rotation on the hb , evolutionary models in all cases preserve a rapidly rotating core . to test the idea that hb stars contain such a core
, one can appeal to detailed computations of trace element abundences and rotational mixing .
however , a more direct probe is available to test these limiting cases of angular momentum evolution . some of the hottest horizontal branch stars are members of the pulsating sdb class .
they frequently show rich pulsation spectra characteristic of nonradially pulsating stars .
thus their pulsations probe the _ internal _ rotation of these stars , and should show the effects of rapid rotation in their cores . using models of sdb stars that include angular momentum evolution ,
we explore this possibility and show that some of the sdb pulsators may indeed have rapidly rotating cores . |
microsoft kinect@xmath0 ( hereafter , simply ` kinect ' ) @xcite , a low - cost , portable motion - sensing hardware device , was developed by the microsoft corporation ( microsoft , usa ) as an accessory to the xbox @xmath5 video - game console ( 2010 ) .
the sensor is a webcamera - type , add - on peripheral device , enabling the operation of xbox via gestures and spoken commands . in 2011 , microsoft released the software - development kit ( sdk ) for kinect , thus enabling the development of applications in several standard programming languages .
the first upgrade of the sensor ( ` kinect for windows v2 ' ) , both hardware- and software - wise , tailored to the needs of xbox one , became available for general development and use in july 2014 .
the present paper is part of a broader research programme , aiming at involving either sensor in the analysis of motion data of subjects walking or running ` in place ' ( e.g. , on a commercially - available treadmill ) . if successful , kinect could become an interesting alternative to marker - based systems ( mbss ) in capturing data for motion analysis , one with an incontestably high benefit - to - cost ratio .
regarding medical applications of this sensor , e.g. , in physiotherapy in the home environment , a number of products are available , e.g. , by ` home team ' ( massachusetts , usa , https://www.hometeamtherapy.com ) and ` reflexion ' ( california , usa , http://www.reflexionhealth.com ) . in particular , it is known that ` reflexion ' aims at increasing the success rates in rehabilitation ; the participation of the us navy in the tests of their product attests to the importance of the availability of such solutions .
the validation of the output of the original kinect sensor in static measurements or in case of slow movements was a popular subject in the recent past .
data acquired from @xmath6 healthy subjects , performing three simple tasks ( forward reach , lateral reach , and single - leg , eyes - closed standing balance ) were analysed in ref . @xcite . in their work , the authors drew attention to systematic effects in the kinect output , in particular for the sternum .
even more optimistic were the results obtained ( and the conclusions drawn ) in ref .
@xcite , which investigated the accuracy in the determination of the joint angles from data acquired with an original kinect sensor and an mbs ; an inclinometer was assumed to provide the reference or baseline solution ( also called ` ground truth ' ) .
the authors concluded that the differences in accuracy and reliability between the two measurement systems were small , thus enabling the use of kinect as `` a viable tool for calculating lower extremity joint angles '' .
in fact , the kinect results and those obtained with the inclinometer were found to agree to better than @xmath7 .
interestingly , the authors also made a point regarding the depth measurements of the kinect sensor , which are subject to increasing uncertainty with increasing distance from the sensor , reaching a maximal value of about @xmath8 cm at the most distal position of the sensor , which ( according to the specifications ) should not exceed about @xmath9 m. naturally , such a dependence introduces bias in the analysis of walking- and running - motion data acquired with a treadmill .
these effects are present regardless of the largeness of the subject s stray motion in depth ; for example , the depth uncertainties are different at the two extreme lower - leg positions , i.e. , ahead of and behind the walker / runner .
another medical application of kinect was investigated in ref .
@xcite , namely its use for home - based assessment of movement symptoms in patients with parkinson s disease . in that study ,
a number of tasks were performed by @xmath10 subjects , ten of which comprised the control group ; parallel data were acquired with kinect and an mbs .
the authors reported that the kinect results were generally ( but not in all cases ) found to correlate well with those obtained from the mbs for a variety of movements . regarding the use of kinect in medical / health - related applications , complacency and optimism were impaired after the paper of bonnech@xmath11re @xcite appeared .
the authors recorded data from @xmath12 subjects , performing four simple tasks ( shoulder abduction , elbow flexion , hip abduction , and knee flexion ) , and compared the results of different sessions pursuing both their reproducibility within each measurement system , as well as an assessment of the differences between the two systems .
the authors concluded that the lower body is not tracked well by kinect .
the conclusions of ref .
@xcite constitute rather disturbing news in terms of applications of the sensor in monitoring walking and running behaviour in a medical / health - related environment .
the literature on the biomechanics of motion is extensive .
some selected works , relevant to the present study , include refs .
earlier scientific works are cited therein , in particular in the review articles @xcite . * cavagna and kaneko @xcite studied the efficiency of motion in terms of the mechanical work done by the subject s muscles . *
cavanagh and lafortune @xcite studied the ground reaction forces in running at about @xmath13 km / h , as well as the motion of the ` centre of the pressure distribution ' during the stance phase of the right foot of @xmath14 subjects .
relatively large variability is seen in their results , partly due to the different characteristics in the motion of rear - foot and mid - foot strikers , partly reflecting the extent of their database in terms of running experience , weekly training , and ( perhaps , more importantly ) habitual individual behaviour .
the vertical component of the ground reaction force , which may be as large as three times the subject s body weight , showed sizeable variability .
* cairns , burdett , pisciotta , and simon @xcite analysed the motion of ten competitive race - walkers in terms of the ankle flexion , of the knee and hip angles , as well as of the pelvic tilt , obliquity , and rotation .
the work discussed the main differences between walking and race - walking , and provided explanations for the peculiarity of the motion in the latter case , invoking the goal of achieving higher velocities ( than in normal walking ) while maintaining double support with fully - extended knee and suppressing the vertical undulations of the subject s centre of mass ( cm ) .
* unpuu @xcite discussed important aspects of the biomechanics of gait , including the variation of relevant physical quantities within the gait cycle . that work may be used as a starting point for those in seek of an overview in the topic .
it must be borne in mind that the subjects used in ref .
@xcite were children .
* novacheck @xcite also provided an introduction to the biomechanics of motion .
5 and 6 of that work contain the variation of the important angles ( projections on the coronal , sagittal , and transverse planes ) within the gait cycle , at three velocities : @xmath15 km / h ( walking ) , @xmath16 km / h ( running ) , and @xmath17 km / h ( sprinting ) .
9 therein provides the variation of the joint moments and powers ( kinesics ) in the sagittal plane within the gait cycle . * in a subsequent article @xcite , schache , bennell , blanch , and wrigley investigated the inter - relations in the movement of the lumbar spine , pelvis , and hips in running , aiming at optimising the rehabilitation process in case of relevant injuries .
it is rather surprising that only one study , addressing the possibility of involving kinect in the analysis of walking and running motion ( i.e. , not in a static mode or in slow motion ) , has appeared so far @xcite . using similar methodology to the one proposed herein
, the authors in this study came to the conclusion that the original sensor is unsuitable for applications requiring high precision ; after analysing preliminary data , we have come to the same conclusion .
of course , it remains to be seen whether any improvement ( in the overall quality of the output ) can be obtained with the upgraded sensor .
our aim in the present paper is to develop the theoretical background required for the comparison of the output of two measurement systems used ( or intended to be used ) in the analysis of human motion ; we will give all important definitions and outline meaningful tests .
although this methodology has been developed for a direct application in the case of the kinect sensors , other applications may use this scheme of ideas in order to obtain suitable solutions in other cases .
the tests we propose in section [ sec : tests ] should be sufficient to identify the important differences in the output of two such measurement systems .
as such , they should ( in a comparative study ) pinpoint the essential differences in the performance of the two kinect sensors or ( if the second measurement system is an mbs ) enable the validation of the kinect sensors .
the material in the present paper has been organised as follows . in section [ sec : systems ] , the output of the two kinect sensors is described ; subsequently , the output , obtained with one popular marker - placement scheme from an mbs , is detailed .
a scheme of association of these two outputs is developed .
the definitions of important quantities , used in the description of the motion , are given in section [ sec : method ] .
section [ sec : acquisition ] describes one possibility for the data acquisition ; in the second part of this section , we explain how one may extract characteristic forms ( waveforms ) from the motion data , representative of the subject s motion within one gait cycle . in section
[ sec : tests ] , we outline our proposal for the necessary tests , to be performed on the waveforms obtained in the previous section .
the last part contains a short summary of the paper and outlines two directions in future research .
to enable the analysis of the data with the same software application , the mbs output , obtained for the specific marker - placement scheme described in subsection [ sec : pig ] , will be transformed into kinect - output format , using reasonable associations between the kinect nodes and the marker locations ; due to the removal of the constant offsets in the data analysis ( see subsection [ sec : analysis3 ] ) , the exact matching between the kinect nodes and the locations at which these markers are placed is not essential . in the original kinect sensor , the skeletal data
( ` stick figure ' ) of the output comprises @xmath6 time series of three - dimensional ( 3d ) vectors of spatial coordinates , i.e. , measurements of the ( @xmath18,@xmath19,@xmath20 ) coordinates of the @xmath6 nodes which the sensor associates with the axial and appendicular parts of the human skeleton . in coronal ( frontal )
view of the subject ( sensor view ) , the kinect coordinate system is defined with the @xmath18 axis ( medial - lateral ) pointing to the left ( i.e. , to the right part of the body of the subject being viewed ) , the @xmath19 axis ( vertical ) upwards , and the @xmath20 axis ( anterior - posterior ) away from the sensor , see fig .
[ fig : kinect ] . the nodes @xmath21 to @xmath8 are main - body nodes , identified as hip_center , spine , shoulder_center , and head .
the nodes @xmath1 to @xmath2 relate to the left arm : shoulder_left , elbow_left , wrist_left , and hand_left ; similarly , the nodes @xmath22 to @xmath23 on the right arm are : shoulder_right , elbow_right , wrist_right , and hand_right .
the eight remaining nodes pertain to the legs , the first four to the left ( hip_left , knee_left , ankle_left , and foot_left ) , the remaining four to the right ( hip_right , knee_right , ankle_right , and foot_right ) leg of the subject
. the nodes of the original sensor may be seen in fig .
[ fig : nodeskinect ] . in the upgraded kinect sensor ,
some modifications have been made in the naming ( and placement ) of some of the nodes .
the original node hip_center has been replaced by spine_base ( and appears slightly shifted downwards ) ; the original node spine has been replaced by spine_mid ( and appears slightly shifted upwards ) ; finally , the original node shoulder_center has been replaced by neck ( and also appears slightly shifted upwards ) .
five new nodes have been appended at the end of the list ( which was a good idea , as this action enables easy adaption of the analysis code processing the kinect output ) , one of which is a body node ( spine_shoulder , node @xmath24 ) , whereas four nodes pertain to the subject s hands , hand_tip_left ( @xmath25 ) , thumb_left ( @xmath26 ) , hand_tip_right ( @xmath27 ) , and thumb_right ( @xmath28 ) .
evidently , emphasis in the upgraded sensor is placed on the orientation of the subject s hands ( i.e. , on gesturing ) . in both versions , parallel to the captured video image
, kinect acquires an infrared image , generated by the infrared emitter ( seen on the left of the original sensor in fig .
[ fig : kinect ] ) ; captured with a ccd camera , this infrared image provides the means of extracting information on the depth @xmath20 .
the sampling rate in the kinect output ( for the video and the skeletal data , for both versions of the sensor ) is @xmath29 hz .
the description of the algorithm , used in the determination of the 3d positions of the skeletal joints of the subject being viewed by the original sensor , may be found in ref .
candidate values for the 3d positions of each skeletal joint are obtained via the elaborate analysis of each depth image separately .
these positions may be used as starting points in an analysis featuring the temporal and kinematic coherence in the subject s motion ; it is not clear whether such a procedure has been hardcoded in the preprocessing ( hardware processing ) of the captured data .
shotton define @xmath30 body segments covering the human body , some of which are used in order to localise skeletal joints , some to fill the gaps or yield predictions for other joints . in the development of their algorithm ,
shotton generated static depth images of humans ( of children and adults ) in a variety of poses ( synthetic data ) .
the application of their method results in the extraction of probability - distribution maps for the 3d positions of the skeletal joints ; their joint proposals represent the modes ( maxima ) in these maps . according to the authors , the probability - distribution maps are both accurate and stable , even without the imposition of temporal or kinematic constraints .
it must be borne in mind that the ` 3d positions of the joints ' of ref .
@xcite are essentially produced from the ` 3d positions of the projections of the joints onto the front part of the human body ' after applying a ` shift ' in depth ( i.e. , from the surface to the interior of the human body ) , namely a constant offset ( @xmath31 ) of @xmath32 mm ( see end of section 3 of ref .
@xcite ) . although the ` computational efficiency and robustness ' of the procedure are praised in ref .
@xcite , it remains to be seen whether results of similar quality can be obtained in dynamic applications ( e.g. , when the subject is in motion ) .
featuring several cameras , viewing the subject from different directions , mbss provide powerful object - tracking solutions , yielding high - quality , low - latency data , at frame rates exceeding that of the kinect sensors .
such systems reliably reconstruct the time series of the spatial coordinates of markers ( reflective balls , flat markers , active markers , etc . )
directly attached to the subject s body or to special attire worn by the subject .
one popular placement scheme of the markers , known as ` plug - in gait ' @xcite , uses a total of @xmath32 markers ( see table [ tab : mbsmarkers ] ) . the mbs output for these markers
may be transformed into kinect - output format ( for simplicity , we refer to the naming of the nodes in the original kinect sensor ) by using the following association scheme . *
the kinect - equivalent head is assigned to the midpoint of the marker positions lfhd and rfhd .
the marker positions lbhd and rbhd , pertaining to the back of the head , are not used . *
the kinect - equivalent shoulder_center is taken to be the marker position clav .
the marker positions c7 and rbak , which are placed on the back part of the body , are not used in comparisons with data acquired with the original kinect sensor ; in the upgraded kinect sensor , spine_shoulder may be identified with c7 . *
the kinect - equivalent spine is estimated as an average of the marker positions t10 , lpsi , and rpsi . *
the kinect - equivalent shoulder_left and shoulder_right are taken to be the marker positions lsho and rsho , respectively . regarding the upper part of the body , the marker positions lupa , lfra , rupa , and rfra
are not used .
* the kinect - equivalent elbow_left and elbow_right are taken to be the marker positions lelb and relb , respectively . *
the kinect - equivalent wrist_left and wrist_right are assigned to the midpoints of the marker positions lwra and lwrb , and of rwra and rwrb , respectively .
* the kinect - equivalent hand_left and hand_right are taken to be the marker positions lfin and rfin , respectively . *
the kinect - equivalent knee_left and knee_right are taken to be the corrected ( according to ref .
@xcite ) marker positions lkne and rkne , respectively . *
the kinect - equivalent ankle_left and ankle_right are taken to be the corrected ( according to ref .
@xcite ) marker positions lank and rank , respectively .
* the kinect - equivalent foot_left and foot_right are taken to be the marker positions ltoe and rtoe , respectively . *
the kinect - equivalent hip_left and hip_right positions are evaluated from those of the marker positions lasi , rasi , lpsi , and rpsi , according to ref .
@xcite . regarding the procedure set forth in that paper , a few comments are due .
the positions of the hips are obtained therein using a model for the geometry of the pelvis , featuring three parameters ( @xmath33 , @xmath34 , and @xmath35 ) , the values of which had been obtained from a statistical analysis of radiographic data of @xmath28 subjects ; however , the values of these parameters are poorly known ( see page 583 of ref .
@xcite ) . a simple analysis of the uncertainties given in ref .
@xcite shows that , when following that method , the resulting uncertainties in the estimation of the positions of the hips are expected to exceed about @xmath36 mm in each spatial direction . as a result
, the positions of the hips , calculated from the mbs output according to that procedure , should not be considered as accurate as the rest of the information obtained from the mbs .
more importantly , it is not evident how the movement of the pelvis reflects itself in the motion of the four markers which are used in the extraction of its position and orientation ; it is arguable whether any markers , placed on the surface of the human body , can capture the pelvic motion accurately . *
the kinect - equivalent hip_center is estimated as an average of the kinect - equivalent hip_left and hip_right , and of the marker position strn . * regarding the lower part of the body , the marker positions lthi , ltib , lhee , rthi , rtib , and rhee
are not used . in regard
to the markers placed on the human extremities , it must be borne in mind that their positions are also affected by rotations , not only by the translational motion of these extremities ; the markers are placed at some distance from the actual rotation axes , coinciding with the longest dimension of the upper- and lower - extremity bones . for instance , rotating the left humerus by @xmath37 around its long axis ( assumed , for the sake of the argument , to align with the vertical axis @xmath19 ) will result in a movement of the marker lelb along a circular arc , thus affecting its @xmath18 and @xmath20 coordinates . on the other hand , the kinect nodes are rather placed _ on _ ( or , in any case , closer to ) the rotation axes ; as a result
, it is expected that they are less affected by such rotations . as such effects
can not be easily accounted for , it is evident that the association scheme , proposed in the present section , can only lead to an approximate comparison of the output of the two measurement systems .
we will next describe how one may obtain estimates of three important angles in the sagittal plane , representing the level of flexion of the trunk , of the hip , and of the knee .
estimates for the left and right parts of the body will be obtained for the hip and knee angles . * * trunk angle*. this angle is obtained from the ( @xmath19,@xmath20 ) coordinates of four points , comprising the nodes @xmath21 ( hip_center ) , @xmath38 ( shoulder_center ) , and two midpoints , namely of the nodes @xmath39 ( hip_left ) and @xmath14 ( hip_right ) , and of the nodes @xmath1 ( shoulder_left ) and @xmath22 ( shoulder_right ) .
an unweighted least - squares fit on the ( @xmath19,@xmath20 ) coordinates of these four points ( spine ) , this node should not be included in estimations involving the @xmath20 coordinate . ]
yields the slope @xmath40 ( with respect to the @xmath19 axis ) of the optimal straight line .
the trunk angle is defined as @xmath41 ; @xmath42 in the upright position , positive for forward leaning . * * hip angle*. two definitions of the hip angle have appeared in the literature : the angle may be defined with respect to the trunk or to the @xmath19 axis ; in the present paper , we adopt the latter definition .
if the relevant hip coordinates are ( @xmath43,@xmath44 ) and those of the knee are ( @xmath45,@xmath46 ) , the hip angle is obtained via the expression : @xmath47 two hip angles will be obtained : the left - hip angle @xmath48 uses the nodes @xmath39 ( hip_left ) and @xmath49 ( knee_left ) ; the right - hip angle @xmath50 uses the nodes @xmath14 ( hip_right ) and @xmath51 ( knee_right ) . *
* knee angle*. this is the angle between the femur ( thigh ) and the tibia ( shank ) .
two definitions of the knee angle have appeared in the literature : the knee angle may be @xmath52 or @xmath53 in the extended position of the knee ; we adopt the latter definition .
it will shortly become clear why we make use of both the sine and the cosine of the knee angle : @xmath54 and @xmath55 where the coordinates of the ankle are denoted as ( @xmath56,@xmath57 ) , and @xmath58 and @xmath59 are the projected lengths of the femur and the tibia onto the sagittal plane , respectively : @xmath60 and @xmath61 we define the knee angle as : @xmath62 two knee angles will be obtained : the left - knee angle @xmath63 uses the nodes @xmath39 ( hip_left ) , @xmath49 ( knee_left ) , and @xmath64 ( ankle_left ) ; the right - knee angle @xmath65 uses the nodes @xmath14 ( hip_right ) , @xmath51 ( knee_right ) , and @xmath10 ( ankle_right ) .
we define four angles in the coronal plane : the lateral trunk , the lateral hip , the lateral knee , and the lateral pelvic angles ; the lateral pelvic angle is also called pelvic obliquity .
estimates for the left and right parts of the body will be obtained for the lateral hip and lateral knee angles . * * lateral trunk angle*. the same four points , which had been used in the evaluation of the trunk angle in the sagittal plane , are also used in extracting an estimate of the lateral trunk angle ; of course , the ( @xmath18,@xmath19 ) coordinates of these points must be used now .
in addition to these nodes , node @xmath66 ( spine ) may also be used .
the lateral trunk angle is defined with respect to the @xmath19 axis ; @xmath67 in the upright position , positive for tilting in the positive @xmath18 direction ( tilt of the subject to his / her right ) .
* * lateral hip angle*. this angle describes hip abduction / adduction in the coronal plane .
similarly to the hip angle in the sagittal plane , two definitions of the lateral hip angle are possible : the angle may be defined with respect to the trunk or to the @xmath19 axis ; herein , we adopt the latter definition .
if the relevant hip coordinates are ( @xmath68,@xmath43 ) and those of the knee are ( @xmath69,@xmath45 ) , the lateral hip angle is obtained via the expression : @xmath70 two lateral hip angles will be obtained : the lateral left - hip angle @xmath71 uses the nodes @xmath39 ( hip_left ) and @xmath49 ( knee_left ) ; the lateral right - hip angle @xmath72 uses the nodes @xmath14 ( hip_right ) and @xmath51 ( knee_right ) . * * lateral knee angle*. this is the projection of the angle between the femur and the tibia onto the coronal plane .
@xmath73 where @xmath58 and @xmath59 are now redefined as the projected lengths of the femur and the tibia onto the coronal plane , respectively : @xmath74 and @xmath75 the angle is defined positive when , with respect to the femur direction , the ankle appears ( in coronal view ) ` further away ' from the subject s body .
of course , two lateral knee angles may be defined , corresponding to the left and right parts of the human body , @xmath76 and @xmath77 , respectively . *
* pelvic obliquity*. this angle is defined as : @xmath78 where ( @xmath79,@xmath80 ) and ( @xmath81,@xmath82 ) are the ( @xmath18,@xmath19 ) coordinates of the left and right hips , respectively . in regard to motion analysis
, a few additional angles may be found in the literature : the pelvic tilt and the angle describing the plantarflexion / dorsiflexion of the foot are defined in the sagittal plane ; the hip , pelvic , and foot rotations in the transverse plane .
we do not believe that the kinect output can yield reliable ( if any ) information on these quantities .
the knee angle , obtained from the 3d vectors ( @xmath83,@xmath84,@xmath85 ) and ( @xmath86,@xmath87,@xmath88 ) , will be called ` knee angle in 3d ' ; it is easily evaluated using expressions analogous to eqs .
( [ eq : eq02])-([eq : eq04 ] ) . in view of the fact that the angle between 3d vectors is invariant under rotations ( so(3 ) rotation group ) and translations in 3d ,
the knee angle in 3d is independent of the details regarding the alignment between the relevant coordinate systems ( e.g. , between the kinect sensor and the mbs coordinate systems ) .
two last comments are due . 1 .
the trunk angle @xmath89 is positive in walking and running ; it is difficult to maintain balance if one leans backwards while moving forwards .
however , the trunk angle , obtained from the kinect output , is frequently negative .
this is due to the fact that the nodes of the kinect output , which enter the evaluation of @xmath89 , do not represent locations on the spine .
2 . due to the properties of the knee joint
, the knee angle is expected to satisfy the condition @xmath90 . in practice , even in the fully - extended position , @xmath91 remains ( for many subjects ) positive ; knee hyperextension is a deformity .
however , owing to the placement of the nodes by kinect , the knee angle ( estimated from the kinect output ) may occasionally come out negative . to examine further such cases ,
we retain eq .
( [ eq : eq04 ] ) in the evaluation of the knee angle .
one possibility to avoid these effects is to extract robust measures for the selected physical quantities from the data .
for instance , one could use the variation of these quantities within the gait cycle or even their range of motion ( rom ) , i.e. , the difference between the maximal and minimal values within the gait cycle .
as long as an extremity moves as one rigid object , such measures ( being differences of two values ) are not affected by a constant bias which may be present in the data .
we propose that the similarity of corresponding waveforms ( representing the variation of a quantity within the gait cycle , see subsection [ sec : analysis3 ] ) be judged on the basis of one ( or more ) of the following scoring options : pearson s correlation coefficient , the zilliacus error metric , the rms error metric , whang s score , and theil s score .
assuming that a ( @xmath92-centred ) waveform from measurement system @xmath21 ( e.g. , from one of the kinect sensors ) is denoted by @xmath93 and the corresponding ( @xmath92-centred ) waveform from measurement system @xmath66 ( e.g. , from the mbs ) by @xmath94 , these five scoring options are defined in eqs .
( [ eq : eq08])-([eq : eq12 ] ) ( for details on the original works , see ref .
@xcite ) ; all sums are taken from @xmath95 to @xmath96 , where @xmath96 stands for the number of bins used in the histograms yielding these waveforms .
( in our analyses , we normally use @xmath97 . )
@xmath98 @xmath99 @xmath100 @xmath101 @xmath102 in case of identical waveforms ( from the two measurement systems ) , @xmath103 ; all other scores vanish ( @xmath104 ) . evidently , whang s score is the symmeterised version of the zilliacus error metric , whereas theil s score is the symmeterised version of the rms error metric .
although the differences between the zilliacus and the rms error metric are generally small ( as are those between whang s and theil s scores ) , we make use of all aforementioned scoring options in our research programme . other ways for testing the similarity of the output of different measurement systems have been put forth .
for instance , some authors favour the use of the ` coefficient of multiple correlation ' ( cmc ) @xcite .
ferrari , cutti , and cappello @xcite define the cmc as : @xmath105^{1/2 } \ , \ , \ , , \ ] ] where the triple array @xmath106 contains the entire data , i.e. , @xmath107 waveforms of dimension @xmath96 ( @xmath96 depends on the gait cycle in ref .
@xcite ) ; @xmath108 is the number of measurement systems being used in the study ( ` protocols ' , in the language of ref .
@xcite ) and @xmath109 denotes the number of waveforms obtained within each measurement system . the averages @xmath110 and @xmath111 in eq .
( [ eq : eq13 ] ) are defined as : @xmath112 @xmath113 unlike pearson s correlation coefficient , ` directional information ' for the association between the tested quantities is lost when using the cmc in an analysis . in its first definition @xcite
, the cmc was bound between @xmath92 and @xmath21 .
however , the quantity cmc , obtained with eq .
( [ eq : eq13 ] ) , is frequently imaginary ( the ratio of the triple sums may be larger than @xmath21 ) ; this is due to the use of @xmath111 , instead of the grand mean ( along with the normalisation factor @xmath114 , instead of @xmath115 ) , in the denominator of the expression . importantly ,
it is unclear how the obtained cmc values relate to the goodness of the association between the tested waveforms .
the association scheme of ref .
@xcite is arbitrary ; there is no theoretical justification for such an interpretation of the cmc results . the basic problem in testing the similarity of the waveforms lies with the fact that the established tests in correlation theory enable the acceptance or the rejection of the hypothesis that the observed effects can be accounted for by an underlying correlation of ` strength ' @xmath116 , where @xmath117 .
the test when @xmath118 involves the transformation : @xmath119 the variable @xmath120 is expected to follow the @xmath120-distribution ( student s distribution ) with @xmath121 degrees of freedom ( dof ) .
the tests when @xmath122 involve fisher s transformation ; the details may be found in standard textbooks on statistics .
no tests are possible when @xmath123 , i.e. , when attempting to judge the _ goodness _ of the association between waveforms , if ideally the waveforms should be identical .
the only tests which can be carried out in such a case are those involving @xmath118 , i.e. , investigating the presence of a statistically - significant correlation between the tested waveforms when the null hypothesis for no such effects is assumed to hold . in practice , the one - sided tests for @xmath124 dof result in the rejection of the null hypothesis at the significance level of @xmath125 when @xmath126 and at the significance level of @xmath127 when @xmath128 .
formal , well - defined ( in the mathematical sense ) ways to compare waveforms do exist . as a general rule , the application of rigorous tests has the tendency to yield significant discrepancies in many cases , even when a judgment based on a visual inspection of the tested quantities is favourable .
a ) one possibility would be to obtain the uncertainties in the histogram bins and make use of a @xmath129 function to assess the goodness of the association .
the variability of the output across different sensors could also be assessed and this additional uncertainty could be taken into account in the tests .
b ) another possibility would be to invoke analysis of variance ( anova ) , defining the reduced ` within - treatments ' variation as @xmath130 and the reduced ` between - treatments ' variation as @xmath131 appearing in these expressions are two average waveforms : the average waveform obtained with measurement system
@xmath132 : @xmath133 and the grand - mean waveform : @xmath134 the ratio @xmath135 is expected to follow fisher s distribution with @xmath136 and @xmath137 dof .
the resulting p - value enables a decision on the acceptance or rejection of the null hypothesis , i.e. , of the observed effects being due to statistical fluctuation .
c ) a third possibility would be to histogram the _ difference _ of corresponding waveforms obtained from the two measurement systems within the same gait cycle @xmath138 ; the decision on whether the final waveform is significantly different from @xmath92 can be made on the basis of a number of tests , including @xmath129 tests for the constancy and shape of the result of the histogram . nevertheless , to retain simplicity in the present paper , we have decided to make use in the data analysis of the simple scoring options introduced by eqs.([eq : eq08])-([eq : eq12 ] ) .
the data acquisition may involve subjects walking and running on commercially - available treadmills .
the placement of the treadmill must be such that the motion of the subjects be neither hindered nor influenced in any way by close - by objects .
prior to the data - acquisition sessions , the two measurement systems must be calibrated and the axes of their coordinate systems be aligned ( spatial translations are insignificant ) .
the measurement systems must then be left untouched throughout the data acquisition .
the original kinect sensor also provides information on the elevation ( pitch ) angle at which it is set . during our extensive tests
, we discovered that this information is not reliable , at least for the particular device we used in our experimentation . to enable the accurate determination of the elevation angle of the kinect sensor , we set forth a simple procedure .
the subject stands ( in the upright position , not moving ) at a number of positions on the treadmill belt , and static measurements ( e.g. , @xmath1 s of kinect data ) at these positions are obtained and averaged .
the elevation angle of the kinect sensor may be easily obtained from the slope of the average ( over a number of kinect nodes , e.g. , of those pertaining to the hips , knees , and ankles ) ( @xmath19,@xmath20 ) coordinates corresponding to these positions .
the output data , obtained from the kinect sensor , must be corrected ( off - line ) accordingly , to yield the appropriate spatial coordinates in the ` untilted ' coordinate system . to prevent kinect from re - adjusting the elevation angle during the data acquisition ( which is a problematic feature ) , we attach its body unto a plastic structure mounted on a tripod .
it is worth mentioning that , as we are interested in capturing the motion of the subject s lower legs ( i.e. , of the ankle and foot nodes ) , the kinect sensors must be placed at such a height that the number of lost lower - leg signals be kept reasonably small .
our past experience dictates that the kinect sensor must be placed close to the minimal height recommended by the manufacturer , namely around @xmath66 ft off the ( treadmill - belt ) floor . placing the sensor higher
( e.g. , around the midpoint of the recommended interval , namely at @xmath8 ft off the treadmill - belt floor ) leads to many lost lower - leg signals leg ( the ankle and foot nodes are not tracked ) , as the lower leg is not visible by the sensor during a sizeable fraction of the gait cycle , shortly after the toe - off ( to ) instant .
the kinect sensor may lose track of the lower parts of the subject s extremities ( wrists , hands , ankles , and feet ) for two reasons : either due to the particularity of the motion of the extremity in relation to the position of the sensor ( e.g. , the identification of the elbows , wrists , and hands becomes problematic in some postures , where the viewing angle of the ulnar bone by kinect is small ) or due to the fact that these parts of the human body are obstructed ( behind the subject ) for a fraction of the gait cycle . assuming that these instances remain rare ( e.g. , below about @xmath139 of the available data in each time series , namely one frame in @xmath29 ) , the missing values may be reliably obtained ( interpolated ) from the well - determined ( tracked ) data .
although , when normalised to the total number of the available values , the untracked signals usually appear ` harmless ' as they represent a small fraction of the total amount of measurements , particular attention must paid in order to ensure that no node be significantly affected , as in such a case the interpolation might not yield reliable results .
a few velocities may be used in the data acquisition : walking - motion data may be acquired at @xmath1 km / h ; running - motion data at @xmath2 and @xmath3 km / h . at each velocity setting , the subject must be given time ( e.g. , @xmath21 min ) to adjust his / her movements comfortably to the velocity of the treadmill belt .
to obtain reliable waveforms from the kinect - captured data , we recommend measurements spanning at least @xmath66 min at each velocity .
the subject s motion is split into two components : the motion of the subject s cm and the motion of the subject s body parts relative to the cm .
of course , the accurate determination of the coordinates of the subject s physical cm from the kinect or mbs output is not possible . as a result ,
the obtained cm should rather be considered to be one reference point , moving synchronously with the subject s physical cm .
ideally , these two points are related via a simple spatial translation ( involving an unknown , yet constant 3d vector ) at all times ; if this condition is fulfilled , the obtained cm may be safely identified as the subject s physical cm , because a constant spatial separation between these two points does not affect the evaluation of the important quantities used in the modelling of the motion . at all time frames , we obtain the coordinates of the subject s cm from seven nodes , namely from the first three main - body nodes @xmath21 to @xmath38 , from the shoulder nodes @xmath1 and @xmath22 , as well as from the hip nodes @xmath39 and @xmath14 . being subject to considerable movement in walking and running motion , the node @xmath8 ( head ) is not included in the determination of the coordinates of the subject s cm . prior to further processing ,
the cm offsets ( @xmath140,@xmath141,@xmath142 ) are removed from the data ; thus , the motion is defined relative to the subject s cm at all times .
( the angles , defined in subsection [ sec : definitions ] , involve differences of corresponding coordinates ; as a result , they are not affected by the removal of the cm offsets from the data . )
the largeness of the ` stray ' motion of the subject may be assessed on the basis of the root - mean - square ( rms ) of the @xmath140 , @xmath141 , and @xmath142 distributions .
to investigate the stability of the motion over time , the data may be split into segments . in our data analysis
, the duration of these segments may be chosen at will ; up to the present time , we have made use of @xmath36 and @xmath23 s segments in the analysis of the kinect - captured data . within each of these segments , information which may be considered ` instantaneous ' is obtained , thus enabling an examination of the ` stability ' of the subject s motion at the specific velocity ( see subsection [ sec : analysis2 ] ) .
the symmetry of the motion for the left and right parts of the human body may be investigated by comparing the corresponding waveforms .
finally , the largeness of the motion of the extremities may be examined on the basis of the roms obtained from these waveforms .
we subsequently address some of these issues in somewhat more detail .
ideally , the period of the gait cycle @xmath143 is defined as the time lapse between successive time instants corresponding to identical postures of the human body ( position and direction of motion of the human - body parts with respect to the cm ) .
( of course , the application of ` identicalness ' in living organisms is illusional ; no two postures can ever be expected to be identical in the formal sense . )
we define the period of the gait cycle as the time lapse between successive most distal positions @xmath20 of the same lower leg ( i.e. , of the ankle or of the ankle - foot midpoint ) .
the arrays of time instants , at which the left or right lower leg is at its most distal position with respect to the subject s instantaneous cm , may be used in timing the waveforms corresponding to the left or right part of the human body .
the period of the gait cycle is related to two other quantities which are used in the analysis of motion data . *
the stride length @xmath144 is the product of the velocity @xmath145 and the period of the gait cycle : @xmath146 . *
the cadence @xmath35 is defined as the number of steps per unit time ; one commonly - used unit is the number of steps per min .
it has been argued ( e.g. , by daniels @xcite ) that the minimal cadence in running motion should be ( optimally ) @xmath147 steps per min , implying a maximal period of the gait cycle of @xmath148 s. to examine the constancy of the period of the gait cycle throughout each session ( according to our definition , each session involves _
one _ velocity ) , the values of the instantaneous period of the gait cycle are submitted to further analysis . the overall constancy is judged on the basis of a simple @xmath129 test , assessing the goodness of the representation of the input data by one overall average value ; the resulting p - value is obtained from the minimal value @xmath149 for the given number of dof , i.e. , for the number of data segments reduced by one unit . to assess statistical significance
, we favour the use of the p - value threshold of @xmath127 , which is a popular choice among statisticians . using the time - instant arrays from the analysis of the left and right lower - leg signals ( as described in subsection [ sec : analysis1 ] ) , each time series ( pertaining to a specific node and spatial direction ) is split into one - period segments , which are subsequently superimposed and averaged , to yield a representative movement for the node and spatial direction over the gait cycle .
finally , one average waveform for each node and spatial direction is obtained , representative of the motion at the particular velocity .
the investigation of the asymmetry in the motion rests on the comparison of the waveforms obtained for corresponding left and right nodes , and spatial directions .
average waveforms for all nodes and spatial directions , representing the variation of the motion of that node ( in 3d ) within the gait cycle , are extracted separately for the left and right nodes of the extremities ; waveforms are also extracted for the important angles introduced in subsection [ sec : definitions ] .
as mentioned in subsection [ sec : analysis1 ] , the time instant at which the subject s left ( right ) lower leg is at its most distal position ( with respect to the subject s cm ) marks the start of each gait cycle ( as well as the end of the previous one ) , suitable for the study of the left ( right ) part of the human body . in case that left / right ( l / r ) information is not available ( as , for example , for the trunk angle ) , the right lower leg may be used in the timing .
all waveforms are subsequently @xmath92-centred .
the removal of the average offsets is necessary , given that the two measurement systems yield output which can not be thought of as corresponding to the same anatomical locations .
for instance , according to the ` plug - in gait ' placement scheme , the markers for the shoulder are placed on top of the acromioclavicular joints ; the kinect nodes shoulder_left and shoulder_right match better the physical locations of the shoulder joints .
the left and right waveforms yield two new waveforms , identified as the ` l / r average ' ( lra ) and the ` right - minus - left difference ' ( rld ) ; if emphasis is placed on the extraction of asymmetrical features in the motion from the kinect output , the validation of the rlds is mandatory .
the comparisons of the waveforms obtained for the nodes of the extremities from the two measurement systems , as well as of those obtained for the important angles defined in subsection [ sec : definitions ] , are sufficient in providing an estimate of the degree of the association of the output of the systems under investigation .
if one of these systems is an mbs , such a comparison enables decisions on whether reliable information may be obtained from the tested kinect sensor ( assumed to be the second system ) ; a common assumption in past studies @xcite is that the inaccuracy of the mbs output is negligible compared to that of the kinect sensor .
( of course , to obtain from the marker positions information on the internal motion , i.e. , on the motion of the human skeletal structure , is quite another issue ; we are not aware of works addressing this subject in detail . )
as already mentioned , the theoretical background , developed in the present paper , also applies to a comparative study of the two kinect sensors , identifying the similarities and the differences in their performance , but ( of course ) it can not easily enable decisions on which of the two sensors performs better . in summary ,
irrespective of whether one of the two measurement systems is an mbs or not , the same tests are performed , but the interpretation of the results is different .
we propose tests as follows : * identification of the node levels of the extremities and spatial directions with the worst association ( e.g. , with a similarity - index value in the first quartile of the distribution ) between the waveforms of the two measurement systems . * determination of the similarity of the association between the waveforms pertaining to the upper and lower parts of the human body . * determination of the similarity of the association between the waveforms pertaining to the three spatial directions @xmath18 , @xmath19 , and @xmath20 . * determination of the similarity of the association between the waveforms obtained from the raw lower - leg signals .
we propose separate tests for the lra and rld waveforms ( see end of subsection [ sec : analysis3 ] ) ; if the reliable extraction of the asymmetry of the motion is not required in a study , one may use only the lra waveforms . after studying the goodness of the association between the waveforms at fixed velocity
, velocity - dependent effects may be investigated .
we will now provide additional details on each of these tests .
the goodness of the association between the waveforms , obtained from the two measurement systems for the eight node levels of the extremities ( shoulder , elbow , wrist , hand , hip , knee , ankle , and foot ) and spatial directions , may be assessed as follows . separately for each of the five scoring options of subsection [ sec : comparison ] , for each velocity setting , and for each spatial direction , the node levels may be ranked according to the goodness of the association of the waveforms of the two measurement systems .
the node level with the worst association may be given the mark of @xmath92 , whereas the one with the best association the mark of @xmath150 .
the sum of the ranking scores over all velocities and scoring options yields an @xmath151 ` matrix of goodness of the association ' ( @xmath2 node levels of the extremities , @xmath38 spatial directions ) ; entries in this matrix are restricted between @xmath92 and @xmath152 , where @xmath153 is the number of the velocities used in the data acquisition ; further analysis of the entries of this matrix yields relative information on the goodness of the association for the node levels of the extremities and spatial directions , e.g. , it identifies those pertaining to the first quartile of the similarity - index distribution ( worst association ) . to assess the similarity of the waveforms of the two measurement systems , obtained for the nodes of the upper and lower extremities ,
one - factor anova tests may be performed , separately for each of the five scoring options of subsection [ sec : comparison ] , on the scores obtained at each velocity setting , for all upper - extremity nodes and spatial directions , and all lower - extremity nodes and spatial directions .
the outlined test should be sufficient in determining whether the performance between the two measurement systems for the lower part of the human body ( in relation to its upper part ) deteriorates .
it must be also investigated whether the aforementioned results are significantly affected after excluding the nodes with the worst association between the waveforms of the two measurement systems .
the goodness of the association between the waveforms , pertaining to the three spatial directions @xmath18 , @xmath19 , and @xmath20 , may be determined after employing anova tests .
similarly to the previous tests , it must be investigated whether the results are significantly affected after the exclusion of the nodes with the worst association between the waveforms of the two measurement systems .
our past experience indicates that the @xmath19 waveforms , corresponding to the raw lower - leg signals ( i.e. , the @xmath19 offsets of the subject s cm are not be removed from the signals ) , must be examined .
this comparison is important for two reasons .
first , the lower - leg signals are used in timing the motion ; second , we intend to use these signals in order to obtain the times ( expressed as fractions of the gait cycle ) of the initial contact ( ic ) and the to @xcite ; the difference of these two values is the stance fraction .
we had noticed in the past that a salient feature in the waveforms obtained from the original kinect sensor is a pronounced peak appearing around the ic ; this peak is less pronounced in the data obtained with the upgraded sensor , e.g. , see figs .
[ fig : lly_l ] and [ fig : lly_r ] .
although it can not influence the timing of the motion ( because of its position ) , this artefact complicates the determination of the stance fractions , at least when using the original sensor .
the goodness of the association between the rld waveforms must be investigated in the case that emphasis is placed on the reliable determination of any asymmetric features in the motion . to establish
whether the differences in the reliability of the lra and rld waveforms are significant , two - sided t - tests may be performed on the score distributions between corresponding lra and rld waveforms , a total of @xmath154 tests ( five scoring options , three tests per scoring option , @xmath153 velocity settings ) .
as it is not clear which type of t - tests is more suitable , we propose that three tests be made per case : paired , homoscedastic , and unequal - variance .
finally , we address the comparison of the roms obtained from the waveforms of the two measurement systems .
it might be argued that one could simply use in a study the roms , rather than the waveforms , as representative of the motion of each node .
of course , given that each waveform is essentially replaced by one number , the information content in the roms is drastically reduced compared to that contained in the waveforms .
plotted versus one another ( scatter plot ) , the ideal relation between the roms obtained from the two measurement systems should be linear with a slope equal to @xmath21 , both for the lra and for the rld waveforms .
the comparison of the two straight - line slopes , obtained in case of the lra and the rld waveforms , provides an independent assessment on the significance of the differences in the reliability of the lra and rld waveforms .
our aim in the present paper was to develop the theoretical background required for the comparison of the output of two measurement systems used ( or intended to be used ) in the analysis of human motion ; important definitions are given in section [ sec : method ] , whereas the data acquisition and the first part of the data analysis are covered in section [ sec : acquisition ] .
a list of meaningful tests , comprising the second part of the data analysis , is given in section [ sec : tests ] .
although this methodology has been developed for a direct application in the case of the microsoft kinect@xmath0 ( ` kinect ' , for short ) @xcite sensors , the use of which in motion analysis is our prime objective , its adaption may yield solutions suitable in other cases
. the outcome of the proposed tests of section [ sec : tests ] should be sufficient in identifying the important differences in the output of two measurement systems .
as such , these tests identify ( in our case ) differences in the performance of the two kinect sensors ( in a comparative study ) or enable conclusions regarding the outcome of the validation of the output of either of the kinect sensors ( if the second measurement system is a marker - based system ( mbs ) ) . as next steps in our research programme
, we first intend to conduct a comparative study of the two kinect sensors , after applying the methodology set forth herein . at a later stage , we will attempt to validate the output of the two kinect sensors on the basis of standard mbss .
99 http://www.xbox.com/en-gb/kinect/ , http://www.microsoft.com/en-us/kinectforwindows/ r.a .
clark , validity of the microsoft kinect for assessment of postural control , gait posture 36 ( 2012 ) 372 - 377 .
a. schmitz , mao ye , r. shapiro , ruigang yang , b. noehren , accuracy and repeatability of joint angles measured using a single camera markerless motion capture system , j. biomech .
47 ( 2014 ) 587 - 591 .
b. galna , accuracy of the microsoft kinect sensor for measuring movement in people with parkinson s disease , gait posture 39 ( 2014 ) 1062 - 1068 .
b. bonnech@xmath11re , validity and reliability of the kinect within functional assessment activities : comparison with standard stereophotogrammetry , gait posture 39 ( 2014 ) 593 - 598 .
cavagna , m. kaneko , mechanical work and efficiency in level walking and running , j. physiol . 268 ( 1977 ) 467 - 481
cavanagh , m.a .
lafortune , ground reaction forces in distance running , j. biomech . 13 ( 1980 ) 397 - 406 .
cairns , r.g .
burdett , j.c .
pisciotta , s.r .
simon , a biomechanical analysis of racewalking gait , med .
sports exercise 18 ( 1986 ) 446 - 453 .
s. unpuu , the biomechanics of walking and running , clin .
sport med .
13 ( 1994 ) 843 - 863 .
novacheck , the biomechanics of running , gait posture 7 ( 1998 ) 77 - 95 .
schache , k.l .
bennell , p.d .
blanch , t.v .
wrigley , the coordinated movement of the lumbo - pelvic - hip complex during running : a literature review , gait posture 10 ( 1999 ) 30 - 47 .
a. pfister , a.m. west , s. bronner , j.a .
noah , comparative abilities of microsoft kinect and vicon 3d motion capture for gait analysis , j. med .
technol . 38 ( 2014 ) 274 - 280 .
j. shotton , real - time human pose recognition in parts from single depth images , computer vision and pattern recognition ( cvpr ) , 2011 ieee conference on , issue date : 20 - 25 june 2011 .
http://www.idmil.org/mocap/plug-in-gait+marker+placement.pdf , http://wweb.uta.edu/faculty/ricard/classes/kine-5350/pigmanualver1.pdf r.b .
davis , s. unpuu , d. tyburski , j.r .
gage , a gait analysis data collection and reduction technique , hum .
movement sci . 10 ( 1991 ) 575 - 587 .
m. mongiardini , m.h . ray , m. anghileri , development of a software for the comparison of curves during the verification and validation of numerical models , @xmath155 european ls - dyna conference , salzburg , austria , may 14 - 15 , 2009 .
kadaba , repeatability of kinematic , kinetic , and electromyographic data in normal adult gait , j. orthop .
res . 7 ( 1989 ) 849 - 860 .
mcginley , r. baker , r. wolfe , m.e .
morris , the reliability of three - dimensional kinematic gait measurements : a systematic review , gait posture 29 ( 2009 ) 360 - 369 .
p. garofalo , inter - operator reliability and prediction bands of a novel protocol to measure the coordinated movements of shoulder - girdle and humerus in clinical settings , med .
comput . 47 ( 2009 ) 475 - 486 .
a. ferrari , a.g .
cutti , a. cappello , a new formulation of the coefficient of multiple correlation to assess the similarity of waveforms measured synchronously by different motion analysis protocols , gait posture 31 ( 2010 ) 540 - 542 .
h. theil , economic forecasts and policy , second edition , north - holland , amsterdam ( 1961 ) .
a. ferrari , first in vivo assessment of `` outwalk '' : a novel protocol for clinical gait analysis based on inertial and magnetic sensors , med .
comput . 48 ( 2010 ) 1 - 15 .
j. daniels , daniels running formula , third edition , human kinetics ( 2013 ) , pp .
the @xmath6 nodes of the original kinect sensor .
the figure has been produced with carmetal , a dynamic geometry free software ( gnu - gpl license ) , first developed by r. grothmann and recently under e. hakenholz @xcite.,width=585 ] preliminary results for the waveforms for the raw @xmath19 coordinate of the left lower leg ( ankle ) obtained from one subject , using both kinect sensors .
the quantity @xmath156 is the fraction of the gait cycle .
the sensors were attached unto a plastic structure mounted on a tripod ; the difference in the @xmath19 values simply reflects the higher position on the mount of the upgraded kinect sensor.,width=585 ] | the present paper aims at providing the theoretical background required for investigating the use of the microsoft kinect@xmath0 ( ` kinect ' , for short ) sensors ( original and upgraded ) in the analysis of human motion .
our methodology is developed in such a way that its application be easily adaptable to comparative studies of other systems used in capturing human - motion data .
our future plans include the application of this methodology to two situations : first , in a comparative study of the performance of the two kinect sensors ; second , in pursuing their validation on the basis of comparisons with a marker - based system ( mbs ) . one important feature in our approach is the transformation of the mbs output into kinect - output format , thus enabling the analysis of the measurements , obtained from different systems , with the same software application , i.e. , the one we use in the analysis of kinect - captured data ; one example of such a transformation , for one popular marker - placement scheme ( ` plug - in gait ' ) , is detailed .
we propose that the similarity of the output , obtained from the different systems , be assessed on the basis of the comparison of a number of waveforms , representing the variation within the gait cycle of quantities which are commonly used in the modelling of the human motion .
the data acquisition may involve commercially - available treadmills and a number of velocity settings : for instance , walking - motion data may be acquired at @xmath1 km / h , running - motion data at @xmath2 and @xmath3 km / h . we recommend that particular attention be called to systematic effects associated with the subject s knee and lower leg , as well as to the ability of the kinect sensors in reliably capturing the details in the asymmetry of the motion for the left and right parts of the human body .
the previous versions of the study have been withdrawn due to the use of a non - representative database .
+ _ pacs : _ 87.85.gj ; 07.07.df , , biomechanics , motion analysis , treadmill , marker - based system , kinect @xmath4e - mail : evangelos[dot]matsinos[at]zhaw[dot]ch , evangelos[dot]matsinos[at]sunrise[dot]ch |
the application of uniaxial strain on graphene has been studied in the tight - binding approximation by pereira and co - workers @xcite .
the major results are i ) the application of strain opens up a gap in the bandstructure ii ) the strain has to cross a threshold value for the gap to open up and iii ) the application of strain along every direction does not result in a bandgap , for instance a strain along the armchair direction never opens a gap .
along the zigzag direction , a threshold strain of @xmath109 is needed for the gap to open up . figs .
[ fig_bs_dos_till_eps_crit ] and [ fig_bs_dos_beyond_eps_crit ] show plots of the bandstructure and the density of states ( dos ) corresponding to various values of strain along the zigzag direction . the plots in fig .
[ fig_bs_dos_till_eps_crit ] are for strain values less than the critical strain required to open a bandgap , whereas in fig .
[ fig_bs_dos_beyond_eps_crit ] the strain is larger than the critical strain .
as mentioned in the main text , in order to study the hubbard model on the honeycomb lattice under uniaxial strain , the method we choose is the two - particle self - consistent approach ( tpsc ) .
this nonperturbative , semi - analytical technique is valid from weak to intermediate values of interaction .
initially developed as an approximation scheme to study the single - band hubbard model on the square lattice @xcite , tpsc was later extended to study the multi - band case of the hubbard model @xcite .
in particular , the semi - metal to antiferromagnet transition of the half - filled hubbard model on the honeycomb lattice was studied using this method @xcite .
tpsc satisfies many important physical constraints like conservation laws , pauli principle and local sum rules for spin and charge susceptibilities .
most importantly , this approach obeys the mermin - wagner theorem , which prevents a finite temperature phase transition in two dimensions , while aptly capturing the physics of long wavelength antiferromagnetic fluctuations in the system .
once the antiferromagnetic correlation length becomes larger than the thermal de broglie wavelength , there is a crossover to the renormalized classical regime .
since the quasiparticles get destroyed by these large antiferromagnetic fluctuations , a pseudogap opens up @xcite . in tpsc , we start with the noninteracting susceptibility and define bethe - salpeter equations for the interacting spin and charge susceptibilities .
although tpsc looks similar to rpa , the irreducible vertices for the spin and charge channels are different , and are different from the bare interaction@xcite . in the case of the honeycomb lattice , for our purposes
we have a non - interacting susceptibility which is a @xmath110 matrix , @xmath111 where @xmath112 , @xmath113 is the momentum and @xmath114 is the bosonic matsubara frequency .
an element of the noninteracting susceptibility in the momentum - imaginary frequency representation is given by , @xmath115 where @xmath116 are the sublattice indices on the honeycomb lattice and the lattice / momentum grid is of size @xmath117 . here
@xmath118 , @xmath119 is the momentum and @xmath27 is the fermionic matsubara frequency .
the spin and charge susceptibilities in the position - imaginary time representation are , @xmath120 here @xmath121 is a shorthand for @xmath122 the position and the imaginary time . from eq.s and , the local spin and charge sum rules
are obtained when we set @xmath123 , @xmath124 the interacting spin and charge susceptibilities are given by @xmath125^{-1 } \bm{\chi}_{0}(q ) \label{eq_spin_susc_graphene } \\
\bm{\chi}_{ch}(q ) = \left [ \bm{1 } + \frac{1}{2 } \bm{\chi}_{0}(q)\mathbf{u}_{ch } \right]^{-1 } \bm{\chi}_{0}(q ) \label{eq_charge_susc_graphene}\end{aligned}\ ] ] in terms of the noninteracting susceptibility .
this is the matrix version of the scalar equation appearing in the single - band case @xcite .
the elements of the matrix spin ( @xmath126 ) and charge ( @xmath127 ) vertices appearing in the above expression can be obtained as functional derivatives of the elements of the self - energy matrix with respect to the elements of the green function matrix . using the tpsc ansatz , we can write a first approximation to the self - energy which is momentum and frequency independent @xcite and from this form of the self - energy we get a local spin vertex where the diagonal @xmath128 and @xmath129 elements are non - zero .
this local spin vertex is given by @xmath130 because of sublattice symmetry @xmath131 and @xmath132 .
although the charge vertex is not local , and involves higher order correlation functions which are hard to compute , we assume that the charge vertex is also local ( @xmath133 ) .
we define ferromagnetic and the antiferromagnetic spin susceptibilities for both noninteracting and interacting cases as , @xmath134 the interacting susceptibilities are related to the noninteracting ones by the relation @xmath135 that resembles the scalar bethe - salpeter equations in the single - band case . focusing only on antiferromagnetic correlations ,
the correlation length @xmath42 can be quantified in terms of the ratio of the maximum of the interacting antiferromagnetic susceptibility to the maximum of the noninteracting antiferromagnetic susceptibility at @xmath136 .
@xmath137 the second approximation to self - energy including the effects of spin and charge fluctuations can be made as @xmath138 \label{eq_se_graphene}\end{aligned}\ ] ] we work at half - filling , and start by finding the noninteracting susceptibility @xmath139 using a combination of fft and cubic splines for a momentum grid @xmath117 for @xmath140 number of matsubara frequencies . for a particular value of interaction @xmath19 ,
a guess value of double occupancy @xmath141 is chosen . using the tpsc ansatz and the spin sum rule , the double occupancy , and thereby the spin vertex ( @xmath142 )
can be determined self - consistently .
once the double occupancy is obtained , using the charge sum rule , we can solve for the charge vertex ( @xmath133 ) .
again by using a combination of ffts and cubic splines we can compute the self - energy using .
the application of strain to the honeycomb lattice changes bond lengths .
this results in a change in the reciprocal lattice vectors as well as a change in the hopping parameters . in our calculations
, we assume the reciprocal lattice remains the same as the one corresponding to the undistorted lattice .
we take into account only the change in the hopping parameters . in the case of unstrained graphene ,
the hopping parameters along all the three bonds are equal . upon the application of strain , the hopping parameters are no longer isotropic . when a strain is applied along the zigzag direction , then @xmath143 .
this in turn changes the dispersion relation .
when we evaluate the noninteracting green functions and the noninteracting susceptibilities , this information is taken into account .
the rest of the procedure remains intact as mentioned in the previous section .
the results for double occupancy @xmath141 , irreducible spin and charge vertices @xmath142 and @xmath133 and antiferromagnetic correlation length ( @xmath42 ) for @xmath144 are shown in fig .
[ fig_prelim_results ] .
these quantities have been plotted as functions of strain @xmath2 for various values of @xmath19 .
as we can see , the antiferromagnetic fluctuations grow in magnitude with increasing interaction , at low values of temperature
. + the imaginary part of self - energy as a function of the fermionic matsubara frequency @xmath145 has been plotted for various values of strain less than the critical strain . as mentioned in the main text , for no strain or very low strain ( @xmath146 ) ,
the dirac landau fermi liquid remains stable till @xmath147 , whereas for @xmath148 , for @xmath149 , the imaginary part of self - energy shows anomalous behavior .
all these details can be seen from the plots in fig .
[ fig_se_till_eps_crit ] .
+ the spin fluctuation contribution to the magneto - volume was discussed in the main text . from the self - consistent value of double occupancy obtained
, we can calculate @xmath150 .
the plots of this quantity as a function of temperature for various values of strain are given in [ fig_z_till_eps_crit ] .
as temperature decreases , the antiferromagnetic fluctuations in the system become large and @xmath151 also increases as a consequence . + similarly , the derivative of @xmath151 with respect to temperature @xmath22 ( @xmath152)is plotted in fig .
[ fig_der_z_till_eps_crit ] for various values of strain . for @xmath84 , for @xmath153 , the derivative tends to diverge beyond @xmath154 .
this , as previously stated in the main text refer to quantum criticality associated with large antiferromagnetic fluctuations .
+ this can be further substantiated by plotting the derivative of @xmath151 with respect to strain @xmath2 ( @xmath155 ) , as a function of strain for various values of temperature as shown in fig .
[ fig_der_z_vs_eps ] . for low values of temperature , beyond @xmath149 , the singularity in the derivative points to the existence of a phase transition driven by the growing antiferromagnetic fluctuations .
for values of strain beyond the critical strain @xmath156 , the system becomes a band insulator .
the main result is that even at low temperatures , the antiferromagnetic fluctuations do not become huge as seen from the plots in fig .
[ fig_prelim_results_beyond_eps_crit ] .
these corroborate the statement in the main text that the features of quantum criticality are influenced by the growing antiferromagnetic fluctuations in the system .
since the antiferromagnetic fluctuations do not become huge , we do not expect the self energy to show any anomalous feature as @xmath157 .
this can be seen clearly from fig .
[ fig_se_beyond_eps_crit ] where we have plotted imaginary part of the self - energy as a function of the fermionic matsubara frequency , for values of strain beyond the critical strain .
the plots of @xmath158 in fig .
[ fig_z_beyond_eps_crit ] and the derivative of @xmath152 in fig .
[ fig_der_z_beyond_eps_crit ] also give us the same information . from the plots of fig .
[ fig_z_beyond_eps_crit ] , we can see that @xmath158 is almost temperature independent , for the same values of interaction at which we see large antiferromagnetic fluctuations in the system below critical strain . and
, @xmath152 does not diverge as @xmath84 , and is almost independent of the value of interaction .
6ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1103/physrevb.80.045401 [ * * , ( ) ] @noop * * , ( ) in @noop _ _ ( , ) pp .
link:\doibase 10.1103/physrevb.87.045113 [ * * , ( ) ] link:\doibase 10.1103/physrevb.92.144511 [ * * , ( ) ] link:\doibase 10.1103/physrevb.92.045111 [ * * , ( ) ] | strain tuning is increasingly being recognized as a clean tuning parameter to induce novel behavior in quantum matter . motivated by the possibility of straining graphene up to @xmath0 percent , we investigate novel quantum criticality due to interplay between strain - induced anisotropic band structure and critical antiferromagnetic spin fluctuations ( afsf ) in this setting .
we detail how this interplay drives @xmath1 a quantum phase transition ( qpt ) between the dirac - semimetal - incoherent pseudogapped metal - correlated insulator as a function of strain ( @xmath2 ) , and @xmath3 critical afsf - driven divergent nematic susceptibility near critical strain ( @xmath4 ) manifesting as critical singularities in magneto - thermal expansion and grneisen co - efficients . the correlated band insulator at large strain affords realization of a two - dimensional dimerized spin - singlet state due to this interplay , and we argue how doping such an insulator can lead to a spin - charge separated metal , leading to anomalous metallicity and possible unconventional superconductivity . on a wider front , our work serves to illustrate the range of novel states realizable by strain - tuning quantum materials . the exciting discovery of single - layer graphene has spawned a tremendous burst of activity @xcite on fundamental and applied grounds .
it is the strongest electronic material , capable of sustaining reversible elastic deformations in excess of @xmath0 percent . @xcite .
besides , graphene possesses rich electronic properties , originating from gapless , linearly dispersing dirac - like excitations @xcite .
attractive features like high mobility , absence of backscattering and strong field effects hold out the promise for future devices , but such possibilities are hindered by the gapless spectrum .
inducement of a gap by quantum confinement is possible , but this also gives rise to edge roughness , with deleterious effect on electronic properties .
recently , strain engineering has been studied @xcite as a way to profitably unite seemingly independent mechanical and electronic properties of graphene .
experimental studies indicate that reversible and controlled strain up to @xmath0 percent can be produced in graphene @xcite , opening a unique opportunity to realize this aspect .
while detailed investigation of this interplay within one - electron band structure has been carried out @xcite , possibility of novel physics arising due to interplay between strain - modified electronic structure and electron - electron interactions has not received much attention .
observed breakdown of the wiedemann - franz law @xcite near the charge neutrality point betrays the effect of sizable @xmath5 interactions in graphene . in this light , one may anticipate that a strain - modified band structure could give rise to novel quantum phase transition(s ) and novel physical responses as a result of interplay between an anisotropic electronic structure and local coulomb interactions . in this letter
, we explore the possibility of inducing novel quantum phase transitions in ( @xmath0 percent ) strained graphene as a consequence of interplay between strong antiferromagnetic spin fluctuations ( afsf ) and anisotropic band structure under strain .
we stress that our analysis has potentially wider consequences for materials with honeycomb lattice structures , but with correlations not so strong as to induce a mott transition ( which requires complementary dmft - like analyses @xcite ) .
our starting point is the electronic structure of graphene under strain @xcite .
we have repeated their earlier analysis . as shown in detail in supplementary information ( si ) @xcite and in fig . [ fig1 ] ,
zig - zag strain induces sizable reconstruction of the electronic structure : it moves the dirac point away from @xmath6 toward @xmath7 for finite strain , and @xmath3 beyond a critical zig - zag strain ( @xmath8 ) , a band - insulating gap develops in graphene and the `` dirac point '' gets shifted to the @xmath7 point . in contrast , armchair strain has very little effect on the electronic structure , so we will restrict ourselves to zig - zag strain hereafter .
though scarcely studied @xcite , @xmath5 interactions are not negligible in graphene .
realistic estimates put the intra - orbital hubbard @xmath9 ev within first - principles studies .
our strategy is to vary @xmath10 from small to realistic values , where @xmath11 is the tight - binding hopping integral between nearest - neighbor @xmath12 orbitals on the honeycomb lattice . under zig - zag strain ,
the zig - zag hoppings @xmath13 progressively reduce , while the @xmath14 hopping @xmath15 increases .
thus , interplay with hubbard correlations is , very generally , expected to @xmath1 enhance strain - induced anisotropy , @xmath3 produce a tendency to mott localization beyond a critical @xmath16 , and @xmath17 perhaps most importantly , to induce exotic semi - metallic / insulating states around critical strain @xmath18 when @xmath19 is varied across a critical @xmath20 . in the @xmath21 case that applies in graphene , long - range afsf should be very relevant .
while dynamic - mean - field theory ( dmft ) @xcite or its cluster extensions can reveal `` mottness '' effects , they can not , by construction , access the effects of long - ranged critical afsf and possible af quantum criticality at low @xmath22 . here
, we investigate the role of long - range critical afsf in strained graphene .
we employ the two - particle self - consistent ( tpsc ) theory : tpsc exploits the pauli principle and exact sum rules @xcite relating wave - vector dependent susceptibilities to _ local _ correlators to formulate a non - trivially renormalized version of the random - phase approximation ( rpa ) .
it is similar in spirit to the self - consistent renormalization ( scr ) theory of moriya and co - workers @xcite .
use of the renormalized hubbard u ( different for charge and spin channels ) , @xmath23 leads to a negative feedback in rpa , preventing the divergence of the spin susceptibility at _ any _ finite @xmath22 , in accord with stringent mermin - wagner restrictions in @xmath21 .
further , tpsc also recovers quantum critical ( qc)-``renormalized classical ( rc ) '' and qc - quantum disordered ( qd ) crossovers at low but finite @xmath22 , and yields low-@xmath22 af correlation lengths in good accord with non - linear sigma model ( nlsm ) predictions @xcite .
it is thus an excellent approximation for @xmath21 when coupling of carriers to long - range magnetic fluctuations is dominant .
indeed , tpsc has been extensively employed @xcite within the itinerant spin - fluctuation theory to study af quantum criticality in magnetic metals within the framework of hubbard models .
we now present our results for strained graphene , relegating technical details to si@xcite . for each value of strain ( @xmath2 ) , we vary the hubbard @xmath19 in the interval @xmath24 ev choosing an unstrained value of @xmath25 ev . in fig .
[ fig2 ] , we show the imaginary part of the self - energy , im@xmath26 @xcite at the strain shifted `` dirac '' point as a function of matsubara frequency @xmath27 . for zero or small ( @xmath28 ) strain ,
the dirac lfl metal remains stable for the range of @xmath10 we consider , as seen by the fact that -im@xmath29 for small @xmath19 @xcite . for @xmath30 , however , we find im@xmath31 , implying correlated landau fermi liquid ( lfl ) , or dirac liquid metallicity , for @xmath32 .
@xcite however , for larger @xmath2-dependent @xmath33 , im@xmath26 shows anomalous behavior : -im@xmath26 _ increases _ as @xmath34 , reflecting appearance of a _ pseudogap _ in the one - particle excitation spectrum . in the inset ,
we show how @xmath35 develops an anomaly around @xmath36 .
further , since gapless electronic states exist only at the `` dirac '' point , im@xmath37 also reflects the @xmath22-dependence of the @xmath38 resistivity : thus , given @xmath39 in the lfl regime , we expect the @xmath40-co - efficent of the resistivity to show a similar anomaly , reflecting a change from lfl metal to an incoherent non - lfl metal as @xmath41 around the critical strain . within tpsc
, the origin of this feature is associated with the coupling of itinerant @xmath12-band carriers to long - ranged , critical af spin fluctuations ( afsf ) . in fig .
[ fig3 ] , we show that while the af correlation length ( @xmath42 ) is finite for @xmath43 , it indeed diverges ( for low @xmath22 , @xmath44 ) at @xmath45 for @xmath46 , confirming this view .
it is known that @xmath47 e@xmath48 in the quantum critical region within tpsc ( see fig .
[ fig3 ] ) , in accord with predictions of the nlsm approach .
these features are thus manifestations of the crossover from qd to the qc regime of afsf as @xmath19 increases : it is the coupling of itinerant @xmath12 carriers ( encoded in the bare one - electron propagator , @xmath49 ) to nearly critical afsf , encoded in the spin susceptibility @xmath50^{-1}$ ] with @xmath51 as above that drives the qd - qc crossover . as expected , in fig .
[ fig3 ] , we show that increasing strain ( toward @xmath4 ) reduces @xmath20 for the qd - qc crossover .
this is due to the fact that reduction in @xmath52 along the zig - zag direction under strain enhances the effective @xmath10 anisotropically ( it becomes larger for the zig - zag relative to @xmath14 bonds ) , resulting in enhanced afsf for smaller @xmath53 .
this also implies that the renormalized afsf under strain will be enhanced by nearly - divergent @xmath54 , and that the feedback effect of the anisotropic afsf will in turn enhance the electronic anisotropy .
while this issue requires a self - consistent extension of tpsc in its current form , we consider this issue ( see below ) while investigating strain - enhanced nematicity .
anisotropic electronic structure modification under strain has profound consequences .
since uniaxial strain acts as a field conjugate to the nematic order parameter , electronic nematic ( en ) criticality itself will be washed out by strain , ruling out en itself as a candidate for the quantum phase transition ( qpt ) under strain . however , since strong afsf directly feed back on the tpsc self - energy @xcite and hence the renormalized propagator , the non - trivial renormalization effects do show up in @xmath55^{-1}$ ] .
computing the average ( renormalized ) kinetic energies along @xmath56 bonds from @xmath57 thus allows for a direct estimate of the influence of strong afsf in enhancing strain - induced electronic anisotropy , and hence on _ bond _ en state . in fig .
[ fig4 ] , we show variation of @xmath58 , with @xmath59 with @xmath60 without and with the hubbard @xmath19 .
it is rather clear that en `` order '' is significantly enhanced by strong afsf , the origin of which is anisotropic suppression of @xmath61 relative to @xmath62 , the latter itself being a direct consequence of afsf - driven anisotropic renormalization of @xmath63 under strain .
finally , notwithstanding the absence of quantum criticality associated with en order , we find that @xmath64 shows an abrupt change precisely around @xmath65 .
thus , there _ is _ evidence for nematic criticality as a function of strain : to characterize it , we observe that @xmath1 if bond en order would exist without strain , it would break the discrete ( @xmath66 for graphene ) rotational symmetry of the crystal lattice , implying spontaneous breaking of an ising - like symmetry , but @xmath3 strain washes out this criticality .
however , since strain is a field conjugate to @xmath67 , nematic critiality can still occur @xcite at finite strain , as in fe arsenides .
such a criticality should show up as a divergent nematic susceptibility , @xmath68 , near @xmath20 . in the inset to fig .
[ fig4 ] , we show that this is indeed what we find .
this leads us to the central conclusion of this work : _ a nematic quantum criticality of the quantum liquid - gas variety , driven by strong afsf around the critical @xmath20 , underlies the `` transition '' from dirac - lfl to an incoherent metal near critical strain_. lack of lfl quasiparticles for @xmath33 as above implies that one should expect an ( anisotropic ) breakdown of the wiedemann - franz ( wf ) law .
this has recently been shown to occur for neutral graphene , leading to possibility of a novel `` dirac liquid '' @xcite with transport due to collective excitations .
thermal and electrical transport under strain could help in determining the fate of the wf law under strain .
however , use of strain as a tuning parameter suggests that studying magnetic - fluctuation contributions to magneto - volume , thermal expansion and grneisen co - efficents could more naturally facilitate observation of the above qpt . that it is possible to reversibly tune the strain to about @xmath0 percent
could make the above scenario realistically possible .
motivated hereby , we now propose that critical divergences in thermal expansion and magnetic grneisen co - efficients due to divergent afsf near @xmath4 can be fruitfully employed to unearth the novel criticality proposed here . in our picture , coupling of strain to afsf results in a magneto - elastic interaction .
this will have novel consequences for spin fluctuation contribution to the magneto - volume , defined as @xmath69 and thermal expansion co - efficient , @xmath70 .
such effects have long been well - documented in @xmath71-band transition metals and compounds and explicated by the well - known moriya - usami approach @xcite within scr theory . given that tpsc is an advanced variant of scr theory , we can use the scr formulation to investigate magneto - elastic effects in strained graphene .
moriya - usami theory asserts that electronic correlation or spin fluctuation contribution to the magneto - volume in the quantum paramagnetic state ( @xmath72 in @xmath21 ) is related , apart from parameters @xmath73 ( related to the strain dependence of the band - width via @xmath74 ) and @xmath75 , the bulk modulus , to the magnetic fluctuations , @xmath76 , the latter evaluated in the quantum paramagnetic phase .
this can be evaluated using @xmath77 , with @xmath78 , the double occupancy evaluated from tpsc ( from @xmath79 ) using the tpsc sum rule , providing a direct way to study magnetoelastic effects near the qcp above . in fig.[fig5 ] , we show @xmath80 and @xmath81 for a range of @xmath19 near critical strain to illustrate the main features of interest .
@xmath73 is negative and of the order of 1 , while @xmath82 .
interestingly , though the magnetic fluctuation contribution to @xmath83 is sizable at small strain , we find that _ both _ contributions vanish as @xmath84 . beyond a strain - dependent @xmath65 ,
however , marked changes show up : most interestingly , the influence of long - range critical afsf reveal themselves in a critical divergence in @xmath85 .
further , this should have direct bearing on the divergence of the _ magnetic _ grneisen co - efficient , defined as @xmath86 , where @xmath87 is the electronic contribution to the specific heat .
it is numerically involved to compute @xmath88 within tpsc : however , we expect @xmath89 in @xmath21 up to logarithmic corrections ( in fact , within the field - theoretic rg work of sheehy _ et al _
@xcite , one finds @xmath90log@xmath91 with @xmath92 a constant .
if this were to qualitatively hold right up to @xmath93 from the lfl side , it would imply that @xmath94 would also show a critical divergence .
furthermore , these hallmarks of quantum criticality disappear beyond critical strain @xcite , pinning them to the influence of critical afsf .
anomalies in @xmath83 and @xmath95 have recently received intense attention in the context of quantum criticality in @xmath71- and @xmath96-band materials @xcite .
we can now rationalize the critical divergence of @xmath97 near critical strain at @xmath19 close to @xmath65 .
these are directly related to singular @xmath22-dependence of @xmath98 due to diverging afsf at sizable @xmath19 in the vicinity of @xmath4 as shown above . in light of inducement of reversible strain
@xcite up to @xmath0 percent , such a novel qcp with diverging nematic susceptibility at finite strain is potentially realizable in practice . more generally , _
metallic _ honeycomb materials involving @xmath99 in @xmath100 configuration @xcite ( whence modelled by one - band hubbard model ) could exhibit such anomalies as we find near a pressure - driven mott transition between a spin - liquid insulator and ( incoherent ) metal .
+ we conclude by discussing the novel prospect of fractionalization of added electrons ( holes ) in heavily strained graphene .
since @xmath63 is enhanced by afsf under strain , it is reasonable to expect that at sizable @xmath101 @xmath102 , in the correlated `` band '' insulator beyond @xmath4 , the combined effect will be to readily promote spin - singlet valence - bond pairing on @xmath103-bonds . as shown in fig .
[ fig7 ] , it then follows that this state can be viewed as a @xmath21 generalization of ( dimerized ) polyacetylene . in our case , it is the interplay between strain and strong afsf ( which also promotes the nematic state ) which produces the dimerized state in the insulator . in the spinful case of relevance here , a single doped hole corresponds to a missing state in the valence band , and unoccupied , singly and doubly occupied states correspond @xcite to charge and spin quantum numbers @xmath104 , @xmath105 and @xmath106 . in panel ( b ) of fig .
[ fig7 ] , we schematically show how a doped hole results in generation of an extended `` string defect '' of singlets emanating from the bond hosting the hole : clearly the unpaired spin , originally in the vicinity of the hole , separates from it by creating such a string . thus , a doped hole will fractionalize into a spinless _ holon _ and a spin-@xmath107 spinon ! in the continuum limit , this could be seen by solving the bogoliubov - de gennes like equation for a doped hole , which results in only _
one _ renormalizable solution for a vortex or antivortex .
this bares an alternative route to create fractionalized excitations by strain tuning in honeycomb lattice models .
these solitonic excitations are defects in the dimerized ground state , and have exciting consequences .
specifically , beginning with such an insulator , doping should lead to a fractionalized state , wherein electrical current is carried by holons but the hall current by spinons , as is implicit in the two - relaxation rate scenario @xcite .
further , if two doped holes sit on different dimers , the hole pair , dressed by the dynamically fluctuating short - range singlet correlations ( see the dashed region in fig .
[ fig7 ] , panel ( c ) ) could hop as a composite pair without scrambling the background spin configuration , leading to unconventional superconductivity in a spin - gapped background ( this would be a @xmath21 version of the luther - emery scenario in @xmath108 ) .
our study points toward such exotic possibilities , but we leave more detailed exploration of these themes for the future . to conclude , we have presented a specific scenario to realize a range of unconventional states of matter by use of `` clean '' strain - tuning in graphene .
our work motivates study of other interesting materials like phosphorene , silicene , among other honeycomb systems , whose unstrained structures bear some resemblance to graphene .
it should also be an attractive tool to investigate novel strain - induced qpts in topological and weyl systems , where the role of sizable coulomb interactions remains largely unexplored , and should be a theme of potential interest .
60ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1103/revmodphys.81.109 [ * * , ( ) ] link:\doibase 10.1103/physrevb.76.064120 [ * * , ( ) ] link:\doibase 10.1038/nmat1849 [ * * , ( ) ] link:\doibase 10.1103/physrevb.80.045401 [ * * , ( ) ] link:\doibase 10.1021/nn800459e [ * * , ( ) ] ,
, link:\doibase 10.1126/science.aad0343 [ * * , ( ) ] , link:\doibase 10.1103/physrevb.94.165168 [ * * , ( ) ] `` see supplementary information for ( i ) tight binding analysis of graphene under zig - zag strain , sec ( i ) , ( ii ) tpsc method and formalism for graphene under uniaxial strain , sec ( ii a ) , ( iii ) tpsc as applied to graphene under strain , sec ( ii b ) , ( iv ) more results for strain less than critical strain sec ( ii c ) and ( v ) scenario beyond critical strain , sec .
( iii ) . ''
link:\doibase 10.1103/physrevlett.106.236805 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.99.226803 [ * * , ( ) ] `` , '' in http://www.springer.com/materials/book/978-3-642-21830-9[__ ] , , vol . ,
( ) chap .
@noop _ _ , vol .
( , ) link:\doibase 10.1103/physrevb.39.2344 [ * * , ( ) ] link:\doibase 10.1103/physrevb.92.144511 [ * * , ( ) ] link:\doibase 10.1103/physrevb.76.064402 [ * * , ( ) ] link:\doibase 10.1126/science.1190482 [ * * , ( ) ] , link:\doibase 10.1038/nature11178 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/0038-1098(80)91241-7 [ * * , ( ) ] link:\doibase 10.1038/nphys892 [ * * , ( ) ] http://dx.doi.org/10.1038/srep06451 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.98.186809 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.67.2092 [ * * , ( ) ] * supplementary information for ` exploring novel quantum criticality in strained graphene ' * |
we fit the md data of @xmath35 with two functional dependencies : @xmath107 and @xmath108 for a range of @xmath109 , for which the @xmath46-dependence of @xmath35 has a monotonic evolution . for a given @xmath0 and @xmath1
, we start our fitting procedure by choosing a rather small value of @xmath110 , and fit the entire @xmath35 curve with either of @xmath111 or @xmath112 depending on @xmath46 smaller or larger than the chosen @xmath110 . for each region
( @xmath46 smaller or larger than @xmath110 ) , we get a value for @xmath113 , measuring the goodness of the fit @xcite in that region . adding its values from the two regions , we get the total @xmath113 of the entire fit for that specific @xmath110 .
now , we increase @xmath110 in small incremental steps , and calculate total @xmath113 for each such @xmath110 .
finally , the optimal values of @xmath110 and other parameters , e.g. , @xmath114 were obtained by considering the @xmath110 for which total @xmath113 is minimum . the data in fig .
2(c , d ) of the main paper are generated by repeating above procedure for the entire range of @xmath109 .
the adjoining fig .
[ fig : fit_gs ] depicts one such case , where we highlight the above procedure .
we emphasize that a value of @xmath115 implies that the whole @xmath35 curve has a single gaussian fall .
while it is also possible to get @xmath115 from a multiple - gaussian @xcite fall with different width in different ranges of @xmath46 , our results of @xmath115 do not correspond to such possibilities .
curve with two functional dependencies : @xmath116 for @xmath117 , and @xmath118 for @xmath119 .
@xmath110 denotes the optimal distance for which total @xmath113 is minimum .
, width=340 ]
in the main text , we have presented the evolution of @xmath37 with data limited to @xmath120 , and @xmath121 . while extrapolation of the same analysis is not possible at large @xmath0 for the lowest temperatures , we here present @xmath36 up to @xmath122 for @xmath123 .
the two main features to note here are : + ( a ) for @xmath20 , the traces of @xmath36 rises with @xmath1 for @xmath124 , where as initial increase of @xmath1 for @xmath125 to @xmath126 lowered them progressively ( fig .
2(a ) in the main text ) . interestingly , @xmath127 marks the solid to liquid crossover according to static properties @xcite , for which @xmath37 attains the minimum value ( over all @xmath1 ) .
( b)for @xmath20 , the monotonic fall of @xmath36 for all @xmath1 persists , and shows a tendency toward saturation for large @xmath128 .
for @xmath33 , on the other hand , @xmath37 approaches a value of @xmath38 , indicating gaussian behavior for large @xmath0 , and @xmath1 .
the data presented in fig . [ fig : expk ] has results from a single molecular dynamics simulation resulting into large fluctuations , though we made sure that they represent the true situation . in the limit of large time @xmath129 and temperature @xmath130 for the irregular ( panel
a ) and circular ( panel b ) confinements .
while , for circular confinement , @xmath36 is mostly bounded between the values @xmath39 and @xmath38 , it goes below @xmath39 and shows a tendency towards saturation for iregular confinement for @xmath124.,width=680 ]
[ fig : disp_100](a , b ) shows the displacements @xmath131 of the particles for irregular and circular confinements , respectively , at intermediate time @xmath132 for @xmath11 , illustrating the caging effect . for both confinements , most particles ( red dots ) perform rattling motion around their equilibrium positions .
these particles are ` trapped ' within the cage formed by their nearest neighbors , and such an arrest causes @xmath35 traces to collapse ( inset of fig .
1(a ) in the main text ) on each other for intermediate times . from the alignment of dots and the connecting lines ( fig .
[ fig : disp_100](a ) ) , we see that the coherent motion sets up randomly due to the absence of spatial symmetry for iwm , while such commencement of the coherent motion in cwm ( fig .
[ fig : disp_100](b ) ) is guided by the circular symmetry .
[ fig : disp_100](c , d ) shows the displacements of the particles for the two confinements in the same time regime for @xmath133 . here , we see that the displacements are random and relatively large in magnitude , for both the confinements
. one can expect such random motion at this temperature because of the higher thermal energy . of the particles are shown at @xmath11 ( panel ( a ) , and ( b ) ) , and @xmath133 ( panel ( c ) , and ( d ) ) for iwm and cwm at @xmath134 .
red , green , blue , and brown dots represent initial positions of the particles with increasing magnitude of displacement .
the length of the connecting black lines correspond to the magnitude of the displacements of the particles at the given time @xmath0 .
, width=566 ]
in our paper , we claimed ( fig .
2 ( c - f ) ) that @xmath35 evaluated from eq . ( 1 ) shows broad agreement with its direct calculation from the md simulation . here , in fig .
[ fig : pd ] , we quantify such ` broad ' agreement further , for both the confinements . at @xmath11 , while we have already seen @xmath35 evaluated from md simulation shows good match with its evaluation from eq .
( 1 ) for all time regimes for iwm ( fig . 2 ( c ) in the main paper ) , cwm shows small deviation at intermediate times ( @xmath136 , in fig .
2 ( d ) in the main paper ) . here , we substantiate this point in fig . [
fig : pd ] ( a ) , and ( b ) for iwm and cwm , respectively . at intermediate times
@xmath137 , we see good agreement for iwm ( fig . [
fig : pd ] ( a ) ) , while small deviation persists for cwm ( fig .
[ fig : pd ] ( b ) ) . at @xmath133
, we found that our phenomenological formalism could not capture the large @xmath46 trend of the @xmath35 curve for cwm ( fig .
2 ( f ) in the main paper ) . here , from fig .
[ fig : pd ] ( c ) , and ( d ) , we see that , while the match between the two methods of evaluating @xmath35 is not so good for intermediate times for both the confinements , match gets better for iwm as time progresses . from direct md simulations with its evaluation from eq .
@xmath138 ( see main paper ) for iwm ( panel ( a),(c ) ) and cwm ( panel ( b),(d ) ) at t=0.006 and 0.030 , respectively .
thick dots represent @xmath35 evaluated from eq .
@xmath138 , and solid lines are the same evaluated from md simulation .
insets show the distribution of diffusivities @xmath68 used to evaluate @xmath35 for the given @xmath0.,width=529 ]
we show the distinct part of vhcf , @xmath79 , at @xmath133 in fig . [
fig : dvh_03](a , b ) for the iwm and cwm respectively .
as expected in a liquid , the large @xmath46 bragg - peaks are already washed out at smallest @xmath0 .
its relaxation with time smoothens out even the weak modulations leaving only the strong peak at @xmath139 . with @xmath46 for several
@xmath0 in the ` liquid ' , depicting the temporal relaxation in an ( a ) iwm and ( b ) cwm .
the ` correlation - hole ' in @xmath76 fills up very quickly for both confinements.,width=566 ]
in order to illustrate the complementary role of the fast and slow particles at low @xmath1 , we identified in the main text ( inset of fig .
3(a ) ) the sections of a single @xmath35 contributed by them individually for an iwm at intermediate @xmath0 .
their independent roles can also be seen in circular trap and at large @xmath0 , which we demonstrate here in fig .
[ fig : slowfast ] . | the dynamical response of coulomb - interacting particles in nano - clusters are analyzed at different temperatures characterizing their solid- and liquid - like behavior . depending on the trap - symmetry ,
both the spatial and temporal correlations undergo slow , stretched exponential relaxations at long times , arising from spatially correlated motion in string - like paths .
our results indicate that the distinction between the ` solid ' and ` liquid ' is soft : while particles in a ` solid ' flow producing dynamic heterogeneities , motion in ` liquid ' yields unusually long tail in the distribution of particle - displacements .
a phenomenological model captures much of the subtleties of our numerical simulations .
study of nano - systems demands refinement of fundamental concepts , in addition to their technological promises .
for example , a sharp transition @xcite between the solid and liquid turns into a smooth crossover in nano - clusters @xcite .
while the relaxation in simple solids and liquids produce phonon and diffusive modes , particles motion in confinements show anomalous behavior @xcite .
experimental evidences from dipolar colloids @xcite and dusty plasma @xcite , point towards a rich interplay of long - range interactions and confinements yielding subtle dynamics , and has attracted theoretical attention @xcite .
then , how good are these ` solid ' and ` liquid ' in traps from a dynamical perspective ? we address these questions focusing on the dynamical properties of coulomb - interacting particles in confinements with and without symmetries .
the ` solid ' phase of coulomb - interacting particles in a nano - cluster is called a wigner molecule ( wm ) because it mimics the physics of a wigner crystal @xcite .
the wm ` melts ' into a ` liquid ' in circular @xcite and irregular @xcite confinements , termed circular and irregular wigner molecule ( cwm and iwm ) .
a comparison between the two elucidates the role of ` disorder ' , coming from the irregularities in the trap , on the crossover .
analysis of melting using spatio - temporal correlations yields crucial dynamical information @xcite .
the interplay of interactions and brownian motion ( consistent with the trap - geometry ) often leads to frustrated motion and dynamic heterogeneities ( dh ) @xcite , i.e. multiple time - scale relaxations .
these are ubiquitous near the transitions in colloids @xcite , gels @xcite , glasses @xcite , and bio - molecule suspensions @xcite .
we consider simple models for iwm and cwm and analyze the equilibrium motion of the constituents using iso - kinetic molecular dynamics ( md ) simulations @xcite .
our main results can be summarized as follows : \(1 ) the motion in traps display a variety of non - gaussian behavior of the distribution of particle displacements , including its stretched - exponential _ spatial _ decay .
such an ultra - slow relaxation is consistent with preliminary experimental signatures @xcite .
+ ( 2 ) the dynamics at long @xmath0 is intriguing : at low @xmath1 the solid ` flows ' producing dh , whose nature depends on the confinement .
though particle displacements are random at large @xmath1 , their distribution defies central limit theorem @xcite .
+ ( 3 ) our results demonstrate how the innate dynamics sets apart the manifestation of wigner physics @xcite based on the trap symmetries .
we model a wm using the hamiltonian : @xmath2 , where the first term represents coulomb repulsion between particles of charge @xmath3 and @xmath4 , and @xmath5 is the dielectric constant . here , @xmath6 is the confinement potential .
we compare the results from two confinements : ( a ) irregular @xmath7 @xcite , and ( b ) circular @xmath8 , where @xmath9 . , normalized by @xmath10 for clarity , for several @xmath0 in the ` solid ' ( @xmath11 ) : ( a ) iwm , and ( d ) cwm .
@xmath12 is gaussian up to @xmath13 and displays non - gaussian trends for @xmath14 , including multi - peaks at large times .
bunching - up of the traces of @xmath12 for @xmath15 is shown in the insets demonstrating caging ( weaker in cwm ) .
corresponding @xmath16 plots are shown ( b , c : iwm and e , f : cwm ) for @xmath17 and @xmath18 , illustrating the motional coherence .
the thick dots represent the initial position and the connecting lines signify the displacements of the same particle at @xmath0 .
red , green , blue , and purple dots represent particles with increasing @xmath19.mobile particles reside in the central region for @xmath20 , leaving nearly immobile ones in the narrow limbs near corners .
in contrast , the mobile particles in cwm circle around the periphery , while the slow ones flock towards the origin . ,
width=321 ] we rescale the length ( @xmath21 ) and energy ( @xmath22 ) , that transform @xcite the cwm hamiltonian to : @xmath23 . the time - scale is set by @xmath24 @xcite .
the overall factor @xmath25 and @xmath26 are chosen to keep mean density in the two traps same for a fixed total number of particles , @xmath27 .
the irregularity parameters @xmath28 , @xmath29 $ ] ( controls chaoticity ) , and @xmath30 $ ] ( breaks reflection symmetry ) @xcite are adjusted to access the standard features of disordered systems .
distances are measured in the unit of @xmath31 , the mean inter - particle spacing between the neighbors .
the statistics of the results from @xmath20 were improved by ` disorder averaging ' @xcite over five configurations , each identified by a specific @xmath32 . for @xmath33 , averaging is done over five independent md simulations . in the ` liquid ' ( @xmath34 ) of an (
a ) iwm and ( b ) cwm , shows deviations from the gaussian behavior ( except for small @xmath0 ) in both traps though @xmath16 remains random .
the nature of the evolution of @xmath35 is analyzed in ( c ) and ( d ) by studying the exponent @xmath36 ( see text ) .
@xmath37 decays continuously from @xmath38 to nearly @xmath39 at low @xmath1 .
the circular trap yields @xmath40 .
@xmath37 keeps degrading with @xmath0 and @xmath1 in irregular confinements .
the inset in ( d ) shows the gaussian fit of all the three peaks of a representative @xmath35 in cwm .
@xmath0-dependence of msd at different temperatures are shown for irregular ( e ) , and circular ( f ) traps .
the dots represent msd by integrating @xmath41 ( see text ) , and are shown for @xmath42 .
the perfect match between the solid lines and corresponding dots persists for all @xmath1 .
, width=321 ] a thermal crossover from a ` solid ' to ` liquid ' in wm asserts @xcite a ` solid ' at @xmath11 and ` liquid ' at @xmath34 for our chosen parameters . the spatio - temporal evolution of the system is studied from the van hove correlation function ( vhcf ) @xcite , whose self - part is @xcite : @xmath43 \rangle$ ] . here
@xmath44 denotes average over up to @xmath45 independent time origins .
@xmath35 thus measures the probability of a particle to move ( on average ) a distance @xmath46 in time @xmath0 . focusing on iwm at @xmath11 in fig .
1(a ) , the dynamics can be broadly classified into three temporal regimes @xcite : ( i ) the particles move ballistically for @xmath47 , and the resulting @xmath35 is gaussian .
they start feeling each other only for @xmath48 implying that our unit time is the time scale of interaction .
( ii ) particles get ` arrested ' in the cage formed by neighbors in the intermediate time @xmath49 , leading to a slowly decaying @xmath35 whose traces bunch up , see the inset of fig .
1(a ) .
( iii ) the dynamics for @xmath50 develop multi - peak structure in @xmath35 . in order to develop a deeper understanding ,
we next analyze the particle displacements , @xmath51 .
most particles only rattles around their equilibrium positions for @xmath52 .
motion that begins by breaking a cage causes avalanches of such events correlated in space , which leads to a string - like path of delocalization @xcite .
fig .
1(b ) presents @xmath53 for an iwm at @xmath11 , showing such an alignment of neighboring particles leading to a slow relaxation that yields a long tail in @xmath35 .
once the collective motion subsumes large part of the system , particles undergo little relative displacements , but an overall drift , producing secondary peak in @xmath35 . with further increase in @xmath0 ,
the string - like path degenerates into sub - clusters , each performing independent collective motion ( fig .
1(c ) ) , and contribute to the multi - peak @xmath35 of fig .
1(a ) . the motion in the cwm at @xmath11 in fig .
1(d ) is qualitatively similar to iwm insofar as the temporal regimes are concerned , albeit the caging regime is narrow ( @xmath54 ) and untenable , causing weaker agglomeration of @xmath35 ( inset , fig .
1(d ) ) .
however , the differences between the two traps stand out in the nature of coherent motion . while the circular symmetry forces such motion along its azimuth ( fig .
1(e , f ) ) , the collective dynamics in iwm is guided by the irregularity producing nonequivalent regions ( fig .
1(b , c ) ) .
unlike the sub - clusters with independent motions in iwm , a single ( azimuthal ) coherence grows in the cwm , see fig .
1(f ) .
this results into a rather broad hump of @xmath55 in fig .
1(d ) .
interestingly , the mobile particles travel by several lattice spacings riding on the coherence even in this ` solid ' phase @xcite
. signatures of spatially correlated motion of particles , have been realized in circular plasma crystals @xcite , and also in a flocking transition of a driven granular matter in confinements @xcite . in the ` liquid ' ( @xmath34 ) ,
@xmath16 are large and random .
yet , @xmath35 deviates from gaussian and differences persist in the two traps ( fig .
2(a , b ) ) . while the displacements are nearly gaussian in circular traps , it undergoes a progressively slower decline for @xmath20 .
it is not possible to discern sharp temporal regimes in such a ` melted ' state .
next , we quantify the large-@xmath46 decay of @xmath35 by fitting it to @xmath56 ( see supplementary material ) . fig .
2 ( a : iwm , b : cwm ) presents the time - evolution of the best fit parameter @xmath37 . for low @xmath1
, @xmath37 degrades close to unity ( both traps ) establishing a near - exponential decay for intermediate times . with an increase in @xmath1 , the long - time value of @xmath37 rises in @xmath33 , however , broadly remains within @xmath57 . in a striking contrast
, @xmath37 decreases monotonically for @xmath20 , causing a _ stretched _ exponential spatial decay . above analysis
does not extend beyond @xmath58 due to the non - monotonic evolution of @xmath35 at larger @xmath0 and low @xmath1 .
the monotonic fall of @xmath36 in the irregular trap , however , continues for large @xmath1 , and hints towards a saturation only for @xmath59 .
interestingly , @xmath37 reaches its minimum for @xmath60 , a temperature that marks the crossover between the ` solid ' and ` liquid ' @xcite .
the traces of @xmath36 rises for @xmath61 .
the system resembles a conventional liquid ( @xmath62 for large @xmath0 ) only at a much larger @xmath63 . in contrast , a circular confinement yields @xmath64 at large @xmath0 for @xmath65 ( see supplementary material ) .
the classification of temporal regimes is also revealed from the mean - square displacement ( msd ) , @xmath66 , shown in fig . 2 ( e : iwm , f : cwm ) .
the low-@xmath1 evolution of msd is surprisingly flat for iwm showing strong caging .
the corresponding cwm show a weak brownian behavior .
the plateau at intermediate @xmath0 gradually shrinks with @xmath1 leading to a smooth crossover between ballistic and fickian motion .
it follows that , @xmath67 , and its estimation is shown as dots in fig .
2(e - f ) , confirming the ` fickian yet non - gaussian ' motion @xcite in the wm at large @xmath0 .
msd saturates when displacements approach the system size .
its occasional jumps in iwm fades away with @xmath1 .
such jumps are not discernible in a cwm even at the lowest @xmath1 .
iwm , thus , resembles a supercooled system @xcite more than what cwm does .
obtained from direct md simulations with its evaluation using eq .
( 1 ) .
the insets show corresponding @xmath68 and the parameters @xmath0 and @xmath1 appear as legends .
the nature of @xmath35 appears largely controlled by @xmath68 , from its rich structures due to heterogeneous diffusion , and from its long tail .
, width=321 ] a phenomenological description expresses @xmath35 as a convolution of gaussian displacements and the distribution of diffusivities , @xmath68 @xcite .
we generalize this to accommodate the multi - peak nature of @xmath35 : @xmath69,\ ] ] where @xmath70 is the number of peaks observed in @xmath35 . by fitting the individual peaks ( see the inset of fig .
2(d ) ) , we obtain @xmath71 and the peak widths .
we determine @xmath72 from the statistics of particles whose @xmath73 falls within the peak width corresponding to @xmath71 @xcite . a comparison of eq .
( 1 ) with the direct simulation is shown in fig .
3(a - d ) .
the distinct @xmath71 signals dynamic heterogeneity for @xmath50 . the inset of fig .
3(c ) shows a stretched - exponential fall @xcite of @xmath68 for @xmath34 in @xmath74 , whereas , its near exponential decay characterizes @xmath75 . a full @xmath68 corresponding to a multi - peak @xmath35 displays rich structures ( insets of fig .
3(c - f ) ) , and the higher peaks contribute near homogeneous weight to @xmath68 .
with @xmath46 for several @xmath0 in the ` solid ' , depicting the temporal relaxation in ( a ) iwm and ( b ) cwm .
the ` correlation - hole ' in @xmath76 fills up with time producing strong peak at large @xmath0 .
the inset of ( a ) identifies the contribution of slow ( brown ) and fast ( blue ) particles in @xmath35 .
the inset of ( b ) shows that the fast particles contribute to @xmath77 , while the slow ones trace the frozen ` solid ' .
_ lower panels : _ a strong bond - orientation at @xmath11 reflects in @xmath78 , for all @xmath0 for both iwm ( panel c ) and iwm ( panel d ) . its decay in the melted state ( @xmath34 ) follow an exponential and stretched - exponential @xmath0-dependences for circular and irregular confinements respectively .
, width=321 ] the multiple relaxations is also probed by the distinct - part of vhcf , @xmath79 @xcite : @xmath80^{-1}\sum_{i\neq j}\delta \left [ r - |\vec{r}_{i}(t ) - \vec{r}_{j}(0)|\right ] \rangle$ ] .
it represents the probability of finding at @xmath0 a different particle at a distance @xmath46 from a location where there was a particle initially .
fig .
4(a ) shows the lack of positional order at @xmath11 in the iwm @xcite through depleted bragg - peaks @xcite of @xmath81 , while a cwm displays sharper peaks ( fig .
4(b ) ) due to the azimuthal order .
we substantiate the heterogeneity defining mobility @xcite as @xmath82 .
a particle is qualified as ` fast ' ( ` slow ' ) if its mobility lies within the top ( bottom ) @xmath83 ( @xmath84 ) .
the long - time accumulation of @xmath85 ( fig .
4(a , b ) ) arises from the fast particles occupying the location from where another particle moved out , developing spatially correlated motion . in contrast , the slow particles trace out bragg - peaks .
such distinct behavior is illustrated as the inset of fig .
4(b ) for @xmath33 .
a similar picture emerges from the self - part too while the slow particles produce the small-@xmath46 gaussian behavior , the fast particles contribute to the long tail of @xmath35 , see inset of fig.4(a ) .
the ` solid ' state of a wm possesses bond - orientational order .
each particle has predominantly six neighbors , and each of these when connected to the given particle makes an angle of @xmath86 with the line connecting successive neighbors .
the corresponding bond - orientational correlation function : @xmath87\rangle$ ] @xcite is unity for a perfect wigner crystal .
its decay describes how the propensity of the six co - ordinated neighborhood corrodes in time . here
@xmath88 is the angle fluctuation of a fixed bond .
we present @xmath89 in fig .
4(c ) and ( d ) for @xmath20 and @xmath33 respectively .
a bulk 2d isotropic liquid shows an exponential fall of @xmath89 with @xmath0 signaling the loss of this order according to the kthny theory @xcite . while our results in fig .
4(d ) broadly capture such a trend in @xmath33 , the ` liquid ' in @xmath20 ( fig .
4(c ) ) shows a slower , stretched - exponential decay : @xmath90 $ ] with @xmath91 . in conclusion , we find non - gaussian distribution of the particle displacements , correlated motion of the constituents in string - like paths , distinct relaxation of slow and fast particles , and stretched - exponential decay of both spatial and temporal correlations in irregular confinement with coulomb particles , whereas the circular traps produce different motional signatures .
bulk supercooled liquids , which are also disordered systems with short - range inter - particle interactions , share much of these dynamic footprints . such similarities hint towards these findings to constitute the robust dynamical features of disordered systems
. it will be interesting to explore the response functions @xcite in different temporal regimes .
our results will provide guidance in designing nano - technology devices . _
we thank deepak dhar and chandan dasgupta for valuable discussions .
ba acknowledges his doctoral fellowship from the university grant commission ( ugc ) , india .
99 k. huang _ statistical mechanics _ , second edition , wiley ( 1987 ) . c. n. yang , and t. d. lee , phys .
rev .
, * 87 * , 404 ( 1952 ) .
v. m. bedanov , and f. m. peeters , phys .
rev .
b , * 49 * , 2667 ( 1994 ) . a. ghosal , a. d. gl , c. j. umrigar , d. ullmo , and h. u. baranger , nat .
phys . ,
* 2 * , 336 ( 2006 ) .
d. bhattacharya , and a. ghosal , eur .
phys .
j. b , * 86 * , 499 ( 2013 ) .
a. v. filinov , m. bonitz , and yu .
e. lozovik phys .
rev .
lett . * 86 * , 3851 ( 2001 ) .
m. d. ediger , annu .
rev .
phys .
chem .
, * 51 * , 99 ( 2000 ) .
k. h. nagamanasa , s. gokhale , r. ganapathy , and a. k. sood , proc .
natl .
acad .
sci .
usa , * 108 * , 11323 ( 2011 ) .
r. bergman , and j. swenson , nature , * 403 * , 283 ( 2000 ) .
t. ott , m. bonitz , z. donk , and p. hartmann , phys .
rev .
e * 78 * , 026409 ( 2008 ) .
a. schella , a. melzer , c. julyb , and c. bechingerb , soft matter , * 11 * , 1197(2015 ) .
y. ivanov , a. melzer phys .
plasmas * 12 * , 072110 ( 2005 ) .
e. yurtsever , f. calvo , and d. j. wales , phys .
rev .
e * 72 * , 026110 ( 2005 ) .
j. p. hernandez - ortiz , c. g. stoltz , and m. d. graham , phys .
rev .
lett . * 95 * , 204501 ( 2005 ) .
e. wigner , phys .
rev .
, * 46 * , 1002 ( 1934 ) .
k. zahn , and g. maret , phys .
rev .
lett . , * 85 * , 3656 ( 2000 ) .
ludovic berthier , phys . , * 4 * , 42 ( 2011 ) .
j. kim , c. kim , and b. j. sung , phys . rev . lett .
, * 110 * , 047801 ( 2013 )
. y. gao , and m. l. kilfoil , phys .
rev .
lett . , * 99 * , 078301 ( 2007 ) .
p. i. hurtado , l. berthier , and w. kob , phys .
rev .
lett . , * 98 * , 135503 ( 2007 ) .
p. chaudhuri , l. berthier , and w. kob , phys .
rev .
lett . , * 99 * , 060604 ( 2007 ) .
b. wang , s. m. anthony , s. c. bae , and s. granick , proc .
natl .
acad .
sci .
u.s.a .
, * 106 * , 15160 ( 2009 ) . d. frenkel , and b. smit , _ understanding molecular simulation _ , second edition , academic press ( 2001 ) . k. he @xmath92 @xmath93 , acs nano , * 7 * , 5122 ( 2013 ) .
o. kallenberg , _ foundations of modern probability _ , new york : springer - verlag ( 1997 ) .
o. bohigas , s. tomsovic , and d. ullmo , phys . rep . , * 223 * , 43 ( 1993 ) .
our scaled energy and time unit takes the value @xmath94 mev and @xmath95 fs , for electrons in circular quantum dot . here
, bare parameters are taken for gaas heterostructures ( @xmath96 , @xmath97 , the mass of an electron , @xmath98 mev , @xmath99 ) @xcite m. a. reed , and w. p. kirk , _ nanostructure physics and fabrication _ , ( academic press , boston , 1989 ) .
h. jiang , h. u. baranger , and w. yang , phys .
rev .
lett . , * 90 * , 026806 ( 2003 ) .
d. ullmo , t. nagano , and s. tomsovic , phys .
rev .
lett . , * 90 * , 176801 ( 2003 ) .
l. van hove , phys .
rev .
, * 95 * , 249 ( 1954 ) .
j. p. hansen , and i. r. mcdonald , _ theory of simple liquids,_third edition , academic press ( 2006 ) .
though the classification of temporal regimes holds for all realizations of irregularity ( combination of @xmath32 ) , the width of each regime show statistical fluctuations over realizations .
thus , figs . 1 , 2(a , b ) and 3 display results for a specific @xmath100 and @xmath101 . c. donati , j. f. douglas , w. kob , s. j. plimpton , p. h. poole , and s. c. glotzer phys .
rev .
lett . * 80 * , 2338 ( 1998 ) .
the coalescing azimuthal motion in the cwm at large @xmath0 is different from a rigid rotation , which are easily tracked in the periodic modulation of the mean - square displacemt , and subtracted from all our results .
nitin kumar , harsh soni , sriram ramaswamy , a. k. sood nature comm , * 5 * , 4688 ( 2014 ) .
m. v. chubynsky , and g. w. slater , phys .
rev .
lett . , * 113 * , 098302 ( 2014 ) .
b. wang , j. kuo , c. bae , and s. granick , nat .
mater . , * 11 * , 481 ( 2012 ) . we obtain @xmath68 using msd , @xmath102 for all particles @xmath103 .
also @xmath104 for a fixed @xmath0 and the equivalence was used for convenience .
note that @xmath105 , where @xmath106 is the pair correlation function .
d. r. nelson _ defects and geometry in condensed matter physics _ , cambridge university press ( 2002 ) .
s. karmakar , c. dasgupta , and s. sastry , annu .
rev .
cond .
matt .
, * 5 * , 255 ( 2014 ) .
* spatio - temporal correlations in wigner molecules : supplementary information * biswarup ash^1^ , j. chakrabarti^2^ , and amit ghosal^1^ + we begin this supplementary material elaborating the procedure of obtaining @xmath37 as a function of @xmath0 , as presented in fig .
2(c , d ) of the main paper . |
as indicated in @xcite , astronomical photometry is about the measurement of the brightness of radiating objects in the sky .
many factors , such as those coming from the instrument limitation , from a fixed measurement strategy , or the limitation from the mean through which the measurement is taking place , make this area of the science relatively imprecise .
the improvement in the dectectors technology plays a key role in the area of optimizing the resulting astronomical photometric measurements . in this sense , a signal - to - noise ratio capable of being configured as part of an optimization framework of the measurement system seems to be a useful input .
+ charge - coupled devices ( ccds ) constitutes the state - of - the - art of detectors in many observational fields .
@xcite enumerates the areas involved in the recent advances of the ccds systems , which are : * manufacturing standards that provide higher tolerances in the ccd process leading directly to a reduction in their noise output . *
increased quantum efficiency , especially in the far red spectral regions .
* new generation controll electronics with the ability for faster readout , low noise performance , and more complex control functions . * new types of scientific grade ccds with some special properties . any data , in general , is always limited in accuracy and incomplete , therefore , deductive reasoning does not seem to be the proper way to prove a theory .
however , and as said in @xcite , statistical inference provides a mean for estimating the model parameters and their uncertainties , which is known as data analysis .
it also allows assessing the plausability of one or more competing models .
the use of a bayesian approach here is also justified in @xcite where it is stated that for data with a high signal - to - noise ratio for example , a bayesian analysis can frequently yield many orders of magnitude improvement in model parameter estimation , through the incorporation of relevant prior .
this is exactly what we intend through the implementation of what will be described in this paper , and detailed in the following section .
the problem to be resolved here consists in the implementation of bayesian inference for a set of configurable parameters which affect the signal - to - noise ratio of a measurement . in comparisson with other methodologies , such as anova , which shows serious weakness
when outliers are present in the measured data , bayesian parameter inference offers a robust method against outliers .
it also allow to improve the results of inference by using the posterior probability density distribution ( pdf ) of one execution as the prior pdf for another execution in a recursive framework .
this will lead to an adaptive measurement strategy which can be addressed as a calibration refinement .
professional surveys plan the measurement strategy well in advance , taking into account all the relevant factors impacting on the measurement ; this involves the set of specified fix parameters from the detector and also a set of parameters which configure the measurement , such as integration time , diameter of the aperture , etc .
once a measurement has finished , the data are archived and their analysis and processing begin .
the problem proposed here is to establish a link between the results of a measurement under a specific detector configuration and the refinement , by application of parameter bayesian inference , of the configuration parameters to be applied in a further measurement .
the result of this bayesian inference at parameter level is proposed to be injected as additional knowledge for a model selection problem in the context of measurement data analysi ( i.e , photometric cross - matching of multiwavelength astronomical sources ) .
for example , let us imagine that we have performed colour measurement in a multi - wavelength survey with ten different instruments , each one under a specific configuration .
let us imagine that in the process of model comparison for the cross - matching scenarios , the existence of a source inferred in a bandwith which is not detecting it , is plausible .
then , based on this result , a new configuration for that instrument can be inferred in such a way that allows us to explore the refined plausability indicated by the model comparison from the data obtained in the first measurement loop .
figure [ fig1 ] shows a block diagram reflect in general lines the idea of the problem proposed here + [ fig1 ]
the great majority of detectors used in the astronomical field are silicon - based ones ; this means that the electronics involved in the specification , manufacturing and operational life of these detectors are relvant to the outcome obtained .
as detailed in @xcite , the excitations of electrons responding to incident photons constitutes the fundamentals for practical flux measurement in almost all nowadays photometric systems .
+ in nowadays , ccds are used in many instruments involved in the main current astronomical surveys ; an extensive and increasing bibliography is currently being publishing .
therefore our paper will focussed on this type of instruments , however we have tried to retain the generic aspect of the characterization of any other type of detector and obviously the methology presented here is fully valid with any other set of specific parameters . a summary from @xciteon the relevant information related to detector parameters has been included in this subsection for the sake of clarity .
a more detailed information can be found in @xcite + in general terms and for the context of our problem , a detector can be characterized by the following parameters , as indicated in @xcite : + * * quantum efficiency ( q ) * : it is the ratio between registered events and the incident photon . * * information transfer efficiency ( e ) * : it is the ration between the square of the signal - to - noise ratio of the output and the square of the signal - to - noise ratio of the input .
+ @xmath0 * * noise equivalent power ( nep)*:it is defined as the optical power producing an output equal to the noise level of the detector . @xmath1 * * linearity and saturation * : one of the most relevant characteristics of a ccd detector is its linear property .
this means that , under ideal scenario ( ignoring effects such as , noise , dark current , polarization current , sky background contribution etc ) , the intentsity registered in each pixel ( as elecrons ) is proportional to the incident light .
however this linear behaviour has its limits .
the most obvious one is the * saturation threshold * , which is measured by the * full - well capacity * * * full - well capacity*:this parameter measures the limit of the accumulated charge before saturation begins .
this value is normally included in the technical specification of the detector s supplier . * * event or pixel capacity * : it is defined as how many events can be usually accumulated before some saturation effect takes place . * * working range * : also named * dynamic range * , is essentially the same as event capacity for a noiseless accumulative single detector , but more generally interpreted as the difference between useful maximum and minimum event counts . * * gain(g ) * : the digital image consists of a table of numbers which indicate the intensity registered in each pixel .
however , the numbers stored do not mean the quantity of electrons found in each electrode , as this quantity can be huge and this would make the resultant storage files too big .
therefore what is normally done is to divide the quantity of electrons by a certain number , named * gain * ; thus , what we register in the file is the number of * counts * obtained when performed the above division .
sometimes counts are called * adu * ( analog - to - digital units ) or * dn * ( data number ) .
the gain is therefore measured in electrons per count . + some cameras allow the user to choose the gain
. then we could choose a small value for faint detections or bit to measure correctly sources of various brightness , but all this without overpassing the limit done by : + @xmath2 * * counts(c ) * : the number of photons that fall on a pixel is related to the counts by : + @xmath3 detector parameters such as quantum efficiency , linearity event capacity and so on , are often characterized by * figures of merit * , which manufacturers quote about their products .
the error which all scientific measurements should carry means really uncertainty and it is due to noise . following the line of discussion presented at the begining of this section , once an electron has been excited by a photon , the next step consists in registering this event by the electronics of the detector . in this process ,
handling and reducing the number of extraneous electron activity which does not come from the source subject to the pure measurement ( noise ) consitutes a delicate and complex step of this process .
a summary from @xciteon the relevant information related to sources of noise has been included in this section for the sake of clarity . for more detailed information about the functional block description of any detector can be found in @xcite + there are numerous sources of noises in the ccd images .
the following list identify those sources of noise which are more relevant to the problem described in section [ sect:2 ] : * * dark current * ; also named * thermal noise * , it is produced by the spontaneous generation of electrons in the silicium ( si ) due to the themal excitation of the material .
+ the noise associated with dark current in a ccd is regarded as having a spatial dependence , in that it relates to minor irregularities in the solid state molecular bonding lattice , associated with material nterfaces and impurities .
each pixel generates a slightly different level of dark current , so the noise depends on non - uniformity of the response over the surface as well as the dark current s inherent quasi - poissonian contribution to the electron counts .
dark current , following a generally richardson law type dependence on temperature , could fill the potential wells of an uncooled ccd at @xmath4 in , typically , a few seconds , but cooling to @xmath5 say , reduces this to a tolerable few electrons per pixel per second .
the standards approach to spatial non - uniformity of response in ccd is through * flat fielding * , although it should be rememberd that there is an additional shot noise contribution to the adopted flat field contribution .
+ within the ccd , each pixel has a slightly different gain or quantum efficiency when compared with its neighbors . in order to flatten the relative response for each pixel to the incoming radiation , a flat field image
is obtained and used to perform this calibration .
ideally , a flat field image would consist of uniform illumination of every pixel by a light of source of identical spectral response to that of the object frames .
once a flat field image is obtained , one them simply divides each object frame by it implementing then instant removal of pixel - to - pixel variations . * * cosmic rays * : they are part of the inhabitants of the interstellar space ; cosmic rays are sub - atomic particles which go at very high speed , near to the speed of light .
cosmic rays are very annoying for the astronomers because they interact with the silicium of the coupled charged devices .
these electronic micro - aluds , concentrated in a few pixels appear in the images as points , or sometimes , lines very bright . as much the exposition time is , more quantity of cosmic rays will deteriorate it . * * read noise * ; this is a very important contribution to the toal noise of the images . this noise is due to the random and unavoidable errors that are produced during the reading of the image , in the process of amplification and counting of the electrons captured in each pixel .
these errors are intrinsic to the nature of the detector device .
+ the existance of read noise has always to be taken into account because it affects to all steps towards the obtainment and traitment of the digital images .
each camera must have its own level of read noise , documented in its technical specifications .
* * shot noise * : this is a statistical noise due to the inherently non - steady photon influx .
a poissonian distribution is normal for the arrival of the primary photon stream from a source of constant emissive power . * * clocking noise * ; this noise comes from the various high frequency oscillators involvd in the gating circuitry .
this noise rises with load and clocking frequency , but it can normally be controlled by manufacturers to a negligible level for astronomical applications . * * atmosphere * : the total noise of detection is not just that of the photoelectric effect on the detector , since the signal has already been deteriorated by the atmosphere .
further reading on the problems with atmosphere in the astronomical measurements can be found in @xcite for simplification purposes in the resolution of the problem describe in [ sect:2 ] , only the following list of noise sources will be considered representative for the context of our problem .
however the methodology can be extended to a more exhaustive list of noises : dark current , read noise , background noise .
as detailed in @xcite , a careful understanding of the main sources of uncertainties can suggest ways to improve our measurement strategy , this means , the observation , reduction and analysis processes .
+ a crucial concept in photometry is the * signal - to - noise ratio ( s / n ) * , which is equivalent to the concept of percentage error @xmath6 + it is very important to assess and , if possible , to reduce the noise of the images.in this direction , the parameter @xmath7 is very useful in the assessment of the feasibility , reliability and quality of the detection . as a general rule , and based on the considerations expressed in @xcite , to get reliable photometry and/or astrometry measurements , the minimum threshold must be : @xmath8 + as a first preliminary simplification , under the asumption of photon noise dominating the noise , the counting statistics of the number of photons impacting on a given area per second can be modelled by a gaussian distribution , where the scatter is the square root of the number of photons , therefore : @xmath9 where @xmath10 includes the photon counts for the sky foreground and the sky background , both of them carrying noise components . to obtain the count from the source alone ,
the sky background contribution has to be substracted . then , considering this two contributions , we can write the following : @xmath11 the s / n ratio changes with the integration time , @xmath12 and with the telescope aperture,@xmath13 , therefore these two parameters encompass the configuration domain by which the s / n ratio can be optimized in the process explained in section [ sect:2 ] .
+ for the telescope aperture , we know that increasing the diameter of the telescope primary,@xmath13 , by a factor of 2 increases the collecting area by factor of 4 , thus for a given integration time , @xmath12 we get : @xmath14 + regarding the integration time dependency with the signal - to - noise ratio , we can write : @xmath15 where @xmath16 is th count rate expressed in @xmath17 .
therefore , @xmath18 similarly the number of exposures impact on the noise of the resulting image . according to @xcite if the exposure is broken into @xmath10 equal short exposures , the error in the mean of the measurements would be : @xmath19 where @xmath20 is the scatter in the individual short exposures and @xmath21 is the number of such short exposures.it is important to keep in mind that if we add @xmath10 images resulting from @xmath10 short exposures , the resulting signal is @xmath22 , and total noise of the resulting image will be @xmath23 as a conclusion , the signal - to - noise ratio of an addition of images from the same object is bigger and therefore better than the signal - to - noise ration of each individual image .
so far , photon noise has been assumed to be the dominant contribution of the noise , and therefore the other noise sources contributions have been reduced to zero for the @xmath7 equations above . however , in the case of faint sources detection the dark current and the read noise can play a key role in the @xmath7 , therefore , we will develop the @xmath7 equation , with the integration time dependency and including at least the following sources of noise : photon noise , dark current and read noise .
for each of the noise sources , valid approximations will be considered in order to obtain a final @xmath7 equation which is computational cost affordable for a intel mac core 2 duo computer . therefore , the equation for the @xmath7 of a measurement made with a ccd can be given by : @xmath24 where : * @xmath25 is the total number of photons which compound the signal detected from the object * @xmath26 is the total number of photons coming from the background also called sky * @xmath27 is the total number of dark current electrons per pixel * @xmath28 is the total number of electrons per pixel resulting from the read noise .
the noise terms in equation [ eq8 ] can be modelled by poisson distributions .
the term @xmath29 is used to apply each noise term on a per pixel basis to all the pixels involved in the @xmath7 measurement .
a more complete equation taking into account digitization noise within the a / d converter can be found in @xcite + as explained in @xcite , and using the fact that @xmath30 , a standard error for the measurement can be obtained as : @xmath31 where @xmath32 is equal to the noise terms indicated in [ eq8 ] , and @xmath33 is the correction term between an error in flux ( electrons ) and that same error in magnitudes ( howell , 1993 ) .
the equation [ eq8 ] can also be expressed in terms of count rate and integration time , as follows : @xmath34 and in line with the text above , each terms of equation [ eq8 ] can be expressed as follows : @xmath35 @xmath36 @xmath37 where @xmath38,@xmath39 and @xmath40 are proporcionality constants .
in general , as described in @xcite , a bayesian probability density function is a measure of our state of knowlege of the value of the parameter .
when we acquire some new data , bayes theorem provides a means for combining the information about the parameter coming from the data , through the likelihood function , with the prior probability , to arrive at a posterior probability density , @xmath41 , for the parameter .
+ let us be @xmath42 the known model for the signal - to - noise ratio of a multi - wavelength measurement system which identifies cross - matching sources , @xmath43 the set of configurable parameters,@xmath44 , being @xmath12 the integration time and @xmath13 the configurable aperture diameter of the instrument , and @xmath45 the data of the measurement .
@xmath46 is the information associated to the model . by application of bayesian inference , the configurable parameters for a established model @xmath42 can be estimated as follows : @xmath47 this equation can be written in the following way : @xmath48 therefore it becomes evident that the denominator of [ eq2 ] is just a normalization factor and we can focus on just the numerator , where @xmath49 is named the probability a priori and @xmath50 is the likelihood . strictly speaking , as it is explained in @xcite , bayesian inference
does not provide estimates for parameters ; rather , the bayesian solution to the parameter estimation problem is the full posterior pdf , @xmath51 and not just a single point in the parameter space .
it is useful to summarize this distribution and one possible candidate of the best - fit value is the posterior mean , @xmath52 we will develop here the bayesian estimation parameter for the integration time @xmath12 , however the consideration of more than one parameter is immediate and in that case multiple integrals will be considered for the marginalization of the corresponding nuisance parameters . a preliminary choice for the prior probability will be a uniform distribution , therefore : @xmath53 where @xmath54 and @xmath55 are the maximum and minimum integration time to be defined for the observation under consideration . in general ,
the difference between the data and the model is called the error , therefore : @xmath56 the model here @xmath42 is the one proposed in [ eq9],however for the sake of clarity in the following expressions , we will consider that equation [ eq9 ] can be simplified as follows : @xmath57 being @xmath58 a constant the likelihood for all the channels @xmath59 under consideration in a multiwavelength observation , can be expressed as follows : @xmath60 now we can compute @xmath61 as indicated in [ eq14 ] , and considering a normalizing unity constant for the denominator we obtain the following result : .
the main purpose of the methodology presented here is to enable the capability of an observational system to adapt its configurable parameters depending on the results from previous observations . in this way , it is clear that the integration time has a considerably strong impact on the faint sources detections , therefore the approach covered in the previous chapter of this paper intends to infer the optimum integration time in order to increase the probability of detection for the next observation to be planned and defined .
the inclusion in the model of the main contributions of noise leads to a more refined parametric inference .
a toy example has been built in order to show some preliminary results derived from the approach presented in this paper .
figure [ fig2 ] represents the data in terms of integration times from several observations and figure [ fig3 ] shows the probability density corresponding to the posterior probability solved as detailed in the previous section .
a real example is planned to be implemented using a model of signal - to - noise ratio in line with the expression presented in equation [ eq20 ] [ fig1 ] [ fig1 ]
the authors would like to thank dr .
luis manuel sarro baro and dr .
tamas budavari for his support and valuable insights in the area of bayesian inference , and mr .
georges bernede for his continuous support in the effective conciliation of my professional and academic activities .
h. kopka and p. w. daly , _ a guide to latex _
, 3rd ed.1em plus 0.5em minus 0.4emharlow , england : addison - wesley , 1999 .
steve b. howell , _ handbook of ccd astronomy _ , 2nd ed.1em plus 0.5em minus 0.4emnational optical astronomy observatory and wiyn observatory : cambridge university press .
edwin budding and osman demircan , _ introduction to astronomical photometry _ , 2nd ed.1em plus 0.5em minus 0.4emcanakkale university , turkey : cambridge university press . w. romanishin , _ an introduction to astronomical photometry using ccds _ ,
september 16 , 2000.1em plus 0.5em minus 0.4emuniversity of oklahoma .
david galadi enriquez and ignasi ribas canudas , _ manual practico de astronomia con ccd _
, 1998.1em plus 0.5em minus 0.4emdepartment of astronomy and metrology of university of barcelona : omega . j. v. wall and c. r. jenkins , _ practical statistics for astronomers _
, 2003.1em plus 0.5em minus 0.4emuniversity of oxford and schlumberger cambridge research : cambridge university press .
michael p .hobson , andrew h. jaffe , andrew r. liddle , pia mukherjee and david parkinson , _ bayesian methods in cosmology _
, 2010.1em plus 0.5em minus 0.4emcavendish laboratory , university of cambridge , imperial college , london , university of sussex : cambridge university press .
phil gregory , _
bayesian logical data analysis for the physical sciences _
, 2005.1em plus 0.5em minus 0.4emdepartment of physics and astronomy , university of british columbia : cambridge university press | key words : signal ; noise ; gain ; quantum efficiency ; count ; rad noise ; dark current ; nuisance parameters + calibration is nowadays one of the most important processes involved in the extraction of valuable data from measurements .
the current availability of an optimum data cube measured from a heterogeneous set of instruments and surveys relies on a systematic and robust approach in the corresponding measurement analysis . in that sense , the inference of configurable instrument parameters can considerably increase the quality of the data obtained .
+ any measurement devoted to scientific purposes contains an element of uncertainty .
the level of noise , for example , determines the limit of usability of an image .
therefore , a mathematical model representing the reality of the measured data should also include at least the sources of noise which are the most relevant ones for the context of that measurement . + this paper proposes a solution based on bayesian inference for the estimation of the configurable parameters relevant to the signal to noise ratio .
the information obtained by the resolution of this problem can be handled in a very useful way if it is considered as part of an adaptive loop for the overall measurement strategy , in such a way that the outcome of this parametric inference leads to an increase in the knowledge of a model comparison problem in the context of the measurement interpretation .
+ the context of this problem is the multi - wavelength measurements coming from diverse cosmological surveys and obtained with various telescope instruments . as a first step ,
, a thorough analysis of the typical noise contributions will be performed based on the current state - of - the - art of modern telescope instruments , a second step will then consist of identifying configurable parameters relevant to the noise model under consideration , for a generic context of measurement chosen .
then as a third step a bayesian inference for these parameters estimation will be applied , taking into account a proper identification of the nuisance parameters and the adequate selection of a prior probability .
finally , a corresponding set of conclusions from the results of the implementation of the method proposed here will be derived |
to date , eight separate luminosity relations have been identified for long duration gamma - ray bursts ( grbs ) ( schaefer and collazzi , 2007 ) .
these relations correlate a burst s peak bolometric luminosity with various light curve and spectral parameters .
the possible utility of these relations is that we can then use grbs as tracers of the high - redshift universe .
one of these eight relations is the @xmath9 relation proposed by firmani et al .
( 2006 , hereafter named the firmani relation ) . here
, @xmath2 describes the peak of the @xmath10 curve ( proportional to @xmath11 ) , which is the photon energy of the peak spectral flux ; @xmath1 is the isotropic luminosity of the burst measured bolometrically ( 1 - 10,000 kev in the burst rest frame ) .
@xmath3 is the reichart definition of a grb time duration ( reichart et al .
2001 ) where the duration is the total time interval of the brightest bins in the light curve that contains 45@xmath12 of the burst fluence .
the firmani relation was presented as an improvement over the @xmath0 relation .
nineteen grbs were used to demonstrate a tight correlation with a reduced chi - square of 0.7 over 16 degrees of freedom ; the resulting luminosity relation being @xmath13 .
the reported scatter in the firmani relation is substantially smaller than those of most other grb luminosity relations .
this result offers the hope of substantial improvement in the accuracy of grbs for cosmological distance measures .
we have no physical reason to expect that the addition of a duration should make for a tighter relation .
nonetheless , we should look to see if we can get tighter luminosity relations from duration definitions other than @xmath3 , as @xmath3 may not be the optimal duration to use .
we have no physical reason to expect that any one definition of duration is best , while the particular choice of the reichart definition was used only for historical reasons no longer of any relevance .
for example , we could still use the reichart definition , but measure the duration over a different percentage of the burst fluence .
so perhaps the use of a duration based on 30% or 60% ( @xmath7 or @xmath8 ) might be better .
alternative definitions of duration can be considered instead of the reichart formulation .
for example , we can try to define the duration as equalling the total fluence divided by the peak flux to get a sort of equivalent width ; alternatively , we could use the familiar @xmath4 or @xmath5 durations
. a wide variety of alternative durations can be defined , and we do not know which one will be optimal . in this section 2 ,
we first reproduce the firmani relation for the original data of firmani et al .
( 2006 ) as a test that we are using identical fitting procedures .
then we test the firmani relation with a set of independent data for the same 19 bursts .
a further test of the firmani relation is made with a much larger sample of 60 bursts . in section 3 ,
we test many duration definitions in a firmani - like relation to see which produces the ` tightest ' correlation .
section 4 contains a simple and forced derivation of the firmani relation from two other luminosity relations .
we first started with the model as stated in firmani et al .
( 2006 ) : @xmath14 we generalized this equation , and put it into logarithmic form : @xmath15 here @xmath16 , @xmath17 , and @xmath18 are the fit parameters derived from fitting a set of grb data , which can then predict model values of @xmath19 .
for the one - sigma uncertainty used in evaluating the chi - square , we used an elliptical error box to account for the errors in the measured quantities on both axes .
specifically , @xmath20 with the model values for the luminosity and its uncertainty as well as the observed luminosity ( all in logarithmic space ) , we can calculate the reduced chi - square for the model fit . to perform this test on the 19 bursts , we used data as reported in firmani et al .
( 2006 ) : their values for @xmath3 were the observed durations , and thus needed to be corrected for time dilation into the burst frame via dividing by @xmath21 .
their values of @xmath2 were already in the burst frame , and did nt need to be corrected .
@xmath1 was obtained by taking their value of @xmath22 and multiplying it by their value @xmath23 .
we also used firmani s reported redshift values .
our best fit had a @xmath24 with an root - mean - square ( rms ) scatter of the observed values of @xmath19 about the best fit model equal to @xmath25 . with 19 burst and 3 fit parameters , we have 16 degrees of freedom .
this is in agreement with the reported value of @xmath26 reported by firmani et .
al . ( 2006 ) .
we also agree with the best fit parameters and their uncertainties as reported by firmani et al .
( 2006 ) and as given in eq .
thus , we are able to reproduce their reported result
. we will use these procedures for all subsequent fits .
figure 1 shows the firmani relation as we have reproduced it .
given the nature of observational data , there are inevitably differences in the various published values of all these burst properties . for instance , firmani et al .
( 2006 ) report @xmath2= 685 @xmath27 133 kev for grb971214 , while jimenez et al .
( 2001 ) reports a value of 840 @xmath27 88 kev . for the same burst ,
the peak flux is slightly different for different detectors , so fimani et al .
( 2006 ) report @xmath19 equals 52.86 @xmath27 0.08 , while we derive a value of 52.92 @xmath27 0.01 ( schaefer 2007 ) .
the firmani relation should be robust on the use of independent measures from different published sources , so we should be able to obtain the same result with these different reported values .
for this test , we have collected independent measures for the luminosity , peak energy , and @xmath3 from various published reports .
we have no reason to think that either set of values for these 19 grbs is better or worse in accuracy .
the independent values for these burst qualities are in table 1 .
column 1 is the grb designation ; column 2 is the redshift ; column 3 is the log of the isotropic bolometric luminosity ; column 4 is the photon energy of the observed peak spectral energy ( which needed to be blueshifted into the bursts frame ) ; and column 5 is the reichart duration @xmath3
. the values and their references for the luminosities , peak energies , and redshifts can be found in schaefer ( 2007 ) .
we measured the durations for @xmath3 from the light curves referenced in schaefer ( 2007 ) .
the formal measurement error bars quoted for the @xmath3 values in table 1 are fairly small , but the real total uncertainties are substantially larger . we know this because various groups have reported measures of @xmath3 for many of the same grbs and the scatter is much larger than anyone s quoted error bars . in particular ,
we have collected independent measures reported by guidorzi ( 2005 ) , firmani et al .
( 2006 ) , rizzuto et al .
( 2007 ) , rossi et al .
( 2008 ) , table 1 of this paper , and independent values calculated from our own group , with an average of five values for each of our bursts in table 1 .
we find that the median scatter is 17%
. this value changes little if we look at bursts measured with one satellite ( _ swift _ ) alone or if we only look at results from our group .
we think that this variation arises from relatively small changes resulting from differing time bin sizes and time intervals for the calculation .
( similar scatter is found for other duration measures , see for example koshut et al .
1996 and norris et al . 1995 . )
this additional systematic error contributes a small fraction of the extra scatter that we observe for the firmani relation .
the reason for this small contribution is that the extra systematic error on @xmath3 is @xmath2812% and the values are included nearly as a square root ( see eq .
1 ) , so the extra contribution to @xmath29 is 0.026 ( see eq .
3 ) . all this is to say that the systematic errors in measuring @xmath3 are much larger than has ever been realized , yet even these additional uncertainties are negligibly small . upon optimizing the fit with the independent values ,
the equation for the best fit is : @xmath30 a comparison with eq .
1 shows that the two best fits are similar , with the exponent for the @xmath2 being moderately different . the firmani relation for this independent data
is displayed in figure 2 .
the obvious difference between figures 1 and 2 is that figure 2 has a much larger scatter than in figure 1 .
the rms value for the independent data was 0.35 , whereas the rms for the data from firmani et al .
( 2006 ) is 0.14 .
the reduced chi - square for the 19 bursts about this best fit model is @xmath31 for the independent data .
this is greatly larger than the value of @xmath32 obtained from the data from firmani et al .
( 2006 ) . with this large reduced chi - square
, we realize that there must be some additional source of systematic uncertainty that is beyond that from ordinary measurement errors .
we can introduce an additional figure of merit which quantifies the scatter about the best fit firmani relation .
this is the systematic error required to be added in quadrature to the measurement error such that the resulting reduced chi - square equals unity .
a desirable fit with little scatter will have a small required systematic contribution to the uncertainties , whereas a poor fit with large scatter will have a large required systematic contribution .
for this , we will assume that the systematic error is a constant , even though the reality is likely more complex in ways that we can not now see .
in essence , we are calculating the uncertainty in the chi - square calculation as @xmath33 for the case of firmani s 19 grbs with his data , there is no required additional systematic uncertainty ( as indicated by the reduced chi - square being less than unity ) . but for the independent data for the same bursts , we require a systematic error of 0.34 ( in logarithmic units ) to be added in quadrature so as to get an acceptable fit with a reduced chi - square of unity .
so in all , we can quantify the quality of the firmani relation for any data set by three parameters , @xmath34 , rms , and @xmath35 .
table 2 summarizes these parameters for the firmani relations with various data sets .
our next step was to add more bursts .
that is , the firmani relation should be robust when applied to a much larger sample of bursts .
specifically , we used 60 of the bursts as given in table 1 , with the observed burst properties as collected in schaefer ( 2007 ) based on references reported therein .
schaefer ( 2007 ) tabulates 69 bursts in all , but several had to be omitted for various reasons .
grb 980613 , grb 990712 , grb 011211 , and grb 020903 were not used as we did not have the light curves for duration calculations .
grb 050824 was omitted because the value of @xmath2 is only an upper limit .
grb 050319 , grb 050408 , grb 050802 , and grb 051111 were omitted due to the reported value of @xmath2 in schaefer 2007 not having been directly measured . to remain consistent , we only used data reported in schaefer ( 2007 ) for the peak energy and redshift .
the values for both @xmath1 and @xmath36 were derived from values for the bolometric peak flux ( @xmath37 ) reported by schaefer ( 2007 ) . with the standard inverse square law , we get @xmath38 the luminosity distance ( @xmath39 ) to the grb is calculated with the measured spectroscopic redshift , assuming the concordance cosmology ( @xmath40=0.27 in a flat universe with @xmath41 ) . with this independent data set for 60 grbs , we fitted the model in eq .
the equation for the best fit for this extended sample is : @xmath42 this best fit model is similar to the best fits with the 19 grb subsample ( cf .
eqs 1 and 4 ) .
the resulting firmani relation is plotted in figure 3 .
again , the immediate reaction is that the figure displays a lot of scatter , and much more scatter than in either figures 1 or 2 .
quantitatively , the comparisons are presented in table 2 .
we see that the rms scatter has risen to 0.41 , which is greatly larger than in the earlier figures .
the reduced chi - square of the fit is @xmath43 , which shows that there is some source of scatter that is much larger than produced by the simple measurement uncertainties in the input parameters .
the systematic error for the 60 burst sample is 0.38 , which is substantially larger than for either data set in the 19 burst subsample .
the primary result from this section is that the firmani relation is neither robust to the use of independent data nor robust to the extension to many more bursts .
now that we have seen how the firmani relation behaves for our sample of 60 grbs , we should try the same test procedures for various different duration definitions . for this
, we use a generalized form of eq . 2 :
@xmath44 here , the duration has been generically labeled as @xmath45 , and the denominators inside the logarithms are constants equal to the average @xmath45 and @xmath2 values for the data set .
the reason to have these averages in the denominator is to improve the convergences of the fits by avoiding long thin error regions with strong correlations between fit parameters .
there are many alternative ways to measure duration .
for example , we could use the reichart definition , but with a different percentage of the total fluence to take the duration over .
in other words , we could expand the reichart definition of duration out to say @xmath8 , or contract it to @xmath7 .
again , there is no reason to believe that using the exact duration as proposed by reichart ( @xmath3 ) would be any more effective than the others .
other duration definitions are reasonable , and indeed much easier to calculate .
we could adopt a duration defined as the time a burst spends above @xmath46 of the peak flux of the burst ( @xmath47 ) .
the well - known definitions of duration , @xmath4 and @xmath5 , should also be included . here ,
the two durations are the time interval containing the central 90 or 50 percent of the fluence of the burst , respectively .
we also take the bolometric fluence @xmath48 and divide it by the bolometric peak flux @xmath37 to get a sort of ` equivalent width ' . as a control ,
we have also adopted a case where all the burst durations are set equal to a constant , which we arbitrarily take to be @xmath45=10 seconds .
the chosen value does not matter , as different choices will merely result in a different @xmath16 value that will not change the quality of the fit . by taking a constant duration
, we are transforming the firmani relation ( @xmath9 ) into the old @xmath0 relation .
a comparison of the scatter in the @xmath45=10 seconds relation versus the generalized firmani relations will tell us whether the addition of a time scale has substantially improved the quality of the luminosity indicator .
so far , the alternative durations have all been measures of the total duration of the burst .
however , the physics of the luminosity relations points to the correlations as being with the individual peak pulse and not the overall set of pulses that make up the entire light curve ( schaefer 2003 ; 2004 ) .
so we should here consider various pulse duration measures .
a simple and reasonable means of doing this is to take all of the overall - burst - duration measures and divide by the number of pulses in the light curve ( @xmath49 , as defined in schaefer 2007 ) .
this immediately doubles the number of trial definitions considered .
we are left with a total of 32 duration definitions , all of which are listed in table 3 .
for all 60 bursts in our sample , we have measured the durations according to all 32 definitions .
we have calculated these durations from the light curves as given on the batse , hete , _ konus _ , and _ swift _ websites as well as the values reported in schaefer ( 2007 ) . for @xmath4 and @xmath5 , we have almost always been able to find values reported by the instrument teams ( see schaefer 2007 for references ) .
we fitted eq .
8 for all 60 grbs for all 32 duration measures . for each best fit relation
, we calculated the rms and @xmath35 values as quantitative figures of merit .
these are summarized in table 3 .
the results indicate that while there are certainly differences between duration definitions , the differences tend to be small .
the smallest values of rms and @xmath35 occurs for durations defined as @xmath50 .
the scatter in this best relation is somewhat smaller than for the original firmani relation ( with @xmath3 in the third line of table 3 ) .
however , we do not think that the differences are significant .
our reason is that there will be noise in the figures of merit which will inevitably produce one duration definition as being the best even if the values were uncorrelated or random , and we see the scale for such variations by comparing the values in table 2 .
that is , the variations of the figures of merit in table 3 are consistent with the case where duration information is not correlated with @xmath1 , and a different set of 60 grbs would randomly produce a different best definition .
as such , we do not think that we should replace the firmani relation with a luminosity relation involving @xmath50 .
a particularly important comparison is between the firmani relation and the @xmath0 relation ( represented by the line with the durations all taken to be a constant of 10 seconds ) .
we see that the @xmath0 relation is the third poorest relation in the table .
nevertheless , the difference is not large enough to evaluate as being significant .
that is , the differences in the figures of merit ( 0.09 in the rms and 0.08 in @xmath35 ) are too small to view as necessarily arising from a physical effect in the bursts . we know this because the variation caused by simple sampling effects ( see the last two lines of table 2 ) are of order 0.06 in the rms and 0.04 in @xmath35 . as such
, we view the firmani relation as having a similar scatter as the @xmath0 relation .
in a recent independent preprint , rossi et al . ( 2008 ) also examined the firmani relation , in particular with a comparison to the amati relation .
they use an extended sample of 40 _ bepposax _ and _ swift _ bursts , with little overlap with our sample of 60 grbs .
their best fit is somewhat different from those in eqs 1 , 4 , or 7 ; with their fitted firmani relation scaling as @xmath51 . they realized that this firmani relation is essentially identical to the amati relation ( amati et al .
2002 ) , which gives the isotropic energy emitted in gamma radiation over the whole burst duration as @xmath52 . with the reasonable approximation that the total energy in the light curve equals the peak luminosity times the duration ( @xmath53 ) , the amati relation ( @xmath52 ) is transformed into their firmani relation ( @xmath51 ) .
while our exponents in the firmani relation are somewhat different , the rossi derivation demonstrates that the fermani relation has a physical basis that is close to that of the amati relation .
rossi et al . ( 2008 ) further go on to show that the scatter in their firmani relation is comparable to that in the amati relation , which is another way of saying that the two relations are not independent . here
, we will directly derive the firmani relation from _ both _ the @xmath0 and amati relations .
let us start with the relation @xmath54 as given in schaefer ( 2007 ) .
this can be rearranged as @xmath55 .
the amati relation ( @xmath52 ) can be inserted to get @xmath56 .
now we can select one of our duration definitions with @xmath57 .
the ratio of fluence to peak flux will equal to the ratio of the burst energy and the peak luminosity , so we have @xmath58 .
this can now be substituted to obtain @xmath59 .
we see that we have just derived eq . 8 with @xmath60 and @xmath61 , values which are characteristic of the fitted firmani relation ( cf .
eq . 7 ) . with this
, we see that the firmani relation has no independent existence because it is only a combination of two simpler relations .
thus , given any two of these three relations , we can derive the third .
a question is which of these is more fundamental .
the inherent problem with making this assessment is that it comes down to how one identifies the more fundamental of relations that really address different physics . by occam
s razor , the more fundamental relation would be one that has the fewest parameters , while accurately and efficiently yielding a luminosity .
thus , the firmani relation is less fundamental as it uses more parameters for the fit .
of the two relations that remain , the one that has the most ` utility ' will be the one with the least amount of scatter in its calibration curve .
we were able to reproduce firmani s results over his small sample of 19 bursts .
however , when we substitute independent values for @xmath1 , @xmath2 and @xmath3 , we see a substantial broadening around the model . that is , the firmani relation is not robust on the use of alternative input data .
in addition , when we extend the test to a larger sample of 60 bursts , the scatter becomes substantially larger again .
indeed , this scatter is comparable to the scatter in the original @xmath62 relation .
that is , the firmani relation is not robust for the use of additional bursts .
these failures of the firmani relation have dashed our hopes raised by the tight calibration curves displayed in firmani et al .
it also suggests that the addition of a duration does not significantly improve the @xmath0 relation .
the larger point of interest is that no duration shows a significant advantage over the @xmath62 . while it might be possible
that that a relation involving @xmath50 might really have a smaller scatter than the firmani relation , the improvements are small and not significant .
this leads us to conclude that the addition of a duration is not doing enough to improve the @xmath0 relation to be considered to be a separate luminosity relation .
we conclude that the firmani relation is not useful for several reasons : first , the firmani relation is simply derived by putting together two well - known , simpler , and independent luminosity relations , and thus it has no separate existence .
second , it is not robust for the inclusion of independent input data or for the extension to many more grbs . third , the real scatter for the firmani relation does not live up to the hope generated by the original report , with the scatter being comparable to those of the luminosity relations from which it is derived . in all , we can see no utility or advantage to using the firmani relation .
amati , l. et al .
2002 , a&a , 390 , 81 .
firmani , c. , ghisellini , g. , avila - reese , v. , & ghirlanda , g. 2006 , mnras , 370 , 185 .
guidorzi , c. , frontera , f. , montanari , e. , rossi , f. , amati , l. , gomboc , a. , hurley , k. , & mundell , c. g. 2005 , mnras , 363 , 315 .
jimenez , r. , band , d. , piran , t. , 2001 , apj , 561 , 171 .
koshut , t. m. , paciesas , w. s. , kouveliotou , c. , van paradijs , j. , pendleton , g. n. , fishman , g. j. , & meegan , c. a. 1996 , apj , 463 , 570 .
norris , j. p. , bonnell , j. t. , nemiroff , r. j. , scargle , j. d. , kouveliotou , c. , paciesas , w. s. , meegan , c. a. , fishman , g. j. 1995 , apj , 439 , 542 .
reichart , d. e. , et al .
2007 , mnras , 379 , 619 .
rizzuto , d. et al .
2001 , apj , 552 , 57 .
rossi , f. et al .
2008 , preprint , astro - ph/0802.0471 schaefer , b.e .
2003 , apjlett , 583 , l71 .
schaefer , b.e .
2004 , apj , 602 , 306 .
schaefer , b.e .
2007 , apj , 660 , 16 .
schaefer , b.e . ,
collazzi , a.c .
, 2007 , 656 , l53 .
ccccc 970228 & 0.70 & 52.20 @xmath27 0.03 & @xmath63 & 2.37 @xmath27 0.44 + 970508 & 0.84 & 52.04 @xmath27 0.04 & @xmath64}_{-[40 ] } $ ] & 4.03 @xmath27 0.24 + 970828 & 0.96 & 52.68 @xmath27 0.05 & @xmath65}_{-[30 ] } $ ] & 10.50 @xmath27 0.45 + 971214 & 3.42 & 52.92 @xmath27 0.01 & @xmath66}_{-[20 ] } $ ] & 6.72 @xmath27 0.09 + 980703 & 0.97 & 51.78 @xmath27 0.01 & @xmath67}_{-[25 ] } $ ] & 18.00 @xmath27 1.80 + 990123 & 1.61 & 53.36 @xmath27 0.02 & @xmath68}_{-[60 ] } $ ] & 16.58 @xmath27 0.05 + 990506 & 1.31 & 53.05 @xmath27 0.01 & @xmath69}_{-[30 ] } $ ] & 12.67 @xmath27 0.63 + 990510 & 1.62 & 52.76 @xmath27 0.02 & @xmath70}_{-[10 ] } $ ] & 5.06 @xmath27 0.25 + 990705 & 0.84 & 52.36 @xmath27 0.02 & @xmath71 & 9.54 @xmath27 0.27 + 991208 & 0.71 & 52.67 @xmath27 0.04 & @xmath66}_{-[20 ] } $ ] & 4.80 @xmath27 0.24 + 991216 & 1.02 & 53.36 @xmath27 0.01 & @xmath72}_{-[30 ] } $ ] & 3.58 @xmath27 0.05 + 000131 & 4.5 & 53.20 @xmath27 0.05 & @xmath73 & 4.54 @xmath27 0.09 + 000210 & 0.85 & 52.85 @xmath27 0.04 & @xmath74 & 1.73 @xmath27 0.06 + 000911 & 1.06 & 53.05 @xmath27 0.04 & @xmath75}_{-[100 ] } $ ] & 6.46 @xmath27 0.32 + 000926 & 2.07 & 52.97 @xmath27 0.04 & @xmath76 & 4.67 @xmath27 0.45 + 010222 & 1.48 & 53.51 @xmath27 0.01 & @xmath77 & 6.46 @xmath27 0.14 + 010921 & 0.45 & 51.14 @xmath27 0.04 & @xmath78 & 5.74 @xmath27 0.58 + 020124 & 3.2 & 52.76 @xmath27 0.07 & @xmath79 & 12.14 @xmath27 0.58 + 020405 & 0.7 & 52.20 @xmath27 0.02 & @xmath80 & 10.18 @xmath27 0.38 + 020813 & 1.25 & 52.56 @xmath27 0.03 & @xmath81 & 17.36 @xmath27 0.23 + 021004 & 2.32 & 52.00 @xmath27 0.10 & @xmath82 & 6.89 @xmath27 0.41 + 021211 & 1.01 & 52.08 @xmath27 0.03 & @xmath83 & 0.66 @xmath27 0.12 + 030115 & 2.5 & 52.22 @xmath27 0.07 & @xmath84 & 4.26 @xmath27 0.30 + 030226 & 1.98 & 51.89 @xmath27 0.08 & @xmath85 & 8.86 @xmath27 0.87 + 030323 & 3.37 & 52.11 @xmath27 0.22 & @xmath86 & 6.89 @xmath27 0.70 + 030328 & 1.52 & 52.38 @xmath27 0.03 & @xmath87 & 20.83 @xmath27 0.70 + 030329 & 0.17 & 51.14 @xmath27 0.02 & @xmath88 & 4.43 @xmath27 0.23 + 030429 & 2.66 & 52.08 @xmath27 0.12 & @xmath89 & 2.13 @xmath27 0.35 + 030528 & 0.78 & 50.66 @xmath27 0.09 & @xmath90 & 9.99 @xmath27 0.93 + 040924 & 0.86 & 51.97 @xmath27 0.05 & @xmath91 & 0.49 @xmath27 0.02 + 041006 & 0.71 & 51.76 @xmath27 0.03 & @xmath92 & 4.26 @xmath27 0.12 + 050126 & 1.29 & 51.03 @xmath27 0.05 & @xmath93 & 6.59 @xmath27 0.32 + 050318 & 1.44 & 51.85 @xmath27 0.05 & @xmath94 & 2.88 @xmath27 0.20 + 050401 & 2.9 & 53.19 @xmath27 0.05 & @xmath95 & 4.61 @xmath27 0.23 + 050406 & 2.44 & 51.32 @xmath27 0.11 & @xmath96 & 1.92 @xmath27 0.18 + 050416 & 0.65 & 50.99 @xmath27 0.07 & @xmath97 & 0.58 @xmath27 0.07 + 050502 & 3.79 & 52.79 @xmath27 0.12 & @xmath98 & 4.60 @xmath27 0.28 + 050505 & 4.27 & 52.79 @xmath27 0.07 & @xmath99 & 9.02 @xmath27 0.41 + 050525 & 0.61 & 51.90 @xmath27 0.01 & @xmath100 & 2.24 @xmath27 0.11 + 050603 & 2.82 & 53.84 @xmath27 0.03 & @xmath101 & 1.47 @xmath27 0.12 + 050820 & 2.61 & 52.28 @xmath27 0.07 & @xmath102 & 6.46 @xmath27 0.41 + 050904 & 6.29 & 53.08 @xmath27 0.06 & @xmath103 & 59.46 @xmath27 2.97 + 050908 & 3.35 & 52.02 @xmath27 0.07 & @xmath104 & 4.86 @xmath27 0.14 + 050922 & 2.2 & 52.88 @xmath27 0.02 & @xmath105 & 1.15 @xmath27 0.05 + 051022 & 0.8 & 52.53 @xmath27 0.03 & @xmath106 & 10.30 @xmath27 0.23 + 051109 & 2.35 & 52.54 @xmath27 0.05 & @xmath107 & 3.78 @xmath27 0.35 + 060108 & 2.03 & 51.54 @xmath27 0.43 & @xmath108 & 3.20 @xmath27 0.14 + 060115 & 3.53 & 52.21 @xmath27 0.05 & @xmath109 & 15.42 @xmath27 0.54 + 060116 & 6.6 & 53.03 @xmath27 0.24 & @xmath110 & 21.76 @xmath27 1.15 + 060124 & 2.3 & 52.66 @xmath27 0.05 & @xmath111 & 5.12 @xmath27 0.18 + 060206 & 4.05 & 52.86 @xmath27 0.02 & @xmath112 & 1.86 @xmath27 0.09 + 060210 & 3.91 & 52.93 @xmath27 0.02 & @xmath113 & 23.62 @xmath27 1.00 + 060223 & 4.41 & 52.64 @xmath27 0.08 & @xmath114 & 2.82 @xmath27 0.18 + 060418 & 1.49 & 52.35 @xmath27 0.02 & @xmath115}_{-[20 ] } $ ] & 12.48 @xmath27 0.59 + 060502 & 1.51 & 51.75 @xmath27 0.19 & @xmath116 & 7.42 @xmath27 0.28 + 060510 & 4.9 & 52.42 @xmath27 0.07 & @xmath117}_{-[30 ] } $ ] & 70.21 @xmath27 1.63 + 060526 & 3.21 & 52.36 @xmath27 0.06 & @xmath118}_{-[5 ] } $ ] & 9.54 @xmath27 1.03 + 060604 & 2.68 & 51.75 @xmath27 0.08 & @xmath119}_{-[5 ] } $ ] & 9.28 @xmath27 0.63 + 060605 & 3.8 & 52.25 @xmath27 0.19 & @xmath120}_{-[30 ] } $ ] & 8.83 @xmath27 0.22 + 060607 & 3.08 & 52.36 @xmath27 0.13 & @xmath121 & 13.12 @xmath27 0.54 + [ zlandep ] ccc @xmath122 & 0.41 & 0.37 + @xmath7 & 0.40 & 0.36 + @xmath3 ( firmani relation ) & 0.41 & 0.38 + @xmath123 & 0.41 & 0.37 + @xmath8 & 0.41 & 0.37 + @xmath124 & 0.41 & 0.38 + @xmath125 & 0.46 & 0.39 + @xmath126 & 0.41 & 0.38 + @xmath127 & 0.37 & 0.33 + @xmath128 & 0.38 & 0.38 + @xmath129 & 0.41 & 0.36 + @xmath130 & 0.40 & 0.36 + @xmath131 & 0.56 & 0.49 + @xmath4 & 0.46 & 0.41 + @xmath5 & 0.47 & 0.43 + @xmath132 ( @xmath133 ) & 0.50 & 0.46 + @xmath134 & 0.33 & 0.29 + @xmath135 & 0.35 & 0.30 + @xmath136 & 0.34 & 0.30 + @xmath137 & 0.34 & 0.30 + @xmath138 & 0.35 & 0.31 + @xmath139 & 0.35 & 0.31 + @xmath140 & 0.37 & 0.30 + @xmath50 & 0.31 & 0.27 + @xmath141 & 0.32 & 0.28 + @xmath142 & 0.35 & 0.30 + @xmath143 & 0.38 & 0.34 + @xmath144 & 0.38 & 0.35 + @xmath145 & 0.41 & 0.33 + @xmath6 & 0.42 & 0.38 + @xmath146 & 0.41 & 0.37 + @xmath147 & 0.52 & 0.51 + [ rms ] | firmani et al . proposed a new gamma ray burst ( grb ) luminosity relation that showed a significant improvement over the @xmath0 relation .
( @xmath1 is the isotropic peak luminosity and @xmath2 is the photon energy of the spectral peak for the burst . )
the new proposed relation simply modifies the @xmath2 value by multiplying it by a power of @xmath3 , where @xmath3 is a particular measure of the grb duration .
we begin by reproducing the results of firmani for his 19 bursts .
we then test the firmani relation for the same 19 bursts except that we use independently measured values for @xmath1 , @xmath3 , and @xmath2 , and we find that the relation deteriorates substantially .
we further test the relation by using 60 grbs with measured spectroscopic redshifts , and find a relation that has a comparable scatter as the original @xmath0 relation .
that is , a much larger sample of bursts does not reproduce the small scatter as reported by firmani et al .
finally , we investigate whether the firmani relation is improved by the use of any of 32 measures of duration ( e.g. , @xmath4 , @xmath5 , @xmath6 , the fluence divided by the peak flux , @xmath7 , and @xmath8 ) in place of @xmath3 . the quality of each alternative duration measure is evaluated with the root mean square of the scatter between the observed and fitted logarithmic @xmath1 values .
although we find some durations yield slightly better results than @xmath3 , the differences between the duration measures are minimal .
we find that the addition of a duration does not add any significant improvement to the @xmath0 relation .
we also present a simple and direct derivation of the firmani relation from _ both _ the @xmath0 and amati relations . in all we conclude that the firmani relation neither has an independent existence nor does it provide any significant improvement on previously known relations that are simpler . |
since the fundamental suggestion made by anderson in 1987 , the two - dimensional ( 2d ) @xmath0 model ( @xmath1 m ) has been regarded as an effective hamiltonian description of the basic properties of the superconducting copper oxides.@xcite essential physical features of the insulating copper - oxide planes of these compounds are successfully reproduced by the spin-@xmath2 antiferromagnetic heisenberg model ( ahm),@xcite to which the @xmath1 m reduces when holes are absent . on the other hand , such insulating planes can turn superconducting if doped with mobile holes , whose presence is accounted for in the @xmath1 m by adding to the ahm a nearest - neighbor hole - hopping term .
hence , the @xmath0 hamiltonian is @xmath3 where the summations run over nearest - neighbor sites on the square lattice , @xmath4 is the hole hopping amplitude , and @xmath5 the antiferromagnetic coupling between site @xmath6 and @xmath7 ; @xmath8 creates an electron , at site @xmath6 , of spin projection ( e.g. , along the @xmath9 axis ) @xmath10 avoiding double occupancy .
therefore , in terms of the usual fermionic creation operator @xmath11 ; @xmath12 and @xmath13 , with @xmath14 , are the spin-@xmath2 and the number operator respectively . despite the simple form of the above hamiltonian , its study constitutes , even in presence of a single hole , a very difficult problem .
exact diagonalization ( ed ) , possible only for small system sizes ( i.e. , up to @xmath15 sites),@xcite approximated analytical treatments,@xcite as well as numerical approaches of various type,@xcite have been largely employed to gain insight into the physics of the one - hole @xmath1 m .
specifically , quantum monte carlo ( qmc ) techniques based on imaginary time projection furnish , even for large systems , accurate ( exact in principle ) results at half filling ( i.e. , no - hole case ) or for @xmath4=0 ( i.e. , static - hole case).@xcite however , when @xmath16 these methods are affected by the sign problem , for which only essentially uncontrolled workarounds ( e.g. , fixed - node approximation ) exist .
furthermore , being a 2d one , the model of our interest can not be tackled , due to the unfavorable scaling of the required computational resources , by using the density matrix renormalization group ( dmrg),@xcite the accepted method of choice for 1d lattice models .
variational monte carlo ( vmc ) schemes do not suffer from the sign problem ; on the other hand they are not strictly exact , their accuracy depending on the particular choice of the trial wave function ( wf ) . for the @xmath1 m with a single hole , specific
variational wf s @xcite have been designed affording estimates of the hole dispersion relation in reasonable agreement with those , generally more accurate , obtained by qmc .
recently , the suggestion was made of combining dmrg and monte carlo in order to design novel variational algorithms for tensor - network based ansatze applicable to systems of spatial dimension higher than one.@xcite the wf introduced by us,@xcite founded on entangled - plaquette states ( eps ) , has been successfully employed to study ground - state ( gs ) properties of lattice spin models with open or periodic boundary conditions ( pbc ) .
such ansatz has been proven to yield estimates of physical observables of accuracy at least comparable to that obtainable with the best alternative techniques or trial wfs.@xcite the eps wf can be efficiently optimized for plaquettes of a given size , as well as systematically improved by adding more plaquettes and/or increasing their size.@xcite in this paper we apply the eps ansatz to investigate the dynamics of a hole in a quantum antiferromagnet described in eq .
( [ eq : ham ] ) .
specifically , we compute the hole energy dispersion curve @xmath17 , as well as the spectral weight of the quasi - hole state , for lattice sizes up to @xmath18 sites in the parameter - range @xmath19 ( the @xmath1 m is believed to be relevant to the cuprate superconductors for a value of @xmath20 ) . from the methodological standpoint , this study is considerably more challenging than that of the various ( un)frustrated spin models to which the eps ansatz has been recently applied with success.@xcite the main goal of this work is to demonstrate the accuracy of the eps wf in describing the property of a mobile hole in the @xmath1 m .
this is achieved by a systematic comparison of our results for the energy dispersion curve and the hole spectral weight with estimates computed by different approaches .
furthermore , besides the mentioned methodological aspects we offer extremely accurate results for the hole spectral weight in the so called `` physical region '' ( i.e.,@xmath21 ) where the accuracy of qmc data , the most reliable ones obtained so far , becomes less satisfactory.@xcite assessing the reliability of the eps ansatz in a case ( i.e. , the @xmath1 m with one mobile hole ) for which a large number of results obtained with alternative trial states , or other methodologies are available in literature , is a fundamental step also in view of future applications of our wf to the long - standing problem of understanding the phase diagram of the @xmath1 m at finite hole concentration .
the accuracy of the estimates presented here is clearly not a proof of the reliability of our variational choice when more holes are present , it constitutes , however , the necessary condition to consider the eps wf , with no need of essential adjustments , a promising option for the investigation of the finite hole concentration scenario .
a direct study of the latter problem , given the paucity of reliable results obtained in the past , would render the assessment of the trustability of our ansatz problematic leaving , on the other hand , room for dubitative , definitely well - posed , questions such as : _ why results at finite hole concentration should be taken seriously if the reliability of the eps wf has not even been proved in the simpler single hole case ? _ on using a minimal eps ansatz based on @xmath22 @xmath23 entangled plaquettes , we obtain estimates of the hole dispersion curve in qualitative and semi - quantitative agreement with those computed by qmc .
we achieve quantitative agreement , also for other observables , with an eps wf based on plaquettes comprising @xmath24 lattice sites .
the accuracy of our findings for one hole confirms the flexibility of the eps ansatz , which can be applied with no substantial modifications to a large class of lattice models .
more importantly , the eps wf used in this work is easily generalizable to the case of a finite hole concentration , where the sign problem renders qmc inapplicable , and the generalization of different trial states employed for one or two holes appears quite complicated.@xcite
let us consider a square lattice of @xmath22 sites comprising one electron per site , a generic wf can be written as @xmath25 where @xmath26 , @xmath27 is the eigenvalue of @xmath28 and @xmath29 is the weight of a configuration of the system . the eps ansatz is constructed as follows : \i ) cover the systems with @xmath30 plaquettes in such a way that @xmath31 sites labelled by @xmath32 belong to the @xmath33 plaquette .
\ii ) express the weight of a global spin configuration as a product of variational coefficients @xmath34 in biunivocal correspondence to the particular configuration ( e.g. , spin state along the @xmath9 axis of the plaquette sites ) of the plaquettes .
hence , taking into account explicitly the bipartite nature of the spin-@xmath2 square heisenberg antiferromagnet at half filling : @xmath35 being @xmath36 , according to the marshall - sign rule , the number of `` down '' spins in one of the two sublattices@xcite and @xmath37 the crucial part of the ansatz .
analogously , in presence of a single hole , our eps wf is defined via @xmath38 here @xmath39 refers to the generic configuration containing one hole and @xmath40 electrons , @xmath41 and @xmath42 are the hole momentum and the spatial position of the empty site respectively .
it is worth mentioning that the trial states adopted in this work naturally incorporate , for plaquettes including more than one site , electron - electron as well as electron - hole correlations .
we estimate gs energies of the hamiltonian in eq .
( [ eq : ham ] ) via variational optimization of the states ( [ coeffh ] ) at half filling and ( [ coeffk ] ) in presence of a single hole .
independent optimizations are carried on for each @xmath41 value by monte carlo sampling both the energy and its derivatives with respect to the varational parameters ( i.e. , the @xmath43 s ) .
further details concerning the variational family of eps and the numerical technique adopted to minimize the energy are given in refs . and .
although the eps ansatz is exact when a single plaquette as large as the system is employed , accurate results are already obtainable by using @xmath22 entangled ( i.e. , overlapping ) plaquettes comprising @xmath44 sites , the accuracy of the estimates being sensibly improvable increasing the plaquette size .
specifically , we compute the hole energy dispersion curve @xmath45 defined by the difference between the @xmath46-dependent gs energy of the one - hole @xmath1 m and that of the ahm , and the hole spectral weight @xmath47 where @xmath48 ( @xmath49 ) is the normalized gs of the one ( zero ) -hole system . we compare the eps results with exact ones for a @xmath50 lattice and with the most accurate available in literature , for larger systems . in this case
the main features of the hole dispersion relation , namely the characteristic shape with a global minimum at @xmath51 and a nearly flat region around @xmath52 , are very well reproduced with an ansatz based on @xmath53 plaquettes , while the increasing of the plaquette size to @xmath54 gives rise to an approximately rigid shift , towards lower energies , of the band and an almost perfect agreement ( not reachable , to our knowledge , with other variational wf s ) with previous qmc studies for all the @xmath55 values considered in this work .
moreover , with the @xmath54 eps , accurate estimates of the hole spectral weight are obtained , on the same footing , via the monte carlo method .
the gs energies of the one - hole @xmath1 m on a @xmath50 lattice can be obtained , given the small system size , by using a single plaquette comprising all the lattice sites . in this case the eps ansatz is exact , the error bar of our estimates being only due to the limited number of configurations sampled , and reproduces , as expected , ed results.@xcite clearly , for larger systems , the use of a single plaquette is not feasible , therefore , it is important to asses the accuracy of our wf for smaller plaquette size .
the error relative to the exact result : @xmath56 of our estimate of the global energy minimum at @xmath57 is @xmath58% for the @xmath53 eps ansatz , decreasing down to less than @xmath59% when @xmath54 plaquettes are used . for this plaquette size ,
our results compare favorably to those obtained with different trial states , even when larger lattices are considered .
for example , the eps energy when @xmath51 on a @xmath60 lattice at @xmath61 is @xmath62 , while the wf including both spin - spin and momentum dependent spin - hole correlations adopted in ref .
yields a value of @xmath63 ( i.e. , @xmath64% higher ) .
sites with pbc at @xmath65 .
eps estimates ( triangles ) are computed using square plaquettes of 9 sites .
qmc results ( crosses ) are also shown for comparison.@xcite , title="fig : " ] sites with pbc at @xmath66 .
eps estimates ( triangles ) are computed using square plaquettes of 9 sites .
qmc results ( crosses ) are also shown for comparison.@xcite , title="fig : " ] sites with pbc at @xmath61 . eps estimates are computed using square plaquettes of 4 ( circles ) or 9 ( triangles ) sites .
gfmc results ( crosses)@xcite are also shown for comparison.,title="fig : " ] as opposed to the case in which more holes are present and the sign problem is particularly severe , qmc techniques based on imaginary time projection , for the one - hole @xmath1 m , provide extremely accurate results , to our knowledge the most accurate in literature , when , due to the large system size , the ed of the hamiltonian is not possible .
hence , a comparison of the eps wf against these approaches is in order . the hole energy dispersion curve , defined in eq .
( [ eq : disp ] ) , is shown in fig .
[ fig:1 ] for the @xmath67 lattice at @xmath65 .
our estimates ( triangles ) , computed with the @xmath54 eps ansatz , are indistinguishable ( taking into account the error bars ) from available qmc ones@xcite ( crosses ) .
around @xmath68 the curve is nearly flat , in agreement with other studies.@xcite it has to be mentioned that the degeneracy observed on the @xmath50 lattice between @xmath69 and @xmath70 is due to geometrical reasons@xcite and is no longer present when the lattice size increases .
the hole dispersion relation for lower values of @xmath55 is plotted in figs .
[ fig:2 ] and [ fig:3 ] .
the main qualitative features of the curve ( e.g. , the position of the minimum ) appear independent of the value of @xmath55 .
to illustrate further the accuracy of the eps wf , estimates of @xmath71 obtained with @xmath53 plaquettes ( circles in fig .
[ fig:3 ] ) are also shown for @xmath72 . despite the relatively small number of variational coefficients utilized for this ansatz , the resulting dispersion relation is in close agreement with that found with @xmath54 plaquettes as well as with green function monte carlo ( gfmc ) results@xcite ( triangles and crosses respectively in the same figure ) and qmc ones.@xcite specifically , the hole band predicted by using such a minimal eps ansatz differs from the more accurate prediction achievable with @xmath54 plaquettes only by a rigid energy shift towards lower energy values .
this energy shift ( albeit quantitatively different ) occurs for all the three values of @xmath55 considered here , pointing out how the eps ansatz , even in its simplest and economical implementation ( i.e. , when @xmath53 plaquettes are used ) , can provide reliable estimates of observable such as the effective mass , the band width , and generally all those obtainable from band energy differences .
.[tab:2]band width ( @xmath73 ) and energy difference ( @xmath74 ) between @xmath69 and @xmath75 , in units of @xmath5 , for a system of @xmath76 sites with pbc .
error bars are in parenthesis . [ cols="^,^,^",options="header " , ] values of the band width @xmath73 and the energy gap @xmath74 respectively defined as @xmath77 and @xmath78 are reported , as a function of @xmath55 , for a @xmath67 lattice in tab .
[ tab:2 ] .
our results of the band width are in excellent agreement with qmc@xcite and gfmc@xcite calculations . at @xmath65
our estimate differs from the vmc one : @xmath79 reported in ref . .
moreover we find that @xmath74 , at least in the cases examined here , varies linearly with @xmath4 according to the relation @xmath80 .
next we discuss our findings for the hole spectral weight @xmath81 ( see eq .
( [ eq : zeta ] ) ) . on a @xmath50 lattice
we obtain , for @xmath61 , @xmath82 , in agreement with the ed value@xcite of @xmath83 .
eps values of @xmath84 as a function of @xmath55 on a @xmath67 lattice are shown in tab .
[ tab:3 ] .
our estimates , computed with plaquettes of @xmath24 sites , are in agreement with qmc ones@xcite for all the values of @xmath55 ; agreement is found with self consistent born approximation ( scba ) calculations@xcite at @xmath61 , this approach providing , as well as the wf used in ref . ,
higher values of the spectral weight when @xmath55 increases .
the momentum dependence of the spectral weight for @xmath61 is shown in fig .
[ fig:4 ] along particular cuts of the brillouin zone .
@xmath85 is approximately @xmath86% at @xmath87 and , going in the @xmath88 direction , reaches @xmath89 at @xmath68 ; it then decreases , slightly , down to @xmath90 at @xmath91 , and more pronouncedly , in the @xmath92 direction , beyond this @xmath46 value .
lattice with pbc at @xmath61 .
the dashed line is a guide to the eye.,title="fig : " ]
the two dimensional @xmath0 model in presence of a single mobile hole has been investigated by means of variational monte carlo simulations based on the eps trial wave function . for systems comprising up to @xmath93 sites ( i.e. , the maximum size considered in this work ) , an eps ansatz consisting of @xmath22 entangled plaquettes of @xmath44 sites provides estimates of the dispersion curve in qualitative and semi - quantitative agreement ( up to a rigid energy shift ) with the most accurate alternative numerical approaches . by increasing the plaquette size to @xmath24 sites , with a minimal additional computational effort , quantitative agreement
is easily recovered and accurate results for the hole spectral weight obtained .
given the accuracy of the results presented in this work , as well as the simplicity of the variational algorithm used to minimize the energy , the eps ansatz appears a very promising option for future investigations of the @xmath0 model in the case of finite hole concentration , a problem of great physical interest to which exact quantum monte carlo techniques based on imaginary time projection are not applicable due to the sign problem .
the author acknowledges discussions with j. i. cirac . this work has been supported by the dfg ( for 635 ) and the eu project quevadis .
20 p. w. anderson , science * 235 * , 1196 ( 1987 ) .
f. c. zhang and t. m. rice , phys rev .
b * 37 * , 3759 ( 1988 ) .
e. manousakis , rev .
phys . * 63 * , 1 ( 1991 ) .
e. dagotto and j. r. schrieffer , phys .
b * 43 * , 8705 ( 1991 ) .
t. barnes , a. e. jacobs , m. d. kovarik and w. g. macready , phys .
b * 45 * , 256 ( 1992 ) .
d. poilblanc , h. j. schulz , and t. ziman , phys .
b * 47 * , 3268 ( 1993 ) . c. l. kane , p. a. lee and n. read , phys . rev .
b * 39 * , 6880 ( 1989 ) .
f. marsiglio , a. e. ruckenstein , s. schmitt - rink and c. m. varma , phys .
b * 43 * , 10882 ( 1991).t z. liu and e. manousakis , phys . rev .
b * 44 * , 2414 ( 1991 ) . g. martinez and p. horsch , phys .
b * 44 * , 317 ( 1991 ) . m. boninsegni and e. manousakis , phys
b * 43 * , 10353 ( 1991 ) .
m. boninsegni and e. manousakis , phys .
b * 45 * , 4877 ( 1992 ) .
m. boninsegni , phys .
a * 188 * , 330 ( 1994 ) .
m. brunner , f. f. assaad and a. muramatsu , phys .
b * 62 * , 15480 ( 2000 ) .
s. r. white , phys .
lett . * 69 * , 2863 ( 1992 ) .
a. w. sandvik and g. vidal , phys .
. lett . * 99 * , 220602 ( 2007 ) .
n. schuch , m. m. wolf , f. verstraete and j. i. cirac , phys .
* 100 * , 040501 ( 2008 ) ; a. sfondrini , j. cerrillo , n. schuch and j. i. cirac , phys .
b * 81 * , 214426 ( 2010 ) .
f. mezzacapo , n. schuch , m. boninsegni and j. i. cirac , new j. phys .
* 11 * , 083026 ( 2009 ) .
j. changlani , j. m. kinder , c. j. umrigar and g. k - l chan , phys . rev .
b * 80 * , 245116 ( 2009 ) .
f. mezzacapo and j. i. cirac , new j. phys .
* 12 * , 103039 ( 2010 ) .
e. neuscamman , h. j. changlani , j. m. kinder and g. k - l chan , arxiv:1008.4945v1 .
w. marshall , proc .
london ser .
a * 232 * , 48 ( 1955 ) . c. j. hamer , z. weihong and j. oitmaa , phys .
b * 58 * , 15508 ( 1998 ) . | we study the properties of a mobile hole in the @xmath0 model on the square lattice by means of variational monte carlo simulations based on the entangled - plaquette ansatz .
our energy estimates for small lattices reproduce available exact results .
we obtain values for the hole energy dispersion curve on large lattices in quantitative agreement with earlier findings based on the most reliable numerical techniques .
accurate estimates of the hole spectral weight are provided . |
systems of imaging atmospheric cherenkov telescopes , such as the hegra stereoscopic telescope system ( daum et al .
1997 , konopelko et al .
1999 ) , allow to reconstruct the directions of air showers over the full field of view , with a radius of about 1.5@xmath6 , and can therefore be used for sky surveys . here
, we report on two surveys , one along the galactic equator ( glp , fig . [ glp_scan ] ) with the focus on the search for point sources and a second one perpendicular to the galactic equator ( gls , fig .
[ gls_scan ] ) with the focus on diffuse emission .
the glp scan , covering one quarter of the galactic disk , from the galactic center to the cygnus region , includes 105 h of observation time and is described in detail in phlhofer et al . , 1999 .
here we give results on individual point sources .
+ the second scan ( gls ) covers a rectangular patch of the sky of roughly 80 deg@xmath7 , centered on the galactic plane at longitude @xmath8 and includes 42 h of observation .
the motivation for this survey was mainly the search for diffuse gamma - ray emission from the galactic plane .
+ diffuse emission results from the interactions of charged cosmic rays with interstellar gas confined to the plane or with photons .
diffuse emission in the energy range from tens of mev to tens of gev has been studied recently by egret ( hunter et al . 1997 ) .
the basic features can be modeled assuming @xmath9 decay as the dominant mechanism , with the gamma - ray emission proportional to the product of the gas column density and the cosmic - ray density ( see , e.g. , hunter et al .
1997 ) . above 1 gev
, data show an excess in gamma - ray flux over model predictions ( hunter et al .
1997 ; see however aharonian and atoyan 2000 ) . at these energies ,
contributions from inverse compton scattering of electrons start to become relevant ( see , e.g. , porter & protheroe 1997 ) . in response ,
revised models speculate that the local measurements of the electron flux may not be representative for the entire galaxy ( porter & protheroe 1997 ; pohl & esposito 1998 ) . in case
that the solar system is in an `` electron void '' , diffuse gamma - ray emission at high energies could be an order of magnitude above predictions based on local electron spectra ( pohl & esposito 1998 ) .
another diffuse gamma - ray flux component arises from the hard energy spectrum of those galactic crs that are still confined in the ensemble of their unresolved sources like snrs ( berezhko & vlk , 2000 )
. + [ cols="<,>,>,<,>,>",options="header " , ] [ pointsearch44_ul ] @xmath10 & [ cu ] + g29.7 - 0.3 & -0.7 & 0.28 & g45.7 - 0.4@xmath11 & -1.7 & 0.15 + g30.7 + 1.0@xmath11&-0.4 & 0.46 & g46.8 - 0.3@xmath7 & 1.0 & 0.35 + g31.5 - 0.6@xmath7&-0.9 & 0.19 & g49.2 - 0.7@xmath11 & -0.5 & 0.21 + g31.9 + 0.0@xmath12&0.1 & 0.1 & g54.1 + 0.3 & -0.3 & 0.17 + g32.1 - 0.9@xmath13&0.8 & 0.6 & g54.4 - 0.3@xmath13 & -0.6 & 0.27 + g32.8 - 0.1@xmath7&1.1 & 0.45 & g55.0 + 0.3@xmath7 & -0.1 & 0.21 + g33.2 - 0.6@xmath7&1.4 & 0.25 & g57.2 + 0.8@xmath12 & 0.3 & 0.11 + g33.6 + 0.1@xmath12&0.4 & 0.15 & g59.5 + 0.1@xmath12 & 1.4 & 0.26 + g34.7 - 0.4@xmath13&-0.7 & 0.3 & g59.8 + 1.2@xmath7 & 0.6 & 0.4 + g36.6 - 0.7@xmath11&1.1 & 0.45 & g63.7 + 1.1@xmath12 & 0 & 0.35 + g39.2 - 0.3@xmath14&1.04 & 0.08 & g69.7 + 1.0@xmath7 & 0.2 & 0.17 + g40.5 - 0.5@xmath15&-0.9 & 0.06 & g73.9 + 0.9@xmath11 & 0 & 0.18 + g41.1 - 0.3@xmath16&1.0 & 0.1 & g74.9 + 1.2@xmath12 & 0.8 & 0.15 + g42.8 + 0.6@xmath15&1.1 & 0.26 & g76.9 + 1.0@xmath12 & 0.4 & 0.13 + g43.3 - 0.2 & -0.7 & 0.14 & & & + + j1856 + 0115@xmath11 & -2.7 & 0.11 & j2020 + 3658@xmath13 & 0.4 & 0.27 + j1907 + 0557@xmath17 & -0.3 & 0.78 & j2020 + 4023@xmath7 & 0 & 0.15 + j1957 + 2859@xmath17 & -2.3 & 0.2 & & & + the galactic plane is also a region rich in potential gamma - ray point sources like snrs as well as pulsar driven nebulae and unidentified egret sources .
theoretical models predict that typical gamma - ray fluxes from the majority of these objects are below the detection threshold of the current generation of instruments ( see , e.g. , aharonian et al . 1997 ; drury et al .
however , both the lack of knowledge of the individual source parameters as well as approximations used in the modeling result in large uncertainties in the predictions for individual objects , by an order of magnitude or more .
point sources were searched in both scans on a grid where the bin position is taken to be the source position .
a grid spacing below the angular resolution of @xmath18 is chosen of @xmath19 in case of the gls - scan and of @xmath20 in case of the glp - scan .
the radius of the source bin for point sources is chosen according the angular resolution of the telescope configuration to optimize @xmath21 . in case of
moderately extendet objects the radius of the source bin was adapted such , that the complete source fits in the search bin . to estimate backgrounds ,
three regions of the same size as the source region were used , rotated by 90@xmath6 , 180@xmath6 and 270@xmath6 around the telescope axis , relative to the source .
significances for a detection are then calculated according to li & ma ( 1983 ) . since no high significant source bin is found in both scans upper limits are calculated using the procedure of o. helene ( 1983 ) . to convert from counts to fluxes in the case of the glp - scan crab data
are used for calibration , in case of the gls - scan the expected number of counts are derivied from mc simulations .
the simulations show that the result on the differential flux upper limit depends only weak on the spectral index at 1 tev ( see aharonian et al . , 1995 )
[ glp_scan ] shows a map of resulting upper limits for point sources for the glp - scan .
analysis is carried out for sources with @xmath22 corresponding to zenith angels @xmath23 and results on individual sources are given in tab .
[ point_sources ] .
results on smaller galactic longitude are preliminary and no individual upper limits are given .
for the gls - scan results are shown in fig .
[ gls_scan ] and in tab .
[ point_sources ] .
to 50@xmath6 , together with a fit by a sum of two gaussians .
data points taken from hunter et al .
in addition a gaussian profiles with a @xmath10 of @xmath24 , is shown.,width=313 ] ) and for the off - region ( @xmath25 ) .
the dashed lines indicate the expected gamma - ray region.,width=313 ] as background data and assuming the spatial distribution measured by the ergret instrument ( @xmath26 ) ( 3 ) . for detailed explanations
also shown is the egret flux for @xmath27 ( reynolds et al .
1993 , lebohec 2000 ) and the tibet upper limit ( amenomori , 1997 ) .
the dotted line indicates an extrapolation of the egret data with an index of 2.5 .
the dashed line indicates the scaled leaky box model prediction by berezhko & vlk , 2000 ( bv2000).,width=313 ] the search for diffuse gamma - ray emission from the galactic plane is complicated by the extended structure of the emission region .
the structure may be extended in latitude beyond the field of view .
[ fig_difflat ] illustrates the profile in galactic latitude as measured by egret .
+ to achive low threshold , observations were carried out under small zenith angles ( @xmath28 ) in june , july and august 1999 .
emphasis was on the control of systematic effects like changing weather conditions .
+ the most robust and model - independent but also least sensitive technique to derive limits on the diffuse flux simply selects 5-telescope - events according to their shapes as gamma - ray candidates .
a cut on @xmath29 less than 1.0 keeps about 1/2 of the gamma - rays , but rejects cosmic rays very efficiently .
assigning all 428 events after cuts as diffuse gamma - rays a 99% upper limit on the diffuse gamma - ray flux at 1 tev of 23.4 @xmath30 ph @xmath4 s@xmath5 sr@xmath5 mev@xmath5 results , for @xmath31 , assuming a flux from the crab nebula of @xmath32 ph @xmath4 s@xmath5 mev@xmath5 at 1 tev ( aharonian et al . , 2000 ) .
+ the limit obtained by this technique can be improved by subtracting , on a statistical basis the background , using an experimental background region , sufficiently far away from the galactic plane .
data sets can be normalized to each other using the rates of events with large @xmath29 ( @xmath33 ) , well outside the gamma - ray region .
unfortunately , the availability of suitable background data samples with the same telescope configuration is limited to 4.1 h of data , limiting the sensitivity . after a @xmath34 cut
1928 events survived compared to 141 events in the reference sample .
after subtraction of isotropic components , we find a 99% limit on the diffuse flux in the region @xmath31 of 10.4 @xmath30 ph @xmath4 s@xmath5 sr@xmath5 mev@xmath5 at 1 tev .
+ the final , and most sensitive analysis makes the assumption that diffuse gamma - ray emission from the galactic plane is limited to the central parts of the scan region , and uses the outer parts of the scan region to estimate backgrounds .
the signal region was considered @xmath35 . to ensure optimum quality of the events , only four- and five - telescope events were used , and the field of view
was restricted to 1.5@xmath6 from the optical axis .
a cut at 1.1 on the @xmath29 was applied to reject cosmic - ray background . to account for a possible zenith - angle dependence of background rates
, data were grouped into four different ranges in zenith angle , 20@xmath6 - 24@xmath6 , 24@xmath6 - 28@xmath6 , 28@xmath6 - 32@xmath6 and 32@xmath6 - 36@xmath6 . for each range in zenith angle and each scan band ,
the expected number of events in the signal region was calculated based on the number of events observed in the corresponding areas of the camera for the background region .
the expected and observed numbers of events were then added up for all zenith angles and scan bands . with a total number of 2387 gamma - ray events in the signal region , compared to 2353 expected events , there is no significant excess ( see fig .
[ fig - diff - mscw ] ) .
+ in order to translate the limit in the number of events into a flux limit , one now has to make assumptions concerning the distribution in galactic latitude of the diffuse radiation , since a spill - over of diffuse gamma - rays into the background region @xmath25 will effectively reduce the signal . for a profile with a width less or comparable to the egret profile a correction of 12%
is applied and one finds a limit @xmath36 ph @xmath4
s@xmath5 sr@xmath5 mev@xmath5 for the diffuse gamma - ray flux at 1 tev , averaged over the @xmath35 region .
the limit refers to an assumed spectral index of -2.6 , and changes by + 13% for an index of -2 , and by -5% for an index of -3 . for wider distributions of @xmath37 and @xmath38 rms ,
the limit changes to @xmath39 ph @xmath4 s@xmath5 sr@xmath5 mev@xmath5 and @xmath40 ph @xmath4 s@xmath5 sr@xmath5
mev@xmath5 , respectively . results are summarized in fig .
[ fig - uplimits ] .
no strong tev - gamma - ray emitter is detected in roughly one quater of the galactic disc coverd by both scans .
limits derived for known sources are on the level or lower than 20% of the crab flux . +
a search for diffuse gamma - ray emission resulted in an upper limit of @xmath36 ph @xmath4 s@xmath5 sr@xmath5 mev@xmath5 at 1 tev averaged over the region @xmath26 and assuming the spatial emission profile measured by the egret instrument .
since the analysis used to derive this limit is only sensitive to the variation of the diffuse flux with @xmath41 , rather than its absolute value , a distribution significantly wider than at egret energies will increase the limit .
other variants of the data analysis are sensitive to the absolute flux , but give less stringent limits of @xmath42 and @xmath43 ph @xmath4 s@xmath5 sr@xmath5 mev@xmath5 .
the limit on the tev gamma - ray flux can be used to derive a lower limit on the spectral index of the diffuse radiation of 2.5 , and to exclude models which predict a strong enhancement of the diffuse flux .
aharonian et al . , 1995 , j. phys .
g 21 , 419 aharonian , f. et al .
, 1997 , mnras 291 , 162 aharonian et al . ,
2000 , apj , 539 , 317 aharonian , f. a. & atoyan , a. m. 2000 , a&a , 362 , 937 amenomori , m. et al . , 1997 ,
25th icrc , durban , 3 , 117 berezhko , e.g , vlk , h.j . , 2000 ,
apj 540 , 923 borione , a. , et al . , 1997 , apj 493 , 175 daum , a. , et al . , 1997 , astropart . phys . 8 , 1 drury , l.o .
et al.,1994 , a&a , 287 , 959 rowell , g. , 2001 , this proceedings green , d.a.,mullard radio astronomy observatory , 1998 helene , o. , 1983 , nim 212 , 319 hunter et al .
, 1997 , apj 481 , 205 konopelko , a. , et al . , 1999 ,
10 , 275 lamb r.c . ,
macomb d.j . , 1997 ,
apj , 488 , 872 lebohec , s. , et al . , 2000 ,
apj 539 , 209 li , t. , ma , y. , 1983 , apj 272 , 317 pohl , m. , esposito , j.a . , 1998 ,
apj 507 , 327 porter , t.a . ,
protheroe , r.j . , 1997 , j. phys .
g 23 , 1765 phlhofer , g , et al .
1999 , proc .
26th icrc , 77 reynolds , p.t .
et al . , 1993 ,
apj 404 , 206 tayler , j.h .
et al . , 1993 ,
apjss 88 , 529 | the hegra - iact - system with a fov of @xmath0 deg radius has been used for surveying one quater of the galactic disc in respect to point sources , moderately extended sources and for diffuse emission in the energy range above 1 tev . in total 140 h of good observation time
were accumulated .
no new source has been discoverd .
limits on the level of 20% or lower of the crab flux on about 87 potential sources like snr , pulsars and egret sources are derived .
a limit on the diffuse emission is given on the level of @xmath1(e= 1 tev ) = 6.1 @xmath2 10@xmath3 ph @xmath4 s@xmath5 sr@xmath5 mev@xmath5 resulting in a lower limit of 2.5 on the spectral index for the extrapolation of the meassured egret flux for the diffuse emission . |
a @xmath5-uniform hypergraph is a pair @xmath15 of vertices @xmath16 and ( hyper)edges @xmath17 , where @xmath18 .
let @xmath12 denote the random @xmath5-uniform hypergraph with vertex set @xmath19:=\{1 , 2 , \ldots , n\}$ ] , where each of the @xmath20 potential edges is present with probability @xmath7 , independently of all other potential edges .
the similar model @xmath21 is a random @xmath5-uniform hypergraph on @xmath19 $ ] where @xmath14 edges are chosen uniformly at random among all sets of @xmath14 potential edges . for @xmath22 ,
these models are the typical random graph models , @xmath23 and @xmath24 as customary , we say that for a given @xmath25 ( @xmath14 resp . )
some graph property holds _ with high probability _ , denoted _
_ , if the probability that @xmath12 ( @xmath26 resp . )
has this property tends to 1 as @xmath27 .
existence problems of hamilton cycles have a rich history in random graphs .
first posed by erds and rnyi @xcite , even a correct threshold of hamiltonicity resisted the efforts of researchers until a breakthrough by psa @xcite and korshunov @xcite , who found that w.h.p .
@xmath28 has a hamilton cycle if @xmath29 is sufficiently large .
later korshunov @xcite , komls and szemerdi @xcite , and bollobs @xcite established the sharp theshold for hamiltonicity as well as the probability of hamiltonicity within the narrow window enclosing the critical @xmath7 ( resp .
@xmath14 ) in @xmath23 ( resp .
@xmath30 ) ; in particular , if @xmath31 , then the probability that @xmath23 is hamiltonian tends to @xmath11 .
the probability that each vertex in @xmath23 has degree at least 2 , a trivial necessary condition for hamiltonicity , also tends to this limit . as a culmination point
, bollobs @xcite showed w.h.p .
a hamilton cycle is born at the first moment that the random graph process , where edges are added uniformly at random one after another , has minimum degree at least @xmath32 . as for hypergraphs ,
the first problem that arises is how to even define a cycle .
there has been many fine results on the so - called @xmath33-overlapping cycles .
see khn and osthus @xcite for a great survey of results , and in particular , see dudek and frieze @xcite , where they establish the threshold values for the appearance of many of these cycles .
however , we use a different twist on the classical notion of a hypergraph cycle , defined by berge in @xcite . for @xmath34 , a _ ( berge ) cycle , @xmath35 of length @xmath33 _
is an alternating sequence of vertices and edges such that @xmath36 are distinct , @xmath4 are distinct , and @xmath37 a _ weak _ cycle is defined similarly except we do not require the edges to be distinct ; see figure [ fig : weak h cycle ] for an example . for the graph case @xmath38
, these edges must be distinct , and thus weak cycles are cycles .
in proper hypergraphs ( @xmath8 ) , one crucial difference between cycles and weak cycles is that a vertices with degree 1 can be in weak cycles , but necessarily these vertices can not be in cycles .
( v1 ) at ( 2,2 ) ; ( v2 ) at ( 4 , 2 ) ; ( v3 ) at ( 6 , 0 ) ; ( v4 ) at ( 4 , -2 ) ; ( v5 ) at ( 2 , -2 ) ; ( v6 ) at ( 0 , -1 ) ; ( v7 ) at ( 0 , 1 ) ; ( v8 ) at ( 6 , 2.5 ) ; ( v9 ) at ( -1,-1 ) ; ( v10 ) at ( -2,-.5 ) ; ( v11 ) at ( -.5 , -1.9 ) ; ( 2,2 ) ( 4,2 ) ( 6 , 0 )
( 4 , -2 ) ( 2 , -2 ) ( 0 , -1 )
( 0 , 1 )
( 2 , 2 ) ; plot [ smooth cycle ] coordinates ( @xmath39 ) ( @xmath40 ) ( @xmath41 ) ( @xmath42 ) ( @xmath43 ) ( @xmath44 ) ( @xmath45 ) ( @xmath46 ) ; plot [ smooth cycle , tension=.2 ] coordinates ( @xmath47 ) ( @xmath48 ) ( @xmath49 ) ( @xmath50 ) ; plot [ smooth cycle , tension=.5 ] coordinates ( @xmath51 ) ( @xmath52 ) ( @xmath53 ) ( @xmath54 ) ( @xmath55 ) ; plot [ smooth cycle , tension=.4 ] coordinates ( @xmath56 ) ( @xmath57 ) ( @xmath58 ) ( @xmath59 ) ; ( v1 ) circle ( 0.1 ) ; ( v2 ) circle ( 0.1 ) ; ( v3 ) circle ( 0.1 ) ; ( v4 ) circle ( 0.1 ) ; ( v5 ) circle ( 0.1 ) ; ( v6 ) circle ( 0.1 ) ; ( v7 ) circle ( 0.1 ) ; ( v8 ) circle ( 0.1 ) ; ( v9 ) circle ( 0.1 ) ; ( v10 ) circle ( 0.1 ) ; ( v11 ) circle ( 0.1 ) ; we ll say that the weak cycle @xmath60 _ spans _ the vertices @xmath61 ; note that the edges making up @xmath60 possibly contain more than just the spanned vertices ( again , see figure [ fig : weak h cycle ] ) .
a _ weak hamilton cycle _ is a cycle spanning the entire vertex set , and a hypergraph is _ weak hamiltonian _ if it contains a weak hamilton cycle .
trivially , each vertex in a weak hamiltonian hypergraph must have degree at least 1 .
however , we found that the main barrier to weak hamiltonicity in @xmath12 is this local " obstruction about whether or not this hypergraph has this trivial necessary condition .
[ theorem : 10.6.1 ] * ( i ) * for @xmath8 and @xmath62 , where @xmath63 , @xmath64 * ( ii ) * if @xmath65 then @xmath66 consequently , for any @xmath67 , w.h.p @xmath12 and @xmath26 are not weak hamiltonian and for any @xmath68 , w.h.p . @xmath12 and @xmath26 are weak hamiltonian .
first , since weak hamiltonicity is an increasing graph property , part * ( ii ) * can be easily shown to follow from * ( i ) * using a standard random graphs argument .
as such , we omit the proof of * ( ii)*. this theorem appears to be in contrast with the hamiltonicity results of @xmath23 and @xmath30 , where the main barrier is the presence of vertices of degree less than 2 .
although this seems like a behavioral difference between the graph and the hypergraph models , notice that a vertex in a hypergraph ( @xmath8 ) with degree at least 1 already has at least 2 neighbors ; so in both random graphs and hypergraphs , the main barrier for weak hamiltonicity is the presence of vertices with less than 2 neighbors .
having found the threshold of weak hamiltonicity of @xmath26 and @xmath12 , we naturally wondered whether the @xmath5-uniform hypergraph process , where edges are added to @xmath69 $ ] uniformly at random one after another , satisfies a similar result .
that is , we believe that w.h.p .
the moment this process loses its last isolated vertex , denoted @xmath70 , is also the moment that the hypergraph becomes weak hamiltonian , denoted @xmath71 .
trivially , @xmath72 , and by our theorem , w.h.p .
we feel that our method here could be significantly refined to show that @xmath74 , but by its nature will not be able to prove our suspected result that w.h.p .
if true , this conjecture would generalize to hypergraphs the weak hamilton version of bollobs @xcite hitting time result .
we actually prove a stronger result than theorem [ theorem : 10.6.1 ] , which says that for any @xmath7 in the critical range , @xmath12 has as large a weak cycle as can be expected .
let @xmath76 and @xmath77 .
@xmath12 has a weak cycle spanning all non - isolated vertices .
first , this theorem establishes the first equality in statement * ( i ) * of theorem [ theorem : 10.6.1 ] . to get a handle on the probability that there are no isolated vertices , we have the following lemma . [
lemma : converge in distribution ] suppose @xmath63 .
let @xmath78 denote the number of isolated vertices in @xmath79 .
then @xmath78 converges in distribution to a poisson random variable with mean @xmath80 .
in particular , @xmath81 one can prove this lemma using standard techniques by computing the factorial moments of @xmath78 and finding that @xmath82 \sim ( e^{-c})^r$ ] for any fixed @xmath83 as such , the proof is omitted . to find this sharp threshold ,
we will follow a hypergraph version of bollobs proof ( see @xcite ) that w.h.p .
@xmath84 is hamiltonian with a few key differences .
we outline the proof of the theorem here , hoping that it will help following the later arguments . 1 . in section [ sec :
de la vega ] : via an analogue of de la vega s theorem , it will be shown that there is likely a path of length @xmath85 for @xmath7 sufficiently close to but less than the sharp threshold of connectivity .
2 . in section [ sec : posa ] : we obtain a necessary condition for an extremal path starting at a fixed vertex , which actually says that the set of endpoints of all paths obtained via rotations is non - expanding . " 3 . in section [ sec : nonexpanding ] : then , we show that all non - expanding sets are sufficiently large , which means that the set of endpoints of rotations of a longest path must be large as well
. this will imply that for any such graph , there are sufficiently many non - present edges ( on the order @xmath86 ) such that the addition of any one of these absent edges will connect two endpoints of a longest path creating a cycle .
as long as this cycle is not isolated , we can break apart the cycle to create a longer path than before .
4 . in section [ sec : increment ] : by increasing @xmath7 in small increments , we will be assured of likely adding one of these beneficial edges from step ( 3 ) .
we keep increasing our edge probability until we finally end with a weak cycle on the set of non - isolated vertices .
the first , and most obvious , difference from bollobs proof , is that certain enumerative issues that are relatively simple for graphs become much more difficult when switching to hypergraphs .
for instance , in the proof that all non - expanding sets must be large , we encounter the probability that each vertex of a set of @xmath87 vertices , denoted @xmath88 , is adjacent to at least one of some @xmath89 vertices , denoted @xmath90 , using only edges contained within @xmath91 . for @xmath23 , this probability is precisely @xmath92 , since for different vertices of @xmath88 , the events that these vertices are adjacent to @xmath90 are independent .
however , for the hypergraph case , these events are definitely not independent , and we introduce a greedy edge finding algorithm , which we analyze to establish a useful bound on this probability ( that we feel is interesting in its own right ) .
second , we are also concerned with @xmath7 in the critical range , where the limiting probability that @xmath12 is weak hamiltonian is strictly between 0 and 1 . in this range , there are possibly isolated vertices , and we need to be more careful about the lack of small " non - expanding sets .
further , when we increase our edge probability @xmath7 in small increments , we will end up being forced to do it in 2 large steps rather than just 1 .
just prior to the typical window where the random hypergraph becomes connected , we will show that there is a path of length @xmath85 .
an analogous result , established by fernandez de la vega @xcite , was used to determine the sharp threshold of hamiltonicity in @xmath93 . here , we prove a hypergraphic version of de la vega s theorem , which in fact is established by the graph version after a key reduction argument . the following statement of de la vega s theorem is in bollobs @xcite . [ de la vega ]
let @xmath94 , then w.h.p . there is a path in @xmath95 of length at least @xmath96 .
we will be using the following immediate corollary of de la vega s theorem .
[ cor de la vega ] let @xmath97 , then with high probability , there is a path in @xmath98 of length at least @xmath99 in the following hypergraph analogue of de la vega s theorem , we do not try to obtain the best possible bounds on a likely long path , instead we care only to find sufficient bounds that we will later use to prove our ultimate result .
[ hyper de la vega ] suppose @xmath100 as @xmath27 so that @xmath101 .
let @xmath102 $ ] and @xmath103 w.h.p . in @xmath104
, there is a path of length at least @xmath105 .
our essential argument is that we decompose @xmath12 into 3 ( random ) hypergraphs @xmath106 , @xmath107 , and @xmath108 .
first , on @xmath106 , we will prove there is a path of length approximately @xmath109 along only the first @xmath109 vertices ; second , we prove that in @xmath107 , there is a path along only the last @xmath109 vertices ; finally , using the edges of @xmath108 , we concatenate these long paths together .
let @xmath110 and let @xmath111 , for @xmath112 , be random hypergraphs on @xmath19 $ ] , which are _ independent _ and distributed as @xmath113 .
note that @xmath114 is distributed as @xmath115 , where the probability that a generic edge in @xmath116 is present , @xmath117 , is equal to the probability that at least one of @xmath118 have this edge ; in particular @xmath119 .
moreover , @xmath120 and so @xmath116 can be naturally coupled with @xmath12 such that @xmath116 is a subgraph of @xmath12 .
therefore , it suffices to show that w.h.p .
@xmath116 has a path of the desired length .
let s begin with finding a long path in @xmath106 along only the first @xmath109 vertices .
we construct a random graph , @xmath121 , with vertex set @xmath122 $ ] by the following : for each possible edge @xmath123 , i \neq j$ ] , we define @xmath124 if and only if there is some hyperedge in @xmath125 that contains @xmath126 and @xmath127 but no other vertex of @xmath128 $ ] , i.e. @xmath129 each potential edge in @xmath121 considers distinct potential hyperedges in @xmath125 , and thus the potential edges in @xmath121 are present independently of one another .
further , the potential edge @xmath130 is not in @xmath121 iff all @xmath131 potential hyperedges of the form @xmath132 are not present .
hence @xmath133 consequently , @xmath134 is equal in distribution to @xmath135 and @xmath136 for large enough @xmath137 , we have that @xmath138 in particular , we meet the conditions to apply the corollary to de la vega s theorem ( corollary [ cor de la vega ] ) to @xmath121 ; w.h.p .
there is a path in @xmath121 of length at least @xmath139 which corresponds to a path in @xmath106 of same length spanning only vertices in @xmath140 ( note that the hyperedges that make up this path will include larger index vertices ) .
now for finding long paths in @xmath107 , we construct another random graph , @xmath141 , on @xmath142 $ ] by the following : for each possible edge @xmath143 $ ] , we define @xmath144 if and only if there is some hyperedge in @xmath145 of the form @xmath146 where @xmath147 $ ] .
by the same argument as before , w.h.p . there is a path in @xmath141 of length at least @xmath139 which corresponds to a path in @xmath107 of the same length spanning only vertices in @xmath148 . using @xmath149 , we will concatenate these paths together .
let @xmath150 denote the event that these long paths exists in @xmath125 and @xmath145 .
on @xmath150 , let @xmath90 be the last @xmath151 vertices of one such long path in @xmath125 and @xmath88 be the first @xmath151 vertices of one such long path in @xmath145 ( both of these are defined because the length of these paths is @xmath152 ) . by the independence of potential hyperedges of @xmath108 from @xmath106 and @xmath153
@xmath154 which tends to zero .
hence , with high probability these two long paths on distinct vertices exist and there is a hyperedge that contains one of the last @xmath155 vertices , say @xmath89 , of the first path and one of the first @xmath155 vertices , say @xmath87 , of the second path .
we construct our path in @xmath116 by following the long path in @xmath125 until we reach the vertex @xmath89 , then choosing this connecting hyperedge containing @xmath89 and @xmath87 , then following the path in @xmath145 until it ends .
this path has length at least @xmath156 as desired .
now that we have established that paths of length @xmath85 are likely to exist near the connectedness threshold of @xmath12 , our next goal is to show that the sets of endpoints of longest paths are relatively large ( on the order of @xmath137 ) . in the proof of finding the hamiltonicity threshold of @xmath23 ( see @xcite ) , psa s lemma is indispensable in describing the set of endpoints of these longest paths ; in particular , the set of endpoints is non - expanding " .
it plays a similar role here as well .
we begin with a couple of definitions . for a hypergraph , @xmath116 , on @xmath19 $ ] , and a set of vertices @xmath16 , we define @xmath157 as the set of neighbors of @xmath16 . formally ,
@xmath158 \setminus v : \exists e \in e(h ) , \exists v \in v \text { such that } w , v \in e\}\ ] ] we say that a set of vertices , @xmath90 , is _ non - expanding _ if @xmath159 the hypergraph version of psa s lemma is effectively the same as the graph version .
let @xmath116 be a hypergraph and let @xmath160 be a longest path starting from @xmath161 .
note that any neighbor of an endpoint of a longest path must necessarily be in said path or else we could extend this supposedly longest path and reach a contradiction .
now suppose there is some present hyperedge @xmath162 containing both @xmath163 and @xmath164 for some @xmath165 .
we say that the path @xmath166 is a _ rotation _ of @xmath167 by @xmath168 ; see figure [ fig : informativefigure ] for an illustration of these paths . at ( -5,.5 ) before rotation ; ( 0,0 )
( 10,0 ) ; ( 10,-0.4 ) to[out=270 , in=270 ] ( 4,-0.4 ) ; at ( 0,-.2 ) @xmath161 ; at ( 1,1.2 ) @xmath169 ; at ( 5,1.2 ) @xmath170 ; at ( 4,-.2 ) @xmath164 ; at ( 10,-.2 ) @xmath171 ; at ( -1,.5 ) @xmath172 ; at ( -5 , -3.5 ) after rotation ; ( 0,-4 ) ( 1 , -3 ) ( 2 , -4 ) ( 3 , -3 ) ( 4 , -4 ) ; ( 10,-4.4 ) to [ out=270 , in=270 ] ( 4 , -4.4 ) ; ( 5,-3 ) ( 6,-4 ) ( 7,-3 ) ( 8,-4 ) ( 9,-3 ) ( 10,-4 ) ; at ( 0,-4.2 ) @xmath161 ; at ( 1,-2.8 ) @xmath169 ; at ( 4,-4.2 ) @xmath164 ; at ( 5,-2.8 ) @xmath170 ; at ( 10,-4.2 ) @xmath171 ; at ( -1 , -3.5 ) @xmath173 ; at ( 7,-5.7 ) @xmath162 ; note that the vertex set of the path is unchanged after rotation . as opposed to graphs
, each hyperedge may give rise to more than one such rotation for a given path ; in fact , @xmath162 could have already been present within the path @xmath167 ! the hypergraph version of psa s lemma requires that we allow the possibility of repeated hyperedges in the paths which in turn gives rise to our definition of a _ weak _ cycle allowing duplicate hyperedges as well .
in fact , this is the precise moment in which we turned to _ weak _ cycles . to proceed , let @xmath174 be the set of paths obtained by the path @xmath167 , starting at @xmath161 via any number of rotations and let @xmath175 be the set of all endpoints of @xmath176 excluding @xmath161 .
this set @xmath177 is referred to as a _
psa set_. by allowing repeated edges as we do , the proof of psa s lemma for graphs can be naturally extended to hypergraphs .
in fact , the proof of psa s lemma in bollobs @xcite can be followed word for word here .
consequently , we omit the proof . in order to show that the set of endpoints formed by rotations of a longest path must likely be large
, we will show that any non - trivial " non - expanding set is large . since any set of isolated vertices trivially form a non - expanding set , we will focus on non - expanding sets of non - isolated vertices . for a hypergraph @xmath116 ,
let @xmath179 be the size of the smallest non - expanding set of non - isolated vertices .
formally , if @xmath180 denotes the set of non - isolated vertices of @xmath116 , then @xmath181 as a consequence of psa s lemma , for any longest path , @xmath167 , starting at non - isolated @xmath161 , we must have that @xmath182 .
an important fact that we ll later use is that if @xmath116 is connected , then the addition of edges to @xmath116 can only increase this measure @xmath183 .
our eventual goal is to show that in @xmath6 for @xmath7 near the connectivity threshold , w.h.p .
@xmath183 is on the order of @xmath137 .
a very useful corollary of psa s lemma for graphs , due to bollobs @xcite , gives a lower bound on the number of absent edges whose addition necessarily extends a longest path .
we give an analogous hypergraph version .
[ absent ] let @xmath167 be a longest path in hypergraph , @xmath184 on @xmath19 $ ] , and suppose the length of this path is @xmath185 .
if @xmath116 does not have a weak cycle of length @xmath186 , then there are at least @xmath187 absent edges of @xmath116 such that the addition of any one of these non - present edges creates a weak cycle of length @xmath186 .
let @xmath161 be one of the two endpoints of @xmath167 and @xmath188 .
we define the set of endpoints formed by rotations of @xmath167 by @xmath177 as before . by psa s lemma , we know that @xmath177 is non - expanding and so @xmath189 .
let @xmath190 be distinct vertices of @xmath177 with some largest paths @xmath191 @xmath192 @xmath193 @xmath194 starting at @xmath161 that end in @xmath195 @xmath196 @xmath193 @xmath197 , respectively .
now for each @xmath198 , let @xmath199 be the set of endpoints formed by rotations of a longest path @xmath200 starting from @xmath201 ; in other words , let @xmath202 again by psa s lemma , we have that @xmath203 and there are some @xmath204 vertices in @xmath199 . if there is some edge containing @xmath201 and at least one vertex of @xmath199 , then there would be a weak cycle of length @xmath186 ; by hypothesis , no such cycle exists . to finish off the proof ,
we enumerate these missing edges .
let @xmath205 be the collection of absent edges @xmath162 which contain @xmath201 and some @xmath206 .
we wish to determine a lower bound on @xmath207 .
note that latexmath:[\ ] ] these latter no events are independent since they consider different groups of potential edges .
further , the induced hypergraph on @xmath513 $ ] is distributed as @xmath514 .
therefore , @xmath515 just as in , we have that @xmath516 this bound holds as long as @xmath517 , which is definitely true in our case . taking on the @xmath283-term , we see that @xmath518 hence , for @xmath137 sufficiently large , we have that @xmath519 consequently , we have that @xmath520 to show that this latter sum tends to zero , it suffices to show that @xmath521 tends to zero and that the ratios of consecutive terms also ( uniformly ) tend to zero . | a _ weak ( berge ) cycle _ is an alternating sequence of vertices and ( hyper)edges @xmath0 such that the vertices @xmath1 are distinct with @xmath2 for each @xmath3 , but the edges @xmath4 are not necessarily distinct .
we prove that the main barrier to the random @xmath5-uniform hypergraph @xmath6 where each of the potential edges of cardinality @xmath5 is present with probability @xmath7 , developing a weak hamilton cycle is the presence of isolated vertices .
in particular , for @xmath8 fixed and @xmath9 , the probability that @xmath10 has a weak hamilton cycle tends to @xmath11 , which is also the limiting probability that @xmath12 has no isolated vertices . as a consequence , the probability that the random hypergraph @xmath13 where @xmath14 potential edges are chosen uniformly at random to be present , is weak hamiltonian also tends to @xmath11 .
* keywords : * random hypergraphs ; hamilton cycles |
the particle identification ( pid ) is an important tool to study the hot and dense matter created in relativistic heavy ion collisions . at intermediate @xmath6(2 - 8
@xmath4 ) hadrons can be produced not only by fragmentation but also by quark recombination .
it allows us to learn how the medium affects the particle composition @xcite . in section 2
we describe the analysis to measure the production of @xmath7 as a function of @xmath6(3 - 20 @xmath4 ) .
the discussion of the results and conclusions are presented in sections 3 and 4 , respectively .
the measurement of the yields of charged pions , kaons and ( anti ) protons in the range @xmath8 @xmath4is statistically performed exploiting the peculiarity of the mean energy loss , @xmath9 , of particles traversing the the tpc gas , in the relativistic rise regime ( @xmath10 ) of the bethe - bloch ( bb ) curve .
the tpc response was calibrated using a clean sample of @xmath11 ( @xmath12 ) identified through the weak decay topology of @xmath13 ( @xmath14 ) .
primary pions and electrons were tagged using the time of flight ( tof ) detector .
after parametrization of the bb curve and of the @xmath15resolution as a function of the calibrated @xmath9 , the measured resolution of minimum ionizing particles were @xmath16 and @xmath17 for low ( @xmath0 ) and high ( @xmath10 - 5% ) multiplicity environment , respectively .
the yields were extracted by fitting the quantity @xmath18@xmath19 to a sum of four gaussian functions ( @xmath20 ) in each @xmath6interval . the means and sigmas were constrained using the aforementioned parametrization and the @xmath6spectra obtained via factorization , as described in @xcite .
the main source of systematic uncertainty comes from the estimation of the parameters of the fitted functions .
a range of variation for those parameters was evaluated running the analysis without exploiting the information from topological and tof pid .
the parameters were randomly varied in the estimated ranges , then the yields were extracted using the same algorithm as in data . at high
@xmath6(@xmath21 @xmath4 ) the systematic uncertainties were 5% , @xmath22% and @xmath23% for @xmath11 , @xmath24 and @xmath12 , respectively .
also , we added the systematic errors from the spectra of inclusive charged particles @xcite .
figure [ fig : ptopi ] shows @xmath25 as a function of @xmath6for @xmath1and @xmath0collisions at @xmath26 2.76 tev . for completeness the lower @xmath6results are also displayed for @xmath1 @xcite . at intermediate @xmath6the ratio
is enhanced , this phenomenon was first observed at rhic @xcite . the maximum is located at 3 @xmath4and reaches @xmath27 ( @xmath28 ) for the most central ( peripheral ) events . for higher
@xmath6(@xmath29 @xmath4 ) , the ratio is consistent with fragmentation in vacuum .
figure [ fig : ptopi7 ] shows a comparison with the ratio in @xmath0collisions at @xmath30 7 tev .
it is interesting that even the @xmath0data exhibit a maximum in the same @xmath6region as observed in @xmath1collisions .
ratio a a function of @xmath6compared to @xmath0at 2.76 tev .
statistical and systematic uncertainties are displayed as error bars and boxes , respectively .
the low @xmath6results come from the combined pid of its , tpc and tof @xcite . ]
ratio as a function of @xmath6between @xmath1and @xmath0collisions at @xmath31 2.76 tev and 7 tev , respectively .
statistical and systematic uncertainties are displayed as error bars and boxes , respectively .
the low @xmath6results come from the combined pid of its , tpc and tof @xcite .
in addition the @xmath0result includes the information of the cherenkov detector , hmpid . ]
alice previously reported the nuclear modification factor ( @xmath32 ) for @xmath11 and @xmath33 @xcite .
now figure [ raa : k ] shows the @xmath32as a function of @xmath6for pions , kaons and protons , the results are presented for 0 - 5% and 60 - 80% .
for the most central events and at intermediate @xmath6(2 - 8 @xmath4 ) , @xmath32@xmath34 @xmath35 @xmath32@xmath36 and @xmath32@xmath34 @xmath37 @xmath32@xmath38 , no significant species dependence is seen at high @xmath6 . as a function of @xmath6for @xmath7 in 0 - 5% ( left ) and 60 - 80% ( right ) @xmath1collisions .
statistical and systematic uncertainties are displayed as error bars and boxes , respectively.,title="fig : " ] as a function of @xmath6for @xmath7 in 0 - 5% ( left ) and 60 - 80% ( right ) @xmath1collisions .
statistical and systematic uncertainties are displayed as error bars and boxes , respectively.,title="fig : " ]
at intermediate @xmath6 , we observe an enhancement of the @xmath39 ratio . within systematics ,
@xmath7 are equally suppressed at high @xmath6which suggests that the fragmentation is not modified by the medium .
00 k. yagiet _
et al . , _
`` quark - gluon plasma '' , cambridge , ( 2005 ) .
j. alme _ et al . , _
( alice collaboration ) , nucl .
instrum .
a * 622 * ( 2010 ) 316 .
peter christiansen , for the alice collaboration .
`` high @xmath6identified particle production in alice '' , arxiv:1208.5368 .
b. abelev _
et al_. , ( alice collaboration ) , `` pion , kaon , and proton production in central @xmath1collisions at @xmath41 tev '' , arxiv:1208.1974v1 .
leonardo milano , for the alice collaboration , `` identified charged hadron production at the lhc with the alice experiment '' , this proceedings . | in this work we report on the production of charged pions , kaons , and ( anti ) protons in @xmath0and @xmath1collisions at @xmath2 = 2.76 tev for @xmath3 @xmath4 . the evolution of the nuclear modification factor , @xmath5 , with collision centrality and transverse momentum is discussed .
lhc , alice , identified particles , high @xmath6,heavy ions . |
the ten orbital magnetic excitation of the form @xmath72 are all time reversal odd and spin rotational invariant . in the absence of time reversal symmetry breaking they form a subspace within the space of all orbital excitations .
it is thus sufficient to restrict our consideration in this subspace .
the rpa correction to the orbital magnetic response is contributed by the inter - orbital coulomb term , the hund s rule coupling term and the pair hopping term .
for example , the inter - orbital coulomb term has the following mean field decoupling(@xmath73 ) , @xmath121 when expressed in terms of @xmath85 , we have @xmath122 thus the rpa kernel is diagonal in the subspace of @xmath85 .
following the same steps , it can be shown that the rpa correction contributed by the last two terms in eq.([eqn2 ] ) cancels with each other .
[ [ the - form - of - the - symmetry - breaking - perturbation - in - the - orthogonal - phase ] ] the form of the symmetry breaking perturbation in the orthogonal phase ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ the form of the symmetry breaking perturbation in the orthogonal phase can be determined from the following group theoretical arguments .
we first consider the form of the on - site symmetry breaking term .
the point group around each fe ion in the orthogonal phase is @xmath25 and has four one dimensional irreducible representations . among the five mlwfs , @xmath123 and @xmath124
both belong to the identity representation , @xmath125 belongs to the @xmath126 representation , the linear combinations @xmath127 and @xmath128 belong to the @xmath129 and @xmath130 representation .
thus symmetry allowed on - site fermion bilinear terms have the general form of @xmath131 in the tetragonal phase , the local symmetry around each fe ion is promoted to @xmath24 , which has four one dimensional representations and a two dimensional representation . among the five mlwfs , @xmath123 belongs to the identity representation , @xmath124 and @xmath125 belong to the @xmath126 and @xmath129 representation , the linear combinations @xmath127 and @xmath128 form the two components of the two dimensional representation . for this reason ,
the bilinear form @xmath132 , @xmath133 , @xmath134 , and @xmath135 all belong to the identity representation of @xmath24 . when these symmetric perturbations are removed from eq.([eqn8 ] ) , we get the symmetric breaking perturbation in the orthogonal phase , which now takes the form of @xmath136 in which @xmath137 , @xmath138 .
the above argument can be easily generalized to determined the form the symmetry breaking perturbation on various bonds .
in particular , we find there are in total 13 independent symmetry breaking perturbations on nearest neighboring fe - fe bonds .
the form of these terms are @xmath139 in which @xmath140 @xmath141 and @xmath142 here @xmath143 , @xmath144 are p - wave form factors , @xmath94 is the d - wave form factor .
the value of these form factors are illustrated in fig.[sfig1 ] if we restrict our consideration to the subspace spanned by the @xmath145 and @xmath146 orbital , then up to nearest neighboring hopping term , the only allowable symmetry breaking perturbation has the following form @xmath147 in which @xmath148 , @xmath149 , @xmath150 . unlike the spin magnetic response
, the orbital magnetic response is contributed by both intra - band and inter - band process . as a result ,
the orbital magnetic response is much less sensitive to the variation of temperature and doping concentration of the system . in fig.[sfig2 ] , we present the temperature and doping dependence of the rpa - corrected in - plane orbital magnetic susceptibility and the ratio between the in - plane and the out - of - plane orbital magnetic susceptibility . from the figure it is clear that both quantities have only small temperature and doping dependence .
more specifically , the relative change of the in - plane orbital magnetic susceptibility for @xmath151 is only about 5 percent .
the change in the anisotropy ratio is less than 0.1 . | the orbital fluctuation of the conduction electrons in the iron - based superconductors is found to contribute significantly to the magnetic response of the system . with the use of a realistic five - band model and group theoretical analysis
, we have determined the orbital magnetic susceptibility in such a multi - orbital system . at @xmath0 , the in - plane orbital magnetic susceptibility
is predicted to be about 10@xmath1 , which is more than @xmath2 of the observed total susceptibility around 200 k in 122 systems(of about 14@xmath1 or @xmath3@xcite ) .
we find the in - plane orbital magnetic response is sensitive to the breaking of the tetragonal symmetry in the orbital space .
in particular , when the observed band splitting(between the @xmath4 and the @xmath5-dominated band ) is used to estimate the strength of the symmetry breaking perturbation@xcite , a 4.5% modulation in the in - plane orbital magnetic susceptibility can be produced , making the latter a useful probe of the orbital ordering in such a multi - orbital system . as a by product , the theory also explains the large anisotropy between the in - plane and the out - of - plane magnetic response observed universally in susceptibility and nmr measurements .
an unresolved issue in the study of the iron - based superconductors is the role of their multi - orbital nature . in most other superconductors ,
the orbital degree of freedom is quenched at low energy in the crystal field environment .
however , both lda calculation and arpes measurement@xcite indicate that in the iron - based superconductors all the five fe @xmath6 orbital play essential role in forming the low energy degree of freedom around the fermi surface .
many novel properties of the iron - based superconductors , especially those in the name of electronic nematicity@xcite , have been argued to be related to the orbital ordering in these systems@xcite .
most recently , a two - fold modulation of the magnetic susceptibility in the fe - fe plane is found to develop around a temperature that is significantly higher than the structural phase transition point@xcite .
however , it is still a mystery how the observed electronic nematicity is related to the orbital ordering of the system .
another puzzle about the iron - based superconductors is the strong anisotropy in their magnetic response observed universally in susceptibility and knight shift measurements@xcite .
the susceptibility in the fe - fe plane is found to be significantly larger than that perpendicular to it .
this is very unusual , since the magnetic response of a transition metal is usually attributed to the spin of its valence electron and is essentially isotropic . the orbital magnetic response , on the other hand ,
is usually quenched as a result of the crystal field effect .
however , since the crystal field splitting in the iron - based superconductors is very small and all the five @xmath6 orbital are involved in the low energy physics@xcite , the orbital angular momentum of the conduction electron can contribute to the magnetic response of these systems .
such a contribution is intrinsically anisotropic and depends on the electronic structure of the system , especially on the symmetry breaking in the orbital space .
the purpose of this paper is to evaluate orbital magnetic response of the iron - based superconductors from a realistic model and to explore the relation between orbital ordering and the electronic nematicity observed in recent torque magnetometry measurement@xcite .
we find the orbital magnetic susceptibility in these multi - orbital systems is comparable in magnitude with the measured total magnetic susceptibility . more specifically , the in - plane orbital magnetic susceptibility is predicted to be about 10@xmath1 , which accounts for more than @xmath2 of the observed susceptibility at 200 k in 122 systems@xcite .
furthermore , the in - plane orbital magnetic response is found to be sensitive to the breaking of the tetragonal symmetry in the orbital space , making it a useful probe of orbital ordering in these multi - orbital systems . as a by product ,
the observed strong anisotropy between the in - plane and out - of - plane magnetic susceptibility also find a natural explanation from our calculation .
the iron - based superconductors have a very complicated band structure . in this study
, we adopt the five - band tight - binding model derived from fitting the lda band structure@xcite of the lafeaso system .
following the notations of ref . , the band model reads , @xmath7 + \sum_{i,\nu,\sigma}\varepsilon_{\nu}n_{i,\nu,\sigma } , \nonumber \\
\label{eqn1}\end{aligned}\ ] ] where @xmath8 is the index for the five maximally localized wannier functions(mlwfs ) on the fe site , namely , @xmath9 , @xmath10 , @xmath11 , @xmath12 and @xmath13 .
@xmath14 denotes the hopping integral between the @xmath15-th and @xmath16-th orbital at site @xmath17 and site @xmath18 . here
an unfolded scheme is adopted as in ref .. the @xmath19 and @xmath20-axis for the wannier functions , which are in the fe - as bond direction , are rotated by 45 degree from the @xmath21 and @xmath22-axis of the fe - fe square lattice(see fig.[fig1 ] ) .
@xmath23 is the on - site energy of the @xmath15-th orbital .
the hopping integral is truncated at the fifth neighbor and the values of the model parameters can be found in ref .. is the angle between the @xmath19-axis and the direction in which magnetic susceptibility is measured . in the tetragonal phase ,
the point group symmetry around the fe ion is @xmath24 , which is broken down to @xmath25 in the orthogonal phase .
, width=340 ] the interaction of electron has the following general form @xmath26 here we have included the intra- and inter - orbital coulomb repulsion , the hund s rule coupling and the pair hopping term and have assumed that @xmath27 .
@xmath28 is the number density operator of the electron and @xmath29 . since the five fe @xmath6 orbital @xmath30 are all real functions
, they can not carry current and thus their orbital angular momentum are quenched in the static limit .
however , since the orbital content varies on the fermi surface , fluctuation in the orbital character and orbital angular momentum survives in the low energy limit and can contribute to the magnetic response of the system . in the following
we will calculate such a magnetic response in the rpa scheme .
the orbital magnetic susceptibility is defined through the correlation function of the orbital magnetic moment in the following way @xmath31 in which @xmath32 denotes the fourier component of the orbital magnetic moment density in the @xmath33 direction and @xmath34 .
here we use @xmath35 as the unit of susceptibility .
the operator for the orbital magnetic moment on a given site is defined as @xmath36 , where @xmath37 is the matrix element of the orbital magnetic moment in the basis spanned by the five mlwfs .
the matrix element @xmath37 can be determined in principle from a first principle calculation . here
we will be satisfied with the result of a semi - quantitative analysis , for which much simplification can be achieved when symmetry arguments are adopted . in the following , we will illustrate the steps for @xmath38 .
first , since @xmath39 is time reversal odd and the five @xmath6 orbital are all real , @xmath38 must be purely imaginary .
second , since @xmath39 is odd under the action of the three generators of the @xmath24 point group around each fe ion , namely @xmath40 , @xmath41 and @xmath42@xcite , while the five @xmath6 orbital transform as @xmath43 @xmath44 and @xmath45 the only none - zero matrix elements are @xmath46 and @xmath47 .
thus @xmath39 can be generally written as @xmath48+h.c.,\nonumber\ ] ] in which @xmath49 and @xmath50 are two real numbers .
following the same line of reasoning one find that @xmath51+h.c.\nonumber\\ \hat{l}^{y}&=-i&\sum_{\sigma}[\gamma_{3 } c^{\dagger}_{1,\sigma}c_{2,\sigma}+\gamma_{4 } c^{\dagger}_{3,\sigma}c_{5,\sigma}-\gamma_{5 } c^{\dagger}_{2,\sigma}c_{4,\sigma}]+h.c.\nonumber,\end{aligned}\ ] ] with the three real coefficients @xmath52 left undetermined . to have an estimate of the values of the five coefficients @xmath53 , we approximate the five mlwfs @xmath30 , @xmath54 , with the five fe @xmath6 orbital in the atomic limit .
these atomic orbital are related to the spherical harmonics of @xmath55 in the following ways(apart from the radial part of the wave function which is not used in determining the matrix element of @xmath56 ) @xmath57 where @xmath58 are the spherical harmonics of @xmath55 . since @xmath59 and @xmath60(here @xmath61 ) , we have @xmath62 we will use these values in the following calculation .
the bare orbital magnetic susceptibility is readily obtained as @xmath63 here @xmath64 is the band energy of the @xmath65-th band(@xmath66 ) and @xmath67 is the chemical potential . @xmath68 and @xmath69 is the @xmath65-th eigenvector of the band hamiltonian at momentum @xmath70 . as a comparison ,
the pauli spin susceptibility is given by @xmath71\nonumber.\end{aligned}\ ] ] unlike the orbital magnetic susceptibility , the pauli spin susceptibility has contribution only from intra - band process .
thus at low temperature the spin susceptibility is solely determined by the electronic state around the fermi surface , while the orbital magnetic susceptibility depends on electronic states both on and far away from the fermi energy . as a result , both the temperature and the doping dependence of the orbital magnetic response should be much weaker than that of the spin magnetic response .
now we consider the rpa correction of the orbital magnetic susceptibility .
the orbital magnetic excitation of the system has the general form of @xmath72 . without losing generality , we assume @xmath73
. there are in total 10 such excitations and all of them are time reversal odd and spin singlet .
the correlation function between these excitations can be defined in the following way @xmath74 and the corresponding bare susceptibility in the static limit @xmath75 is given by an expression similar to eq.([eqn6 ] ) , except that the matrix element @xmath76 should be replaced by @xmath77 the rpa correction of @xmath78 is contributed by the inter - orbital coulomb repulsion , the hund s rule coupling and the pair hopping term .
the rpa kernel is extremely simple and is given by @xmath79(see supplementary material a ) .
the rpa corrected susceptibility can be written formally as @xmath80 in which @xmath81 , @xmath82 and @xmath83 are all to be understood as @xmath84 matrix(we note while @xmath83 is a diagonal matrix in the space of @xmath85 , @xmath82 is not ) .
the orbital magnetic susceptibility can be obtained from the combinations of the matrix element of @xmath81 .
for example , @xmath86 the orbital magnetic susceptibility in other direction can be obtained in a similar way .
the observation of the two - fold modulation in the in - plane magnetic susceptibility indicates that the tetragonal symmetry of the system is broken down to orthogonal .
this can happen either through orbital ordering , or through nematicity in spin correlation@xcite .
here we assume it happens through orbital ordering , since the orbital magnetic response is much more sensitive to it than to spin nematicity .
the form of the symmetry breaking perturbation in the orthogonal phase can be largely determined by group theoretical arguments . among the five @xmath6
orbital , the @xmath87 , @xmath88 and @xmath89 orbital each form a one dimensional representation of the @xmath24 point group .
the @xmath90 and @xmath91 orbital form a two - dimensional representation which becomes reducible when the symmetry is lowered to orthogonal .
we thus focus on symmetry breaking terms in the space spanned by the @xmath90 and @xmath91 orbital .
a group theoretical analysis then shows that up to nearest neighboring hopping terms , the only allowable symmetry breaking perturbation in the orthogonal phase takes the form ( see supplementary information b ) @xmath92 in which @xmath93 is the vector between nearest neighboring fe sites .
@xmath94 is the d - wave form factor and @xmath95 , @xmath96 . here
, @xmath97 is the strength of the on - site symmetry breaking perturbation . @xmath98 and @xmath99 are the strengths of the d - wave intra - orbital and s - wave inter - orbital hopping terms between nearest neighboring fe sites . from arpes measurement@xcite
, it is found that the splitting between the @xmath4 and the @xmath5-dominated band is zero at the @xmath100 point and maximizes at the x and y point . among the three perturbations in eq.([eqn7 ] ) , only the d - wave intra - orbital hopping term is consistent with such a momentum dependence .
for example , both the @xmath97 or @xmath99-type perturbation would result in an nonzero band splitting at the @xmath100 point , which is not observed .
furthermore , the @xmath99-type perturbation has no effect at the x and y point , where the observed band splitting reaches its maximum .
we thus set @xmath101 .
this leaves us @xmath98 as the only undetermined parameter .
we are now at the position to present the numerical results .
our calculation is done at a fixed band filling of @xmath0 .
the chemical potential is determined by solving the mean field particle number equation at each temperature .
we have set @xmath102ev , @xmath103ev , as is chosen in ref .. to estimate the value of @xmath98 from the observed band splitting , we note that the band width of the iron - based superconductors is significantly smaller than the prediction of band structure calculation .
we thus fit the relative rather than the absolute magnitude of the band splitting . according to arpes measurement , the maximal band splitting between the @xmath5 and @xmath4-dominated band is about one half of the dispersion of the @xmath5-dominated band between the @xmath100 and x point@xcite .
to fit such a splitting , we set @xmath104 mev .
the calculated band dispersion along the @xmath105 and @xmath106 direction is shown in fig.[fig2 ] , which looks very similar to the experimental result@xcite .
the temperature dependence of @xmath98 is modeled by the mean field form of @xmath107 , in which @xmath108 is to be understood as the mean field critical temperature of orbital ordering .
we set @xmath109k in our calculation@xcite . and @xmath106 direction in the orthogonal phase .
the orbital character is indicated by the color of the lines and the dispersion in the tetragonal phase is plotted in thin lines for reference . in the calculation
we have set @xmath104 mev .
the dashed line indicates the fermi level at @xmath0 .
, width=302 ] in the tetragonal phase , the orbital magnetic susceptibility is found to be isotropic in the fe - fe plane and is almost temperature and doping independent for @xmath110(see supplementary material c ) .
this is reasonable since the orbital magnetic response is contributed by the whole band , rather than the electronic state near the fermi level only . at @xmath0 , the bare orbital magnetic susceptibility in the fe - fe plane is found to be about 7.3@xmath1 , which is enhanced to 10@xmath1 after rpa correction .
this is already comparable to the observed total in - plane magnetic susceptibility at 200k in 122 systems , which is about @xmath3 ( or 14@xmath1)@xcite . as a comparison ,
the bare pauli spin susceptibility is only about 2@xmath1 .
when a symmetry breaking perturbation of the @xmath98-type is turned on , a two - fold modulation shows up in the in - plane orbital magnetic susceptibility .
the angular dependence of the in - plane susceptibility at @xmath111 is shown in fig.[fig3]a . here
@xmath112 denotes the angle between the @xmath19-axis and the direction in which the magnetic susceptibility is measured .
the relative strength of the modulation is about 2.6% before rpa correction and is enhanced to 4.5% after rpa correction .
the principle axes of the modulation are along the direction of the nearest fe - fe bond , which is just what we should expect from our model construction .
the temperature dependence of the susceptibility in the principle axes are shown in fig.[fig3]b .
these predictions are in good agreement with the result of the recent torque magnetometry measurement@xcite .
thus the magnetic anisotropy provide a realistic probe of the orbital ordering in the iron - based superconductors .
a robust prediction of our theory is the strong anisotropy between the in - plane and the out - of - plane orbital magnetic susceptibility . at @xmath0 , the bare orbital magnetic susceptibility in @xmath113 direction
is found to be about 3.8@xmath1 , which is enhanced to 4.5@xmath114 after rpa correction .
this is only about the half of the value of the in - plane orbital magnetic susceptibility .
we find the ratio between the in - plane and out - of - plane orbital magnetic susceptibility is also almost temperature and doping independent for @xmath110 and is always close to 2 .
according to experiments , both the in - plane and the out - of - plane magnetic susceptibility exhibit linear temperature dependence with almost the same slope .
however , the intercept of the in - plane magnetic susceptibility is always much larger than that of the out - of - plane magnetic susceptibility@xcite .
this behavior can be easily understood if we decompose the measured magnetic susceptibility into an isotropic component that is linearly temperature dependent and a temperature independent component that is anisotropic , or , @xmath115 it is then quite natural to associate the anisotropic component @xmath116 with the orbital magnetic response , which is essentially temperature independent .
the isotropic component @xmath117 should then be attributed to the spin magnetic response , whose linear temperature dependence is still an unresolved issue in the field.@xcite in our calculation , we have used a five - band model derived from the band structure of the lafeaso system .
however , the best known susceptibility data on single crystalline sample are all taken from the 122 system .
it is thus better to perform the calculation with a material - specific band structure for the 122 systems . while this is an interesting possibility and
should be pursued in the future , we note that the basic structure of the bands in both the 1111 and the 122 systems are quite similar . since the orbital magnetic response is contributed by the whole band rather than the electronic state near the fermi level only , we expect the 1111 and 122 system to exhibit similar orbital magnetic response .
another way to improve our calculation is to use the matrix element of @xmath56 calculated from first principle code , rather than approximating them with those in the basis spanned by the atomic orbital .
however , since the form the matrix element is largely determined by symmetry , we do not expect such more advanced calculation to change the conclusion of this paper in a qualitative way .
indeed , we find that our results are not sensitive to the small variation of the parameters @xmath118 . in summary
, we have shown that the orbital angular momentum of the conduction electrons in the iron - based superconductors contributes significantly to the magnetic response of the system .
in particular , the theory predicts that the orbital magnetic susceptibility accounts for more than @xmath2 of the observed magnetic susceptibility at 200 k in 122 systems .
we show that the orbital magnetic response is sensitive to symmetry breaking in the orbital space , which makes it a useful probe of the orbital ordering in these multi - orbital systems .
a large and temperature independent anisotropy between the in - plane and the out - of plane susceptibility is predicted , which provides a natural understanding on the behavior of the magnetic response of these systems .
yuehua su is support by nsfc grant no .
10974167 and tao li is supported by nsfc grant no .
10774187 , no . 11034012 and
national basic research program of china no .
2010cb923004 .
we are grateful to k. kuroki for clarifying the phase convention used in ref .. 99 m. yi , d. lu , j.h .
chu , j. analytis , a. sorini , a. kemper , b. moritz , s.k .
mo , r.g .
moore , m. hashimoto , w.s .
lee , z. hussain , t. devereaux , i.r .
fisher , and z.x .
shen , proc .
natl .
acad .
sci . * 108 * , 6878 ( 2011 ) .
y. zhang , f. chen , c. he , b. zhou , b. p. xie , c. fang , w. f. tsai , x. h. chen , h. hayashi , j. jiang , h. iwasawa , k. shimada , h. namatame , m. taniguchi , j. p. hu , d. l. feng , phys .
rev .
b * 83 * , 054510 ( 2011 ) .
t. shimojima , k. ishizaka , y. ishida , n. katayama , k. ohgushi , t. kiss , m. okawa , t. togashi , x .- y .
wang , c .- t .
chen , s. watanabe , r. kadota , t. oguchi , a. chainani and s. shin , phys . rev .
lett . * 104 * , 057002 ( 2010 ) .
t.m .
chuang , m.p .
allan , j. lee , y. xi , n. ni , s. budko , g.s .
boebinger , p.c .
canfield , and j.c .
davis , science * 327 * , 181 ( 2010 ) .
j.h .
chu , j.g .
analytis , d. press , k. de greve , t.d .
ladd , y. yamamoto , i.r .
fisher , phys .
rev .
b * 81 * , 214502 ( 2010 ) .
j.h .
chu , j.g .
analytis , k. de greve , p.l .
mcmahon , z. islam , y. yamamoto , and i.r .
fisher , science * 329 * , 824 ( 2010 ) .
j.h .
chu , h.h .
kuo , j.g .
analytis and i.r .
fisher , science * 337 * , 710 ( 2012 ) .
s. kasahara , h. j. shi , k. hashimoto , s. tonegawa , y. mizukami , t. shibauchi , k. sugimoto , t. fukuda , t. terashima , a.h .
nevidomskyy and y. matsuda nature * 486 * , 382 ( 2012 ) .
c.c .
lee , w.g .
yin and w. ku , phys .
rev .
lett . * 103 * , 267001 ( 2009 ) .
w. lv , j. wu , and p. phillips , phys .
rev .
b * 80 * , 224506 ( 2009 )
. c.c .
chen , j. maciejko , a.p .
sorini , b. moritz , r. singh and t. p. devereaux , phys .
rev .
b * 82 * , 100504 ( 2010 ) .
w. lv , f. kruger , and p. phillips , phys .
rev .
b * 82 * , 045125 ( 2010 ) .
a. h. nevidomskyy , arxiv.org:1104.1747 ( 2011 ) .
g. wu , h. chen , t. wu , y.l .
xie , y.j .
yan , r.h .
liu , x.f .
wang , j.j .
ying and x.h .
chen , j. phys .
: cond .
matter * 20 * , 422201 ( 2008 ) .
x.f .
wang , t. wu , g. wu , h. chen , y.l .
xie , j.j .
ying , y.j .
yan , r.h .
liu , and x.h .
chen , phys .
rev .
lett .
* 102 * , 117005 ( 2009 ) .
j.q .
yan , a. kreyssig , s. nandi , n. ni , s.l .
budko , a. kracher , r.j .
mcqueeney , r.w .
mccallum , t.a .
lograsso , a.i .
goldman , and p.c .
canfield , phys .
rev .
b * 78 * , 024516 ( 2008 ) .
z. li , d.l .
sun , c.t .
lin , y.h .
su , j.p .
hu , g.q .
zheng , phys .
rev .
b * 83 * , 140506 ( 2011 ) . c. xu , m. muller , and s. sachdev , phys . rev .
b * 78 * , 020501 ( 2008 ) .
r. m. fernandes , a. v. chubukov , j. knolle , i. eremin , and j. schmalian , phys .
rev .
b * 85 * , 024534 ( 2012 ) .
k. kuroki , s. onari , r. arita , h. usui , y. tanaka , h. kontani , and h. aoki , phys .
rev . lett.*101 * , 087004 ( 2008 ) . here
@xmath40 and @xmath119 are the @xmath120 rotations along the @xmath21 and @xmath22 axis , @xmath41 and @xmath42 are the mirror planes normal to the @xmath19 and @xmath20 axis .
@xmath40 , @xmath41 and @xmath42 form a set of generators of the symmetry group @xmath24 of the tetragonal phase and @xmath40 and @xmath119 form a set of generators of the symmetry group @xmath25 of the orthogonal phase .
r. klingeler , n. leps , i. hellmann , a. popa , u. stockert , c. hess , v. kataev , h.j .
grafe , f. hammerath , g. lang , s. wurmehl , g. behr , l. harnagea , s. singh , and b. bchner , phys .
rev .
b * 81 * , 024506 ( 2010 ) .
g.m . zhang , y.h .
su , z.y .
weng , d.h .
lee , and t. xiang , europhys . lett . * 86 * 37006 ( 2009 ) .
s.p .
kou , t. li , and z.y .
weng , europhys . lett . * 88 * 17010 ( 2009 ) .
m.m .
korshunov , i. eremin , d.v .
efremov , d.l .
maslov , and a.v .
chubukov , phys .
rev .
lett .
102 , 236403 ( 2009 ) . |
the current paradigm for the electronic structure of manganites ( a@xmath0b@xmath7mno@xmath2 ) with x @xmath8 1/2 is a lattice of mn sites with d@xmath9 or d@xmath10 orbital occupancy , with the proportion of each decided by the value of x and orbital ordering ( oo ) of the occupied e@xmath11 orbital on d@xmath10 sites .
the corresponding double - exchange model hamiltonian has been studied extensively @xcite .
manganites with x @xmath8 1/2 usually exhibit a phase transition @xcite which has been assumed to be charge ordering ( co ) of the d@xmath9 and d@xmath10 sites at a temperature well below the paramagnetic transition temperature .
the conventional picture of the orbital and charge ordered phase with the ce magnetic structure @xcite is shown schematically in fig .
[ fig : fig1]a . b@xmath12mno@xmath2 compounds with a ce magnetic structure : ( a ) the conventional double exchange picture ; ( b ) the polaronic picture ; ( c ) a possible change in orbital order in a small ferromagnetic domain when an electron is added at the oxygen ion site indicated by an unshaded circle
. spins are indicated by arrows , mn d@xmath10 sites by double - lobed orbitals , mn d@xmath9 sites by small shaded circles , o@xmath13 ions by larger shaded circles and o@xmath3 ions by black circles.,height=14 ] the validity of this picture for the manganites has been questioned recently @xcite ; its contradictions were pointed out by goodenough in 1955 @xcite .
an alternative picture , the zener polaron @xcite , which challenges the ideas of double - exchange and co in the manganites , was recently proposed to account for the structure of pr@xmath5ca@xmath6mno@xmath2 determined by neutron scattering @xcite . in the conventional co and oo picture ,
ordering of the e@xmath11 electron on d@xmath10 mn sites is expected to induce a jahn - teller ( jt ) distortion ; the long mn - o bond distance in lamno@xmath2 exceeds 2.15 @xcite while the mn - o distance in cubic perovskite camno@xmath2 is 1.88 @xcite .
the lesser modulations of bond length in doped manganites with x @xmath14 1/2 ( 1.92 and 2.06 in la@xmath12ca@xmath12mno@xmath2 @xcite and 1.98 and 2.01 in pr@xmath5ca@xmath6mno@xmath2 @xcite ) suggest an intermediate valence .
in this communication we report results of unrestricted hartree - fock ( uhf ) calculations on la@xmath0ca@xmath1mno@xmath2 for x = 1/2 and x = 2/3 .
the uhf electronic structure for these compounds is interpreted in terms of ferromagnetic polarons containing two ( x = 1/2 ) or three ( x = 2/3 ) mn ions . the main difference between results from uhf calculations and
the conventional double exchange picture is that all mn ions are essentially d@xmath10 .
consequently electrons must transfer from oxygen ions to every other mn ion in the x = 1/2 compound , resulting in an ordered array of o@xmath3 ions between pairs of mn ions .
the structures of the polaron phases for x = 1/2 and x = 2/3 are shown schematically in figs .
[ fig : fig1]b and [ fig : fig2 ] .
ca@xmath15mno@xmath2 with the experimentally observed magnetic structure .
spins are indicated by arrows , mn d@xmath10 sites by double - lobed orbitals , o@xmath13 ions by shaded circles and o@xmath3 ions by black circles.,height=4 ] mn - o@xmath3-mn and mn - o@xmath3mn - o@xmath3mn chains constitute the polarons for x = 1/2 and x = 2/3 , respectively .
mn magnetic moments are in the range 3.8 to 3.9 @xmath4 for either compound while oxygen ions within the polarons have charges closer to o@xmath3 than o@xmath13 and magnetic moments of -0.7 @xmath4 .
all other oxygen ions are essentially o@xmath13 ions .
each polaron has a net magnetic moment of 7.0 @xmath4 , in good agreement with recently reported magnetisation measurements from electron holography and fresnel imaging@xcite .
uhf calculations predict that magnetic coupling within polarons is strong and ferromagnetic ( fm ) while coupling between polarons is much weaker and can be antiferromagnetic ( af ) or fm ; these observations lead to a natural explanation for the observed ce or a - type af magnetic ground states of manganites with x = 1/2 @xcite and the magnetic ground state of la@xmath16ca@xmath15mno@xmath2 , which contains polarons with the magnetic moment roughly aligned along the polaron axis @xcite .
given that the magnetic coupling within the polarons is strong and fm , the appropriate model hamiltonian for these systems at low temperature is a heisenberg hamiltonian on a triangular lattice where each polaron is a single magnetic entity , rather than the conventional double exchange hamiltonian .
results of uhf our calculations are at odds with density functional theory ( dft ) calculations in the local spin density approximation ( lsda ) reported recently @xcite and with lsda calculations done as part of this work using the same electronic structure code @xcite in that uhf calculations predict a magnetic moment on oxygen ions within the polarons whereas dft calculations do not .
all electron uhf calculations were performed for la@xmath12ca@xmath12mno@xmath2 and la@xmath16ca@xmath15mno@xmath2 using the low temperature structures refined using x - ray data by radaelli and coworkers @xcite in the full ( 80 and 120 atom ) magnetic unit cells .
spin densities of la@xmath12ca@xmath12mno@xmath2 ( fig .
[ fig : fig3 ] ) and la@xmath16ca@xmath15mno@xmath2 ( fig .
[ fig : fig4 ] ) in the ac planes of the crystal structures clearly show the polaronic nature of the electronic structures , including the magnetic moment of the o@xmath3 ions opposed to those of the neighbouring mn ions .
the magnetic unit cells of either compound each contain four polarons in the ac planes shown .
ca@xmath12mno@xmath2 in the ce - af state.,height=6 ] ca@xmath15mno@xmath2 in af state.,height=6 ] magnetic moments and mulliken populations on mn and oxygen ions in the polarons are given in table [ tab : tab1 ] .
c d d & mn moment @xmath4 & o moment @xmath4 + uhf ( ce - af ) & 3.78 , 3.91 & -0.67 + expt . & 2.57 , 2.98 & - + expt . & 3.4 @xmath17 0.2 & - + & mn charge & o charge + uhf ( ce - af ) & 2.18 , 2.17 & -1.24 + [ tab : tab1 ] + + + electron holography and fresnel imaging@xcite measurements show that below the af nel temperature the x = 1/2 compound actually consists of both fm and af domains and that the magnetic moment per mn ion in the fm domains was 3.4 @xmath17 0.2 @xmath4 .
uhf calculations predict a net magnetic moment of 7.0 @xmath4 for the polaron , i.e. the polarons are fully spin polarised , in agreement with these measurements of the magnetisation of fm domains @xcite .
earlier neutron scattering data for this compound ( table [ tab : tab1 ] ) indicated magnetic moments somewhat smaller than these fully polarised values @xcite but the trend , where larger moments are found on the sites denoted mn@xmath18 , is reproduced .
mulliken populations from uhf calculations on la@xmath0ca@xmath1mno@xmath2 indicate total mn ion populations in the range 2.25 ( x = 0 ) to 2.13 ( x = 1 ) with a monotonic variation for intermediate values of x. mn d populations have an even smaller relative variation across the range of x with a d population of 4.66 in lamno@xmath2 , a range of d populations from 4.72 to 4.73 in la@xmath12ca@xmath12mno@xmath2 , and 4.70 in camno@xmath2 .
this is consistent with a bonding picture in which the t@xmath19 shell on each mn ion is half - filled and a pair of e@xmath11 orbitals is combined with a set of four empty sp@xmath9 orbitals to form a set of equivalent d@xmath20sp@xmath9 octahedral hybrid orbitals for polar - covalent mn - o bonding .
the consistency of both the mn ion population and d population across the range of values of x contradicts the conventional double exchange picture where mn ions are assumed to have their formal mn@xmath18 and mn@xmath21 charges .
total energies were calculated for the fm state and four different low energy af states of the x = 1/2 compound .
low energy states are found by flipping entire polaron magnetic moments ; flipping the spin of one of the mn ions in a polaron results in an increase of the total energy by around 400 mev .
energies of the various low energy magnetic states of the x = 1/2 compound are well - fitted by an ising - like , nearest - neighbour hamiltonian of the form given in eq .
[ eqn : eqn1 ] .
@xmath22 labelling of exchange couplings is shown schematically in fig .
[ fig : fig5 ] and exchange constants obtained by fitting total energies of uhf calculations are given in table [ tab : tab2 ] .
the spin magnitude in eq .
[ eqn : eqn1 ] was chosen to be _ s _ = 2 .
exchange couplings within the plane shown in fig .
[ fig : fig5 ] along zig - zag chains are fm and roughly equal in magnitude whereas they are af perpendicular to the zig - zag chains .
exchange coupling between polarons in different planes is af .
the magnetic ground state is expected to be ce - type when j@xmath23 exceeds ( j@xmath24 + j@xmath25)/2 in magnitude and a - type otherwise .
both are observed to be the magnetic ground state for various x = 1/2 compounds , depending on the a and b ion types @xcite .
c d d & exchange constant ( mev ) + j@xmath23 & 5 + j@xmath26 & 8 + j@xmath24 & -14 + j@xmath25 & -12 + [ tab : tab2 ] our calculations actually predict an a - type af ground state whereas the ce - af structure is the ground state in la@xmath12ca@xmath12mno@xmath2 @xcite .
however , in similar uhf calculations on lamno@xmath2 @xcite , it was found that af couplings are significantly underestimated compared to either experiment or results of configuration interaction ( ci ) cluster calculations , whereas fm couplings are in better agreement with both .
the polaronic picture for manganites with x @xmath8 1/2 which has been presented is consistent with a wide range of experimental data and allows various observations , such as the unusual magnetisation in la@xmath16ca@xmath15mno@xmath2 , to be explained : it accounts for the observation of a - af or ce - af ground states for various combinations of counterion in a@xmath12b@xmath12mno@xmath2 @xcite ; it is consistent with full spin polarisation of mn ions in la@xmath12ca@xmath12mno@xmath2 @xcite .
one must be cautious in using uhf calculations to estimate magnitudes of magnetic moments on the oxygen ions .
an analogy can be drawn between electron transfer from the o ion in a polaron to a neighbouring mn ion and separation of the electron pair in a hydrogen molecule as the proton - proton distance is increased above the molecular equilibrium bond length : uhf wave functions for hydrogen molecules at large bond distances consist of a spin - up electron on one proton and an spin - down electron on the other ; the additional configuration in which spins on protons have been exchanged is omitted , owing to the simple functional form of the wave function .
uhf ( and lsda ) wave functions incorporate strong electron correlation effects at the cost of a proper treatment of electron spin .
spin symmetry is restored in ci methods , but these can only be applied to finite clusters owing to the large number of electronic configurations encountered in that approach .
ci calculations on the pair of mno@xmath27 octahedra in the polaron in la@xmath12ca@xmath12mno@xmath2 show that the overall spin configuration is a linear combination of several determinants @xcite with a magnetic moment on the central oxygen ion .
the la@xmath12ca@xmath12mno@xmath2 polaron electronic structure therefore resembles that in @xmath28-nav@xmath29o@xmath30 @xcite ; it is unlike the electronic structure which results in fm coupling in lamno@xmath2 , which is very well described by a single spin configuration @xcite .
an obvious question which arises within the polaron picture is , what happens when electrons are added to the x = 1/2 phase? that is , what is the electronic structure predicted to be in the fm region of the phase diagram with 0.2 @xmath31 x @xmath31 0.5? since the bottom of the conduction band in la@xmath12ca@xmath12mno@xmath2 is comprised largely of vacant oxygen 2p states on the o@xmath3 ion in the polaron , one would naturally expect to add the extra electron here . the spin density plots in figs .
[ fig : fig3 ] and [ fig : fig4 ] show that oo in the x = 1/2 compounds is _ parallel _ to the polaron axis , whereas it is t - shaped in lamno@xmath2 ( long mn - o bonds in jt distorted octahedra in lamno@xmath2 are _ perpendicular _ to each other ) .
if the extra electron is added to the polaronic oxygen site , it is expected that there will be reorientation of e@xmath11 orbitals ( fig .
[ fig : fig1]c ) and adjustment of mn - o - mn bond lengths .
oo in the vicinity of the added electron is expected to resemble that in lamno@xmath2 and one might expect to nucleate a small fm patch as further electrons are added .
the p21/m structure used for la@xmath12ca@xmath12mno@xmath2 was obtained by x - ray diffraction at 20 k ( table i ref .
the pnma wigner crystal structure used for la@xmath16ca@xmath15mno@xmath2 was obtained by powder neutron diffraction at 1.5 k ( table v of ref . | unrestricted hartree - fock calculations on la@xmath0ca@xmath1mno@xmath2 ( x = 0.5 and x = 0.67 ) in the full magnetic unit cell show that the magnetic ground states of these compounds consist of ferromagnetic molecules or polarons ordered in herring - bone patterns .
each polaron consists of either two or three mn ions separated by o@xmath3 ions with a magnetic moment opposed to those of the mn ions .
ferromagnetic coupling within the polarons is strong while coupling between them is relatively weak .
magnetic moments on the mn ions range between 3.8 and 3.9 @xmath4 in the x = 0.5 compound and moments on the o@xmath3 ions are -0.7 @xmath4 .
each polaron has a net magnetic moment of 7.0 @xmath4 , in good agreement with recently reported magnetisation measurements from electron microscopy .
the polaronic nature of the electronic structure reported here is obviously related to the zener polaron model recently proposed for pr@xmath5ca@xmath6mno@xmath2 on the basis of neutron scattering data . |
with advances in technology , it will become indispensable to determine numerous relativistic effects in the propagation of light beyond the first order in the newtonian gravitational constant @xmath0 , in particular in the area of space astrometry .
the global astrometric interferometer for astrophysics ( gaia , perryman _
2001 ) is already planned to measure the positions and/or the parallaxes of celestial objects with typical uncertainties in the range 1 - 10 @xmath1arcsecond ( @xmath1as ) whereas the laser astrometric test of relativity ( lator ) mission will measure the bending of light near the sun to an accuracy of 0.02 @xmath1as ( turyshev _ et al .
_ 2004 ) . in this last case , it is clear that the effects of the second order in @xmath0 must be taken into account . to obtain a modelling of the above - mentioned projects , it is necesssary to determine the deflection of light rays between two points @xmath2 and @xmath3 of space - time . in almost all of the theoretical studies devoted to this problem ,
the properties of light rays are determined by integrating the differential equations of the null geodesics .
this procedure is workable as long as one contents oneself with analyzing the effects of first order in @xmath0 , as it is proven by the generality of the results obtained in the litterature ( klioner 1991 , klioner & kopeikin 1992 , kopeikin 1997 , kopeikin & schfer 1999 , kopeikin & mashhoon 2002 , klioner 2003 ) .
unfortunately , analytical solution of the geodesic equations requires cumbersome calculations when terms of second order in @xmath0 are taken into account , even in the case of a static , spherically symmetric space - time ( richter & matzner 1982 , 1983 ) . however , an alternative approach exists and seems to be promising .
based on the synge s world function and variational properties of geodesic , it precisely does not require the knowledge of the geodesic and directly provides the time delay of light and the direction of a ray at the point of reception , i.e. at the observation point .
in this work we derive the general expression of the angular separation between two point light sources as measured by a given observer in arbitrary motion .
we show that the angular distance is fully determined if we calculate several ratios which can be obtained from the knowledge of the synge s world function . throughout this paper , @xmath4 is the speed of light in a vacuum and @xmath0 is the newtonian gravitational constant .
the lorentzian metric of space - time @xmath5 is denoted by @xmath6 .
we adopt the signature @xmath7 .
we suppose that space - time is covered by some global coordinate system @xmath8 .
we assume that the curves of equations @xmath9 const are timelike in the neighbourhood of the observer .
this condition means that @xmath10 in the vicinity of the observer .
we employ the vector notation @xmath11 in order to denote either the ordered set @xmath12 , or the orderer set @xmath13 .
given @xmath14 , for instance , @xmath15 denotes @xmath16 if @xmath17 and @xmath18 if @xmath19 ) , the einstein convention of summation on repeated indices being used in each case . the quantity @xmath20 denotes the ordinary euclidean norm of @xmath11 : @xmath21 if @xmath14 , and @xmath22 if @xmath23 .
the indices in parentheses characterize the order of a term in a perturbative expansion .
theses indices are set up or down , depending on the convenience .
to begin with , let us consider a light ray @xmath24 received at point @xmath25 and let us recall how is defined the direction of this ray as measured by an observer @xmath26 moving at @xmath27 with a unit 4-velocity @xmath28 .
the three - space relative to the observer @xmath26 at point @xmath27 is the subspace @xmath29 of tangent vectors orthogonal to @xmath28 ( see figure below ) . , width=510 ] an arbitrary vector @xmath30 at @xmath27 admits one and only one decomposition of the form v = v__u + v__u , where @xmath31 is colinear to the unit vector @xmath28 and @xmath32 is a vector of the three - space @xmath29 .
since @xmath32 and @xmath28 are orthogonal , one has v__u = ( u .
v ) u and v__u = v - ( u .
v ) u . the vector @xmath32 is called the ( orthogonal ) projection of @xmath30 onto the three - space relative to the observer @xmath26 .
its magnitude @xmath33 is given by v__u= .
the direction of vector @xmath30 as seen by the observer @xmath26 is the direction of the unit spacelike vector @xmath34 defined as v__u^ = = .
consider now a light ray @xmath24 received at @xmath27 and denote by @xmath35 a vector tangent to @xmath24 at @xmath27 . in this work
, we always assume that a vector tangent to a light ray is a null vector and is future oriented , so that l^2 = 0 , u.l > 0 .
the direction of the ray @xmath24 as measured by the observer @xmath26 is the direction of the vector @xmath36 . by using eq .
( [ di1 ] ) , and then taking into account eqs .
( [ nul ] ) , it is easy to see that l__u^ = - u .
let @xmath37 be an other light ray received at @xmath27 .
if @xmath38 denotes a vector tangent to @xmath37 at @xmath27 , the direction of @xmath37 as observed by @xmath26 is given by eq .
( [ di2 ] ) in which @xmath38 is substituted for @xmath35 . as a consequence , the angular separation between @xmath24 and @xmath37 as measured by @xmath26 may be defined as the angle @xmath39 between the two vectors @xmath36 and @xmath40 belonging to the same subspace @xmath29 ( see soffel 1988 , brumberg 1991 ) . angle @xmath39 may be characterized without ambiguity by relations as follow _
u = - l__u^ .
l__u^ , 0 _ u < .
taking into account properties of a light ray , we have the following relations @xmath41 as @xmath28 is an unitary vector , we can express @xmath42 with @xmath43 as follows = g_00 + 2 g_0i ^i + g_ij^i^j , where ^i = = .
finally , substituting equations ( [ llll]-[hlb ] ) into equation ( [ an1 ] ) yields the fundamental formula ^2 = - _ x_o , where _ 0 = 1 , _
i = , _ 0 = 1 , _
the determination of the angular distance thus requires explicit computations of the ratios @xmath44 and @xmath45
. they can be obtained by the integration of null geodesic equations .
however we will show that they result easily from the knowledge of the synge s world function .
synge s world function is a scalar function of the base point @xmath46 and the field point @xmath47 .
it is defined by ( synge , 1964 ) @xmath48 and the integral is evaluated on the unique geodesic @xmath49 that links @xmath46 to @xmath47 , @xmath50 being an affine parameter . a fundamental property of @xmath51 is to give an important information concerning the covariant components of the vectors tangent to @xmath49 at @xmath46 and @xmath47 respectively : @xmath52 if @xmath49 is a light ray , we can consider @xmath46 as the observation point . moreover , in this case , we have @xmath53 we recently show that explicit determination of @xmath51 can be obtained from the integration of hamilton - jacobi equations without the knowkledge of @xmath49 ( le poncin - lafitte _ et al .
all this means that determination of ratios @xmath44 and @xmath45 require the following steps : * to determine @xmath51 by solving hamilton - jacobi equations , * to impose the condition @xmath54 , * to compute the following relation @xmath55
in this paper , we give the general and rigorous expression of the observed angular distance between two point sources of light .
the fundamental formulae ( [ an7 ] ) and ( [ cov ] ) show that the theoretical calculation of the angular distance can be carried out when the world function is known . in this idealized sense
, one can say that the problem of space astrometry involves one and only one unknown function .
an other important point is that the aberration and the gravitational deflection of light can not be treated as completely distinct phenomena . | almost all of the studies devoted to relativistic astrometry are based on the integration of the null geodesic differential equations .
however , the gravitational deflection of light rays can be calculated by a different method , based on the determination of the bifunction giving half the squared geodesic distance between two arbitrary points - events , the so - called synge s world function .
we give a brief review of the main results obtained by this method . |
some of the theoretically most fascinating aspects of crack propagation in amorphous materials are the instabilities that are observed in well controlled laboratory experiments @xcite . besides some exceptions , ( see for example @xcite and also @xcite ) , it would be fair to say that the observed instabilities are still poorly understood .
it is the opinion of the present authors that the reason for the relative lack of understanding is that the theory of crack propagation did not treat cracks as moving free boundaries whose instabilities stem from the dynamics of the free boundary itself . instead ,
crack tip dynamics " were replaced by energy balance within the theory of linear elastic fracture mechanics @xcite , together with an ad - hoc `` law '' of one nature or another as to where a crack is supposed to move . in principle
this undesirable state of affairs can be greatly improved within the shear - transformation - zones ( stz ) theory of amorphous materials @xcite .
this theory treats developing cracks or growing cavities as free boundaries of a material in which both elasticity and plasticity are taken into account , preserving all the symmetries and conservation laws that promise a possibly correct theory of amorphous materials driven out of mechanical equilibrium .
this theory in its various appearances was compared to a number of experiments and simulations ( see below ) , with a growing confidence that although not final , stz theory is developing in the right direction . indeed , the application of a highly simplified version of stz theory to crack propagation resulted in physically interesting predictions , explaining how plasticity can intervene in blunting a crack tip and resulting in velocity selection @xcite .
the application of the full fledged theory of stz to crack propagation is still daunting ( although not impossible ) due to the tensorial nature of the theory and the need to deal with an extremely stiff set of partial differential equations with a wide range of time - scales and length - scales involved .
for that reason it seemed advantageous to apply the full theory to a situation in which the symmetries reduce the problem to an effectively scalar theory ; this is the problem of a circular cavity developing under circular symmetric stress boundary conditions @xcite .
while this problem does not reach the extreme conditions of stress concentration that characterizes a running slender crack , it still raises many physical issues that appear also in cracks , in particular the give - and - take between elasticity and plasticity , the way stresses are transmitted to moving boundaries ( in apparent excess of the material yield stress ) and most importantly for this paper , the possible existence of dynamical instabilities of the moving free boundary .
this last issue might also be connected to the difference between ductile and brittle behaviors . in the former ,
a growing cavity is likely to remain rather smooth , whereas in the latter , one may expect an instability resulting in the growth of `` fingers '' , possibly ending up being cracks .
it is one of the challenges of the present paper to examine whether the theory may predict a transition , as a function of material parameters or a constitutive relation , between these two types of behavior .
note that we have chosen to study the problem in a purely 2-dimensional geometry ; recently quasi 2-dimensional systems exhibited interesting failure dynamics in laboratory experiments , where the 3rd dimension appears irrelevant for the observed phenomena @xcite .
our motivation here is however theoretical , to reduce the unnecessary analytic and numerical complications to a minimum and to gain insight as to the main physical effects under the assumption that the thin third dimension in real systems does not induce a catastrophic change in behavior . when this assumption fails , as it does in some examples c.f .
@xcite , the analysis must be extended to include the third dimension .
this is beyond the scope of this paper . in sec .
[ ebc ] we present the equations that describe the problem at hand and specify their boundary conditions .
particular attention is paid to distinguish between the general eulerian formulation which is model - independent ( subsec .
[ general ] ) and the constitutive relations involving plasticity where the stz model is explained ( subsec .
[ stz ] ) .
this section finishes with the presentation of the unperturbed problem , preparing the stage for the linear stability analysis which is discussed in sec .
[ lsa ] . in this section
we present a general analysis where inertia and elastic compressibility effects are taken into account . in appendix
[ qs ] we complement the analysis by considering the quasistatic " ( when the velocity of the boundary is sufficiently small ) and incompressible case ( when the bulk modulus is sufficiently large ) and show that both formulations agree with one another in the relevant range .
the results of the stability analysis are described in detail in sec .
[ results ] and a few concluding remarks are offered in sec .
[ discussion ] .
we start by writing down the full set of equations for a general two - dimensional elasto - viscoplastic material .
a basic assertion of this theory is that plastic strain tensors in such materials are not state variables since their values depend on the entire history of deformation .
thus , one begins by introducing the total rate of deformation tensor @xmath0 \ , \label{d_tot}\ ] ] where @xmath1 is the material velocity at the location @xmath2 at time @xmath3 and @xmath4 denotes here the transpose of a tensor .
this type of eulerian formulation has the enormous advantage that it disposes of any reference state , allowing free discussion of small or large deformations .
as is required in an eulerian frame we employ the full material time derivative for a tensor @xmath5 , @xmath6 where @xmath7 is the spin tensor @xmath8 \ . \label{omega}\ ] ] for a scalar or vector quantity @xmath9 the commutation with the spin tensor vanishes identically .
the eulerian approach allows a natural formulation of moving free boundary problems ; this will be shown to lead to a significant advance compared to more conventional treatments .
the plastic rate of deformation tensor @xmath10 is introduced by assuming that the total rate of deformation tensor @xmath11 can be written as a sum of a linear elastic and plastic contributions @xmath12 we further assume that @xmath10 is a traceless tensor , corresponding to incompressible plasticity .
all possible material compressibility effects in our theory are carried by the elastic component of the deformation .
the components of the linear elastic strain tensor @xmath13 are related to the components of stress tensor , whose general form is @xmath14 according to @xmath15 where @xmath16 and @xmath17 are the two dimensional bulk and shear moduli respectively .
the tensor @xmath18 is referred to hereafter as the deviatoric stress tensor " and @xmath19 as the pressure .
the equations of motion for the velocity and density are @xmath20 in order to prepare the general set of equations for the analysis of a circular cavity we rewrite the equations in polar coordinates . for that aim
we write @xmath21 where @xmath22 and @xmath23 are unit vectors in the radial and azimuthal directions respectively .
these expressions enable us to represent the divergence operator @xmath24 in the equations of motion and the covariant derivative @xmath25 in the material time derivative of vectors and tensors .
some care should be taken in evaluating these differential operators in polar coordinates since the unit vectors themselves vary under differentiation according to @xmath26 we then denote @xmath27 , @xmath28 , @xmath29 and using eqs .
( [ sig ] ) we obtain @xmath30 in this notation the equations of motion ( [ eqmot1 ] ) can be rewritten explicitly as @xmath31 where @xmath32 is calculated explicitly in appendix [ polar ] .
equations ( [ el_pl ] ) can be rewritten in components form as @xmath33 here the components of the total rate of deformation tensor are related to the velocity according to eqs .
( [ d_tot ] ) as @xmath34 \ ,
\label{eq : totalrate}\end{aligned}\ ] ] where the components of the spin tensor @xmath35 in eq .
( [ omega ] ) are given by @xmath36 \
.\end{aligned}\ ] ] the calculation of the tensor @xmath37 is presented in appendix [ polar ] ; the linear elastic strain components of eqs .
( [ linear ] ) are given by @xmath38 since most of the materials of interest have a large bulk modulus @xmath16 , i.e. they are almost incompressible , we assume that the density is constant in space and time @xmath39 therefore , eq .
( [ eqmot2 ] ) is omitted .
finally , the existence of a free boundary is introduced as the following boundary conditions @xmath40 where @xmath41 is the unit normal vector at the free boundary . up to now
we have considered mainly symmetries and conservation laws . a general theoretical framework for the elasto - viscoplastic deformation dynamics of amorphous solids should be supplemented with constitutive equations relating the plastic rate of deformation tensor @xmath10 to the stress and possibly to other internal state fields .
we use the constitutive equations of the recently proposed athermal shear transformation zones ( stz ) theory @xcite .
this theory is based on identifying the internal state fields that control plastic deformation .
the basic observation is that stressing a disordered solid results in localized reorganizations of groups of particles .
these reorganizations occur upon surpassing a local shear threshold , and when they involve a finite irreversible shear in a given direction , we refer to them as an stz transition " .
once transformed , due to a local redistribution of stresses , the same local region resists further deformation in that direction , but is particulary sensitive to shearing transformation if the local applied stress reverses its direction .
thus an stz transition is conceived as a deformation unit that can undergo configurational rearrangements in response to driving forces .
furthermore , the stress redistribution that accompanies an stz transition can induce the creation and annihilation of other local particle arrangements that can undergo further localized transitions ; these arrangements are formed or annihilated at a rate proportional to the local energy dissipation ( recall that thermal fluctuations are assumed to be absent or negligible ) . in this sense
the interesting localized events need not depend on pre - existing " defects in the material , but can appear and disappear dynamically in a manner that we describe mathematically next .
this picture is cast into a mathematical form in terms of a scalar field @xmath42 that represents the normalized density of regions that can undergo stz transitions , a tensor @xmath43 that represents the difference between the density of regions that can undergo a transition under a given stress and the reversed one , and an effective disorder temperature @xmath44 that characterizes the state of configurational disorder of the solid @xcite .
the present state of the theory relates these internal state fields , along with the deviatoric stress tensor @xmath18 , to the plastic rate of deformation tensor @xmath10 according to @xmath45 this equation represents the dependence of the plastic rate of deformation on the current stress @xmath46 and the recent history encoded by the internal state tensorial field @xmath43 .
this field acts as a back - stress , effectively reducing the local driving force for stz transitions , up to the possible state of jamming when the whole parentheses vanishes .
the parentheses provides information about the orientation of the plastic deformation .
the function @xmath47 determines the magnitude of the effect , and is re - discussed below .
the field @xmath42 appears multiplicatively since the rate of plastic deformation must be proportional to the density of stz .
the second equation describes the dynamics of the internal back stress field @xmath48 this equation captures the dynamical exchange of stability when the material yields to the applied stress .
the equation has a jammed fixed point when the plastic deformation vanishes , in agreement with the state of stz being all in one orientation , without the production of a sufficient number of new ones in the other orientation .
the jammed state is realized when the applied stress is below the yield stress .
when the stress exceeds the threshold value the stable fixed point of this equation corresponds to a solution with non - vanishing plastic rate of deformation .
this state corresponds to a situation where enough stz are being created per unit time to allow a persistent plastic flow .
the quantity @xmath49 represents the rate of stz production in response to the flow @xmath10 .
the next equation , for the density of stz @xmath42 , is an elementary fixed point equation reading @xmath50 the unique fixed point of this equation is the equilibrium solution @xmath51 where @xmath44 is a normalized temperature - like field which is not necessarily the bath temperature when the system is out of thermal and/or mechanical equilibrium .
the last equation is for this variable , reading @xmath52,\nonumber\\
\hbox{with}\quad \bar{d}^{pl}&\equiv & \sqrt{\frac{d^{pl}_{ij}d^{pl}_{ij}}{2 } } \ .\end{aligned}\ ] ] this is a heat - like equation for the configurational degrees of freedom ; it is discussed in detail below . here and elsewhere
we assume that quantities of stress dimension are always normalized by the yield stress @xmath53 ; this is justified as the stz equations exhibit an exchange of dynamic stability from jamming to flow at @xmath54 , i.e. at a stress that equals to @xmath53 @xcite .
the set of eqs .
( [ eq : dpl])-([eq : chi ] ) is a tensorial generalization of the effectively scalar equations derived in @xcite ; such a generalization can be obtained by following the procedure described in ref .
@xcite . in these equations
, @xmath55 is the elementary time scale of plasticity , @xmath56 is a dimensionless constant and @xmath57 is a specific heat in units of @xmath58 per particle .
a weak point of the theory is the lack of a first - principle derivation that determines the function @xmath59 in eq .
( [ eq : dpl ] ) , which lumps together much of the microscopic physics that controls the stress - dependent rate of stz transitions .
our theory constrains it to be a symmetric function of @xmath60 that vanishes with vanishing derivatives at @xmath61 , due to the athermal condition that states that no transitions can occur in a direction opposite to the direction of @xmath60 @xcite .
this constraint is not sufficient , however , to determine @xmath59 . to appreciate the uncertainties , recall that stz transitions are relaxation events , where energy and stress are expected to re - distribute . even without external mechanical forcing ,
aging in glassy systems involves relaxation events that are poorly understood @xcite .
the situation is even more uncertain when we deal with dynamics far from mechanical equilibrium .
the best one can do at present is to choose the function @xmath62 by examining its influence on the resulting macroscopic behaviors @xcite .
thus in this paper we will examine the sensitivity of the stability of the expanding cavity to two different choices of @xmath59 . at present
we use the one - parameter family of functions , @xmath63 , proposed in @xcite @xmath64 the integral is over a distribution of transition thresholds whose width is controlled by a parameter @xmath65 ( and see @xcite for details ) . for finite values of @xmath65 there can be nonzero sub - yield plastic deformation for @xmath66 .
this behavior is well documented in the literature cf .
@xcite in the context of experimental stress - strain relations and plastic deformations .
we note that for @xmath60 very small or very large , @xmath67 in sec .
[ rate_function ] we propose a different one - parameter family of functions @xmath68 and study in detail the implications of this different choice on the stability of the expanding cavity .
( [ eq : chi ] ) deserves special attention .
it is a heat - like equation for the effective disorder temperature @xmath44 with a fixed - point @xmath69 which is attained under steady state deformation .
this reflects the observations of ref .
@xcite , where the effective temperature @xmath44 was shown to attain a unique value in the limit @xmath70 , where @xmath71 was the particles vibrational time scale . indeed , in most applications , realistically
_ imposed _ inverse strain rates are much larger than the elementary time scale @xmath71 , i.e. @xmath72 . if we identify our @xmath55 with the vibrational time scale @xmath71 ( see for example @xcite ) , we conclude that @xmath69 can be taken as a constant , independent of the plastic rate of deformation .
this assumption was adopted in all previous versions of stz theory .
note also that a low plastic rate of deformation is associated with @xmath73 , i.e. a deviatoric stress that approaches the yield stress from above .
however , the situation might be very different in free boundary evolution problems , where high stresses concentrate near the boundary , reaching levels of a few times the yield stress .
estimating @xmath44 in the typical range of @xmath74 @xcite , @xmath75 is in the range @xmath76 .
therefore , estimating the other factors in eq .
( [ eq : dpl ] ) , for the high stresses near the free boundary , in the range @xmath77 , we conclude that @xmath78 can reach values in the range @xmath79 .
very recent simulations @xcite demonstrated convincingly that in this range of normalized plastic rates of deformation , @xmath69 shows a considerable dependence on this rate , see fig .
[ hl ] . since @xmath44 affects plastic deformation through an exponential boltzmann - like factor , even small changes of @xmath69 in eq .
( [ eq : chi ] ) can generate significant effects @xcite .
this issue is of particular importance for the question of stability ( and localization ) under study since the strain rate sensitivity of @xmath69 might incorporate an instability mechanism ; fluctuations in the plastic rate of deformation , caused for example by fluctuations in @xmath44 , can induce , through @xmath69 , a further localized increase in plastic deformation and so on .
this intuitive idea will be studied in the analysis to follow .
the set of eqs .
( [ eq : dpl])-([eq : chi ] ) @xcite ( and slight variants ) was shown to capture viscoelastic behavior in a variety of examples .
these include small stress and finite plasticity at intermediate stresses @xcite , a transition to flow at the yield stress ( as discussed above ) @xcite , the deformation dynamics of simulated amorphous silicon @xcite , the necking instability @xcite , the deformation dynamics near stress concentrations @xcite , the cavitation instability @xcite and strain localization @xcite . in this work we focus on the implications of these constitutive equations on the stability of propagating free boundaries in relation to the failure modes of amorphous solids .
in this subsection we adapt the general theory to the circular symmetry of the unperturbed expanding cavity problem .
we consider an infinite medium with a circular cavity of radius @xmath80 , loaded by a radially symmetric stress @xmath81 at infinity .
the superscript @xmath82 in all the quantities denotes the fact that they correspond to the perfectly symmetric case that is going to be perturbed later on . for the perfect circular symmetry the velocity field @xmath83
is purely radial and independent of the azimuthal angle @xmath84 , i.e. @xmath85 this symmetry also implies that @xmath86 and all the diagonal components are independent of @xmath84 .
( [ el_pl ] ) , after a simple manipulation , can be rewritten as @xmath87 where we have defined @xmath88 the equations of motion ( [ eom ] ) reduce to @xmath89 the boundary conditions are given by @xmath90 the initial conditions are chosen to agree with the solution of the static linear - elastic problem , i.e. @xmath91 this choice reflects the separation of time scales between elastic and plastic responses .
this separation of time scales can be written explicitly in terms of the typical elastic wave speed , the radius of the cavity and the time scale of plasticity : @xmath92 finally , the rate of the cavity growth is simply determined by @xmath93 for the circular symmetry , the stz equations ( [ eq : dpl])-([gamma ] ) reduce to @xmath94 \ . \nonumber\end{aligned}\ ] ] note that the @xmath44 and @xmath95 equations contain a factor of the small stz density @xmath96 , which implies they are much stiffer than the @xmath97 and @xmath42 equations . therefore , whenever the advection terms can be neglected this separation of time scales @xcite allows us to replace the equations for @xmath98 and @xmath99 by their stationary solutions @xmath100 and @xmath101 note that eq .
( [ eq : m0 ] ) has two stable fixed - point solutions given by eq .
( [ m0_fix ] ) , where we used eq .
( [ lam0_fix ] ) and omitted the advection term .
the transition between these two solutions corresponds to a transition between a jammed and a plastically flowing state for a deviatoric stress below and above the yield stress respectively @xcite .
( [ eq : dpl0 ] ) exhibits the corresponding solutions in terms of the plastic rate of deformation , zero and finite , below and above the yield stress respectively .
the unperturbed problem was studied in detail in ref .
it was shown that for stresses @xmath102 smaller than a threshold value @xmath103 the cavity exhibits transient dynamics in which its radius approaches a finite value in a finite time .
when this happens the material is jammed . on the other hand , for @xmath104
the cavity grows without bound , leading to a catastrophic failure of the material , accompanied by large scale plastic deformations .
we stress that to our knowledge this mode of failure by propagating a plastic solution is new , apparently not related to other recently discovered failure fronts @xcite
. one major goal of the present study is to analyze the stability of the unbounded growth modes that result from this cavitation .
however , we are also interested in the range @xmath105 where the unperturbed theory predicts no catastrophic failure . in this range ,
a failure can still occur if the cavity , prior to jamming , loses its perfect circular symmetry in favor of relatively slender propagating `` fingers '' . in that case ,
stress localization near the tips of the propagating `` fingers '' can lead to failure via fracture .
such a scenario is typical of brittle fracture where the stress localization due to the geometry of the defect drives crack propagation that might lead to macroscopic failure .
we derive here a set of equations for the linear perturbations of the perfect circular symmetry where both inertia and elastic compressibility effects are taken into account . in appendix
[ qs ] we complement the analysis by considering the quasi - static and incompressible case .
this case is mathematically more involved as it contains no explicit time evolution equation for the velocity and the pressure fields . by comparing the results of the two formulations we test for consistency and obtain some degree of confidence in the derivation and the numerical implementation of the equations presented in this section .
the quantities involved in the problem are the tensors @xmath106 as well as the pressure @xmath107 , the velocity @xmath108 and the location of the free boundary @xmath109 .
we start by expanding all these quantities as follows @xmath110 here all the quantities with the superscript @xmath111 are assumed to be much smaller than their @xmath82 counterparts and @xmath112 is the discrete azimuthal wave - number of the perturbations . the small perturbation hypothesis results in a formal linear decomposition in which each linear mode of wave - number @xmath112 is decoupled from all the other modes . when nonlinear contributions are non - negligible , all the modes become coupled and the formal linear decomposition is invalid .
we expand then the equations of motion ( [ eom ] ) to first order to obtain @xmath113 we proceed by expanding eqs .
( [ el_pl ] ) to first order , which after a simple manipulation yields @xmath114 + 2{d^{pl}}^{(1 ) } \ , \nonumber\\ & & \frac{1}{2 } \left[\frac { \partial v^{(1)}_\theta}{\partial r } + \frac{n v^{(1)}_r - v^{(1)}_\theta}{r } \right ] = \\ & & \frac{1}{2 \mu } \left [ \frac{\partial \tau^{(1 ) } } { \partial t } + v^{(0)}_r \frac{\partial \tau^{(1)}}{\partial r } -\frac{2 s^{(0 ) } v_\theta^{(1)}}{r}\right ] + { d^{pl}_{r \theta}}^{(1)}\nonumber \ . \label{last_eq}\end{aligned}\ ] ] at this point we derive an evolution equation for the dimensionless amplitude of the shape perturbation @xmath115 . to that aim
we note that @xmath116 \ .\ ] ] expanding this relation using eqs .
( [ basic_perturbations ] ) , we obtain to zeroth order eq .
( [ edge_velocity0 ] ) and to first order @xmath117 therefore , we obtain @xmath118 \ .\nonumber\end{aligned}\ ] ] this is an important equation since a linear instability manifests itself as a significant increase in @xmath115 such that nonlinear terms become non - negligible .
note that the two last terms in the square brackets are always negative , therefore an instability can occur only if the first term in the square brackets is positive with absolute value larger than the sum of the two negative terms .
moreover , recall that the problem is non - stationary , implying that all the zeroth order quantities depend on time . in order to derive the boundary conditions for the components of the stress tensor field we expand to linear order the normal unit vector @xmath41 ( not to be confused with the discrete wave - number @xmath112 ) and tangential unit vector @xmath119 at the free boundary , obtaining @xmath120 eqs .
( [ stressbc ] ) , expanded to first order , translate to @xmath121\ , \\
\label{taub } \tau^1(r^{(0)})\!\!&=&\!\ !
n \left[s^{(0)}(r^{(0)})-p^{(0)}(r^{(0)})\right ] \frac{r^{(1)}}{r^{(0)}}.\end{aligned}\ ] ] in addition , all the first order fields decay as @xmath122 .
the initial conditions are determined by the perturbation scheme that is being studied . to avoid dealing with an infinite and time - dependent domain we applied the following time - dependent coordinate transformation @xmath123 this transformation allows us to integrate the equations in the time - independent finite domain @xmath124 $ ] , with the price of introducing new terms in the equations . controlling the equations at small distances required the introduction of an artificial viscosity on the right - hand - side ( rhs ) of eq .
( [ eqmot1 ] ) .
the term introduced is @xmath125 , with @xmath126 chosen of the order of the square of space discretization over the time discretization .
this introduces zeroth order contributions on the rhs of eq .
( [ eom0 ] ) and first order contributions on the rhs of eqs .
( [ eom1])-([eom2 ] ) .
the only missing piece in our formulation is the perturbation of the tensorial stz equations .
in addition to the fields considered up to now , the analysis of the stz equations includes also the internal state fields @xmath127 therefore , in addition to eqs .
( [ basic_perturbations ] ) we have @xmath128 we then expand systematically eqs .
( [ eq : dpl])-([c_s ] ) .
first , we have @xmath129 accordingly we expand @xmath130 ( assuming @xmath131 ) in the form @xmath132 where @xmath133 substituting the last three equations into ( [ eq : dpl ] ) and expanding to first order , we obtain @xmath134 \ , \\
\label{dploff1 } \tau_0 { d^{pl}_{r\theta}}^{(1 ) } & = & \epsilon_0 \lambda^{(0)}{{\mathcal{c}}}\left(s^{(0)}\right)\left(\frac{\tau^{(1)}}{|s^{(0)}|}-m_{r\theta}^{(1)}\right ) \ .\end{aligned}\ ] ] we then expand @xmath49 in the form @xmath135 \ .\nonumber\\\end{aligned}\ ] ] eq .
( [ eq : m ] ) is now used to obtain @xmath136 ^ 2}- \frac{\lambda^{(1)}}{\lambda^{(0)}}\right ) \right ] \ , \nonumber\end{aligned}\ ] ] and @xmath137 using eq .
( [ eq : lambda ] ) we obtain @xmath138 ^ 2}\!-\!\lambda^{(1 ) } \right)\!+\!\gamma^{(1)}\left(e^{-1/\chi^{(0)}}\!-\!\lambda^{(0 ) } \right ) \ . \nonumber\end{aligned}\ ] ] expanding @xmath139 , similarly to eq .
( [ sbar_exp ] ) , we obtain @xmath140 accordingly we expand @xmath141 ( with @xmath142 ) in the form @xmath143 then , using eq .
( [ eq : chi ] ) we obtain @xmath144 thus , eqs .
( [ dpl1])-([dploff1 ] ) , ( [ m1])-([moff1 ] ) , ( [ lambda1 ] ) and ( [ chi1 ] ) constitute our equations for the dynamics of the first order stz quantities .
these equations already reveal some interesting features .
first note that the coupling between @xmath145 ( which is the quantity that is expected to be of major importance in determining @xmath146 in eq .
( [ smallness ] ) through eqs .
( [ first_eq])-([last_eq ] ) ) and @xmath147 , @xmath148 depends on @xmath149 .
this means that the strength of the coupling depends on @xmath65 .
similarly , the coupling between @xmath145 and @xmath150 depends on @xmath151 which is also a function of @xmath65 .
these observations demonstrate the importance of the precise form of the function @xmath152 .
this issue is further discussed in sec .
[ rate_function ] .
finally , note that whenever the advection terms can be neglected , the known separation of time scales @xcite allows us to use eqs .
( [ m0_fix])-([lam0_fix ] ) and to replace the equations for @xmath148 , @xmath153 and @xmath154 by their stationary solutions @xmath155 ^ 2 } & if $ s^{(0 ) } > 1 $ } \ , \ ] ] @xmath156 ^ 2 } & if $ s^{(0 ) } > 1$}\ ] ] and @xmath157 ^ 2}e^{-1/\chi^{(0 ) } } \ . \label{lam1_fix}\ ] ] in the next section we summarize the results of our analysis of the equations derived in sec .
[ zeroth ] , [ inertial ] and [ perturbstz ] .
we are now ready to present and discuss the results of the stability analysis of the expanding circular cavity .
the full set of equations was solved numerically as discussed above .
time and length are measured in units of @xmath55 and @xmath158 respectively . @xmath42 and @xmath97 are set initially to their respective fixed - points .
the material - specific parameters used are @xmath159 , @xmath160 , @xmath161 , @xmath162 , @xmath163 , @xmath164 , @xmath165 and @xmath166 , unless otherwise stated . in subsec .
[ pert_shape ] we study perturbations of the shape of the cavity and of the effective temperature @xmath44 . in subsec . [ strain_rate ] we study the effect of the rate dependence of @xmath69 on the stability analysis and in subsec .
[ rate_function ] we analyze the effect of the stress - dependent rate function @xmath152 . studying the linear stability of the expanding cavity
can be done by selecting which fields are perturbed and which are left alone . in practice
each of the fields involved in the problem may experience simultaneous fluctuations , including the radius of the cavity itself .
therefore , one of our tasks is to determine which of the possible perturbation leads to a linear instability .
to start , we perturb the radius of the expanding cavity at @xmath167 while all the other fields are left alone . in fig .
[ pertr ] we show the ratio @xmath115 as a function of time for various loading levels @xmath102 ( both below and above the cavitation threshold ) and wave - numbers @xmath112 .
the initial amplitude of the perturbation was set to @xmath168 .
the observation is that the ratio @xmath115 does not grow in time in all the considered cases were the radius was perturbed , implying that here the circular cavity is stable against shape perturbations .
note that @xmath115 decays faster for larger n and for larger @xmath102 .
also note that for @xmath169 .
i.e. for unbounded zeroth order expansion , the ratio @xmath115 decays to zero while below the cavitation threshold this ratio approaches a finite value .
the latter observations means that when the material approaches jamming ( with @xmath170 attaining a finite value in a finite time ) the perturbation have not yet disappeared entirely .
we stress at this point the non - stationary nature of the problem in which @xmath80 is an increasing function of time .
thus , even if the absolute magnitude of the amplitude of the shape perturbation @xmath171 increases with time , an instability is not automatically implied ; @xmath171 should increase sufficiently faster than @xmath80 in order to imply an instability . to exemplify this feature of the problem
, we present in fig .
[ onlyr1 ] @xmath171 for @xmath172 and @xmath173 .
it is observed that even though @xmath174 increases , the smallness parameter @xmath115 decreases , see fig .
[ pertr ] .
note also that @xmath174 does not increase exponentially as expected in stationary linear stability analysis , but rather tends to asymptote to a constant .
next we have tested the stability of the expanding cavity against initial perturbations in the velocity field or in the stress field .
the results were quantitatively similar to those for the shape perturbations summarized in figs . [ pertr ] and [ onlyr1 ] , all implying linear stability . in light of these results , we concentrated then on the effect of perturbations in the stz internal state fields . since the dynamics of the tensor @xmath43
are mainly determined by the deviatoric stress field @xmath60 , we focus on fluctuations in the effective disorder temperature @xmath44 . this may be the most liable field to cause an instability . indeed , in ref .
@xcite it was shown that @xmath44 perturbations control strain localization in a shear banding instability .
qualitatively , an instability in the form of growing `` fingers '' involves strain localization as well ; plastic deformations are localized near the leading edges of the propagating `` fingers '' . in ref
@xcite , based on the data of ref .
@xcite , it was suggested that the typical spatial fluctuations in @xmath44 have an amplitude reaching about @xmath175 of the homogeneous background @xmath44 .
obviously we can not treat such large perturbations in a linear analysis and must limit ourselves to smaller perturbations . in fig .
[ pertchi ] we show the ratio @xmath115 as a function of time for a perturbation of size @xmath176 , introduced at time @xmath167 .
the wave - number was set to @xmath173 and @xmath102 was set both below and above the cavitation threshold .
first , note that for both loading conditions @xmath115 increases on a short time scale of about @xmath177 , a qualitatively different behavior compared to the system s response to shape perturbations .
second , note the qualitatively different response below and above the cavitation threshold . in the former case ,
@xmath115 increases monotonically , approaching a constant value when @xmath170 attains a finite value ( i.e. jamming ) . in the latter case
, @xmath115 increases more rapidly initially , reaches a maximum and then decays to 0 in the large @xmath3 limit .
therefore , in spite of the initial growth of @xmath115 , for this magnitude of @xmath44 perturbations , the expanding circular cavity is linearly stable ; below the cavitation threshold the relative magnitude of the deviation from a perfect circular symmetry @xmath115 tends to a finite constant , i.e. a shape perturbation is `` locked in '' the material , while above the threshold the cavity retains its perfect circular symmetry in the large @xmath3 limit .
nevertheless , in light of the significant short time increase in @xmath115 ( here up to @xmath178 ) , we increased the initial ( @xmath167 ) @xmath44 perturbation to the range @xmath179 , in addition to shape perturbations of a typical size of @xmath180 . in these cases
@xmath115 grows above @xmath181 ; even more importantly , the field @xmath182 ( as well as other fields in the problem ) becomes larger than @xmath183 near the boundary of the cavity , _ invalidating _ the small perturbation hypothesis behind the perturbative expansion and signaling a linear instability .
naturally , this breakdown of the linearity condition takes place firstly near a peak of the ratio @xmath115 , similar to the one observed in fig .
[ pertchi ] .
we thus propose that sufficiently large perturbations in the shape of the cavity and the effective disorder temperature @xmath44 , but still of formal linear order , may lead to an instability .
this dependence on the magnitude of the perturbations in a linear analysis is a result of the non - stationarity of the growth .
another manifestation of the non - stationarity is that even in cases where we detect an instability , it was not of the usual simple exponential type where an eigenvalue changes sign as a function of some parameter ( or group of parameters ) .
combined with the evidence for the existence of large fluctuations in @xmath44 @xcite , the present results indicate that it will be worthwhile to study the problem by direct boundary tracking techniques where the magnitude of the perturbation is not limited . we conclude that the issue of the stability of the expanding cavity can be subtle .
sufficiently small perturbations are stable , though there is a qualitative difference in the response to perturbations in the effective disorder temperature @xmath44 , where the ratio @xmath115 increases ( at least temporarily ) , and other perturbations , where @xmath115 decays .
we have found that for large enough @xmath44 perturbations combined with initial shape perturbations , but still within the formal linear regime , the growth of @xmath115 takes the system beyond the linear regime , making nonlinear effects non - negligible and signaling an instability .
this observation is further supported by the existence of large @xmath44 fluctuations discussed in @xcite . note
that none of these conclusions depend significantly on variations in @xmath56 and @xmath57 . moreover , perturbing the expanding cavity at times different than @xmath167 or introducing a pressure inside the cavity instead of a tension at infinity did not change any of the results . the analysis of sec .
[ pert_shape ] indicates the existence of a linear instability as a result of varying the magnitude of the perturbations , mainly in @xmath44 , and not as a result of varying material parameters . here , and in sec .
[ rate_function ] , we aim at studying the effect of material - specific properties on the stability of the expanding cavity . up to now we considered @xmath69 as a constant parameter .
however , as discussed in detail in sec .
[ stz ] , the plastic rate of deformation near the free boundary can reach values in the range where changes in @xmath69 were observed
. therefore , we repeated the calculations using the function @xmath184 plotted in fig .
[ hl ] . in fig .
[ strainrate ] we compare @xmath115 as a function of time with and without a plastic rate of deformation dependence of @xmath69 , both above and below the cavitation threshold .
the initial perturbation has @xmath176 and @xmath173 . both below and above the cavitation threshold the plastic rate of deformation
dependent @xmath184 induces a stronger growth of @xmath115 , though the effect is much more significant above the threshold .
this is understood as significantly higher rate of deformation is developed above the cavitation threshold , where unbounded growth takes place @xcite , compared to below the threshold where the rate of deformation vanishes at a finite time .
we note that the dependence of @xmath69 on @xmath185 affects both the zeroth and first order solutions such that @xmath170 and @xmath174 increase .
our results show that @xmath174 is more sensitive to this effect than @xmath170 , resulting in a tendency to lose stability at yet smaller perturbations .
we conclude that the tendency of @xmath69 to increase with the rate of deformation plays an important role in the stability of the expanding cavity and might be crucial for other strain localization phenomena as the shear banding instability @xcite .
moreover , this material - specific dependence of @xmath69 , that was absent in previous formulations of stz theory , might distinguish between materials that experience catastrophic failure and those that do not , and between materials that fail through a cavitation instability @xcite and those who fail via the propagation of `` fingers '' that may evolve into cracks .
this new aspect of the theory certainly deserves more attention in future work .
we note in passing that recently an alternative equation to eq .
( [ eq : chi ] ) for the time evolution of the effective temperature @xmath44 was proposed in light of some available experimental and simulational data @xcite .
preliminary analysis of the new equation in relation to the stability analysis performed in this paper indicates that the circular cavity _ does _ become linearly unstable @xcite .
a more systematic study of this effect may be a promising line of future investigation . here
we further study the possible effects of details of the constitutive behavior on the macroscopic behavior of the expanding cavity . in this subsection
we focus on the material function @xmath152 .
this phenomenological function , as discussed in sec .
[ stz ] , describes the stress - dependent stz transition rates .
it is expected to be symmetric and to vanish smoothly at @xmath61 in athermal conditions @xcite .
the plastic rate of deformation for @xmath186 can be measured in a steady state stress - controlled simple shear experiment .
for such a configuration the deviatoric stress tensor is diagonal and the stable fixed - points of eqs .
( [ eq : m])-([eq : chi ] ) imply that the steady state plastic rate of deformation of eq .
( [ eq : dpl ] ) reads @xmath187 therefore , if the steady state relation @xmath188 is known , @xmath152 can be determined from measuring the steady state value of @xmath95 for various @xmath186 , see for example @xcite .
the idea then is to interpolate the @xmath189 behavior to the @xmath186 behavior with a single parameter that controls the amount of sub - yield deformation in the intermediate range .
in fact , a procedure to measure @xmath152 at intermediate stresses was proposed in ref .
up to now we used the one - parameter family of functions @xmath190 of eq .
( [ c_s ] ) , where @xmath65 controls the sub - yield deformation .
we now aim at studying the effect of choosing another function @xmath59 . here
we specialize for @xmath191 , with @xmath192 in fig .
[ changingc ] we show @xmath152 according to the previous choice of eq .
( [ c_s ] ) with @xmath166 and also @xmath152 according to the present choice of eq .
( [ c_s1 ] ) with @xmath193 .
the different behaviors of @xmath152 and @xmath194 near @xmath54 might affect differently @xmath170 and @xmath174 , thus influencing the stability of the expanding cavity . in fig .
[ changingc1 ] we compare @xmath115 as a function of time for @xmath152 of eq .
( [ c_s ] ) ( previous choice ) with @xmath166 and @xmath152 of eq .
( [ c_s1 ] ) ( present choice ) with @xmath193 , both for a constant @xmath69 .
an effective temperature perturbation with @xmath176 and @xmath173 was introduced at @xmath167 for @xmath195 .
we observe that @xmath115 grows faster for the present choice compared to the previous one .
for the sake of illustration we added the result of a calculation with the plastic rate of deformation dependent @xmath69 as discussed in sec .
[ strain_rate ] . as expected , the effect is magnified .
we conclude that the material - specific function of the stress dependence of the stz transition rates @xmath152 can affect the stability of the expanding cavity , possibly making it unstable for smaller perturbations .
again , the relations between this constitutive property and the macroscopic behavior should be further explored in future work . bringing into consideration explicit macroscopic measurements
, one can constrain the various phenomenological features of the theory of amorphous plasticity .
this philosophy provides a complementary approach to obtaining a better microscopic understanding of the physical processes involved .
we presented in this paper a detailed analysis of the linear stability of expanding cavity modes in amorphous elasto - viscoplastic solids .
the stability analysis is somewhat delicate due to the non - stationarity of the problem , thus a perturbation may grow leaving the problem stable if this growth is slower than the growth of the radius of the cavity .
the radial symmetry of the expanding cavity makes it surprisingly resilient to perturbations in shape , velocity , external strains and pressure . on the other hand
the radial symmetry may be lost due to perturbations in the internal state fields , especially @xmath44 , and is also sensitive to details of the constitutive relations that are employed in the stz theory . in this respect
we highlight the role of the plastic rate of deformation dependent @xmath184 and of the stress - dependent rates of stz transitions @xmath152 .
it is difficult to reach conclusive statements , since growth of perturbations beyond the linear order invalidate the approach taken here , calling for new algorithms involving surface tracking , where the size of perturbations is not limited .
nevertheless the results point out that instabilities are likely , motivating further research into the nonlinear regime .
of particular interest is the possibility to select particular forms of constitutive relations by comparing the predictions of the theory to macroscopic experiments .
this appears as a promising approach in advancing the stz theory towards a final form .
* acknowledgements * we thank t. haxton and a. liu for generously sharing with us their numerical data , and chris rycroft for pointing out an error in an early version of the manuscript .
this work had been supported in part by the german israeli foundation and the minerva foundation , munich , germany .
e. bouchbinder acknowledges support from the center for complexity science and the lady davis trust .
the aim of this appendix is to derive some expressions in polar coordinates that were used earlier in the paper .
specifically , our goal is to calculate the divergence and covariant derivative of a tensor in polar coordinates .
we represent a general second order tensor @xmath5 in polar coordinates as @xmath196 where @xmath197 and @xmath198 are unit vectors in polar coordinates and @xmath199 denotes a tensor product . using eqs .
( [ polaro])-([unit_vectors ] ) we obtain @xmath200 substituting eqs .
( [ sig_p_s ] ) for @xmath5 , we obtain the right - hand - sides of eqs .
( [ eom ] ) .
we proceed now to calculate the covariant derivative of a tensor @xmath201 . using eqs .
( [ polaro ] ) , ( [ unit_vectors ] ) and ( [ tcomp ] ) we obtain @xmath202 substituting eqs . ( [ eq : stress - strain ] ) for @xmath5 we obtain the needed expressions for @xmath203 in eq . ( [ eq : da_ij ] ) .
the aim of this appendix is to derive independently the linear perturbation theory for a quasi - static and incompressible case and to compare to the inertial and compressible case in the limit of large bulk modulus @xmath16 and small velocities @xmath204 .
we show that the results in this limit agree , giving us some degree of confidence in the derivation and the numerical implementation of the equations in both cases .
the unperturbed problem in the quasi - static and incompressible limit was discussed in detail in @xcite and is obtained by taking the quasi - static and the incompressible limits in the equations of sec .
[ zeroth ] . before considering the linear stability problem
, we stress that the linear perturbation theory of the stz equations , presented in sec .
( [ perturbstz ] ) remains unchanged in the present analysis .
only the equations of motion and the kinematic equations are being modified . in the absence of inertial terms , the equations of motion ( [ eqmot1 ] ) become @xmath205 @xmath206 to first order we obtain @xmath207 @xmath208 the boundary conditions of eqs .
( [ spb])-([taub ] ) can be further simplified by using the force balance equation to zeroth order and the zeroth order boundary conditions of eq .
( [ bc0 ] ) @xmath209 substituting into ( [ spb ] ) and ( [ taub ] ) we obtain @xmath210 in addition , all the first order fields decay as @xmath122 . in principle , the initial conditions for the partial differential equations for the first order fields depend on the type of perturbation under consideration . for explicit perturbations in the shape of the cavity ,
i.e. @xmath211 , we can determine the initial stress field by assuming it is simply the quasi - static linear elastic solution corresponding to the perturbed circle . in order to obtain this solution
we start with the bi - laplace equation for the airy stress function @xmath212 @xcite @xmath213 where the stress tensor components are given by @xmath214 we then expand the solution in the form @xmath215 the general solutions for @xmath216 , that also decay at infinity , are given by @xmath217 with @xmath218 . substituting in eqs .
( [ airy_components ] ) , using the boundary conditions to first order and the following zeroth order solution @xmath219 one obtains @xmath220^n\!\ ! \sigma^{\infty}\frac{r^{(1)}}{r^{(0)}}\ , \quad b=\left[r^{(0 ) } \right]^{n+2}\!\ ! \sigma^{\infty}\frac{r^{(1)}}{r^{(0)}}\ .\ ] ] the resulting stress components are easily calculated , from which we obtain @xmath221\ .\nonumber\\\end{aligned}\ ] ] these are the initial conditions for the first order stress tensor components in terms of the initial @xmath174 . to proceed we
expand eqs .
( [ el_pl ] ) to first order , assuming @xmath222 , @xmath223 \!-\ ! { d^{pl}}^{(1)}\!,\nonumber\\ \label{kinematic_1st_b } & & \frac{\!-\!n v^{(1)}_\theta \!+\ ! v^{(1)}_r}{r } = \\ & & \frac{1}{2 \mu } \left[\ ! \frac{\partial s^{(1)}}{\partial t } + v_r^{(0 ) } \frac{\partial s^{(1)}}{\partial r } \!+\ !
v^{(1)}_r \frac{\partial s^{(0)}}{\partial r}\right ] \!+\ ! { d^{pl}}^{(1)},\nonumber\\ \label{kinematic_1st_c } & & \frac{1}{2 } \left[\!\frac { \partial v^{(1)}_\theta}{\partial r } \!+ \!\frac{n v^{(1)}_r\!-\ ! v^{(1)}_\theta}{r } \right ] = \\ & & \frac{1}{2 \mu } \left[\ ! \frac{\partial \tau^{(1 ) } } { \partial t } + v^{(0)}_r \frac{\partial \tau^{(1)}}{\partial r}-\frac{2 s^{(0 ) } v_\theta^{(1)}}{r } \right ] + { d^{pl}_{r \theta}}^{(1 ) } \nonumber\ .\end{aligned}\ ] ] in order to propagate @xmath150 and @xmath224 in time according to these equations we need to know @xmath225 and @xmath226 at each time step . however , a basic feature of the quasi - static problem is that there is no evolution equation for the velocity field .
therefore , we must calculate @xmath225 and @xmath226 in a different way .
we now discuss the major mathematical difficulty in the quasi - static formulation , i.e. the absence of an explicit evolution equation for the velocity field @xmath227 . to overcome this difficulty
, we should derive new _
ordinary differential equations _ for @xmath225 and @xmath226 such that their time evolution is inherited from the other fields in the problem . the first equation can be obtained readily by adding ( [ kinematic_1st_a ] ) to ( [ kinematic_1st_b ] ) @xmath228 from which we can extract @xmath226 @xmath229 in order to obtain the second equation , we eliminate @xmath230 from the equations by operating with @xmath231 on eq .
( [ force2 ] ) , adding the result to eq .
( [ force1 ] ) and taking the partial time derivative to obtain @xmath232 here and elsewhere the dot denotes partial time derivative . using eqs .
( [ kinematic_1st_a ] ) and ( [ kinematic_1st_c ] ) we obtain @xmath233\nonumber\\ & & -v_r^{(0)}\frac{\partial\tau^{(1)}}{\partial r}+\frac{2s^{(0)}v_\theta^{(1)}}{r}\ , \\ \label{s_dot}\!\!\!\!&\dot s^{(1)}&=-\!2\mu\ !
\left(\!\frac{\partial v_r^{(1)}}{\partial r}\!+\!{d^{pl}}^{(1)}\right)-\!v_r^{(1)}\frac{\partial s^{(0)}}{\partial r}\nonumber\\ & & -v_r^{(0)}\frac{\partial s^{(1)}}{\partial r } \
.\end{aligned}\ ] ] substituting the last two relations in eq .
( [ elim_p ] ) and using eq .
( [ theta ] ) , we obtain a _
fourth order linear ordinary differential equation _ for @xmath146 .
since it is straightforward to obtain , but very lengthy , we do not write it explicitly here .
it is important to note that the coefficients in this equation depend on time and therefore by solving it at each time step we effectively have a time evolution for the velocity field .
once one solves for @xmath146 , eq .
( [ theta ] ) can be used to calculate @xmath234 .
the forth order linear differential equation requires four boundary conditions .
the first boundary condition is obtained by using eq .
( [ taub1 ] ) , with eqs .
( [ kinematic_1st_c ] ) , ( [ edge_velocity1 ] ) and ( [ theta ] ) we obtain a linear relation between @xmath235 , @xmath236 and @xmath237 , which is the required boundary condition .
another boundary condition is obtained by multiplying eq .
( [ force2 ] ) by r , operating with @xmath238 on the result and using eq .
( [ spb1 ] ) .
additional simple manipulations result in a linear relation between @xmath235 , @xmath236 , @xmath237 and @xmath239 .
this is a second boundary relation .
two other boundary conditions are obtained from the requirement that @xmath146 vanishes at @xmath240 with vanishing derivative @xmath241 with these four boundary conditions the forth order differential equation can be solved in the following way : at each step we guess @xmath242 and @xmath243 and use the first two boundary conditions to calculate @xmath244 and @xmath245
. then we use the forth order differential equation to calculate @xmath146 and @xmath246 at @xmath240 .
we improve our guess until the solution satisfies eqs .
( [ bcinfty ] ) ( the shooting method ) .
thus , we have a complete solution procedure ( assuming that the plastic rate of deformation is known , see section [ perturbstz ] ) ; for a given @xmath247 and @xmath248 we solve the forth order differential equation for @xmath249 following the procedure described above .
having @xmath249 we use eq .
( [ theta ] ) to obtain @xmath250 .
then we use eqs .
( [ kinematic_1st_a ] ) , ( [ kinematic_1st_c ] ) and ( [ edge_velocity1 ] ) to propagate @xmath251 , @xmath252 and @xmath174 in time .
we follow the same procedure at each time step to obtain the full time evolution of the perturbation .
we note that we have eliminated @xmath253 from the problem , though we can calculate it at every time step using eq .
( [ force1 ] ) or ( [ force2 ] ) .
we are now able to compare the quasi - static and incompressible case to the inertial and compressible counterpart in the limit of small velocities and large bulk modulus @xmath16 .
we introduced at @xmath167 a perturbation of magnitude @xmath168 to the radius of the cavity , with a discrete wave - number @xmath254 , and solved the dynamics in both formulations .
we chose @xmath105 such that the velocities are small and @xmath255 in the inertial case in order to approach the incompressible limit . in fig .
[ comparer ] we compare @xmath170 and @xmath174 for both the quasi - static and the inertial formulations .
the agreement is good .
note that the stability of the expanding cavity depends on the time dependence of the ratio @xmath115 ; however , we do not discuss the stability yet , but focus on the comparison between the two formulations . in fig .
[ comparefields ] we further compare the predictions of the two formulations for the zeroth and first order deviatoric stress field @xmath60 and effective disorder temperature @xmath44 at a given time . in all cases the differences are practically indistinguishable .
we thus conclude that the quasi - static formulation and the inertial one agree with one another , giving us some confidence in the validity of both . in particular we conclude that the inertial formulation can be used with confidence also for high velocities where the quasi - static counterpart becomes invalid .
note that for situations where @xmath69 can become very large one should rewrite eq .
( [ eq : chi ] ) such that term in the square brackets reads @xmath256 $ ] .
this modification is not needed here as we do not consider such situations . | recently , the existence and properties of unbounded cavity modes , resulting in extensive plastic deformation failure of two - dimensional sheets of amorphous media , were discussed in the context of the athermal shear - transformation - zones ( stz ) theory .
these modes pertain to perfect circular symmetry of the cavity and the stress conditions . in this paper
we study the shape stability of the expanding circular cavity against perturbations , in both the unbounded and the bounded growth regimes ( for the latter the unperturbed theory predicts no catastrophic failure ) .
since the unperturbed reference state is time dependent , the linear stability theory can not be cast into standard time - independent eigenvalue analysis .
the main results of our study are : ( i ) sufficiently small perturbations are stable , ( ii ) larger perturbations within the formal linear decomposition may lead to an instability ; this dependence on the magnitude of the perturbations in the linear analysis is a result of the non - stationarity of the growth , ( iii ) the stability of the circular cavity is particularly sensitive to perturbations in the effective disorder temperature ; in this context we highlight the role of the rate sensitivity of the limiting value of this effective temperature .
finally we point to the consequences of the form of the stress - dependence of the rate of stz transitions .
the present analysis indicates the importance of nonlinear effects that were not taken into account yet .
furthermore , the analysis suggests that details of the constitutive relations appearing in the theory can be constrained by the modes of macroscopic failure in these amorphous systems . |
scanning tunneling microscopy ( stm ) experiments on surfaces of the high temperature cuprate superconductor bi@xmath3sr@xmath3cacu@xmath3o@xmath4 ( bscco ) provide a powerful tool for probing strong electronic correlations in this material.@xcite since the stm measures a local density of states ( ldos ) with a high real - space resolution it is widely used for studying impurity effects in @xmath0-wave superconductors ( sc ) . @xcite
extensive stm studies have been performed on both magnetic ni and non - magnetic zn impurities in bscco .
particularly interesting is the zn resonant state , i.e. , a large peak in the differential conductance ( i.e. , in the ldos ) at small but finite energy which is mainly localized near the zn site.@xcite despite the fact that there are numerous theoretical studies dealing with the resonant state induced by non - magnetic impurities in @xmath0-wave cuprates , no consensus on the theoretical description has been reached yet.@xcite the zn@xmath5 impurity has spin @xmath6 and it might be natural to describe it as a delta - function potential scatterer within a potential scattering ( ps ) model.@xcite in the ps model the resonance peak can be identified with a quasibound state whose energy can be tuned by varying the strength of the potential .
however , a low - energy state appears only for large potential values , i.e. , in the unitary limit .
moreover , some experimental findings can not be easily explained within the ps model like the spatial dependence of the resonance peak which shows a global maximum on the zn site and local maxima on the next - nearest neighbor cu sites.@xcite nuclear magnetic resonance ( nmr ) experiments are another powerful probe of strongly correlated systems like cuprates . in a series of nmr experiments
it has been established that each non - magnetic impurity ( e.g. , zn@xmath5 or li@xmath7 ) induces a net local @xmath8 moment that is distributed on the cu ions in its vicinity.@xcite furthermore , the effective induced moment interacts with the elementary excitations of the bulk material giving rise to kondo - like physics which has indeed been observed in nmr experiments.@xcite consequently , one can treat a non - magnetic zn impurity in the @xmath0-wave sc within the kondo model and identify the low - energy impurity state in the stm spectra with the kondo resonance.@xcite however , it is still not completely settled whether the potential scattering or the kondo - like description is more appropriate for modeling the non - magnetic impurities in the cuprates and for explaining all experimental facts.@xcite the main objective of the present study is to resolve this question . in principle , kondo - like behavior should be suppressed by an external magnetic field which would allow to identify the origin of the observed impurity stm signal .
so far , no such experiments have been performed and it has been shown that for this purpose rather high magnetic fields would be needed.@xcite recently the stm measurements on bscco have revealed the presence of nanoscale electronic inhomogeneities which show up as spatial modulation of both the gap amplitude and the ldos.@xcite these electronic inhomogeneities and variations of the ldos attracted numerous theoretical and experimental studies.@xcite random out - of - plane dopant atoms which change the pair interaction locally@xcite and a competing order like antiferromagnetism@xcite have been proposed to be a possible origin of the ldos modulations in the cuprates .
importantly , the stm studies on bscco have identified two distinct electronic domains : ( i ) the domains with a small - gap magnitude and with well - developed sc coherence peaks and ( ii ) the large - gap domains without sc coherence peaks .
remarkably , in the presence of impurities it was observed that impurity resonances are present only in the small - gap domains and no impurity resonances can be detected in the large - gap domains.@xcite using this experimental fact and motivated by the question about the origin of the impurity resonance in @xmath0-wave cuprate sc we investigate how the resonance state is influenced by putting the impurity into different domains of the sc host in both the ps model and in the kondo model . to get quantitative results we employ a @xmath1-matrix and a large-@xmath2 approach for the ps and the kondo model , respectively .
we conclude that the zn impurity resonance in bscco has predominantly a kondo - like character .
very recently a new possibility for the origin of the zn resonance peak in bscco has been proposed.@xcite namely , in @xmath0-wave sc short - ranged off - diagonal ( @xmath9 ) impurities generate low - energy andreev states which are then associated with the impurity resonance seen in the stm spectra .
if the usual @xmath10 impurity scattering is absent , the andreev resonant state exists symmetrically around the fermi level . however , by including a weak @xmath10 potential this symmetry is broken .
this gives results which bear closer resemblance to the stm data .
below , we will comment on some of these results .
the remainder of the paper is organized as follows . in sec .
[ sec : model ] we introduce the relevant models for describing a non - magnetic impurity embedded in a @xmath0-wave superconductor . in sec .
[ sec : method ] the @xmath1-matrix approach and the large-@xmath2 formalism are explained .
our numerical results and a comparison with experimental data are presented in sec .
[ sec : results ] .
we conclude the paper in sec .
[ sec : concl ] .
to describe a single impurity atom in a superconducting material we will consider a hamiltonian consisting of a bulk and an impurity part , @xmath11 . both terms are specified below . as a model for the cuprate superconductor we use the usual bcs hamiltonian [ eq : ham_bcs ] & h_^= _ * k*_*k*^ _ * k*^ , where @xmath12 is a nambu spinor with momentum @xmath13 , @xmath14 ( @xmath15 ) creates ( annihilates ) an electron with momentum @xmath16 and spin @xmath17 , and the three pauli matrices in particle - hole space
are denoted by @xmath18 .
we use a tight - binding dispersion of the form & _ * k*^= ( k_x+k_y)+t_2k_xk_y + ( 2k_x+ + & + 2k_y)+(2k_xk_y+ k_x2k_y)+ + & + t_52k_x2k_y , with values for the hopping elements that fit the arpes data:@xcite @xmath19ev , @xmath20ev , @xmath21ev , @xmath22ev , @xmath23ev , and @xmath24ev for optimal doping ( @xmath25 ) .
the @xmath0-wave pairing function is given by @xmath26 where @xmath27mev . in order to model the nanoscale electronic inhomogeneities seen by the stm ( i.e. , the small- and large - gap domains ) it is necessary to modify the pure bcs state ; this will be discussed in sec .
[ sec : ldos ] .
we consider an effective impurity hamiltonian which consists of a potential scattering and a magnetic term , @xmath28 .
the potential scattering term describes electrostatic interactions between the impurity atom and the conduction electrons whereas the magnetic term represents an impurity moment embedded in the sc host .
we assume that the potential scattering is completely localized at the impurity site @xmath29 , i.e. , @xmath30 .
consequently , the scattering occurs only in the isotropic @xmath31-wave channel and the scattering term is given by @xmath32 , where @xmath33 , and @xmath34 is the number of sites . in the ps model
the bulk hamiltonian is defined as @xmath35 . to specify the magnetic term in the impurity hamiltonian we use the so - called bose - fermi kondo model.@xcite it describes a spin-@xmath36 impurity , @xmath37 , coupled both to the electrons of the conduction band and to a bosonic bath representing bosonic collective excitations in the host material .
( apart from its application to the impurity moments in the cuprates,@xcite the bose - fermi kondo model has attracted considerable interest in the context of local quantum criticality within an extended dynamical mean - field theory,@xcite and also in certain mesoscopic systems like a quantum dot coupled to ferromagnetic leads.@xcite ) in the @xmath0-wave cuprate sc it is reasonable to separate the low - energy degrees of freedom into _ fermionic _ bogoliubov quasiparticles around the nodal points , and into _ bosonic _
antiferromagnetic spin-1 collective fluctuations near the ordering wave - vector @xmath38.@xcite the effective impurity moment induced by non - magnetic impurity interacts with both the fermionic and the bosonic bulk excitations .
therefore , a non - magnetic impurity in cuprates is expected to be well described by the bose - fermi kondo model@xcite [ eq : kondo_ham ] & h_^= j_^ * s*_^_0^+ _ 0^ * s*_^ _ 0^. here , @xmath39 is the conduction electron spin operator at the impurity site , @xmath40 are the pauli matrices in spin space , and @xmath41 is the usual kondo coupling .
the impurity moment is coupled to the bosonic bath via a coupling constant @xmath42 .
the local field @xmath43 represents the local orientation of the antiferromagnetic order parameter and it is given by @xmath44 , where @xmath45 is the bulk exchange constant , @xmath46 is a dimensionless function containing the geometry of the system , and @xmath47.@xcite the simplest bosonic bath consists of free vector bosons , @xmath48 , with the dispersion @xmath49 , where the momentum @xmath50 is measured relative to the ordering wave - vector @xmath51 and @xmath52 is the velocity of the spin excitations.@xcite at zero temperature , the effective bosonic mass @xmath53 is equivalent to the bulk spin gap @xmath54 ( as seen in neutron scattering experiments@xcite ) .
a spin gap @xmath55 corresponds to a bulk quantum - critical point controlling a transition between a quantum paramagnet ( @xmath56 ) and the antiferromagnetically ordered phase.@xcite in the present study the kondo hamiltonian describes the physics of the so - called pseudogap ( i.e. , softgap ) bose - fermi kondo model.@xcite there , a fermionic ldos follows a power - law close to the fermi energy , @xmath57 , where @xmath58 is measured from the fermi level .
clearly , the @xmath0-wave superconductor has @xmath59 .
in addition , at zero temperature and for @xmath55 , the bosonic ldos also follows a power - law , @xmath60 , where @xmath61 , and @xmath0 is the dimension of the system ( for cuprates @xmath62 ) .
it is important to note that for @xmath63 the bose - fermi kondo model reduces to the extensively studied fermionic pseudogap kondo model.@xcite depending on the value of @xmath64 and the presence or absence of particle - hole symmetry , the fermionic pseudogap kondo model exhibits a quantum phase transition between a phase with kondo screening for @xmath65 , and a phase where the magnetic moment is free ( i.e. , decoupled from the bath ) for @xmath66 .
it is also known that the presence of an additional bosonic bath for @xmath67 , leads to a suppression of the kondo screening.@xcite note that in the present analysis we will use the bose - fermi kondo model with the bulk hamiltonian @xmath68.@xcite to describe non - magnetic impurities like zn ( or li ) in the copper oxide planes it is appropriate to use the so - called extended magnetic impurity.@xcite nmr experiments have shown that the effective moment induced by a non - magnetic impurity is spatially distributed around the impurity site , @xmath69 , and mainly resides on the four nearest neighbor copper sites.@xcite therefore , the kondo model with an extended magnetic impurity is required in order to explain both the stm and nmr measurements on cuprates with non - magnetic impurities.@xcite in the present calculation we consider a four - site kondo model in @xmath0-wave superconductors where the effective magnetic moment is coupled to the four cu sites @xmath70 adjacent to the impurity .
one should bare in mind that in this four - site model the effective ldos of the fermionic bath seen by the impurity changes ; this will be shortly discussed in sec .
[ sec : fermk ] .
we start with the definition of the conduction electron green s function in the absence of impurities , which is given by @xmath71 .
( in our notation all underlined symbols denote matrices . )
its fourier transform is given by @xmath72 , where @xmath73 represents the fermionic matsubara frequencies and @xmath74 is the inverse temperature .
note that the bare green s function in real space is simply given by @xmath75^{-1}$ ] . the single scatterer problem , @xmath76 , can be solved exactly with the usual result for the perturbed green s function@xcite [ eq : tmat_green ] & _ c(i_n,*r*,*r*)= _ c^0(i_n,*r*-*r*)+ + & + _ c^0(i_n,*r*-*r*_0 ) ( i_n ) _ c^0(i_n,*r*_0-*r * ) , where the @xmath1-matrix is given by & ( i_n)=v_0^z ^-1 .
the @xmath1-matrix depends on the strength of the scattering potential and on the local conduction electron green s function , @xmath77 .
it is important to note that this local green s function is the only input quantity representing the bulk material ; it differs significantly in the small - gap and the large - gap regions of the sc host ( more details are given in sec .
[ sec : ldos ] ) . in order
to obtain quantitative results in the bose - fermi kondo model we use the large-@xmath2 formalism .
the large-@xmath2 theory can be constructed by generalizing the spin symmetry of the impurity moment , the conduction electrons , and the vector bosons from su(2 ) to su(@xmath2 ) .
an antisymmetric representation of the impurity spin is given by @xmath78 with the constraint @xmath79 , where @xmath80 ( @xmath81 ) creates ( annihilates ) auxiliary fermions with @xmath82 spin flavors .
the constraint is enforced by introducing a lagrange multiplier @xmath83 .
the generalization of the spin symmetry leads to @xmath84 triplet bosons denoted by @xmath85.@xcite in what follows we consider only the particle - hole symmetric case for which @xmath86 .
therefore , in the large-@xmath2 formalism the impurity hamiltonian is given by [ eq : kondo_hamn ] & h_= _
f_^f_^ c_0^c_0^+ _ f_^f_^ _ 0^+ + & + _ 0(_f_^f_^
-q_0^ ) .
it is important to emphasize that in the limit @xmath87 physics of both the fermionic kondo model and the bose kondo model is controlled by a corresponding saddle point.@xcite we proceed with defining the pseudo - fermion green s function @xmath88 and its fourier transform @xmath89 .
we drop the spin indices since the su(@xmath2 ) symmetry is preserved .
the full pseudo - fermion green s function is determined from the dyson equation [ eq : fullgf ] & _
f^-1(i_n)= ^-1- _ f(i_n ) , where the self - energy @xmath90 has to be calculated .
in addition , the lagrange multiplier , @xmath83 , can be determined by employing the occupation constraint for the pseudo - fermions , which yields [ eq : half_sc ] & q_0=f^f , f^f= _
f(=0 ^ - ) . in order to determine @xmath90 in the bose - fermi kondo model we first briefly present the known results for two limiting cases , namely for ( i ) the fermionic kondo , @xmath63 , and ( ii ) the bose kondo model , @xmath91 . for @xmath63 by using a simple decoupling approximation
one can obtain the non - interacting form of the kondo hamiltonian , i.e. , @xmath92 $ ] with the self - consistently determined condensation amplitude @xmath93 .
this scheme , the so - called slave - boson mean - field approximation , becomes exact in the limit @xmath87 when the physics is dominated by a static saddle point .
the known result for the pseudo - fermion self - energy is given by@xcite [ eq : sigmaf_fer ] & _ f^()(i_n)=v^2_c^0(i_n,*r*_0 ) , where @xmath94 $ ] is the local green s function of the fermionic bath ; for @xmath95 one should use instead .
the condensation amplitude , @xmath96 , measures the interaction between the conduction electrons and the impurity spin .
the non - zero value of @xmath96 signals kondo screening whereas for @xmath97 the moment is unscreened ( free ) .
@xmath96 can be determined by calculating the corresponding expectation value of the bosonic operator @xmath98 which is related to the mixed propagator @xmath99 .
hence , [ eq : v_sc ] & v = j_^ c^_0 f , c^_0 f= _ fc(=0 ^ - ) .
a simple diagrammatic analysis leads to [ eq : fullgfc ] & _ fc(i_n)=-v _ f(i_n)_c^0(i_n,*r*_0 ) .
the factor @xmath96 arises due to the fact that no bare mixed propagator exists and the minus sign as a result of the anticommutation relations .
a non - zero condensation amplitude , @xmath96 , at @xmath100 , decreases with increasing temperature and vanishes continuously at the so - called kondo temperature , @xmath101 .
the equations - are a self - consistent system of equations for determining the full @xmath102-propagator , the condensation amplitude , @xmath96 , and the lagrange multiplier , @xmath83 .
these equations can be solved for a given kondo coupling , @xmath41 , and a local conduction green s function at the impurity site . from the full @xmath102-propagator
we can easily obtain the pseudo - fermion spectral function , @xmath103 . a trivial solution of the self - consistent system of equations is given by @xmath97 and @xmath104 , which corresponds to an impurity spin which is completely decoupled from the conduction electrons . at high enough temperature
this is the only solution .
with decreasing temperature another non - trivial solution with @xmath105 can appear .
the non - trivial solution signaling the kondo screening arises only below the kondo temperature and for @xmath65 . note that @xmath101 can be determined for a given @xmath41 using in the limit @xmath106 , @xmath107 ; at zero temperature we use in the same limit to calculate the critical value of the kondo coupling , @xmath108 .
it is known that the slave - boson mean - field method has numerous artifacts at finite temperatures or in the vicinity of the quantum critical point for @xmath109 which is the case of physical interest .
nevertheless , this approach has been extensively used in the context of the pseudogap kondo model@xcite and is known to capture the qualitative physics of the kondo screened phase .
we emphasize that in the four - site kondo model , which we use in the present study , the ldos seen by the impurity is given by the effective propagator @xmath110 $ ] .
the green s function @xmath111 is defined in eq .
( i.e. , if the potential scattering is present , @xmath95 , we assume that it is located at the impurity site , @xmath69 ) and @xmath1121 $ ] for @xmath113 $ ] . for a more detailed discussion
see ref . . for @xmath91
the bose - fermi kondo model reduces to the so - called bose kondo model . in the large-@xmath2 limit
the pseudo - fermion self - energy is given by@xcite [ eq : sigmaf_bos ] & _
f^()(i_n)= -_i_n _ f(i_n+i_n)_^0(i_n,*r*_0 ) , where @xmath114 is the local green s function of the bosonic bath . the self - energy @xmath115 is derived by summing all diagrams with non - crossing bosonic lines ( the so - called rainbow " diagrams ) which is known as the non - crossing approximation ( nca).@xcite the self - consistency is obtained using , , and .
interestingly , in the bose kondo model there is always a non - trivial interaction between the impurity and the bosonic bath , i.e. , @xmath116 .
at zero temperature and at the bulk quantum critical point ( @xmath117 and @xmath118 ) the self - consistent solution yields a power - law behavior for the pseudo - fermion spectral function , @xmath119 .
this behavior does not correspond to the usual kondo screening which occurs in the fermionic kondo model , but rather reflects non - trivial fluctuations that are usually referred to as the bosonic fractional ( fluctuating ) phase . for finite but small @xmath54
there are no bulk spin excitations at low energies .
for @xmath120 the impurity moment is decoupled from the bath while the case @xmath121 is analogous to the case @xmath55 ; the crossover between these two regimes has been studied in detail in ref . .
a generalization of the large-@xmath2 approach for the bose - fermi kondo model can not be done in a straightforward manner and further approximations are necessary .
( we note , however , that the large-@xmath2 analysis has been applied to a _ multi - channel _ version of the bose - fermi kondo model.@xcite ) guided by the fact that the physics of both the fermionic and the bosonic kondo model is controlled by a saddle point in the @xmath87 limit , we combine both effects through the pseudo - fermion self - energy [ eq : sigmaf ] & _ f^(i_n)= _ f^()(i_n)+_f^()(i_n ) .
the fermionic part obtained within the slave - boson mean - field approach and the bosonic nca contribution are only formally decoupled .
their mutual effect can be easily understood through the self - consistency which we present below .
namely , diagrams containing both the conduction electron and the bosonic propagators are included through the full @xmath102-propagator in @xmath115 .
in addition , according to eq .
the mixed propagator , @xmath122 , also contains the full pseudo - fermion propagator .
therefore , the interaction with the bosonic bath entering the mixed propagator through @xmath123 influences the condensation amplitude , i.e. , kondo screening .
the system of equations - can be solved self - consistently in order to obtain the full @xmath102-propagator , @xmath96 , and @xmath83 for given couplings @xmath41 and @xmath42 , and for given local green s functions of the fermionic and the bosonic bath , @xmath124 and @xmath125 , respectively ; the results obtained by numerically solving this self - consistent system are presented sec .
[ sec : kondoresp ] .
before we show our numerical results in the next section , we discuss some limiting cases . for small coupling @xmath41 and the fermionic bath exponent
@xmath126 we expect no kondo screening , @xmath97 , and that the problem reduces to the bose kondo model .
on the other hand , for large @xmath41 and small @xmath42 the pseudo - fermion self - energy is dominated by its fermionic term and we should recover the results of the fermionic pseudogap kondo model .
consequently , the two interactions of the impurity with the fermionic and the bosonic bath ( i.e. , the fermionic and the bosonic kondo physics ) compete ; the same conclusion has been reached within the renormalization group approach.@xcite for @xmath117 there is a quantum phase transition between a fermionic kondo - dominated phase , for @xmath127 , and a bosonic fluctuation - dominated phase , for @xmath128 . note that there is no zero temperature free moment phase as long as @xmath67 and @xmath117 ; this is the so - called bosonic fractional moment phase.@xcite an interesting quantity is @xmath129 where the kondo screening sets in . it can be derived using eq .
provided that @xmath106 and @xmath107 . in our calculation
we determine the critical value of the kondo coupling , @xmath108 , in the @xmath130 limit .
in this section we present our numerical results which are obtained for a system at zero temperature .
first we explain the way in which the ldos of the host sc is modified .
afterwards , we show the results for both ( i ) the pure ps model and ( ii ) the kondo model with and without the potential scattering at the impurity site .
finally , we compare them with the existing stm experimental data .
as already noted recent stm measurements of the ldos in cuprates have led to numerous studies ; they deal with the possible origin of the nanoscale spatial variation@xcite or identify the processes contributing to the tunneling spectra.@xcite in the present calculation we do not attempt to explain the origin of the nanoscale inhomogeneities but we model the host ldos in a way to get close resemblance to the stm data.@xcite the stm experiments have established that the ldos shows a nanoscale spatial modulation with two types of regions:@xcite ( i ) the first type corresponds to domains with small gaps and well - defined coherence peaks . in this region
we assume that the bcs hamiltonian describes the bulk superconductor and we use the conduction electron green s function defined in eq . to obtain the ldos .
( ii ) the second type of domains have large gaps and very broad gap - edge peaks whose amplitude is relatively low . in the following
we will refer to these regions as the pseudogaps " . in order to describe the host superconductor in the pseudogap " domains we use a rather phenomenological approach .
we write the conduction electron green s function as @xmath131^{-1}= [ \underline{\mathcal{g}}_{\,c}^0(i\nu_n,{\bf k})]^{-1}- \underline{\sigma}_{\,c}(i\nu_n,{\bf k})$ ] , where the bare propagator is the propagator in the pure bcs state .
the main approximation enters through the form of the conduction electron self - energy , @xmath132 , which we do not calculate but we employ a simple model for it .
( for a thorough discussion about the conduction electron self - energy we refer the reader to ref . .
) it is important to note that the self - energy is chosen in such a way that the corresponding ldos obtained from the perturbed green s function , @xmath133 , mimics the ldos measured by the stm on bscco.@xcite as already mentioned , the local green s function of the host sc is used further as an input ( i.e. , a given ) quantity for the @xmath1-matrix and the large-@xmath2 calculations .
we assume that the imaginary part of the self - energy is given by @xmath134 for @xmath135 , and @xmath136 otherwise.@xcite the @xmath137-dependence ensures that the low - energy region of the ldos ( e.g. , for @xmath138 ) is not modified in all nanoscale domains which is in accordance with the stm measurements.@xcite the real part of the self - energy is obtained by using the kramers - kronig relation .
in addition , for simplicity we assume that @xmath139 .
-matrix and the large-@xmath2 calculations . by going from a pure bcs state ( denoted by 1 `` ) towards a pseudogap '' state ( denoted by 6 " )
the gap value increases and the superconducting coherence peaks disappear .
the low - energy region remains unchanged . ] the host ldos which we use in the present calculation are shown in fig .
[ fig : ldos ] and are denoted by @xmath140 .
the case 1 " corresponds to the pure bcs state with well - defined coherence peaks and a small gap while in other spectra the gap increases and the coherence peaks disappear . in this way we can model the stm spectra with small ( large ) gap regions with ( without ) sc coherence peaks
furthermore , the stm measurements show that the low - energy region close to the fermi level remains unchanged;@xcite this feature is also captured in fig .
[ fig : ldos ] . note that larger gap values correspond to a lower hole doping level ; we use the gap values @xmath141 for the spectra @xmath140 , respectively . by adjusting the chemical potential @xmath142
one obtains the correct doping ; more precisely we fixed @xmath143(@xmath144,@xmath145,@xmath146,@xmath147,@xmath148,@xmath149)@xmath150mev for the doping levels @xmath151(@xmath152,@xmath153,@xmath154,@xmath155,@xmath156,@xmath157)@xmath158.@xcite it is important to emphasize that the doping level does not significantly change our results and it can be in principle completely ignored .
nevertheless , we include the different dopings in order to model the experimental situation as close as possible .
we start our discussion with the pure ps model ( @xmath159 ) .
it is known that the delta - function potential scatterer generates a resonant state in the superconductor ldos and the resonance can be tuned towards the fermi level by increasing @xmath160 ; the low - energy resonant state is possible only in unitary limit , i.e. , when @xmath160 is the largest energy scale in the problem . however ,
the ps model is not able to explain the spatial distribution of the resonance which is seen in stm experiments.@xcite these results are summarized , for instance , in fig . 1 of ref . . here
, we consider a non - magnetic impurity as a pure potential scatterer placed in different regions of the @xmath0-wave superconductor ; those regions are characterized by the corresponding host ldos ( shown in fig .
[ fig : ldos ] ) . in fig .
[ fig : ps ] we show the ldos in a @xmath0-wave sc at the impurity site obtained within the @xmath1-matrix approach for a given @xmath160 and different host ldos .
the potential scattering strength is chosen such that one gets a resonance in the low - energy region .
ev and different host ldos .
a resonance appears at a low - energy scale @xmath161 ; it is slightly suppressed and moved away from the fermi level by modifying the superconducting host , i.e. , by going from 1 `` to 6 '' in fig .
[ fig : ldos ] . ] modifying the pure bcs state towards the pseudogap " state does not change the resonance significantly .
the resonance is slightly moved away from the fermi level and its amplitude is somewhat suppressed .
we conclude that if the zn resonance in bscco were of the ps type then it would be present in all regions ( both small- and large - gap regions ) .
this result is not surprising if we recall that in the @xmath1-matrix approach only the low - energy part of the host ldos can influence the resonance close to the fermi level . according to the stm experiments
the low - energy part of the host ldos remains unchanged ( as in fig .
[ fig : ldos ] ) so that the zn resonance in the ps model survives in all regions which is in disagreement with the stm data .
note that for larger @xmath160 ( i.e. , for lower energy of the resonant state ) the change in the position and the amplitude of the resonance state is even less pronounced . here
we present numerical results for the bose - fermi kondo model obtained within the large-@xmath2 approach . in fig .
[ fig : rho_f ] the typical impurity spectrum in the kondo screened phase ( @xmath105 ) is shown ; it is obtained by solving the self - consistent system of equations - presented above . in the kondo screened phase the impurity spectral function , @xmath162 ,
develops a pronounced peak at small but finite energy , typically of the order of @xmath163 .
such a peaked feature does not develop in the local moment phase where the kondo screening is completely suppressed and @xmath97 . ,
and the sc host is in the pure bcs state ( 1 " ) .
the impurity spectral function develops a peaked structure at a characteristic kondo energy scale @xmath164 . ] in the present large-@xmath2 analysis @xmath101 and @xmath164 are not equivalent , i.e. , they do not behave in the same manner upon approaching the transition .
nevertheless , @xmath164 can be used to characterize kondo screening since ( i ) the local magnetic susceptibility changes its behavior at @xmath164 and ( ii ) the peaked structure at @xmath164 also appears in the conduction electron @xmath1-matrix as seen both in the large-@xmath2 and the nrg calculations.@xcite ( note , however , that in the nrg calculations the characteristic energy scale , @xmath164 , can be identified with the kondo temperature up to a numerical factor of order unity.@xcite ) in the present work we perform a zero temperature calculation so that @xmath164 serves as a natural choice to characterize the kondo screening .
furthermore , we associate the peaked structure in the impurity spectral function with the zn resonance in the stm data.@xcite in other words , the zn resonance is likely caused by a kondo - like behavior of the induced effective magnetic moment with the kondo temperature as a natural low - energy scale .
as already noted the resonance peak seen by stm does not decay monotonically with distance from the zn site but rather oscillates , producing local minima and maxima in the dos map .
remarkably , the four nearest neighbor cu sites around the impurity have no dos local maxima associated with them ( in contrast to the next - nearest and next - next - nearest neighbor cu sites).@xcite this non - monotonic spatial dependence can not be naturally explained within the ps model . on the other hand , the four - site kondo model for an impurity in a @xmath0-wave superconductor can capture this feature.@xcite here
we analyze how the characteristic energy scale changes with the kondo coupling when the impurity is placed in different regions of the sc host . in fig .
[ fig : tk_jk ] we show how @xmath164 depends on the kondo coupling when a host ldos is modified from the pure bcs to the pseudogap `` case ( the corresponding ldos are shown in fig .
[ fig : ldos ] ) .
we see that a larger value of the kondo coupling is required in order to screen the impurity moment in the large - gap regions ( i.e. , the large - gap regions have larger critical kondo coupling , @xmath165 , below which the moment is free and above which the moment is screened ) .
in particular , for certain values of @xmath166 the screening is possible only in the small - gap regions and not possible in the large - gap regions . in other words ,
the kondo - like impurity resonance is suppressed by going from the pure bcs to the pseudogap '' state ; this is an important result which is supported by the experimental data.@xcite this fact can not be explained within the ps model but naturally arises in the kondo model . ) .
the potential scattering and the bosonic bath are absent : @xmath167 , @xmath63 .
the pure bcs state ( small - gap domain ) has lowest @xmath108 . ]
note that the values of @xmath163 also agree well with the kondo temperatures estimated in nmr experiments @xmath168.@xcite in ref .
it has already been shown that the presence of weak to moderate potential scattering does not change the results of the four - site fermionic kondo model .
in contrast , large values of the scattering potential induce a large dos on the neighboring sites of the zn impurity . as a consequence the critical kondo coupling
is strongly reduced and the kondo scale is increased up to temperatures @xmath169 , which is in disagreement with the nmr data.@xcite on the other hand , the kondo temperature is suppressed in the presence of the collective bosonic modes which are included in the bose - fermi kondo model . in what follows we present the phase diagram of the bose - fermi kondo model obtained using the large-@xmath2 approach and also show the suppression of the kondo screening in the presence of the collective bosonic modes .
the projection of the phase diagram onto the @xmath166-@xmath42 plane is shown in fig .
[ fig : jkgamma0_mass ] where the critical coupling @xmath165 is determined for different values of the spin gap and the host sc in the pure bcs state ( 1 " ) .
for @xmath63 we recover a fermionic pseudogap kondo model with a finite value of @xmath165 which is , as expected , independent of the spin gap ; below @xmath165 the impurity moment is free . for @xmath67 and @xmath170
we distinguish two different cases : ( i ) for the gapless bosonic bath , @xmath55 , the impurity moment is not completely free but rather placed into a bosonic fractional ( fluctuating ) phase.@xcite ( ii ) a coupling to a gapped bosonic bath , @xmath171 , leaves the impurity completely decoupled from both baths .
furthermore , fermionic and bosonic kondo physics compete since the larger @xmath42 values require larger @xmath165 in order to screen the moment .
( note that the same conclusion was reached within the rg approach for the bose - fermi kondo model with @xmath55.@xcite ) also , for larger spin - gap values the bosonic bath is less effective in destroying the fermionic kondo screening .
( this fact was proposed earlier to explain the nmr data showing a suppression of the kondo screening by underdoping.@xcite ) -@xmath42 plane ; the critical coupling @xmath108 is determined for different spin - gap values and the host sc in the pure bcs state ( 1 " ) . for @xmath55 and @xmath63 ( @xmath67 )
there is a transition between the free ( fractional ) moment phase and the kondo screened phase .
for @xmath171 there is only a transition between the screened and the free moment phase .
the potential scattering is absent , @xmath167 . ] in fig .
[ fig : jkmass_gamma0 ] we show the projection of the phase diagram onto the @xmath41-@xmath54 plane .
the critical kondo coupling is determined for different values of the bosonic coupling . for @xmath55 ( @xmath171 )
there is a transition between the fractional ( free ) moment phase for @xmath66 , and the kondo screened phase for @xmath65 . by increasing the spin gap the number of low - energy bosonic excitations is reduced which makes the bosonic bath less effective in destroying the kondo screening and as a consequence @xmath108 decreases .
the bosonic spin gap is doping dependent and decreases with underdoping.@xcite as a consequence , one expects that the bosonic bath stronger suppresses the kondo screening in the region with lower hole doping which corresponds to the region with the larger - gap in ldos .
-@xmath54 plane ; the critical coupling @xmath108 is determined for different values of the bosonic coupling and the host sc in the pure bcs state ( 1 " ) . for @xmath55 ( @xmath171 ) there is a transition between the fractional ( free ) moment phase and the kondo screened phase .
the potential scattering is absent : @xmath167 . ] as mentioned above in the presence of the potential scattering the dos on the neighboring sites of the zn impurity increases which results in a decrease of the critical kondo coupling .
for example , for @xmath172mev and @xmath173mev and the host sc in the region 1 " , we have @xmath174mev for @xmath175ev , respectively . here
we discuss how the characteristic energy scale changes by modifying the superconducting host .
figure [ fig : tk_delta0 ] shows the kondo scale , @xmath164 , as a function of an average energy gap @xmath176 .
the average energy gap is defined as @xmath177 where the quantity @xmath178 ( @xmath179 ) is the energy of the first peak below ( above ) the fermi level in the host ldos without impurity . for
the pure bcs state @xmath176 corresponds to a superconducting gap @xmath180 while in the pseudogap " state one can estimate it from the ldos shown in fig .
[ fig : ldos ] .
mev , @xmath181ev , @xmath182mev , @xmath183mev ( open squares ) ; @xmath184mev , @xmath185ev , @xmath182mev , @xmath173mev ( solid circles ) and @xmath183mev ( open circles ) ; @xmath184mev , @xmath167 , @xmath63 ( solid squares ) ; the lines are guides to the eye . by going from the small - gap region to the large - gap region the kondo screening of the impurity moment vanishes . ]
all sets of data points in fig .
[ fig : tk_delta0 ] indicate that the kondo effect is more suppressed in the regions with a larger energy gap .
furthermore , the suppression of the kondo screening is a monothonic function of the average energy gap and it is , in principle , possible to identify the maximum value of the energy gap above which the impurity resonance can not appear in the ldos ; this agrees with the stm measurements.@xcite ( note that the amplitude of the andreev resonant state in ref . is a non - monotonic function of the energy gap
; the largest resonance amplitude one obtains for intermediate gap values . )
it is important to note that the kondo energy scale @xmath161 can be tuned to zero by simply modifying the host ldos ; we have shown that this is not possible in the ps model ( fig .
[ fig : ps ] ) . on the other hand , in the kondo model the low - energy region of the bath ldos is important for the kondo effect , but through the self - consistent procedure also the ldos away from the fermi level influences the kondo screening .
since the high - energy part of the bath ldos differs significantly in small- and large - gap regions , it is natural to expect that the kondo resonance will be affected when the host ldos is modified .
figure [ fig : tk_delta0 ] also suggests that the presence of the bosonic bath additionally suppresses the kondo screening . for the spin gap , @xmath54
, we use a constant value regardless of the specific host ldos , i.e. , regardless of doping level .
we could , however , associate different spin gaps to different host ldos since the spin gap is doping dependent;@xcite the underdoped regions ( i.e. , the regions with a larger energy gap @xmath176 ) have smaller @xmath54 . in this case , the suppression of @xmath164 with @xmath176 would be even more pronounced .
motivated by the question about the origin of the impurity resonance in @xmath0-wave sc we have compared two frequently discussed scenarios , i.e. , the potential scattering model and the kondo model .
we have investigated how the stm impurity resonant state is influenced when the sc host is chosen in such a way that it resembles the nanoscale electronic inhomogeneities seen by stm.@xcite ( i ) if the impurity atom is assumed to be a pure potential scatterer we have shown within the @xmath1-matrix approach that the impurity resonance appears as a robust feature .
the resonant state is rather insensitive to changes of the sc host .
it is present in both small- and large - gap regions which is in disagreement with the experimental data .
( ii ) the suppression of the impurity resonance in the large - gap domains observed by stm can be naturally explained by assuming that the impurity atom gives rise to kondo - type physics .
we have used the bose - fermi kondo model and the large-@xmath2 method to show that the characteristic kondo energy scale is affected by changing the host properties ( i.e. , by changing the corresponding host ldos ) .
the impurity resonance resulting from the kondo screening can appear in the host small - gap regions and can be completely suppressed in the large - gap regions ; this is in agreement with the stm experimental data for zn - doped bi@xmath3sr@xmath3cacu@xmath3o@xmath4 .
therefore , we conclude that the kondo spin dynamics of the impurity moment is the origin of the impurity resonant state and that the pure potential scattering model is not sufficient to capture the physics of the non - magnetic impurity in @xmath0-wave superconductors .
q. si , s. rabello , k. ingersent , and j. l. smith , nature * 413 * , 804 ( 2001 ) ; m. t. glossop and k. ingersent , preprint cond - mat/0607566 ; j .- x .
zhu , s. kirchner , r. bulla , and q. si , preprint cond - mat/0607567 . in principle , we could introduce a @xmath16-dependent factor in the expression for the self - energy to account for the difference in the nodal and antinodal regions ; a suitable choice would be @xmath186 or @xmath187 .
however , after the momentum summation this change would not affect the low - energy local dos because the @xmath137-dependence in the self - energy is subdominant in comparison with the linearly vanishing dos for the bare @xmath0-wave superconductor .
experimentally it is not possible to determine the local hole doping and associate it with the certain gap value.@xcite nevertheless , one can measure the bulk hole doping and then associate it with the spatially averaged gap value , @xmath188 . | we investigate the impurity resonant state induced by non - magnetic impurities in @xmath0-wave cuprate superconductors in two different impurity models : ( i ) in a pure potential scattering model within the @xmath1-matrix approach and ( ii ) in a bose - fermi kondo model within the large-@xmath2 formalism .
we modify the superconducting host to resemble the nanoscale electronic inhomogeneities seen in the scanning tunneling microscopy ( stm ) spectra ( i.e. , the small- and large - gap domains ) and study how this influences the impurity resonance . in the pure potential scattering model the resonant state is found to be a rather robust feature appearing in all host domains . on the other hand , in the kondo - like model the impurity resonance can appear in the small - gap regions of the host and be completely suppressed in the large - gap regions ;
this is in agreement with the stm experimental data for zn - doped bi@xmath3sr@xmath3cacu@xmath3o@xmath4 .
our results imply that the kondo spin dynamics of the impurity moment is the origin of the impurity resonant state in @xmath0-wave superconductors , rather than the pure potential scattering . |
consider the directional hilbert transform in the plane defined for a fixed direction @xmath1 as h_v f(x , y):=p.v .
_ f(x -t , y- u t)t for any test function @xmath2
. by the dilation symmetry , the length of the vector @xmath3 is irrelevant for the value of @xmath4 , which explains our normalization of the first component . by an application of fubini s theorem and the @xmath5 bounds for
the classical hilbert transform one obtains a priori @xmath5 bounds for @xmath4 . on the other hand ,
the corresponding maximal operator @xmath6 for varying directions is well known to not satisfy any a priori @xmath5 bounds .
+ in @xcite , bateman and thiele proved that @xmath7 for the range @xmath8 .
this result is half of a maximal bound since the supremum is squeezed between the two @xmath5 norms on the left hand side , compared to being completely inside or outside as in our previous remarks .
the case @xmath9 of ( [ bt ] ) goes back to coifman and el kohen ( see @xcite for the detailed discussion ) , who noticed that a fourier transform in the @xmath10 direction makes @xmath11 for @xmath9 equivalent to @xmath0 bounds for carleson s operator .
estimate ( [ bt ] ) highlights a bi - parameter structure of the directional hilbert transform .
the bi - parameter structure arises since the kernel is a tensor product between a hilbert kernel in direction v and a dirac delta distribution in the perpendicular direction .
if one considers the linearized maximal operator @xmath12 for some function @xmath13 , then inequality ( [ bt ] ) can be rephrased as a bound for the linearized maximal operator under the assumption that @xmath14 is constant on every vertical line @xmath15 for all @xmath16 .
such vector fields @xmath3 of the form @xmath17 for some measurable function @xmath18 are call one - variable vector fields in @xcite
. + the purpose of the present paper is to relax this rigid assumption on @xmath14 , and prove an analogue of ( [ bt ] ) for vector fields which are constant along suitable families of lipschitz curves .
to formulate such a result , we perturb ( [ bt ] ) by a bi - lipschitz horizontal distortion , that is @xmath19 with @xmath20 for every @xmath21 and every @xmath10 , such that the transformation ( [ bilip ] ) maps vertical lines to near vertical lipschitz curves : latexmath:[\[\label{liplines } two conditions can be rephrased as 1/a_0_1ga_0 |_2g|b_0 under these assumptions , @xmath5 norms are distorted boundedly under the transformation ( [ bilip ] ) .
namely , ( [ hordis ] ) implies for every @xmath10 that [ mm1.8 ] a_0 ^ -1f(x , y)_l^p(x)^p f(g(x , y),y)_l^p(x)^pa_0 f(x , y)_l^p(x)^p and we may integrate this in @xmath10 direction to obtain equivalence of @xmath5 norms in the plane .
hence the change of measure is not the main point of the following theorem , but rather the effect of the transformation on the linearizing function @xmath14 , which is now constant along the family of lipschitz curves which are the images of the lines @xmath15 under the map ( [ bilip ] ) .
let @xmath23 satisfy assumption ( [ hordis ] ) for some @xmath24 and assumption ( [ liplines ] ) for some @xmath25 .
then for any @xmath26 , we have @xmath27 here @xmath28 is a constant depending only on @xmath24 and @xmath29 .
the constant @xmath28 is independent of @xmath25 due to the unisotropic scaling symmetry @xmath30 . in view of the implicit function theorem ( see @xcite for recent developments ) , our result covers a large class of vector fields which are of the critical lipschitz regularity .
indeed , it implies the following [ corollary1.2 ] for a measurable unit vector field @xmath31 , suppose that \i ) there exists a bi - lipschitz map @xmath32 s.t .
[ e1.8 ] v_0(g_0(x , y ) ) y ; \ii ) there exists @xmath33 s.t .
@xmath34 , [ scaling ] ( _ 2 g_0(x , y ) , v_0(g_0(x , y ) ) ) d_0 ya.e . .
then the associated hilbert transform , which is defined as h_v_0f(x , y):=_f((x , y)-t v_0(x , y))dt / t , is bounded in @xmath0 , with the operator norm depending only on @xmath35 and the bi - lipschitz norm of @xmath36 .
[ nnremark1.4 ] the structure theorem for lipschitz functions by azzam and schul in @xcite states exactly that any lipschitz function @xmath37 ( any lipschitz unit vector field @xmath38 in our case ) can be precomposed with a bi - lipschitz function @xmath39 such that @xmath40 is lipschitz in the first coordinate and constant in the second coordinate , when restricted to a `` large '' portion of the domain . without the assumption that @xmath33 , the operator @xmath41 might be unbounded in @xmath5 for any @xmath42 .
the counter - example is based on the besicovitch - kakeya set construction , which will be discussed at the end of the proof of the corollary . to our knowledge , this is the first result in the context of the directional hilbert transform with a lipschitz assumption in the hypothesis .
lipschitz regularity is critical for the directional hilbert transform as we will elaborate shortly . to use the assumption that @xmath3 is constant along lipschitz curves
, we apply an adapted littlewood - paley theory along the level lines of @xmath3 .
this is a refinement of the analysis of coifman and el kohen who use a fourier transform in the @xmath10 variable and the analysis of bateman and thiele who use a classical littlewood - paley theory in the @xmath10 variable .
this adapted littlewood - paley theory is the main novelty of the present paper .
it is in the spririt of prior work on the cauchy integral on lipschitz curves , for example @xcite , but it differs from this classical theme in that it is more of bi - parameter type as it is governed by a whole fibration into lipschitz curves .
we crucially use jones beta numbers as a tool to control the adapted littlewood - paley theory . to our knowledge
this is also the first use of jones beta numbers in the context of the directional hilbert transform .
+ in this paper we focus on the case @xmath0 , since our goal here is to highlight the use of the adapted littlewood - paley theory and jones beta numbers in the technically most simple case .
we expect to address the more general case @xmath5 with a range of @xmath43 as in the bateman - thiele theorem in forthcoming work . while coifman and el kohen use the difficult bounds on carleson s operator as a black box , bateman and thiele have to unravel this black box following the work of lacey and li @xcite-@xcite and use time - frequency analysis to prove bounds for a suitable generalization of carleson s operator . luckily , in the present work we do not have to delve into time - frequency analysis as we can largely recycle the work of bateman and thiele for this aspect of the argument .
an upper bound such as @xmath44 is necessary in our theorem . by a limiting argument
we may recover the theorem of bateman and thiele , using the scaling to tighten the lipschitz constant @xmath25 at the same time as relaxing the condition @xmath44 .
an interesting open question remains whether the same holds true for @xmath45 .
we do not know of a soft argument to achieve this relaxation , as the norm bounds arising from our argument grow unlimited as @xmath29 approaches @xmath46 .
this question suggests itself for further study .
+ part of our motivation is a long history of studies of the linearized maximal operator ( [ linmax ] ) under various assumptions on the linearizing function @xmath14 .
if one truncates @xmath47 as @xmath48 then it is reasonable to ask for pure regularity assumptions on @xmath14 to obtain boundedness of @xmath49 .
it is known that lipschitz regularity of @xmath14 is critical , since a counterexample in @xcite based on a construction of the besicovitch - kakeya set shows that no bounds are possible for @xmath50 regularity with @xmath51 .
however , it remains open whether lipschitz regularity suffices for bounds for @xmath49 . on the regularity scale ,
the only known result is for real analytic vector fields @xmath3 by stein and street in @xcite .
a prior partial result in this direction appears in @xcite .
it is our hope that our result corners some of the difficulties of approaching lipschitz regularity in the classical problem .
further substantial progress ( including the case @xmath45 ) is likely to use lipschitz regularity not only of the level curves of @xmath14 but also of @xmath14 itself across the level curves .
for example , one possibility would be to cut the plane into different pieces by the theorem of azzam and schul stated in remark [ nnremark1.4 ] , and to analyze each piece separately by the main theorem in the present paper .
we leave this for the future study .
+ related to the directional hilbert transform and thus additional motivation for the present work is the directional maximal operator @xmath52 which arises for example in lebesgue type differentiation questions and has an even longer history of interest than the directional hilbert transform .
hilbert transforms and maximal operators share many features , in particular they have the same scaling and thus share the same potential @xmath5 bounds .
the maximal operator is in some ways easier as it is positive and does not have a singular kernel .
for example , bounds for the maximal operator under the assumption of real analytic vector fields were proved much earlier by bourgain in @xcite .
an instance of bounds satisfied by the maximal operator but not the hilbert transform arises when one restricts the range of the function @xmath14 instead of the regularity . for certain sets of directions characterized by bateman in @xcite
there are bounds for the maximal operator ( for example for the set of lacunary directions ) , while karagulyan proves in @xcite that no such bounds are possible for the hilbert transform .
on the other hand , the hilbert transform is easier in some other aspects , most notably it is a linear operator .
for example , bounds for the bilinear hilbert transform mapping into @xmath53 were known ( lacey and thiele @xcite , @xcite ) before the corresponding maximal operator bounds ( lacey @xcite ) , due to the fact that orthogonality between different tiles is preserved under the hilbert transform but not the maximal operator
. in particular we do not know at the moment whether the analogue of our main theorem holds for the directional maximal operator .
this may be an interesting subject for further investigation .
+ * outline of paper : * in section 2 we will prove corollary [ corollary1.2 ] by reducing it to the main theorem .
the reduction will also be used later in the proof of the main theorem . in section 3
we will state the strategy of the proof for the main theorem . as it appears that our result is a lipschitz perturbation of the one by bateman and thiele , this turns out also to be the case for the proof :
if we denote by @xmath54 a littlewood - paley operator in the @xmath10-variable , the main observation in bateman and thiele s proof is that @xmath4 commutes with @xmath54 . in our case , this is no longer true .
however , we can make use of an adapted version of the littlewood - paley projection operator @xmath55 ( see definition [ definition1 ] ) to partially recover the orthogonality .
we split the operator @xmath4 into a main term and a commutator term _
k h_v p_k(f)=_k(h_v p_k(f)-_k h_v p_k(f)+_k h_v p_k(f ) ) .
the boundedness of the main term @xmath56 is essentially due to lacey and li @xcite , with conditionality on certain maximal operator estimate . in section 4
we modify bateman s argument in @xcite and @xcite to the case of vector fields constant on lipschitz curves and remove the conditionality on that maximal operator .
the main novelty is the boundedness of the commutator term _
k(h_v p_k(f)-_k h_v p_k(f ) ) , which will be presented in section 5 . to achieve this , we will view lipschitz curves as perturbations of straight lines and use jones beta number condition for lipschitz curves and the carleson embedding theorem to control the commutator . here
we shall emphasis again that the commutator estimate is free of time - frequency analysis .
+ * notations : * throughout this paper , we will write @xmath57 to mean that @xmath58 , @xmath59 to mean that there exists a universal constant @xmath28 s.t . @xmath60 , and @xmath61 to mean that @xmath59 and @xmath62 .
@xmath63 will always denote the characteristic function of the set @xmath64 . +
* acknowledgements .
* this work is done under the supervision of prof .
christoph thiele , to whom the author would like to express his most sincere gratitude , for suggesting him such an interesting topic , for sharing with him his deep insight into this problem , and for numerous suggestions on the exposition of this paper .
in this section we prove corollary [ corollary1.2 ] , by reducing it to the main theorem .
some parts of the reduction will also be used in the proof of the main theorem in the rest of the paper .
+ we first divide the unit circle @xmath65 into @xmath66 arcs of equal length , with the angle of each arc being @xmath67 .
choose n > 6/d_0 s.t . @xmath68 .
denote these arcs as @xmath69 .
for each @xmath70 , define @xmath71 if we were able to prove that @xmath72 is bounded by a constant @xmath28 which is independent of @xmath73 , then we conclude that h_v_0_22c n(d_0 )
. now fix one @xmath70 , we want to show the boundedness of @xmath74 .
choose a new coordinate such that the @xmath75-axis passes through @xmath70 and bisects it .
then all the vectors in @xmath70 form an angle less than @xmath76 with the @xmath75-axis . as we assume that ( _ 2 g_0 , v_0(g_0))d_0>0 , we see that the vector @xmath77 forms an angle less than @xmath78 with the @xmath10-axis .
renormalize the unit vector @xmath38 such that the first component is 1 , i.e. write @xmath79 , then by the fact that @xmath38 forms an angle less than @xmath76 with the @xmath75-axis , we obtain |u_0|(d_0/6 ) .
next we construct the lipschitz function @xmath80 in the main theorem from the bi - lipschitz map @xmath36 , and the coordinate we will use here is still the one associated to @xmath70 as above . under this linear change of variables , we know that @xmath36 is still bi - lipschitz .
we renormalize the bi - lipschitz map in such a way that g_0(x,0)=(x,0 ) , x. fix @xmath81 , the map @xmath36 , when restricted on the vertical line @xmath82 , is still bi - lipschitz .
we denote by @xmath83 the image of this bi - lipschitz map , i.e. _ x:=\{g_0(x , y ) : y}. define the function @xmath80 by the following relation ( g(x , y),y)=g_0(x , y ) , for some @xmath84 . by the fact that @xmath36 is bi - lipschitz , we know that such @xmath84 exists and is unique . + from the above construction and the fact that @xmath77 forms an angle less than @xmath78 with the @xmath10-axis
, we see easily that [ ee2.9 ] |g(x , y_1)-g(x , y_2)|(d_0/2 ) |y_1-y_2| , x , y_1 , y_2 . hence what is left is to show that condition is also satisfied with a constant @xmath24 depending only on @xmath35 and the bi - lipschitz constant of @xmath36 .
one side of the equivalence @xmath85 is quite clear from the picture below : the bi - lipschitz map @xmath36 sends the points @xmath86 to @xmath87 separately , then by definition of bi - lipschitz map , there exists constant @xmath24 s.t .
g(x_1 , y)-g(x_2 , y)|p - q|(x_1-x_2 ) . for the other side
, we argue by contradiction .
if for any @xmath88 large , there exists @xmath89 s.t .
g(x_1 , y)-g(x_2 , y)m(x_1-x_2 ) , then together with , this implies that dist(k , _
x_1)m(d_0/2)(x_1-x_2 ) . but this is not allowed as by the definition of the bi - lipschitz map @xmath36 and the lipschitz function @xmath80 , @xmath90 must be comparable to @xmath91 .
so far , we have verified all the conditions in the main theorem with b_0=(d_0/2 ) c_0=(d_0/6)/(d_0/2)<1 .
hence we can apply the main theorem to obtain the boundedness of @xmath92 .
+ in the end , as claimed in the corollary , we still need to show the blowing up of the operator norm without the assumption that @xmath33 .
indeed , the blowing up happens in @xmath5 not only for @xmath9 but also for all @xmath42 . for the range @xmath93 ,
the counter example is simply a knapp example : take the function @xmath94 , let @xmath95 be the upper cone which forms an angle less than @xmath96 with the vertical axis .
first define the vector field @xmath97 for @xmath98 , then extend the definition to the whole plane properly such that @xmath3 satisfies the condition .
it is then easy to see that |h_v f(x)|~ , xb_1(0 ) , which does not belong to @xmath99 for @xmath100 for the range @xmath101 , the counter example is given by the standard besicovitch - kakeya set construction , which can be found in @xcite and @xcite .
if we linearize the maximal operator in the main theorem , what we need to prove turns to be the following [ mm3.1 ] _
f(g(x , y)-t , y - t u(x))dt
/ t_2f_2 , where @xmath18 is a measurable function such that @xmath102 .
the change of coordinates ( x , y)(g(x , y ) , y ) in also changes the measure on the plane .
however , we still want to use the original lebesgue measure for the littlewood - paley decomposition .
hence we invert and denote the inversion by [ mm3.3 ] ( x , y)(p(x , y ) , y ) , where `` p '' stands for `` projection '' , and the reason of calling so can be illustrated by the following picture : the change of coordinates in turns the estimate into the following equivalent form _ f(x - t , y - t u(p(x , y)))dt / t_2 f_2 .
moreover , we will denote [ mm3.5 ] h_v f(x , y):=_f(x - t , y - t u(p(x , y)))dt / t . in the rest of the paper , we want to make the convention that whenever @xmath4 appears , it denotes the hilbert transform along the vector field @xmath103 , which is the above , just to distinct it from the various @xmath4 that have appeared in the introduction .
+ to prove the above estimate , we first make several reductions : by the unisotropic scaling xx , yy , we can w.l.o.g .
assume that @xmath104 . by a similar cutting and pasting argument to that in the proof of corollary [ corollary1.2 ]
, we can assume that @xmath105 , i.e. the vector field @xmath3 is of the form @xmath106 with @xmath107 .
+ now we start the detail of the proof .
it was already observed in bateman @xmath108 that under the assumption @xmath107 , we can w.l.o.g .
assume that supp @xmath109 lies in a two - ended cone which forms an angle less than @xmath110 with the vertical axis , as for functions @xmath2 with frequency supported on the cone near the horizontal axis , we have that h_vf(x , y)=h_(1 , 0)f(x , y ) , which is the hilbert transform along the constant vector field @xmath111 .
but @xmath112 is bounded by fubini s theorem and the @xmath0 boundedness of the hilbert transform . for the frequencies outside the cone near the horizontal axis , the proof consists of two steps .
in the first step we will prove the boundedness of @xmath4 when acting on functions with frequency supported in one single annulus . to be precise ,
let @xmath95 be the cone which forms an angle less than @xmath110 with the vertical axis , @xmath113 be the projection operator on @xmath95 , i.e. _ f:=^-1 _ f , where @xmath114 stands for the fourier transform and @xmath115 the inverse transform .
let @xmath54 be the @xmath116-th littlewood - paley projection operator in the vertical direction , as we are always concerned with the frequency in @xmath95 , later for simplicity we will just write @xmath54 instead of @xmath117 for short
. then what we will prove first is [ thm1 ] under the same assumptions as in the main theorem , we have for @xmath118 that h_v p_k ( f)_pp_k(f)_p , with the constant being independent of @xmath119 . in order to prove the boundedness of @xmath4
, we need to put all the frequency pieces together . in the case of @xmath120 vector fields for any @xmath121 , lacey and li s idea in @xcite
is to prove the almost orthogonality between different frequency annuli . in the case where the vector field is constant along vertical lines ,
an important observation in the paper of bateman and thiele is that @xmath4 and @xmath54 commute , which then makes it possible to apply a littlewood - paley square function estimate . in our case bateman and
thiele s observation is no longer true .
we need to take into account that the vector field is constant along lipschitz curves , which gives rise to an adapted littlewood - paley projection operator ( the following definition [ definition1 ] ) . before defining this operator ,
we first need to make some preparation .
fix one @xmath122 , take the curve @xmath123 which passes through @xmath124 , recall that @xmath123 is given by the set @xmath125 , where @xmath80 is the lipschitz function in the main theorem . by the definition of the operator @xmath4 we know that the vector field @xmath3 is equal to the constant vector @xmath126 along @xmath123 . change the coordinate s.t .
the horizontal @xmath127-axis is parallel to @xmath126 .
the following lemma says that in the new coordinate , the curve @xmath123 can still be realized as the graph of a lipschitz function .
[ lem3.2 ] for any fixed @xmath122 , there exists a lipschitz function @xmath128 s.t .
@xmath123 can be reparametrized as @xmath129 .
moreover , we have that @xmath130 , where @xmath25 is the constant in the main theorem
. * proof of lemma [ lem3.2 ] : * denote by @xmath131 the angle between the vector @xmath126 and the @xmath75-axis as in the picture below .
+ the new coordinate of the point @xmath132 will be given by [ e3.4 ] ( x^ , y^)= ( + g ( , ) , -g ( , ) ) .
look at the identity for the second component y^= -g ( , ) , we want to solve @xmath133 by @xmath134 by using the implicit function theorem .
as = - , by the fact that @xmath107 and @xmath135 , we obtain that , from which it is clear that the implicit function theorem is applicable . after solving @xmath133 by @xmath134 , we just need to substitute @xmath133 into the identity for the first component in , which is x^= + g
( , ) , to get an implicit expression of @xmath127 in terms of @xmath134 , which we will denote as @xmath128 . to estimate the lipschitz norm of the function @xmath136 , we just need to observe that when doing the above change of variables , we have rotated the axis by an angle @xmath131 which satisfies @xmath137 . together with the fact that @xmath138
, we can then derive that || , which finishes the proof of lemma [ lem3.2 ] .
@xmath139 ( adapted littlewood - paley projection)[definition1 ] select a schwartz function @xmath140 with support on @xmath141\cup [ -\frac{5}{2},-\frac{1}{2}]$ ] such that _ k _ 0(2 ^ -kt)=1 , t0 . for @xmath142 , for every fixed @xmath122 , define the adapted ( one dimensional ) littlewood - paley projection on @xmath123 by _ k(f)(x^ , y^):=_f(g_(z ) , z ) _ k(y^-z)dz = p_k()(y^ ) , where @xmath143 denotes one point in @xmath123 , @xmath144 and we use @xmath145 to denote the function @xmath146 , and @xmath54 the one dimensional littlewood - paley projection operator .
now it is instructive to regard the lipschitz curves as perturbation of the straight lines , or equivalently , to think that @xmath147 still has frequency supported near the @xmath116-th frequency band , which has already been used by lacey and li in their almost orthogonality estimate for @xmath120 vector fields in @xcite .
we then subtract the term @xmath148 from @xmath149 , and estimate the commutator . to be precise ,
we first write [ e3.5 ] _ k h_v p_k(f)=_k(h_v p_k(f)-_k h_v p_k(f)+_k h_v p_k(f ) ) , then by the triangle inequality , we have _ k h_v p_k(f)_2_k(h_v p_k(f)-_k h_v
p_k(f))_2+_k_k h_v p_k(f)_2 .
we call the second term the main term , and the first term the commutator term
. the @xmath0 boundedness of the main term will follow from orthogonality argument , which is the following adapted littlewood - paley theorem .
[ lemma3 ] for @xmath150 , we have the following variants of the littlewood - paley estimates : [ ee3.9 ] ( _ k|_k(f)|^2)^1/2_p~f_p , [ e3.10 ] ( _ k|_k^*(f)|^2)^1/2_p~f_p , with constants depending only on @xmath24 . * proof of lemma [ lemma3 ] : * in from the introduction , we have already explained the following coarea formula : [ e3.11 ] _
^2|f(x , y)|dxdy~ _ [ _ _ [ e3.9 ] ( _ k|_k(f)|^2)^1/2_p^p ~ _ _ _ ( _ k |_k ( f)|^2)^p/2ds _ d. for every fixed @xmath151 , by definition [ definition1 ] , the right hand side of turns to _ [ _ ( _ k |p_k(_)(y^)|^2)^p/2dy^]d , where @xmath152
. then the classical littlewood - paley theory applies and we can bound the last expression by _
f_l^p(_)^p d f_l^p^p . for the boundedness of the adjoint operator
, it suffices to prove that [ e3.12 ] _ k _ k^*(f ) , f_kf_l^p ( _ k |f_k|^2)^1/2_l^p^. first by linearity and hlder s inequality , we derive _ k _ k^*(f ) , f_k = f , _ k _ k ( f_k)f_l^p _ k _ k ( f_k)_l^p^. applying the coarea formula , we obtain _ k _ k ( f_k)_l^p^~(_(__|_k_k(f_k)|^pds_)d)^1/p. by the definition [ definition1 ] , for every fixed @xmath122 , the inner integration in the last expression turns to [ e3.25 ] _
|_kp_k(_k , ) ( y^)|^pdy^ , where @xmath153 .
now the classical littlewood - paley theory applies and we bound the term in by [ e3.26 ] _ ( _ k|_k , ( y^)|^2)^p/2dy^__(_k|f_k|^2)^p/2ds_(_k|f_k|^2)^1/2_l^p(_)^p. then to prove , we just need to integrate @xmath154 in and apply the coarea formula to derive @xmath155 thus we have finished the proof of lemma [ lemma3].@xmath139 + now we will show how to prove the @xmath0 boundedness of the main term by lemma [ lemma3 ] and proposition [ thm1 ] : first by duality , we have @xmath156 applying the cauchy - schwartz inequality and hlder s inequality , we can bound the last term by _
g_2=1 ( _ k|h_v p_k(f)|^2)^1/2_2 ( _ k @xmath157 for the latter term , lemma [ lemma3 ] implies that ( _ k |_k^*(g)|^2)^1/2_2 g_2 .
thus we have proved the @xmath0 boundedness the main term , modulo proposition [ thm1 ] .
+ as the second step , we will prove the @xmath0 boundedness of the commutator , which is _
k(h_v p_k(f)-_k h_v p_k(f))_2 f_2 .
to do this , we first split the operator @xmath4 into a dyadic sum : select a schwartz function @xmath140 such that @xmath140 is supported on @xmath141 $ ] , let _
l(t):=_0(2 ^ -l t ) , by choosing @xmath140 properly , we can construct a partition of unity for @xmath158 , i.e. _ ( 0,)=_l _ l. let [ e4.24 ] h_lh(x , y):= _ l(t)h(x - t , y - tu(p(x , y)))dt , then the operator @xmath4 can be decomposed into the sum h_v=-+ 2 _ l h_l . hence to bound the commutator , it is equivalent to bound the following _ k _ l(h_l p_k f- _ k h_l p_k
f ) . notice that by definition , @xmath159 vanishes for @xmath160 , which simplifies the last expression to _ l0 _ k ( h_k - l p_k f- _ k h_k - l p_k f ) . by the triangle inequality , it suffices to prove [ thm2 ] under the same assumption as in the main theorem , there exists @xmath161 such that _ k ( h_k - l p_k f- _ k h_k - l p_k f)_22 ^ -lf_2 , with the constant independent of @xmath162 so far , we have reduced the proof of the main theorem to that of proposition [ thm1 ] and proposition [ thm2 ] , which we will present separately in the following sections .
lacey and li in their prominent work @xcite have reduced the @xmath0 boundedness of the operator @xmath163 to the boundedness of one new operator they introduced , which is the so called lipschitz - kakeya maximal operator .
as soon as this operator is bounded , we can then repeat the argument in @xcite to obtain proposition [ thm1 ] as a corollary . here
we follow @xcite , where a slightly different version of the lipschitz - kakeya maximal operator is used , see the following lemma [ lem4.8 ] .
the only place in @xcite where the one - variable vector field plays a special role is lemma 6.2 in page 1037 .
hence to prove proposition [ thm1 ] , we just need to replace this lemma by lemma [ lem4.8 ] , and leave the rest of the argument unchanged .
+ in this section we make an observation that both the boundedness of the lipschitz - kakeya maximal operator ( corollary [ lemma1 ] ) and its variant ( lemma [ lem4.8 ] ) can be proved by adapting bateman s argument in @xcite to our case where the vector fields are constant only on lipschitz curves . before defining the lipschitz - kakeya maximal operator ,
we first need to introduce several definitions .
( popularity ) for a rectangle @xmath164 , with @xmath165 its length , @xmath166 its width , we define its uncertainty interval @xmath167 to be the interval of width @xmath168 and centered at slope(@xmath169 ) .
then the popularity of the rectangle @xmath169 is defined to be pop_r:=|\{(x , y)^2 : u(p(x , y))ex(r)}|/|r| . given two rectangles
@xmath170 and @xmath171 in @xmath172 , we write @xmath173 whenever @xmath174 and @xmath175 , where @xmath28 is some properly chosed large constant , and @xmath176 is the rectangle with the same center as @xmath171 but dilated by the factor @xmath28 .
denote @xmath177 , pop_r\ge \delta , w(r)=\omega\}$ ] , where @xmath178 is the collection of all the rectangles in @xmath172
. then the lipschitz - kakeya maximal function is defined as [ maxoperator ] m _ _ , ( f)(x):=_xr _ , _
r |f| [ lem4.8 ] let @xmath14 and @xmath179 be the functions given in the definition of the operator @xmath4 in .
suppose @xmath180 is a collection of pairwise incomparable ( under `` @xmath181 '' ) rectangles of uniform width such that for each @xmath182 , we have , ( pop_r ) and _ r _ f .
then for each @xmath42 ,
_ r_0|r| .
the same covering lemma argument as in lemma 3.1 @xcite shows the boundedness of lacey and li s lipschitz - kakeya maximal operator as a corollary of lemma [ lem4.8 ] .
[ lemma1 ] for all @xmath183 we have the following bound m _ _ , _
l^pl^pc(p , a_0 ) * proof of lemma [ lem4.8 ] : * the proof is essentially due to bateman @xcite . most of the argument in @xcite remains , with just one minor modification in order to adapt to the family of lipschitz curves on with the vector field is constant .
[ definition4](rectangles adapted to the vector field ) for a rectangle @xmath184 , with its two long sides lying on the parallel lines @xmath185 and @xmath186 for some @xmath187 $ ] and @xmath188 , define @xmath189 to be the adapted version of @xmath169 , which is given by the set \{(x , y ) : p(x , y)p(r)}\{(x , kx+b):x , b } , where @xmath179 is the projection operator in .
what we need to do is just to replace the rectangles @xmath169 in @xcite by @xmath189 , and observe that the two key quantities length and popularity of rectangles are both preserved under the projection operator @xmath179 up to a constant depending on the constant @xmath24 in the main theorem .
hence we leave out the details and refer to @xcite.@xmath139
this section consists of three subsections . in the first subsection we will introduce some notations ,
most of which we adopt from bateman s paper @xcite , with minor changes for our purpose . in the second
we will use jones beta numbers and the carleson embedding theorem to prove proposition [ thm2 ] , modulo one crucial lemma which will be presented afterwards in the third subsection .
the content of this subsection is basically taken from bateman s paper @xcite , with minor changes as we are now dealing with all frequencies instead of one single frequency annulus . +
* discretizing the functions : * fix @xmath190 , we write @xmath191 as the collection of the dyadic intervals of length @xmath192 contained in @xmath193 $ ] .
fix a smooth positive function @xmath194 s.t .
( x)=1 , |x|1 ; ( x)=0 , |x|2 .
also choose @xmath195 such that @xmath196 is a smooth function .
then fix an integer @xmath197(whose exact value is unimportant ) , for each @xmath198 , define _
( x)= ( 2^l+c(x - c__1 ) ) , where @xmath199 is the right half of @xmath200 and @xmath201 is its center .
define _ l(x)=__l _ ( x ) , note that _
l(x+2 ^ -l)=_l(x ) , x. define _ l=_-1 ^ 1 _ l(x+t)dt , because of the above periodicity , we know that @xmath202 is constant for @xmath203 $ ] , independent of @xmath204 .
say @xmath205 , hence _
l(x ) _ [ -1,1](x)=_[-1,1](x ) . define another multiplier @xmath206 with support in @xmath207 $ ] and @xmath208 for @xmath209 $ ] .
we define the corresponding multiplier on @xmath172 : @xmath210 then what we need to bound can be written as @xmath211 where the terms @xmath212 for @xmath160 in the sum vanish as explained before .
* constructing the tiles : * for each @xmath119 and @xmath198 with @xmath190 , let @xmath215 be a partition of @xmath172 by rectangles of width @xmath216 and length @xmath217 , whose long side has slope @xmath131 , where @xmath218 , which is the center of the interval @xmath200 . if @xmath219 , we will write @xmath220 , and @xmath221 to be the right half of @xmath200 , @xmath222 the left half . this subsection is devoted to the proof of proposition [ thm2 ] , which is motivated in large by the proof of the @xmath234 theorem and the boundedness of the paraproduct , see @xcite and @xcite for example . in our case
, unlike bateman and thiele s proof for the one - variable vector fields , it s no longer true that @xmath147 still has frequency in the @xmath116-th annulus .
in order to get enough orthogonality for the term @xmath147 to apply the littlewood - paley theory , we need to subtract the term @xmath235 , which should be viewed as a family of paraproducts .
+ we proceed with the details of the proof .
if we expand the summation on the left hand side of proposition [ thm2 ] with , what we need to bound can be rewritten as _
k _ _ l_s_k , f , _ s(_s-_k
_ s)_22 ^ -lf_2 .
in order to use the orthogonality of different wave packets , we will prove the @xmath0 bound for the dual operator , which is _ k _ _ l_s_k , h , _ s-_k _
notice that for @xmath236 and @xmath237 with @xmath238 , we have [ orthogonality ] _
s_1 , _ s_2=0 by the definition of the wavelet function @xmath227 in . also if we know that @xmath239 are in the same @xmath215 , for some @xmath116 and @xmath200 , then we can find @xmath240 s.t .
c(s_2)=c(s_1)+(ml(s_1 ) , nw(s_1 ) ) where @xmath225 is the center of the tile @xmath226 , @xmath241 its length and @xmath242 its width . then by the non - stationary phase method we know for any @xmath243
, there exists a constant @xmath244 depending only on @xmath66 s.t .
[ almostorthogonality ] |_s_1 , _
s_2| . here
we want to make a remark that the exact value of @xmath66 is not important , it just denotes some large number which might vary from line to line if we use the same notation later .
+ applying the above two estimates , we obtain @xmath245 as we know for any @xmath246 there exists @xmath240 s.t .
[ e1.66 ] c(s_2)=c(s_1)+(ml(s_1 ) , nw(s_1 ) ) , the above sum can be rewritten as _ m , n _
k__l_s_1_k , h , _ s_1-_k _
s_1_s_1,_s_2h , _ s_2-_k _
s_2with @xmath247 satisfying the relation .
+ now fix @xmath240 , by the estimate in , we know that @xmath248 by the cauchy - schwartz inequality , the last term is bounded by _
k__l_s_k,|h , _
s|^2 , then it suffices to prove that [ e5.23 ] _ k _ _ l_s_k , h , _ s-_k
_ s^2 2 ^ -lh_2 ^ 2 .
first to estimate every single term @xmath249 for a fixed tile @xmath226 : denote @xmath250 to be the shift of @xmath226 by @xmath251 units , i.e. s_m , n:=\{(x , y)^2 : ( x - ml(s ) , y - nw(s))s } , then by the triangle inequality we know that |h , _ s-_k _
s|_m , n|_s_m ,
n h(_s-_k _ s)dydx| . recall that in definition [ definition4 ] we use @xmath189 to denote the adapted version of the rectangle @xmath169 to the family of lipschitz curves , then clearly @xmath252 .
thus now for the inner integration along the curve @xmath83 , we do the same change of coordinates and the same parametrization of @xmath83 as in definition [ definition1 ] , i.e. we choose the coordinates s.t .
the horizontal axis is parallel to @xmath255 , and represent the curve @xmath83 by the lipschitz function @xmath256 .
if we let @xmath257 denote the projection of @xmath258 on the new vertical axis , the last expression becomes [ ee5.27 ] _ m , n _
p(s_m , n)_j(x , s_m , n)|h(g_x(y ) , y ) ( _ s(g_x(y ) , y ) - p_k [ _ s(g_x(y ) , y)])|dy dx .
( @xcite ) for a lipschitz function @xmath259 , we first take the caldern decomposition of @xmath260 , which yields the representation a(x)=_i dyadic a_i _
i(x ) , where @xmath261 is some mean zero function supported on @xmath262 , @xmath263 for each dyadic interval @xmath264 , let _
i=_|j||i| a_i _
j(c_i ) , where @xmath265 stands for the center of @xmath264 , denote the `` average slope '' of the lipschitz curve near @xmath264 , and define the beta number @xmath266 and the @xmath267-th beta number @xmath268 [ claim1 ] for @xmath271 , we have the following estimate : @xmath272)|dy\\ & \lesim \sum_{j_0\in \n}\frac{2^{-3l/2}}{<|j_0|+|m|+|n|>^n } \beta_{j_0}(x , s_{m , n } ) [ h]_{x , s_{m , n } } \mathbbm{1}_{\{-u(x)\in \omega_{s,2}\}}(x)\end{aligned}\ ] ] where @xmath273 is the @xmath267-th beta number for the lipschitz curve @xmath256 on the interval @xmath257 , @xmath274_{x , s_{m , n}}$ ] is the average of the function @xmath275 on the interval @xmath257 , i.e. _ x , s_m , n:=_j(x , s_m , n)|h(g_x(y ) , y)|dy .
the proof of lemma [ claim1 ] will be postponed to the next subsection .
substitute the estimate in lemma [ claim1 ] into the estimate for the term @xmath249 , we then have that @xmath276)|dy dx\\ & \lesim \sum_{m , n}\int_{p(s_{m , n})}\sum_{j_0\in \n}\frac{2^{-3l/2}}{<|j_0|+|m|+|n|>^n } \beta_{j_0}(x , s_{m , n } ) [ h]_{x , s_{m , n}}\mathbbm{1}_{\{-u(x)\in \omega_{s,2}\}}(x)dx\end{aligned}\ ] ] hence @xmath277_{x , s_{m , n}}\mathbbm{1}_{\{-u(x)\in \omega_{s,2}\}}(x)dx|^2\\ & \lesim \sum_{m , n , j_0}\frac{2^{-2l}}{<|j_0|+|m|+|n|>^n } ... \\ & \text { } \text { } \text { } \text { } \text { } \text { } ... \sum_k \sum_{\omega\in \mathcal{d}_{l}}\sum_{s\in \mathcal{u}_{k,\omega } } w(s ) \int_{p(s_{m , n } ) } \beta_{j_0}^2(x , s_{m , n } ) [ h]^2_{x , s_{m , n}}\mathbbm{1}_{\{-u(x)\in \omega_{s,2}\}}(x)dx\end{aligned}\ ] ] * proof of lemma [ lemma7 ] : * this lemma is akin to the carleson embedding theorem , as we have the following carleson type condition _
s_m , n_s^_m , n : j(x , s^_m , n)j(x , s_m , n ) _
j_0 ^ 2(j(x , s^_m , n ) ) w(s^_m , n)j_0 ^ 3 lip^2(_x ) , where the term @xmath279 plays such a role that , originally there are @xmath280 groups of dyadic rectangles _ k _ _ l_s_k , \{s_m , n } in the summation @xmath281 , which means that there are also @xmath280 groups of dyadic intervals _ k _ _ l_s_k , \{j(x , s_m , n ) } which are the projections of the intersection of the dyadic rectangles with @xmath83 on the vertical axis , the term @xmath279 just guarantees that there is just one such collection which has contribution , i.e. which has the right orientation in the sense of lemma [ lemma5 ] .
we assume that @xmath283 , which means the vector @xmath255 is roughly parallel to the long side of @xmath250 , otherwise the left hand side in lemma [ claim1 ] will also vanish due to lemma [ lemma5 ] .
after the change of variables in , the vector @xmath255 turns to @xmath111 .
+ * proof by ignoring the tails : * in order to explain how jones @xmath195-number appears , we first sketch the proof by ignoring the tails of the wavelet functions and the tail of the kernel of the littlewood - paley projection operator @xmath54 .
s ) ) [ h]_x , s. for fixed @xmath75 , we denote by @xmath285 the line of `` average slope '' we picked in the definition of the beta number for the lipschitz curve @xmath256 on the interval @xmath286 , for the sake of simplicity we assume @xmath287 . moreover , as both @xmath75 and @xmath226 are fixed , we will also just write @xmath288 instead of @xmath289
. then we make the crucial observation that [ aa5.38 ] p_k[_s^x(y , y)]=_s^x(y , y ) , where _ s^x(y , y):=__s(t)_s(y - t , y)dt , due to the fact that for any function @xmath227 with frequency supported on the @xmath116-th annulus , if we restrict the function to a straight line , it will still have frequency supported on the @xmath116-th annulus ( with one dimension less ) . in comparison with the definition of @xmath290 in ,
@xmath291 is defined as the hilbert transform along the vector @xmath255 ( which is ( 1 , 0 ) after the change of the variables we made in lemma [ lem3.2 ] and in the expression ) instead of the direction of the vector field @xmath3 at the point @xmath292 . as we have also ignored the tails of the kernel of @xmath54
, it is easy to see that the former and the latter terms in the last expression can essentially be handled in the same way .
hence in the following we will only consider the former term , which corresponds to the term [ aa5.41 ] & |_s(g_x(y),y)-_s^x(y , y)| + & = | _ _ k - l(t)_s(g_x(y)-t , y)dt- _ _ k - l(t)_s(y - t , y)dt| + & = 2^k - l |__0(2^k - lt)_s(g_x(y)-t , z)dt- _ _ 0(2^k - lt)_s(y - t , y)dt| + & = 2^k - l |_[_0(2^k - l(t+g_x(y)-y))-_0(2^k - lt)]_s(y - t , z)dt| . by the definition of the beta numbers
, we have that |g_x(y)-y|_0(x , s ) 2 ^ -k , which implies that |_0(2^k - l(t+g_x(y)-y))-_0(2^k - lt)|2 ^ -l _ 0(x , s ) by the fundamental theorem . in the end , by substituting the above estimate into and we obtain the desired estimate .
the full proof : * the main idea is still the same , and the difference is that we need to be more careful with the tails of the wavelet functions and the kernel of @xmath54 . for fixed @xmath75 , fixed @xmath294 and @xmath295 , denote @xmath296 as the line of `` average slope '' for the lipschitz curve @xmath256 on the interval @xmath257 , for the sake of simplicity we assume @xmath287 .
then the crucial observation becomes [ ee5.37 ] p_k[_s^x((x , s_m , n)y , y)]=_s^x((x , s_m , n)y , y ) . hence similar to , we obtain from that @xmath297\\ & = \phi_s(g_x(y),y)-\phi_s^x(\tau(x , s_{m , n})y , y)-p_k[\phi_s(g_x(y),y)-\phi_s^x(\tau(x , s_{m , n})y , y)].\end{aligned}\ ] ] denote i_s_m , n=|_j(x , s_m , n)h(g_x(y),y ) ( _ s(g_x(y),y)-_s^x((x , s_m , n)y , y))dy| and also ii_s_m , n= |_j(x , s_m , n)h(g_x(y),y ) p_k[_s(g_x(y),y)-_s^x((x , s_m , n)y , y)]dy| . let us first continue the proof of lemma [ claim1 ] : for the first term @xmath300 , we take @xmath267 in lemma [ lemma9 ] to be zero , then |_s(g_x(z ) , z)-_s^x((x , s_m , n)z , z)| , which implies that i_s_m , n _
0(x , s_m , n ) [ h]_x , s_m , n . for the second term @xmath301 , by the definition of @xmath54 , @xmath302|\\ & = |\int_{\r } ( \phi_s(g_x(z),z)-\phi_s^x(\tau(x , s_{m , n})z , z))2^k \check{\psi}_0(2^k(y - z))dz|\\ & \le |\sum_{j_0\in \z
} \int_{j(x , s_{m , n})+j_0 2^{-k } } ( \phi_s(g_x(z),z)-\phi_s^x(\tau(x , s_{m , n})z , z))2^k \check{\psi}_0(2^k(y - z))dz|.\end{aligned}\ ] ] for @xmath303 and @xmath304 , by the non - stationary phase method , we have that |_0(2^k(y - z))| , together with the estimate in lemma [ lemma9 ] , we arrive at @xmath302|\\ & \lesim \sum_{j_0\in \z } \frac{\beta_{|j_0|}(x , s_{m , n } ) 2^{k } 2^{-3l/2}}{<\min\{|m|+|n| , |m|+|n|-|j_0|\}>^n } \frac{1}{<j_0>^n}\\ & \lesim \sum_{j_0\in \z } \frac{\beta_{|j_0|}(x , s_{m , n } ) 2^{k } 2^{-3l/2}}{<|m|+|n|+|j_0|>^n}.\end{aligned}\ ] ] substitute the last expression into the estimate for @xmath301 , we get the desired estimate .
so far we have finished the proof of lemma [ claim1 ] except the lemma [ lemma9 ] , which we will do now .
+ * proof of lemma [ lemma9 ] : * as @xmath75 and @xmath250 are fixed now , later for simplicity we will just write @xmath288 instead of @xmath305 .
notice that in the new coordinate we chose for @xmath83 , the vector field along @xmath83 points in the direction of @xmath111 . then by the definition of @xmath290 and @xmath293
, we have @xmath306\varphi_{s}(\tau z - t , z)dt|.\end{aligned}\ ] ] by the definition of the beta numbers , we have that |g_x(z)-z|_|j_0|(x , s_m , n ) 2 ^ -k , which implies that |_0(2^k - l(t+g_x(z)-z))-_0(2^k - lt)|2 ^ -l _ |j_0|(x , s_m , n ) by the fundamental theorem . in the end , non - stationary phase method leads to the final estimate : @xmath307\varphi_{s}(\tau z - t , z)dt|\\ & \lesim \frac{2^{-l } \beta_{|j_0|}(x , s_{m , n } ) 2^{\frac{k}{2 } } 2^{\frac{k - l}{2}}}{<\min\{|m|+|n| , |m|+|n|-|j_0|\}>^n } .\end{aligned}\ ] ] thus we have finished the proof of lemma [ lemma9 ] and hence lemma [ claim1 ] .
20 p. auscher , s. hofmann , c. muscalu , t. tao and c. thiele : carleson measures , trees , extrapolation , and @xmath234 theorems .
46 ( 2002 ) , no . 2 , 257 - 325 .
j. azzam and r. schul : hard sard : quantitative implicit function and extension theorems for lipschitz maps .
geom . funct .
22 ( 2012 ) , no . 5 , 1062 - 1123
. m. bateman : @xmath5 estimates for maximal averages along one - variable vector fields in @xmath172 .
( 2009 ) , no . 3 , 955 - 963 .
m. bateman : single annulus @xmath308 estimates for hilbert transforms along vector fields .
rev . mat .
iberoam . 29
( 2013 ) , no . 3 , 1021 - 1069
. m. bateman : kakeya sets and directional maximal operators in the plane
. duke math .
j. 147 ( 2009 ) , no . 1 , 55 - 77 . m. bateman and c. thiele : @xmath309 estimates for the hilbert transforms along a one - variable vector field .
to appear in anal .
pde , arxiv : 1109.6396 ( 2011 )
. j. bourgain : a remark on the maximal function associated to an analytic vector field .
analysis at urbana .
i ( urbana , il , 1986 - 1987 ) , 111 - 132 .
m , christ , a. nagel , e. stein and s. wainger : singular and maximal radon transforms : analysis and geometry .
ann . of math .
( 2 ) 150 ( 1999 ) , no . 2 , 489 - 577 .
r. coifman , p. jones and s. semmes : two elementary proofs of the @xmath0 boundedness of cauchy integrals on lipschitz curves .
soc . 2 ( 1989 ) , no . 3 , 553 - 564 .
a. cordoba : the kakeya maximal function and the spherical summation multipliers .
99 ( 1977 ) , no . 1 , 1 - 22 .
jones : square functions , cauchy integrals , analytic capacity , and harmonic measure .
harmonic analysis and partial differential equations ( el escorial , 1987 ) , 24 - 68 , lecture notes in math . , 1384 , springer , berlin , 1989 . g. karagulyan : on unboundedness of maximal operators for directional hilbert transforms .
( 2007 ) , no .
10 , 3133 - 3141 ( electronic )
. m. lacey : the bilinear maximal functions map into @xmath5 for @xmath310 .
ann . of math .
( 2 ) 151 ( 2000 ) , no . 1 , 35 - 57 . m. lacey and x. li : maximal theorems for the directional hilbert transform on the plane
358 ( 2006 ) , no . 9 , 4099 - 4117
. m. lacey and x. li : on a conjecture of e. m. stein on the hilbert transform on vector fields .
( 2010 ) , no .
965 , viii+72 pp .
isbn : 978 - 0 - 8218 - 4540 - 0 m. lacey and c. thiele : @xmath5 estimates for the bilinear hilbert transform .
94 ( 1997 ) , no . 1 , 33 - 35 . m. lacey and c. thiele : on caldern s conjecture for the bilinear hilbert transform
usa 95 ( 1998 ) , no .
9 , 4828 - 4830 ( electronic )
. o. perron : ber einen satz von besicovitch .
z. 28 ( 1928 ) , no . 1 , 383 - 386 .
e. m. stein and b. street : multi - parameter singular radon transforms iii : real analytic surfaces .
( 2012 ) , no . 4 , 2210 - 2238 . | we prove the @xmath0 boundedness of the directional hilbert transform in the plane relative to measurable vector fields which are constant on suitable lipschitz curves . |
in most publications the integrable @xmath0-dimensional euler top is represented as a flow on the cotangent bundle @xmath7 or on the coalgebra @xmath1 .
recently , an alternative description of this problem as a system on a symplectic subvariety of the group product @xmath8 was proposed in @xcite .
a first discretization of the free @xmath0-dimensional top on @xmath7 was constructed in @xcite by the method of factorization of matrix polynomials .
this discretization is represented by a second order lagrangian correspondence , which does not explicitly involve a time step , it is determined by initial data ( a choice of two subsequent points on @xmath9 ) .
on the other hand , in @xcite ( see also @xcite ) suris introduced a concept of an integrable discretisation of a finite - dimensional hamiltonian system as a one parameter family of integrable poisson maps parameterized by a time step @xmath10 , which differ from the identity map by @xmath11 , and whose poisson structure and the integrals of motion differ at most by @xmath11 from those of the continuous - time system . in the special case where the discretization preserves exactly both the poisson structure and the integrals of motion , one speak of an `` exact discretization ''
: one has a family of bcklund transformations , which map solutions into solutions and are interpolated by a hamiltonian flow generated by some function of the integrals of motion of the continuous system .
a class of implicitly defined poisson maps @xmath12 discretizing the classical euler top in the space of the angular momentum was indicated in @xcite .
the maps preserve the energy and momentum integrals of the continuous problem and contain explicitly a time step parameter .
it was shown that such a map preserves the standard lie
poisson structure on @xmath6 if and only if its restriction onto complex invariant manifolds , open subsets of elliptic curves , is a shift , which is constant on each curve .
recently , another integrable discretization of the top on @xmath6 , which is _ explicit _ , but does not preserve the integrals of the continuous problem was found in @xcite by applying the hirota method .
[ [ contents - of - the - paper . ] ] contents of the paper .
+ + + + + + + + + + + + + + + + + + + + + + our aim is twofold .
first , in section 2 , we propose yet another description of the continuous @xmath13-dimensional euler manakov top as a reduction of a hamiltonian system on so called extended stiefel variety @xmath2 , a symplectic submanifold of dimension @xmath14 in @xmath15 , where @xmath16 is an even integer .
we present a lax representation of this system with a rational parameter , which , in a sense described below , is dual to manakov s lax pair found in @xcite .
the system possesses @xmath17 commuting symmetry fields @xmath18 generated by hamiltonians @xmath19 .
its marsden weinstein reduction with respect to the action of the fields gives rise to a hamiltonian system on a rank @xmath20 coadjoint orbit @xmath21 in the coalgebra @xmath1 , whereas the original poisson structure in @xmath2 is a pull - back of the standard lie
poisson structure of @xmath1 restricted onto the orbit .
the reduced hamiltonian system coincides with the euler manakov system on @xmath21 . in case of the maximal rank @xmath20 ,
the level variety @xmath22 is the group @xmath9 , and the restriction of the original system onto the group yields a flow describing the motion of the @xmath13-dimensional top in space .
second , in section 3 , we present an intertwining relation ( discrete lax pair ) generating a explicit @xmath23-depended family of two - valued complex bcklund transformations @xmath24 of the variety @xmath4 , which preserve the above poisson structure and the first integrals of the continuous hamiltonian system ( formula ( [ 4.10 ] ) ) .
the restricton of @xmath24 onto the group @xmath5 provides a discretization of the motion of the classical euler top in space and has a transparent geometric interpretation , which , in turn , can be regarded as a discrete version of the celebrated poinsot model of motion and which inherits some properties of another discrete integrable system , the elliptic billiard ( figure 1 ) . on the other hand
, the reduction of @xmath24 onto the coalgebra @xmath6 gives a new explicit discretization of the classical euler top , which also preserves its first integrals ( formula ( [ diff ] ) ) . like the moser
veselov correspondence , the both discretizations do not explicitly involve a time step and their continuous limits depend on the parameter @xmath23 .
recall that the free motion of an @xmath0-dimensional rigid body is described by the euler
frahm equations ( @xcite ) @xmath25 , \label{2.1}\ ] ] where @xmath26 is the angular velocity , @xmath27 the angular momentum of the body in the moving frame .
following @xcite , these equations are hamiltonian with respect to the degenerate lie
poisson bracket on @xmath1 @xmath28 and @xmath29 .
the restriction of @xmath30 onto orbits of coadjoint action of @xmath9 in @xmath1 is nondegenerate .
a generic orbit @xmath31 parameterized by @xmath32 $ ] independent casimir functions of the bracket is thus a symplectic variety of dimension @xmath33 $ ] .
equations ( [ 2.1 ] ) are known to be integrable provided @xmath34 and @xmath35 are related as @xmath36=[\omega , b]$ ] , where @xmath37 are constant commuting matrices , and all the eigenvalues of @xmath38 and @xmath39 are distinct .
the integrability follows from the lax representation with a rational spectral parameter found by manakov in @xcite , or from a hyperelliptic lax pair indicated in @xcite .
these lax pairs provide a complete set of integrals of motion , whose involutivity can be proved by applying @xmath40-matrix theory . for the concreteness , in the sequel we consider the case @xmath41 , @xmath42 .
then @xmath43 , and equations ( [ 2.1 ] ) take the form @xmath44\ ] ] apart from this `` basic '' system , there exists a whole hierarchy of `` higher manakov systems '' , which are defined by different relations between @xmath35 and @xmath34 , and which commute with ( [ fm_again ] ) . below
we show that the restrictions of the frahm manakov system on rank @xmath20 orbits of coadjoint representation of @xmath9 in @xmath1 are closely related to certain hamiltonian dynamical systems on extended stiefel varieties .
recall that the standard stiefel variety @xmath45 is the variety of ordered sets of @xmath20 orthogonal vectors in @xmath46 ( @xmath47 ) having fixed squares .
it is a smooth variety of dimension @xmath48 ( see e.g. , @xcite ) .
namely , as follows from ( [ 2.1 ] ) , the angular momentum in space is a constant matrix .
hence , due to the darboux theorem , in the case rank @xmath49 there exist @xmath20 mutually orthogonal and _ fixed in space _
vectors @xmath50 , @xmath51 such that @xmath52 and the momentum @xmath34 can be represented in form @xmath53 under the above conditions , the set of @xmath54 matrices @xmath55 forms the _ extended _ stiefel variety @xmath2 .
in contrast to the standard stiefel variety , for each index @xmath56 , the absolute values @xmath57 are not fixed .
thus , @xmath2 is of dimension @xmath14 , and the @xmath54 components of @xmath58 play the role of excessive coordinates on it .
let @xmath59 be the canonical symplectic structure on the space @xmath60 and let @xmath61 denote the restriction of 2-form @xmath62 onto @xmath63 .
the latter subvariety is defined by conditions @xmath64 which consist of @xmath65 independent scalar equations @xmath66 .
the matrix of standard poisson brackets of the constraint functions @xmath67 in @xmath15 is nondegenerate .
it follows that 2-form @xmath61 is also nondegenerate and the extended stiefel variety is symplectic .
since the vectors are fixed in space , in the frame attached to the top they satisfy the poisson equations @xmath68 ,
@xmath69 , @xmath70 , which imply @xmath71 as above , we put @xmath72 , @xmath73 and define a dynamical system on @xmath2 , which is generated by ( [ r5.88 ] ) , ( [ r5.87 ] ) : @xmath74 , \\ \dot{\cal y } & = { \cal y } [ a ( { \cal x}^t{\cal y}-{\cal y}^t{\cal x } ) -{\cal y}^t{\cal x } a ] \ , . \end{aligned } \label{r588}\ ] ] [ tr5.25 ] \1 ) under the substitution @xmath75 solutions of the system @xmath76 give rank @xmath20 solutions of the frahm manakov system ( [ fm_again ] ) on @xmath1 .
\2 ) up to the action of the discrete group generated by reflections @xmath77 , the system @xmath76 is described by @xmath78 lax pair with rational parameter @xmath79 @xmath80\ , , \qquad l , a\in { \rm sp}(k/2 ) , \quad \lambda\in{\mathbb c } \ , , \label{r5.89 } \\ l=\begin{pmatrix}-{\cal x } ( \lambda{\bf i}-a)^{-1 } { \cal y}^t & -{\cal x } ( \lambda{\bf i}-a)^{-1 } { \cal x}^t \\ { \cal y } ( \lambda{\bf i}-a)^{-1 } { \cal y}^t & { \cal y } ( \lambda{\bf i}-a)^{-1 } { \cal x}^t \end{pmatrix } \equiv \sum_{i=1}^m \frac { { \cal n}_i } { \lambda - a_i}\ , , \label{dual1 } \\ { \cal n}_i=\begin{pmatrix } { \bar x}_i { \bar y}_i^t & -{\bar x}_i { \bar x}_i^t \\ -{\bar y}_i { \bar y}_i^t & -{\bar y}_i { \bar x}_i^t \end{pmatrix } , \nonumber \\ a=\begin{pmatrix } { \cal x } ( a+\lambda{\bf i}){\cal y}^t & { \cal x } ( a+\lambda{\bf i}){\cal x}^t \\ -{\cal y } ( a+\lambda{\bf i } ) { \cal y}^t & -{\cal y } ( a+\lambda{\bf i}){\cal x}^t \end{pmatrix}\ , , \label{aa}\end{gathered}\ ] ] where @xmath81 ( respectively @xmath82 is @xmath83-th column of @xmath84 ( respectively of @xmath85 , and @xmath86 is the unit @xmath87 matrix .
_ the first statement follows directly from the derivation of the system @xmath76 .
further , we calculate the derivative @xmath88 by virtue of equations ( [ r588 ] ) . in view of matrix relations @xmath89 and @xmath90 , the derivative coincides with the commutator in @xmath91 . @xmath92
* remark 2.1 . *
notice that the entries of matrices @xmath93 are polynomials in @xmath79 , and , under the substitution ( [ r5.87 ] ) , the coefficients of the characteristic polynomial @xmath94 can be expressed in terms of @xmath95 only as follows @xmath96 where @xmath97 are diagonal minors of order @xmath56 corresponding to multi - indices @xmath98 , @xmath99 notice that the leading coefficients @xmath100 form a complete set of casimir functions on @xmath1 .
the @xmath78 matrix @xmath101 in ( [ dual1 ] ) belongs to a wide class of lax operators of the form @xmath102 where @xmath103 is a constant matrix and @xmath104 are @xmath105 matrices .
such lax matrices can be regarded as images of moment maps to the loop algebra @xmath106 , and integrable systems generated by them have been studied in the series of papers @xcite in connection with the duality to so called rank @xmath20 perturbations of constant diagonal matrices of dimension @xmath107 ( following moser @xcite ) . in particular , the @xmath78 lax matrix ( [ dual1 ] ) is dual to the @xmath107 lax matrix in the manakov representation , @xmath108 in the sense that under the relation ( [ r5.87 ] ) the spectral curves @xmath109 and @xmath110 are birationally equivalent and the parameter @xmath79 plays the role of the eigenvalue parameter for ( [ 2.17 ] ) .
the characteristic polynomials of the dual lax matrices are related by the weinstein
aronjan formula ( see @xcite ) .
* remark 2.2 .
* the matrix @xmath111 in ( [ aa ] ) can be represented in form @xmath112_+ + ( a_1+\cdots + a_n ) l_0 , \qquad l_0=\begin{pmatrix } 0 & { \cal x } { \cal x}^t \\
-{\cal y } { \cal y}^t & 0 \end{pmatrix}\ ] ] where @xmath113_+$ ] denotes the polynomial part in @xmath79 of the expression .
notice that the lax equation @xmath114 $ ] describes the vector flow @xmath115 for each index @xmath18 , equations ( [ rot ] ) generate rotations @xmath18 in 2-planes spanned by the vectors @xmath116 , which leave the momentum @xmath34 invariant .
let @xmath117 be the poisson bracket on @xmath2 obtained as the dirac restriction of the standard bracket in @xmath15 .
symplectic properties of our system are descibed by [ stand ] the dynamical system @xmath76 on @xmath2 is hamiltonian with respect to @xmath117 with the hamilton function @xmath118 . in the abundant
coordinates @xmath119 it admits the canonical representation @xmath120 _ proof . _
the equivalence of equations ( [ canonical ] ) and ( [ r588 ] ) on @xmath2 is verified by direct calculations .
next , according to the dirac formalizm , the standard bracket and @xmath117 are different by terms containing @xmath121 .
the latter equal zero since the constraint functions @xmath67 given by ( [ constraints ] ) are invariants of the flow generated by @xmath122 on @xmath15 .
hence , equations ( [ r588 ] ) or ( [ canonical ] ) are hamiltonian with respect to @xmath117 .
@xmath92 rotations @xmath18 given by ( [ rot ] ) are generated by the hamiltonians @xmath123 , the restrictions of the functions @xmath124 on @xmath2 . clearly , these functions are first integrals of the system ( [ r588 ] ) and moreover they commute with @xmath125 .
let us fix the values of the hamiltonians by putting @xmath126 these conditions define the customary stiefel variety @xmath45 . under the substitution ( [ r5.87 ] ) ,
the factor variety @xmath127 coincides with a rank @xmath20 coadjoint orbit @xmath128 of dimension @xmath129 , which is parameterized by the constants @xmath130 .
notice that @xmath131 .
\1 ) under the map @xmath132 , the lie poisson bracket on @xmath128 is the push - forward of the bracket @xmath117 .
\2 ) the poisson ( marsden weinstein ) reduction of the system ( [ r588 ] ) obtained by fixing values of @xmath123 and by factorization by @xmath18 , @xmath51 coincides with the restriction of the frahm manakov system with hamiltonian + @xmath133 onto the orbit @xmath21 . _ proof . _ 1 ) . in view of ( [ poinsot ] ) , ( [ r5.87 ] ) , @xmath134 i.e.
, the canonical bracket on @xmath15 is the pull - back of the bracket @xmath135 on @xmath128 . on the other hand , on @xmath2 , @xmath136 since for any @xmath137 , @xmath138 .
this proves item 1 ) . 2 ) .
by item 1 ) and proposition [ stand ] , the poisson reduction of system ( [ r588 ] ) onto @xmath21 is described by the lie
poisson bracket @xmath135 and the hamiltonian @xmath139 , i.e. , it is the corresponding restriction of the frahm manakov system . @xmath92
the reduced system on the orbit @xmath21 is integrable and its generic invariant manifolds are tori of dimension @xmath140 ( see , e.g. , @xcite ) . on the other hand ,
the preimage of a generic point @xmath141 in @xmath45 is a @xmath17-fold product of circles @xmath142 ( in the complex case @xmath143 ) .
this implies that the original system on @xmath2 has generic invariant tori of dimension @xmath144 , i.e. , a half of dimension of the symplectic manifold @xmath2 .
hence , the original system @xmath76 is also integrable . to get a global view on the above manifolds ,
we represent them in the following diagram , with the dimension indicated above , where arrows denote the corresponding relations ( restrictions or factorizations ) .
@xmath145 @xmath146 * remark 2.3 . * in the case of maximal rank @xmath20 ( @xmath147 or @xmath148 ) , when @xmath21 is a generic coadjoint orbit @xmath31 , the stiefel variety @xmath45 is isomorphic to the group @xmath9
. then the following commutative diagram holds @xmath149 where the values @xmath150 of nonzero casimir functions @xmath151 correspond to the constants @xmath152 .
the mapping @xmath153 can be regarded as a multi - dimensional analog of the hopf fibration @xmath154 .
the restriction of the system ( [ r588 ] ) onto @xmath45 yields an integrable flow on the group @xmath155 which describes the motion of the frahm manakov top in space for the chosen angular momentum .
for @xmath156 such a flow was considered in @xcite from the point of view of its hydrodynamical interpretation .
[ [ a - generalization - of - the - chasles - theorem . ] ] a generalization of the chasles theorem .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + if the rank @xmath20 is not maximal , then the components of @xmath119 themselves are not sufficient to form a complete set of coordinates on @xmath9 and to determine the position of the top in space uniquely . however , in this case one can make use of the following geometric property described in @xcite .
let us fix a part of constants of motion by putting in ( [ cal_i ] ) @xmath157 and consider family of confocal cones in @xmath158 @xmath159 let @xmath160 be a @xmath20-plane spanned by the orthogonal vectors @xmath161 .
( @xcite ) .
[ prop5.6 ] 1 ) . under the motion of the frahm manakov top with constants ( [ c s ] )
the @xmath20-plane @xmath162 is tangent to the fixed cones @xmath163 . 2 ) .
let @xmath164 be a normal vector of the cone @xmath165 at a point of the contact line @xmath166 .
then the vectors @xmath167 together with @xmath168 form an orthogonal frame in @xmath169 which is fixed in space . for fixed polynomial @xmath170 , the vectors @xmath171 can be calculated in terms of @xmath172 and , thereby , the position of the top in space is completely determined .
proposition [ prop5.6 ] defines a single - valued map @xmath173 under which generic invariant tori of dimension @xmath174 on @xmath2 become tori of the same dimension on the group @xmath9 . note that the above proposition generalizes the celebrated chasles theorem on the propery of the tangent line to a geodesic on a quadric .
[ [ the - rank-2-case . ] ] the rank 2 case .
+ + + + + + + + + + + + + + + + in the simplest case @xmath175 the angular momentum can be represented in form @xmath176 and equations ( [ r588 ] ) describe a hamiltonian system on the extended stiefel variety @xmath177 , @xmath178 with the hamiltonian @xmath179 equivalently , this system describes the evolution of fixed orthogonal vectors @xmath180 in a frame attached to the @xmath0-dimensional body .
the system admits the following @xmath181 lax pair arising from ( [ r5.89 ] ) , @xmath182\ , , \label{r5.99 } \\
l(\lambda)=\phi(\lambda ) \begin{pmatrix } -\sum_{i=1}^m \frac{x_i y_i}{\lambda - a_i } & -\sum_{i=1}^m \frac{x_i^2}{\lambda - a_i } \\
\sum_{i=1}^m \frac{y_i^2}{\lambda - a_i } & \sum_{i=1}^m \frac{x_i
y_i}{\lambda - a_i } \end{pmatrix}\ , , \nonumber \\
\medskip \nonumber { \cal a } ( \lambda)=\begin{pmatrix } -\sum_{i=1}^m ( \lambda+ a_i){x_i y_i } & -\sum_{i=1}^m ( \lambda + a_i)x_i^2 \\ \sum_{i=1}^m ( \lambda + a_i)y_i^2 & \sum_{i=1}^m ( \lambda + a_i ) { x_i y_i } \end{pmatrix}\ , , \nonumber \\ \phi(\lambda)=(\lambda - a_{1})\cdots(\lambda - a_{m } ) , \nonumber \end{gathered}\ ] ] the lax representation ( [ r5.99 ] ) was first indicated in @xcite , where it was shown to be dual to an @xmath107 lax pair for the rank 2 case found by moser in @xcite . in view of relation ( [ decom ] ) , the characteristic polynomial @xmath183 for ( [ r5.99 ] ) can be written in form @xmath184 , where @xmath185 is the family of quadratic integrals defined in ( [ cal_i ] ) , @xmath186 notice that @xmath187 is a casimir function of the standard lie poisson bracket on @xmath1 .
with respect to the poisson bracket @xmath188 on @xmath189 , this function generates permanent rotations of the top in the fixed 2-plane @xmath190 , which leave the components of @xmath34 invariant .
let us fix the constants of motion by putting @xmath191 this defines hyperelliptic spectral curve in @xmath192 of genus @xmath193
@xmath194 as noticed in @xcite , the real generic @xmath195-dimensional invariant tori of the system can be extended to open subsets of generalized jacobian varieties jac@xmath196 , which are extensions of the customary @xmath197-dimensional jacobian jac@xmath198 by @xmath199 and which can be regarded as the factor of @xmath200 by the lattice generated by @xmath201 independent period vectors of @xmath197 holomorphic differentials @xmath202 and a meromorphic differential of the third kind @xmath203 having a pair of simple poles at the infinite points @xmath204 on the curve @xmath205 .
the coefficients of the matrix polynomial @xmath101 are meromorphic functions on jac@xmath196 , whereas the components of the momentum @xmath95 and the normal vectors @xmath164 are meromorphic on a covering of the jacobian jac@xmath198 itself ( the @xmath199-extension is factored out by the action of @xmath206 ) . in the classical case
@xmath156 the curves @xmath205 become elliptic ones and generic invariant tori in @xmath207 and in @xmath5 are 2-dimensional .
an explicit solution for the components of the rotation matrix in terms of theta - functions and exponents was first given in @xcite ( see also @xcite ) .
since now rank @xmath2082 in the generic case , the above commutative diagram takes the form @xmath209 @xmath210 being the coadjoint orbit ( 2-dimensional sphere ) corresponding to the constant @xmath211 .
a first integrable discretization of the @xmath0-dimensional euler manakov top was constructed in @xcite by the method of factorization of matrix polynomials .
it was represented by the correspondence @xmath212 , @xmath213 , which , in our notation reads @xmath214 given @xmath215 , the new matrix @xmath216 is found from equation @xmath217 , whose solution is not unique . in given section we describe a symplectic map @xmath218 , @xmath219 governed by an arbitrary parameter @xmath220 , which preserves the first integrals of the continuous system ( [ bg ] ) and whose restriction to each generic complex torus , generalized jacobian jac@xmath196 , is given by shift by the 2-dimensional vector @xmath221 @xmath222 being involutive points on the elliptic spectral curve , the simplest case of ( [ hypp ] ) , @xmath223 note that @xmath224 is a correctly defined vector in the generalized jacobian : under a change of integration path on @xmath205 it increases by a period vector of jac@xmath196 .
we also emphasize that here @xmath23 is a constant parameter , whereas the conjugated coordinate @xmath225 depends on the equation of the curve . as shown in @xcite by applying an addition theorem for a class of meromorphic functions on generalized hyperelliptic jacobians ,
such a map admits intertwining relation ( discrete lax pair ) @xmath226 where @xmath101 is defined in ( [ r5.99 ] ) and @xmath227 depends on the new variables @xmath228 in the same way as @xmath101 depends on @xmath229 . in view of ( [ m2.13 ] ) for @xmath156 ,
@xmath230 now putting in ( [ bak4.8 ] ) subsequently @xmath231 and calculating the matrices @xmath232 , we find @xmath233 from here the new variables can be recovered up to the action of the group generated by reflections @xmath234 .
imposing the condition of the existence of a continuous limit ( see below ) , we choose the following relations @xmath235 these expressions together with ( [ ab ] ) , ( [ * ] ) describe the map @xmath218 in an _ explicit _ form . since a generic parameter @xmath23 corresponds to two values of @xmath225 ,
the map is generally two - valued .
the restriction of the map onto the group @xmath5 admits a transparent geometric interpretation , which can be regarded as a `` discrete version '' of the kinematic poinsot model ( see , e.g. , @xcite ) .
namely , let @xmath236 and let @xmath237 be rotation matrix defining a position of a rigid body in space .
we attach to the body a cone @xmath238 , which is fixed in the body frame @xmath239 , and assume that @xmath240 under these conditions the cone is real and regular .
let @xmath241 be 2-plane spanned by @xmath180 , which is thus fixed in space and orthogonal to the momentum vector @xmath242 .
assume also that @xmath180 are such that @xmath241 has a nonempty real intersection with the cone @xmath243 along lines @xmath244 .
one can show that under this condition the coordinates @xmath225 defined in ( [ * ] ) and the parameters @xmath245 are real .
[ proj_bill ] let @xmath246 , @xmath247 be a unique cone attached to the body such that it is confocal to @xmath243 and tangent to the fixed plane @xmath241 .
then then new position of the body defined by the rotation matrix @xmath248 and expressions ( [ 4.10 ] ) is obtained from the original position by rotating the cones @xmath249 about axis @xmath250 or @xmath251 until @xmath252 again touches @xmath241 .
this geometric construction is illustrated on figure 1 . in the new position @xmath253 determined by rotation about @xmath251 ,
the cone @xmath243 intersects @xmath241 along @xmath251 and another line @xmath254 .
then the next iteration is generated by rotation about @xmath251 or @xmath254 .
the two - valuedness of the map @xmath255 is now related to the possibility of rotation about two different axes in @xmath256 .
it follows that @xmath257-th iteration of the map is only @xmath258-valued , not @xmath259-valued . by fixing a sign of @xmath225 in ( [ * ] ) , @xmath255 becomes single - valued and generates a sequence of points on @xmath5 .
* remark 3.1 .
* as follows from ( [ 4.10 ] ) , the map @xmath255 admits particular solutions , for which the vector @xmath242 remains to be an eigenvector of the inertia tensor @xmath260 , whereas @xmath229 themselves rotate by a fixed angle in the plane @xmath261 .
such solutions can be regarded as analogs of stationary permanent rotations of the classical euler top about its principal intertia axes .
note that when @xmath180 are chosen such that @xmath241 is ( almost ) tangent to the cone @xmath243 ( @xmath262 ) , @xmath252 and @xmath243 confluent and , according to the above model , the cone @xmath263 is rolling without sliding over the fixed plane thus giving a continuous limit motion on the group @xmath5 . from the algebraic geometrical point of view , in the above limit the points @xmath264 on the spectral curve @xmath205 come together to a branch point @xmath265 and the shift vector @xmath224 on the generalized jacobian tends to zero .
let @xmath10 be a small complex parameter .
setting @xmath266 , @xmath267 in ( [ ab ] ) , we have the expansions @xmath268 where @xmath269 is a real nonzero constant .
now we set @xmath270 substituting the above expansions into ( [ 4.10 ] ) , then matching the coefficients at @xmath10 in both sides and taking into account relation @xmath271 we obtain the following differential equations describing the limit flow on a subset of @xmath207 @xmath272 \equiv \frac 1\varkappa x\wedge a^*(x\wedge y ) , \\
y & = \frac { \det a^*}\varkappa \bigg[\langle y , ( a^*)^{-1}y\rangle ( a^*)^{-1}x - \langle x , ( a^*)^{-1}y\rangle ( a^*)^{-1}x \bigg ] \equiv \frac 1\varkappa y\wedge a^*(x\wedge y ) , \end{aligned } \\ a^*=a-\lambda^ * { \bf i } .
\nonumber \end{gathered}\ ] ] these equations are hamiltonian with the hamilton function @xmath273 \equiv \frac 1{2\varkappa } \sum_{i=1}^3 ( a_i-\lambda^ * ) m_i^2.\ ] ] notice that this function equals zero on the limit continuous flow .
the restriction of this flow on jac@xmath196 is tangent to the curve @xmath274 at the point @xmath275 .
_ proof of theorem _ [ proj_bill ] .
the condition for @xmath241 to be tangent to the cone + @xmath246 has the form @xmath276 comparing this with the family of integral ( [ i_2 ] ) for @xmath156 , we conclude that @xmath277 , which is constant under the map , hence the plane spanned by @xmath228 is again tangent to @xmath252 . next ,
any translation in @xmath5 is represented as a finite rotation about an axis in @xmath256 . as follows from relations ( [ 4.10 ] ) , @xmath278 , hence the line along the vector @xmath279 is invariant of the action of @xmath255 on @xmath256 and therefore represents the axis of such a rotation . finally ,
in view of ( [ ab ] ) , we have @xmath280 , \end{aligned}\ ] ] which equals zero by virtue of ( [ * ] ) .
hence @xmath281 , which imply that the vector @xmath282 lies on the cone @xmath243 .
this establishes the theorem . @xmath92
* remark 3.2 .
* when the attached cone @xmath283 does not have real intersection with @xmath284 , the coordinate @xmath225 is imaginary and , according to ( [ 4.10 ] ) , ( [ ab ] ) , the new values @xmath228 are complex . as a result ,
under the reality conditions ( [ real ] ) the map @xmath255 is real only on the subset @xmath285 defined by unequality @xmath286 on the boundary of @xmath287 , the map tends to the identical one . under the factorization by rotations of @xmath206
, the transformation @xmath255 induces a map @xmath288 such that @xmath289 the latter map is correctly defined , i.e. , it does not depend on a concrete choice of vectors @xmath180 giving the same @xmath34 .
it preserves the first integrals of the classical euler top on @xmath6 and its generic invariant manifolds are open subsets of 4-fold unramified coverings of the complex torus jac@xmath290 .
the restriction of @xmath291 onto jac@xmath198 is given by shift by the holomorphic integral @xmath292 , which thus depends only on the constants @xmath293 . according to a theorem in @xcite
, this implies that the map @xmath291 preserves the standard lie
poisson structure on @xmath6 .
_ proof of proposition _ [ reflection ] . in view of relations ( [ 4.10 ] ) , we find @xmath297 where , as above , @xmath298 , @xmath299 note that vector @xmath300 is normal to the cone @xmath243 at a point of the intersection line @xmath251 or @xmath250 . hence , @xmath301 is orthogonal to @xmath302 and @xmath282 .
next , since @xmath282 lies in the planes @xmath303 , this vector is orthogonal to @xmath294 .
this , together with the equality @xmath304 implies that the sum @xmath305 is parallel to @xmath300 and @xmath306 is parallel to @xmath282 . as a result , ( [ times ] ) implies ( [ b - l - s ] ) . to find factor @xmath269 ,
we first introduce angle @xmath307 between vectors @xmath215 and @xmath34 .
since @xmath304 , the vectors @xmath301 and @xmath308 are orthogonal , and we have @xmath309 on the other hand , since @xmath310 is orthogonal to @xmath311 , from ( [ b - l - s ] ) and the properties of the vector product we deduce @xmath312 comparing the right hand sides of the above two relations , we obtain ( [ kappa ] ) . the first equality in ( [ bill ] ) holds because the map @xmath291 preserves the first integrals of the euler top .
next , since the vector @xmath306 lies on the cone @xmath243 , we have @xmath313 .
expanding this and using the first equality in ( [ bill ] ) yields the second equality . @xmath92 * remark 3.3 .
* the fact that the difference @xmath301 is orthogonal to @xmath302 and @xmath282 implies that _ the angle between the normal vector @xmath302 and the plane @xmath241 equals the angle between @xmath302 and _ @xmath314 .
this property can be regarded as a projective version of the birkhoff condition of elastic impacts , hence the geometric construction of theorem [ proj_bill ] illustrated in figure 1 describes a projective analog of the plane elliptic billiard .
( note that no any plane section of the cones and of the sequence of @xmath241 gives such a plane billiard . ) to obtain the map @xmath315 in an explicit form , we use the fact that the vector @xmath282 satisfies the system of homogeneous equations @xmath316 one of its solutions , @xmath317 , normalized by the condition @xmath318 , has the form @xmath319 substituting this instead of @xmath282 into ( [ times ] ) and symmeterising obtained expressions , we arrive at the following relations @xmath320 where @xmath321 and @xmath322 is a common factor .
next , multiplying the both sides of ( [ diff ] ) by @xmath323 respectively , then summing and using the second equality in ( [ bill ] ) , we find the factor @xmath322 in form @xmath324}\ , , \ ] ] where the summation in the denominator ranges over the three terms obtained by the cyclic permutations of indices ( 1,2,3 ) .
thus the right hand sides of ( [ diff ] ) are homogeneous expressions of degree -1 in @xmath325 .
the map @xmath315 given by ( [ diff ] ) , ( [ chi ] ) , ( [ d ] ) preserves the first integrals of the continuous euler top , as well as the standard lie
poisson structure on @xmath6 . under reality conditions ( [ real ] ) , the map is real inside the conical domain @xmath326 its restriction onto each regular elliptic curve @xmath327 is represented by the shift by the holomorphic integral @xmath292 . near the boundary of @xmath328 the energy constant @xmath329 tends to @xmath23 and
the shift @xmath330 tends to zero .
the continuous limit of @xmath291 coincides with the euler equations @xmath331 $ ] . since generic solutions @xmath332 of the classical euler top
are proportional to the elliptic functions @xmath333 , with @xmath334 and the map @xmath291 results in adding @xmath330 to the argument @xmath335 , the relations ( [ diff ] ) , ( [ d ] ) , ( [ chi ] ) can be regarded as a set of explicit addition formulae for these functions . however , in this case neither the moduli of the curve @xmath205 nor the shift parameter are fixed : they depend on initial position of @xmath34 in @xmath6 .
fedorov yu .
integrable systems , lax representations , and confocal quadrics . in : _ dynamical systems in classical mechanics _ , 173199 ,
ser . 2 , * 168 * , amer .
, providence , ri , 1995 fedorov yu . discrete versions of some algebraic integrable systems related to generalized jacobians .
_ side iii symmetries and integrability of difference equations _ ( sabaudia , 1998 ) , 147160 , crm proc .
lecture notes * 25 * , amer . math .
soc . , providence , ri , 2000 kozlov v. v. hydrodynamics of noncommutative integration of hamiltonian systems . in : _
dynamical systems in classical mechanics _ , 227238 , amer .
ser . 2 , * 168 * , amer .
soc . , providence , ri , 1995 | we show that the @xmath0-dimensional euler manakov top on @xmath1 can be represented as a poisson reduction of an integrable hamiltonian system on a symplectic extended stiefel variety @xmath2 , and present its lax representation with a rational parameter .
we also describe an integrable two - valued symplectic map @xmath3 on the 4-dimensional variety @xmath4 .
the map admits two different reductions , namely , to the lie group @xmath5 and to the coalgebra @xmath6 .
the first reduction provides a discretization of the motion of the classical euler top in space and has a transparent geometric interpretation , which can be regarded as a discrete version of the celebrated poinsot model of motion and which inherits some properties of another discrete system , the elliptic billiard .
the reduction of @xmath3 to @xmath6 gives a new explicit discretization of the euler top in the angular momentum space , which preserves first integrals of the continuous system .
_ short title _
: flows and backlund transformations on extended stiefel varieties |
a primary goal of supernova research is an understanding of the progenitor stars and explosion mechanisms of the different types of supernovae ( sne ) .
unfortunately , a sn leaves few traces of the star that underwent the catastrophic event .
inferences concerning the nature of the progenitor can be made from the ejecta at late times ( fransson & chevalier 1989 ; leibundgut et al .
1991 ) and by probing the circumstellar emission from the blast wave ( e.g. , fransson et al . 1989 ) .
unambiguous information can be derived if the progenitors can be directly identified in pre - explosion images , but to date this has been possible for only three sne : sn 1987a in the lmc ( west et al .
1987 ; walborn et al . 1987 , 1989 ) , and probably sn 1978k in ngc 1313 ( ryder et al .
1993 ) and sn 1993j in m81 ( aldering , humphreys , & richmond 1994 ; cohen , darling , & porter 1995 ) .
( the progenitor of sn 1961v has also been identified , but this event was most likely an @xmath0 carinae - like outburst , rather than a genuine sn ; see goodrich et al .
1989 and filippenko et al . 1995a . ) in the absence of direct information about the progenitor stars , scrutiny of the host galaxies and local environments of sne continues to yield valuable clues to their nature .
sne occur in at least three , and possibly four or more , spectroscopically distinct varieties ( e.g. , harkness & wheeler 1990 ; filippenko 1997 ) .
type i sne are defined by the absence of hydrogen in their optical spectra . the classical sne ia are characterized by the presence of a deep si ii absorption trough near 6150 in their early - time photospheric spectra ( @xmath1 month past maximum brightness ) .
members of the ib and ic subclasses , in contrast , do not show this line .
moderately strong he i lines distinguish sne ib from sne ic .
sne ii all exhibit hydrogen in their optical spectra , but the strength and profile of the h@xmath2 line vary widely among these objects ( e.g. , filippenko 1991 ; schlegel 1996 ) . at late times
( @xmath310 months past maximum brightness ) , sne ia show strong blends of co and fe emission lines .
sne ib and ic , on the other hand , are dominated by relatively unblended lines of intermediate - mass elements , such as o and ca , with sne ic exhibiting larger line widths than sne ib ( filippenko et al .
sne ii are dominated by the strong h@xmath2 emission line , but otherwise they spectroscopically resemble sne ib / c at late times .
photometrically , the light curves of sne i are all broadly similar , although sne ib and ic are somewhat fainter and redder than sne ia , whereas those of sne ii exhibit much dispersion ( filippenko 1997 , figure 3 ; patat et al .
sne ii are subclassified into `` plateau '' ( sne ii - p ) and `` linear '' ( sne
ii - l ) , based on the shape of their light curves ( barbon , ciatti , & rosino 1979 ; doggett & branch 1985 ) .
previous ground - based studies of sn host galaxies and environments have primarily concentrated on statistical results .
presumably caused by the core collapse of massive stars , sne ii have been associated with a young stellar population ( e.g. , van dyk 1992 ) .
the properties of sne ib and ic favor massive progenitors which lose their hydrogen envelopes prior to explosion ( e.g. , porter & filippenko 1987 ; filippenko 1997 ) , but their progenitors have not yet been unambiguously identified .
wolf - rayet stars have been proposed ( see branch , nomoto , & filippenko 1991 and references therein ) , but a closer association with giant h ii regions _ might _ be expected for consistency with this hypothesis ( panagia & laidler 1991 ; van dyk , hamuy , & filippenko 1996 ) .
other possibilities include the off - center explosion of a white dwarf ( branch & nomoto 1986 ) or the explosion of he stars in binary systems ( e.g. , uomoto 1986 ) .
the current paradigm of white dwarfs as progenitors of sne ia rests partly on their location in the parent galaxies .
sne ia are not _ clearly _ concentrated toward star - forming regions ( maza & van den bergh 1976 ; van dyk 1992 ; mcmillan & ciardullo 1996 ) . on the other hand , a correlation between the global h@xmath2 brightness of the galaxy and the sn ia rate has been established ( oemler & tinsley 1979 ; kennicutt 1984 ; van den bergh 1991 ) .
bartunov , tsvetkov , & filimonova ( 1994 ) conclude that sne ia occur in spiral arms with a frequency very similar to that of the massive sne ii , but this is controversial .
investigation of the local stellar and gaseous environments of sne can , in favorable cases , yield useful constraints on the ages and masses of progenitor stars and therefore resolve ambiguities and contradictions in the progenitor models .
however , most studies of this kind have been hampered by the limited spatial resolution of ground - based observations ( e.g. , thompson 1982 ; richter & rosa 1984 ; huang 1987 ; van den bergh 1991 ; panagia & laidler 1991 ; van dyk 1992 ; van dyk et al .
the superior angular resolution of the _ hubble space telescope _
( _ hst _ ) offers the potential for greater understanding of sn environments .
an investigation using data from the _ hst _ archive was begun , as reported by barth et al .
( 1996 ) , with the following goals : 1 ) to study the stellar populations in the immediate environments of sne in cases where individual stars or clusters are resolved , and to use this information to help constrain the age and mass of the progenitor star ; 2 ) to search for progenitor stars in images taken prior to sn explosions ; 3 ) to determine whether old sne are still visible ; 4 ) to augment the ground - based data on the statistical association of the different sn types with star - forming regions ; and 5 ) to measure magnitudes for sne observed `` accidentally '' while still bright , adding to existing ground - based light curves .
we can also determine how sources in the fields of sne affect the late - time sn light curves ( boisseau & wheeler 1991 ) , and the photometry can be corrected for any contamination .
as barth et al .
( 1996 ) point out , although _ hst _ images offer the tremendous advantage of higher angular resolution over similar ground - based data , the _ hst _ images were almost exclusively obtained for purposes other than the analysis of sn environments , making it difficult to assemble a statistically meaningful sample ( although , as the present study demonstrates , this issue becomes less severe as larger numbers of host galaxies are imaged ) .
additionally , one of the major limitations to a study of this kind is the small field of view of the _ hst _ cameras ; we find that many potentially useful images of sn host galaxies do not contain the sites of sne at all .
also , exposure times are often chosen for the purpose of imaging galactic nuclear regions , and the underexposed sn sites are not particularly useful .
finally , many galaxies are imaged through only one _ hst _ filter .
the environments of several sne in pre - refurbishment hst images were studied by barth et al .
( 1996 ) , but the results of this work were limited by the aberrated _ hst _ optics and by the small size of the pre - refurbishment archival database .
van dyk et al .
( 1999 ) recently examined the environment of the type ii - l sn 1979c in ngc 4321 ( m100 ) in post - refurbishment _ hst _ archive images . as a result
, they were able to place more rigorous constraints on the mass of the sn progenitor , which may have had an initial mass @xmath418 @xmath5 . from additional
_ hst _ multi - band imaging obtained 17 years after explosion , they also recovered and measured the brightness of sn 1979c .
in this paper we report on the analysis of post - refurbishment _ hst _ archive images of some additional sn environments . as compared to the barth et al .
sample , the number of environments is now significantly larger , to the point where we can begin to establish meaningful statistical results for the various sn types based on these _ hst _ data .
additionally and most importantly , a number of galaxies hae been imaged deeply and through two or more filters , primarily for cepheid - based distance measurements , and several sn sites are in these images .
therefore , for the first time , we can obtain photometry of individual stars and small clusters in sn environments , in order to constrain the ages and masses of the progenitor stars .
we obtained from the _ hst _ data archive all non - proprietary post - refurbishment images of the host galaxies of all known sne available by 1996 july 9 , except when the nominal position of the image indicated that the sn site under consideration was definitely outside of the field of view .
additionally , some datasets were not considered because the sne had very poorly known positions or unknown spectroscopic types .
we also excluded from our sample all sn environments imaged as part of the _ hst _ supernova intensive study ( sins ) program ( pi : r. p. kirshner ) .
only wfpc2 images from the archive met our selection criterion , and the vast majority were obtained as part of cycle 5 programs .
all of the images had been processed by the routine wfpc2 calibration pipeline at the space telescope science institute ( stsci ) .
the images used in this study were taken on or after 1994 april 23 , when the wfpc2 ccd temperature was adjusted . because most of the photometry we performed was in high background regions , the correction for any charge transfer efficiency effects would be on the order of , at most , a few percent , and we ignore these effects on the data . also , the near - ir quantum efficiency was stable for all observations ( see holtzman et al .
1995 ) .
table 1 lists the sne , their spectral types , host galaxies , wfpc2 detector containing the sn site , filters , observation dates , and total exposure times for the images which we have analyzed .
the spectral types were generally obtained from the online asiago observatory sn catalog ( http://athena.pd.astro.it/@xmath6supern/snean.txt ; see barbon , cappellaro , & turatto 1989 ) , although other sources , such as iau circulars and the sternberg astronomical institute online catalog ( http://www.sai.msu.su/sn/sncat/ ; see tsvetkov & bartunov 1993 ) , were also consulted . as can be seen from table 1 , a sn site is often found on one image taken through only one filter , in which case we can only describe the general appearance of the sn environment . a single image , in most cases , actually consisted of two or more exposures , which were combined to remove cosmic rays , using the combine task in the stsdas package of iraf . in cases
where the image consists of only a single exposure , we used a routine by m. dickinson which interpolates around cosmic - ray hits to remove them from the image .
in other cases , images in more than one filter are available . of these
, several cases discussed below allow us to resolve and perform photometry on individual stars and stellar objects in the sn environment .
when stellar objects could be measured in the sn environments , we employed point spread function ( psf ) fitting photometry performed by daophot ( stetson 1987 ) and allstar within iraf , which is designed to handle conditions of severe crowding , ill - defined local background , and low signal - to - noise ( s / n ) ratio .
stars were generally located on the images using daofind , with a detection threshold of 3@xmath7 , determined by the gain and read noise parameters for the image .
the principal element of the psf fitting is an accurate model psf , ideally derived from the same image as the sn environment , using a sufficient number of bright isolated stars .
this minimizes the problem of changing focus , chip distortions , spacecraft jitter , or pointing drifts , that causes psfs to vary with time and position .
however , because of the severe crowding and lack of stars of sufficient s / n ratio on the archival images we analyzed , it was generally impossible to build a good model psf from field stars in these images .
instead , we used the _
tiny tim _ routine ( krist 1995 ) to produce an artificial psf , simulated to have stellar spectral type a and psf radius @xmath8 .
the artificial psf was created at the center of the region of interest , to minimize possible systematic effects caused by the dependence of psf shape on position .
a single model is sufficiently constant over the generally small ( @xmath9 ) region to fit all stars . in order to better match the width of our model psfs with stars in the image
, we convolved the tiny tim psf with a gaussian .
however , a subtle psf mismatch would not contribute as substantially to photometric uncertainties as would low s / n ratio , crowding , and background brightness fluctuations .
in order to determine the possible systematic uncertainties in crowded field photometry , we added artificial stars with known brightnesses similar to our real objects .
then , comparing their known magnitude with the outcome of the psf fit we correct the real object magnitudes accordingly .
in general about @xmath100.1 mag is added to correct for systematic uncertainties .
count rates were converted to magnitudes using either the synthetic photometric zeropoints listed by holtzman et al .
( 1995 ; their table 9 ) or hill et al .
( 1998 ; their tables 4 and 5 ) , appropriate for a @xmath8 aperture radius .
( much of the data we could analyze using psf - fitting photometry were from the _ hst _ extragalactic distance scale key project , so for consistency we used the hill et al .
zeropoints for these key project data . )
the holtzman et al .
zeropoints were determined for the 14 @xmath11 adu@xmath12 gain state , while the hill et al .
zeropoints were determined for the 7 @xmath11 adu@xmath12 gain state .
a correction for the gain ratio to either set of zeropoints shows that they are quite consistent with each other .
when necessary , we corrected for throughput degradation , due to buildup of contaminants on the camera optics .
the psf - fitting photometry often resulted in color - color and color - magnitude diagrams ( cmds ) for the stars in the sn environment . throughout this paper
we express the magnitudes and colors in the wfpc2 synthetic magnitude system . to analyze these diagrams
, we used the theoretical isochrones for solar metallicity from bertelli et al .
( 1994 ; hereafter b94 ) , in order to constrain the ages and masses of the stars .
these isochrones were computed for johnson - cousins @xmath13 and therefore require conversion into the wfpc2 synthetic magnitude system .
we produced conversions via synthetic photometry of model spectra , using the task synphot within iraf / stsdas and the bruzual synthetic spectral atlas within synphot . in table 2
we list our conversion formulae for the magnitudes and colors of interest ( those listed in the tables in b94 ) ; the agreement with the holtzman et al .
( 1995 ) empirical conversions is generally quite good . as noted in holtzman et al . , the passbands for the various filters in both filter systems are significantly different , and some filters , such as f336w , suffer from red leaks .
magnitudes in the two filter systems tend to agree reasonably well for intermediate colors , but they diverge substantially for extremes in color . since many of the resolved stars are either very blue or very red , it is absolutely essential to perform these filter conversions . additionally , as noted in holtzman et al .
( 1995 ) , because of the difference in passbands , the extinction appropriate for the filters in each of the two systems differ .
we determined the conversions for reddening and extinction in the johnson - cousins system to the wfpc2 synthetic filter system , assuming the reddening laws of cardelli , clayton , & mathis ( 1989 ) .
these conversions are given in table 3 .
in some cases , where no resolved stars are apparent in the sn environment , we performed aperture photometry , applying the magnitude zeropoint for infinite aperture .
the aperture photometry provides a larger - scale measurement of the magnitude , and where possible , color , of the general sn environment .
additionally , in the case of some sn ia environments where stars could be resolved and subtracted , we obtained the magnitude and color of the unresolved diffuse background through aperture photometry in order to place a lower limit on the ages of the old stellar populations .
van dyk ( 1992 ) and van dyk et al .
( 1996 ) point out that a major factor affecting the results of a study of this kind is the overall uncertainty in sn positions within their parent galaxies .
this problem is particularly severe for these high - resolution _ hst _ images . in all cases
, we verified the _ hst _ image astrometry either against astrometry derived from ground - based images of the sn host galaxies or from careful visual inspection of available sn finding charts , or both .
we found that , especially when the _ hst _ fine guidance sensor lock mode was either `` coarse '' or `` gyro '' , the image reference coordinates can be grossly in error ( by 5 10 or more ) . in `` fine lock ''
mode , in particular , image coordinates are generally good to , or better than , @xmath142 ; for these images , the pixel positions of the sne sites could be determined using the image header coordinate information , correcting for known wfpc2 geometrical distortion .
for the coarse - lock and gyro mode images , we estimated the pixel positions of the sn sites either by accurate offsets from a fiducial point , such as the galactic nucleus or an unsaturated star in the _ hst _ image , or from available sn finding charts , or both . in all cases , we assign an estimated error centered on the nominal sn position and regard the error circle as the sn environment . this uncertainty radius varied from about @xmath15 in the best cases , to 10 in cases where
the sn positions are based only on nuclear offsets , adopting marsden s ( 1982 ) estimate of these errors , even if the positional accuracy of the _ hst _ image is good .
( it is our experience , from the work in van dyk et al .
1996 , that the nuclear offset positions are actually not greatly discrepant with absolute positions ; we believe that marsden s estimate is a rather conservative , yet appropriate , one . )
it is a pity that , given the high wfpc2 spatial resolution , sn sites can not be more precisely defined in the _ hst _ images .
but , even with this relative lack of precision , the fine image detail afforded by _ hst _ in many of the sn environments considered here provides unprecedented information about these sn progenitors .
here we present and discuss the results for the individual sne of each of the three main sn types .
a pair of 80-s f606w wfpc2 exposures of ngc 4303 were available .
three sne ii occurred in this galaxy : sn 1926a , sn 1961i , and sn 1964f .
unfortunately , the site of sn 1961i is not on these images .
( we consider sn 1964f in 3.1.3 . ) to find the site of sn 1926a on the _ hst _ image , we determined the position of the nucleus of ngc 4303 and used the nuclear offset , 11 w and 69 n , given in the online asiago sn catalog ; the uncertainty is assumed to be 10 .
van dyk et al . (
1996 ) found that this sn was associated with an h ii region of @xmath64 radius , offset from the nucleus by 8 w and 70 n. within the large error circle , we find that the sn appears to have occurred near two spiral arms , with several small clusters and associations of stars near the site . a small cluster to the west of the sn s nominal position contains stars with f606w magnitudes which range from about 23.4 to 24.4 . assuming a distance modulus @xmath16 ( tully 1988 ) , these imply @xmath17 to @xmath106.5 , corresponding to bright ( presumably massive ) supergiant stars .
this galaxy was observed for the _ hst _ extragalactic distance scale key project , and the archive contains a large number of f555w images , and a lesser number of f439w and f814w images .
many more images in these bands were planned , but we have used only those available before our cutoff date . we combined the individual images into a single image in each band .
this galaxy has been host to two sne : sn ia 1969h , which is not in these images , and sn ii 1940b , whose environment is shown in figure 1 .
the position of this sn is not accurate ( based on the nuclear offset , 95 e and 118 n ) , and we show the 10 error circle around the nominal position .
the sn occurred near the edge of what appears to be two associations of bright , blue stars ( also see zwicky 1965 , his plate iv ) .
we performed psf - fitting photometry on the combined images in all three bands , although we also analyzed the individual images , as a check on our results .
we show in figure 2 the color - magnitude diagrams ( cmds ) for the measured objects within the error circle ; most of them are in the bright associations . also shown
are unreddened isochrones from b94 for various ages , corrected by the distance modulus @xmath18 to ngc 4725 ( gibson et al .
the object at @xmath19 , f439w@[email protected] , f555w@[email protected] mag , in particular , is too bright to be a single star and is probably a compact star cluster .
( although we have left the isochrones unreddened , examination of the cmds shows that reddening may be appropriate for some stars ; sn 1940b itself may have been somewhat reddened ; see minkowski 1964 , de vaucouleurs 1974 , and patat et al . 1993 . )
the environment of sn 1940b contains many bright , young , blue stars , particularly in the large ob associations .
several red stars , presumably red supergiants , with possible ages of @xmath6820 myr , are also seen on the cmds .
the detected stars generally have ages , depending on the actual reddening , in the range @xmath6630 myr .
sn 1940b may have been associated with the populations of young stars ; if the sn progenitor was a red supergiant , it may have had an age similar to the detected red supergiants , but could have been as young as 6 myr .
the lack of a precise sn position , in particular , prevents us from being more restrictive about the nature of the progenitor .
we have already discussed in 3.1.1 this image , on which we also find the site of sn 1964f .
this sn has been identified in the literature as a sn i , but based on r. barbon s re - examination of the sn spectra , the asiago sn catalog lists it as a sn ii . from the nuclear offset in the asiago catalog ( 28 w and 1 s ) , we derive a sn position , near which are several bright stars or clusters .
these stars have f606w magnitudes which range from about 23.1 to 23.7 .
again , adopting @xmath16 ( tully 1988 ) , these correspond to approximate @xmath21 magnitudes of @xmath107.8 to @xmath107.2 , which are likely bright ( massive ) supergiants or , possibly , small clusters of stars ( which may be true for the brightest objects ) .
only a single f606w image of ngc 5597 is available .
the sn position ( uncertainty radius 10 ) was derived from its nuclear offset ( 17 w and 13 s ) .
the sn occurred near a spiral arm , with a number of bright stellar associations visible within the error circle , although the sn position does not closely coincide with any of them .
one pair each of a medium - band f547 m and narrow - band f658n ( h@xmath2 ) and f502n ( [ o iii ] ) images of ngc 4258 are available . sn was not spectroscopically classified , but is very likely a type ii , based on the properties of its radio emission ( van dyk et al .
it appears to have occurred along the edge of the northeast arm in this galaxy , near several large h ii complexes ( figure 3a ) .
given the observed correlation of radio emission in sne ii with late - time optical emission ( chevalier & fransson 1994 ) , we might expect sn 1981k to have still been a relatively luminous optical emission source in 1995 , when these images were made .
a very faint point - like source of possible h@xmath2 emission is seen near the bottom of the error circle , with @xmath22 mag ( adopting the stmag zeropoint from holtzman et al .
1995 ; their table 9 ) , which could be sn 1981k ( or , possibly , a faint h ii region ) . using the online `` wfpc2 exposure time calculator '' to convert a point - source count rate into flux
, we find @xmath23 erg @xmath24 s@xmath12 . for a distance to ngc 4258 of 7.3 mpc ( herrnstein et al .
1997 ) , this corresponds to a luminosity of @xmath25 erg s@xmath12 ( not corrected for reddening ) . for comparison , fesen et al .
( 1999 ) recently estimated the dereddened late - time h@xmath2 luminosity for sn 1979c to be @xmath26 erg s@xmath12 .
if this is not the sn , but instead an unrelated emission source in the environment , then we can place an upper limit on the h@xmath2 emission at the position of the sn : @xmath27 erg @xmath24 s@xmath12 , or @xmath28 erg s@xmath12 .
we see from figure 3b that the [ o iii ] emission in the field appears to be particularly associated with the cores of the bright h ii complexes .
the [ o iii ] emission within the error circle , even at the position of the faint h@xmath2 source , is extremely faint . if the h@xmath2 emission is from sn 1981k , the relative lack of [ o iii ] emission would be consistent with the shock / circumstellar matter interaction model by chevalier & fransson ( 1994 ) .
figure 3c shows the continuum counterpart of the faint h@xmath2 source , as well as some associated nebulosity or unresolved stars .
the source has @xmath29 mag , or , for the distance modulus @xmath30 , @xmath31 .
a nearby small cluster of stars can also be seen inside the northwest edge of the error circle .
the brighter stars among them have @xmath32 , @xmath33 , and @xmath34 mag ; the faint star nearby has @xmath35 mag .
these stars have @xmath36 , and @xmath37 mag , respectively .
although we do not have any color information for this environment , these magnitudes are consistent with those of supergiant stars .
the fact that little h@xmath2 emission is associated with these stars implies that they are non - ionizing , cooler , possibly yellow or red supergiants .
thus , sn 1981k may be associated with a population of evolved supergiants , with no recent star formation occurring in its immediate environment .
given that the radio emission for the sn possibly arises from the sn shock interacting with the progenitor s red supergiant wind ( van dyk et al .
1992 ) , this would also imply that the sn s progenitor was a red supergiant of relatively low mass ( @xmath38 ) .
a pair of f606w images of the highly - inclined galaxy ngc 4490 were available . from the nuclear offset ( 35 e and 20 s ) for sn ii 1982f ( from the asiago sn catalog ) , we derive a position for the sn near the edge of the pc chip .
sn 1982f occurred just outside the main body of ngc 4490 .
the stars nearest the sn position have @xmath39 , @xmath40 , and @xmath41 mag . given the distance modulus ( @xmath42 ; tully 1988 ) , these correspond to @xmath43 , @xmath44 , and @xmath45 mag , and are probably bright supergiant stars .
this sn ii belongs to a peculiar class , known as sne iin ( schlegel 1990 ) , which show narrow emission profiles atop broader bases in their spectra , particularly at h@xmath2 , possibly arising from the sn shock interacting with very dense circumstellar matter .
sn 1986j was first discovered in the radio and subsequently followed optically ( rupen , van gorkom , & gunn 1987 ; leibundgut et al .
1991 ) ; it was missed in optical sn searches near the time of explosion .
the sn is a very luminous radio source ( weiler , panagia , & sramek 1990 ) and should also still be a luminous optical emission - line source .
weiler et al .
( 1990 ) provide an accurate radio position for the sn , but , unfortunately , the coordinates for the pair of short - exposure f606w archival images of ngc 891 are greatly in error .
we located the sn site on the coadded image pair using a finding chart for sn 1986j ( rupen et al .
a star is seen in figure 4 with @xmath46 mag on 1994 dec .
1 , which corresponds to the same object seen in the rupen et al .
image , and is therefore almost certainly the sn . only faint , diffuse emission is visible around the sn , so it is not possible to study any associated stellar population .
images in f218w , f547 m , f555w and f814w are available for ngc 4651 in multiple exposures , except for the f547 m band .
we used the finding chart for the sn iib 1987k from filippenko ( 1988 ) to aid in locating the sn environment on the images .
we show the sn environment in the f555w image in figure 5 . adopting an error of 2 in the sn position
, we find that the sn occurred along a faint northern spiral arm in the galaxy .
no individual stars or clusters are detected within the error circle in the f555w and f814w images .
we measure a color f555w@xmath10f814w@xmath47 mag for the sn environment , but we note that the environment appears dusty in both bands . no emission in the sn environment
is seen in the f218w image , but it is of a low s / n ratio . in the f547 m image a bright single star with @xmath48 mag and some diffuse emission can be seen within the error circle . for a distance modulus @xmath49 to ngc 4651 ( tully 1988 ) , the star has @xmath50 mag
. this star could be the sn , seen at late times .
however , it is more likely a red supergiant star in the environment . sn 1988 m , discovered by filippenko , shields , & sargent ( 1988 ) , occurred in ngc 4496b ( @xmath51 km s@xmath12 ) , an optical double to ngc4496a ( @xmath52 km s@xmath12 ) , the host galaxy of sn ia 1960f ( see 3.3.7 ) .
ngc 4496a , b were imaged as part of a project to measure a cepheid - based distance to sn 1960f ( saha et al .
1996a ) , so a large number of f555w and f814w images are available . these were combined to provide very deep images in both bands .
we show the sn environment in the f555w image in figure 6 .
the position is based on a nuclear offset ( @xmath53 s ) , measured accurately from a two - dimensional spectrum including the sn by filippenko et al .
( 1988 ) ; the uncertainty in the measurement is likely @xmath543 .
however , comparing to ground - based images of these galaxies , the sn environment is very near the edge of the wfpc2 chip , so we conservatively assign an uncertainty of 5 to the sn position in the _ hst _ image . the sn appears to have occurred along a faint extension of a spiral arm in ngc 4496b .
several very faint stars can be barely resolved by daophot in the environment , and they appear to be quite red ; however , the extinction is unknown .
1988 m may have occurred among these red stars , either supergiants or reddened ob stars , with ages possibly as young as @xmath620 myr .
ngc 3960 ( mk 171 , arp 299 ) has been host to perhaps five sne in the past decade : 1990al ( huang et al .
1990 ) , 1992bu ( van buren et al .
1994 ) , 1993 g , 1998 t ( see 3.2.5 ) , and 1999d ( see 3.1.12 ) .
sne 1993 g ( type ii ) , 1998 t ( type ib / c ) , and 1999d ( type ii ) have been spectroscopically classified ; the other two possible sne ( 1990al and 1992bu ) have been identified only in the radio and infrared , respectively .
a single short - exposure f606w image is available for ngc 3690 . adopting the absolute position for sn 1993 g from forti ( 1993 ;
we assign an uncertainty radius 2 ) , we find that the sn appears to have occurred well outside the main body of the interacting system , in a region of faint emission with no resolvable stars . a pair of archival f606w images of ngc 3627 , which is also host to sn 1989b ( see 3.3.13 ) , were obtained .
these images were taken well before the sn iin 1997bs was discovered ( treffers et al .
although sn 1997bs has an absolute position ( cavagna 1997 ) , we used ground - based images , _ hst _ images from later cycles , and the discovery images to locate the sn environment , which we show in figure 7a .
we are able to assign a 1 radius uncertainty in the position . within the error circle
is a star with @xmath55 mag on 1994 dec .
28 . for a distance modulus @xmath56 ( saha et al . 1997 ) , this corresponds to @xmath57 mag for the star . in figures 7b and
7c we show the same field in subsequent wfpc2 f555w images made as part of program go 6549 . on 1997 nov .
12 ( figure 7b ) , nearly seven months after the sn s discovery , the star s position is coincident with that of the sn , which had @xmath58 mag . on 1998 jan .
10 ( figure 7c ) the star had faded to @xmath59 mag , below its original brightness .
thus , we conclude that this star was the progenitor of sn 1997bs .
although we have no color information for this star , its absolute magnitude is consistent with it having been an extremely luminous supergiant star .
this is only likely the fourth sn progenitor to be identified in pre - explosion images ( see 1 ) .
goodrich et al . ( 1989 ) and filippenko et al . ( 1995 ) analyze the case of sn 1961v , concluding that it was a super - outburst of a luminous blue variable star , resembling the enormous eruptions sometimes experienced by @xmath0 car .
sne with relatively similar spectra and low luminosities include sn 1999bw and sn 1997bs ; they might be additional examples of such outbursts ( filippenko , li , & modjaz 1999 ) . unlike normal sne ii , in which the progenitor destroys itself and creates a compact remnant ( neutron star or black hole ) , here the progenitor survives the explosion reasonably unharmed .
thus , in a sense these are not `` genuine '' sne .
the detection of a very luminous progenitor to sn 1997bs provides some evidence for this hypothesis , but the real test will be whether the star is still visible in future _ hst _ images obtained years after the outburst .
the sn ii 1999d recently occurred ( qiu , qiao , & hu 1999 ) in the same interacting galaxy as sn 1993 g ( 3.1.10 ) .
the sn occurred near the outer edge of the galaxy , in a region of faint diffuse emission . within a 2 error circle around the sn s position
are two unresolved star clusters , with @xmath60 and 19.8 mag , respectively , along a more inward spiral arm .
images of ngc 1365 in the f160bw , f336w , f547 m , f555w , and f814w bands are available from two different programs , one of which is the _ hst _ extragalactic distance scale key project .
unfortunately , the key project images for this galaxy did not include the site of the sn ic 1983v .
the exposure times for the two sets of uv images were too short to be useful ; we did not consider them further .
the f547 m , f555w , and f814w images have reasonable s / n ratio after coaddition , such that we could produce limited color - magnitude diagrams for the sn environment .
the absolute position from lindblad & grosbol ( 1983 ) was used to isolate the sn site , and we show the sn environment in figure 8 .
van dyk et al . (
1996 ) had found that the sn occurred within a bright h ii region of @xmath65 radius . from the _ hst _ image we see the stars and small clusters which are presumably ionizing the region .
we performed psf - fitting photometry on the f555w and f814w images ( the f547 m band does not provide any additional color information and is less sensitive in the same wavelength range than the f555w band ) . in figure 9
we show the resulting cmd .
the reddening to sn 1983v may be between @xmath61 and 0.4 ( clocchiatti et al .
we assume both values of the reddening , applying them to the theoretical isochrones .
we also apply to the isochrones the distance modulus to ngc 1365 determined from cepheids using _ hst _ , @xmath62 ( silbermann et al .
1999 ) .
the positions of the stars on the cmd in figure 9 appear to agree better with the lower reddening than the higher .
one can see that the three objects in or very near the error circle around the sn position appear too bright to be individual stars and are probably compact blue star clusters .
the resolved stars near the sn site have ages @xmath6410 myr .
we can age - date the compact clusters by using the color evolution models for starbursts from leitherer & heckman ( 1995 ) . in figure 10
we show the colors of these probable star clusters , compared to the color evolution models for an instantaneous starburst from leitherer & heckman , transformed to wfpc2 synthetic colors , and reddened by @xmath63 .
the line conventions in figure 10 are the same as figure 1 of leitherer & heckman .
these clusters are evidently bluer than the model predictions .
ideally , one would make this comparison having more color information .
however , we find that the cluster colors are consistent with very young starbursts , of ages @xmath645 myr . if the progenitor of sn 1983v was associated with these clusters , then it may have had a comparably young age , implying a very high initial mass of @xmath65 .
stars of this mass generally evolve to the wolf - rayet stage toward the end of their lives .
thus , analysis of the environment of sn 1983v is consistent with the wolf - rayet progenitor model for sne ib / c , but the data are of limited sensitivity and color information .
a pair of 80-s f606w exposures are available .
the image header coordinate information is erroneous , so we used ground - based images and the sn position given by filippenko & sargent ( 1986 ) to isolate the sn environment , which we show in figure 11 .
the sn occurred along the galaxy bar , on or near a bright h ii region ( see figure 1 in filippenko et al .
we assign an uncertainty in the sn position of 2 radius . within this error circle
are a number of bright stars .
we performed psf - fitting photometry on the immediate field , including stars somewhat outside the error circle .
the brightest stars have @xmath66 to 23.6 mag .
assuming @xmath67 for the galaxy ( tully 1988 ) , these correspond to @xmath68 to @xmath69 mag , which are consistent with , although somewhat brighter than , most luminous known supergiant and wolf - rayet stars .
the faintest stars detected within the error circle have @xmath70 mag , corresponding to @xmath71 mag , consistent with the absolute magnitudes of red supergiants .
the brightest of these objects could be multiple stars in compact clusters , since the two brightest objects within the error circle have somewhat broader profiles than those of single stars in the image .
the brightest few objects could be multiple stars in compact clusters , since they have somewhat broader profiles than those of single stars in the image .
unfortunately , we do not have any color information for this environment , to better constrain the stellar populations , and hence the age and mass of the progenitor .
the present evidence suggests that the progenitor of sn 1985f was a massive star , and is consistent with the exploding wolf - rayet star model ( e.g. , begelman & sarazin 1986 ) , but our data do not eliminate other possible models .
only a single 500-s f606w image is available for ngc 6221 .
we adopt the absolute position for the sn ic 1990w from the sternberg sn catalog ( tsvetkov & bartunov 1993 ) . in figure 12
we show the sn environment , with a 2 radius uncertainty .
the sn occurred in a very bright cluster of stars to the northeast of the nuclear region .
we performed psf - fitting photometry on these stars , taking care to distinguish cosmic - ray residuals from real stars .
the brightest star near the sn has @xmath72 mag ; the star closest to the sn position has @xmath73 mag . assuming @xmath74 for the galaxy ( tully 1988 ) , these correspond to @xmath75 and @xmath76 mag , respectively almost certainly too bright to be single stars .
fainter stars near the sn have @xmath7723.5 mag , which correspond to @xmath78 to @xmath79 mag , consistent with the brightest known supergiants and wolf - rayet stars .
unfortunately , we do not have any color information for this environment , but it appears that this sn occurred very near a massive , bright , and likely recent star formation region in ngc 6221 , implying that the sn progenitor itself was also probably a young , massive star .
only a single 500-s f606w image is available for ngc 1614 .
we used the nuclear offset , @xmath80 e ( drissen et al . 1996 ) , with a 2 radius uncertainty , to isolate the environment of the sn ic 1996d , which is shown in figure 13 .
the sn occurred along a bright spiral arm , in the vicinity of a number of stellar objects with median @xmath81 mag and several unresolved objects with @xmath82 to 20.8 mag . for a distance modulus to ngc 1614 of @xmath83 ( assuming @xmath84 km s@xmath12 and @xmath85 km s@xmath12 mpc@xmath12 ) , these apparent magnitudes correspond to @xmath86 mag for the fainter objects , and @xmath87 to -13.2 mag for the brighter probable star clusters .
the image data do not allow us to be more specific about the stellar populations in the sn site , other than that it is likely these are recently - formed young , massive stars and star clusters .
the sn ib 1998 t occurred ( li , li , & wan 1998 ) in the same interacting host galaxy as sne 1993 g and 1999d ; we show the sn environment in figure 14 .
we tried to isolate the progenitor in this pre - sn image . unfortunately , given the distance to ngc 3690 ( @xmath642 mpc , for @xmath85 km s@xmath12 mpc@xmath12 and @xmath88 km s@xmath12 ) and the short exposure time , no individual stars can be resolved .
however , from the accurate position of the sn ( @xmath89 @xmath90 , j2000 , from an @xmath91-band image taken with the lick observatory 0.75-m katzman automatic imaging telescope ; see also yamaoka et al .
1998 ) with a 2 radius uncertainty , we see that the sn occurred among the bright , dense concentrations of star formation occurring in this galaxy .
this single image of the sn environment implies that the sn progenitor was likely a massive star .
several f255w images and a single 500-s f606w image are available for this galaxy before our cutoff date .
the galaxy has been the host of two bright sne ia , 1895b and 1972e .
unfortunately , the site of sn 1972e was not in either set of images .
we consider here the environment of sn 1895b , adopting the absolute position ( uncertainty 2 radius ) of the sn from caldwell & phillips ( 1989 ) . in the f255w image
the sn is far from the very active star formation taking place near the center of the galaxy , as traced by the uv - bright stars . from psf - fitting photometry of the f606w image ( see gorjian 1996 ) we find that the stars near the sn site have magnitudes @xmath92 mag ; the faintest among these have @xmath93 mag . the fact that these stars have no detectable counterparts on the deeper uv image implies that they are probably red .
assuming the distance modulus ( @xmath94 ) to ngc 5253 determined using cepheids by saha et al .
( 1995 ) , these stars have @xmath95 to @xmath96 mag , consistent with bright red giants and fainter supergiants .
a pair of 80-s f606w exposures of ngc 1003 are available .
we used the nuclear offset position for sn 1937d from the online asiago catalog ( 48 e and 1 s ) .
the sn appears to have occurred along a spiral arm between several stellar associations , although the inclination is high ( 68 ) for this galaxy .
three images of ngc 4621 in each of f555w and f814w were obtained .
we used the nuclear offset position for sn 1939b ( 53
s ) from the online asiago catalog ; it occurred toward the edge of the galaxy , away from the central region . several faint stars or clusters , as well as an unresolved background of stars ,
are within the large ( 10 radius ) error circle . using psf fitting we have subtracted the foreground stars and measured the color of the unresolved light within the error circle , which is f555w@xmath10f814w=1.37 mag , consistent with a background of k - giant stars .
it is therefore most likely that this sn i was a sn ia , rather than a sn ib / c ; the host galaxy is of early type , and the sn did not occur in or near a region of recent star formation .
the progenitor of sn 1939b , based on the color of the environment , was likely old and evolved .
a pair of f656n ( figure 15 ) and f675w images of ngc 253 were available .
we used the nuclear offset position for the sn i 1940e ( 51 w and 17 s ) from the online asiago catalog .
the sn occurred near the edge of dusty lanes or patches , among a field of diffuse starlight .
the fact that this sn site is not near a dense stellar association or a region of h@xmath2 emission makes it likely that the sn was a type ia . a single 230-s f547 m image and a pair of 700-s f218w images are available for ngc 5195 .
we used the nuclear offset position for the sn i 1945a ( 6 w and 4 s ) from the online asiago catalog .
dust lanes cut across the diffuse starlight in the sn environment .
the deeper f218w image is completely blank in this field , indicating no detectable uv emission from stars or nebulae in this environment , and implying either a possible lack of recent star formation or appreciable extinction , or both .
the extinction toward ngc 5195 has been estimated to be between @xmath97 ( thronson , rubin , & ksir 1991 ) and @xmath98 ( smith et al .
1990 ) , at least partly due to dust from the arm of ngc 5194 obscuring the nucleus of ngc 5195 ( zwicky 1957 ; van dyk 1987 ) .
based on the environmental evidence , it is not certain to which subclass sn 1945a belongs ; however , its probable lack of association with massive star formation suggests a sn ia . as is the case for sn 1979c ( van dyk et al .
1999 ) , the environment of the sn i 1959e in m100 can be analyzed using the deep _ hst _ key project images in f555w and f814w . in figure 16
we show the sn environment , using the absolute position from porter ( 1993 ) .
the sn site is seen against a dust lane in this deep image . within the @xmath15 radius error circle
four stars are detected with magnitudes @xmath99 from 24.1 to 25.5 ( @xmath100 to @xmath101 for distance modulus @xmath102 ; ferrarese et al .
1996 ) and colors f555w@xmath10f814w from 0.1 to 1.3 mag ( these stars are likely cool or reddened supergiants ) .
we performed psf - fitting photometry on the region of the galaxy beyond the error circle , out to @xmath67 from the sn position ; in figure 17 we show the resulting cmd .
we also show the b94 isochrones , reddened only by the galactic foreground ( burstein & heiles 1984 ) , although the reddening must clearly be larger than this .
a large number of populations are revealed from these deep images , from quite blue to very red , including hot main sequence stars , red supergiants , and some bright red giants .
reddening and crowding are undoubtedly affecting the observed stellar magnitudes and colors .
however , we see that the photometry does not probe very deep in age ( only to @xmath6100 myr ) , in part because of the dust and the relatively bright background .
most of the measured stars appear to have ages @xmath62050 myr .
after all resolved stars were subtracted away , we estimated the magnitude and measured the color of the diffuse background and find @xmath103 mag ( the magnitude of pixels 1@xmath7 above the mean background ) and f555w@xmath10f814w@xmath104 mag .
this is consistent with a background of g stars , presumably giants with ages @xmath105 myr , or possibly fainter , reddened early - type stars ( considering the extensive dust lanes in this environment ) . from these results , we can not rule out that sn 1959e was a sn ib / c , with a massive progenitor ; however , we consider it more likely that it was a sn ia , based on its environment .
as is the case for sn 1988 m in 3.1.9 , the environment of the sn ia 1960f is in the deep f555w and f814w images , taken as part of a project to measure a cepheid - based distance to sn 1960f ( saha et al .
seventeen epochs of f555w imaging and four epochs of f814w imaging were obtained , each set of images consisting of a pair of 2000-s exposures .
the method for analyzing these images was the same as for the environment of sn 1979c ( see van dyk et al .
1999 ) : a model psf was produced , stars were located , photometry was performed with daophot and allstar , and the results were averaged over all epochs .
faint stars that were not found in the individual images were located in summed images , and the resulting photometry was appended to the results for the individual images .
we performed the photometry over a large area of the image .
although crowding was severe in some regions , the photometry with a tinytim psf led to satisfactory results .
the sn environment on the summed f555w image is shown in figure 18 .
we used the accurate absolute position for the sn from porter ( 1993 ) .
the sn occurred in a region between two spiral arms in ngc 4496a .
we show in figure 19 the cmd produced by our photometry of the larger environment around the sn , along with the b94 isochrones , adjusted for both the reddening ( @xmath106=0.04 $ ] ) and the distance modulus ( @xmath107 ) from saha et al .
( 1996a ) , and transformed to wfpc2 synthetic magnitudes .
a large number of populations is revealed in these deep images , from quite blue to very red , including hot main sequence stars , red supergiants , and the tip of the red giant branch .
reddening and crowding are undoubtedly affecting the observed stellar magnitudes and colors
. however , even these images are not deep enough to see populations with ages @xmath108500 myr , from which sne ia are thought to arise .
after all resolved stars were subtracted away , we estimated a magnitude and measured a color for the environment , through a @xmath15 aperture centered on the sn position , of @xmath109 mag ( the magnitude of pixels 1@xmath7 above the mean background ) and f555w@xmath10f814w@xmath110 mag , consistent with a background of late f - stars , presumably giants with ages @xmath111 myr , or possibly fainter , reddened early - type stars . four f606w exposures of 3c 84 ( ngc 1275 ) are available . using the nuclear offset , 7 e and 24 s , from the online asiago catalog , the site of the sn i 1968a is near the outer edge of the galaxy and the wfpc2 chip .
only diffuse emission is seen near the sn site , and only two stellar - like objects are seen within the 10 radius error circle .
these two objects have magnitudes @xmath112 and 23.3 mag ; for a distance of 70.2 mpc ( from @xmath113 km s@xmath12 and @xmath85 km s@xmath12 mpc ) , these objects of @xmath114 and @xmath1014.0 mag are almost certainly compact clusters , more likely globular clusters .
we do not have any color information , but from the lack of detected recent star formation at the sn site , it is most likely that this sn was a type ia , rather than type ib / c .
only a single f606w image of ngc 5427 is available .
we used the nuclear offset position for the sn ( 35 e and 34 n ) from the online asiago catalog .
the sn site is very near the edge of the chip , near a spiral arm . within the 10 radius error circle
are a number of bright nebulae , presumably h ii regions and star clusters .
the site of this well - studied sn ia is not in the deep sandage et al .
_ hst _ cepheid images ( saha et al .
1996b ) of ngc 4536 , but instead is in shallower f555w , f675w , f814w , and f658n exposures taken for another project .
sn 1981b occurred well beyond the galaxy s spiral arms , near several faint h ii regions ( figure 20 ) .
we have performed psf - fitting photometry of the stars in the three broad - band images within @xmath610 of the sn site .
unfortunately , the resulting cmds do not go sufficiently deep to constrain the progenitor population for this sn ia , since the oldest population detected is apparently only @xmath115 myr .
after the stars were subtracted away from the broad - band images , we estimated the colors and magnitude of the environment within the error circle .
we find that f555w@xmath10f675w@xmath116 mag , f555w@xmath10f814w@xmath117 mag , and @xmath118 mag .
the f555w@xmath10f675w color almost certainly must be contaminated by h@xmath2 emission in the environment .
the f555w@xmath10f814w color is consistent with late f - type stars , indicating a possible mixture of older red and younger blue stars in the environment .
only a single f606w pc image of ngc 3227 is available . using the absolute position of the sn ia 1983u from the sternberg sn catalog , we find that the sn site is along a faint extension of a spiral arm , far from the brighter stellar emission near the galaxy nucleus , and near ( but not in ) the dust lanes .
a pair of 80-s f606w exposures of ngc 4772 are available .
sn 1988e was possibly a peculiar sn i ( pearce et al .
1990 ) . using the absolute sn position from the sternberg sn catalog ,
we find that the sn site is in a region of faint , diffuse emission on the outskirts of this elliptical galaxy , with no stellar objects within the 2 radius error circle .
a pair of f606w images of ngc 3627 were obtained ( see 3.1.11 ) .
we show the sn ia 1989b environment in figure 21 , based on the absolute sn position from the sternberg sn catalog .
the sn occurred along an inner spiral arm in an environment of bright stellar objects , diffuse emission , and dust , consistent with the relatively large reddening ( @xmath119=0.37 $ ] mag ) found by wells et al .
( 1994 ) from the sn light curves .
the resolved stars within the 2 radius error circle have @xmath120 to 20.9 mag , which for a distance modulus @xmath56 ( 3.1.11 ) , correspond to @xmath121 to @xmath109.4 mag ; these are either extremely luminous single stars or , more likely , small compact star clusters .
images in f218w , f300w , f439w , and f547 m of ngc 4639 are available .
unfortunately , the deeper and more numerous sandage et al .
cepheid project images of this galaxy ( saha et al .
1997 ) do not contain the sn site , due to the site s proximity to a very bright foreground star . however , from the images we obtained , using the absolute sn position from the sternberg sn catalog , we find that the sn occurred near the edge of the galaxy along an outer spiral arm
. the images in each band are not deep enough to resolve any individual bright stars or clusters within the @xmath15 radius error circle , nor for us to measure the colors of the environment .
deep images in f336w , f555w , f675w , and f814w are available for ngc 7742 .
sn 1993r is classified as peculiar type i ( filippenko & matheson 1993 ) , with substantial differences from the late - time ( nebular ) spectra of normal sne ia .
the spectrum of sn 1993r somewhat resembles that of the subluminous sn ia 1991bg ( filippenko et al . 1992 ; leibundgut et al .
1993 ) , yet the ca ii near - infrared triplet is strong in sn 1993r and very weak in sn 1991bg .
the spectrum differs from the nebular spectra of sne ib and sne ic , in that the [ o i ] 6300 emission is weak in sn 1993r .
in addition , the sn appeared to be photometrically subluminous near maximum brightness .
an approximate absolute position was given for sn 1993r by treffers et al .
( 1993 ) . we have measured a more accurate absolute position from an @xmath122-band image taken by b. leibundgut and w. vacca at the lick observatory nickel 1-m telescope on 1993 august 18 ut . using the _ hst _ _
guide star catalog _
positions for foreground stars seen on the image , we derive a position of @xmath123 , @xmath124 ( 2000.0 ) .
the image header coordinate information for the _ hst _ images is erroneous , so , given the sn s absolute position and the absolute position ( also from the lick image ) of an unsaturated foreground star on the _ hst _ images , we located the sn site on the images using the offset between the sn and the star .
we adopt a positional uncertainty of 1 for this offset .
we show the sn environment in figure 22 . from the measured position , the sn occurred along the bright circumnuclear ring in the galaxy , in a region of patchy emission and dust .
filippenko & matheson ( 1993 ) reported that the sn is superposed on a very bright h ii region .
this region appears as a bright knot just outside the error circle for the sn in the _ hst _ images and is not directly coincident with the sn location .
this likely compact cluster has @xmath125 mag , f336w@[email protected] , f555w@xmath10f675w=0.71 , and f555w@xmath10f814w=0.01 mag . for a distance modulus @xmath126 ( tully 1988 )
, it has @xmath127 mag and colors consistent with young , blue stars ( the f555w@xmath10f675w color is almost certainly contaminated by h@xmath2 emission ) .
two stellar objects just outside the error circle have , respectively , @xmath128 mag , f336w@[email protected] , f555w@xmath10f675w=0.54 , and f555w@xmath10f814w=0.71 mag ; and , @xmath129 , f336w@[email protected] , f555w@xmath10f675w=0.15 , and f555w@xmath10f814w=0.11 mag .
the former object appears to be either a very bright red star or reddened compact cluster ; the latter is likely a blue compact cluster . removing these stellar objects from the images and measuring the colors and magnitude for the environment ,
we find @xmath130 , f336w@xmath10f555w=0.10 , f555w@xmath10f675w=0.67 , and f555w@xmath10f814w=0.93 mag . these colors are significantly redder than the individual objects and are consistent with a mixture of a through k stars , but also may be due to reddened fainter blue populations .
a set of short- and long - exposure f555w and f814w images are available .
sn ia 1994d is clearly seen on the _ hst _ images ; it occurred along the edge of the central dusty disk of this host galaxy , just beyond or north of the prominent dust lanes ( figure 23 ) .
although the sn is saturated in the long - exposure images , only the central pixels of the sn profile are saturated in the shorter exposures .
we have measured the magnitudes of the sn in the two _ hst _ filters by using the wings of the stellar profile and interpolating the flux lost in the saturated portion of the profile .
we used the unsaturated magnitudes of the field star ( which is `` star c '' in richmond et al .
1995 ) , also seen on the image , as a reference .
we find @xmath131 and @xmath132 mag for sn 1994d on 1994 may 9 ut .
these values agree well with @xmath133 and @xmath134 mag for 1994 may 8 ut by richmond et al .
( our measurements for the field star magnitudes , @xmath135 and @xmath136 mag , also agree well with richmond et al.s magnitudes for this star . )
the sn environment is dominated by diffuse starlight in the host galaxy .
we averaged the measurements of the magnitude and color of the sn environment on the deeper images through two @xmath8-radius apertures , placed on either side of ( and outside ) the wings of the saturated sn psf .
we find that @xmath137 and f555w@xmath10f814w=1.52 mag in the environment , which is quite red .
undoubtedly , dust must be contributing to the colors of this environment , but taken at face value , the color implies the presence of k - type stars , presumably giants , in the sn environment in this early - type host galaxy .
we have conducted an analysis of sn environments in post - refurbishment _ hst _ archival data . due to the superior resolution of _ hst _ , the results of this study are a significant improvement over those involving ground - based data or our previous study of pre - refurbishment _ hst _ data ( barth et al . 1996 ) .
we have accomplished several of our main goals set out for this project .
( 1 ) for the first time , through deep multicolor imaging undertaken by the _ hst _ cepheid projects , in particular ,
we have been able to produce color - color and cmds for several sn environments , providing constraints on the ages and masses of stellar populations associated with the sne , and allowing us to indirectly infer the ages and masses of the sn progenitors ; this is true for sne 1940b , 1983v , 1987k , and , to a lesser degree , 1988 m .
( 2 ) we have identified the stellar progenitor of sn 1997bs , clearly visible on a pre - sn _ hst _ image , and we find that the progenitor was a luminous supergiant star .
( 3 ) we have searched for old sne on the images , being successful in the cases of sne 1979c ( see van dyk et al .
1999 ) , 1986j , and possibly 1981k , as expected , since these three objects showed evidence of interaction with circumstellar gas .
if the late - time light curves of other sne declined exponentially , many or most should be well below detectability . also , several sne had substantially more complicated environments and imprecise positions , thwarting our attempts to detect them at late times .
( 4 ) we can , based on the archival data , advance the statistical association of the different sn types with star - forming regions .
( 5 ) in the case of sn 1994d , which was caught `` by accident '' on _ hst _ images , we can measure its brightness , augmenting existing ground - based light curves for the sn .
the immediate environments of four sne ia considered above contained bright clusters and h ii regions with which these sne ia might have been associated .
however , van dyk ( 1992 ) has shown that sne ia are more likely chance superposed on recent star formation regions than physically associated .
we expect that sn ia progenitors , based on the current models , should be considerably older and fainter . in three cases
, we were able to do detailed analyses , based on cmds , of the younger stars in the environments . removing the younger stars , through psf fitting , allowed us to measure the colors of the underlying diffuse stellar emission , which may be due to populations similar to those of the sn ia progenitors .
the colors of the diffuse emission provide suggestive indications that the stellar populations are generally old and red , consistent with red giants .
since we are unlikely to measure the intrinsically faint individual stars in the stellar populations associated with sn ia progenitors , regardless of the depth of the _ hst _ exposures ( except possibly for the cases of local group galaxies ) , we hope in the future to continue to amass statistics which can place more stringent constraints on the nature of sn ia progenitors .
the four sne ib and ic that we have included in this study seem more closely associated with brighter , more massive stellar regions than is the case for the sne ii in our sample , which tend to be associated with smaller , fainter stellar associations .
van dyk et al .
( 1996 ) , from their ground - based survey , found no difference in the association of the two sn types with massive stellar regions .
we caution that the conclusions in the current paper are based on very few sne ib / c and a small number of sne ii ; we clearly need to accumulate better statistics to confirm or refute this trend .
but , taken at face value , the high - resolution _ hst _ images suggest that the sne ib / c progenitors may be more massive , in general , than the sne ii progenitors .
exploding wolf - rayet stars have been proposed as possible progenitors for sne ib / c , and our limited results are consistent with this scenario .
fortunately , the environments of several sne in our study have recently been imaged as part of other _ hst _ projects , for longer exposure times and in various bands .
we will continue to acquire images for these , and other environments , as they become available in the future and will build on the current study in future papers . with the ever - increasing number of images filling the _ hst _ archive , we intend to enhance our sample significantly .
moreover , as new relatively nearby sne occur , we will hunt directly for the sn progenitors on pre - sn images .
financial support for this work was provided by nasa through grant nos .
ar-5793 , ar-6371 , ar-8006 , and go-6043 from the space telescope science institute , which is operated by aura , inc .
, under nasa contract no . nas 5 - 26555 ; we also acknowledge nsf grant ast-9417213
. we are grateful to jay anderson for assisting us with measuring the brightness of sn 1994d .
we appreciate fruitful consultations with craig sosin and adrienne cool .
we also appreciate assistance from john biretta at stsci on wfpc2 image header coordinate information .
svd is grateful to the division of astronomy & astrophysics at ucla , especially to jean turner .
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 .
aldering , g. , humphreys , r. m. , & richmond , m. 1994 , , 107 , 662 barbon , r. , ciatti , f. , & rosino l. 1979 , , 72 , 287 barbon , r. , cappellaro , e. , & turatto , m. 1989 , , 81 , 421 barth , a. j. , van dyk , s. d. , filippenko , a. v. , leibundgut , b. , & richmond , m. w. 1996 , , 111 , 2047 bartunov , o. s. , tsvetkov , d. yu .
, & filimonova , i. v. 1994 , , 106 , 1276 begelman , m. c. , & sarazin , c. l. 1986 , , 302 , l59 bertelli , g. , bressan , a. , chiosi , c. , fagotto , f. , & nasi , e. 1994 , , 106 , 275 ( b94 ) boisseau , j. r. , & wheeler , j. c. 1991 , , 101 , 1281 branch , d. , & nomoto , k. 1986 , , 164 , l13 branch , d. , nomoto , k. , & filippenko , a. v. 1991 , comm . ap . , xv , 221 burstein , d. , & heiles , c. 1984 , , 54 , 33 caldwell , n. , & phillips , m. m. 1989 , , 338 , 789 cavagna , m. 1997 , 6632 cardelli , j. a. , clayton , g. c. , & mathis , j. s. 1989 , , 345 , 245 chevalier , r. a. , & fransson c. 1994 , , 420 , 268 clocchiatti , a. , et al .
1997 , , 483 , 675 cohen , j. g. , darling , j. , & porter , a. c. 1995 , , 110 , 308 de vaucouleurs , g. 1974 , in supernovae & supernova remnants , ed . c. b. cosmovici ( dordrecht : d. reidel ) , 203 doggett , j. b. , & branch , d. 1985 , , 90 , 2303 drissen , l. , robert , c. dutil , y. , roy , j .-
r . , filippenko , a. v. , cappellaro , e. , & patat , f. 1996 , 6317 ferrarese , l. , et al .
1996 , , 464 , 568 fesen , r. a. , et al . 1999 , , 117 , 725 filippenko , a. v. 1988 , , 96 , 1941 filippenko , a. v. 1991 , in sn 1987a and other supernovae , ed .
i. j. danziger and k. kjr ( garching : eso ) , 343 filippenko , a. v. 1997 , , 35 , 309 filippenko , a. v. , barth , a. j. , bower , g. c. , ho , l. c. , stringfellow , g. s. , goodrich , r. w. , & porter , a. c. 1995a , , 110 , 2261 [ erratum : 112 , 806 ( 1996 ) ] filippenko , a. v. , & matheson , t. 1993 , 5842 filippenko , a. v. , porter , a. c. , sargent , w. l. w. , & schneider , d. p. 1986
, , 92 , 1341 filippenko , a. v. , & sargent , w. l. w. 1986 , , 91 , 691 filippenko , a. v. , shields , j. c. , & sargent , w. l. w. 1988 , , 100 , 1233 filippenko , a. v. , et al .
1992 , , 104 , 1543 filippenko , a. v. , et al .
1995b , apj , 450 , l11 filippenko , a. v. , li , w. d. , & modjaz , m. 1999 , 7152 forti , g. 1993 , 5719 fransson , c. , & chevalier , r. a. 1989 , , 343 , 323 fransson , c. , et al .
1989 , apj , 336 , 429 gibson , b. k. , et al . 1999 , , 512 , 48 goodrich , r. w. , stringfellow , g. s. , penrod , g. d. , & filippenko , a. v. 1989 , , 342 , 908 gorjian , v. 1996 , , 112 , 1886 harkness , r. p. , & wheeler , j. c. 1990 , in supernovae , ed .
a. petschek ( springer - verlag ) , 1 herrnstein , j. , moran , j. , greenhill , l. , inoue , m. , nakai , n. , miyoshi , m. , & diamond , p. 1997
, , 191 , 2507 hill , r. j. , et al .
1998 , , 496 , 648 holtzman , j. a. , burrows , c. j. , casertano , s. , hester , j. j. , trauger , j. t. , watson , a. m. , & worthey , g. 1995 , , 107 , 1065 huang , y .- l .
1987 , , 99 , 461 huang , z. p. , condon , j. j. , yin , q. f. , & thuan , t. x. 1990 , 4988 kennicutt , r. c. 1984 , , 277 , 361 krist , j. 1995 , in calibrating _ hubble space telescope _ : post servicing mission ( baltimore : stsci ) , 311 leibundgut , b. , kirshner , r. p. , pinto , p. a. , rupen , m. p. , smith , r. c. , gunn , j. e. , & schneider , d. p. 1991
, , 372 , 531 leibundgut , b. , et al .
1993 , , 105 , 301 leitherer , c. , & heckman , t. m. 1995 , , 96 , 9 li , w .- d . , li , c. , & wan , z. 1998 , 6830 lindblad , p. o. , & grosbol , p. 1983 , 3897 marsden , b. g. 1982 , 3746 maza , j. , & van den bergh , s. 1976 , , 204 , 519 mcmillan , r. j. , & ciardullo , r. 1996 , , 473 , 707 minkowski , r. 1964 , , 2 , 247 oemler , a. , & tinsley , b. m. 1979 , , 84 , 985 panagia , n. , & laidler , v. , 1991 , in supernovae , ed . s. woosley ( new york : springer - verlag ) , 559 patat , f. , barbon , r. , cappellaro , e. , & turatto , m. 1993 , , 98 , 443 patat , f. , barbon , r. , cappellaro , e. , & turatto , m. 1994 , , 282 , 731 pearce , g. , patchett , b. , murdin , p. , & allington - smith , j. 1990 , , 164 , 59 porter , a. c. 1993 , , 105 , 1250 porter , a. c. , & filippenko , a. v. 1987 , , 93 , 1372 qiu , y. l. , qiao , q. y. , & hu j. y. 1999 , 7088 richmond , m. w. , et al . 1995 , , 109 , 2121 richter , o .- g . , & rosa , m. 1984 , , 140 , l1 rupen , m. p. , van gorkom , g. r. , & gunn , j. e. 1987 , , 94 , 61 ryder , s. , staveley - smith , l. , dopita , m. , petre , r. , colbert , e. , malin , d. , & schlegel , e. 1993 , , 416 , 167 saha , a. , sandage , a. , labhardt , l. , schwengeler , h. , tammann , g. a. , panagia , n. , & macchetto , f. d. 1995 , , 438 , 8 saha , a. , sandage , a. , labhardt , l. , tammann , g. a. , macchetto , f. d. , & panagia 1996a , , 107 , 693 saha , a. , sandage , a. , labhardt , l. , tammann , g. a. , macchetto , f. d. , & panagia 1996b , , 466 , 55 saha , a. , sandage , a. , labhardt , l. , tammann , g. a. , macchetto , f. d. , & panagia 1997 , , 486 , 1 schlegel , e. m. 1990 , , 244 , 269 schlegel , e. m. 1996 , , 111 , 1660 silbermann , n. a. , et al . 1999 , , 515 , 1 smith , j. , gehrz , r. d. , grasdalen , g. l. , hackwell , j. a. , dietz , r. d. , & friedman , s. d. 1990 , , 362 , 455 stetson , p. b. 1987 , , 99 , 191 thompson , l. a. 1982 , , 257 , l63 thronson , h. a. , rubin , h. , & ksir , a. 1991 , , 252 , 550 treffers , r. r. , leibundgut , b. , filippenko , a. v. , & richmond , m. w. 1993 , 5812 treffers , r. r. , peng , c. y. , filippenko , a. v. , & richmond , m. w. 1997 , 6627 tsvetkov , d. yu . , & bartunov , o. s. 1993 , bull .
centre donnees stellaires , 42 , 17 tully , r. b. 1988 , nearby galaxies catalog ( cambridge univesity press , cambridge ) uomoto , a. 1986 , , 310 , l35 van buren , d. , jarrett , t. , terebey , s. , beichman , c. , shure , m. , & kaminski , c. 1994 , 5960 van den bergh , s. 1988 , in sn shells and their birth events , ed .
w. kundt ( berlin : springer ) , 184 van den bergh , s. 1991 , , 101 , 845 van dyk , s. d. 1987 , , 99 , 467 van dyk , s. d. 1992 , , 103 , 1788 van dyk , s. d. , weiler , k. w. , sramek , r. a. , & panagia , n. 1992 , , 396 , 195 van dyk , s. d. , hamuy , m. , & filippenko , a. v. 1996 , , 111 , 2017 van dyk , s. d. , et al . 1999 , , 111 , 313 walborn , n. r. , lasker , b. m. , laidler , v. g. , & chu , y .- h . 1987 , , 321 , l41 walborn , n. r. , prevot , m. l. , prevot , l. , wamsteker , w. , gonzalez , r. , gilmozzi , r. , & fitzpatrick , e. l. 1989 , , 219 , 229 weiler , k. w. , panagia , n. , & sramek , r. a. 1990 , , 364 , 611 wells , l. a. , et al .
1994 , , 108 , 2233 west , r. m. , lauberts , a. , jorgensen , h. e. , & schuster , h. e. 1987 , , 177 , l1 yamaoka , h. , kato , t. , filippenko , a. v. , & van dyk , s. d. 1998 , 6859 zwicky , f. 1957 , morphological astronomy ( berlin : springer ) , 200 zwicky , f. 1965 , in stars and stellar systems , vol .
viii , ed .
l. h. aller & d. b. mclaughlin ( chicago : univ .
chicago press ) , 367 lcccclc ngc 253 & 1940e & i & wf4 & f675w & 1994 sep 17 & 400 & & & & f656n & & 2400 ngc 891 & 1986j & iin & wf2 & f606w & 1994 dec 1 & 160 ngc 1003 & 1937d & ia & wf4 & f606w & 1994 nov 5 & 160 ngc 1275 & 1968a & i & wf2 & f606w & 1994 mar 31 & 560 ngc 1365 & 1983v & ic & wf3 & f547 m & 1995 jan 15 & 200 & & & & f555w & & 300 & & & & f814w & & 300 ngc 1614 & 1996d & ib / c & pc & f606w & 1994 dec 11 & 500 ngc 3227 & 1983u & ia & pc & f606w & 1995 feb 23 & 500 ngc 3627 & 1989b & ia & wf2 & f606w & 1994 dec 28 & 160 & 1997bs & iin & wf4 & & & ngc 3690 & 1993 g & ii & wf2 & f606w & 1994 sep 17 & 500 & 1998 t & ib & pc & 1999d & ii & wf4 ngc 4258 & 1981k & ii & wf3 & f547 m & 1995 mar 16 & 1160 & & & & f502n & & 2300 & & & & f658n & & 2300 ngc 4303 & 1926a , & ii - l & wf4 & f606w & 1994 jun 6 & 160 & 1964f & ii & wf2 & & & ngc 4321 & 1959e & i & wf2 & f555w & 1994 apr 23jun 19 & 21600 & & & & f814w & & 7200
ngc 4490 & 1982f & ii - p & pc & f606w & 1994 dec 3 & 160 ngc 4496a / b & 1960f & ia & wf4 & f555w & 1994 may 27aug 7 & 68000 & & & & f814w & & 16000 & 1988 m & ii & wf2 & & & ngc 4526 & 1994d & ia & pc & f555w & 1994 may 9 & 60 , 230 & & & & f814w & & 60 , 230 ngc 4536 & 1981b & ia & wf4 & f555w & 1994 may 12 & 660 & & & & f675w & & 280 & & & & f814w & & 660 & & & & f658n & & 3000 ngc 4618 & 1985f & ib & pc & f606w & 1995 feb 16 & 160 ngc 4621 & 1939b & i & wf4 & f555w & 1995 feb 5 & 1050 & & & & f814w & & 1050 ngc 4651 & 1987k & iib & wf2 & f555w & 1994 may 20 & 600 & & & & f814w & & 600 & & & pc & f218w & 1995 mar 4 & 1800 & & & & f547 m & & 300 ngc 4725 & 1940b & ii - p & wf2 & f439w & 1995 apr 12jun 15 & 2500 & & & & f555w & & 30000 & & & & f814w & & 10000 ngc 4772 & 1988e & ipec & wf3 & f606w & 1994 aug 13 & 160 ngc 5195 & 1945a & i & pc & f218w & 1994 oct 8 & 1400 & & & & f547 m & & 230 ngc 5253 & 1895b & ia & wf3 & f255w & 1995 may 29 & 6900 & & & pc & f606w & 1995 mar 11 & 500 ngc 5427 & 1976d & ia & wf4 & f606w & 1994 jun 25 & 500 ngc 5597 & 1981e & ii & wf2 & f606w & 1994 jul 15 & 500 ngc 6221 & 1990w & ic & pc & f606w & 1995 apr 2 & 500 ngc 7742 & 1993r & ipec & wf3 & f336w & 1995 jul 9 & 2100 & & & & f555w & & 480 & & & & f675w & & 480 & & & & f814w & & 680 cccccc f555w & @xmath21 & @xmath138 & 0.066 & @xmath100.026 & @xmath100.004 f336w@xmath10f439w & & @xmath139 & 1.204 & @xmath100.233 & @xmath100.074 f439w@xmath10f555w & & @xmath138 & 1.024 & @xmath1400.106 & @xmath1400.016 f555w@xmath10f675w & & @xmath141 & 1.329 & @xmath100.152 & @xmath100.003 f555w@xmath10f814w & & @xmath142 & 1.043 & @xmath100.005 & @xmath100.014 ccccc @xmath143 & 4.794(@xmath540.005 ) & @xmath100.011(@xmath540.001 ) @xmath144 & 4.073(@xmath540.006 ) & @xmath100.061(@xmath540.001 ) @xmath145 & 3.148(@xmath540.002 ) & @xmath100.027(@xmath540.001 ) @xmath146 & 2.615(@xmath540.004 ) & @xmath100.052(@xmath540.001 ) @xmath147 & 1.917(@xmath540.001 ) & @xmath100.013(@xmath540.000 ) @xmath148 & 5.591(@xmath540.053 ) & @xmath100.454(@xmath540.011 ) @xmath149 & 4.301(@xmath540.006 ) & @xmath100.054(@xmath540.001 ) @xmath150 & 3.199(@xmath540.004 ) & @xmath100.046(@xmath540.001 ) @xmath151 & 2.474(@xmath540.001 ) & @xmath100.015(@xmath540.000 ) @xmath152 & 1.910(@xmath540.003 ) & @xmath100.033(@xmath540.001 ) @xmath153 & 0.720(@xmath540.005 ) & @xmath1400.051(@xmath540.001 ) @xmath154 & 0.533(@xmath540.001 ) & @xmath1400.025(@xmath540.000 ) @xmath155 & 0.698(@xmath540.003 ) & @xmath100.040(@xmath540.001 ) @xmath156 & 1.231(@xmath540.002 ) & @xmath100.015(@xmath540.000 ) @xmath157 & 1.055(@xmath540.026 ) & @xmath100.165(@xmath540.013 ) @xmath158 & 1.102(@xmath540.004 ) & @xmath100.008(@xmath540.001 ) @xmath159 & 0.725(@xmath540.003 ) & @xmath100.032(@xmath540.001 ) @xmath160 & 0.564(@xmath540.002 ) & @xmath1400.018(@xmath540.000 ) @xmath161 & 1.288(@xmath540.002 ) & @xmath100.013(@xmath540.001 ) | the locations of supernovae in the local stellar and gaseous environment in galaxies contain important clues to their progenitor stars .
access to this information , however , has been hampered by the limited resolution achieved by ground - based observations .
high spatial resolution _ hubble space telescope _
( _ hst _ ) images of galaxy fields in which supernovae had been observed can improve the situation considerably .
we have examined the immediate environments of a few dozen supernovae using archival post - refurbishment _ hst _ images .
although our analysis is limited due to signal - to - noise ratio and filter bandpass considerations , the images allow us for the first time to resolve individual stars in , and to derive detailed color - magnitude diagrams for , several environments .
we are able to place more rigorous constraints on the masses of these supernovae .
a search was made for late - time emission from supernovae in the archival images , and for the progenitor stars in presupernova images of the host galaxies .
we have detected sn 1986j in ngc 891 and , possibly , sn 1981k in ngc 4258 .
we have also identified the progenitor of the type iin sn 1997bs in ngc 3627 . by removing
younger resolved stars in the environments of sne ia , we can measure the colors of the unresolved stellar background and attribute these colors generally to an older , redder population .
_ hst _ images `` accidentally '' caught the type ia sn 1994d in ngc 4526 shortly after its outburst ; we measure its brightness . finally , we add to the statistical inferences that can be made from studying the association of sne with recent star - forming regions . |
the recent discovery at the lhc @xcite of a new particle having mass about 125gev and other properties compatible with those of the standard model ( sm ) higgs boson undoubtedly has far - reaching implications for efforts to search for new physics beyond the sm .
particularly , all new - physics models would have to include such a particle , hereafter denoted by @xmath0 , as one of their ingredients .
in general , different models would have different production rates and decay patterns for @xmath0 because of contributions from and/or mixing with other new particles .
it is therefore important to have a detailed study of the characteristic quantum numbers of @xmath0 and its interactions with sm particles .
it is hoped that through such an analysis , the newly discovered particle will provide us with hints of new physics .
one of the possible scenarios for new physics is the existence of an extra u(1 ) gauge group involving a massive gauge particle , the @xmath1 boson .
such a gauge symmetry may have its origin from various grand unified theories , string - inspired models , dynamical symmetry breaking models , and little higgs models @xcite , just to name a few .
the @xmath1 boson in different representative models has been directly searched for at colliders as well as indirectly probed via a variety of precision data @xcite , putting limits on its gauge coupling and/or mass .
generally speaking , the @xmath1 couplings to sm fermions can be family universal or nonuniversal .
the latter case has especially attracted a lot of interest in recent years due to its many interesting phenomenological implications @xcite . in this work
, we focus on family - nonuniversal interactions of the @xmath1 with the neutrinos and charged leptons and explore constraints on its relevant couplings from a number of experiments on transitions involving leptons in the initial and final states , plus possibly a photon .
these processes suffer less from qcd corrections and hadronic uncertainties than processes involving hadrons .
moreover , many of such experiments have been performed at a high precision , imposing relatively stringent constraints on any possible new interactions .
more specifically , we assume that the @xmath1 boson arises from a new u(1 ) gauge symmetry , interacts with leptons in a family - nonuniversal way , and has negligible mixing with the @xmath2 boson for simplicity , but otherwise adopt a model - independent approach to make the analysis as general as possible . due to the family nonuniversality , such a @xmath1 boson would feature flavor - changing leptonic couplings at tree level , leading to the distinctive signature of lepton - flavor violation .
however , lepton - flavor violating processes have been searched for at colliders with null results .
we therefore examine a number of flavor - conserving and flavor - changing processes to evaluate constraints on the leptonic @xmath1 couplings .
the results are then used to estimate @xmath1 contributions to both flavor - conserving and flavor - changing decays of the higgs boson into a pair of leptons , @xmath4 , at the one - loop level .
as @xmath0 will be probed with increasing precision at the lhc in coming years , and even more so if a higgs factory is built in the future , the acquired data could reveal the signals of the @xmath1 boson which we consider , assuming that @xmath0 is a sm - like higgs boson .
last but not least , we will compute the @xmath1 effects on several four - lepton decays of the @xmath3 and @xmath2 bosons and make comparison with the data if available . the lhc may also be sensitive to such indirect indications of the @xmath1 presence . the information on the @xmath1 gained from the @xmath0 , @xmath3 , and @xmath2 measurements would be complementary to that from the @xmath1 direct searches .
this paper is organized as follows .
we present the interactions of the @xmath1 boson with the leptons in section [ sec : interactions ] , allowing for in particular flavor - nonuniversal couplings . in section [ sec : constraints ] , we study constraints on the couplings of the @xmath1 from @xmath2-pole data , cross sections of @xmath5 scattering into lepton - antilepton pairs measured at lep ii , various experimental limits on low - energy flavor - changing processes involving charged leptons , and measurements of their anomalous magnetic moments . in section [ sec : predictions ] , we proceed to make predictions on both flavor - conserving and flavor - changing decays of @xmath0 into a pair of charged leptons .
we also explore the @xmath1 contributions to @xmath3 and @xmath2 decays into four leptons .
we summarize our findings in section [ sec : summary ] .
the lagrangian describing the interactions of the @xmath1 boson with neutrinos @xmath6 and charged leptons @xmath7 can be expressed as @xmath8 where summation over @xmath9 is implied , the primes of the lepton fields refer to their interaction eigenstates , @xmath10 , and the parameters @xmath11 are generally different from one another , reflecting the family nonuniversality . to denote the triplet of charged leptons , @xmath12 , and @xmath13 to refer to an individual charged lepton in general . ]
the hermiticity of @xmath14 requires these coupling constants to be real . since each of the left - handed neutrinos and its charged counterpart form a sm weak doublet , they share the same @xmath15
. since we are concerned with processes below the electroweak scale , we do not consider right - handed neutrinos in the low - energy spectrum .
the @xmath1 may also have couplings to quarks and other nonstandard fermions , but we do not address them in this analysis .
the interaction states are related to the mass eigenstates @xmath16 and @xmath17 by @xmath18 where @xmath19 for fermion @xmath20 and the 3@xmath213 matrices @xmath22 are unitary . in terms of the mass eigenstates , one can then write @xmath23 where summation over @xmath24 is implied and @xmath25 it follows that @xmath26 where @xmath27 .
hence family nonuniversality implies that the @xmath1 interactions with the leptons can be flavor violating at tree level .
furthermore , @xmath28 and @xmath29 are generally unequal .
we begin with the determination of constraints from the existing data on the @xmath2-boson decays @xmath30 and @xmath31 . for the former , the amplitude takes the form @xmath32 where @xmath33 and @xmath34 contain both sm and @xmath1 contributions and are given by @xmath35 with as usual the weak coupling constant @xmath36 and @xmath37 involving the weinberg angle . in the absence of @xmath2-@xmath1 mixing , the @xmath1 effects modify the @xmath38 vertex and leptonic self - energy diagrams at the one - loop level .
we obtain @xmath39 \nonumber \\ & & \hspace*{7ex } -\ ; i\pi\ , \biggl [ 3 + 2\delta+ 2(1+\delta)^2\,\ln\frac{\delta}{1+\delta } \biggr ] \biggr\ } ~,\end{aligned}\ ] ] where li@xmath40 is the dilogarithm .
the expression for the real part of @xmath41 has been derived previously @xcite .
the relevant observables here are the forward - backward asymmetry at the @xmath2 pole and decay rate @xmath42 where @xmath43 in @xmath31 , the @xmath1-loop contributions are analogous to those in the charged - lepton case , but without the right - handed couplings .
since the unobserved neutrinos in the final state may belong to different mass eigenstates , we can express the amplitude as @xmath44 since @xmath28 and @xmath29 are related according to eq.([bb * ] ) , summing over the final neutrinos then results in the decay rate @xmath45 accordingly , this channel may offer a complementary probe for @xmath29 . from the formula for @xmath46 in eq.([epsilonllz ] ) , we can then see that these @xmath2 decay modes are potentially sensitive to not only the flavor - conserving @xmath1 couplings , but also the flavor - changing ones .
these @xmath2-pole observables have been measured with good precision .
the experimental and sm values of @xmath47 are @xcite , in mev , @xmath48 while those of @xmath49 are @xcite @xmath50 for @xmath31 , we have @xcite , also in mev , @xmath51 numerically , for the sm contributions we employ the tree - level formulas in eqs .
( [ mz2ll])-([gsm ] ) and ( [ mz2nn ] ) along with the effective values @xmath52 which allow us to reproduce the sm numbers in eqs .
( [ xgz2ll ] ) and ( [ xal ] ) within their errors and obtain @xmath53mev in agreement with @xmath54 , indicating that other sm invisible modes are negligible . to extract the upper limit on @xmath55 , we assume it to be the only nonvanishing coupling subject to the 90% confidence - level ( cl ) ranges @xmath56 the rate numbers being in mev .
@xmath57 to be its sm lower ( upper ) value because @xmath58 is above @xmath59 @xmath60 is below @xmath61 by @xmath622 sigmas . ]
we find that the constraint on @xmath63 from @xmath31 is weaker ( stronger ) than that from @xmath30 for @xmath64 @xmath65 .
assuming that only one flavor - conserving coupling is nonzero at a time , we show the results for @xmath66tev in fig.[bll - limits](a ) , where the displayed curves represent the stronger limit in each @xmath13 case .
the blue dotted ( dash - dotted ) curve refers to @xmath67 , the green short - dashed ( long - dashed ) curve @xmath68 , and the red solid curve @xmath69 .
the curve for @xmath70 coincides with that for @xmath71 because of the @xmath31 constraint . the horizontal solid straight line for @xmath72 above 500gev marks the perturbativity limit , @xmath73 , which we have imposed as an extra requirement on the couplings .
we present another view on the results in fig.[bll - limits](b ) , which has the corresponding limits on the couplings divided by the @xmath1 mass . and ( b ) @xmath74 for @xmath75 at 90%@xmath76cl from @xmath2 decay data if only one of the couplings is not zero at a time .
the blue dotted ( dash - dotted ) curve refers to @xmath67 , the green short - dashed ( long - dashed ) curve @xmath68 , and the red solid curve @xmath69 .
the curves for @xmath70 and @xmath71 coincide due to the @xmath31 constraint .
the straight portions of the solid curves denote the perturbativity requirement , @xmath73 .
these results also serve as limits on the flavor - changing couplings @xmath77 , as explained in the text.[bll - limits],title="fig:",width=321 ] and ( b ) @xmath74 for @xmath75 at 90%@xmath76cl from @xmath2 decay data if only one of the couplings is not zero at a time .
the blue dotted ( dash - dotted ) curve refers to @xmath67 , the green short - dashed ( long - dashed ) curve @xmath68 , and the red solid curve @xmath69 .
the curves for @xmath70 and @xmath71 coincide due to the @xmath31 constraint .
the straight portions of the solid curves denote the perturbativity requirement , @xmath73 .
these results also serve as limits on the flavor - changing couplings @xmath77 , as explained in the text.[bll - limits],title="fig:",width=340 ] if instead only one of the flavor - changing couplings is nonvanishing at a time , we can get its upper limit also from fig.[bll - limits ] , in view of @xmath46 in eq.([epsilonllz ] ) . accordingly
, the limits on @xmath78 , @xmath79 , and @xmath80 are the same as those on @xmath81 , @xmath81 , and @xmath82 , respectively , for @xmath83 or @xmath84 . as we will see later , there may be stronger bounds on some of these individual flavor - changing couplings from other measurements , depending on the @xmath1 mass .
since @xmath1-mediated diagrams can contribute to @xmath85 scattering , its data can provide additional restrictions on the @xmath1 couplings . here
we will use lep - ii measurements at various center - of - mass energies above the @xmath2 pole , from 130 to 207 gev @xcite . in the absence of @xmath2-@xmath1 mixing , the amplitude if @xmath86 is @xmath87 where @xmath88 is the electromagnetic fine - structure constant , the lepton masses have been neglected , @xmath89 denote the total widths , the plus sign of the @xmath90-channel term follows from fermi statistics , @xmath91 , and @xmath92 . in the absence of flavor - changing couplings , thus the last line of eq.([mee2ll ] ) , the resulting cross - section @xmath93 and forward - backward asymmetry @xmath94 , with @xmath95 , are known in the literature ( see , _
e.g. _ , @xcite ) . as mentioned in ref.@xcite ( which also has the more general formulas in the presence of @xmath2-@xmath1 mixing ) , the expressions for these observables imply that in a model - independent study their experimental values can not lead to restrictions on the individual flavor - conserving @xmath1 couplings , assumed to be free parameters , but can nevertheless still restrain their products , @xmath96 . since we have left the @xmath1 total width , @xmath97 , unspecified in concentrating on its couplings to leptons and since it would be needed to compute the cross sections if @xmath98 , we evaluate the limits on @xmath96 from the lepii data only for @xmath72 values starting from 210gev up to 3tev . using as before eq.([eff ] ) , along with the effective value @xmath99 , and the 90%@xmath76cl ranges of the experimental numbers @xcite , and @xmath100 measurements disagree with their sm predictions by about @xmath101sigmas or more . in each of those cases , we take the lower ( upper ) bound of the required range of the relevant observable to be its sm lower ( upper ) value if the measurement is above ( below ) the sm prediction . ]
we draw fig.[beebll - limits](a ) for the upper limits on the products @xmath102 ( green long - dashed curve ) , @xmath103 ( green dash - dotted curve ) , @xmath104 ( green short - dashed curve ) , @xmath105 ( red dotted curve ) , @xmath106 ( red dash - double - dotted curve ) , and @xmath107 ( red solid curve )
. since the positive and negative limits on these coupling products are not generally symmetric with respect to zero , we present fig.[beebll - limits](b ) for the negative limits .
the plots on the right ( c and d ) depict the corresponding limits for @xmath108 . and ( b ) @xmath109 for @xmath110 and @xmath111 at 90%@xmath76cl from lepii data if only one of the coupling products is nonvanishing at a time .
the different curves are described in the text .
the right plots ( c and d ) show the corresponding limits on @xmath108 .
[ beebll - limits],title="fig:",width=319 ] and ( b ) @xmath109 for @xmath110 and @xmath111 at 90%@xmath76cl from lepii data if only one of the coupling products is nonvanishing at a time .
the different curves are described in the text .
the right plots ( c and d ) show the corresponding limits on @xmath108 .
[ beebll - limits],title="fig:",width=332 ] + and ( b ) @xmath109 for @xmath110 and @xmath111 at 90%@xmath76cl from lepii data if only one of the coupling products is nonvanishing at a time .
the different curves are described in the text .
the right plots ( c and d ) show the corresponding limits on @xmath108 .
[ beebll - limits],title="fig:",width=319 ] and ( b ) @xmath109 for @xmath110 and @xmath111 at 90%@xmath76cl from lepii data if only one of the coupling products is nonvanishing at a time .
the different curves are described in the text .
the right plots ( c and d ) show the corresponding limits on @xmath108 .
[ beebll - limits],title="fig:",width=332 ] if one of the flavor - changing couplings @xmath112 in eq .
( [ mee2ll ] ) does not vanish , we can constrain it separately , assuming @xmath113 . in that case , the @xmath85 cross - sections have the expressions given in the appendix , and so @xmath97 is not required in the calculation . using the lepii data again
, we obtain the upper limits on @xmath114 for @xmath66tev in fig.[bel - limits ] .
the second plot displays the corresponding limits on @xmath115 .
these results are evidently more stringent than the bounds on @xmath116 inferred from fig .
[ bll - limits ] . and
( b ) @xmath117 for @xmath110 at 90%@xmath76cl from lepii data .
the green long - dashed ( short - dashed ) curves refer to @xmath118 and the red dotted ( solid ) curves @xmath119 .
[ bel - limits],title="fig:",width=313 ] and ( b ) @xmath117 for @xmath110 at 90%@xmath76cl from lepii data .
the green long - dashed ( short - dashed ) curves refer to @xmath118 and the red dotted ( solid ) curves @xmath119 .
[ bel - limits],title="fig:",width=336 ] turning our attention now to constraints on the @xmath1 couplings from low - energy data , we will consider a number of lepton flavor - violating processes and leptonic anomalous magnetic moments .
the relevant formulas are available from the expressions given in ref .
@xcite in the more general case with @xmath2-@xmath1 mixing .
we begin by mentioning an additional process that can restrict @xmath120 separately , namely the muonium - antimuonium conversion , @xmath121 . in this case , the constraints are the same as those determined in ref .
@xcite , @xmath122 these are stricter than their counterparts in fig.[bel - limits ] if @xmath123 goes below 30gev .
we next look at various constraints on the products of a pair of different couplings of the @xmath1 , at least one of them being flavor changing , from the experimental limits @xcite for flavor - changing leptonic 3-body and radiative 2-body decays of the @xmath124 and @xmath125 leptons .
the leptonic decays arise from tree - level @xmath1-mediated diagrams , whereas the radiative decays proceed from loop diagrams containing the @xmath1 and an internal lepton . assuming that only the two couplings in each of the products are present at a time
, we collect the results in table@xmath76[le - limits ] .
.limits on products of @xmath1 couplings determined from low - energy data .
the second column contains 90%@xmath76cl experimental upper - limits on the branching ratios of the listed decay modes.[le - limits ] [ cols="^,^,^ " , ] the anomalous magnetic moments @xmath126 and @xmath127 of the electron and muon have been measured very precisely and therefore can provide extra constraints .
in contrast , the experimental information on @xmath128 is still too limited to provide significant bounds @xcite . for the @xmath1 contributions to @xmath126 and @xmath127
, we will retain only the terms induced by the @xmath125 lepton in the loop , as they are enhanced by the @xmath125 mass compared to the other lepton terms @xcite , @xmath129 the sm prediction for @xmath126 is compatible with its most recent measurement , the difference between them being @xmath130 @xcite .
consequently , we can impose the 90%@xmath76cl range @xmath131 .
in contrast , the sm and experimental values of @xmath127 presently differ by nearly 3 sigmas , @xmath132 @xcite .
this suggests that we may require @xmath133 .
it follows that @xmath134 the @xmath135 limits are complementary to the individual bounds on @xmath136 illustrated in fig.[bel - limits ] .
the results above allow us to explore how the @xmath1 effects may modify the leptonic decays of the newly found particle , @xmath0 , assumed to be a sm - like higgs boson , and also those of the weak bosons ( @xmath2 and @xmath3 ) . we will deal with both flavor - conserving and -violating channels involving two and four leptons in the final states .
we first look at the flavor - violating decay @xmath137 .
it proceeds from a @xmath1-loop diagram having an @xmath138 and two lepton-@xmath1 vertices , at least one of the latter being flavor changing , and one - loop @xmath139 diagrams with flavor - changing couplings .
we calculate the amplitude to be @xmath140 where @xmath141gev is the higgs vacuum expectation value , @xmath142 ~ , & \end{aligned}\ ] ] with @xmath20 being the fermion in the loop .
the rate for @xmath143 is then @xmath144 to predict the largest rates , we observe from the bounds derived in the last section and eq.([ellh ] ) that the most important contribution comes from the internal lepton @xmath145 and that the rates are maximized for final states with one @xmath125 .
thus , we take @xmath146 and @xmath147 from the @xmath148 bounds in table@xmath76[le - limits ] . for definiteness , we take @xmath149gev , compatible with the average @xmath0 mass of @xmath150 gev from the lhc measurements @xcite . with only one nonzero product of couplings being present in each case and
@xmath151mev @xcite , we obtain for @xmath66tev @xmath152 the upper bound occurring at @xmath153tev , and a somewhat smaller number for @xmath154 .
clearly these @xmath1-induced flavor - violating higgs decays will not be observable in the near future .
we next consider the @xmath1 impact on the flavor - conserving decay @xmath155 .
the @xmath1 contribution follows from eq.([h2ll ] ) after setting @xmath156 . combining the result with the sm tree - level contribution
, we then get @xmath157 l\end{aligned}\ ] ] leading to @xmath158 we expect again that the internal lepton @xmath145 in the loop yields the maximal impact . accordingly , for @xmath159 and @xmath160 we take , respectively , the @xmath161 and @xmath162 ranges in eq.([g-2 ] ) , assuming that the coupling products are purely real or dominated by the real part .
the graphs in fig.[deltaem ] depict the resulting fractional change @xmath163 in the rate due to the presence of @xmath1 exclusively via these flavor - changing couplings .
although the @xmath1 contribution can reduce the @xmath159 rate sizeably [ blue - shaded areas in fig.[deltaem](a ) ] , this decay mode , with a branching ratio of @xmath164 in the sm , may be beyond reach for a long time .
much more interesting is @xmath160 , which has a sm branching ratio of @xmath165 and therefore may be measurable in the not - so - distant future with precision possibly sensitive to the @xmath1 effect , indicated by the green - shaded areas in fig.[deltaem](b ) . for ( a ) @xmath64 and ( b ) @xmath166 due to only the coupling products @xmath161 and @xmath162 , respectively.[deltaem],title="fig:",height=257 ] for ( a ) @xmath64 and ( b ) @xmath166 due to only the coupling products @xmath161 and @xmath162 , respectively.[deltaem],title="fig:",height=257 ] as it turns out , potentially more considerable modifications to the @xmath155 rate can be induced by the flavor - conserving couplings @xmath167 , which enter @xmath168 as the product @xmath169 . to estimate their maximal impact
, we focus on the @xmath110 cases . in each of them
, we set all the other couplings to zero and scan the values of @xmath167 satisfying the requirements in eq.([zpole ] ) as well as the perturbativity condition @xmath73 .
we find that the coupling values allowed by these constraints can translate into substantial @xmath170 that are positive or negative . especially , in the @xmath171gev region , the decrease in the rate could reach a few tens percent , whereas the increase could exceed 100% , even up to @xmath62300% , as can be seen in fig.[deltamt ] . and @xmath172 to sm rate of @xmath155 for ( a ) @xmath166 and ( b ) @xmath173 if only the flavor - conserving couplings @xmath167 are nonzero.[deltamt],title="fig:",height=253 ] and @xmath172 to sm rate of @xmath155 for ( a ) @xmath166 and ( b ) @xmath173 if only the flavor - conserving couplings @xmath167 are nonzero.[deltamt],title="fig:",height=253 ] the @xmath174 decay has begun to be observable at the lhc .
the latest signal strengths for this channel reported by the atlas and cms collaborations are @xmath175 and @xmath176 , respectively , at @xmath177gev @xcite . obviously , these early findings already disfavor some parts of the @xmath1 parameter space implied by fig.[deltamt](b ) , although it is still too soon to be quantitative about the exclusion zones in view of the sizable uncertainties of these current data . as their precision continues to improve ,
the upcoming experiments can either uncover a @xmath1 signal or restrain its @xmath125 couplings further .
a similar expectation can be stated regarding the @xmath124 couplings from future measurements of the @xmath160 mode .
the flavor - violating decays @xmath178 are not yet observed , but have been searched for , the experimental upper - limits on their branching ratios being @xcite @xmath179 at 95%@xmath76cl , where @xmath180 .
these modes get @xmath1-loop contributions analogously to those in @xmath30 and hence may provide further limits on the @xmath1 couplings if the experimental limits are less than the predicted values derived from the upper limits on the couplings . to make the predictions for the @xmath181 , @xmath182 , and @xmath183 final states , we take the biggest values @xmath184 from the @xmath185 , @xmath186 , and @xmath187 bounds , respectively , in table@xmath76[le - limits ] .
these translate into @xmath188 hence , although the @xmath181 mode is likely to be undetectable , the @xmath182 and @xmath183 predictions are only 18 and 16 times below their respective experimental bounds .
it is worth noting that these results are comparable to those found in ref.@xcite in the presence of @xmath2-@xmath1 mixing , in accord with the expectation mentioned therein .
the presence of the @xmath1 may also manifest itself in the decays of @xmath0 and the weak bosons into four leptons .
the cms collaboration @xcite has recently reported the first observation of @xmath2 decays into four charged leptons ( @xmath189 and @xmath124 ) consistent with the sm expectations .
this suggests that such decays can be measured well in the future at the lhc , thereby providing another means to look for new - physics hints , such as those of the @xmath1 . on the other hand , since four - lepton final states in @xmath0 decays proceed from the @xmath190 or @xmath191 mode , the @xmath1 impact would expectedly be very small and hard to detect in flavor - conserving decays .
the flavor - violating four - lepton decays of @xmath0 would also have rates that are too tiny to be observable .
we therefore focus on the @xmath1 contributions to a number of 4-lepton decays of the gauge bosons , @xmath192 and @xmath193 for @xmath75 , using the constraints found in section [ sec : constraints ] .
we compute these decays using the calchep 3.4 package @xcite incorporating the new vertices in the model file and assuming @xmath194gev , since we have not specified the @xmath1 total width . in
the flavor - conserving decays @xmath195 , the @xmath1 contributions depend on the @xmath196 couplings . applying the constraints given in fig.[bll - limits ]
, we estimate the possible ranges of the fractional change @xmath197 for @xmath198 , taking only one of @xmath196 to be nonzero at a time .
thus , we obtain @xmath199 where the lowest numbers all belong to @xmath200gev and we have imposed the kinematical cut @xmath201gev in estimating the rates . since the sm predicts @xmath202 for @xmath203 @xcite , a large luminosity is required to observe the changes .
we also compute the possible ranges of the fractional change @xmath204 in the @xmath205 cases where @xmath206 . for @xmath207 and @xmath208 we should take into account the constraints on the products of a pair of different couplings in fig.[beebll - limits ] , whereas for @xmath209 we apply the upper limits on @xmath210 and @xmath211 from fig.[bll - limits ] .
we find that the magnitudes of @xmath212 and @xmath212 are smaller than @xmath213 because of the constraints on @xmath214 and @xmath215 in fig.[beebll - limits ] , while the magnitude of @xmath216 can be comparatively greater , @xmath217 the positive ( negative ) sign of @xmath218 corresponding to the negative ( positive ) sign of @xmath219 .
one can see that the upper bound of @xmath220 is roughly similar in order of magnitude to the lower bounds of @xmath221 and @xmath222 .
we also consider the flavor - violating 4-lepton decays of the @xmath2 , concentrating on the modes @xmath223 , @xmath224 , and @xmath225 , as their amplitudes are determined by @xmath226 , @xmath227 , and @xmath228 , respectively .
the constraints on the first two products of couplings are given in table@xmath76[le - limits ] , and the upper limits on @xmath229 come from fig.[bll - limits ] . applying the biggest values from eq.([biggest ] ) and the upper limit on @xmath230 in fig.[bll - limits ] , we find that the branching ratios are of @xmath231 for the first two modes and @xmath232 for the third one , which will be unobservable in the near future .
the 4-lepton decays of the @xmath3 are treated similarly . in the flavor - conserving case , the fractional change @xmath233 for @xmath75 depend on @xmath234 , in light of the relation between @xmath235 and @xmath236 in eq .
( [ bb * ] ) .
we obtain @xmath237 where the lowest value corresponds to @xmath238gev and we have employed the @xmath201gev cut as before in estimating the rates .
the @xmath239 numbers are almost the same because the upper limits of @xmath240 are the same . since the sm prediction is @xmath241 , again a large luminosity is required to measure the changes , as in the @xmath2-decay case . for the flavor - violating 4-lepton decays of the @xmath3
, we consider the @xmath242 and @xmath243 channels .
we obtain the biggest branching ratios of these modes from the upper limits of @xmath244 and @xmath230 , respectively .
using eq.([biggest ] ) and fig.[bll - limits ] , we find that the branching ratio of the first mode is at most of @xmath245 and that of the second mode @xmath232
. hence they will likely be undetectable for a long time , as the flavor - violating 4-lepton @xmath2 decays .
we have explored a @xmath1 boson originating from a new u(1 ) gauge symmetry and interacting with leptons in a family - nonuniversal way . the @xmath1 is assumed to have no mixing with the @xmath2 boson .
we have studied the effects of the @xmath1 as a virtual particle in various processes with leptons in the initial and/or final states , especially the leptonic decays of the newly discovered @xmath0 , putatively a standard model - like higgs boson .
we first study bounds on the leptonic couplings of the @xmath1 from available experimental data .
we employ the @xmath2-pole observables , including the @xmath246 rates , the associated forward - backward asymmetries , and the invisible @xmath2 decay rate , to put constraints on @xmath247 .
the cross sections of @xmath248 for @xmath249 from lep - ii experiments are used to place bounds on the products of chiral couplings @xmath250 for @xmath251 .
further restrictions are found from various low - energy processes , including the muonium - antimuonium conversion for @xmath252 in particular , several flavor - changing leptonic 3-body and radiative 2-body decays of the muon and tauon , and the anomalous magnetic moments of the electron and muon .
we then apply the constraints on the @xmath1 couplings and mass to make predictions for the flavor - conserving and -violating 2-body leptonic decays of @xmath0 as well as 4-body decays of the @xmath3 and @xmath2 . for the flavor - violating @xmath253 decays ,
the most important contribution comes from the case with one @xmath125 in the final state and an internal @xmath125 running in the loop due to mass enhancement .
unfortunately , such @xmath1-mediated flavor - violating higgs decays are too small to be observable in the near future . on the other hand
, the flavor - conserving @xmath254 decay can be significantly affected by the contribution of an internal @xmath125 . with flavor - changing couplings alone ,
the branching ratio of @xmath255 decay may be enhanced by a mere few percent .
in contrast , with only flavor - conserving couplings , both the @xmath255 and @xmath256 channels can be considerably modified by a few tens to a few hundreds percent .
consequently , upcoming measurements of the latter mode with better precision than its current data will either uncover or constrain the @xmath1 boson .
we have found that the flavor - violating @xmath257 decays , mediated by the @xmath1 at the loop level , are consistent with current experimental upper bounds at 95% cl .
the 4-body leptonic decays of the @xmath2 have started to be measured by the lhc .
our calculations show that a larger luminosity is required to observe modifications in flavor - conserving @xmath258 decays due to the @xmath1 , the maximal changes being of @xmath259 .
a similar situation holds for flavor - conserving @xmath260 decays .
we have also found that the changes in the flavor - violating @xmath258 and @xmath260 decays are unlikely to be observable soon .
we would like to thank xiao - gang he and german valencia for helpful discussions .
this research was supported in part by the national science council of taiwan , r. o. c. , under grant no .
nsc-100 - 2628-m-008 - 003-my4 .
g. aad _ et al . _ [ atlas collaboration ] , phys .
b * 716 * , 1 ( 2012 ) [ arxiv:1207.7214 [ hep - ex ] ] ; s. chatrchyan _ et al .
_ [ cms collaboration ] , phys .
b * 716 * , 30 ( 2012 ) [ arxiv:1207.7235 [ hep - ex ] ] . for the case of heavy @xmath1 bosons ,
see , for example , p. langacker , rev .
phys . * 81 * , 1199 ( 2009 ) [ arxiv:0801.1345 [ hep - ph ] ] and references therein . for the lighter case ,
see , for example , m. williams , c.p .
burgess , a. maharana , and f. quevedo , jhep * 1108 * , 106 ( 2011 ) [ arxiv:1103.4556 [ hep - ph ] ] and references therein .
g. brooijmans , m.c .
chen , and b.a .
dobrescu , in ref . @xcite
. f. del aguila , j. de blas , p. langacker and m. perez - victoria , phys . rev .
d * 84 * , 015015 ( 2011 ) [ arxiv:1104.5512 [ hep - ph ] ] .
v. barger , c.w .
chiang , j. jiang , and p. langacker , phys .
b * 596 * , 229 ( 2004 ) [ arxiv : hep - ph/0405108 ] ; v. barger , c.w .
chiang , p. langacker , and h.s .
lee , phys .
b * 598 * , 218 ( 2004 ) [ arxiv : hep - ph/0406126 ] ; a. arhrib , k. cheung , c.w .
chiang , and t.c .
yuan , phys .
d * 73 * , 075015 ( 2006 ) [ arxiv : hep - ph/0602175 ] ; k. cheung , c.w .
chiang , n.g .
deshpande , and j. jiang , phys .
b * 652 * , 285 ( 2007 ) [ arxiv : hep - ph/0604223 ] ; x.g .
he and g. valencia , phys .
d * 74 * , 013011 ( 2006 ) [ arxiv : hep - ph/0605202 ] ; c.w . chiang , n.g .
deshpande , and j. jiang , jhep * 0608 * , 075 ( 2006 ) [ arxiv : hep - ph/0606122 ] ; x.g .
he and g. valencia , phys .
b * 651 * , 135 ( 2007 ) [ arxiv : hep - ph/0703270 ] ; v. barger , l.l .
everett , j. jiang , p. langacker , t. liu , and c.e.m .
wagner , jhep * 0912 * , 048 ( 2009 ) [ arxiv:0906.3745 [ hep - ph ] ] ; x.g .
he and g. valencia , phys .
b * 680 * , 72 ( 2009 ) [ arxiv:0907.4034 [ hep - ph ] ] ; q. chang , x.q .
li , and y.d .
yang , jhep * 1002 * , 082 ( 2010 ) [ arxiv:0907.4408 [ hep - ph ] ] ; n.g .
deshpande , x.g .
he , and g. valencia , phys .
d * 82 * , 056013 ( 2010 ) [ arxiv:1006.1682 [ hep - ph ] ] ; a.k .
alok , s. baek , and d. london , jhep * 1107 * , 111 ( 2011 ) [ arxiv:1010.1333 [ hep - ph ] ] ; f. yu , phys . rev .
d * 83 * , 094028 ( 2011 ) [ arxiv:1104.0243 [ hep - ph ] ] ; k. cheung and j. song , phys .
lett . * 106 * , 211803 ( 2011 ) [ arxiv:1104.1375 [ hep - ph ] ] ; p. ko , y. omura , and c. yu , phys . lett .
b * 710 * , 197 ( 2012 ) [ arxiv:1104.4066 [ hep - ph ] ] ; p.j .
fox , j. liu , d. tucker - smith , and n. weiner , phys .
d * 84 * , 115006 ( 2011 ) [ arxiv:1104.4127 [ hep - ph ] ] ; s. chang , k.y .
lee , and j. song , arxiv:1104.4560 [ hep - ph ] ; z. liu , p. nath , and g. peim , phys .
b * 701 * , 601 ( 2011 ) [ arxiv:1105.4371 [ hep - ph ] ] ; j.l . hewett and t.g .
rizzo , arxiv:1106.0294 [ hep - ph ] ; j. fan , d. krohn , p. langacker , and i. yavin , arxiv:1106.1682 [ hep - ph ] ; p. ko , y. omura , and c. yu , arxiv:1108.0350 [ hep - ph ] ; j.y
. liu , y. tang , and y.l .
wu , j. phys .
g * 39 * , 055003 ( 2012 ) [ arxiv:1108.5012 [ hep - ph ] ] ; c.x .
yue , y.m .
zhang , and l.j .
liu , phys .
b * 547 * , 252 ( 2002 ) [ hep - ph/0209291 ] ; j.i .
aranda , j. montano , f. ramirez - zavaleta , j.j .
toscano , and e.s .
tututi , phys .
d * 86 * , 035008 ( 2012 ) [ arxiv:1202.6288 [ hep - ph ] ] .
chiang , y.f .
lin , and j. tandean , jhep * 1111 * , 083 ( 2011 ) [ arxiv:1108.3969 [ hep - ph ] ] .
carone and h. murayama , phys .
* 74 * , 3122 ( 1995 ) [ hep - ph/9411256 ] ; phys .
d * 52 * , 484 ( 1995 ) [ hep - ph/9501220 ] ; e. ma and d.p .
roy , phys .
d * 58 * , 095005 ( 1998 ) [ hep - ph/9806210 ] .
j. beringer _
[ particle data group collaboration ] , phys .
d * 86 * , 010001 ( 2012 ) .
j. alcaraz _ et al . _
[ aleph and delphi and l3 and opal and lep electroweak working group collaborations ] , hep - ex/0612034 .
b. kors and p. nath , jhep * 0507 * , 069 ( 2005 ) [ arxiv : hep - ph/0503208 ]
. t. aoyama , m. hayakawa , t. kinoshita , and m. nio , phys .
lett . * 109 * , 111807 ( 2012 ) [ arxiv:1205.5368 [ hep - ph ] ] .
t. aoyama , m. hayakawa , t. kinoshita , and m. nio , phys .
lett . * 109 * , 111808 ( 2012 ) [ arxiv:1205.5370 [ hep - ph ] ] . | though not completely a surprise according to the standard model and existing indirect constraints , the higgs - like particle , @xmath0 , of mass around 125 gev recently observed at the lhc may offer an additional window to physics beyond the standard model .
in particular , its decay pattern can be modified by the existence of new particles .
one of the popular scenarios involves a @xmath1 boson associated with an extra abelian gauge group . in this study
, we explore the potential effects of such a boson with family - nonuniversal couplings on the leptonic decays of @xmath0 , both flavor - conserving and flavor - changing . for current constraints ,
we take into account leptonic decays at the @xmath2 pole , lep ii scattering data , limits on various flavor - changing lepton transitions , and lepton magnetic dipole moments . adopting a model - independent approach and assuming that the @xmath1 has negligible mixing with the @xmath2 boson , we find that present data allow the @xmath1 effects to reach a few percent or higher on @xmath0 decays into a pair of leptons .
future measurements on @xmath0 at the lhc or a linear collider can therefore detect the @xmath1 contributions or impose constraints on its couplings .
we also consider @xmath1-mediated four - lepton decays of the @xmath2 and @xmath3 bosons . |
dark matter prominently appears in galaxy rotation curves beyond a few kpc of the central bulge of spiral galaxies , in velocity dispersion in galaxy clusters and in cosmological evolution based on analysis of the cosmic micro - wave background ( cmb ) .
they have in common dynamics at accelerations below the cosmological scale @xmath18 , defined by the product of the velocity of light @xmath2 and the hubble parameter @xmath19 @xcite . in spiral galaxies
, accelerations @xmath20 appears to harden to the baryonic tully - fisher inverse distance law @xcite , distinct from effectively newtonian gravitational dynamics within at @xmath21 based on baryonic matter alone . given ample evidence of dark matter on cosmological scales @xcite , it is tempting but not imperative to also attribute this apparent anomaly to dark matter concentrations on galactic scales @xcite .
however , the transition acceleration scale @xmath22 is tiny , some four orders below the scales at which newtonian mechanics and gravitational attraction have been tested .
it may reflect a hitherto unappreciated low energy scale in the cosmological vacuum that perturbs motion in possibly unfamiliar ways .
presently , our universe is in a state of accelerated expansion that is effectively approaching a de sitter state driven by a finite density of dark energy @xcite , inferred from a deceleration parameter @xmath23 with fractions of cold dark matter @xmath24 and dark energy @xmath25 , , normalized to closure matter density @xmath26 with newton s constant @xmath27 .
the conclusion ( [ eqn_q ] ) is based on the three - flat friedmann - robertson - walker ( frw ) line - element @xmath28 with dynamical scale factor @xmath29 , @xmath30 and @xmath31 evolving according to the theory of general relativity , assuming a no interaction with any low - energy scales in the background vacuum .
to first order , our vacuum is locally minkowski described by lorentz invariant light cones @xmath32 at each point of spacetime .
on cosmological scales , the vacuum has a small but finite gibbons - hawking temperature @xcite @xmath33 associated with the cosmological horizon at @xmath34 in the limit of lorentz invariant de sitter space , where @xmath35 denotes the boltzmann constant and , currently , @xmath36 km s@xmath12mpc@xmath12 @xcite , at some tension with @xmath37 km s@xmath12mpc@xmath38 from the cmb @xcite . moving at constant acceleration @xmath39 .
it assumes a finite temperature of that derives from a map of phase uniform on a null surface element @xmath40 in the future into a logarithmic distribution on a null surface element x in the past .
vacuum in the past hereby has a corresponding non - zero state in the future , that appears to @xmath41 at the unruh temperature ( [ eqn_a4 ] ) .
( middle panel . ) a rindler observer @xmath41 is trailed by an event horizon @xmath42 at constant distance @xmath43 , whose inertia can be attributed to a thermodynamic potential associated with @xmath42 .
_ strong _ and _ weak _ gravity limits ( middle and , respectively , right panel ) are defined by @xmath43 relative to the distance @xmath44 to the cosmological horizon @xmath3 as seen by inertial observers .
the latter signals a perturbation of @xmath41 inertia below its newtonian value . ] in a newtonian description , stars in galaxies are mutually interacting by gravitational forces that put them on non - inertial trajectories in flat minkowski spacetime at zero temperature vacuum .
observers along such trajectories see things differently . at acceleration @xmath45 ,
the world - line of a rindler observer @xcite has a time - dependent rapidity @xmath46 as a function of eigentime @xmath47 . in a minkowski diagram
, the trajectory @xmath48 appears curved with @xmath49 at a constant distance @xmath50 according to the line - element @xmath51 of 1 + 1 minkowski spacetime @xmath52 .
@xmath41 hereby considers itself at a constant distance @xmath43 to an event horizon @xmath42 , given by a light cone with vertex at the origin ( fig .
[ rindlero ] ) , satisfying @xmath53 thus , @xmath41 s field of view is limited to a wedge minkowski spacetime . in this wedge ,
@xmath41 shares only a finite interval @xmath54 of its eigentime @xmath47 with inertial observers in minkowski spacetime .
according to ( [ eqn_arh ] ) , @xmath41 is trailed by event horizons @xmath42 at distances @xmath43 inversely proportional to @xmath39 . in unitary holography
, a particle of mass @xmath5 at a distance @xmath43 is imaged by information @xcite @xmath55 on two - dimensional screens , that covers an einstein area @xmath56 in terms of the compton phase @xmath57 defined by the wave number @xmath58 and the planck sized elements of area @xmath59 , where @xmath60 denotes the planck constant .
holography @xcite is hereby realized by imposing unitarity @xmath61 on the probabilities for @xmath4 to be inside ( @xmath62 ) or outside ( @xmath63 ) a screen of radius @xmath43 about @xmath5 . a collusion of the rindler and cosmological horizon ( fig .
[ rindlero ] ) occurs at when @xmath50 satisfies @xmath64 in describing the galactic disks of a spiral galaxy of mass @xmath65 with gravitational radius @xmath66 by newton s law of gravitational attraction @xmath67 , @xmath68 , we thus arrive at a transition radius @xcite @xmath69 beyond which we reach weak gravity @xmath70 as the rindler horizon @xmath42 drops beyond @xmath3 .
this expression ( [ eqn_a2 ] ) serves as an order of magnitude estimate .
implications for gravitational attraction requires a detailed consideration on the properties of @xmath42 and @xmath3 , since @xmath42 is defined for @xmath41 while the cosmological horizon is ab initio defined for inertial observers .
nevertheless , ( [ eqn_a2 ] ) suffices to identify a transition between strong and weak gravity , from newtonian to non - newtonian behavior in galaxy rotation curves . in the following ,
we pursue ( [ eqn_a2 ] ) further in weak gravity accompanying baryonic matter .
to start , we identify a holographic origin of inertia associated thermodynamic properties of the rindler event horizon .
rest mass hereby equals the potential energy in the gravitational field associated with the rindler horizon . in minkowski space , inertial and total mass - energy
are hereby identical .
the same holds true in spherical symmetry about the event horizons of black holes in asymptotically flat spacetimes .
we extend this approach to the homogeneous and isotropic de sitter and , more generally , frw cosmologies .
our approach identifies a breakdown of the equivalence principle at non - relativistic vacuum temperatures .
it results in a relation between total mass energy and baryonic mass - energy , based on dispersive behavior in cosmological holography at low vacuum temperatures on the order of ( [ eqn_tds ] ) .
in minkowski spacetime , rindler observers detect the unruh temperature @xcite @xmath71 associated with the surface gravity of @xmath42 .
to highlight its basic premises in common with ( [ eqn_tds ] ) , we recall that ( [ eqn_a4 ] ) derives from a bogoliubov map , by ray tracing between radiative hilbert spaces @xmath72 and @xmath73 in the past and , respectively , future @xcite , here shown in fig .
[ rindlero ] with null surface elements @xmath74 and @xmath40 as indicated .
a choice of uniform distribution of phase on @xmath40 defines a given positive frequency state of @xmath73 relative to @xmath75 .
the ray tracing map thus introduces a fourier transform on @xmath74 of a logarithmic distribution of phase @xmath76 over @xmath77 , giving rise to a thermal spectrum @xmath78 @xcite . in this process
, a rindler observer , according to its eigentime , detects a wave front with exponential divergence in frequency into the future . starting from the true vacuum state of @xmath74
, the inverse of our map gives a thermal distribution of positive energy photon number ( per unit time per unit frequency ) @xmath79 out to the future through @xmath40 @xcite . at a distance @xmath43 to the event horizon
@xmath42 of a rindler observer , changes in @xmath43 introduce a corresponding change in entropy @xmath80 at @xmath81 of @xmath45 on @xmath42 , and @xmath82 is also a thermodynamic temperature .
inertia of rindler observers hereby satisfies @xmath83 with associated potential energy relative to @xmath42 equal to its rest mass energy , @xmath84 by virtue of @xmath45 being constant along @xmath43 . in ( [ eqn_u1 ] ) , inertia
@xmath4 originates in the thermodynamic potential @xmath85 associated with the rindler horizon @xmath42 .
this implies that @xmath4 will be perturbed away from @xmath5 in ( [ eqn_u1 ] ) whenever @xmath42 drops beyond the cosmological horizon @xmath3 .
the quantify this , we next consider the geometry and thermodynamic properties of the latter .
in frw cosmologies , cosmological horizons are trapped surfaces defined in hypersurfaces of constant cosmic time @xmath86 . given a unit normal @xmath87 and an extrensic curvature @xmath88 , they satisfy @xmath89 @xcite . in three - flat frw cosmologies with hubble parameter @xmath3 , this reduces to @xmath90 , i.e. , @xmath1 as stated before .
a covariant derivation of surface gravity obtains from geodesic deviation between a pair of generators in four - dimensionlal spacetime @xmath91 centered about an inertial observer .
a tangent @xmath92 to such generator is null , @xmath93 . a spacelike separation @xmath94 in @xmath95 between such null - generators , initially along the unit vector @xmath96 experiences a change in length with tip - to - tip acceleration , given by the equation of geodesic deviation along @xmath97 .
with cosmic time @xmath86 as an affine parameter , @xmath98 , the acceleration @xmath99 of one tip satisfies @xmath100 where @xmath101 denotes the riemann tensor of the frw line - element @xcite . in magnitude
, @xmath102 satisfies the _ intrinsic _ surface gravity @xmath103 once again , @xmath104 is an associated thermodyamic temperature @xcite .
the same arguments apply to black hole event horizons , upon considering a null - tangent @xmath105 of the event horizon of a schwarzschild black hole with mass @xmath106 .
it has the intrinsic surface gravity @xmath107 equal to the _ extrinsic _ surface gravity @xmath108 in the schwarzschild line - element with @xmath109 and redshift factor @xmath110 , @xmath111 .
this suface gravity is conform hawking s thermodynamic temperature @xcite @xmath112 akin to ( [ eqn_u1 ] ) : a test particle of rest mass @xmath113 taken from @xmath3 to infinity is associated with entropic work @xmath114 , based on the bekenstein - hawking entropy @xmath115 @xcite , where @xmath116 denotes the planck length associated with planck s constant @xmath60 . in the schwarzschild line - element of black holes , the same result obtains from the line - integral of force @xmath117 over distance @xmath118 from @xmath119 to infinfity , i.e. , @xmath120 .
this recovers @xmath121 as the gravitational binding energy of a mass element @xmath4 in the black hole .
hidden behind the event horizon , the source of the gravitational field is again unseen @xcite . in unitary holography ,
the black hole and cosmological event horizons define the minimum , respectively , maximum sized holographic screens with @xmath122 .
in a cartesian coordinate system , @xmath123 at @xmath124 , the geodesic deviation equation ( [ eqn_alpha1 ] ) gives @xmath125 for the acceleration of separation @xmath94 parallelly transported between two null - generators of a horizon null - surface . upon taking the norm of @xmath126
, this defines an harmonic oscillation @xmath127 that , in a holographic approach , extends to a wave equation in spacetime within , @xmath128 where we @xcite @xmath129 the wave equation ( [ eqn_alpha6 ] ) will be recognized to describe electromagnetic and linearized gravitational waves in de sitter space , considered in a lorenz gauge on the u(1 ) and , respectively , so(3,1 ) connection @xcite , by coupling to the ricci tensor @xmath130 .
importantly , ( [ eqn_alpha6 ] ) reveals dispersion at low energies .
as illustrated by rindler observers in minkowski spacetime , the equivalence principle describes sameness of lorentz factors in minkowski space and potentials in static gravitational fields . for an inertial observer in minkowski spacetime , a slice of constant time represents an equipotential hypersurface , while a rindler observer identifies a potential gradient in hypersurfaces of constant eigentime .
their equivalence satisfies conservation of total energy , in that the observable effects of a lorentz factor in the first and a potential drop in the second are equivalent . in ( [ eqn_u1 ] ) and ( [ eqn_u2 ] )
, the equivalence principle applies independent of geometry , in uniform and spherically symmetric gravitational fields alike . in de sitter cosmologies , surfaces of constant cosmic time
are equipotential hypersurfaces as in minkowski space , endowed with geodesic expansion described by the hubble parameter with associated lorentz factor @xmath131 as a function of distance @xmath132 to an inertial observer .
according to aforementioned equivalence principle , spacetime about an inertial observer is equivalently static and spherically symmetric with a gravitational potential @xmath133 , at least in the case of static event horizons .
the equipotential surface of constant cosmic time @xmath86 in ( [ eqn_frw ] ) now becomes equivalent to surfaces on which test particles satisfy @xmath134 .
the net result is the well - known static de sitter universe with an extension to a central mass @xmath4 described by @xmath135 imposing @xmath136 in this equivalence of a static , spherically universe with ( [ eqn_lambda ] ) , we obtain @xmath137 with @xcite @xmath138 in terms of the mass density @xmath139 at a total mass - energy @xmath140 within the volume @xmath141 . with @xmath142 , we note the extrinsic surface gravity at @xmath136 @xmath143 ch = -qch . \label{eqn_alpha2b}\end{aligned}\ ] ] thus , @xmath144 is directly related to the hubble flow through the cosmological horizon , positive for ingoing ( @xmath145 ) and negative for outgoing ( @xmath146 ) flows .
outgoing hubble flows conform to the newtonian regime of negative gravitational binding energy . in de sitter spacetime @xmath147
, @xmath144 equals @xmath148 in ( [ eqn_alpha2 ] ) . by virtue of constant radius ,
its horizon is akin to rindler horizons in minkowski spacetime . in identifying the equivalence principle with @xmath149 leading to ( [ eqn_u1 ] ) and ( [ eqn_u2 ] ) .
the mass - energy @xmath150 of a test particle represents the gravitational binding energy in ( [ eqn_u3 ] ) in taking it out from the horizon at @xmath1 .
_ the discrepancy between intrinsic and extrinsic surfrace gravity , @xmath151 quantifies breakdown of the equivalence principle in non - de sitter cosmologies .
_ likewise , screen temperature @xmath152 is generally not equal to local vacuum temperature @xmath153 .
according to ( [ eqn_alpha2 ] ) and ( [ eqn_alpha6 ] ) , the association of intrinsic surface gravity with horizon temperature and dark energy defines @xmath154 as a positive energy condition .
in weak gravity ( [ eqn_w1 ] ) , ( [ eqn_u1 ] ) is reduced by integration limited to @xmath44 , whereby @xmath155 satisfies @xmath156 . as a holographic condition
, we impose equality in the number of screen modes and radial bulk modes , @xmath157 where @xmath158 refers to a momentum coupling between screen and bulk modes and @xmath159 is the temperature of the cosmological horizon associated with intrinsic curvature ( [ eqn_alpha2b ] ) @xmath160 with @xmath161 , it follows that @xmath162 with @xmath163 given a gravitational attraction @xmath164 by a baryonic mass @xmath106 , we thus arrive at an acceleration @xmath165 at very low momenta corresponding to @xmath166 @xmath167 in ( [ eqn_b0 ] ) with ( [ eqn_lambda ] ) captures the tully - fisher law equivalent to milgrom s inverse distance law @xmath168 of gravity with milgrom s constant @xmath8 @xcite , here @xmath169 with @xcite @xmath170 presently , we have @xmath171 where @xmath172 refers to the present hubble parameter normalized to @xmath173 . the weak gravity range ( [ eqn_w2 ] ) of ( [ eqn_b0 ] ) is well known for its description of galaxy rotation curves .
however , a constant milgrom parameter @xmath8 , does not permit a continuous transition to strong gavity in the sense of @xmath174 .
this observation has led to various phenomenological interpolation functions for @xmath8 as a function of acceleration @xcite .
define a transition to weak gravity ( [ eqn_w1 ] ) as rindler and cosmological horizon collude . for @xmath0 , inertia
@xmath4 drops below the newtonian value @xmath5 .
this transition is sharp as manifest in galaxy rotation curves , here plotted as @xmath175 as a function of newtonian gravitational acceleration @xmath176 based on baryonic matter content . shown
are binned data accompanied by @xmath177 uncertainties .
the curved green band is the theoretical curve covering @xmath178 with @xmath179 km s@xmath12mpc@xmath12 .
( data from @xcite ) . ] in the transition region between ( [ eqn_w2 ] ) and ( [ eqn_w1 ] ) , we encounter the momentum dependent coefficient @xmath180 . in associating @xmath181 with an isotropic thermal distribution of vacuum fluctuations ,
we propose a thermal average @xmath182 with @xmath183 . for numerical evaluation , we express ( [ eqn_b ] ) as @xmath184 with @xmath185 , @xmath186 , @xmath187 . with @xmath188 , the thermal average @xmath189 has the property that @xmath190 ( to be precise , @xmath191 for @xmath192 and @xmath193 for @xmath194 ) , effectively ensuring continuity @xmath195 in the transition to weak gravity ( [ eqn_w1 ] ) , while @xmath196 as @xmath197 .
2 shows a confrontation of the perturbed inertia as a function of newtonian acceleration @xmath198 based on the observed baryonic matter in spiral galaxies . the data in fig .
2 show a remarkably sharp transition at @xmath199 , here identified with ( [ eqn_w1 ] ) . with ( [ eqn_b ] ) , ( [ eqn_b0 ] )
gives a well - defined and accurate transition to milgrom s law at @xmath200 .
measurements of hubble parameter @xmath201 and the deceleration parameter @xmath202 provide a powerful probe of cosmic expansion that may be obtained from local surveys independent of cosmological models .
inverting ( [ eqn_a0 ] ) gives measurement of the deceleration parameter , @xmath203 for @xmath204 , we encounter uncertainties in @xmath205 and @xmath206 . a best fit to the data of fig .
2 based on least square errors shows @xmath207 based on the data points in the weak gravity range ( [ eqn_w1 ] ) and @xmath208 when augmented with the first data point about the transition to strong gravity . from local surveys , a typical hubble parameter is @xmath179 km s@xmath12mpc@xmath12 in tension by about 7% with @xmath209 km s@xmath12mpc@xmath12 obtained from cmb analysis .
based on ( [ eqn_qr ] ) , uncertainties in @xmath3 contribute to @xmath210 .
since @xmath211 in ( [ eqn_qr ] ) , ( [ eqn_qr ] ) gives a mild constraint @xmath212 on @xmath213 .
alternatively , we may turn to data from the cmb .
the dark matter density @xmath24 in ( [ eqn_u3b ] ) is similar but not identical to a cold dark matter density @xmath214 in @xmath16cdm .
in particular , @xmath215 and @xmath216 in the matter dominated era @xmath217 .
this precludes a direct comparison of @xmath24 with dark matter measurements obtained by fitting @xmath16cdm to planck data .
in fact , their somewhat different evolution , here attributed to a dynamical ( [ eqn_lambda ] ) versus constant dark energy density , is conceivably part of the notorious tension in , e.g. , the hubble parameter in @xmath16cdm extracted from the cmb or from the local universe @xcite .
planck estimates of @xmath25 and the baryonic matter content @xmath9 are likewise somewhat model dependent . since @xmath16cdm provides a leading order approximation to better than 10% , we next consider the following
. for @xmath124 , ( [ eqn_da ] ) and ( [ eqn_a1 ] ) define @xmath218 and @xmath219 , satisfying @xmath220 with matter density @xmath221 , where @xmath222 denotes the total mass energy @xmath223 .
. then @xmath225
. consistency with ( [ eqn_m ] ) obtains when @xmath226 , that is @xmath227 the planck value @xmath228 gives @xmath229 and a deceleration parameter @xmath230 since @xmath231 in ( [ eqn_qc ] ) , ( [ eqn_qc ] ) gives a relatively secure constraint on @xmath213 , more so than ( [ eqn_qr ] ) for the reasons stated above .
high resolution galaxy rotation curve data @xcite reveal a remarkably sharp transition in inertia ( fig .
2 ) at the collusion of the rindler and cosmological horizon ( [ eqn_a2a ] ) . in practical terms
, it defines a transition to an inverse distance law of gravity in ( [ eqn_w1 ] ) , the asymptotic regime ( [ eqn_w2 ] ) of which is known as the tully - fisher law or , equivalently , milgrom s law ( [ eqn_b0 ] ) with ( [ eqn_a0 ] ) .
the transition of ( [ eqn_a2a ] ) between ( [ eqn_a2a ] ) and the latter is described by a running @xmath232 value , modeled by a thermal average @xmath233 that takes @xmath234 from effectively @xmath148 down to @xmath8 as a function of @xmath235 .
the outcome agrees with the galaxy rotation curves within measurement uncertainties . in identifying inertia with a thermodynamic potential that is perturbed in weak gravity ( [ eqn_a1 ] ) ,
galaxy rotation curves are sensitive to the cosmological background . in particular
, sensitivity to @xmath236 in ( [ eqn_a0 ] ) gives rise to ( [ eqn_qr ] ) , that may be probed for redshift dependence in future galaxy rotation surveys to determine @xmath237 , @xmath238 with @xmath239 .
if sufficiently well resolved , measurement of @xmath240 in ( [ eqn_qz1 ] ) suggests a direct test of dynamical versus static dark energy , i.e. , ( [ eqn_lambda ] ) versus @xmath16cdm , with @xcite @xmath241 in our model of weak gravity , the mass scale @xmath242 ev sets a minimum to that of dark matter . while pairing in condensations
may conceivably increase its mass ( e.g. * ? ? ?
* ) , this low mass is expected to effectively limit clustering to the scale of galaxy clusters ( e.g. * ? ? ?
* ) . as such
, it will be undetectable by laboratory experiments seeking direct interactions with baryonic matter .
instead , it may be probed by ( [ eqn_qb ] ) and ( [ eqn_qr ] ) , and the associated dark energy by ( [ eqn_qz1]-[eqn_qz2 ] ) . alternatively , satellite based
, free fall cavendish type experiments may probe ( [ eqn_a2 ] ) , rescaled to laboratory size and mass @xmath106 to @xmath243 .
99 ade , p.a.r .
, aghanim , n. , alves , m.i.r . , et al . , 2014 , a&a , 571 , a1
; planck collaboration xvi 2014 , a&a , 571 , a16 bekenstein , j.d . , 1973 ,
d 7 , 2333 bekenstein , j.d .
d , 23 , 287 ; t hooft , g. , 1993 , arxiv : gr - qc/9310026 ; susskind , l. , 1995 , j. math .
( n.y . ) , 36 , 6377 ; van putten , m.h.p.m . , 2012 ,
d , 85 , 064046 bekenstein , j. , 2009 , nucl .
a , 826 , 555 ; van dokkum , p. , abraham , r. , brodie , j. , et al . , 2016 , apj , 828 , l6 cai r .-
g . , kim s. p. , 2005
, j. high energy phys . , 2 , 50 birrell , n.d . , & davies , p.c.w . , 1982 , _ quantum fields in curved space _ , cambridge university press , cambridge . cook , g.b . , 2000 , living rev .
relativity 3 , 5 , http://www.relativity.livingreviews.org/articles/lrr-2000-5 ; wald , r.m . , & iyer , v. , 1991 , phys .
d 44 , r3719 ; brewin , l. , 1988 , phys .
d 38 , 3020 ; cook , g.b . , & abrahams , a.m. , 1992 , phys . rev .
d 46 , 702 ; thornburg , j. , 2007 , living rev .
relativity 10 , 7 , http://relativity.livingreviews.org/articles/lrr-2007-3 york , j.w .
, 1989 , in frontiers in numerical relativity , edited by c. r. evans , l. s. finn , and d. w. hobill ( cambridge university press , cambridge ) famae , b. , & mcgaugh , s.s . , 2012 , living reviews , 10 gibbons g. w. , hawking s. w. , 1977 , phys . rev .
d , 15 , 2738 hawking , s. , 1975 , commun .
43 , 199 ; wald , r.m .
phys . , 1975 , 45 , 9 lelli , f. , mcgaugh , s.s . ,
schombert , j.m . ,
pawlowski , m.s .
, 2016 , arxiv:1610.08981v1 ; http://astroweb.case.edu/sparc/rarbins.mrt mcgaugh , s.s .
, apj , 632 , 859 ; mcgaugh , s.s . , &
shombert , j.m . , 2015 , apj , 802 , 18 ;
mcgaugh , s.s . , 2011a , astron . j. , 143,40 ; mcgaugh , s.s .
, 2011b , phys .
lett . , 106 , 121303 perlmutter , s. , et al .
, 1999 , apj , 517 , 565 ; riess , a. , et al . , 1998 ,
apj , 116 , 1009 mcgaugh , s.s . , 2011a , astron . j. , 143,40 mcgaugh , s.s . ,
2011b , phys .
, 106 , 121303 milgrom , m. , 1983 , apj , 270 , 365 narnhofer , h. , peter , i. , & thirring , w. 1996 , ijmpb , 10 , 1507 ; jacobson , t. 1998 , cqgra , 15 , 151 ; deser , s. , & levin , o. 1997 , gen .
quantum grav , 14 , l163 pirani , f.a.e .
, 1957 , 105 , 1089 penrose , r. , 1965 , phys .
14 , 57 riess , a.g . ,
macri , l.m . ,
hoffmann , s.l . ,
scolnic , d. , stefano , c. , et al . , 2016 ,
apj , 826 , 56 van putten , 2015 , mnras , 450 , l48 unruh , w. g. , 1976 , phys .
d , 14 , 870 van putten , m.h.p.m . , & eardley , d.m .
, 1996 , phys .
d. , 53 , 3056 van putten , m.h.p.m .
, 2010 , phys .
lett . a , 374 , 3346 van putten , m.h.p.m .
, 2016 , apj , 824 , 43 van putten , m.h.p.m . , 2015 ,
ijmp , 24 , 1550024 verlinde , e. 2011 , jhep , 4 , 29 vikram , v. , chang , c. , & jain , b. , et al . , 2015 ,
d , 92 , id.022006 wald , r.m .
, 1984 , general relativity university of chicago press , chicago ) | we consider weak gravity at accelerations @xmath0 when rindler and cosmological horizon collude at @xmath1 , where @xmath2 is the velocity of light and @xmath3 is the hubble parameter .
this is manifest in reduced inertia @xmath4 , below the value @xmath5 in newtonian gravity .
striking evidence for a sharp transition to weak gravity is found in galaxy rotation curves .
their sensitivity to the cosmological background is expressed by correlations to the deceleration parameter @xmath6 and @xmath7 , where @xmath8 is milgrom s scale in the baryonic tully - fisher relation of spiral galaxies and @xmath9 is the baryonic matter density .
the planck value @xmath10 with @xmath11 km s@xmath12 mpc@xmath12 shows @xmath13 .
future surveys may determine @xmath14 to provide a direct test for dynamical dark energy ( @xmath15 ) versus @xmath16cdm ( @xmath17 ) . |
the maximally supersymmetric yang mills theory , @xmath0 , is an extremely special four dimensional field theory . in this letter
we demonstrate that the colour ordered one - loop scattering amplitudes are remarkably close to a universal form : @xmath4 where @xmath2 is helicity - independent and @xmath5 is a helicity - dependent correction .
@xmath2 contains the exact @xmath6 and @xmath7 infrared singular terms .
we find that the error is surprisingly small , being less than @xmath8 for the explicit six , seven and eight point amplitudes we study .
the exact analytic forms of many one - loop amplitudes are known , however approximate universality is not manifest from an analytic study of these expressions .
instead it is observed when we numerically evaluate these analytic forms . in the next section we review the known analytic forms of @xmath0 one - loop @xmath9-point amplitudes and in section 3
we present a numerical study of these expressions for six and seven gluon amplitudes together with the nmhv eight - point amplitudes .
the one - loop amplitudes for gluon scattering in @xmath0 super - yang mills are particularly simple , being heavily constrained by the large symmetry .
they can be expressed entirely in terms of scalar boxes @xcite : @xmath10 the amplitude is completely defined by specifying the rational coefficients @xmath11 .
this is a considerable simplification relative to a generic yang
mills amplitude which would contain scalar triangle functions , scalar bubble functions and additional rational pieces .
the set of scalar box functions is @xmath12 with the labeling as indicated in figure [ boxesfigure ] .
+ such amplitudes are `` cut - constructible '' meaning they may be reconstructed using cutting rules with on - shell four dimensional tree amplitudes @xcite .
the coefficients @xmath11 may be determined in various ways , originally via the two - particle cuts @xcite , but perhaps most conveniently using `` quadruple cuts '' @xcite .
the quadruple cuts give the box - coefficients as a product of four tree amplitudes , @xmath13 the sum is over all allowed intermediate configurations and particle types @xcite . using the expansion of eq .
( [ boxexpansion ] ) together with the quadruple cuts makes the computation of one - loop amplitudes considerably more tractable than the corresponding qcd amplitudes .
it is convenient to define rescaled box functions @xmath14 by @xmath15 the @xmath16 functions are the box functions with the appropriate momentum prefactors removed ; explicit formulae may be found in , for example , ref.@xcite
. the momentum invariants are @xmath17 and @xmath18 .
analytic results will be given in the four - dimensional helicity regularisation scheme ( fdh ) @xcite with the box integrals defined in @xmath19 dimensions .
the amplitudes in other schemes such as t hooft veltman are simply related to those in fdh via @xmath20 which leads to a small modification of @xmath2 but leaves @xmath5 unchanged . in any supersymmetric theory
the gluon scattering amplitudes with all or all - but - one helicities of the same type vanish @xmath21 and the simplest non - vanishing amplitudes are the `` mhv '' ( maximally helicity - violating ) amplitudes which have exactly two negative helicity gluons . at tree level
these amplitudes are given by the parke
taylor formula @xcite , @xmath22 for a partial amplitude where @xmath23 and @xmath24 label the legs with negative helicity .
we use the notation @xmath25 , @xmath26 , with @xmath27 being a massless weyl spinor with momentum @xmath28 and chirality @xmath29 @xcite .
the spinor products are related to momentum invariants by @xmath30 with @xmath31 .
the one - loop mhv amplitudes in @xmath0 have the form @xcite : @xmath32 where the sum of @xmath33 functions is over all possible inequivalent functions with the appropriate cyclic ordering of legs .
the form @xmath2 is universal in that it does not depend upon which two legs have negative helicity .
the sum over box coefficients can be expressed in terms of logarithms and dilogarithms via @xmath34 where all indices are understood to be mod @xmath9 and @xmath35 is the renormalisation scale .
the forms of @xmath36 and @xmath37 depend upon whether @xmath9 is odd or even .
for @xmath38 , @xmath39 \biggr)\ , \equn\ ] ] @xmath40 whereas for @xmath41 , @xmath42 \biggr ) -\sum_{i=1}^{n/2 } \li \biggl [ 1- { \tn{m-1}{i } \tn{m+1}{i-1 } \over \tn{m}{i } \tn{m}{i-1 } } \biggr]\ , \equn\ ] ] @xmath43 the nmhv amplitudes have also been calculated @xcite and have very different forms from the mhv amplitudes . for example
let us examine the six - point amplitudes .
the three independent nmhv amplitudes are constructed from the one - mass box functions , @xmath44 and the two - mass - hard boxes @xmath45 in the very special combinations @xmath46 } } \right)^\e - \ln\left({-t_{i}^{[3]}\over -t_i^{[2]}}\right ) \ln\left({-t_{i}^{[3 ] } \over -t_{i+1}^{[2]}}\right ) \cr & \quad - \ln\left({-t_{i}^{[3 ] } \over -t_{i+3}^{[2]}}\right ) \ln\left({-t_{i}^{[3 ] } \over -t_{i+4}^{[2]}}\right ) + \ln\left({-t_{i}^{[3 ] } \over -t_{i+2}^{[2]}}\right ) \ln\left({-t_{i}^{[3 ] } \over -t_{i+5}^{[2]}}\right ) \cr & \quad + { 1\over 2}\ln\left({-t_{i}^{[2 ] } \over -t_{i+3}^{[2]}}\right ) \ln\left({-t_{i+1}^{[2 ] } \over -t^{[2]}_{i+4}}\right ) + { 1\over 2}\ln\left({-t_{i-1}^{[2 ] } \over -t_{i}^{[2]}}\right ) \ln\left({-t_{i+1}^{[2 ] } \over -t_{i+2}^{[2]}}\right ) \cr & \quad + { 1\over 2}\ln\left({-t_{i+2}^{[2]}\over -t_{i+3}^{[2]}}\right ) \ln\left({-t_{i+4}^{[2 ] } \over -t_{i+5}^{[2 ] } } \right ) + { \pi^2\over3}\ . \cr } \equn\label{wdef}\ ] ] as we can see the dilogarithms drop out of this expression .
this feature of the @xmath0 nmhv six - point amplitudes persists for amplitudes involving external states other than gluons @xcite but not beyond six points . the first nmhv @xmath0 amplitude is given by @xmath47 where the coefficients @xmath48 are given in terms of the @xmath49 function @xmath50 with @xmath51 and @xmath52 for convenience we use @xmath53 and @xmath54\langle 2\:4 \rangle + [ 1\:3]\langle3\:4\rangle$ ] , etc .
the other amplitudes are : @xmath55 , \cr } \equn\ ] ] where @xmath56 and @xmath57 , \cr } \equn\ ] ] where @xmath58 the seven - point `` split helicity '' nmhv amplitude , @xmath59 , was computed in ref .
@xcite and the remaining seven - point nmhv amplitudes in ref .
@xcite , with the generalisation to all-@xmath9 being given in ref .
the detailed formulae are given in these references , however the pattern for the nmhv is one of increasing complexity .
for example the seven - point amplitude is @xmath60 which is a sum over all possible seven - point box functions .
the formulae for the 35 coefficients are given in two different forms in ref .
@xcite and ref .
@xcite . very few coefficients are absent .
for example the amplitude @xmath59 has 29 non - vanishing coefficients while in the amplitude @xmath61 all 35 coefficients are non - zero .
we have implemented the expressions of ref .
@xcite and used them to study the seven- and eight - point nmhv amplitudes numerically .
in this section we present a numerical study of approximate universality in @xmath0 one - loop amplitudes .
since four- and five - point amplitudes must be mhv ( or conjugate to mhv ) these amplitudes have an exact universality ; therefore we start our study with the six - point amplitudes .
the functional forms of the different components of the six - point amplitudes are gathered together in @xcite together with the implementation of most of these in mathematica .
these components have also been calculated at a specific kinematic point and evaluated numerically by ellis , giele and zandereghi ( egz ) @xcite . for convenience
we first examine the amplitudes at this kinematic point using the published numerical values . in table
[ sixpointn4 ] we extract the @xmath0 amplitudes and split them into their @xmath6 , @xmath7 and @xmath62 pieces .
we also look at the combination @xmath63 .
if the amplitudes were exactly universal this combination would be identical for each helicity configuration .
the numerical value of the tree amplitude is embedded within this table since the coefficient of @xmath6 is just @xmath64 .
we have only included a single mhv amplitude since the combination @xmath63 is identical for the different mhv amplitudes . as we can see , despite the very different analytic forms of the nmhv expressions , the amplitudes are numerically almost universal .
.the almost universal nature of the six - gluon @xmath0 amplitudes at the egz kinematic point . [ cols="^,^,^,^,^",options="header " , ]
there are many surprises but no accidents in quantum field theory .
we have been surprised by the almost universal nature of the one - loop amplitudes we have studied here .
we have seen approximate universality in maximally supersymmetric gauge theory but not in pure gauge theory , minimally supersymmetric theories , nor @xmath65 supergravity .
the main contributions to the scattering appear completely determined by the tree amplitude .
explanations for this could be simply the amplitude being largely determined by its infrared behaviour , or more intriguingly , an indication of a broken symmetry such as dual conformal symmetry @xcite
. it would be interesting to explore this behaviour beyond one loop and possibly relate it to some of the other remarkable results in the @xmath66-matrix of @xmath0 super - yang
mills @xcite , and to explore whether approximate universality continues , or indeed becomes exact , in theories related to @xmath0 yang
mills such as twistor string theory @xcite .
n. e. j. bjerrum - bohr , d. c. dunbar , h. ita , w. b. perkins and k. risager , jhep * 0612 * ( 2006 ) 072 [ hep - th/0610043 ] ; + z. bern , n. e. j. bjerrum - bohr and d. c. dunbar , jhep * 0505 * ( 2005 ) 056 [ hep - th/0501137 ] ; + z. bern , j. j. carrasco , d. forde , h. ita and h. johansson , phys .
d * 77 * ( 2008 ) 025010 [ arxiv:0707.1035 [ hep - th ] ] ; + n. e. j. bjerrum - bohr and p. vanhove , jhep * 0810 * ( 2008 ) 006 [ arxiv:0805.3682 [ hep - th ] ] . | a study of the colour - ordered one - loop amplitudes in @xmath0 sym reveals a surprising property : numerically , the amplitudes have an approximately universal form @xmath1 , where @xmath2 is helicity independent . this form is exact if the amplitude is a mhv amplitude , but has an `` error '' which is typically less than 1% for the six , seven and eight gluon nmhv one - loop amplitudes . + hep - th / yymmnnn + 2.cm * approximate universality of @xmath0 super - yang
mills one - loop amplitudes * 2.cm 2.cm david c. dunbar@xmath3 , james h. ettle and warren b. perkins 0.5 cm 0.5 cm _ department of physics + swansea university + swansea , sa2 8pp , uk _ .3 cm |
random matrix models @xcite ( @xcite for a review ) have attracted great amount of interest both due to its mathematical structure and for its manifold physical applications as well .
when studying matrix models , one is especially interested in the solution of matrix models in the large @xmath2 approximation , which corresponds to the planar feynman diagrams @xcite .
these models are characterized by the @xmath3 by @xmath3 matrix variable @xmath4 and by an hamiltonian @xmath5 after diagonalization of the matrix , one is able to work in the eigenvalue space , and to consider the following expression for the partition function : @xmath6 with @xmath7 or @xmath8 , depending on the symmetry .
equivalently , from the perspective of random matrix theory one studies the joint probability distribution function for the @xmath3 eigenvalues of the matrix @xcite .
it has the well - known general form : @xmath9,\end{aligned}\ ] ] @xmath10 and @xmath11 are named weight function and confining potential , respectively . if the elements of the random matrix are believed to be statistically independent from each other , one obtains the quadratic confinement potential @xmath12 , leading to the gaussian ensembles of random matrices @xcite . in the usual physical applications of matrix models
@xcite , the confining potentials @xmath13 are such that the weight function @xmath14 $ ] is determined by the knowledge of all of its positive integer moments @xmath15 where @xmath16 in this paper , we want to study some general properties of matrix models with a weight function that does not satisfy this property .
interestingly enough , a very particular model in this category shows considerable interest in the context of chern - simons theory @xcite and topological strings @xcite .
we briefly remind now the formalism of orthogonal polynomials . recall that the relevant quantities , such as the density of states and the correlation functions , are obtained from the two - point kernel @xcite : @xmath17 with @xmath18 , where @xmath19 are the orthogonal polynomials associated to the weight function @xmath20this is the fundamental object , due to the following classical result @xcite : @xmath21 _ { q , p=1,2, ... ,n}.\ ] ] where the correlation functions are defined as : @xmath22 for hermitian ensembles ( @xmath23 ) ,
the density of states is the kernel at the origin : @xmath24 the second expression follows when applying the christoffel - darboux formula @xcite . for gaussian ensembles for example
, this quantity tends to the well - known semi - circle law in the limit @xmath25 @xcite . in this work
we shall consider hermitian ensembles , and this expression for the density of states will be useful to understand relevant conceptual points in our discussion .
as mentioned , we are concerned with the special properties of matrix models with an indeterminate weight function .
our discussion is mainly based on the classical moment problem @xcite ( see @xcite for a recent review , that we will follow ) .
the connection with random matrix theory can be readily guessed by the fact that only the moments play a role in the orthogonalization procedure .
therefore the two sets , the moments and the orthogonal polynomials , are essentially equivalent . the paper is organized as follows : in the next section , we give a short introduction to chern - simons theory and , focussing on these recent developments , we show how the matrix models in @xcite are directly related with an hermitian matrix model with a log - normal weight function . in this way , one can do exact computations through the associated orthogonal polynomials . to illustrate this , we compute the partition function of chern - simons theory on @xmath1 with gauge group @xmath26 interestingly enough , the use of @xmath0-deformed orthogonal polynomials readily leads to the natural parameter of chern - simons theory @xmath27 .
therefore , the orthogonal polynomials approach is a method for non - perturbative solutions in chern - simons theory . in section 3 ,
we show that there are actually infinitely many matrix models with the same partition function .
finally , we show that this is a general feature of models with very weak confining potential . to conclude , we present a summary and possible directions for further research .
we begin by outlining basic and well - known facts about chern - simons theory , emphasizing recent results and trends that connect chern - simons theory with matrix models @xcite . as it is well - known , chern - simons theory is a topological quantum field theory whose action is built out of a chern - simons term involving as gauge field a gauge connection associated to a group @xmath28 on a three - manifold @xmath4 .
the action is : @xmath29 with @xmath30 an integer number .
the natural associated observables are the correlators of wilson loops , and its main interest come from the fact that these correlators lead to quantum - group polynomial invariants of knots and links @xcite . for a review of the field since the seminal work @xcite ,
see @xcite .
we mention now some few developments related to chern - simons theory . as advanced in @xcite ,
chern - simons gauge theory has a string description in the sense of t hooft @xcite .
indeed , as it is well - known , gauge theories with the @xmath31 group admit a large @xmath3 expansion . in these expansions ,
correlators are expanded in powers of @xmath32 while keeping the t hooft coupling fixed @xmath33 , with @xmath34 the coupling constant of the gauge theory . in the case of chern - simons theory ,
this large @xmath3 expansion is reminiscent of a string theory expansion .
this connection between chern - simons and topological strings was first pointed out by witten @xcite and has been extended in @xcite .
for example , in @xcite it is shown that if one wraps @xmath3 @xmath35-branes on @xmath4 in @xmath36 , then the associated topological a - model is a @xmath37 chern - simons theory on the three manifold @xmath4 .
on the other hand , regarding matrix models , in @xcite it was already pointed out that the structure of the partition function of chern - simons theory on @xmath38 with gauge group @xmath39 resembles the usual expression for the partition function of a one matrix model in terms of its associated orthogonal polynomials . moreover , in @xcite , it is shown that the partition function of chern - simons theory on @xmath1 and with gauge group @xmath37 , is given by : @xmath40 that describes open topological @xmath41 strings on @xmath42 with @xmath3 branes wrapping @xmath1 ( see the details in @xcite ) .
these type of models have been further considered in @xcite . in @xcite
, it was shown that topological strings for @xmath43-branes are equivalent to hermitian matrix models , then the idea in @xcite is to obtain the results in @xcite , by applying mirror symmetry to obtain @xmath43-brane matrix models .
we will comment further on the results in @xcite later on .
now , we mainly focus on the study of @xmath44 . for recent reviews along these lines see @xcite .
we remind that , as usual in chern - simons theory , the string coupling constant @xmath45 is related with the @xmath30 in @xmath46 by : @xmath47 as explained in @xcite , the limit of the parameter @xmath48 lead to the usual gaussian unitary ensemble .
one can argue , for example , that in this limit , the gaussian becomes a dirac delta function , and therefore the results are independent of the correlation factor , or level repulsion . in @xcite , this model
is essentially studied through the consideration of averages ( in a gaussian unitary ensemble ) of the following quantities : @xmath49 this method leads to a great amount of perturbative information .
nevertheless , we note that the connection with the usual hermitian matrix models -that is , of the type given by @xmath50 - is much simpler .
let us consider the following simple change of variables : @xmath51 then , @xmath44 , reads : @xmath52 the factor @xmath53 can be readily absorbed by considering a simple but remarkable property of the log - normal function @xmath54 @xmath55 note that this functional equation is just a particular case of the following elementary identity : @xmath56 therefore , the chern - simons matrix model is directly related to an ordinary hermitian ensemble with log - normal weight function @xmath57 equivalently , one could have used the change of variables @xmath58 .
notice that the term @xmath59 is necessary to cancel , together with the term coming from the jacobian , the contribution @xmath60 actually , this method was already employed in @xcite , where the same matrix model seems to appear in a very different context .
one of our goals is to make contact between this model and chern - simons theory .
as we shall see , this will comes out rather easily , due to the existence of a closed system of orthogonal polynomials associated to the log - normal weight .
these polynomials are @xmath0-deformed polynomials known as stieltjes - wigert polynomials @xcite .
this opens the possibility of a complete solution for the model @xmath44 of @xcite ( see also @xcite ) , using the classical results for hermitian ensembles and these polynomials .
secondly , we shall show -giving explicit examples- that there exists infinitely many matrix models -deformations of @xmath61- with exactly the same ( chern - simons ) partition function . using the previous results and the stieltjes - wigert orthogonal polynomials @xcite
, we can find a complete solution of the matrix model considered in @xcite , and , in particular , for the partition function .
for this , we can use the following well - known result for the partition function in terms of the orthogonal polynomials @xcite : @xmath62 where the coefficient @xmath63 is : @xmath64 the first step is to compute the @xmath65 associated to the stieltjes - wigert orthogonal polynomials .
the coefficients are given by @xcite : @xmath66 then @xmath67 and @xmath68 therefore : @xmath69 but we are interested in : @xmath70 making the identification @xmath71 , and therefore @xmath72 .
it is remarkable that we have naturally obtained , from this @xmath0 deformed orthogonal polynomials , the usual @xmath0 parameter in chern - simons theory .
the partition function with the usual stieltjes - wigert weight , corresponds to : @xmath73 therefore : @xmath74 to make explicit contact with the typical expressions for the chern - simons partition functions ( as in @xcite for example ) we make some transformations on the product term .
@xmath75 the final expression for the partition function is @xcite : @xmath76 and since @xmath77 , we finally find : @xmath78 it was already mentioned in @xcite , that the structure of the partition function resembled very much the general expression for the partition function of a one matrix model in terms of the associated orthogonal polynomials . here ,
based on the result in @xcite , we are making this statement precise through the explicit computation . notice that the differences between the @xmath31 and @xmath37 comes essentially from the contribution of the partition function of @xmath79 ( see , for example , the first reference in @xcite for more details ) .
in this second part we shall show that the matrix models involved that appear in chern - simons theory present a novelty in comparison with most of known matrix models ( like the ones that are relevant in 2d quantum gravity ) . broadly speaking ,
its weight function is so broad that there is a non - uniqueness issue that leads to very practical and concrete results for the matrix model : there are infinitely many models with identical properties .
in particular , identical ( chern - simons ) partition function .
we begin by briefly reviewing some introductory results in the theory of the moment problem .
the two basic moment problems are @xcite : given a sequence of reals @xmath80 , when is there a measure , @xmath81 on @xmath82 so that @xmath83 and if such measure exists , is it unique ? given a sequence of reals @xmath80 , when is there a measure , @xmath81 on @xmath84 so that @xmath85 and if such measure exists , is it unique ? in our case , we are not concerned with the question regarding existence , since we always refer to the weight function @xmath10 and its associated moments .
then , the main point is to know whether we are in the determinate or indeterminate case . to fix ideas ,
let us mention that there is a natural boundary for functions of the form @xmath86 with @xmath11 of polynomial form . in the hamburger case , the boundary is given by @xmath87 with @xmath88 , while in the stieltjes case it is given by @xmath87 with @xmath89 .
then , in the usual matrix models considered ( as reviewed in @xcite for example ) , one is essentially always in the determined case .
consider now the following generalization of the log - normal weight : @xmath90 .
\label{sti}\ ] ] the moments of this function are @xcite : @xmath91 note that all the integer moments are completely independent of the parameter @xmath92 .
this means that all the functions in the family @xmath93 have the same moments .
thus , they are all undetermined by them .
conversely , one can say that the set of moments @xmath94 is an indeterminate set .
we present now a detailed consideration of this family of functions .
the main point is the computation of its mellin transform .
first , note that for @xmath95 we have the log - normal distribution @xmath96 its mellin transform is : @xmath97 = \mathrm{e}^{\pi ^{2}/4}\mathrm{erf}\left ( -\frac{s}{2}\right ) .\ ] ] the other part of the stieltjes function is @xmath98 .
from @xmath99 we know that all of its integer moments are zero .
the most enlightening possibility is to compute its mellin transform : ] @xmath100 we use the change of variables @xmath101 the integral becomes : @xmath102 dy \\ & = & \sqrt{\pi } \mathrm{e}^{\frac{1}{4}\left ( s^{2}+1\right ) } \sin \left ( \pi \left ( 1+s\right ) \right ) .
\notag\end{aligned}\ ] ] the change of variables is the one employed by stieltjes @xcite , with a generic , complex @xmath103 instead of an integer @xmath30 . clearly , after the change of variables , since @xmath104 is a periodic and odd function , its value is zero at @xmath105 nevertheless , the result is even clearer when looking at the full mellin transform : the contribution of the additional function is zero only for integer values of @xmath106 as a consequence , for example , we have that any of the following functions have the same integer moments . ] ( solid ) and @xmath107 ( dashed ) ] consequently , it is manifest now that there are infinitely many matrix models with exactly the same partition function .
therefore , it turns out that the partition function of @xmath108 chern - simons theory on @xmath1 can be given by an infinitely many different matrix models . from what we have seen ,
one possibility is : @xmath109 as always , with @xmath110 .$ ] notice that due to the correspondence between @xmath111 and @xmath112 through @xmath113 , the density of states of @xmath114 does _ not _ satisfy the property : @xmath115 ( that is , the moments are different for each matrix model ) .
but rather : @xmath116 to extract further physical consequences of these results and its implications for the connection between matrix models and topological quantum field theories seems an interesting open question .
in this concluding section we show that the properties exhibited by the chern - simons matrix model are typical of models with weakly confining -_soft_- potentials .
the mathematically precise description is that the potential is such that @xmath117 is a function that is not determined by its moments .
this feature has already been discussed in the context of random matrix theory and also in the particular case of @xmath0-deformed matrix models @xcite . in this last case ,
the one we are mainly interested , the non - uniqueness is strong since the potentials are asymptotically @xmath118 for @xmath119 @xcite and we have already mentioned that we are in the soft regime when the confinement provided by the potential is weaker than @xmath120 . employing basic results on indeterminate functions @xmath121 and the classical random matrix formulae , one can prove that the density of states of a matrix model is an indeterminate function if its weight function is indeterminate by its moments .
this can be readily shown if we consider the expression for the density of states ( for any @xmath3 ) @xmath122 and krein proposition @xcite , that essentially says that if @xmath123 is convergent , then the corresponding moment problem is undetermined .
an analogous case holds for the stieltjes case . from @xmath122
we have : @xmath124 with @xmath125 a polynomial .
we write the expression in the second way , to emphasize that we just have the addition of a polynomial term . then : @xmath126 and @xmath127 therefore , the convergence or divergence of @xmath128 and @xmath129 is exclusively given by the convergence or divergence of @xmath130 and @xmath131 respectively . then , taking into account krein proposition , it turns out that the matrix models characterized by an indeterminate weight function , have different ( but indeterminate ) density of states but the same partition function .
thus , we have seen that the density of states is an indeterminate function .
let us show explicitly that their moments are all equal : @xmath132 that is clearly @xmath133 dependent and the moments : @xmath134 are @xmath135independent , since the polynomial @xmath136 is always the same and then we are lead to a sum of moments of the indeterminate function @xmath137 , each one of them @xmath135independent .
now , we can show that the partition functions are identical .
we use the following expression : @xmath138 that is , the partition function as the @xmath139-moment of the density of states .
alternatively , the usual expression for the partition function in terms of the orthogonal polynomials can be used @xcite : @xmath140 where @xmath63 are the following coefficients of the associated orthogonal polynomials : @xmath141 this expression is valid for any weight function and only depends on the associated orthogonal polynomials , that are the same for any @xmath142 then , for example , all matrix models with the weight functions @xmath143 , have the same orthogonal polynomials ( essentially the stieltjes - wigert @xcite , see above ) .
nevertheless , it is plain that each one will have a different density of states , since we have a different weight function ( as we have seen in the previous figures ) . in any case
, it is clear that approaching matrix models only from the perspective of orthogonal polynomials , there is no a priori reason to choose the log - normal weight , since any member of the infinite family has identical orthogonal polynomials .
furthermore , the family of functions above discussed is by no means the only one . from our analysis with the mellin transform ,
is clear that many other functions instead of @xmath104 , can lead to the same phenomena .
in addition , examples such as @xcite : @xmath144 can be relatively easily found in the literature .
this weight function also leads to the stieltjes - wigert polynomials , that are actually terminating basic hypergeometric series and have intimate relationships with theta functions @xcite . in this work
, we have discovered their remarkable connection with chern - simons theory and quantum topological invariants . in general , it turns out that the set @xmath145 of solutions to an indeterminate moment problem contains very different types of measures : measures @xmath146 with a @xmath147density , discrete measures and measures which are continuous singular @xcite
. explicit cases of these last types are found in @xcite .
we have seen that the matrix models that appear in chern - simons theory are exactly solvable and that have many interesting properties , which are novel in comparison with the matrix models that appear in 2d quantum gravity .
therefore , the orthogonal polynomials technique constitute an interesting non - perturbative method .
we have checked this for the simple case of @xmath1 and gauge group @xmath37 and , interestingly enough , the use of @xmath0-deformed orthogonal polynomials readily leads to the usual parameter of chern - simons theory .
it seems feasible , and certainly an interesting task , to try to extend this result to other geometries -like lens spaces- and to other gauge groups as well @xcite . on the other hand , the non - uniqueness property that we have established does not only seem to be interesting from a mathematical perspective , but it may have some physical meaning as well .
we have emphasized the fact that the original chern - simons matrix model can be deformed in infinitely many ways , leaving the partition function always unmodified .
since this is a chern - simons partition function , we know it is a topological invariant .
a well - known result in topology says that the topological invariant of a three - manifold is independent of the triangulation employed to compute the invariant .
thus , one is tempted to try to interpret all these matrix models as representing different triangulations of the manifold .
thus , in a sense , it would be interesting to try to obtain an analogous interpretation to the role of more ordinary matrix models in 2d quantum gravity , but in three dimensions .
this is a bit speculative line of research , but it may still be worth . in any case , the non - uniqueness properties are a distinctive mathematical feature of the models and the way the it appears seems also rather interesting in itself , since the resulting logarithmic oscillations are quite typical of physical models where a quantum group structure is present
. a more detailed account of this will appear elsewhere .
m. aganagic , a. klemm , m. mario and c. vafa , `` matrix model as a mirror of chern - simons theory , '' jhep * 0402 * , 010 ( 2004 ) [ arxiv : hep - th/0211098 ] .
akhiezer , _ the classical moment problem _ , hafner publishing co. , new york , 1965 . j. shohat and j.d .
tamarkin , _ the problem of moments . _ american mathematical society , providence , 1950 .
r. askey , _ orthogonal polynomials and theta functions _ , in theta functions , bowdoin ( 1987 ) ( part 2 ) .
proceedings of symposia in pure mathematics , vol 49 , amer .
soc . , procidence , ri , 1989 , 299 - 321 .
e. brezin , c. itzykson , g. parisi and j. b. zuber , `` planar diagrams , '' commun .
phys . * 59 * , 35 ( 1978 ) .
t. s. chihara , a characterization and a class of distribution functions for the stieltjes - wigert polynomials , canad .
* 13 * , 529 ( 1970 ) .
chiara , on generalized stieltjes - wigert and related polynomials , j. comput .
appl . math . * 5 * , 291 ( 1979 ) .
r. dijkgraaf and c. vafa , `` matrix models , topological strings , and supersymmetric gauge theories , '' nucl .
b * 644 * , 3 ( 2002 ) [ arxiv : hep - th/0206255 ] .
p. di francesco , p. h. ginsparg and j. zinn - justin , `` 2-d gravity and random matrices , '' phys .
* 254 * , 1 ( 1995 ) [ arxiv : hep - th/9306153 ] .
forrester , vicious random walkers in the limit of a large number of walkers , j. stat .
phys . * 56 * , 767 ( 1989 ) .
r. gopakumar and c. vafa , `` topological gravity as large n topological gauge theory , '' adv .
theor . math .
* 2 * , 413 ( 1998 ) [ arxiv : hep - th/9802016 ] .
r. gopakumar and c. vafa , `` on the gauge theory / geometry correspondence , '' adv .
theor . math .
* 3 * , 1415 ( 1999 ) [ arxiv : hep - th/9811131 ] ; h. ooguri and c. vafa , `` knot invariants and topological strings , '' nucl .
b * 577 * , 419 ( 2000 ) [ arxiv : hep - th/9912123 ] .
n. halmagyi and v. yasnov , chern - simons matrix models and unoriented strings , jhep * 0402 * , 002 ( 2004 ) [ arxiv : hep - th/0305134 ] ; the spectral curve of the lens space matrix model , arxiv : hep - th/0311117 ; n. halmagyi , t. okuda and v. yasnov , large n duality , lens spaces and the chern - simons matrix model , jhep * 0404 * , 014 ( 2004 ) [ arxiv : hep - th/0312145 ]
. g. t hooft , `` a planar diagram theory for strong interactions , '' nucl .
b * 72 * , 461 ( 1974 ) .
j. m. f. labastida , `` chern - simons gauge theory : ten years after , '' arxiv : hep - th/9905057 .
m. mario , chern - simons theory , matrix integrals , and perturbative three - manifold arxiv : hep - th/0207096 .
m. mario , `` enumerative geometry and knot invariants , '' arxiv : hep - th/0210145 ; a. grassi and m. rossi , `` large n dualities and transitions in geometry , '' arxiv : math.ag/0209044 ; j. m. f. labastida , `` knot theory from the perspective of field and string theory , '' arxiv : hep - th/0201038 ; c. vafa , `` unifying themes in topological field theories , '' arxiv : hep - th/0005180 .
muttalib , y. chen and m.e.h .
ismail , @xmath148random matrix ensembles , in symbolic computation , number theory , special functions , physics and combinatorics ; developments in mathematics , vol 4 , eds .
f.g . garvan and m.e.h .
ismail ( kluwer , 2001 ) .
e. witten , `` quantum field theory and the jones polynomial , '' commun .
phys . *
121 * , 351 ( 1989 ) .
e. witten , `` chern - simons gauge theory as a string theory , '' prog .
math . * 133 * , 637 ( 1995 ) [ arxiv : hep - th/9207094 ] . | we study the properties of matrix models with soft confining potentials .
their precise mathematical characterization is that their weight function is not determined by its moments .
we mainly rely on simple considerations based on orthogonal polynomials and the moment problem .
in addition , some of these models are equivalent , by a simple mapping , to matrix models that appear in chern - simons theory .
the models can be solved with @xmath0 deformed orthogonal polynomials ( stieltjes - wigert polynomials ) , and the deformation parameter turns out to be the usual @xmath0 parameter in chern - simons theory . in this way
, we give a matrix model computation of the chern - simons partition function on @xmath1 and show that there are infinitely many matrix models with this partition function . |
spin glasses are magnetic materials with strange properties that distinguish them from ordinary ferromagnets . in statistical physics , the study of spin glasses originated with the works of edwards and anderson @xcite and sherrington and kirkpatrick @xcite in 1975 . in the following decade
, the theoretical study of spin glasses led to the invention of deep and powerful new methods in physics , most notably parisi s broken replica method .
we refer to @xcite for a survey of the physics literature . however , these physical breakthroughs were far beyond the reach of rigorous proof at the time , and much of it remains so till date .
the rigorous analysis of the sherrington - kirkpatrick model began with the works of aizenman , lebowitz and ruelle @xcite and frhlich and zegarliski @xcite in the late eighties ; the field remained stagnant for a while , interspersed with a few nice papers occasionally ( e.g. @xcite , @xcite ) .
the deepest mysteries of the broken replica analysis of the s - k model remained mathematically intractable for many more years until the path - breaking contributions of guerra , toninelli , talagrand , panchenko and others in the last ten years ( see e.g. @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite ) .
arguably the most notable achievement in this period was talagrand s proof of the parisi formula @xcite .
however , in spite of all this remarkable progress , our understanding of these complicated mathematical objects is still shrouded in mystery , and many conjectures remain unresolved . in this article
we attempt to give a mathematical foundation to some aspects of spin glasses that have been well - known in the physics community for a long time but never before penetrated by rigorous mathematics .
let us now embark on a description of our main results .
further references and connections with the literature will be given at the appropriate places along the way .
consider the following simple - looking probabilistic question : suppose @xmath1 are i.i.d .
standard gaussian random variables , and we define , for each @xmath2 ,
the quantity @xmath3 then is it true that with high probability , there is a large subset @xmath4 of @xmath5 such that @xmath6 and any two distinct elements @xmath7 of @xmath4 are nearly orthogonal , in the sense that @xmath8 ( in the spin glass literature , the quantity @xmath9 is called the ` overlap ' between the ` configurations ' @xmath10 and @xmath11 . ) to realize the non - triviality of the question , consider a slightly different gaussian field @xmath12 on @xmath5 , defined as @xmath13 where @xmath14 are i.i.d .
standard gaussian random variables .
then clearly , @xmath12 is maximized at @xmath15 , where @xmath16 .
note that for any @xmath17 , @xmath18 it is not difficult to argue from here that if @xmath17 is another configuration that is near - maximal for @xmath12 , then @xmath17 must agree with @xmath15 at nearly all coordinates .
thus , the field @xmath12 does not satisfy the ` multiple peaks picture ' that we are investigating about @xmath19 .
this is true in spite of the fact that @xmath20 and @xmath21 are approximately independent for almost all pairs @xmath22 .
we have the following result about the existence of multiple peaks in the field @xmath19 .
it says that with high probability , there is a large collection @xmath4 of configurations satisfying and , that is , @xmath23 for any two distinct @xmath24 , and @xmath25 for each @xmath26 .
[ multisk ] let @xmath19 be the field defined in , and define the overlap @xmath27 between configurations @xmath7 by the formula .
let @xmath28 then there are constants @xmath29 , @xmath30 , @xmath31 , and @xmath32 such that with probability at least @xmath33 , there is a set @xmath34 satisfying 1 .
@xmath35 , 2 .
@xmath36 for all @xmath37 , @xmath38 , and 3 .
@xmath39 for all @xmath26 .
quantitatively , we can take @xmath40 , @xmath41 , @xmath42 and @xmath43 , where @xmath44 is an absolute constant
. however these are not necessarily the best choices .
let us now discuss the implication of this result in spin glass theory .
the sherrington - kirkpatrick model of spin glasses , introduced in @xcite , is defined through the hamiltonian ( i.e. energy function ) @xmath45 the s - k model at inverse temperature @xmath46 defines a probability measure @xmath47 on @xmath5 through the formula @xmath48 where @xmath49 is the normalizing constant .
the measure @xmath47 is called the gibbs measure . according to the folklore in the statistical physics community , the energy landscape of the s - k model has ` multiple valleys ' .
although no precise formulation is available , one way to view this is that there are many nearly orthogonal states with nearly minimal energy . for a physical discussion of the ` many states ' aspect of the s - k model
, we refer to @xcite , chapter iii .
a very interesting rigorous formulation was attempted by talagrand ( see @xcite , conjecture 2.2.23 ) , but no theorems were proved . although our achievement is quite modest , and may not be satisfactory to the physicists because we do not prove that the approximate minimum energy states correspond to significantly large regions of the state space in fact , one may say that it is not what is meant by the physical term ` multiple valleys ' at all because an isolated low energy state does not necessarily represent a valley it does seem that theorem [ multisk ] is the first rigorous result about the multimodal geometry of the sherrington - kirkpatrick energy landscape .
we may call it ` multiple valleys in a weak sense ' .
theorem [ multisk ] can be generalized to the following corollary , which shows that weak multiple valleys exist at ` every energy level ' and not only for the lowest energy .
[ multisk2 ] let all notation be the same as in theorem [ multisk ] . fix a number @xmath50 $ ] .
then for all sufficiently large @xmath51 , with probability at least @xmath52 there exists a set @xmath53 satisfying conditions and of theorem [ multisk ] , such that @xmath54 for all @xmath55 . the variables @xmath1 in the hamiltonian @xmath56 are collectively called the ` couplings ' or the ` disorder ' .
our proof of theorem [ multisk ] is based on the chaotic nature of the s - k model under small perturbations of the couplings ; this is discussed in the next subsection .
the relation between chaos and multiple valleys follows from a general principle outlined in @xcite , although the proof in the present paper is self - contained .
recall the gibbs measure @xmath47 of the s - k model , defined in .
suppose @xmath10 and @xmath11 are two configurations drawn independently according to the measure @xmath47 , and the overlap @xmath9 is defined as in .
it is known that when @xmath57 , @xmath58 with high probability @xcite .
however , it is also known that @xmath9 can not be concentrated near zero for all @xmath59 , because that would give a contradiction to the existence of a phase transition as established in @xcite .
in fact , it is believed that the limiting distribution of @xmath9 in the low temperature phase is given by the so - called ` parisi measure ' , a notion first made rigorous by talagrand @xcite .
now suppose we choose @xmath11 not from the gibbs measure @xmath47 , but from a new gibbs measure @xmath60 , based on a new hamiltonian @xmath61 which is obtained by applying a small perturbation to the hamiltonian @xmath56 .
( we will make precise the notion of a small perturbation below . )
is it still true that @xmath9 has a non - degenerate limiting distribution at low temperatures ?
the conjecture of disorder chaos ( i.e. chaos with respect to small fluctuations in the disorder @xmath1 ) states that indeed that is not the case : @xmath9 is concentrated near zero if @xmath10 is picked from the gibbs measure and @xmath11 is picked from a perturbed gibbs measure .
this is supposed to be true at all temperatures .
to the best of our knowledge , disorder chaos for the s - k model was first discussed in the widely cited paper of bray and moore @xcite ; a related discussion appears in the earlier paper @xcite .
the phenomenon of chaos itself was first conjectured by fisher and huse @xcite in the context of the edwards - anderson model , although the term was coined in @xcite . again , to the best of our knowledge , nothing has been proved rigorously yet .
for further references in the physics literature , let us refer to the recent paper @xcite .
note that this idea of chaos should not be confused with temperature chaos ( also discussed in @xcite ) , which says that spin glasses are chaotic with respect to small changes in the inverse temperature @xmath59 .
we shall consider two kinds of perturbation of the disorder .
the first , what we call ` discrete perturbation ' , is executed by replacing a randomly chosen small fraction of the couplings @xmath62 by independent copies . here
small fraction means a fraction @xmath0 that goes to zero as @xmath63 .
discrete perturbation is the usual way to proceed in the noise - sensitivity literature ( see e.g. @xcite ) .
in fact , it seems that the following result is intimately connected with noise - sensitivity , although we do not see any obvious way to use the standard noise - sensitivity techniques to derive it .
[ chaosdisc ] consider the s - k model at inverse temperature @xmath59 .
take any @xmath51 and @xmath64 $ ] .
suppose a randomly chosen fraction @xmath0 of the couplings @xmath62 are replaced by independent copies to give a perturbed gibbs measure .
let @xmath10 be chosen from the original gibbs measure and @xmath11 is chosen from the perturbed measure .
let the overlap @xmath9 be defined as in .
then @xmath65 where @xmath44 is an absolute constant and the expectation is taken over all randomness .
this theorem shows that the system is chaotic if the fraction @xmath0 goes to zero slower than @xmath66 .
the derivation of this result is based on the ` superconcentration ' property of the free energy in the s - k model that we present in the next subsection .
the notion of perturbation in the above theorem , though natural , is not the only available notion .
in fact , in the original physics papers ( e.g. @xcite ) , a different manner of perturbation is proposed , which we call continuous perturbation . here
we replace @xmath67 by @xmath68 , where @xmath69 is another set of indepenent standard gaussian random variables and @xmath70 so that the resultant couplings are again standard gaussian .
when @xmath71 , we say that the perturbation is small . a convenient way to parametrize the perturbation is to set @xmath72 , where @xmath73 is a parameter that we call ` time ' .
this nomenclature is natural , because perturbing the couplings up to time @xmath73 corresponds to running an ornstein - uhlenbeck flow at each coupling for time @xmath73 , with initial value @xmath67 .
the following theorem says that the s - k model is chaotic under small continuous perturbations .
[ chaoscont ] consider the s - k model at inverse temperature @xmath59 .
take any @xmath74 .
suppose we continuously perturb the couplings up to time @xmath73 , as defined above .
let @xmath10 be chosen from the original gibbs measure and @xmath11 be chosen from the perturbed measure .
let the overlap @xmath9 be defined as in .
then there is an absolute constant @xmath44 such that for any positive integer @xmath75 , @xmath76 the expectation is taken over all randomness .
again , the achievement is very modest , and does not come anywhere close to the claims of the physicists .
but once again , this is the first rigorous result about chaos of any kind in the s - k model .
to the best of our knowledge , the only other instance of a rigorous proof of chaos in any spin glass model is in the work of panchenko and talagrand @xcite , who established chaos with respect to small changes in the external field in the spherical s - k model .
disorder chaos in directed polymers was established by the author in @xcite .
a deficiency of both theorems in this subsection is that they do not cover the case of zero temperature , that is , @xmath77 , where gibbs measure concentrates all its mass on the ground state . in principle
, the same techniques should apply , but there are some crucial hurdles that can not be cleared with the available ideas .
the notion of superconcentration was defined in @xcite .
the definition in @xcite pertains only to maxima of gaussian fields , but it can be generalized to roughly mean the following : a lipschitz function of a collection of independent standard gaussian random variables is superconcentrated whenever its order of fluctuations is much smaller than its lipschitz constant .
this definition is related to the classical concentration result for the gaussian measure , which says that the order of fluctuations of a lipschitz function under the gaussian measure is bounded by its lipschitz constant ( see e.g. theorem 2.2.4 in @xcite ) , irrespective of the dimension .
the free energy of the s - k model is defined as @xmath78 where @xmath56 is the hamiltonian defined in .
it follows from classical concentration of measure that the variance of @xmath79 is bounded by a constant multiple of @xmath51 ( see corollary 2.2.5 in @xcite ) .
this is the best known bound for @xmath80 . when @xmath57 , talagrand ( theorems 2.2.7 and 2.2.13 in @xcite ) proved that the variance can actually be bounded by an absolute constant
this is also indicated in the earlier works of aizenman , lebowitz and ruelle @xcite and comets and neveu @xcite .
therefore , according to our definition , the free energy is superconcentrated when @xmath57 .
the following theorem shows that @xmath81 is superconcentrated at any @xmath59 .
[ superconc ] let @xmath79 be the free energy of the s - k model defined above in .
for any @xmath59 , we have @xmath82 where @xmath44 is an absolute constant .
this result may be reminiscent of the @xmath83 improvement in the variance of first passage percolation time @xcite . however , the proof is quite different in our case since hypercontractivity , the major tool in @xcite , does not seem to work for spin glasses in any obvious way . in that sense ,
the two results are quite unrelated .
our proof is based on our chaos theorem for continuous perturbation ( theorem [ chaoscont ] ) and ideas from @xcite . on the other hand ,
theorem [ superconc ] is used to derive the chaos theorem for discrete perturbation , again drawing upon ideas from @xcite .
this equivalence between chaos and superconcentration is one of the main themes of @xcite , which in a way shows the significance of superconcentration , which may otherwise be viewed as just a curious phenomenon .
incidentally , it was shown by talagrand ( @xcite , eq . ( 10.13 ) ) that the lower tail fluctuations of @xmath79 are actually as small as order @xmath84 under an unproven hypothesis about the parisi measure .
let @xmath85 be an undirected graph .
the edwards - anderson spin glass @xcite on @xmath86 is defined through the hamiltonian @xmath87 where @xmath62 is again a collection of i.i.d .
random variables , often taken to be gaussian .
the s - k model corresponds to the case of the complete graph , up to normalization by @xmath88 . for a survey of the ( few ) rigorous and non - rigorous results available for the edwards - anderson model
, we refer to newman and stein @xcite . unlike the s - k model , there are two kinds of overlap in the e - a model .
the ` site overlap ' is the usual overlap defined in .
the ` bond overlap ' between two states @xmath10 and @xmath11 , on the other hand , is defined as @xmath89 we show that the bond overlap in the e - a model is not chaotic with respect to small fluctuations of the couplings at any temperature .
this does not say anything about the site overlap ; the site overlap in the e - a model can well be chaotic with respect to small fluctuations of the couplings , as predicted in @xcite .
[ nochaos ] suppose the e - a hamiltonian on a graph @xmath86 is continuously perturbed up to time @xmath90 , according to the definition of continuous perturbation in section [ chaossec ] .
let @xmath10 be chosen from the original gibbs measure at inverse temperature @xmath59 and @xmath11 is chosen from the perturbed measure .
let the bond overlap @xmath91 be defined as in .
let @xmath92 where @xmath93 is the maximum degree of @xmath86 .
then @xmath94 where @xmath44 is a positive absolute constant .
moreover , the result holds for @xmath77 also , with the interpretation that the gibbs measure at @xmath95 is just the uniform distribution on the set of ground states .
an interesting case of the above theorem is when @xmath96 .
the result then says that if two configurations are drawn independently from the gibbs measure , they have a non - negligible bond overlap with non - vanishing probability .
the fact that this holds at any finite temperature is in contrast with the mean - field case ( i.e. the s - k model ) , where there is a high - temperature phase ( @xmath57 ) where the bond overlap becomes negligible . however , while theorem [ nochaos ] establishes that the bond overlap does not become zero for any amount of perturbation , it does exhibit a sort of ` quenched chaos ' , in the following sense .
[ quenchedchaos ] fix @xmath97 and let @xmath91 be as in theorem [ nochaos ]
. then @xmath98 that is , if we perturb the system by an amount @xmath99 , the bond overlap between two configurations drawn from the two gibbs measures is approximately equal to the quenched average of the overlap .
in physical terms , the overlap ` self - averages ' . the combination of the last two theorems brings to light a surprising phenomenon . on the one hand ,
the perturbation retains a memory of the original gibbs measure , because the overlap is non - vanishing in theorem [ nochaos ] . on the other hand
, the perturbation causes a chaotic reorganization of the gibbs measure in such a way that the overlap concentrates on a single value in theorem [ quenchedchaos ] .
the author can see no clear explanation of this confusing outcome .
the proof of theorem [ nochaos ] is based on the following result , which says that the free energy is not superconcentrated in the e - a model on bounded degree graphs .
this generalizes a well - known result of wehr and aizenman @xcite , who proved the analogous result on square lattices .
the relative advantage of our approach is that it does not use the structure of the graph , whereas the wehr - aizenman proof depends heavily on properties of the lattice .
[ nosuper ] let @xmath100 denote the free energy in the edwards - anderson model on a graph @xmath86 , defined in .
let @xmath93 be the maximum degree of @xmath86 . then for any @xmath59 , including @xmath95 ( where the free energy is just the energy of the ground state )
, we have @xmath101 the above result is based on a formula ( theorem [ varformula ] ) for the variance of an arbitrary smooth function of gaussian random variables .
it will clear from our proofs that the chaos and superconcentration results hold for the @xmath0-spin versions of the s - k model for even @xmath0 .
( see chapter 6 of @xcite for the definition of these models and various results . ) in fact , a generalization of theorem [ chaoscont ] is proven in theorem [ pspin ] later , which includes the @xmath0-spin models for even @xmath0 .
it will also be clear that the lack of superconcentration is true in the random field ising model on general bounded degree graphs .
( again , the lattice case is handled in @xcite .
we refer to @xcite for the definition of the rfim . )
the absence of superconcentration in the rfim implies that the _ site _ overlap is stable under perturbations , instead of the bond overlap as in the e - a model .
a simple model where our techniques give sharp results is the random energy model ( rem ) .
this is discussed in subsection [ rem ] . in spite of the progress made in this paper over @xcite ,
many key issues are still out of reach .
some of them are as follows : 1 .
improve the multiple valley theorem ( theorem [ multisk ] ) so that @xmath102 is a negative power of @xmath51 , preferably better than @xmath103 , which will prove ` strong multiple valleys ' in the sense of @xcite .
another possible improvement to theorem [ multisk ] can be achieved by increasing @xmath104 to something of the form @xmath105 .
3 . prove the chaos theorems ( theorems [ chaosdisc ] and [ chaoscont ] ) for the ground state ( @xmath95 ) of the s - k model .
4 . improve the superconcentration result ( theorem [ superconc ] ) so that the right hand side is @xmath106 for some @xmath107 .
this is tied to the improvement of the chaos result .
if the above is not possible , at least prove a version of the superconcentration result where the right hand side does not depend on @xmath59 , or has a better dependence than @xmath108 .
this will solve the question of chaos for @xmath95 .
prove that the site overlap in the edwards - anderson model is chaotic with respect to fluctuations in the disorder , even though the bond overlap is not .
prove disorder chaos in the s - k model with nonzero external field , that is , if there is an additional term of the form @xmath109 in the hamiltonian .
the general nature of the s - k model indicates that any result for @xmath110 may be substantially harder to prove than for @xmath111 .
( reportedly , a sketch of the proof in this case will appear in the new edition of @xcite . )
show that in the e - a model , the variance of @xmath112 tends to zero and the graph size goes to infinity .
establish temperature chaos in any of these models .
the rest of the paper is organized as follows . in section
[ sketches ] , we sketch the proofs of the main results . in section [ proofs ] , we present some general results that cover a wider class of gaussian fields .
all proofs are given in section [ proofs ] .
in this section we give very short sketches of some of the main ideas of this paper .
suppose we choose @xmath10 from the gibbs measure @xmath47 at inverse temperature @xmath59 and @xmath11 from the measure @xmath60 obtained by applying a continuous perturbation up to time @xmath73 .
let @xmath56 and @xmath61 be the two hamiltonians .
suppose @xmath113 and @xmath114 sufficiently slowly so that chaos holds ( i.e. @xmath115 as @xmath116 ) .
clearly this is possible by theorem [ chaoscont ]
. then due to chaos , @xmath10 and @xmath11 are approximately orthogonal .
since @xmath117 , @xmath10 nearly minimizes @xmath56 and @xmath11 nearly minimizes @xmath61 .
but , since @xmath118 , @xmath119 .
thus , @xmath10 and @xmath11 both nearly minimize @xmath56 .
this procedure finds two states that have nearly minimal energy and are nearly orthogonal . repeating this procedure , we find many such states .
the details are of this argument are worked out in subsection [ multval ] .
let @xmath120 denote @xmath121 when @xmath10 is drawn from the unperturbed gibbs measure at inverse temperature @xmath59 and @xmath11 is drawn from the gibbs measure continuously perturbed up to time @xmath73 .
let @xmath79 be the free energy defined in .
then we show that @xmath122 the proof of this result ( theorem [ supergauss ] ) is simply a combination of the heat equation for the ornstein - uhlenbeck process and integration - by - parts .
the formula directly shows that @xmath123 whenever @xmath120 falls of sharply to zero , which is a way of saying that chaos implies superconcentration . in subsection [ chaosgauss ]
, we show that @xmath124 is a nonnegative and decreasing function .
this proves the converse implication , since the integral of a nonnegative decreasing function can be small only if the function drops off sharply to zero .
suppose @xmath10 is drawn from the gibbs measure of the s - k model at inverse temperature @xmath59 , and @xmath11 from the measure continuously perturbed up to time @xmath73 .
let @xmath9 be the overlap of @xmath10 and @xmath11 , as usual , and let @xmath125 we have to show that for all @xmath73 , @xmath126 where @xmath44 is some constant that depends only on @xmath59 . by repeated applications of differentiation and gaussian integration - by - parts ,
we show that @xmath127 for all @xmath73 and @xmath128 . here
@xmath129 denotes the @xmath128th derivative of @xmath130 .
such functions are called completely monotone .
now , by a classical theorem of bernstein about completely monotone functions , there is a probability measure @xmath131 on @xmath132 such that @xmath133 by hlder s inequality and the above representation , it follows that for @xmath134 , @xmath135 in other words , _ chaos under large perturbations implies chaos under small perturbations_. thus , it suffices to prove that @xmath136 for sufficiently large @xmath137 .
the next step is an ` induction from infinity ' .
it is not difficult to see that when @xmath138 , after integrating out the disorder , @xmath10 and @xmath11 are independent and uniformly distributed on @xmath5 . from this
it follows that @xmath139 .
we use this to obtain a similar bound on @xmath140 for sufficiently large @xmath137 , through the following steps .
first , we show that for any @xmath75 and @xmath137 , @xmath141 thus , we have a _ chain of differential inequalities_. it is possible to manipulate this chain to conclude that @xmath142 the right hand side is bounded by @xmath143 if and only if @xmath137 is sufficiently large .
( this is related to the fact that when @xmath144 is a standard gaussian random variable , @xmath145 if and only if @xmath146 . )
this completes the proof sketch .
the details of the above argument are worked out in subsection [ chaosgauss ] .
the proof of theorem [ nochaos ] , again , is based on the representation of the variance of the free energy and the representation of the function @xmath124 ( both of which hold for the e - a model as well ) . from
, it follows that there is a nonnegative random variable @xmath147 such that for all @xmath90 , @xmath148 from this and it follows that @xmath149 next , we prove a simple analytical fact : suppose @xmath150 is a nonnegative random variable and let @xmath151 . then for any @xmath90 , @xmath152 using this inequality for the random variable @xmath147 and the lower bound on the variance from theorem [ nosuper ] , it is easy to obtain the required lower bound on the function @xmath120 , which establishes the absence of chaos .
the details of this argument are presented in subsection [ nochaosproof ] .
the proof of theorem [ quenchedchaos ] involves a new idea .
let @xmath153 , and let @xmath154 be independent copies of @xmath155 . for each @xmath73
, let @xmath156 for each @xmath157 , let @xmath158 denote a configuration drawn from the gibbs measure defined by the disorder @xmath159 . for @xmath160 , we assume that @xmath158 and @xmath161 are independent given @xmath162
. define @xmath163 by a similar logic as in the derivation of , one can show that @xmath124 is a completely monotone function . also , @xmath124 is bounded by @xmath84 .
thus , for any @xmath97 , latexmath:[\[\label{pbd0 } @xmath73 and let @xmath165 it turns out that @xmath166 and @xmath167 where @xmath168 is the bond overlap between @xmath158 and @xmath169 . combining these two identities with the inequality , it is easy to complete the proof of theorem [ quenchedchaos ] .
the details are in subsection [ quenchedproof ] .
let @xmath170 , and let @xmath171 be an independent copy of @xmath155 . for any @xmath172 , let @xmath173 be the array whose @xmath174th component is @xmath175 let @xmath81 be the free energy , considered as a function of @xmath155 .
suppose @xmath176 and @xmath102 are constants such that for all @xmath177 , @xmath178 fix @xmath179 , and let @xmath4 be a subset of @xmath180 , chosen uniformly at random from the collection of all subsets of size @xmath75 .
let @xmath10 be chosen from the gibbs measure at inverse temperature @xmath59 defined by the disorder @xmath155 , and let @xmath11 be drawn from the gibbs measure defined by @xmath173 .
let @xmath9 denote the overlap of @xmath10 and @xmath11 , as usual .
the key step is to prove that for some absolute constant @xmath44 , @xmath181 this inequality is the content of theorem [ discgauss ] .
the proof is completed by showing that we can choose @xmath102 and @xmath176 such that @xmath182 , and using the superconcentration bound ( theorem [ superconc ] ) on the variance of @xmath81 .
the details of the proof are given in subsection [ chaosdiscproof ] .
although this result was already proven in @xcite for the e - a model on lattices , it may be worth sketching our argument for general bounded degree graphs here .
our proof is based on a general lower bound for arbitrary functions of gaussian random variables .
the result ( theorem [ varlowbd ] ) goes as follows : suppose @xmath183 is an absolutely continuous function such that there is a version of its gradient @xmath184 that is bounded on bounded sets .
let @xmath155 be a standard gaussian random vector in @xmath185 , and suppose @xmath186 and @xmath187 are both finite .
then @xmath188 where @xmath189 denotes the usual inner product on @xmath185 .
we apply this result to the gaussian vector @xmath153 , taking the function @xmath190 to be the free energy @xmath100 .
a few tricks are required to get a lower bound on the right hand side that does not blow up as @xmath117 .
incidentally , the above lower bound on the variance of gaussian functionals is based on a multidimensional plancherel formula that may be of independent interest : @xmath191 versions of this formula have been previously derived in the literature using expansions with respect to the multivariate orthogonal hermite polynomial basis ( see subsection [ plancherel ] for references ) .
we give a different proof avoiding the use of the orthogonal basis .
the results of section [ intro ] are applications of some general theorems about gaussian fields .
these are presented in this section , together with the proofs of the theorems of section [ intro ] . unlike the previous sections , we proceed according to the theorem - proof format in the rest of the paper .
let @xmath192 be a finite set and let @xmath193 be a centered gaussian random vector .
let @xmath194 let @xmath195 be an independent copy of @xmath196 , and for each @xmath90 , let @xmath197 fix @xmath198 .
for each @xmath199 , define a probability measure @xmath200 on @xmath201 that assigns mass @xmath202 to the point @xmath174 , for each @xmath203 .
the average of a function @xmath204 under the measure @xmath200 will be denoted by @xmath205 , that is , @xmath206 we will consider the covariance kernel @xmath207 as a function on @xmath201 , defined as @xmath208 .
alternatively , it will also be considered as a square matrix .
[ main ] assume that @xmath209 for all @xmath177 . for each @xmath210 ,
let @xmath211 let @xmath212 be any convergent power series on @xmath132 all of whose coefficients are nonnegative .
then for each @xmath90 , @xmath213 moreover , @xmath214 is a decreasing function of @xmath73 .
roughly , the way to apply this theorem is the following : prove that the right hand side is small for some large @xmath73 using high temperature methods , and then use the infimum to show that the smallness persists for small @xmath73 as well .
since the application of theorem [ main ] to the s - k model seems to yield a suboptimal result ( theorem [ chaoscont ] ) , one can question whether theorem [ main ] can ever give sharp bounds . in subsection [ rem ]
we settle this issue by showing that theorem [ main ] gives a sharp result for derrida s random energy model .
let us first extend the definition of @xmath217 to negative @xmath73 .
this is done quite simply .
let @xmath218 be another independent copy of @xmath196 that is also independent of @xmath195 , and for each @xmath90 , let @xmath219 let us now recall gaussian integration by parts : if @xmath220 is an absolutely continuous function such that @xmath221 has finite expectation , then for any @xmath222 , @xmath223 where @xmath224 denotes the partial derivative of @xmath225 along the @xmath128th coordinate ( see e.g. @xcite , appendix a.6 ) .
the following lemma is simply a reformulated version of the above identity . for each @xmath90 ,
define @xmath228 a simple computation gives @xmath229 ( note that issues like moving derivatives inside expectations are easily taken care of due to the assumption that @xmath230 . )
one can verify by computing covariances that @xmath231 and the pair @xmath232 are independent .
moreover , @xmath233 so for any @xmath210 , gaussian integration by parts gives @xmath234 the proof is completed by combining the last two steps .
[ complete ] let @xmath235 be the class of all functions @xmath236 on @xmath132 that can be expressed as @xmath237 for some nonnegative integer @xmath238 and nonnegative real numbers @xmath239 , and functions @xmath240 in @xmath215 . for any @xmath241 , there is a probability measure @xmath242 on @xmath132 such that for each @xmath90 , @xmath243 in particular , for any @xmath244 , @xmath245 note that any @xmath241 must necessarily be a nonnegative function , since @xmath246 and @xmath217 are independent and identically distributed conditional on @xmath196 , which gives @xmath247 now , if @xmath248 , then @xmath249 for all @xmath73 , and there is nothing to prove .
so let us assume @xmath250 .
since @xmath207 is a positive semidefinite matrix , there is a square matrix @xmath44 such that @xmath251 .
thus , given a function @xmath225 , if we define @xmath252 then by lemma [ basic ] we have @xmath253 from this observation and the definition of @xmath235 , it follows easily that if @xmath241 , then @xmath254 .
proceeding by induction , we see that for any @xmath75 , @xmath255 is a nonnegative function ( where @xmath256 denotes the @xmath75th derivative of @xmath236 ) .
such functions on @xmath132 are called ` completely monotone ' . the most important property of completely monotone functions ( see e.g. feller @xcite , vol .
ii , section xiii.4 ) is that any such function @xmath236 can be represented as the laplace transform of a positive borel measure @xmath242 on @xmath132 , that is , @xmath257 moreover , @xmath258 . by taking @xmath259 ,
this proves the first assertion of the theorem . for the second ,
note that by hlder s inequality , we have that for any @xmath244 , @xmath260 this completes the proof .
the next lemma is obtained by a variant of the gaussian interpolation methods for analyzing mean field spin glasses at high temperatures .
it is similar to r. lataa s unpublished proof of the replica symmetric solution of the s - k model ( to appear in the new edition of @xcite ) . for each @xmath210
, define a function @xmath263 as @xmath264 note that @xmath265 where @xmath266 if @xmath267 and @xmath268 otherwise .
since @xmath269 is bounded , this proves in particular that @xmath270 .
take any nonnegative integer @xmath271 . since @xmath272 is a positive semidefinite matrix ,
so is @xmath273 .
( to see this , just note that @xmath274 are independent copies of @xmath196 , then @xmath275 . )
therefore there exists a matrix @xmath276 such that @xmath277 .
define the functions @xmath278 in the following we will denote @xmath279 and @xmath280 by @xmath281 and @xmath282 respectively , for all @xmath283 . let @xmath284 by lemma [ basic ] we get @xmath285 our objective is to get a lower bound for @xmath286 . for this
, we can delete the two middle terms in the above expression because they contribute a positive amount .
for the fourth term , note that by hlder s inequality , @xmath287 thus , by lemma [ complete ] and the above inequalities , we have @xmath288 now let @xmath289 for @xmath290
. then @xmath291 the inequality simply becomes @xmath292 fix @xmath293 , @xmath294 . for each @xmath295 ,
let @xmath296 using ( [ diffineq2 ] ) , we see that @xmath297 inductively , this implies that for any @xmath295 , @xmath298 again , for any @xmath299 , @xmath300 where @xmath301 .
thus , @xmath302 . finally , observe that @xmath303 and @xmath304 are independent .
this implies that for any @xmath238 , @xmath305 combining , we conclude that @xmath306 the result now follows easily by taking @xmath307 and summing over @xmath271 , using the fact that @xmath124 has nonnegative coefficients in its power series .
let @xmath308 be as in the proof of lemma [ hightemp ] . as noted in the proof of lemma [ hightemp ] ,
the matrix @xmath309 is positive semidefinite for every nonnegative integer @xmath271 .
since @xmath124 has nonnegative coefficients in its power series , it follows that the matrix @xmath310 is also positive semidefinite .
let @xmath311 be a matrix such that @xmath312
. then @xmath313 therefore , the function @xmath314 belongs to the class @xmath235 of lemma [ complete ] .
the proof is now finished by using lemma [ complete ] and lemma [ hightemp ] , and the observation that @xmath315 ( since @xmath316 has the same law as @xmath317 ) . the claim that @xmath318 is a decreasing function of @xmath73 is automatic because @xmath319 .
we are now ready to give a proof of theorem [ chaoscont ] using theorem [ main ] .
in fact , we shall prove a slightly general result below , which also covers the case of @xmath0-spin models for even @xmath0 , as well as further generalizations .
let @xmath51 be a positive integer and suppose @xmath320 is a centered gaussian random vector with @xmath321 where @xmath322 is some function on @xmath323 $ ] that does not depend on @xmath51 and @xmath324 is the overlap defined in .
let us fix @xmath198 .
the hamiltonian @xmath56 defines a gibbs measure on @xmath5 by putting mass proportional to @xmath325 at each configuration @xmath17 .
this class of models was considered by talagrand @xcite in his proof of the generalized parisi formula . for the s - k model , @xmath326 , while for the @xmath0-spin models , @xmath327 .
( we refer to chapter 6 in @xcite for the definition of the @xmath0-spin models and related discussions . )
let @xmath61 be an independent copy of @xmath56 , and for each @xmath90 , let @xmath328 given a function @xmath236 on @xmath329 , we define the average @xmath330 as the average with respect to the product of the gibbs measures defined by @xmath331 and @xmath332 , that is , @xmath333 the following result establishes the presence of chaos in this class of models under some restrictions on @xmath322 .
it is easy to see that the result covers all @xmath0-spin models for even @xmath0 , and in particular , the original sk model .
[ pspin ] suppose @xmath322 is nonnegative , @xmath334 and there is a constant @xmath335 such that @xmath336 for all @xmath337 $ ] . then there is a constant @xmath44 depending only on @xmath335 such that for all @xmath90 and @xmath80 , and any positive integer @xmath75 , @xmath338 by symmetry , it is easy to see that for each @xmath17 , @xmath339 again , it follows from elementary combinatorial arguments that there are positive constants @xmath340 and @xmath44 that do not depend on @xmath51 , such that for any positive integer @xmath75 and any @xmath51 , @xmath341 choosing @xmath137 so that @xmath342 , and @xmath343 , we see from theorem [ main ] ( and the assumption that @xmath344 ) that for any @xmath345 @xmath346 where @xmath44 is a constant that does not depend on @xmath51 .
this proves the result for @xmath347 .
for @xmath348 we use the last assertion of theorem [ main ] to conclude that @xmath349 is decreasing in @xmath73 . finally , observe that @xmath350 for some constant @xmath44 that depends only on @xmath335 and @xmath340 . in this subsection
we use theorem [ main ] to prove a multiple valley result for general gaussian fields .
let all notation be the same as in subsection [ chaosgauss ] .
the idea of the proof is borrowed from the proof of theorem 3.7 in @xcite , although there are added complications resulting from the fact that we are trying to derive a result about @xmath95 from a result about finite @xmath59 ( i.e. theorem [ main ] ) . [ multigauss ] let @xmath271 be a positive integer , and let @xmath351 .
choose any @xmath352 and @xmath353 .
let @xmath354 , @xmath355 and @xmath356 .
define @xmath357 where the gibbs average in the last term is taken at inverse temperature @xmath59 .
then with probability at least @xmath358 , there exists a set @xmath359 of size @xmath271 such that for any distinct @xmath360 , we have @xmath361 , and for all @xmath362 , @xmath363 . given @xmath196 ,
let @xmath364 be a random variable chosen from the set @xmath192 such that @xmath365 next , let @xmath354 and define a random function @xmath366 then we have @xmath367 thus , @xmath368 an easy verification shows that @xmath369 therefore @xmath370 is an increasing function of @xmath59 and hence @xmath371 combining this with the observation that @xmath372 , we have @xmath373 now let @xmath374 be i.i.d .
copies of @xmath196 .
let @xmath375 and @xmath376 then @xmath377 and @xmath378 are independent ( jointly gaussian and all covariances vanish ) , and @xmath379 let @xmath380 be a random variable on @xmath192 whose conditional distribution given @xmath377 is the same as that of @xmath364 given @xmath196 .
in particular , by the independence of @xmath377 and @xmath378 , @xmath380 and @xmath378 are also independent . from this observation and the above representation of @xmath196 , we have @xmath381 the last equality holds because @xmath382 and @xmath383 have the same unconditional distribution .
for the same reason , we have @xmath384 combining the last two inequalities , we see that @xmath385 thus , @xmath386 now , we clearly have that for any @xmath387 , @xmath388 thus , @xmath389 finally , by gaussian concentration ( see proposition 1.3 in @xcite ) , we have @xmath390 . combining all steps ,
we get @xmath391 and @xmath392 putting together the last two bounds , we see that the set @xmath393 satisfies the requirements of the theorem .
[ multigauss2 ] let all notation be the same as in theorem [ multigauss ] .
fix any @xmath394 .
let @xmath395 then with probability at least @xmath396 , there exists a set @xmath359 of size @xmath271 such that for any distinct @xmath360 , we have @xmath361 , and for all @xmath362 , @xmath397 .
let @xmath195 be an independent copy of @xmath196 , and let @xmath398 note that @xmath399 has the same distribution as @xmath196 .
let @xmath4 be a set as in theorem [ multigauss ] .
let @xmath400 .
then for any @xmath362 , @xmath401 by gaussian concentration ( see e.g. proposition 1.3 in @xcite ) , we have @xmath402 and @xmath403 moreover , by the independence of @xmath195 and @xmath196 and a standard result for gaussian random variables ( see e.g. lemma 2.1 in @xcite ) , we get @xmath404 therefore , @xmath405 from the above steps and theorem [ multigauss ] , we have that with probability at least @xmath396 , there is a set @xmath359 of size @xmath271 such that for any distinct @xmath360 , we have @xmath361 , and for each @xmath362 , @xmath406 . since @xmath399 and @xmath196 have the same distribution , this completes the proof
. these are direct applications of theorem [ multigauss ] and corollary [ multigauss2 ] .
consider the gaussian field @xmath320 defined in , and choose @xmath407 , \
\delta = ( \log n)^{-1/8 } , \\ & t = ( \log n)^{-1/3 } , \
\epsilon = e^{-(\log n)^{1/8}}.\end{aligned}\ ] ] note that @xmath408 and @xmath409 .
note that the quantity @xmath410 , according to the notation of theorem [ chaosgauss ] , is just @xmath411 .
thus , with the above value of @xmath73 , theorem [ chaoscont ] says that for some absolute constants @xmath412 , @xmath413 again , by the sudakov minoration technique ( see e.g. lemma 2.3 in @xcite ) it is not difficult to prove that @xmath414 for some positive absolute constant @xmath335 .
invoking theorem [ multigauss ] , we now get @xmath415 where @xmath44 and @xmath335 denote arbitrary absolute constants .
this completes the proof of theorem [ multisk ] . to prove corollary [ multisk2 ] , note that the quantity @xmath416 in corollary [ multigauss2 ] can be bounded by @xmath417 since @xmath418 as noted before , this completes the proof of corollary [ multisk2 ] .
carrying on with the notation of subsection [ chaosgauss ] , we have the following formula for the variance of the free energy associated with a gaussian field at inverse temperature @xmath59 .
this is a direct analog of lemma 3.1 in @xcite .
we follow the notation of subsection [ chaosgauss ] .
note that by lemma [ basic ] , for any smooth @xmath225 we have @xmath421 taking @xmath422 , we get @xmath423 again , since @xmath424 has the same joint law as @xmath425 , we see that @xmath426 . combining the steps ,
we get @xmath427 this completes the proof .
the goal of this subsection is to prove that in the absence of superconcentration , we do not have chaos either .
this is an improved version of theorem 3.2 in @xcite , where the absence of chaos was proved only up to a finite time , but not for all @xmath73 .
note that @xmath430 now , for any @xmath431 , we have @xmath432 thus , if @xmath433 , then @xmath434 now @xmath435 if and only if @xmath436 .
combining the steps , we see that @xmath437 this completes the proof of the lemma . by theorem [ supergauss ] , @xmath420 by lemma [ complete ]
, we see that there is a non - negative random variable @xmath147 such that for all @xmath73 , @xmath442 combined with the formula for the variance derived above , this gives @xmath443 the result now follows from lemma [ lowlmm ] .
in this subsection we present a general formula for the variance of a function of independent standard gaussian random variables .
after that , we derive a useful lower bound for the variance using this formula .
the variance formula looks similar to those in hour and kagan @xcite and houdr @xcite but it is not the same .
various versions of the formula have appeared in houdr and prez - abreu ( @xcite , remark 2.3 ) and houdr , prez - abreu and surgailis ( @xcite , proposition 10 ) .
essentially , this is the parseval identity for the @xmath444 norm of a gaussian functional expressed as a sum of squares of its fourier coefficients in the orthogonal basis of multidimensional hermite polynomials .
we present a direct proof that does not involve the multivariate hermite polynomial basis . yet
another proof , based on heat kernel expansions , was suggested to the author in a private communication by michel ledoux .
[ varformula ] let @xmath445 be a vector of i.i.d .
standard gaussian random variables , and let @xmath225 be a @xmath446 function of @xmath155 with bounded derivatives of all orders . then @xmath447 the convergence of the infinite series is part of the conclusion . let @xmath171 and @xmath448 be i.i.d .
copies of @xmath155 , and for each @xmath90 , define @xmath449 let @xmath450 then by lemma [ basic ] , we have @xmath451 for @xmath452 , define @xmath453 , where @xmath454 .
then @xmath455 repeating this step @xmath75 times shows that @xmath456 as in the proof of lemma [ complete ] , we observe that the expectations on the right hand side are always nonnegative
. we can continuously extend @xmath457 to the closed interval @xmath458 $ ] by defining @xmath459 .
then @xmath457 is a continuous function on @xmath458 $ ] that is @xmath446 in @xmath460 with all derivatives non - negative .
such functions are known as absolutely monotone ( see feller @xcite , p. 223 ) , and their most important property is that they can be represented as a power series @xmath461 , where the coefficients are non - negative and sum to @xmath462 . from this
one can easily deduce that for any @xmath463 , @xmath464 since @xmath465 and @xmath466 , this completes the proof .
a great advantage of theorem [ varformula ] is that we can extract lower bounds for the variance just by collecting a subset of the terms in the infinite sum .
this is exactly what we do to get the following theorem .
we do not actually need the theorem in its full generality ( with respect to the smoothness conditions on @xmath225 ) , but prove it in the general form nonetheless .
[ varlowbd ] suppose @xmath183 is an absolutely continuous function such that there is a version of its gradient @xmath184 that is bounded on bounded sets .
let @xmath155 be a standard gaussian random vector in @xmath185 , and suppose @xmath186 and @xmath187 are both finite
. then @xmath467 where @xmath189 denotes the usual inner product on @xmath185 .
first assume that @xmath230 .
theorem [ varformula ] gives @xmath468 integration by parts gives @xmath469 thus , for any @xmath470 function @xmath225 , @xmath471 let us now show that the above inequality holds for any bounded lipschitz function @xmath225
. for each @xmath472 and @xmath473 , define @xmath474 then we can write @xmath475 since @xmath225 is a bounded function , it is clear from the above representation that @xmath476 for any @xmath472 , and hence holds for @xmath477 . again , since @xmath225 is lipschitz , @xmath478 where @xmath479 is the lipschitz constant of @xmath225 .
this shows that we can take @xmath480 and obtain for @xmath225 .
next , we want to show whenever @xmath225 is absolutely continuous and square - integrable under the gaussian measure , and the gradient of @xmath225 is bounded on bounded sets .
take any such @xmath225 .
let @xmath481 $ ] be a lipschitz function that equals @xmath84 in the ball of radius @xmath84 centered at the origin , and vanishes outside the ball of radius @xmath482 .
for each @xmath483 , define @xmath484 then note that each @xmath485 is bounded and lipschitz ( with possibly increasing lipschitz constants ) .
thus , holds for each @xmath485 .
since @xmath486 everywhere , and @xmath487 pointwise , and @xmath225 is square - integrable under the gaussian measure , it follows that we can take @xmath488 and get for @xmath225 .
finally , we wish to show that if @xmath184 is square - integrable under the gaussian measure , we have @xmath489 ( note that @xmath225 is almost surely an absolutely continuous function of @xmath490 if we fix @xmath491 .
this follows from fubini s theorem . )
the above identity follows from the univariate identity @xmath492 that holds when @xmath144 is a standard gaussian random variable and @xmath124 is any absolutely continuous function such that @xmath493 , @xmath494 and @xmath495 are all finite .
the identity is just integration - by - parts when @xmath124 is absolutely continuous and vanishes outside a bounded set . in the general case ,
let @xmath496 , where @xmath497 $ ] is a lipschitz function that is @xmath84 on @xmath323 $ ] and vanishes outside @xmath498 $ ] .
then the above identity holds for each @xmath499 , and we can pass to the limit using the dominated convergence theorem .
( actually , it can be shown that the finiteness of @xmath495 suffices . ) as a last step , we observe that by the cauchy - schwarz inequality , @xmath500 this completes the proof . note that @xmath503 and therefore @xmath504 now , it is easy to verify that @xmath505 is a convex function of @xmath59 , and hence for each @xmath506 , @xmath507 is an increasing function of @xmath59 .
thus , @xmath508 is also an increasing function of @xmath59 .
moreover , @xmath509 and therefore @xmath510 for all @xmath511 . finally note that by integration by parts , @xmath512 combined with theorem [ varlowbd ] , this completes the proof .
we are now ready to complete the proof of theorem [ nosuper ] .
consider an undirected graph @xmath513 , and the edwards - anderson spin glass model on @xmath86 as defined in subsection [ ea ] .
let @xmath514 denote the average with respect to the gibbs measure at inverse temperature @xmath59 .
first , we will work with @xmath515 .
let @xmath516 be as in theorem [ nosuper ] .
by lemma [ varlmm ] , with @xmath517 , @xmath153 and @xmath518 , we get @xmath519 now , under the gibbs measure at inverse temperature @xmath520 , the conditional expectation of @xmath521 given the rest of the spins is @xmath522 , where @xmath523 is the neighborhood of @xmath210 in the graph @xmath86 . using this fact and the inequality @xmath524 , we get @xmath525 thus ,
@xmath526 taking @xmath527 , we get @xmath101 finally , to prove the lower bound for @xmath95 , just note that @xmath528 almost surely , and the quantities are all bounded , so we can apply the dominated convergence theorem to get convergence of the variance .
this completes the proof of theorem [ nosuper ] . for @xmath515 ,
this is just a combination of theorem [ nosuper ] and theorem [ supchaos ] .
( note that the notations of the two theorems are related as @xmath529 ; also note that @xmath530 in this case , and therefore @xmath531 . ) next , note that as @xmath117 , the gibbs measure at inverse temperature @xmath59 converges weakly to the uniform distribution on the set of ground states .
the same holds for the perturbed gibbs measure .
thus , @xmath532 where @xmath533 denotes the gibbs average at inverse temperature @xmath59 . since all quantities are bounded by @xmath84 , we can take expectations on both sides and apply dominated convergence .
let @xmath153 , and let @xmath154 be independent copies of @xmath155 .
for each @xmath73 , let @xmath156 for each @xmath157 , let @xmath158 denote a configuration drawn from the gibbs measure defined by the disorder @xmath159 . for @xmath160 , we assume that @xmath158 and @xmath161 are independent given @xmath162 .
define @xmath163 by lemma [ complete ] , it follows that @xmath124 is a completely monotone function on @xmath132 . also , @xmath124 is bounded by @xmath84 .
thus , for any @xmath97 , @xmath534 then by lemma [ basic ] we know that @xmath535 now fix @xmath73 , and let @xmath536 .
then @xmath537 and so by , we have @xmath538 now let @xmath539 and define @xmath540 and @xmath541
. then @xmath542 and @xmath543 are both uniformly bounded by @xmath84 , and so @xmath544 since @xmath545 , an application of the cauchy - schwarz inequality and to the above bound gives @xmath546 to complete the proof , note that @xmath547 where @xmath168 is the bond overlap between @xmath158 and @xmath169 .
our goal in this subsection is to prove that superconcentration implies chaos under discrete perturbations .
accordingly , let us first set the stage for discrete perturbation . henceforth , we deviate from the notation of subsection [ chaosgauss ] .
the result of this subsection and its proof are inspired by lemma 2.3 in @xcite ; we follow the same notation as in @xcite .
let @xmath548 be a vector of independent random variables with @xmath549 for each @xmath210 .
let @xmath195 be an independent copy of @xmath196 .
for any @xmath550 : = \{1,\ldots , n\}$ ] , let @xmath551 be the vector whose @xmath210th component is @xmath552 let @xmath183 be a twice differentiable function .
let @xmath553 and @xmath554 be the first and second partial derivatives of @xmath225 in the direction of the @xmath210th coordinate . [ discgauss ] suppose @xmath555 and @xmath556 are constants such that for all @xmath210 , @xmath557 and @xmath558 everywhere in the closed convex hull of the support of @xmath196 .
fix @xmath559 , and let @xmath4 be a subset of @xmath560 $ ] , chosen uniformly at random from the collection of all subsets of size @xmath75 . define @xmath551 as above .
let @xmath561 .
then @xmath562 the proof of theorem [ discgauss ] is divided into a series of lemmas .
first , let us introduce some further conventions . to simplify notation , we will write @xmath563 for @xmath564 . when @xmath565 , we will simply write @xmath225 . for any @xmath210 and @xmath4 such that @xmath566 , let @xmath567 as usual , when @xmath565 , we will simply write @xmath568 . let @xmath569 denote the collection of all subsets of @xmath560\backslash \{i\}$ ] of size @xmath75 . for @xmath570 , efine @xmath571
the above quantity is a discrete proxy for the left hand side in .
our first result is an exact formula for the variance in terms of @xmath572 .
this is actually a restatement of lemma 2.3 from @xcite . by exchangeability of @xmath574 and @xmath575 , it is easy to see that the pair @xmath576 has the same distribution as the pair @xmath577 , and therefore @xmath578 we claim that @xmath579}.\ ] ] to see this , consider any @xmath580 $ ] such that @xmath581 and @xmath582 $ ]
. let @xmath583 . on the left hand side in the above display ,
if we write out the definition of @xmath584 as @xmath585 and regroup terms , then the coefficient of @xmath586 in the expansion is @xmath587 similarly , the coefficient of @xmath225 is @xmath84 and the coefficient of @xmath588}$ ] is @xmath589 .
this proves . combining with
, we see that @xmath590 } ) ) = \frac{1}{2n}\sum_{i=1}^n \sum_{k=0}^{n-1}\frac{1}{{n-1 \choose k } } \sum_{a\in \ma_{k , i } } \ee(\delta_i f\delta_i f^a).\ ] ] this completes the proof of the lemma .
take any @xmath4 and @xmath566 .
it is easy to see that given @xmath593 and @xmath575 , the random variables @xmath568 and @xmath584 are i.i.d .
therefore , @xmath594 from this and jensen s inequality , it is clear that @xmath595 , and for any @xmath596\backslash \{i\}$ ] , @xmath597 thus , if @xmath598 , we have @xmath599 where the sum is taken over all @xmath600 such that @xmath601 for some @xmath602 . since any @xmath603 can be obtained by adding one element to @xmath4 for exactly @xmath604 many @xmath605 , we have @xmath606 this can be rewritten as @xmath607 this completes the proof of the lemma .
since @xmath557 and @xmath558 everywhere on the closed convex hull of the support of @xmath196 , by taylor expansion we have @xmath612 thus , @xmath613 now let @xmath614 be another independent copy of @xmath574 , that is also independent of @xmath575 .
let @xmath615 denote @xmath553 with @xmath574 replaced by @xmath614 and define @xmath616 similarly .
since @xmath549 and @xmath617 is independent of @xmath618 , we have @xmath619 again , @xmath620 and now , combining the above bound with , we have @xmath621 we also have to consider the case when @xmath362 .
let @xmath622 .
then by jensen s inequality we have @xmath623 now take @xmath624 and let @xmath625 denote the set of all subsets of @xmath560 $ ] of size @xmath75 . using and
, we get @xmath626 from this and lemma [ disc3 ] , we conclude that for @xmath624 , @xmath627 the same conclusion can be drawn for @xmath628 and @xmath629 by defining @xmath630 and verifying that all steps hold .
this completes the proof .
consider the s - k hamiltonian @xmath56 defined in as a function of the disorder @xmath170 .
fix @xmath59 , and define @xmath631 , where @xmath79 is the free energy defined in .
let @xmath171 be an independent copy of @xmath155 , and define @xmath173 as we defined @xmath551 in theorem [ discgauss ] .
let @xmath632 ( and assume that @xmath75 is an integer ) , and define a perturbed hamiltonian using the disorder @xmath173 , where @xmath4 is chosen uniformly at random from the set of all subsets of @xmath633 of size @xmath75 .
let @xmath10 be sampled from the original gibbs measure , and @xmath11 from the perturbed gibbs measure .
an easy verification shows that @xmath634 where @xmath635 is the derivative of @xmath225 with respect to the @xmath174th coordinate . on the other hand , by theorem [ superconc ]
we know that @xmath636 finally , note that for any @xmath174 , @xmath637 therefore , we can take @xmath638 and @xmath639 while applying theorem [ discgauss ] . using all the above information , we can now apply theorem [ discgauss ] to conclude that @xmath640 where @xmath44 is an absolute constant .
since @xmath641 , we can ignore the second term on the right after replacing @xmath642 by @xmath643 in the first term .
this completes the proof .
the random energy model ( rem ) , introduced by derrida @xcite , is possibly the simplest model of a spin glass .
the state space is @xmath5 as usual , but here the energies of states @xmath644 are chosen to be i.i.d .
gaussian random variables with mean zero and variance @xmath51 .
we show that theorem [ main ] gives a sharp result in the low temperature regime ( @xmath645 ) of this model .
we follow the notation of theorem [ main ] .
suppose @xmath10 is drawn from the original gibbs measure of the rem and @xmath11 from the gibbs measure perturbed continuously up to time @xmath73 , in the sense of subsection [ chaossec ] .
if @xmath645 , there are positive constants @xmath646 and @xmath647 depending only on @xmath59 such that for all @xmath51 and @xmath73 , @xmath648 in the notation of theorem [ main ] , we have @xmath649 if @xmath650 , and @xmath651 if @xmath652 . also , clearly , @xmath653 for each @xmath17 .
suppose @xmath10 is drawn from the original gibbs measure and @xmath11 from the gibbs measure perturbed continuously up to time @xmath73 .
taking @xmath654 in theorem [ main ] , we get @xmath655 now choose @xmath137 so large that @xmath656 .
the above inequality shows that for @xmath347 , @xmath657 and for @xmath658 , @xmath659 a simple computation via theorem [ supergauss ] now gives @xmath660 where @xmath646 is a constant depending only on @xmath59 .
now suppose @xmath645 .
let @xmath661 , where @xmath662 solves @xmath663 let @xmath664 denote the numbers @xmath665 when enumerated in non - increasing order .
it follows from arguments in section 1.2 of talagrand @xcite that this point process converges in distribution , as @xmath63 , to a poisson point process @xmath666 with intensity @xmath667 on @xmath132 , where @xmath668 .
it is not difficult to extend this argument to show that @xmath669 we skip the details , which are somewhat tedious .
( here @xmath645 is required to ensure that the infinite sum @xmath670 converges almost surely . )
however , @xmath671 .
thus , there is a positive constant @xmath647 depending only on @xmath59 such that for any @xmath51 , @xmath672 we can now use theorem [ supchaos ] to prove that for some positive constant @xmath647 depending only on @xmath59 , we have that for any @xmath51 and @xmath73 , @xmath673 however , we also have by theorem [ main ] that @xmath674 is a decreasing function of @xmath73 , and hence @xmath675 combined with , and , this completes the proof .
.2 in * acknowledgments . *
the author thanks michel talagrand , persi diaconis , daniel fisher , victor prez - abreu , christian houdr , michel ledoux , rongfeng sun , tonci antunovic and partha dey for helpful discussions and comments , and itai benjamini for asking the question that led to theorem [ chaosdisc ] . | we prove that the sherrington - kirkpatrick model of spin glasses is chaotic under small perturbations of the couplings at any temperature in the absence of an external field .
the result is proved for two kinds of perturbations : ( a ) distorting the couplings via ornstein - uhlenbeck flows , and ( b ) replacing a small fraction of the couplings by independent copies .
we further prove that the s - k model exhibits multiple valleys in its energy landscape , in the weak sense that there are many states with near - minimal energy that are mutually nearly orthogonal .
we show that the variance of the free energy of the s - k model is unusually small at any temperature .
( by ` unusually small ' we mean that it is much smaller than the number of sites ; in other words , it beats the classical gaussian concentration inequality , a phenomenon that we call ` superconcentration ' . )
we prove that the bond overlap in the edwards - anderson model of spin glasses is _ not _ chaotic under perturbations of the couplings , even large perturbations .
lastly , we obtain sharp lower bounds on the variance of the free energy in the e - a model on any bounded degree graph , generalizing a result of wehr and aizenman and establishing the absence of superconcentration in this class of models .
our techniques apply for the @xmath0-spin models and the random field ising model as well , although we do not work out the details in these cases . |
the gravitational theory with higher derivative terms @xcite is one of the most interesting subjects to study in physics for many years .
the treatment of einstein gravity with perturbative field quantization naturally leads to inclusion of such nonlinear terms in the spacetime curvatures in effective actions .
the generalization of einstein gravity has been developed extensively by many authors .
the lanczos lovelock gravity @xcite , which is dictated by the action with the dimensionally continued euler forms , is defined in a generic dimensional spacetime and it describes a massless spin - two fluctuation despite of the existence of higher order terms in its action . in a few decades ago , it is discovered that the effective field theory arising from string theory involves similar terms at least the lowest - order correction to the einstein - hilbert term @xcite .
the other specific extension is found in the study on so - called critical gravity @xcite ( although the word `` critical '' originally implied the emergence of log gravity @xcite ) .
in such a theory of gravity , the behavior of spin - two modes is described by the lagrangian of special combination of higher order terms in the ricci tensor and the scalar curvature .
there is a ghost field but no spin - zero mode in the theory . in the three dimensional spacetime ,
the model governed by a similar lagrangian expresses the new massive gravity @xcite . in this paper
, we consider incorporation of maxwell fields to the ( next - to ) critical gravity .
the critical gravity has been studied under the assumption of the maximally symmetric background spacetime .
the higher derivative theory on the other type of background fields is worth studying .
inclusion of a non - vanishing flux field brings about the partial compactification with a maximally symmetric extra space and might provide an interesting scenario for a very early stage of the universe .
the assumption of the partially symmetric solution can yield another possibility .
it is feasible to attach the other type of higher order terms to the lagrangian which realizes the spontaneous compactification .
because of the symmetry , use of the riemann tensor is possible in the additional terms .
they are expressed by special combinations of gauss - bonnet terms and horndeski s generalized maxwell term @xcite .
this can induce a second order field equation displaying the cosmological evolution of the background geometry .
this paper is organized as follows . in sec .
[ sec2 ] , we build the lagrangian of our first model , which involves the quadratic terms of the ricci tensor , the ricci scalar and the two - form flux field , in a similar way revealed in the study of critical gravity . the spontaneous compactification in the first model is studied in sec .
[ sec3 ] and cosmological solutions are examined in sec .
[ sec4 ] . in sec .
[ sec5 ] , we present the second model containing the quadratic term of the riemann tensor .
we also investigate the cosmological solution qualitatively in this model .
the last section is devoted to summary and prospects . throughout the present paper
, we deal with our models at the classical level .
in the work on critical gravity , it is found that the lagrangian of the specific combination of curvatures yields a spin - two massless graviton and a ghost mode ; there appears no scalar mode as in einstein gravity . in this section
, we propose a model of higher order extension of maxwell einstein theory .
we will follow the method for constructing the lagrangian examined in our previous work @xcite . first , we start with the action for the einstein gravity and electromagnetism in @xmath0 dimensions : @xmath1 \,,\ ] ] where @xmath2 and @xmath3 denote the scalar curvature obtained from the metric @xmath4 and the cosmological constant , respectively .
the field strength @xmath5 is defined by @xmath6 with a vector field @xmath7 and @xmath8 .
the suffices @xmath9 vary over @xmath10 . applying the variational principle to the action results in the following equations of motion : @xmath11 with @xmath12 where @xmath13 , and @xmath14 next
, we construct a lagrangian of our model as follows : @xmath15 where @xmath16 note that the solution of @xmath17 is the solution for the equation of motion derived from @xmath18 , provided that the field equation for the vector field is satisfied ( though the stability is not necessarily guaranteed ) . to our lagrangian , but we restrict ourselves to the case without such a term in the present paper . ]
the lagrangian @xmath18 can then be written as @xmath19 \nonumber \\ & -&\beta\left[r_{mn}r^{mn}-\frac{d}{4(d-1)}r^2 - 4f^2_{mn}r^{mn } + \frac{d+2}{2(d-1)}f^2r\right.\nonumber \\ & & \quad+4f^2_{mn}f^2{}^{mn}-\frac{d+8}{4(d-1)}(f^2)^2\nonumber \\ & & \quad\left.+\frac{\lambda}{2(d-1)}\left\{(d-2)r-(d-4)f^2-\frac{d}{2 } \lambda\right\}\right ] % \nonumber \\&+&\gamma ( \nabla_mf^{mn})^2 \ , .
\label{ours}\end{aligned}\ ] ] incidentally , we can pick up some ` critical ' cases : if @xmath20 , the cosmological constant is zero in the @xmath0 dimensional spacetime ; if @xmath21 , the einstein hilbert term @xmath22 is absent from the @xmath0 dimensional action ; if @xmath23 , the pure maxwell term @xmath24 is absent from the @xmath0 dimensional action .
we should note that the auxiliary field method @xcite , which can be used in critical higher order gravity , also leads to the additional terms , such as @xmath25 where @xmath26 is an auxiliary symmetric tensor field and @xmath27 .
varying the auxiliary field @xmath26 , we obtain @xmath28 or , solving that with respect to @xmath29 , we get @xmath30 therefore , one can see the equivalence between the lagrangians @xmath18 and @xmath31 .
we can derive the linearized field equation using the action @xmath32 .
now , we investigate the behavior of linear fluctuations around the background geometry , which satisfies @xmath33 .
we decompose the metric as follows : @xmath34 the indices are raised and lowered by the background metric @xmath35 .
then the trace of the fluctuation is expressed as @xmath36 for conveniences , we write down the curvature tensors up to the linear order in @xmath37 here @xcite : @xmath38 where the nabla stands for the covariant derivative and the barred symbols show that they are constructed from the background metric @xmath35 . assuming the background fields satisfies @xmath39 , we can use @xmath40 to obtain the field equation at the linearized order : @xmath41=0\ , , \label{dg}\end{aligned}\ ] ] noting that @xmath26 is in the linear order ( @xmath42 ) , and we have used the condition @xmath43 for the background field .
the generalized maxwell equation in the linear order in fluctuations takes the form @xmath44=0\,,\ ] ] where @xmath45 .
the linearized equation obtained from ( [ ae ] ) is @xmath46\nonumber \\ & & + 2\bar{f}^m{}_p\bar{f}^n{}_qh^{pq}+\frac{1}{2}\bar{f}^2h^{mn } -\bar{f}^2_{pq}h^{pq}\bar{g}^{mn}\nonumber \\ & & -2\bar{f}^{pm}f_{p}{}^n-2\bar{f}^{pn}f_{p}{}^m + \bar{f}^{pq}f_{pq}\bar{g}^{mn}=s^{mn}-\bar{g}^{mn}s\,.\end{aligned}\ ] ] the important observation is that , using ( [ ae ] ) , the field equation ( [ dg ] ) becomes @xmath47 in the @xmath0 dimensional minkowski vacuum , this equation is just the fierz - pauli equation @xcite for the spin - two wave with mass - squared @xmath48 .
therefore we recognize that @xmath26 corresponds to the massive ghost field .
the postulate of no tachyonic ghost requires @xmath49 in the present model .
the linear fluctuation from the background fields can be argued by using above equations . in the next section ,
we consider the solution for spontaneous compactification of spacetime as the background .
in this section , we study the partial compactification of space with a non - trivial flux in our model .
the extra dimensions are considered to be compactified into a sufficiently small size .
a maximally symmetric solution in our model is compactification of the form @xmath50 , where @xmath51 is the @xmath52 dimensional minkowski spacetime and @xmath53 is a two dimensional sphere , as in the work of randjbar - daemi , salam and strathdee ( rss ) @xcite where @xmath54 is considered .
we consider the solution for flat @xmath55 dimensional spacetime and study its classical stability in this section , though other maximally symmetric spacetimes are also interesting .
our notation is that the suffices @xmath56 run over @xmath57 and the suffices @xmath58 are used for the extra dimensions @xmath55 and @xmath59 .
as stated in the previous section , the solution of @xmath17 and @xmath60 is a solution in our model . a suitable ansatz for the flux field is @xmath61 where @xmath62 is the antisymmetric symbol and the constant @xmath63 is the strength of the two - form flux .
this monopole flux yields the relation @xmath64 then , the equation @xmath17 can be decomposed to @xmath65 for the spacetime of a direct product . as for the flat @xmath55 dimensional spacetime ,
where @xmath66 , we must take a fine - tuning of parameters as @xmath67 and therefore , the solution for the background fields reads @xmath68 substituting this compactified background solution into eq .
( [ fp ] ) , we obtain @xmath69 to show complete particle spectrum for the ghost field is rather complicated , but we can see that the lowest mass - squared of the ghost is @xmath70 for symmetric tensor fields ( which do not couples to the ricci curvature of the extra space ) and vector bosons of the kaluza - klein origin ( which come from the zero modes of the lichnerowicz operator on @xmath53 , @xmath71 ) .
it is also remarkable that the other graviton and vector field fluctuations is absent in the linearized equation ( [ ss ] ) .
thus , taking the analysis by rss for other fluctuation modes @xcite into account , in the classical and linearized analysis , we find that the rss background solution @xmath50 is stable if @xmath72 , because there is no growing mode . before closing this section
, we note that the effective @xmath55 dimensional newton constant , which can be found by the coefficient of the ricci scalar of the @xmath55 dimensional spacetime in the action , is not affected by the higher order terms in our model and is independent of @xmath73 .
it might be related to the absence of the graviton and the ghost modes at the linearized level mentioned above .
in this section , we perform the analysis on the stability of the compactification and the cosmological solution in our model .
we consider an effective potential @xmath74 for the radius of extra space @xmath53 @xcite in order to study them .
we can express the riemann tensor of the @xmath53 as @xmath75 where @xmath76 denotes the radius of the compactified sphere @xmath53 .
then , the equations ( [ eom1],[eom2 ] ) with @xmath66 impose @xcite @xmath77 where the first equation is the fine - tuning condition between the parameters . the effective potential @xmath74 as a function of @xmath76 , the scale of the extra space , is obtained from replacing the background fields by functions of @xmath76 in @xmath78 with @xmath79 and @xmath80 .
the effective potential for the rss model is calculated from @xmath81 and is given as @xmath82 and then , the effective potential for our model of which lagrangian is described by ( [ ours ] ) is @xmath83\nonumber \\ & & + \frac{\beta\lambda}{4(d-1 ) } \left[4(d-2)-d\,y+\frac{2d}{y}-\frac{12(d-2)}{y^2}+\frac{7d-16}{y^3}\right ] \nonumber \\ & = & \frac{(y-1)^2}{4(d-1)y^3}[(7d-16)\beta+2(d-4)\beta y+(4(d-1)\alpha - d\beta)y^2]\,,\end{aligned}\ ] ] with @xmath84 .
a minimum of the potential is found at @xmath85 and attains @xmath86 .
the stability of the radius of the extra sphere is determined by the sign of the second derivative of the potential : @xmath87 if @xmath88 and @xmath89 , the scale @xmath90 is found to be stable .
we also find that it is impossible to create another potential minimum satisfying @xmath91 by any tuning of @xmath92 in the model . at a large value of @xmath76
, @xmath74 can be indefinitely negative provided that @xmath93 . in this case , the @xmath0 dimensional cosmological constant becomes negative .
this may not indicate instability of compactification against a large fluctuation , because the higher derivative terms cause the coupling between the compactification scale and the first and second derivatives of an expanding scale factor of the rest of space .
therefore , we investigate the cosmological equation including the dynamical scale factor and the compactification scale .
the metric is usually assumed as @xmath94 where @xmath95 denotes the radius of the compactified sphere @xmath53 and @xmath96 is the line element an the unit sphere .
for the maximally symmetric @xmath97 dimensional space , the line element is denoted @xmath98 here , and its riemann tensor is normalized as @xmath99 with @xmath100 , where @xmath101 .
@xmath102 is the scale factor of @xmath103 dimensional homogeneous and isotropic space .
the riemann curvature is computed with the metric as @xmath104 where the dot @xmath105 denotes the derivative with respect to the cosmic time @xmath106 .
now we can obtain the effective action of the scale factor @xmath102 and the compactification scale @xmath95 for the rss model as @xmath107\nonumber \\ & \sim&\int dt\ , a^{d-3}b^2\left [ -4(d-3)\frac{\dot{a}}{a}\frac{\dot{b}}{b } + ( d-3)(d-4)\frac{-\dot{a}^2+k}{a^2}\right.\nonumber \\ & & \left.\qquad\qquad\qquad+2\frac{-\dot{b}^2 + 1}{b^2 } -\frac{1}{\lambda b^4}-\lambda \right]\equiv \tilde{i}_0\,,\end{aligned}\ ] ] where the over - all constant which comes from the volume of space has been omitted . the action has been integrated by parts after the symbol ` @xmath108 ' .
the effective action for our model can be written , by adopting the auxiliary fields , as @xmath109\right.\nonumber \\ & & + 2(d-3)s_a\left[-(d-4)\frac{\ddot{a}}{a}-2\frac{\ddot{b}}{b}-2(d-4 ) \frac{\dot{a}}{a}\frac{\dot{b}}{b}-\frac{(d-4)(d-5)}{2 } \frac{\dot{a}^2+k}{a^2}-\frac{\dot{b}^2 + 1}{b^2}+\frac{1}{2\lambda b^4}+\frac{1}{2}\lambda\right]\nonumber \\ & & + 4s_b\left[-(d-3)\frac{\ddot{a}}{a}-\frac{\ddot{b}}{b}-(d-3 ) \frac{\dot{a}}{a}\frac{\dot{b}}{b}-\frac{(d-3)(d-4)}{2 } \frac{\dot{a}^2+k}{a^2}-\frac{1}{2\lambda b^4}+\frac{1}{2}\lambda\right]\nonumber \\ & & \left.-\left[s_0 ^ 2+(d-3)s_a^2 + 2s_b^2\right]+\left [ s_0+(d-3)s_a+2s_b\right]^2\right\}\,,\end{aligned}\ ] ] where we set @xmath110 , @xmath111 and @xmath112 . here
, the equations of motion are omitted , but they can be obtained from the above action .
we seek for the de sitter ( @xmath113 ) solution , where @xmath114 , @xmath115 with constant @xmath116 , and @xmath117constant , are taken as an ansatz . in this case , all the equations of motion become algebraic equations . with the ansatz ,
the auxiliary variables are expressed as @xmath118\lambda b_0 ^ 4}{2(d-1)\lambda b_0 ^ 4}\,,\nonumber \\ s_b&=&\frac{5 - 2d+2(d-2)\lambda b_0 ^ 2-[(d-2)(d-3)h_0 ^ 2+\lambda]\lambda b_0 ^ 4}{2(d-1)\lambda b_0 ^ 4}\,.\end{aligned}\ ] ] substituting these auxiliary relations to the other equations , we obtain the following two equations at last : @xmath119\beta\right\ } \lambda b_0 ^ 4\nonumber \\ & & + 4\left\{2(d-1)\alpha+[d(d-3)(d-4)h_0 ^ 2-(d-2)\lambda ] \beta\right\}\lambda^2 b_0 ^ 6\nonumber \\ & & + \left\{(d-3)^2(d-6)(d^2 - 6d+4)h_0 ^ 4\beta+2(d-3)(d-4)h_0 ^ 2 [ 2(d-1)\alpha-(d-2)\beta\lambda]\right.\nonumber \\ & & \quad\left.+[-4(d-1)\alpha+d\beta\lambda ] \lambda\right\}\lambda^2 b_0
^ 8=0\ , , \label{se1}\end{aligned}\ ] ] and @xmath120\beta\right\ } \lambda b_0 ^ 4\nonumber \\ & & + \left\{(d-3)^2(d-2)(d^2 - 6d+4)h_0 ^ 4\beta+2(d-2)(d-3)h_0 ^ 2 [ 2(d-1)\alpha-(d-2)\beta\lambda]\right.\nonumber \\ & & \quad\left.+[-4(d-1)\alpha+d\beta\lambda ] \lambda\right\}\lambda^2 b_0 ^ 8=0\ , .
\label{se2}\end{aligned}\ ] ] the two sets of the solution for @xmath121 and @xmath116 can be found ; one is @xmath122 and another is @xmath123 these values for solutions are the same as those of the rss model @xcite , as expected ( if @xmath54 ) . in both cases , the values for @xmath124 , @xmath125 and @xmath126 are equal to zero . because the simultaneous equations ( [ se1],[se2 ] ) are the fourth order algebraic equations , there are other two sets of solutions , @xmath125 and @xmath126 are not zero for such solutions . ] at most .
one can find , for sufficiently small value for @xmath73 , only two original solutions ( [ s1],[s2 ] ) are real solutions .
for @xmath54 , only two real solutions ( [ s1],[s2 ] ) exist for @xmath127 . as a model of the universe , the multiple de sitter phases would have some significance in cosmology .
we now study the behavior of the small fluctuation around the above exact solutions . in the neighborhood of ( i ) , we set @xmath128 and @xmath129 , and so on . then , the linearized equations of motion read @xmath130 \delta s_a=0\ , , \label{res1}\\ & & 2(d-1)\alpha\left[(d-2)\delta\ddot{b}+2(d-4)\lambda\delta b\right ] \nonumber \\ & & \quad+\frac{d-3}{\sqrt{\lambda}}\left[(d-1)(d-2)\alpha+4\beta\lambda \right]\delta s_a=0\,,\label{res2 } \\ & & ( d-1)\alpha\left[(d-2)\delta\dot{h}-4\lambda^{3/2}\delta b\right]\nonumber \\ & & \quad-\left[(d-1)(d-2)\alpha-2(d-3)\beta\lambda\right]\delta s_a=0\ , , \\ & & ( d-1 ) \delta s_0= -2(d-3 ) \frac{\beta\lambda}{\alpha } \delta s_a\,.\end{aligned}\ ] ] note that for @xmath131 , the equations give @xmath132 , @xmath133 and @xmath134 , which are equivalent to the equations for the cosmology of the rss model @xcite .
the oscillatory behavior of @xmath135 as well as @xmath136 is enough interesting .
the variables @xmath137 , @xmath136 and @xmath138 oscillate with a different frequency from that of @xmath135 , as seen from ( [ res1],[res2 ] ) .
is the square root of @xmath139 in ( [ v2 ] ) . ]
thus , the tuning of initial conditions realizes enough time for accelerating expansion and a sufficient inflation can be achieved , though we can not tell the choice of initial state in the present classical analysis of the model . since the analysis of the perturbation from ( ii )
can be considered and results in finding exponentially growing modes as found in @xcite , we omit the indication of equations here .
some fine tuning gives an enough expansion time for a sufficient inflation as the de sitter phase also in this case .
in the solution for the spontaneous compactification in previous sections , one has found the following relation in the background fields : @xmath140 similarly to the consideration in the previous sections , we come to an idea that the minkowski compactification of the rss model is attained in a higher derivative theory if the relation ( [ rff ] ) is reflected in the higher order correction term .
we then propose an additional higher order term @xmath141 where the generalized kronecker s delta is defined by @xmath142 now , we write the model 2 by the lagrangian ( therefore , we call the previous model the model 1 ) @xmath143 note that @xmath144 consists of the lanczos lovelock lagrangian ( gauss bonnet term ) @xcite and the horndeski s vector - tensor term @xcite and a nonlinear maxwell term .
it is easy to see that adding @xmath144 to the rss lagrangian @xmath145 yields no modification of linearized equation of motion on the rss background fields after gauge fixing .
therefore , the kaluza - klein modes remain unchanged from those in the rss model .
it is worth noting that the effective @xmath55 dimensional newton constant in model 2 is not affected by the value of @xmath146 .
now , we study the cosmological evolution of the classical background in the model .
although the kaluza - klein cosmologies with the gauss bonnet term or the generalized euler invariant have been studied in many works including @xcite , the case with the rss - type compactification with the gauss bonnet term has been studied in a few papers including @xcite . as previously
, we characterize the cosmological background as : @xmath94 and @xmath147 then , the lagrangian @xmath144 reduces to @xmath148\nonumber \\ & & \qquad+8\frac{\ddot{b}}{b}\left[2(d-3)\frac{\dot{a}}{a } \frac{\dot{b}}{b}+ ( d-3)(d-4)\frac{\dot{a}^2+k}{a^2}\right]\nonumber \\ & & \qquad+(d-3)(d-4)(d-5)(d-6)\left(\frac{\dot{a}^2+k}{a^2}\right)^2 \nonumber \\ & & \qquad+8(d-3)(d-4)(d-5)\frac{\dot{a}}{a}\frac{\dot{b}}{b}\frac{\dot{a}^2+k}{a^2 } + 8(d-3)(d-4)\frac{\dot{a}^2}{a^2}\frac{\dot{b}^2}{b^2 } \nonumber \\ & & \qquad + 4(d-3)(d-4)\frac{\dot{a}^2+k}{a^2}\left(\frac{\dot{b}^2 + 1}{b^2 } -\frac{1}{\lambda b^4}\right)\,,\end{aligned}\ ] ] and we obtain the following action for @xmath149 and @xmath76 : @xmath150\right.\nonumber \\ & & \qquad\qquad\qquad+8(d-3)(d-4)(d-5)\left[-\frac{1}{3}\frac{\dot{a}^3}{a^3 } \frac{\dot{b}}{b}-\frac{k}{a^2}\frac{\dot{a}}{a } \frac{\dot{b}}{b}\right ] \nonumber \\ & & \qquad\qquad\qquad + 4(d-3)(d-4)\left[-\frac{\dot{a}^2}{a^2}\frac{\dot{b}^2 + 1}{b^2 } + \frac{k}{a^2}\frac{-\dot{b}^2 + 1}{b^2}\right]\nonumber \\ & & \qquad\qquad\qquad -4(d-3)(d-4)\frac{-\dot{a}^2+k}{a^2}\frac{1}{\lambda b^4}-16(d-3)\left.\frac{\dot{a}}{a}\frac{\dot{b}}{b}\frac{1}{\lambda b^4}\right\}\,.\end{aligned}\ ] ] again , we omit exposition of equations of motion here . we first explore the de sitter phase , or the maximally symmetric product spacetime . to find this , we set @xmath114 , @xmath115 and @xmath151 , where @xmath116 and @xmath121 are constants
we then obtain the simultaneous algebraic equations as follows : @xmath152 + 2\left [ \alpha+2(d-3)(d-4 ) h_0 ^ 2\gamma\right]\lambda b_0 ^ 2\nonumber \\ & & \quad+ \left[\left((d-3)(d-4)h_0 ^ 2-\lambda\right)\alpha+(d-3)(d-4)(d-5)(d-6 ) h_0 ^ 4\gamma\right]\lambda b_0 ^ 4=0\,,\label{t1 } \\ & & \alpha+4(d-2)(d-3)h_0 ^ 2\gamma \nonumber \\ & & \quad+ \left[\left((d-2)(d-3)h_0 ^ 2-\lambda\right)\alpha+(d-2)(d-3)(d-4)(d-5 ) h_0 ^ 4\gamma\right]\lambda b_0 ^ 4=0\label{t2}\,,\end{aligned}\ ] ] of course , the solution ( i ) ( [ s1 ] ) is a trivial solution of above equations , while ( ii ) ( [ s2 ] ) is not a solution for @xmath153 .
one solution of the simultaneous equations ( [ t1],[t2 ] ) with @xmath154 exists for all positive @xmath155 .
the value of @xmath156 increases with the value of @xmath155 .
the existence of the de sitter phase with large values for @xmath116 and @xmath121 is suitable for the possible inflationary phase .
next , we draw the phase diagram of the compactification scale @xmath76 .
we consider the case with @xmath114 and @xmath54 .
1 shows the phase space spanned by @xmath76 and @xmath157 for @xmath158 .
the small arrows in the figures indicate the flow in the phase space .
one can see that @xmath90 is an attractor and the area of attracting region increases with the value of @xmath155 .
we conclude that an appropriate choice of the initial condition , which is in the flow passing through near by the branch point given by the de sitter solution of ( [ t1],[t2 ] ) , might cause a long expansion time . .
we show ( a ) for @xmath159 , ( b ) for @xmath160 , ( c ) for @xmath161 , respectively.,title="fig:",height=188 ] .
we show ( a ) for @xmath159 , ( b ) for @xmath160 , ( c ) for @xmath161 , respectively.,title="fig:",height=188 ] .
we show ( a ) for @xmath159 , ( b ) for @xmath160 , ( c ) for @xmath161 , respectively.,title="fig:",height=188 ] + ( a)(b)(c )
in this paper , we have considered special cases for extension of einstein maxwell theory with higher order terms .
the models proposed in the present paper have the same solution for spontaneous compactification @xmath50 utilizing the internal flux as the rss model .
the stability of the compactified vacuum have been confirmed at the classical level .
nevertheless , the models reveal the different cosmological behaviors because of the additional degrees of freedom or the existence of the additional de sitter phases .
we should investigate evolutional equations by using numerical calculations in future work .
our toy models seem to be quite special , but the analysis of the models will exhibit some typical behaviors including inflation , and we can regard general higher order corrections as a modification from the critical models .
the subjects we should consider are , for instance : the coupling to matter fields ; compactifications on @xmath162 @xcite in our models ; the inclusion of a @xmath163-form flux field such as the freund rubin compactification @xcite ; the treatment of our models in quantum cosmology @xcite ; the initial singularity problems @xcite in the models ; the possible supersymmetrization of the models ; the field content of the model in @xmath164 dimensions . the generalization of our model 1 to more higher order theory is considered straightforwardly .
if we introduce an auxiliary field @xmath165 , the additional term can be replaced by @xmath166 which will be studied extensively elsewhere . finally , we must research the relation to other theories .
we suppose that the structure of the model 1 owes to the matrix @xmath167 , which appears also in the spin - two propagator
. it would be very interesting if our models were the effective theory of another theory .
99 k. s. stelle , phys . rev .
* d16 * ( 1977 ) 953 .
k. s. stelle , gen .
* d16 * ( 1978 ) 353 . c. lanczos , z. phys .
* 73 * ( 1932 ) 147 . c. lanczos
, annals math .
* 39 * ( 1938 ) 842 .
d. lovelock , j. math . phys .
* 12 * ( 1971 ) 498 .
d. lovelock , j. math . phys . * 13 * ( 1972 ) 874 .
t. padmanabhan and d. kothawala , phys . rep .
* 531 * ( 2013 ) 115 .
b. zumino , phys . rep .
* 137 * ( 1985 ) 109 .
b. zwiebach , phys . lett .
* b156 * ( 1985 ) 315 .
h. l and c. n. pope , phys .
* 106 * ( 2011 ) 181302 .
s. deser , h. liu , h. l , c. n. pope , t. .
iman and b. tekin , phys . rev .
* d83 * ( 2011 ) 061502(r ) .
h. l , y. pang and c. n. pope , phys . rev .
* d84 * ( 2011 ) 064001 .
m. alishahiha and r. fareghbal , phys . rev .
* d83 * ( 2011 ) 084052 .
e. a. bergshoeff , o. hohm , j. rosseel and p. k. townsend , phys . rev .
* d83 * ( 2011 ) 104038 .
n. johansson , a. naseh and t. zojer , jhep * 1209 * ( 2012 ) 114 .
e. a. bergshoeff , o. hohm and p. k. townsend , phys .
* 102 * ( 2009 ) 201301 .
e. a. bergshoeff , o. hohm and p. k. townsend , phys .
* d79 * ( 2009 ) 124042 .
g. w. horndeski , j. math . phys . * 17 * ( 1976 ) 1980 .
j. d. barrow , m. thorsrud and k. yamamoto , jhep * 1302 * ( 2013 ) 146
. j. beltrn jimnez , r. durrer , l. heisenberg and m. thorsrud , jcap * 1310 * ( 2013 ) 064 . k. yoshida and k. shiraishi , physica scripta * 43 * ( 1991 ) 147 . n. kan , k. kobayashi and k. shiraishi , isrn mathematical physics * 2013 * ( 2013 ) 651684 .
n. kan , k. kobayashi and k. shiraishi , phys .
* d88 * ( 2013 ) 044035 .
n. h. barth and s. m. christensen , phys .
* d28 * ( 1983 ) 1876 .
m. fierz and w. pauli , proc .
a173 * ( 1939 ) 211 .
s. randjbar - daemi , a. salam and j. strathdee , nucl .
* b214 * ( 1983 ) 491 .
c. wetterich , phys .
b113 * ( 1982 ) 377 .
y. okada , phys .
b150 * ( 1985 ) 103 .
j. madore , phys . lett .
* a110 * ( 1985 ) 289 .
f. mller - hoissen , class .
. grav . * 3 * ( 1986 ) 665 .
a. b. henriques , nucl
. phys . * b277 * ( 1986 ) 621 .
k. shiraishi , prog .
* 76 * ( 1986 ) 321 .
m. arik , nucl . phys .
* b328 * ( 1989 ) 308 .
n. deruelle and l. faria - busto , phys . rev .
* d41 * ( 1990 ) 3696 .
s. mignemi , mod .
* a1 * ( 1986 ) 337 .
k. maeda , class .
. grav . * 3 * ( 1986 ) 233 .
h. l , a. perkins , c. n. pope and k. s. stelle , arxiv:1502.01028 [ hep - th ] .
a. kehagias , c. kounnas , d. lst and a. riotto , arxiv:1502.04192 [ hep - th ] .
a. r. brown , a. dahlen and a. masoumi , phys . rev .
* d90 * ( 2014 ) 045016 .
p. g. o. freund and m. a. rubin , phys .
b97 * ( 1980 ) 233 . j. j. halliwell , nucl . phys .
* b266 * ( 1986 ) 228 .
y. zhong and x .- z .
li , phys . rev .
* d42 * ( 1990 ) 113 .
y. zhong and x .- z .
li , commun .
. phys . * 15 * ( 1991 ) 113 .
a. a. starobinsky , phys . lett .
* b91 * ( 1980 ) 99 . | we discuss the role of a particular combination of higher derivative terms in higher dimensional theories , especially in the background of spontaneous compactification . we find that the special extension of einstein maxwell theory with higher order terms admits interesting cosmological solutions . |
the authors acknowledge the support of the deutsche forschungsgemeinschaft , f.w . by grant sfb509 and c.v.f . by sfb237 . here
, we generalize the sum over @xmath58 in eq .
( 2 ) to include all sites @xmath9 with lattice coordinates @xmath59 and @xmath60 such that @xmath61 and @xmath62 . to achieve a visible segmentation in fig .
1c and 1d we use @xmath63 . | we present a fast and robust cluster update algorithm that is especially efficient in implementing the task of image segmentation using the method of superparamagnetic clustering . we apply it to a potts model with spin interactions that are
are defined by gray - scale differences within the image .
motivated by biological systems , we introduce the concept of neural inhibition to the potts model realization of the segmentation problem .
including the inhibition term in the hamiltonian results in enhanced contrast and thereby significantly improves segmentation quality . as a second benefit we can - after equilibration - directly identify the image segments as the clusters formed by the clustering algorithm . to construct a new spin configuration the algorithm performs the standard steps of ( 1 ) forming clusters and of ( 2 ) updating the spins in a cluster simultaneously . as opposed to standard algorithms , however , we share the interaction energy between the two steps . thus the update probabilities are not independent of the interaction energies . as a consequence
, we observe an acceleration of the relaxation by a factor of 10 compared to the swendson and wang procedure .
the segmentation of images into connected areas or objects is a formidable task and an important step in the process of recognition .
nature provides us with many examples of biological systems that solve this and other tasks related to the recognition problem in highly efficient ways . taken as such
, the problem is ill - defined : one will distinguish different numbers of objects in a noisy picture depending on the level of contrast and resolution .
a physicists answer to the problem has been presented by the method of ` superparamagnetic clustering of data ' @xcite where the pixels of an image are represented by a potts model of spins which interact in such a way that neighboring spins corresponding to similar pixels tend to align .
then the image - segments ( or objects ) may be identified as subsets or clusters of correlated spins at a given temperature . at high temperature
one will find a disordered paramagnetic phase while , when lowering the temperature , superparamagnetic phases occur with clusters of aligned spins . from a theoretical point of view
any method of simulating a given spin system is equivalent as long as it preserves general concepts such as detailed balance . for practical purposes
it is of course desirable to choose a method that is efficient and best adapted to the model .
cluster update algorithms are commonly used to to accelerate the equilibration of large spin systems @xcite .
as opposed to single spin updates following a metropolis procedure , these algorithms provide a method to update connected clusters of aligned spins simultaneously .
our approach to the problem is twofold : on the one hand we introduce to the spin model the concept of ( 1 ) _ global inhibition _ , motivated by the analogy to neural visual systems@xcite , on the other hand ( 2 )
we have developed a novel cluster algorithm that utilizes the energy landscape , which underlies the equilibration process , in a more efficient way .
\(1 ) the concept of global inhibition is found in many biological neural networks and has successfully been applied also in neural computation @xcite .
we implement it by adding a small global penalty for spins to align .
it serves to identify different clusters by different spin labels without need to observe the spin correlations over a longer time period .
\(2 ) in a cluster update algorithm the clusters are formed by `` freezing '' bonds between aligned spins with some probability . commonly the clusters are then updated independently . we update the clusters taking into account also the interactions on bonds that were not frozen .
in addition the inner surface of the larger clusters is reduced by incorporating islands that they might contain .
both of our improvements are implemented while preserving detailed balance . as a result
, we observe a significant increase in quality and speed . without loss of generality in the following
we will use the problem of segmenting an image into individual objects as an example to describe our approach . specifically , given a picture in form of color ( or gray - scale ) values @xmath0 on the @xmath1 sites of a finite @xmath2 lattice , we have the clustering problem : find ` objects ' i.e. clusters of almost the same color .
we define for each pair of nearest neighbors or _ bond _
@xmath3 on the lattice the distance @xmath4 and the mean distance @xmath5 averaged over all bonds . to perform the clustering task
we assign a spin variable @xmath6 to each site @xmath7 and for each bond @xmath3 an interaction strength @xmath8 with the normalization in eq.([gray ] ) the color of sites @xmath9 is assumed to be similar when the gray value distance @xmath10 is smaller than the average .
then the interaction strength is positive with a maximum value of @xmath11 for equal color .
we implement the spin model in such a way that for neighboring sites with similar color the spins have the tendency to align . for this purpose
we use a @xmath12-state potts model with the hamiltonian @xmath13 here , @xmath14 denotes that @xmath9 are nearest neighbors connected by a bond @xmath3 and @xmath15 is the kronecker delta function .
the second term is introduced in analogy to neural systems , where it is generally called `` global inhibition '' .
it serves to favor different spin values for spins in different clusters as explained below .
this is a concept realized in many neural systems that perform recognition tasks .
the segmentation problem is then solved by finding clusters of correlated spins in the low temperature equilibrium states of the hamiltonian @xmath16 .
we perform this task by implementing a clustering algorithm : in a first step the ` satisfied ' bonds , i.e. those that connect nearest neighbor pairs of identical spins @xmath17 are identified .
the satisfied bonds @xmath3 are then ` frozen ' with some probability @xmath18 .
sites connected by frozen bonds define the clusters . each cluster
is then updated by assigning to all spins inside the cluster the same new value .
commonly this is done independently for each cluster @xcite . in that sense
the external bonds connecting the clusters are ` deleted ' .
here , we use a more general cluster algorithm .
when choosing a new spin configuration we take these bonds into account . to preserve detailed balance , we adjust the bond freezing probabilities @xmath18 and the interaction on the external bonds .
our cluster update algorithm , which we call energy - sharing cluster update ( ecu ) is divided in two basic steps .
similar to the swendson & wang cluster algorithm @xcite also in our approach the temperature remains fixed and no annealing takes place between the iterations .
* as for any cluster update we first identify the _ satisfied _ bonds @xmath3 with @xmath17 and freeze these with probability @xmath19 when @xmath20 and @xmath21 . here
@xmath22 is the product of the boltzmann constant @xmath23 and temperature @xmath24 .
+ the additional coefficients @xmath25 with @xmath26 allow us to `` share '' the interaction energy with the following additional steps .
if one chooses @xmath27 then one obtains the usual swendson - wang clusters which may then be updated independently . * in an intermediate step we identify ` invisible ' islands i.e. clusters according to step ( 1a ) that have a boundary only with _
one _ other cluster and have the same spin value .
these islands often delay the spin flip of the larger cluster in step ( 2 ) as their total boundary may be large . for this reason we want to remove them with some finite probability .
this step is not indispensable for our algorithm but it further improves its performance .
we freeze the bonds between an island and the surrounding cluster with probability @xmath28 where @xmath29 if @xmath3 is a bond connecting an island with a surrounding cluster after step ( 1 ) and otherwise @xmath30 .
we impose the condition @xmath31 .
note that we do not increase the bond freezing probability beyond the swendson - wang probability and no size limit for the islands is implied . *
finally we identify the clusters @xmath32 of spins connected by frozen bonds after steps ( 1a ) and ( 1b ) . on this system of clusters that in similar approaches is referred to as a hyperlattice
@xcite we perform a metropolis update that updates all spins in each cluster simultaneously to a common new label .
the metropolis rate is calculated using the modified hamiltonian @xmath33 as has been shown on general grounds in @xcite detailed balance is preserved under the condition that in the modified hamiltonian one uses @xmath34 .
this amounts to sharing the interaction energy between the clustering and updating steps .
note that the inhibition term in eq .
( [ 3 ] ) does not enter the bond freezing probabilities .
for the cluster update it has the effect of favoring a different spin value for each cluster .
we have tested the performance of the proposed segmentation method based on the hamiltonian @xmath16 in eq .
( 2 ) with a finite inhibition of @xmath35 in combination with the ecu cluster update algorithm with energy sharing parameters @xmath36 . to our experience
the efficiency of the algorithm does not depend sensitively on these parameters .
further refinements may be added to improve the segmentation delivered by the ecu algorithm to cope with more delicate recognition problems @xcite .
we have compared the algorithm to the performance of other known segmentation methods . as methods of reference
we have used in particular the method of simulated annealing and the method of superparamagnetic clustering @xcite without inhibition ( @xmath37 ) using the standard swendson&wang ( sw ) update .
in addition we have tested a variant of the sw update that allows to freeze anti - ferromagnetic bonds @xmath3 when @xmath38 .
an example that illustrates the different solutions to the segmentation problem is shown in fig .
1 . let us explain this comparison in some detail .
the gray scale values that define the interactions @xmath39 according to eq .
( 1 ) are taken from fig .
1a .
some noise is included in this input .
all segmentation methods that we consider use @xmath40 state spin variables @xmath41 .
a random initial configuration of the spins is shown in terms of a gray scale picture in fig .
1b . as a first reference
we show the sequence of a simulated annealing procedure in fig .
1c and 1d . here , the hamiltonian @xmath16 in eq .
( 2 ) with @xmath37 is used to define the metropolis rate of local spin updates @xcite . after each sweep of @xmath1 spin updates the temperature
is lowered by a constant factor @xmath42 @xcite .
we started with a temperature @xmath43 and lowered by @xmath44 in 1c and @xmath45 in 1d for each sweep .
the spin configurations at intermediate steps are shown in fig .
1c and 1d . in the slower annealing procedure the two large rectangles in the image
are segmented according to the original picture while the finer structure is not recognized by this algorithm .
when the faster schedule is applied as in 1d then even the larger connected areas are divided into artificial segments .
obviously the simulated annealing method is inefficient for the segmentation task and due to slowing down at low temperatures the local update is very time consuming . even optimizing the annealing rate during the schedule can not change this picture as an extremely slow rate is needed to indentify the fine structure of the thin border line . in fig .
1e - g we compare different cluster update algorithms that avoid the problem of slowing down and we test the influence of the inhibition term and the energy sharing that are included only in fig . 1 g . comparing the series of spin configurations in fig .
1e and 1 g one notices that the inhibition term in 1 g indeed introduces a forced contrast between different segments as compared to 1e , in particular at @xmath46 and @xmath47 .
also the increase in speed is remarkable .
+ in fig .
1f we test a cluster update algorithm @xcite that includes anti - ferromagnetic clustering where in the clustering step ( 1a ) anti - ferromagnetic bonds with @xmath38 and @xmath48 are frozen with probability @xmath49 $ ] .
the clusters defined by ferro- and anti - ferromagnetic bonds are updated while preserving @xmath17 on the ferro- and @xmath48 on the anti - ferromagnetic bonds .
this method introduces additional contrast between areas of different input color but it fails at low temperature where artifacts are generated due to the noise in the input .
the convergence characteristic of the different algorithms is shown in fig .
[ compare]h , where the energy of the spin - lattice is plotted as a function of time at fixed temperature .
the relaxation time of the ecu - algorithm is about 10 times faster than that of the other algorithms .
let us note that the only parameters that enter our segmentation method are the factors of proportionality @xmath50 that determine the share of energy for the bond freezing steps and the inhibition strength @xmath51 .
mainly the @xmath52-parameters are relevant for the efficiency of the segmentation while @xmath51 indroduces some contrast to the representation of the clusters in terms of spin values .
we have not attempted to optimize the choice of the parameters to speed the segmentation of fig .
1a .
rather we are interested in a general purpose algorithm and we have successfully tested the robustness of the ecu segmentation with the present choice of parameters for many different pictures . to demonstrate this robustness we show three examples in fig . 2 .
despite the bad quality of the input a usable segmentation was found within a small number of iterations . a seemingly continuous gray scale background in fig .
2a is segmented into only few clusters identifying the foreground character . in fig .
2b we illustrate that for practical purposes , in this case detection of the license plate , even an intermediate iteration step , here @xmath53 , may be used without need to wait for equilibration ( at @xmath54 ) .
fig.2c shows the quality of segmentation for a highly complicated picture .
data clustering becomes tremendously complicated when the intrinsic correlation between the data points which belong to the same cluster is small .
a situation like this always occurs if the clusters extend into a thin , thread - like shape or an almost fractal structure , for example when dealing with images of biomolecules , polymers , or stellar structures .
the ecu makes better use of the energy landscape which underlies the clustering problem by _ sharing energy _ between the bond - freezing and the spin - update steps of the algorithm .
the additionally introduced global inhibition enhances contrast . as a consequence the quality of the results improves and
, most notably , energy sharing leads to an acceleration of the segmentation by about a factor of ten ( fig .
[ compare ] h ) .
in the course of development of the modern cluster update algorithms similar ideas have been proposed on sometimes more general grounds .
kandel and domany @xcite lay out how to preserve detailed balance for a broad class of algorithms and they show how several other proposed update variants @xcite may be rephrased to comply with this .
the ecu algorithm is also embedded in this framework .
niedermayer @xcite shows that in the clustering step ( 1a ) any function @xmath55 can be used in principle , but proposes for practical purposes to apply @xmath56 $ ] with some appropriately chosen @xmath57 . with this choice the contribution of the non - frozen bonds to the update is clipped at @xmath57 . in our case
we share the energies in a proportional way between the clustering and update steps .
the alignment of clusters is thus enhanced by also including the stronger bonds with higher energy content .
+ in summary , the recognition task of segmenting an image may be performed with high efficiency by a simple cluster update algorithm if global inhibition is implemented .
furthermore , we believe that our cluster update approach may also be useful for the simulation of other spin models as its efficiency is not dependent on the special properties of the potts model we use here . |
although numerical experiments are the most powerful methods to study the formation of structures , the development of analytical or semi - numerical methods is very important , since they help to improve our understanding of the physical processes during the formation .
+ a class of analytical methods is that based on the ideas of press & schechter ( 1974 ) and on their extensions ( bond et al .
1991 , lacey & cole 1993 ) : + the linear overdensity computed at a given point of an initial snapshot of the universe fluctuates when the smoothing scale decreases .
this fluctuation is a markovian process when the smoothing is performed using a top - hat window in fourier space . for any value of the smoothing scale @xmath8 the overdensity field
is assumed to be gaussian .
the mass @xmath5 contained in a given scale @xmath8 depends on the window function used .
for a top - hat window the relation is : @xmath9 , where @xmath10 and @xmath11 are the values of the mean density and the density parameter of the universe , @xmath12 is the gravitational constant and @xmath13 is the hubble s constant .
the index @xmath14 indicates that all the above values are calculated at the initial snapshot .
mass dispersion @xmath15 at scale @xmath8 is a function of mass @xmath5 and is usually denoted by @xmath16 , that is @xmath17 $ ] .
let the random walk of the overdensity cross for first time a given barrier @xmath18 at some value @xmath19 of @xmath16 .
the mass element associated with the random walk is considered to belong to a halo of mass @xmath20 at the epoch with redshift @xmath21 .
the distribution of haloes , at some epoch @xmath21 , is related to the first crossing distributions , by the random walks , of the barrier that corresponds to epoch @xmath21 .
+ the simplest form of the barrier comes from the spherical model .
it is well known that in an einstein - de sitter universe , a spherical overdensity collapses at @xmath21 if the linear extrapolation of its value up to the present exceeds @xmath22 , and this value provides a first reasonable choice for the barrier .
the involved quantities ( density overdensities , dispersion ) are usually extrapolated to the present epoch and thus the barrier in the spherical collapse model is written in the form @xmath23 , where @xmath24 is the growth factor derived by the linear theory with @xmath25 .
the form of the spherical barrier permits the analytical evaluation of the first crossing distribution @xmath26 . + despite the simplicity of the spherical model , its results agree relatively well with the results of n - body simulations ( e.g. white , efstathiou & frenk 1993 ; lacey & cole 1994 ; gelb & bertschinger 1994 ; bond & myers 1996 ) .
deviations appear in the resulting mass functions at both high and low masses .
sheth & tormen @xcite consider a barrier of the form @xmath27 with @xmath28 , where @xmath29 and @xmath30 , in order to describe the ellipsoidal form of collapse .
the first crossing distribution @xmath31 that results from the ellipsoidal model works better than the spherical one .
for example , yahagi et al .
@xcite showed that the multiplicity function resulting from n - body simulations is far from the predictions of spherical model while it shows an excellent agreement with the results of the ellipsoidal model . on the other hand , lin et al .
@xcite compared the distribution of formation times of haloes formed in n - body simulations with the formation times of haloes formed in terms of the spherical collapse model .
they found that n - body simulations give smaller formation times .
hiotelis & del popolo @xcite used merger trees to show that the ellipsoidal collapse model leads to formation times that are shifted to smaller values relative to the spherical collapse model .
thus , a better agreement with the predictions of n - body simulations is achieved .
the distributions of formation times are studied in more detail by other authors ( giocoli et al .
we note that in the cases of a barrier with @xmath32 , or @xmath33 , the corresponding first crossing distribution can be found analytically ( e.g. mahmood & rajesh 2005 ) . for the above non - linear barrier
, sheth & tormen @xcite proposed an analytical expression .
it is shown that this analytical expression approximates well the exact expression found as a numerical solution of an integral equation ( zhang & hui 2008 ) .
+ the constrained first crossing distribution is given by the relation : @xmath34\mathrm{d}s\end{aligned}\ ] ] where @xmath35 , @xmath36 and the function @xmath37 is given by : @xmath38^n}{n!}\frac{\partial ^n}{\partial s^n}b(s , z)\end{aligned}\ ] ] given that a mass element is a part of a halo of mass @xmath39 at redshift @xmath40 the probability that at higher redshift @xmath21 this mass element was a part of a smaller halo @xmath5 is given by eq.(1 ) . the unconstrained expression , @xmath31 , results by setting @xmath41 .
the analytical expression is very useful since it allows the construction of merger trees .
+ the purpose of this paper is : 1 . to construct merger trees able to give merger rates of dark matter haloes .
2 . to compare these merger rates with those predicted by the results of n - body simulations .
3 . to extend the calculations to scales that are not accessible by numerical simulations and finally , 4 . to study the role of some of the main parameters involved . in sect .
2 we give a brief description of the tree code used .
then , the definition of merger rates is presented and the analytical formulae , predicted by other authors from the results of n - body simulations , are given .
+ in sect .
3 we present our results .
merger - trees used in this paper are constructed using eq.(1 ) .
let us assume a descendant halo of mass @xmath42 at redshift @xmath43 .
we study its past by the following procedure : a new larger redshift @xmath44 is chosen .
this is done by solving for @xmath44 the equation @xmath45 , where @xmath46 is a constant ( one of the parameters of the algorithm ) .
then , a value @xmath47 is chosen from the distribution ( 1 ) .
the mass of the progenitor is @xmath48 .
this progenitor is accepted if its mass is larger than a lower limit @xmath49 and smaller than the mass left to be resolved .
if @xmath50 is less than @xmath49 then @xmath50 is added to a sum that is named @xmath51 .
the mass left to be resolved is , at the choice of the k - th progenitor , @xmath52 . if @xmath53 is larger than @xmath49 we proceed to the selection of a next progenitor , while if @xmath53 is smaller than @xmath49 we proceed to the next redshift .
it is obvious that in such construction , the number of progenitors can be larger than two , despite the original assumption of lacey & cole @xcite , and this leads to a better representation of the distribution of progenitors .
our algorithm is based on the `` n - branch '' idea of somerville & kollat @xcite , but extended in order to incorporate aspects of the ellipsoidal collapse results .
a complete description of the tree construction is given in hiotelis & del popolo @xcite .
the comparisons between the predictions of the tree and analytical predictions of the distribution of the number of progenitors show that the tree method is reliable in following the evolution of structures .
various tree - construction algorithms have been presented in the literature ( cole 1991 ; kauffmann & white 1993 ; sheth & lemson 1999 ; cole et al .
2000 ; neinstein & dekel 2008 ) .
the accuracy of these algorithms is usually against simplicity .
for example , the algorithm of neinstein & dekel @xcite requires the solution of several differential equations with nontrivial boundary conditions .
+ we examine one descendant halo from a sample of @xmath54 haloes with masses in the range @xmath55 present at redshift @xmath43 . for a single halo
the procedure is as follows : let @xmath56 be the masses of its @xmath57 progenitors at redshift @xmath58 .
for matter of simplicity we assume that the most massive progenitor is @xmath59 .
we define @xmath60 for @xmath61 and we assume that the descendant halo is formed by the following procedure : during the interval @xmath62 every one of the progenitors with @xmath61 merge with the most massive progenitor @xmath63 and form the descendant halo we examine .
we repeat the above procedure for all haloes in the range @xmath55 found in a volume @xmath64 of the universe .
then , we find the number denoted by @xmath65 of all progenitors with @xmath66 in the range @xmath67 and we calculate the ratio @xmath68 .
we define the merger rate @xmath69 as follows : @xmath70 let the number density of haloes with masses in the range @xmath71 at @xmath43 be @xmath72 .
the ratio @xmath73 measures the mean number of mergers per halo , per unit redshift , for descendant haloes in the range @xmath55 with progenitor mass ratio @xmath74 .
+ we note that the definition of mean merger rate is exactly the same as in fakhouri & ma @xcite ( fm08 hereafter ) .
we also use the assumption that all progenitors merge with the most massive one ( see fm08 for a discussion of this assumption . ) + lacey & cole @xcite showed that in the spherical model the transition rate is given by : @xmath75_{_{z = z_d}}\nonumber\\ \times \frac{1}{\sigma^2(m_d ) } \left[1-\frac{\sigma^2(m_d)}{\sigma^2(m)}\right]^{-3/2}\left[\frac{\mathrm{d}\sigma(m)}{\mathrm{d}m}\right]_{_{m = m_d}}\nonumber\\ \times\exp\left[-\frac{\delta^2_c(t)}{2 } \left(\frac{1}{\sigma^2(m_d)}-\frac{1}{\sigma^2(m)}\right)\right]\mathrm{d}m_d \end{aligned}\ ] ] this provides the fraction of the mass belonging to haloes of mass @xmath5 that merge instantaneously to form haloes of mass in the range @xmath55 at @xmath43 .
the product @xmath76 , where @xmath77 is the unconditional first crossing distribution for the spherical model , gives the above fraction of mass as a fraction of the total mass of the universe and successively multiplying by @xmath78 the number of those haloes is found .
then , by dividing by @xmath79 ( that equals to the number of the descendant haloes ) we find : @xmath80_{_{z = z_d}}\nonumber\\ \times \left[1-\frac{\sigma^2(m_d)}{\sigma^2(m)}\right]^{-3/2}\mathrm{d}{m } \end{aligned}\ ] ] assuming a binary merge , where @xmath74 is the ratio of the small progenitor to the large one ( @xmath81 ) , using @xmath82 and substituting in ( 5 ) we have the final expression for the binary spherical case , that is : @xmath83_{_{z = z_d}}\nonumber\\ \times\left[1-\frac{\sigma^2(m_d)}{\sigma^2(m)}\right]^{-3/2 } \end{aligned}\ ] ] fm08 analyzed the results of the millennium simulation of springel et al .
stewart et al .
@xcite , ( sbbw08 hereafter ) , used their high - resolution n - body simulations to study the merger rates of dark matter haloes .
fitting formulae proposed by the above authors are separable in the three variables , mass @xmath42 , progenitor ratio @xmath74 and redshift @xmath21 .
these formulae are of the form : + @xmath84 fm08 proposed @xmath85 , h(z)=\left(\frac{\mathrm{d}\delta_c}{\mathrm{d}z}\right)^{a_4}_{_{z = z_d}}$ ] . where the values of the parameters are @xmath86 .
+ on the other hand , sbbw08 proposed : @xmath87/{\xi}^{b_2 + 1 } , h(z)=\left(\frac{\mathrm{d}\delta_c}{\mathrm{d}z}\right)^{b_4}_{_{z = z_d}}$ ] where @xmath88 + note that the formulae proposed by the above authors show some significant differences .
first , the dependence on redshift @xmath21 differs in the above two formulae . although the quantity @xmath89 does not vary significantly with redshift , the exponents @xmath90 and @xmath91 can cause significant differences in the merger rates .
second , exponents @xmath92 and @xmath93 that define the dependence on the mass are quite different . in the approximation of fm08 mean merger rates are practically independent on mass .
+ additionally , the above formulae show significant differences in their slope at small and large values of @xmath74 .
the logarithmic slope of @xmath12 , @xmath94 , at @xmath95 is @xmath96 for the fm08 model , and @xmath97 for the sbbw08 model . for @xmath98 the above logarithmic slope is @xmath99 for the fm08 model and it tends to @xmath100 in the formula of sbbw08 .
+ therefore , it is interesting to study merger rates predicted by merger - trees and to compare their characteristics with those of the above fitting formulae .
) for haloes with present day masses @xmath101 , @xmath102 , @xmath103 and @xmath104 respectively , predicted by the formula proposed by fm08 for @xmath105 and @xmath106 .
dashed lines are predicted by the formula proposed by sbbw08 for @xmath107 and @xmath108 .
solid lines are the predictions of merger trees described in the text by eq.(3 ) for @xmath109 and @xmath110 .
we used a sample of @xmath111 present - day haloes for each of the three cases and we evolved the system by a single time - step for @xmath112 .
the minimum mass @xmath49 in every case equals to 0.005 times @xmath113,width=340 ] we used a flat model for the universe with @xmath114 and @xmath115 and a power spectrum proposed by smith et al .
@xcite given by : @xmath116^b}\ ] ] the values for the parameters are : @xmath117 and @xmath118 .
smoothed fields are calculated using the top - hat window function .
the constant @xmath119 of proportionality , is found using the procedure of normalization .
we used two different values for the normalization namely @xmath120 and @xmath121 respectively .
we also use a system of units where @xmath122 and @xmath123 . in this system of units , @xmath124 . for haloes with masses in the range @xmath125 and @xmath126 , respectively .
the merger tree used started at @xmath127 with @xmath112 for a sample of @xmath128 haloes each of mass @xmath102 .
after 22 time steps the redshifts are @xmath129 .
the resolution mass is @xmath130 .
long dashes , big dots and small dashes are the results from the formulae of sbbw08 , fm08 and the binary spherical model for @xmath131 at @xmath132.,width=340 ] we performed a large number of tree realizations in order to study merger rates .
first , we found that good fits are achieved by both of the above formulae but for different values of the parameters than those proposed by the above authors .
we found that merger rates depend on the mass of the descendant halo as @xmath133 and on the redshift as @xmath134^{1.1}$ ] .
so , in the comparisons that follow , the formulae of eq .
7 are used with @xmath135 .
additionally , in the formula of sbbw08 , we use the value of @xmath108 instead of @xmath136 .
+ in fig.1 , comparisons between the predictions of merger trees and those of formulae given by eq . 7 are shown .
the power spectrum used is that given by eq .
8 for @xmath137 .
details are given in the caption of the figure .
we note a very good agreement but also a rapid fall of the predictions of merger trees for small values of @xmath74
. we will show below that this is a matter of resolution that depends on the value of @xmath49 .
+ the predictions of this figure are just after one time step of the tree algorithm . for a higher redshift @xmath138 one
has to use one of the two alternatives : to start with a sample of haloes at @xmath138 and after a single time step to move to a new redshift @xmath139 , to calculate the merger rates , or to start with a sample of haloes at the present epoch @xmath127 and to calculate merger rates after a number of time steps when the redshift has a desirable value .
, for an initial sample of 10000 haloes each with mass @xmath102 after 72 steps ( thin solid line ) , after 144 steps ( thin solid line with large dots ) and after 36 steps ( line with thin and small dashes ) .
these three lines correspond to about the same redshift ( z=3 ) , and to the same resolution mass @xmath140.thick solid line is the prediction for smaller minimum mass , @xmath141 .
big dots are the predictions of fm08 , while large thick dashes , the predictions of sbbw08 . , width=340 ]
obviously , the second approach tests the ability of the merger tree to follow the time evolution of structures and it is close to the nature of n - body simulations .
this approach is followed in our calculations .
+ fig.2 shows merger rates predicted by a merger tree , started at @xmath127 , after 22 time steps , for @xmath112 and @xmath137 .
redshifts are @xmath129 .
the sample of haloes at @xmath127 consists of @xmath128 haloes each with mass @xmath142 .
the number of haloes in the range @xmath143 at @xmath43 is @xmath144 and @xmath145 . in the range @xmath146
there are @xmath147 and @xmath148 haloes .
finally , in the range @xmath149 there are @xmath150 and @xmath151 haloes .
the resolution mass , @xmath49 used is @xmath152 . for @xmath132
long dashes show the formula of sbbw08 , dots the formula of fm08 and small dashes the prediction of the binary spherical model given by eq .
6 . in the following we examine the role of some -of the large number- of the parameters involved in the construction of merger trees .
these are : 1 .
the resolution mass @xmath49 2 .
the step in redshift , defined by the parameter @xmath46 3 .
the value of @xmath153 and 4 . the number of realizations , that is the number of haloes at @xmath127 .
resolution mass @xmath49 is a crucial parameter .
a sample of present day haloes of mass @xmath39 is analyzed to smaller and smaller haloes as the redshift becomes higher .
let @xmath42 be the mass of a descendant halo at some high redshift , @xmath154 its largest progenitor and @xmath155 another progenitor .
obviously , @xmath156 and @xmath157 .
thus @xmath158 and consequently @xmath159 and @xmath49 are related by @xmath160 . since at high redshifts @xmath42 is significantly smaller than @xmath39 , the condition for the merger rate curves to extend to the left up to values as small as @xmath159 is @xmath161 .
we found that for decreasing @xmath49 the value @xmath74 , at which the merger rates curves show their rapid fall , moves to the left .
thus , this rapid fall seen in figs 1 and 2 is clearly a matter of resolution .
+ on the other hand , the results do not seem to be sensitive to the values of the step in redshift @xmath21 .
this step depends on the parameter d described in section 2.1 .
we examined cases , for d=0.025 , d=0.05 and d=0.1 .
we found that the results are the same .
+ fig.3 shows the role of the above two parameters .
it presents merger rates , for descendant haloes with masses in the range @xmath162 , for an initial sample of 10000 haloes , each with mass @xmath102 , after 72 steps ( thin solid line ) , after 144 steps ( thin solid line with large dots ) and after 36 steps ( line with thin and small dashes ) .
all lines correspond to about the same redshift ( z=3 ) , since we used different values for the time step parameter , @xmath112 , @xmath163 and @xmath164 , respectively .
the corresponding redshifts are @xmath165 , respectively .
we see that the resulting merger rates are similar , although the values of @xmath166 used in eq .
3 differ , @xmath167 .
+ at @xmath132 .
dots are the predictions for the power spectrum normalized to @xmath137 , while dashes correspond to @xmath168 .
solid line shows a law @xmath169 for comparison .
deltas correspond to the model of fm08 while solid line with black squares to the model of sbbw08.,width=340 ] in the same figure the thick solid line is the prediction for a resolution mass an order of magnitude smaller , namely @xmath141 and shows the above described role of @xmath49 .
finally , big dots and thick dashes are the predictions of fm08 and sbbw08 , respectively , for values of the parameters given above .
all calculations in fig.3 are for @xmath111 present day haloes and for @xmath137 .
the above described role of the two parameters is the same for the range of present day masses , @xmath0 , we examined .
we note that small values of @xmath49 lead to large numbers of haloes at the past .
for example , @xmath111 present day haloes with mass @xmath170 have for @xmath171 , about @xmath172 progenitors at @xmath173 . for the minimum value of @xmath49 that we have used ,
that is @xmath174 and corresponds to a resolution @xmath175 , at the same redshift , the number of progenitors is larger than @xmath176 . thus , tree construction becomes a computing time consuming procedure .
+ major mergers seem to play an important role in the formation of dark matter haloes .
n - body simulations show that haloes which experienced a recent major merger event , appear to have lower concentrations and steeper inner density profiles ( e.g. ascasibar et al .
2003 , tasitsiomi et al . 2004 ) as well as larger values of spin parameters ( gardner 2001 , peirani et al .
this last result is also supported by semi - numerical results ( e.g. vitvitska et al .
2002 , hiotelis 2008 ) .
thus , it is interesting to study in more detail the behavior of merger rates curves at large values of @xmath74 .
this demands smooth curves and consequently large number of haloes .
so , in fig.4 we present an example of the role of the last two of the parameters we examined .
these parameters are the number of present day haloes and the value of @xmath153 .
all curves of this figure correspond to @xmath132 .
dots are the predictions for a set of @xmath128 present day haloes each with mass @xmath177 for a power spectrum with @xmath137 and for descendant haloes in the range @xmath178 . at @xmath132
, there are @xmath179 and @xmath180 haloes .
dashes are the predictions for a set of @xmath181 present day haloes of the same mass for @xmath168 and for the above range of mass of the descendant haloes . at @xmath132 , there are @xmath182 and @xmath183 haloes .
solid line shows a law @xmath169 for comparison .
deltas are the predictions of the fitting formula of fm08 and finally , the solid line with big dots shows the predictions of sbbw08 .
fig.5 is similar to fig.4 but for @xmath184 and for descendant haloes in the range @xmath185 .
there are @xmath186 and @xmath187 for the tree corresponding to the dashed line while there are @xmath188 and @xmath189 for the tree corresponding to dots .
it is clear from the above two figures , that - at the level of accuracy of the calculations of this paper - no differences can be detected due to the difference between @xmath190 and @xmath191 of the values of @xmath153 .
larger number of present day haloes leads , obviously , to smoother curves but they cause no difference in the overall shape . +
the above study of merger rates that are predicted by merger trees based on the extended ps theory using the ellipsoidal collapse model , leads to the following conclusions that hold for haloes in the range of mass @xmath192 and for redshift @xmath2 . 1 .
merger rates depend on the mass of the descendant halo as @xmath193 with n@xmath194 .
this dependence is in practice indistinguishable from the value @xmath195 proposed from sbbw08 but is far from the value of @xmath196 proposed by fm08 .
2 . we found that merger rates depend on the redshift through the quantity @xmath197^l$ ] , where @xmath198 .
this value of the exponent @xmath199 is close to the value predicted by the binary spherical case ( see eq.6 ) and between the values 0.371 and 2 proposed by fm08 and sbbw08 , respectively .
3 . the results of merger trees are in better agreement ( for the above values of @xmath200 and @xmath199 ) with the predictions of formulae of fm08 and sbbw08 than with the predictions of binary spherical model given by eq.6 .
the binary spherical model underestimates merger rates for small values of @xmath74 while merger trees , for proper resolution , give results much closer to those of n - body simulations .
4 . for large values of @xmath74 ,
merger rate curves scale approximately as @xmath201 .
we examined steps in redshift from @xmath202 to @xmath203 and we found no differences in the results .
smaller values of the resolution mass @xmath49 give a better agreement with the above fitting formulae for small values of @xmath74 .
the construction of reliable merger - trees is a subject under current investigation .
these algorithms usually have fundamental problems regarding mass conservation , accurate representation of the distribution of progenitors etc . as regards n - body simulations , in addition to their resolution problems , it is characteristic that some of their results depend on the techniques used for their derivation .
for example , bett et al .
( 2007 ) showed that the values of spin parameters of haloes and their behavior as a function of mass depends crucially on the halo - finding algorithm .
however , the study of physical parameters -as for example a more accurate power spectrum- seems not to be proper at this stage .
probably , the results are more sensitive to the method used for their analysis than to physical parameters and thus , any detailed description between the results of n - body and those of merger - trees may not be very useful . however , there are characteristic trends that show interesting agreement
. + further improvements of tree construction algorithms as well of the quality of n - body simulations could help to understand better the physical picture during the process of the formation of dark matter haloes .
i would like to thank the _ empirikion foundation _ for its financial support .
ascasibar y. , yepes g. , mller v. , gottlber s. , 2003 , mnras , 346 , 731 bett p. , eke v. , frenk c.s .
, jenkins a. , helly j. , navarro j. , 2007 preprint ( astro - ph/0608607v3 ) bond j.r . , cole s. , efstathiou g. , kaiser n. , 1991 , apj , 379 , 440 bond j.r .
, myers s. , 1996 , apjs , 103 , 41 cole s. , 1991 , apj , 367 , 45 cole s. , lacey c.g . ,
baugh c.m . ,
frenk c.s . , 2000 ,
mnras , 319 , 168 fakhouri o. , ma c - p . , preprint ( astro - ph/0710.4567v2 ) ( fm08 ) gardner j.p . , 2001 , apj , 557 , 616 gelb j. , bertschinger e. , 1994 , apj , 436 , 467 giocoli c. , moreno j. , sheth r.k . , tormen g. , 2007 , mnras , 977 - 983 hiotelis n. , del popolo a. , apss , 2006 , 301,167 hiotelis n. , 2008 , apss , 315 , 191 kauffmann g. , white s.d.m . , 1993 ,
mnras , 261 , 291 lacey c. , cole s. , 1993 , mnras , 262 , 627 lacey c. , cole s. , 1994 , mnras , 271 , 676 lin w.p .
, jing y.p . , lin l. , 2003 , mnras , 344 , 1327 mahmood a. , rajesh r. , 2005 , preprint ( astro - ph/0502513 ) neinstein e. , dekel a. , 2008 , mnras , 388 , 1792 , 2008 peirani s. , mohayaee r. , pachero j. , 2004 , mnras , 348 , 921 press w.h . , schechter p. , 1974 ,
apj , 187 , 425 sheth r.k . , lemson g. , 1999 , mnras , 305 , 946 sheth r.k . , tormen g. , 1999 , mnras , 308 , 119 sheth r.k . , tormen g. , 2002 , mnras , 329 , 61 smith c.c .
, klypin a. , gross m.a . k. , primack j.r . , holtzman j. , 1998 , mnras , 297 , 910 somerville r.s . , kolatt t.s . , 1999 ,
mnras , 305 , 1 springel v.,et al . , 2005 , nat , 435 , 629 stewart k.r . , bullock j.s .
, barton e.j . ,
wechsler r.h . , 2008 ,
preprint ( astro - ph/0811.1218v1 ) ( sbbw08 ) tasitsiomi a. , kratsov a.v . , gottlbber s. , klypin a.a . , 2004 ,
apj , 607,125 vitvitska m. , klypin a.a . , kravtsov a.v . , wechsler r. , primack j. , bullock j. , 2002 , apj , 581 , 799 white s.d.m . , efstathiou g. , frenk c. , 1993 , mnras , 262 , 1023 yahagi h. , nagashima m. , yoshii y. , 2004 , apj , 605 , 709 zhang j. , hui l. , 2006 , apj , 641 , 641 , 2005 | we construct merger trees based on the extended press - schechter theory ( eps ) in order to study the merger rates of dark matter haloes over a range of present day mass ( @xmath0 ) , progenitor mass @xmath1 ) and redshift ( @xmath2 ) .
we used the first crossing distribution of a moving barrier of the form @xmath3 , proposed by sheth & tormen , to take into account the ellipsoidal nature of collapse .
we find that the mean merger rate per halo @xmath4 depends on the halo mass @xmath5 as @xmath6 and on the redshift as @xmath7 .
our results are in agreement with the predictions of n - body simulations and this shows the ability of merger - trees based on eps theory to follow with a satisfactory agreement the results of n - body simulations and the evolution of structures in a hierarchical universe . [ firstpage ] cosmology : theory dark matter , galaxies : haloes structure formation , methods : analytical |
to understand the properties of materials one has to understand surfaces and interfaces : materials interact with their environment through their surfaces , and material properties are profoundly influenced by internal interfaces .
processes related to surfaces and interfaces therefore play a critical role in nature and in a variety of technological applications , such as electronic , magnetic , and optical devices , sensors , catalysts , coatings , and many other industrial systems and processes .
@xcite surfaces also play a vital role in biology and medicine , since most biological reactions occur at surfaces and interfaces .
the applications of surface science in medicine range from the growth of bio - compatible surfaces for tissue cultures , through medical implants , to the design of innumerable medical devices.@xcite it is therefore crucial to understand the fundamental processes occurring at surfaces and interfaces in order to provide means of controlling and manipulating such systems .
the large - scale properties of growing interfaces have been the object of an enormous amount of work in recent years,@xcite but much less attention has been paid to interfacial structure on a microscopic scale .
this is unfortunate because many important interface properties , such as mobility and catalytic and chemical activity , are largely determined by the microscopic interface structure .
the microscopic structure limits the interfacial propagation velocity under an external driving force , such as an external field for a magnetic or dielectric domain wall , or a difference in chemical potential between the bulk phases for a crystal surface .
since the detailed microscopic mechanism of the interface motion is often not known , one standard way to gain insight in the process is by constructing a stochastic model that mimics its essential features .
dynamics that conserve the order parameter , or ones that do not , must be chosen according to the physical characteristics of the system .
once this is decided , there are several dynamics in each category to choose among .
it is well established that structures arising from different dynamics that obey detailed balance and respect the same conservation laws exhibit universal asymptotic large - scale features .
however , recent studies@xcite show that there are important differences between the _ microstructures _ of field - driven interfaces obtained with different dynamics , and that these differences very significantly influence important interface properties such as mobility . a mean - field , nonlinear - response theory developed in refs .
indicates that there are two different classes of nonconservative dynamics that lead to significantly different surface microstructures .
these are the _ soft _ dynamics in which the single - site transition rates can be factorized into one part that depends only on the applied force and a second part that depends only on the interaction energies , and the _ hard _ dynamics for which this factorization is not possible.@xcite by this classification , the widely used glauber and metropolis dynamics are hard .
soft dynamics are appropriate for solidification or adsorption problems where the driving force is a chemical - potential difference.@xcite in previous papers@xcite two of us introduced a dynamic mean - field approximation for the microstructure , based on the burton - cabrera - frank solid - on - solid ( sos ) approximation.@xcite for soft dynamics , interface structures should remain independent of the applied force , while there should be a clear dependence on the force for hard dynamics .
monte carlo ( mc ) simulation results with the soft glauber@xcite and the standard hard glauber dynamics@xcite confirm these predictions .
moreover , sos surfaces generated with the hard glauber dynamics exhibit a skewness that is absent for the soft glauber dynamics .
since the mean - field results depend on the absence of short - range correlations along the interfaces , it remains an open question to which extent these characteristics are shared by all soft and hard dynamics . for the present study we have considered two dynamics that include a local energy barrier representing a transition state inserted between individual ising or lattice - gas states .
such arrhenius dynamics , as they are often called , are appropriate in kinetic mc simulations of discrete ising or lattice - gas models in which the discrete states serve as approximations for high - probability configurations in an underlying continuous potential.@xcite examples are the study of diffusion in a lattice - gas model in a continuous corrugation potential,@xcite the relaxation process from the high - spin state of molecular bistable solids,@xcite or the approximation of a continuous spin model with strong uniaxial anisotropy by an ising model.@xcite when applied to kinetic ising lattice - gas models , arrhenius dynamics give nucleation rates quite different from the ones given by the standard glauber dynamics;@xcite another warning about the importance of choosing the right dynamics for specific physical or chemical systems .
the two arrhenius dynamics that we consider here are the common one - step - dynamics ( osd)@xcite and the two - step transition dynamic approximation ( tda).@xcite the osd dynamics are soft and the tda dynamics are hard , according to the definitions given above .
their transition rates are defined in sec .
the osd dynamics are commonly used in studies of adsorption , such as the electrosorption of halides on single - crystal silver electrodes.@xcite among the experimental systems that have been described with the tda dynamics are the diffusion of h atoms on single - crystal tungsten surfaces.@xcite sos surfaces belong to the kardar - parisi - zhang ( kpz ) dynamic universality class,@xcite in which the macroscopic , stationary distribution for flat , moving interfaces is gaussian , corresponding to a random walk with independent increments .
nevertheless , the step heights in several discrete models in this class are correlated at _ short _ distances.@xcite in the mean - field approximation used here , these short - range correlations are ignored .
the resulting discrepancies will be apparent when we compare the theoretical results with kinetic mc simulations .
the remainder of this paper is organized as follows . in sec .
[ sec : model ] we introduce the sos interface model and give the transition rates for the tda and osd dynamics . in sec .
[ sec : nlr ] we summarize the mean - field approximation for the time evolution of the single - step probability density function ( pdf ) , as well as its stationary form .
we also give expressions for the class populations and interface velocity in terms of the applied force , the temperature , and the angle of the interface relative to the lattice axes . in sec .
[ sec : mc ] we compare simulations and analytical predictions for the detailed stationary interface structure , including the asymmetry of the simulated nonequilibrium interfaces .
a summary and conclusion are provided in sec .
[ sec : disc ] .
the sos interfaces are described by the nearest - neighbor @xmath1 ising hamiltonian with anisotropic , ferromagnetic interactions @xmath2 and @xmath3 in the @xmath4 and @xmath5 direction , respectively : @xmath6 where @xmath7 , @xmath8 runs over all sites , and the applied field @xmath9 is the driving force .
the interface is introduced by fixing @xmath10 and @xmath111 for large negative and positive @xmath5 , respectively . without loss of generality
we take @xmath12 , such that the interface on average moves in the positive @xmath5 direction .
this ising model is equivalent to a lattice - gas model with local occupation variables @xmath13.@xcite specifically , we identify @xmath14 with @xmath15 ( occupied or `` solid '' ) and @xmath16 with @xmath17 ( empty or `` fluid '' ) .
the interactions in the ising model , @xmath18 , are related to the ones in the lattice - gas model , @xmath19 , as @xmath20 , and the applied field is related to the lattice - gas chemical potential @xmath21 as @xmath22 , where @xmath23 is the coexistence value of @xmath21 . here
we will use ising or lattice - gas language interchangeably as we feel it makes a particular aspect of the discussion clearer .
the sos model considers an interface in a lattice - gas or @xmath1 ising system on a square lattice of unit lattice constant as a single - valued integer function @xmath24 of the @xmath4-coordinate , with steps @xmath25 at integer values of @xmath4 .
a typical sos interface configuration is shown in fig . [
fig : pict ] . in this paper the two possible states of the site @xmath26 are denoted by the two `` spin '' values @xmath27 .
( in order that the step positions and the interface heights be integer as stated above , we place the spins at odd half - integer values of @xmath4 and @xmath5 , i.e. , at the centers of the unit cells separated by dotted lines in fig .
[ fig : pict ] . )
the interface will be made to evolve under two different dynamics that contain a microscopic energy barrier against individual spin flips .
these are the transition dynamics approximation ( tda)@xcite and the so - called one - step dynamics ( osd).@xcite the barrier represents a transition state which is inserted between the states allowed in the hamiltonian , such as a saddle point in a corrugation potential for particle diffusion,@xcite or a high energy associated with a transitional spin state that is not along one of the two directions allowed by the ising hamiltonian.@xcite here we express the transition - state energy by the following approximation,@xcite @xmath28 where @xmath29 and @xmath30 are the initial and final energies , and @xmath31 is the bare , microscopic energy barrier . in electrochemical applications , such as electron- or ion - transfer reactions
, this corresponds to the symmetric butler - volmer approximation.@xcite the construction corresponding to eq .
( [ eq : tener ] ) is illustrated in fig .
[ fig : barrier ] .
both the tda and the osd are single - spin - flip ( nonconservative ) dynamics that satisfy detailed balance .
this ensures the approach to equilibrium , which in this case is a uniformly positive phase with the interface pushed off to positive infinity .
such dynamics are defined by a single - spin transition rate , @xmath32 . here
@xmath33 is the inverse of the temperature @xmath34 ( boltzmann s constant is taken as unity ) , @xmath35 is the energy change corresponding to a successful spin flip , and @xmath31 determines the energy barrier between the two states through eq .
( [ eq : tener ] ) . the detailed - balance condition ( valid for transitions between allowed states )
is expressed as @xmath36 , where the right - hand side is independent of @xmath31 . the transition rates for the tda and the osd dynamics with the transition - state energy @xmath37 given by eq .
( [ eq : tener ] ) are defined as@xcite @xmath38}\frac{1}{1+\exp[\beta(e_f - e_t ) ] } \label{eq : wtda}\ ] ] and @xcite @xmath39=\exp(-\beta u)\exp[-\beta \delta e/2 ] , \label{eq : wosd}\ ] ] respectively .
notice that the tda transition rate can not be factorized into one part that depends only on the interaction energy and another that depends only on the applied field ; thus it belongs to the class of dynamics defined as hard.@xcite the osd dynamics can be factorized this way and thus is classified as soft.@xcite another important difference between the tda and the osd is that @xmath40 is restricted to be between @xmath41 and @xmath42 , while there is no upper bound on @xmath43 . as a consequence ,
the same difference is observed for the propagation velocities .
one may however question the physical realism of cases in which the transition - state energy @xmath37 is below the initial energy @xmath29 so that @xmath44 . in order to preserve the sos configuration at all times ,
flips are allowed only at sites which have exactly one broken bond in the @xmath5 direction . with the ising hamiltonian
there are only a finite number of different values of @xmath45 .
the spins can therefore be divided into classes,@xcite labeled by the spin value @xmath46 and the number of broken bonds between the spin and its nearest neighbors in the @xmath4 and @xmath5 directions , @xmath47 and @xmath48 , respectively .
the ten spin classes consistent with the sos model are denoted @xmath49 with @xmath50 and @xmath51 .
they are shown in fig .
[ fig : pict ] and listed in table [ table : class ] . in the sos model
the heights of the individual steps are assumed to be statistically independent and identically distributed .
the step - height probability density function ( pdf ) is given by the interaction energy corresponding to the @xmath52 broken @xmath2-bonds between spins in the columns centered at @xmath53 and @xmath54 as @xmath55 = z(\phi)^{-1 } x^{|\delta(x)| } \ e^ { \gamma(\phi ) \delta(x ) } \;.
\label{eq : step_pdf}\ ] ] the factor @xmath56 determines the width of the pdf , and @xmath57 is a lagrange multiplier which maintains the mean step height at an @xmath4-independent value , @xmath58 , where @xmath59 is the overall angle between the interface and the @xmath4 axis . in equilibrium , @xmath56 is simply the boltzmann factor , @xmath60 .
the partition function is , @xmath61 where @xmath57 is given by @xmath62^{1/2 } } { 2 x \left ( 1 + \tan \phi \right ) } \label{eq : chgam}\ ] ] ( see details in refs . ) .
simple results are obtained for @xmath63 , which yields @xmath64 and @xmath65 and for @xmath66 ( see ref . ) .
the mean spin - class populations , @xmath67 , are all obtained from the product of the independent pdfs for @xmath68 and @xmath69 + 1 ) .
symmetry of @xmath70 $ ] under the transformation @xmath71 ensures that @xmath72 for all @xmath47 and @xmath48 .
numerical results illustrating the breakdown of this up / down symmetry for large @xmath9 are discussed in sec .
[ sec : mc ] .
the general expressions for the class populations are given in the first column of table [ table : class2 ] ; details of the calculation can be found in ref . .
the results for each of the dynamics are obtained by substituting their respective values of @xmath56 , which will be calculated in the next section . whenever a spin at the interface flips from @xmath111 to + 1 , the corresponding column of the interface advances by one lattice constant in the @xmath5 direction .
conversely , a column ( not necessarily the same one ) recedes by one lattice constant when a spin at the interface flips from + 1 to @xmath111 .
the mean velocity of the interface in the @xmath5-direction , @xmath73 , is the difference between the rates of forward and backward steps , averaged over the whole interface .
the energy changes corresponding to the flips are given in the third column in table [ table : class ] .
the sum over @xmath4 that gives the average velocity can be rearranged into a sum over the spin classes of transition rates , weighted by the average class populations .
since the spin - class populations on both sides of the interface are equal in this approximation , the contribution to @xmath73 from sites in the classes @xmath74 and @xmath75 therefore becomes @xmath76 the mean propagation velocity perpendicular to the interface becomes @xmath77 where the sum runs over the classes included in table [ table : class2 ] .
it has been shown in ref .
that eq .
( [ eq : totalv ] ) reduces to the results for the single - step@xcite and the polynuclear growth@xcite models at low temperatures for large and small @xmath59 , respectively .
with @xmath78 , i.e , independent of @xmath9 , the results in table [ table : class2 ] correspond to a linear - response approximation for the velocity . in previous papers@xcite
an expression for a field - dependent @xmath79 was obtained , based on a dynamic mean - field approximation for the equation of motion for the single - step pdf together with a detailed - balance argument for the stationary state .
this improved non - linear response approximation gives ( see ref . for details of the calculation ) , @xmath80 + e^{2\beta h}w[\beta(2h-4j_x),\beta u ] } { w[\beta(-2h-4j_x),\beta u ] + w[\beta(2h-4j_x),\beta u ] } \right\}^{1/2 } \ ; , \label{eq : xth}\ ] ] which is independent of @xmath57 . here
@xmath81 are the transition rates associated with the reversal of a single spin .
the values of @xmath45 , @xmath82 , and @xmath83 are given in table [ table : class ] for the different spin classes .
equation ( [ eq : xth ] ) shows that @xmath79 depends on the specific dynamics , except for @xmath84 , where it reduces to its equilibrium value , @xmath85 .
it is easy to see from the equation that for soft dynamics , where the field and interaction contributions to the transition rates factorize , the @xmath9-dependence in eq .
( [ eq : xth ] ) cancels out . in ref .
it was demonstrated that the soft glauber dynamics yields an sos interface that is identical to the equilibrium sos interface at @xmath84 and the same temperature , regardless of the value of @xmath9 .
hard dynamics , such as the standard glauber and metropolis dynamics and the tda , lead to a nontrivial field dependence in @xmath56 . inserting the transition rates corresponding to the tda and the osd dynamics , defined by eq .
( [ eq : wtda ] ) and eq .
( [ eq : wosd ] ) , respectively , into eq .
( [ eq : xth ] ) , we explicitly get @xmath86 and @xmath87 we note that @xmath88 is similar , but not identical , to the one for the standard glauber dynamics , eq . ( 18 ) of ref . .
the spin - class populations listed in table [ table : class2 ] can now be calculated explicitly for each of the dynamics by replacing @xmath56 with its corresponding value .
the expressions for the contributions to the mean velocity in the @xmath5 direction , eq .
( [ eq : generalv ] ) , for each class in the tda and the osd dynamics are given in the third and fourth columns of table [ table : class2 ] , respectively .
in the next section we show that the nonlinear - response approximation gives good agreement with mc simulations of driven , flat sos interfaces evolving under the tda and osd dynamics for a wide range of fields and temperatures .
we calculated the step - height distributions , propagation velocities , and spin - class populations , analytically and by kinetic mc simulations , for both the tda and the osd dynamics in the isotropic case , @xmath89 .
the details of our particular implementation of the @xmath90-fold way rejection - free mc algorithm@xcite are essentially the same as described in ref . , except for two points .
the first is that only transitions from the classes with one broken @xmath5-bond ( @xmath91 ) are allowed , so as to preserve the sos interface structure .
the second difference is that the present code uses _
continuous time_@xcite to accommodate the large transition rates that are possible with the osd dynamics . by keeping only the interface sites in memory ,
the algorithm is not subject to any size restriction in the @xmath5 direction , and simulations can be carried out for arbitrarily long times .
the numerical results presented here are based on mc simulations mostly at the two temperatures , @xmath92 and 0.6@xmath93 ( @xmath94 is the critical temperature for the isotropic , square - lattice ising model@xcite ) , with @xmath95 and fixed @xmath59 between 0 and @xmath96 .
the microscopic transition barrier , @xmath31 ( see eq.(2 ) ) , is chosen to be @xmath97 .
this is the same value used in a previous study of nucleation with the osd and tda dynamics.@xcite from eq .
( [ eq : wosd ] ) it is clear that for the osd dynamics @xmath31 only appears in a temperature - dependent scaling factor in the transition rate .
it thus has no influence on the interface structure . for the tda dynamics , on the other hand ,
an increase in @xmath31 leads to a decrease in the local interface width and consequently in the propagation velocity .
also , the observed skewness increases , suggesting increasing short - range correlations between the step heights .
these results are discussed in detail in a separate paper.@xcite in order to ensure stationarity we ran the simulation for 50000 @xmath90-fold way updates per updatable spin ( ups ) before taking any measurements .
stationary class populations and interface velocities were averaged over 50000 ups .
for the stronger fields at @xmath98 we used ten times as many ups .
adequate statistics for one- and two - step pdfs were ensured by the large @xmath99 .
stationary single - step pdfs were obtained by mc simulation at @xmath98 and @xmath100 for @xmath63 and several values of @xmath9 . the simulation data and the theoretical results for @xmath101 $ ]
are shown in fig .
[ fig : pd_tda ] and fig .
[ fig : pd_1sd ] for the tda and the osd dynamics , respectively .
the theoretical results are calculated with eq .
( [ eq : step_pdf ] ) , with @xmath79 from eq .
( [ eq : xtda ] ) for the tda dynamics and from eq .
( [ eq : xosd ] ) for the osd dynamics .
for the tda dynamics , fig .
[ fig : pd_tda ] , the agreement is excellent at the higher temperature for all the values of @xmath9 analyzed ( up to @xmath102 ) .
however , at the lower temperature the agreement is not very good for fields above @xmath103 . for the osd dynamics , fig .
[ fig : pd_1sd ] shows that , contrary to the theoretical mean - field results , @xmath104 depends somewhat on @xmath9 .
this dependence is stronger for small fields and at the lower temperature .
however , although not absent as expected from the mean - field approximation , the field dependence is _ much weaker _ than for the tda dynamics .
another way to compare the analytical and simulation results is by calculating @xmath105 by summation of eq .
( [ eq : step_pdf ] ) , @xmath106 , with @xmath56 from eq .
( [ eq : xtda ] ) for the tda dynamics or @xmath56 from eq .
( [ eq : xosd ] ) ( independent of @xmath9 ) for the osd , and comparing these values with the simulated ones . the simulation values for @xmath105
can be obtained in two ways : directly by summation over the numerically obtained pdf , and also from the probability of zero step height as @xmath107^{-1 } - p[0 ] \right\}/2 $ ] .
this latter expression is obtained by observing that @xmath108 = ( 1-x)/(1+x)$ ] for @xmath63 , then solving this for @xmath56 , and inserting the result in the above equation for @xmath105 in terms of @xmath56 .
the results are shown in fig .
[ fig : dh ] for both dynamics , for @xmath63 at @xmath92 and 0.6@xmath93 calculated theoretically ( solid lines ) and by mc simulation ( symbols ) .
the agreement between the simulation and theoretical results for the tda is excellent except at the low temperature , where for intermediate fields a slight deviation can be seen .
the results are similar to those obtained with the standard glauber dynamics .
( see fig .
5(a ) of ref . .
) again the results for the osd dynamics show that , contrary to the theoretical prediction , there is a clear , albeit weak , dependence of the step height on the field .
the theoretical and simulation results for the osd dynamics only coincide at @xmath109 .
the difference between the two dynamics is evident : the step heights for the osd dynamics are weakly dependent on @xmath9 , particularly for low values of @xmath9 , in contrast with the very strong @xmath9 dependence obtained for the tda dynamics .
this behavior is typical for differences expected between soft and hard dynamics.@xcite in this section we compare the interface velocities obtained for the tda and the osd dynamics . in each case
the velocities are calculated with the analytical approximation , eq .
( [ eq : totalv ] ) , and by simulations . fig .
[ fig : vph](a ) and fig .
[ fig : vph](b ) show the normal velocity vs @xmath9 for @xmath63 for the tda and the osd dynamics , respectively .
there is excellent agreement between the mc results and the nonlinear - response theory for the tda dynamics , except for a slight disagreement seen at @xmath110 between @xmath111 and @xmath112 .
the results are very similar to those obtained with the standard glauber dynamics .
( see fig .
6 of ref . .
) however , for the osd dynamics at the lower temperature , the nonlinear - response approximation underestimates the velocity , especially at higher fields .
one of the main differences between the two dynamics is clearly seen in fig .
[ fig : vph ] : the velocity is bounded by unity for the tda while it increases exponentially with @xmath9 for the osd . the dependence of the normal velocity on the tilt angle @xmath59 is shown in fig .
[ fig : vang_tda ] and fig .
[ fig : vang_osd ] , for the tda and the osd dynamics , respectively .
we show results for several values of @xmath113 at @xmath98 and @xmath114 . for the tda dynamics
, the agreement between the theoretical results and the simulations is excellent .
the results are qualitatively similar to those obtained with the standard glauber dynamics .
( see fig .
7 of ref . .
) for the osd , the agreement between theory and simulation is excellent at the higher temperature .
however , at the lower temperature and higher fields the agreement is only good at large values of @xmath59 . in both cases
it is seen that at @xmath98 in weak fields the velocity increases with @xmath59 , in agreement with the polynuclear growth model at small angles and the single - step model for larger angles .
however , for strong fields the tda dynamics change gradually to the reverse anisotropy of eden - type models.@xcite no such change is observed for the osd dynamic . at @xmath114 ,
on the other hand , both models behave very similarly .
the velocity is nearly isotropic for weaker fields , while becoming eden - like for stronger fields . the temperature dependence of the normal interface velocity is shown in fig .
[ fig : vt ] for several values of @xmath113 . for the tda dynamics ,
[ fig : vt](a ) , the agreement between the simulations and the analytical results is excellent almost everywhere , except for a small discrepancy at intermediate @xmath34 and @xmath9 .
this discrepancy is expected from the results shown in fig .
[ fig : vph ] .
this figure shows that at low @xmath34 , the velocity changes steeply from zero to unity at some value between @xmath115 and @xmath103 , developing a step discontinuity in @xmath9 at @xmath116 .
the results are qualitatively similar to those obtained with the standard glauber dynamics .
( see fig .
8 of ref . .
) for the osd , fig .
[ fig : vt](b ) , the theoretical and simulation results agree only when the temperature is higher than a minimum value that increases as the field increases . in this case
the velocity also has a step discontinuity at @xmath116 : for @xmath113 below @xmath117 the velocity goes to zero . for stronger fields the velocity at @xmath116 increases dramatically with @xmath9 and decreases very rapidly as the temperature increases .
as can be seen from eq .
( [ eq : totalv ] ) and table [ table : class2 ] , the mean - field theory predicts that the contributions to the propagation velocity from each of the classes at low @xmath34 is proportional to @xmath118 $ ] .
thus , there is a discontinuity at @xmath116 for @xmath119 , i.e. , for @xmath120 for our selection of @xmath121 . beyond this value of @xmath9 , the velocity diverges to infinity as @xmath122 .
such a divergence at @xmath116 is not present for the soft glauber dynamics , where the velocity vanishes at @xmath116 for all values of @xmath9.@xcite since the analytical predictions for the class populations are based on the assumption that different steps are statistically independent , a comparison with the simulation results gives a way of testing this assumption .
the six mean class populations
@xmath123 , @xmath124 , and @xmath125 with @xmath126 for @xmath63 at @xmath98 and @xmath100 are shown vs @xmath9 in fig .
[ fig : pop_tda ] and fig .
[ fig : pop_1sd ] for the tda and the osd dynamics , respectively . for the tda at both temperatures , fig .
[ fig : pop_tda ] , the analytical approximations follow the average of the populations for @xmath14 and @xmath16 quite well
. however , at intermediate fields the simulations show that the population in front of the surface ( @xmath127 ) is quite different from the one behind it ( @xmath14 ) .
the mean - field approximation seems to reproduce better the population behind the surface .
the results are qualitatively similar to those obtained with the standard glauber dynamics .
( see fig .
9 of ref . .
) for the osd , fig .
[ fig : pop_1sd ] , the mean - field approximation predicts that the mean class populations should be independent of @xmath9 ( since @xmath56 is independent of @xmath9 ) , while the simulation indicates a weak @xmath9-dependence . for small fields
there is a clear dependence of the population on the field , but as the field increases , the populations tend to fixed values independent of the field .
the @xmath9-dependence is consistent with the results for @xmath105 and @xmath79 , shown in fig .
[ fig : dh](b ) .
the short - range correlations between neighboring steps are responsible for the skewness between the spin populations on the leading and trailing edges of the interface that appears in the simulation results .
this phenomenon is commonly observed in driven interfaces .
it occurs even when the _ long - range _ correlations vanish as they do for interfaces in the kpz dynamic universality class , to which the present models belong for all finite , nonzero values of @xmath9 .
skewness has also been observed in several other sos - type models , such as the body - centered sos model studied by neergaard and den nijs,@xcite the model for step propagation on crystal surfaces with a kink - ehrlich - schwoebel barrier studied by pierre - louis et al.,@xcite and a model for the local time horizon in parallel kinetic mc simulations studied by korniss et al.@xcite no skewness was observed for the sos model with the soft glauber dynamics.@xcite however , a small skewness was observed for the ising model ( whose interfaces include bubbles and overhangs ) with soft glauber dynamics ( about two orders of magnitude smaller than the skewness observed for the hard glauber dynamics).@xcite the correlations associated with the skewness generally lead to a broadening of protrusions on the leading edge ( `` hilltops '' ) , while those on the trailing edge ( `` valley bottoms '' ) are sharpened,@xcite or the other way around.@xcite in terms of spin - class populations , the former corresponds to @xmath128 and @xmath129 . the relative skewness can therefore be quantified by the two functions,@xcite @xmath130 and@xcite @xmath131 these two skewness parameters are shown together in fig . [ fig : skew](a ) for the tda , and in fig .
[ fig : skew](b ) for the osd . for both dynamics
the relative skewness is seen to be considerably stronger at the lower temperature .
the temperature dependence is especially pronounced for @xmath132 , due to the low concentration of sites in the class 21 + at low temperatures .
the osd results contrast with previous results that showed that an sos interface with soft glauber dynamics does not present skewness.@xcite in the present case , the sos interface evolving under the soft osd dynamics presents stronger relative skewness at lower temperature than the one evolving under the hard tda dynamics .
another way to visualize the skewness is to consider the joint two - step pdf , @xmath133 $ ] .
logarithmic contour plots of this quantity for both dynamics at different values of @xmath9 , for @xmath134 at @xmath114 , are shown in fig .
[ fig : cont_tda ] and fig .
[ fig : cont_1sd ] .
it is clearly seen that in both dynamics the contours change with @xmath9 . for @xmath9=0
a symmetric diamond shape with equidistant contours indicates statistical independence with single - step pdfs given by eq .
( [ eq : step_pdf ] ) .
this equilibrium result is correctly observed in both the tda dynamics , fig .
[ fig : cont_tda](a ) , and the osd dynamics , fig .
[ fig : cont_1sd](a ) .
however , for nonzero fields the dynamics show different behavior . for the tda , at stronger fields , the shapes are convex in the second quadrant [ @xmath135 fig .
[ fig : cont_tda](c ) and ( d ) .
this shape indicates that large negative @xmath68 tend to be followed by large positive @xmath136 ( sharp valleys ) . for weak fields ,
the tda shows a weak skewness of the opposite sign , as seen in fig .
[ fig : cont_tda](b ) . for the osd dynamics , for all nonzero fields , the shapes are concave in the second quadrant , fig .
[ fig : cont_1sd](b ) , ( c ) , and ( d ) .
this shape indicates that large negative @xmath68 tend to be followed by smaller positive @xmath136 ( rounded valleys ) . for both dynamics the field dependence in the fourth quadrant , corresponding to the shape of `` hilltops
, '' is much weaker that in the second quadrant .
as expected , the contour plots for interfaces with @xmath134 are always symmetric about the line @xmath137 for both dynamics .
in this work we have continued our studies of the microstructure of an unrestricted sos interface driven far from equilibrium by an applied field .
previous studies indicate that different dynamics can lead to important differences in the microstructure of the moving interface and that extreme care therefore must be taken in selecting stochastic dynamics appropriate for the specific physical system of interest.@xcite for this study we have considered two dynamics that include a local energy barrier representing a transition state inserted between the individual ising or lattice - gas states .
such arrhenius dynamics , as they are often called , are appropriate in kinetic mc simulations of discrete ising or lattice - gas models in which the discrete states serve as approximations for high - probability configurations in an underlying continuous potential .
the two arrhenius dynamics that we considered are the commonly used one - step - dynamics ( osd)@xcite and the two - step transition dynamics approximation ( tda).@xcite the osd belongs to the class of dynamics known as soft and the tda to the class known as hard.@xcite we studied the microstructure and velocity of the sos interface by kinetic mc simulations and by a non - linear mean - field theory developed in previous papers.@xcite we calculated the interface velocity as a function of the driving field , temperature , and angle of the interface relative to the lattice axes .
we also studied the local shape of the interface in terms of the spin - class populations , the average height of a step , and the probability density for individual steps in the interface .
the theory predicts significant differences between interfaces moving under hard and soft dynamics . for soft dynamics
, interface structures should remain independent of the applied field , while there must be a clear dependence on the field for hard dynamics .
for the tda dynamics we found generally very good agreement between the theoretical predictions and the mc simulations .
however , for the osd dynamics we found that , contrary to the theoretical prediction , there is a weak but clear dependence of the interface structure on the field .
this dependence is manifest in the average stationary step height and in the mean stationary class populations , which both show a weak dependence with the field that saturates for strong fields . as a consequence of this dependence
, the theoretical results for the velocities do not match very closely the simulated results for low temperatures , particularly for strong fields and small angles . the interfaces moving under the hard tda dynamics present similar characteristics to those evolving under a hard glauber dynamics .
however , we found significant differences between the structure of surfaces evolving under the soft osd dynamics and the soft glauber dynamics studied in ref . .
the velocities under the osd dynamics present a discontinuity at @xmath116 that is not observed for the soft glauber dynamics .
more interesting is the existence of strong skewness in the osd model .
this indicates that lack of skewness is not a necessary characteristic of soft dynamics , as earlier results seemed to suggest.@xcite within the mean - field approximation used here , individual steps of the interface are assumed to be statistically independent .
short - range correlations are not taken into account by this approximation .
the skewness between the spin population on the leading and trailing edges of the interface is a consequence of such short - range correlations .
for both the tda and the osd dynamics , the interfaces undergo a gradual breakdown of up / down symmetry for increasing fields , which has also been observed in other examples of driven interfaces.@xcite this breakdown is clearly evident for the osd model , and it is probably the reason why the mean - field approximation misses the weak field dependence in the osd interface structure .
it is obviously important to note and eventually to understand the discrepancies between the theoretical mean - field predictions and the simulation results for the osd dynamics .
however , on a qualitative level the theory predicts quite accurately the differences between the structures generated by the two dynamics . the average step height for the tda increases dramatically with increasing field , as accurately predicted by the theory .
for the osd , the step height does increase somewhat with @xmath9 , in contrast to the theoretical prediction of @xmath9-independence , but the increase is very small . in comparison with the tda , the osd surface remains very smooth , with an average step height well below unity .
as in previous studies , our results indicate strong differences between interfaces moving under different dynamics , emphasizing the need for extreme care in selecting the appropriate dynamics for the physical system of interest . even in cases where soft dynamics are the more appropriate choice , as for solidification or adsorption problems where the driving force is a chemical - potential difference,@xcite the results can depend significantly on which soft dynamics are chosen .
g. m. b. appreciates the hospitality of the school of computational science at florida state university .
this research was supported in part by national science foundation grant nos .
dmr-0240078 and dmr-0444051 , by florida state university through the center for materials research and technology and the school of computational science , by the national high magnetic field laboratory , and by the deanship of research and development of universidad simn bolvar . a .-
barabsi and h. e. stanley , _ fractal concepts in surface growth _
( cambridge u. press , cambridge , 1995 ) , detailed discussion and copious references to the theoretical and experimental literature can be found here . .
the spin classes in the anisotropic square - lattice sos model .
the first column contains the class labels , @xmath49 .
the second column contains the total field and interaction energy for a spin in each class , @xmath138 , relative to the energy of the state with all spins parallel and @xmath84 , @xmath139 .
the third column contains the change in the total system energy resulting from reversal of a spin from @xmath46 to @xmath140 , @xmath141 .
the fourth and fifth columns contain @xmath82 and @xmath83 , respectively .
the first three classes have nonzero populations in the sos model , and flipping a spin in any of them preserves the sos configuration .
the other two classes ( marked @xmath142 ) also have nonzero populations in the sos model , but flipping a spin in any of them would produce an overhang or a bubble and is therefore forbidden . not that in this table , @xmath46 represents the spin value _ before _ the spin flip .
[ cols="<,<,<,<,<",options="header " , ] between a positively magnetized phase ( or `` solid '' phase in the lattice - gas picture ) below and a negative ( or `` fluid '' ) phase above .
the step heights are @xmath25 .
interface sites representative of the different sos spin classes ( see table [ table : class ] and table [ table : class2 ] ) are marked with the notation @xmath49 explained in the text .
sites in the uniform bulk phases are @xmath143 and @xmath144 .
this interface was randomly generated with a symmetric step - height distribution , corresponding to @xmath63 . from ref . .
] , for the values of @xmath113 given in the legend .
( a ) @xmath98 .
( b ) @xmath114 . the symbols ( and colors )
have the same interpretations in ( a ) and ( b ) . , title="fig : " ] , for the values of @xmath113 given in the legend .
( a ) @xmath98 .
( b ) @xmath114 . the symbols ( and colors )
have the same interpretations in ( a ) and ( b ) .
, title="fig : " ] .
the data points indicate mc results and the straight lines are the theoretical predicted values ( independent of @xmath9 ) .
the values of @xmath113 are given in the legend .
( a ) @xmath98 .
( b ) @xmath114 .
the symbols and colors have the same interpretations in ( a ) and ( b ) .
note the very different scales from fig .
[ fig : pd_tda ] .
, title="fig : " ] .
the data points indicate mc results and the straight lines are the theoretical predicted values ( independent of @xmath9 ) .
the values of @xmath113 are given in the legend .
( a ) @xmath98 .
( b ) @xmath114 .
the symbols and colors have the same interpretations in ( a ) and ( b ) .
note the very different scales from fig .
[ fig : pd_tda ] . , title="fig : " ] vs @xmath9 for @xmath59=0 at @xmath34=@xmath110 and 0.6@xmath93 .
the curves represent the theoretical results .
the mc data were obtained directly by summation over the simulated single - step pdfs ( asterisks and crosses ) and from the probability of zero step height ( circles and squares ) .
see the text for details .
curve with circles and asterisks : @xmath98 .
curve with squares and crosses : @xmath114 .
( a ) tda dynamics , shown on a logarithmic vertical scale .
( b ) osd dynamics , shown on a linear vertical scale . in this and all the following figures , the statistical uncertainty is much smaller than the symbol size .
, title="fig : " ] vs @xmath9 for @xmath59=0 at @xmath34=@xmath110 and 0.6@xmath93 .
the curves represent the theoretical results .
the mc data were obtained directly by summation over the simulated single - step pdfs ( asterisks and crosses ) and from the probability of zero step height ( circles and squares ) .
see the text for details .
curve with circles and asterisks : @xmath98 .
curve with squares and crosses : @xmath114 .
( a ) tda dynamics , shown on a logarithmic vertical scale .
( b ) osd dynamics , shown on a linear vertical scale . in this and all the following figures , the statistical uncertainty is much smaller than the symbol size .
, title="fig : " ] vs @xmath9 for @xmath63 .
the mc results are shown as data points , circles for @xmath98 and squares for @xmath114 , and the theoretical results as solid curves .
( a ) tda , shown on a linear vertical scale .
( b ) osd , shown on a logarithmic vertical scale .
, title="fig : " ] vs @xmath9 for @xmath63 .
the mc results are shown as data points , circles for @xmath98 and squares for @xmath114 , and the theoretical results as solid curves .
( a ) tda , shown on a linear vertical scale .
( b ) osd , shown on a logarithmic vertical scale .
, title="fig : " ] vs @xmath145 , calculated with the tda dynamics , for @xmath146 ( up triangles ) , @xmath147 ( circles ) , @xmath42 ( squares ) , @xmath117 ( diamonds ) , and @xmath148 ( left triangles ) .
the symbols represent mc data , and the solid curves analytical results .
( a ) @xmath98 .
( b ) @xmath114 .
, title="fig : " ] vs @xmath145 , calculated with the tda dynamics , for @xmath146 ( up triangles ) , @xmath147 ( circles ) , @xmath42 ( squares ) , @xmath117 ( diamonds ) , and @xmath148 ( left triangles ) .
the symbols represent mc data , and the solid curves analytical results .
( a ) @xmath98 .
( b ) @xmath114 .
, title="fig : " ] vs @xmath145 , calculated with the osd dynamics , for @xmath146 ( up triangles ) , @xmath147 ( circles ) , @xmath42 ( squares ) , @xmath117 ( diamonds ) , and @xmath148 ( left triangles ) .
the symbols represent mc data , and the solid curves analytical results .
( a ) @xmath98 .
( b ) @xmath114 .
note the difference in the vertical scales .
, title="fig : " ] vs @xmath145 , calculated with the osd dynamics , for @xmath146 ( up triangles ) , @xmath147 ( circles ) , @xmath42 ( squares ) , @xmath117 ( diamonds ) , and @xmath148 ( left triangles ) . the symbols represent mc data , and the solid curves analytical results .
( a ) @xmath98 .
( b ) @xmath114 .
note the difference in the vertical scales .
, title="fig : " ] vs @xmath34 for @xmath63 and @xmath113 between @xmath147 and @xmath149 .
mc data are represented by data points and analytical results by solid curves . from below to above , the values of @xmath113 are 0.5 ( circles ) , 1.0 ( squares ) , 1.5 ( diamonds ) , 2.0 ( up triangles ) , 2.5 ( left triangles ) , 3.0 ( down triangles ) , and 3.5 ( right triangles ) .
online , the colors of the curves and symbols match .
( a ) tda , on a linear vertical scale .
( b ) osd , on a logarithmic vertical scale .
, title="fig : " ] vs @xmath34 for @xmath63 and @xmath113 between @xmath147 and @xmath149 .
mc data are represented by data points and analytical results by solid curves . from below to above , the values of @xmath113 are 0.5 ( circles ) , 1.0 ( squares ) , 1.5 ( diamonds ) , 2.0 ( up triangles ) , 2.5 ( left triangles ) , 3.0 ( down triangles ) , and 3.5 ( right triangles ) .
online , the colors of the curves and symbols match .
( a ) tda , on a linear vertical scale .
( b ) osd , on a logarithmic vertical scale .
, title="fig : " ] vs @xmath113 for @xmath63 , calculated for the tda dynamics .
the simulation results are indicated by symbols , and the analytic approximations by solid curves .
( a ) @xmath98 .
( b ) @xmath114 .
the symbols have the same interpretations in ( a ) and ( b ) .
note the different vertical scales in the two parts .
, title="fig : " ] vs @xmath113 for @xmath63 , calculated for the tda dynamics .
the simulation results are indicated by symbols , and the analytic approximations by solid curves .
( a ) @xmath98 .
( b ) @xmath114 .
the symbols have the same interpretations in ( a ) and ( b ) .
note the different vertical scales in the two parts .
, title="fig : " ] vs @xmath113 for @xmath63 calculated for the osd dynamics .
the simulation results are indicated by the symbols , and the straight lines indicate the theoretical predicted values ( independent of @xmath9 ) . ( a ) @xmath98 .
( b ) @xmath114 .
the symbols have the same interpretations in ( a ) and ( b ) .
note the different vertical scales in the two parts .
, title="fig : " ] vs @xmath113 for @xmath63 calculated for the osd dynamics .
the simulation results are indicated by the symbols , and the straight lines indicate the theoretical predicted values ( independent of @xmath9 ) . ( a ) @xmath98 .
( b ) @xmath114 .
the symbols have the same interpretations in ( a ) and ( b ) .
note the different vertical scales in the two parts .
, title="fig : " ] ( circles ) and @xmath150 ( squares ) , defined in eqs .
( [ eq : rho ] ) and ( [ eq : epsi ] ) , respectively .
the parameters are shown vs @xmath9 for @xmath63 , at @xmath98 ( empty symbols ) and at @xmath114 ( filled symbols ) .
( a ) tda .
( b ) osd .
note the different vertical scales in the two parts .
, title="fig : " ] ( circles ) and @xmath150 ( squares ) , defined in eqs .
( [ eq : rho ] ) and ( [ eq : epsi ] ) , respectively .
the parameters are shown vs @xmath9 for @xmath63 , at @xmath98 ( empty symbols ) and at @xmath114 ( filled symbols ) .
( a ) tda .
( b ) osd .
note the different vertical scales in the two parts .
, title="fig : " ] $ ] for @xmath134 at @xmath114 for the tda dynamics .
( a ) @xmath151 .
( b ) @xmath152 .
( c ) @xmath153 .
( d ) @xmath154 .
note the different scales in the four parts .
see discussion in the text .
, title="fig : " ] $ ] for @xmath134 at @xmath114 for the tda dynamics .
( a ) @xmath151 .
( b ) @xmath152 .
( c ) @xmath153 .
( d ) @xmath154 .
note the different scales in the four parts .
see discussion in the text .
, title="fig : " ] + $ ] for @xmath134 at @xmath114 for the tda dynamics .
( a ) @xmath151 .
( b ) @xmath152 .
( c ) @xmath153 .
( d ) @xmath154 .
note the different scales in the four parts .
see discussion in the text .
, title="fig : " ] $ ] for @xmath134 at @xmath114 for the tda dynamics .
( a ) @xmath151 .
( b ) @xmath152 .
( c ) @xmath153 .
( d ) @xmath154 .
note the different scales in the four parts .
see discussion in the text .
, title="fig : " ] $ ] for @xmath134 at @xmath114 for the osd dynamics .
( a ) @xmath151 .
( b ) @xmath152 .
( c ) @xmath153 .
( d ) @xmath155 .
see discussion in the text .
, title="fig : " ] $ ] for @xmath134 at @xmath114 for the osd dynamics .
( a ) @xmath151 .
( b ) @xmath152 .
( c ) @xmath153 .
( d ) @xmath155 .
see discussion in the text .
, title="fig : " ] + $ ] for @xmath134 at @xmath114 for the osd dynamics .
( a ) @xmath151 .
( b ) @xmath152 .
( c ) @xmath153 .
( d ) @xmath155 .
see discussion in the text .
, title="fig : " ] $ ] for @xmath134 at @xmath114 for the osd dynamics .
( a ) @xmath151 .
( b ) @xmath152 .
( c ) @xmath153 .
( d ) @xmath155 .
see discussion in the text .
, title="fig : " ] | we study the microscopic structure and the stationary propagation velocity of @xmath0-dimensional solid - on - solid interfaces in an ising lattice - gas model , which are driven far from equilibrium by an applied force , such as a magnetic field or a difference in ( electro)chemical potential .
we use an analytic nonlinear - response approximation [ p. a. rikvold and m. kolesik , j. stat.phys . *
100 * , 377 ( 2000 ) ] together with kinetic monte carlo simulations . here
we consider interfaces that move under arrhenius dynamics , which include a microscopic energy barrier between the allowed ising / lattice - gas states .
two different dynamics are studied : the standard one - step dynamics ( osd ) [ h. c. kang and w. weinberg , j. chem .
phys . * 90 * , 2824 ( 1992 ) ] and the two - step transition - dynamics approximation ( tda ) [ t.ala-nissila , j. kjoll , and s. c. ying , phys .
rev .
b * 46 * , 846 ( 1992 ) ] . in the osd
the effects of the applied force and the interaction energies in the model factorize in the transition rates ( soft dynamics ) , while in the tda such factorization is not possible ( hard dynamics ) . in full agreement with previous general theoretical results
we find that the local interface width under the tda increases dramatically with the applied force .
in contrast , the interface structure with the osd is only weakly influenced by the force , in qualitative agreement with the theoretical expectations .
results are also obtained for the force - dependence and anisotropy of the interface velocity , which also show differences in good agreement with the theoretical expectations for the differences between soft and hard dynamics .
our results confirm that different stochastic interface dynamics that all obey detailed balance and the same conservation laws nevertheless can lead to radically different interface responses to an applied force . |
the frontier field of cold ( @xmath0 k ) and ultracold ( @xmath0 mk ) gas - phase molecular physics has brought forth many innovations @xcite . among motivating challenges is the prospect of studying collision processes , especially chemical reactions , under `` matterwave '' conditions .
reaching that realm , where quantum phenomena become much more prominent than in ordinary `` warm '' collisions , requires attaining _ relative _ velocities so low that the debroglie wavelength becomes comparable to or longer than the size of the collision partners . that has been achieved recently for reactions of alkali atoms with alkali dimers formed from ultracold trapped alkali atoms by photoassociation or feshbach resonances @xcite . with the aim of widening the chemical scope
, much effort has been devoted to developing means to slow and cool preexisting molecules .
( compilations are given in @xcite . ) for chemical reactions , however , as yet it has not proved feasible to obtain sufficient yields at very low collision energies , using either trapped reactants or crossed molecular beams .
the major handicap in such experiments is that _ both _
reactants must contribute adequate flux with very low translational energy .
merged codirectional beams with closely matched velocities offer a way to obtain far higher intensity at very low _ relative _ collision energies , since then _ neither _ reactant needs to be particularly slow .
moreover , many molecular species not amenable for slowing techniques become available as reactants .
merged beams have been extensively used with ions and/or neutrals formed by charge transfer , to perform experiments at relative energies below 1 ev with beams having kev energies @xcite .
a key advantage is a kinematic feature that deamplifies contributions to the relative energy by velocity spreads in the parent beams . by virtue of precise control feasible with ions , the velocity spreads are also quite small , typically @xmath10.1% or less . for thermal molecular beams , such as we consider here , the spreads are usually @xmath110% or more . that enables attaining low relative collision energy , but much lower energies and improved resolution
can be obtained by narrowing the spreads to @xmath11% , which now appears feasible at an acceptable cost in intensity .
surprisingly , application of merged beams to low - energy collisions of neutral molecules has been long neglected .
we have come across only three previous , very brief suggestions @xcite .
our treatment accompanies experiments now underway at texas a&m university @xcite . in sec .
ii , in order to assess the major role of velocity spreads in merged beams , we evaluate the average relative kinetic energy , @xmath2 and its rms spread @xmath3 by integrating over velocity distributions familiar for molecular beams . reduced variable plots are provided that display the dependence on the ratio of most probable velocities in the merged beams , their velocity spreads , and merging angle . in secs .
iii and iv we discuss experimental prospects , limitations , and options .
for beams with lab speeds @xmath4 and @xmath5 intersecting at an angle @xmath6 , the relative kinetic energy is @xmath7 with @xmath8 the reduced mass . for merged beams it is feasible to restrict the angle @xmath6 to a small spread , fixed by geometry and typically only about a degree or so about @xmath9 . here we evaluate the average of @xmath10 over the beam velocity distributions , @xmath11\ ] ] and the rms spread , @xmath12^{1/2}.\ ] ] these require only @xmath13 , with k = 1 - 4 , for the individual beams .
we obtain analytic expressions for averages over velocity distributions for beams formed by effusive flow , by supersonic expansion , and by a rotating supersonic nozzle .
figure 1 illustrates these distributions . for the supersonic beams ,
three widths are shown ; the broadest ( @xmath1 10% ) is typical , the narrowest ( @xmath1 1% ) is near the best achieved in stark or zeeman molecular decelerators exploiting phase stability and transverse focusing @xcite .
results given here pertain to molecular flux distributions ; as noted in an appendix , they are readily adapted for number density distributions . the flux distribution , @xmath14\ ] ]
is governed by a single parameter , @xmath15 , with @xmath16 the boltzmann constant , and @xmath17 the source temperature .
the averaged powers of the velocity are @xmath18\ ] ] with @xmath19 and @xmath20 the gamma function . the most probable velocity , @xmath21 , and the rms velocity spread is @xmath22^{1/2 } = 0.483 $ ] .
the average beam kinetic energy is @xmath23 and the rms kinetic energy spread is @xmath24^{1/2 } = [ \gamma(4)-\gamma^2(3)]^{1/2 } = [ 6 - 4]^{1/2 } = 1.41\ ] ] the spread thus is comparable to the beam kinetic energy . for a merged pair of effusive beams , it is convenient to define @xmath25 , @xmath26 , with @xmath27 , and @xmath28 $ ] .
the relative kinetic energy is @xmath29 with c = cos@xmath30 and s = sin@xmath30 . the rms spread is @xmath31^{1/2}\ ] ] with @xmath32 @xmath33 @xmath34 figure 2 shows how @xmath2 and @xmath3 vary with the ratio of beam velocities , which is proportional to @xmath35 .
the curves given are for intersection angles near zero , pertinent for merged beams . for
matched beam velocities , with @xmath36 , both @xmath37 and @xmath3 are smallest .
there , for small @xmath6 , @xmath2 and @xmath3 are less than the nominal relative kinetic energy , @xmath38 , for a perpendicular collision ( @xmath39 ) by factors of only about 5 and 3 , respectively . as @xmath40 decreases below @xmath41 , @xmath37 and @xmath3 increase in nearly parallel fashion .
accordingly , whether or not the merged beam velocities are closely matched , @xmath3 exceeds @xmath2 appreciably ( by @xmath1 40% when @xmath36 ) .
the resolution of the relative kinetic energy hence is worse than seen in eqs .
( 6 ) and ( 7 ) for the beam kinetic energy itself .
the velocity distribution for effusive beams is the same as for a bulk gas .
thus , @xmath2 and @xmath3 for gas mixtures cooled by cryogenic means can be obtained from the merged beam results by merely setting @xmath42 , equivalent to integrating over all angles of collision . a standard approximation , @xmath43 ^ 2 \}\ ] ] for supersonic beams characterizes the velocity distribution by the flow velocity u along the centerline of the beam , and a width parameter @xmath44 , where @xmath45 , termed the parallel or longitudinal temperature , pertains to the molecular translational motion relative to the flow velocity @xcite . according to the thermal conduction model @xcite , @xmath46
is determined by the pressure within the source , @xmath47 , the nozzle diameter , @xmath48 , and the heat capacity ratio , @xmath49 .
likewise , the flow velocity is given by @xmath50^{1/2}[1 - ( t_{\|}/t_0)]^{1/2}\ ] ] analytic results for the velocity averages are readily obtained , @xmath51 with @xmath52 the ratio of velocity width to flow velocity , and @xmath53 ^ 2\}dt\ ] ] with @xmath54 . in the appendix
we give exact analytic formulas for the @xmath55 functions ; table i provides polynomial approximations ; for @xmath56 these are accurate to better than 0.03% .
the most probable velocity is @xmath57^{1/2}\ } \approx u(1 + x^2)$ ] .
the rms velocity spread is @xmath58^{1/2}$ ] .
the corresponding average beam kinetic energy is @xmath59 and the rms spread in kinetic energy is @xmath60^{1/2}\ ] ] for a pair of merged supersonic beams , the averaged relative kinetic energy is @xmath61 where @xmath62 .
the rms spread is @xmath63^{1/2}\ ] ] with @xmath64 + s^4\left[\langle v_2 ^ 4\rangle-\langle v_2 ^ 2\rangle^2\right]\ ] ] @xmath65\ ] ] @xmath66\ ] ] these expressions are akin to eqs .
( 8) and ( 9 ) , with @xmath41 and @xmath40 replaced by @xmath67 and @xmath68
. here @xmath69 , given by eq.(12 ) , may differ for the two beams ( @xmath70 = 1 , 2 ) if their velocity widths ( @xmath71 ) differ .
figure 3 displays the dependence of @xmath2 and @xmath3 on the ratio of flow velocities of the beams , @xmath72 , and their velocity widths .
curves are shown for @xmath73 and @xmath74 , to illustrate that the dependence on the merging angle is weak if the velocity widths of the beams are fairly large ( _ cf_. fig.2 ) but for @xmath37 becomes significant if the velocity widths are small ( _ cf_. fig . 6 below ) .
the dependence on the magnitude of the flow velocities is included simply by adopting units for @xmath37 and @xmath3 that compare them with the relative kinetic energy for collisions at about the most probable beam velocities ( nearly equal to @xmath67 , @xmath68 ) at right angles ( @xmath75 ) , or equivalently in a bulk gas .
we designate that by @xmath76 . as seen in panel ( a ) , if the merged beam velocities are precisely matched ( @xmath77 ; @xmath78 ) the collision kinetic energy @xmath2 is very sensitive to the velocity widths .
in contrast , when the beam flow velocities are unmatched by more than about 15% ( i.e. , @xmath79 ) , the ratio @xmath80 becomes nearly independent of the velocity widths and grows larger as the unmatch increases .
panel ( b ) shows that regardless of whether the beam velocities are matched or not , the spread in relative kinetic energy , @xmath3 , varies strongly with the velocity widths .
panel ( c ) plots the ratio @xmath81 , which defines the energy resolution . whereas for closely matched beams
@xmath2 is minimal , @xmath3 then approaches its maximal value .
indeed , for matched beams with @xmath82 , the resolution ratio , @xmath81 , is near @xmath83 ; that is just as poor as found in fig . 2 for effusive beams .
to improve the resolution ratio for matched beams requires narrowing the velocity widths .
table ii compares , both for the single beam and matched merged beams , effects of reducing the spread from @xmath84 = 0.1 to 0.01 . for the single beam ,
the change in average kinetic energy is very slight , whereas the rms spread in the kinetic energy shrinks tenfold . for the merged beams ,
the relative kinetic energy is lowered by a factor of 25 , and its rms spread by a factor of 100 ; so the resolution ratio is only improved fourfold . the resolution ratio can be lowered further if the merged beam velocities are unmatched , but that raises the averaged relative kinetic energy .
figure 4 displays the trade - offs involved . to obtain optimally low @xmath2 requires nearly exact matching ; that can provide @xmath85 for @xmath84 = 0.01 or @xmath86 for @xmath84 = 0.02 .
however , for exact matching , the resolution ratio is only @xmath87 = 0.35 for @xmath84 = 0.01 and surges to 0.8 for @xmath84 = 0.02 . to attain resolution of 0.2 or 0.3 even with @xmath84 = 0.01 requires @xmath88 = 0.91 or 0.94 and hence would increase @xmath80 to @xmath89 or @xmath90 , respectively .
the upshot is , to improve the resolution ratio by a factor of less than 2 ( from 0.35 to 0.2 ) by unmatching , requires increasing @xmath2 by a factor of 25 .
g. quemener , n. balakrishnan and a. dalgarno , _
inelastic collisions and chemical reactions of molecules at ultracold temperatures , in cold molecules : theory , experiment , applications _
, r. v. krems , w. c. stwalley , and b. friedrich , eds . , ( taylor and francis , london , 2009 ) , p. 69 . ,
for beams formed ( a ) by effusive flow , ( b ) supersonic expansion from a stationary source , and ( c ) a rotating supersonic source , defined by eqs.(4 ) , ( 10 ) , and ( 18 ) , respectively . parameters for ( a ) are @xmath92 ; @xmath93 m / s ; for ( b ) and ( c ) flow velocities are @xmath94 m / s and @xmath95 m / s and widths @xmath96 = 0.01 , 0.05 , 0.10 . for ( b ) and
( c ) the widths also influence somewhat the most probable velocity . ] .
the abscissa scale pertains to the ratio @xmath88 , which ranges from @xmath100 to @xmath77 .
results are shown for an intersection angle of @xmath73 and four sets of velocity spreads : @xmath101 = 0.01 ; 0.05 ; 0.1 ; and @xmath102 , @xmath103 . dashed curves included for the @xmath104 case
are for @xmath105 . ] , for relative kinetic energy , @xmath106 ( left ordinate scale ) and resolution ratio , @xmath81 ( right ordinate scale ) .
curves pertain to velocity spreads in beams of @xmath107 and 0.02 and merging angle of @xmath73 . ] and @xmath104 . both ( a ) and ( b ) exhibit pronounced minima where the matching condition holds : @xmath108 and hence @xmath109 .
full curves show results for slowing mode , with @xmath110 , @xmath111 , and @xmath112 ; dashed curves are for speeding mode , with @xmath113 , @xmath114 , and 2 . ] , for merged supersonic beams from stationary sources .
values of @xmath2 are indicated by black dots .
abscissa scale is in units of @xmath115 , as used in figs.3 - 5 .
( a ) for merging angle @xmath116 and various ratios of the flow velocities , @xmath117 to 0.85 and velocity widths @xmath107 .
note log - log plot is used .
( b ) for matched flow velocities , @xmath118 and @xmath119 , but merging angle varied to illustrate its role in eq.(20 ) .
( c ) for flow velocities unmatched by @xmath115% ( @xmath120 ) and @xmath104 with @xmath73 , @xmath74 , @xmath121 , to illustrate the reduced role of the merging angle when the first term in eq.(20 ) becomes predominant . ] | molecular collisions can be studied at very low relative kinetic energies , in the millikelvin range , by merging codirectional beams with much higher translational energies , extending even to the kilokelvin range , provided that the beam speeds can be closely matched .
this technique provides far more intensity and wider chemical scope than methods that require slowing both collision partners .
previously , at far higher energies , merged beams have been widely used with ions and/or neutrals formed by charge transfer . here
we assess for neutral , thermal molecular beams the range and resolution of collision energy that now appears attainable , determined chiefly by velocity spreads within the merged beams .
our treatment deals both with velocity distributions familiar for molecular beams formed by effusion or supersonic expansion , and an unorthodox variant produced by a rotating supersonic source capable of scanning the lab beam velocity over a wide range . |
the development of new growing techniques enables control of the potential responsible for confining the two - dimensional electron gas ( 2deg ) in a nanometer scale .
this potential influences the electronic transport properties of the 2deg and is a key feature that allows us to have systems with profiles appropriate for different purposes in the development of new electronic devices .
in addition , when a dc magnetic field is applied to a 2deg it modifies the effective potential felt by the electrons and can be responsible for externally controlled modifications of the electronic properties of the system .
it is expected that the electronic spin affects the transport properties of the system and a tilted or oscillating magnetic field modifies the state of the spin as well as its dynamical characteristics .
although the large number of papers dealing with this subject @xcite , we are still far from achieving a complete understanding of the role of the intrinsic magnetic moment in the transport phenomena .
it is supposed that the control of the transport properties via the spin will allow the substitution of the actual generation of electronics equipments by the spin - based devices @xcite that can be used , for example , for future applications in quantum computation .
the two - dimensional systems studied here are constituted by free electrons confined in a plane with a very narrow thickness compared to its lateral dimensions . in the real world "
these systems might be formed by inversion layers of gaas / algaas structures where the electrons are affected by a periodic potential ( that may be very weak ) as well as spin - orbit interaction @xcite .
these interactions are important in order to fully describe the system , but their main effect is to mix the free electron states ( they couple the landau - level subbands ) . in other words , these interactions will introduce corrections but they will not interfere with the main characteristics of the physical properties studied here .
in this paper we study the effect of a tilted magnetic field on the electrical transport properties of a free 2deg .
we name the two - dimensional region where the electrons are confined the @xmath0 plane , and we set the magnetic field in the @xmath1 plane . in section ii we develop the formalism necessary to describe a two - dimensional electronic system in presence of a tilted magnetic field , taking into account the orbital and spin variables .
since the fermi energy depends on the energy and degeneracy of the states and is one of the key quantities to understand the transport properties of any system , the unique behavior of the landau - level subbands with the spin and field configuration suggests that this quantity should be carefully analyzed .
this is done in section iii where we calculate and discuss the dependence of this quantity with the strength and direction of the dc magnetic field . in order to estimate the influence of the combined effect of the spin and field configuration on the transport properties of 2deg , in section
iv we obtain numerical results for the hall conductivity of low electronic density systems .
the conclusions and general comments are presented in section v.
a noninteracting two - dimensional electronic system may describe a moderate or low electronic density system observed at the interfaces of semiconductors and materials with a wider bandgap such as ingaas / inalas or algaas / gaas . in these structures
the electrons are confined in a narrow quantum well that can be less than @xmath2 @xmath3wide . in these cases
we may write the @xmath4 dependence of the wave function as a gaussian function of half - width @xmath5 equal to the thickness of the system . roughly speaking
@xmath5 is the width of the quantum well that confines electrons in the @xmath0 plane , and here we take @xmath6 @xmath3 for all numerical purposes .
the procedure described above gives a result similar to those obtained many years ago by several authors @xcite for spin independent solutions of this system .
here we go further and we analyze the contribution of the spin ( via zeeman interaction ) . as will be seen , in presence of tilted magnetic fields the spin not only split the levels , but
also produces a singular mixture of the landau - levels subbands with significative consequences on the transport properties of the system .
the one electron hamiltonian for this system is writen as : @xmath7 where the first term is the interaction between the particle s spin and the magnetic field , and @xmath8 denotes the pauli matrices .
the second term is the electron kinetic energy with effective mass @xmath9 and charge @xmath10 in the presence of a dc magnetic field @xmath11 .
the magnetic field is just @xmath12 , where the vector potential is @xmath13 .
we have assumed the @xmath4 dependence of the wave functions as a gaussian function and then the effective hamiltonian ( the total hamiltonian averaged in the @xmath4 direction ) may be written as : @xmath14 where @xmath15 . since @xmath16 is a constant ( it commutes with @xmath17 ) and @xmath18 ( @xmath19 ) is also a constant , a little algebra reduces the search for the one electron eigenstates @xmath20 to the search of @xmath21 and @xmath22 that are solutions of @xmath23 from eq .
( [ lab02a ] ) we find the eigenstates and eigenenergies given by @xmath24 , @xmath25 respectively . here
@xmath26 are the eigenstates of @xmath27 ( @xmath28 ) , @xmath29 and @xmath30 .
the solutions of eq .
( 4 ) are landau like states given by : @xmath31 and then , the eigenstate @xmath32 might be represented by the eigenfunction : @xmath33 \label{lab022c}\end{gathered}\ ] ] where @xmath34 is the length of the system along the @xmath35-direction , and @xmath36 ^ 2}}{\sqrt { \pi ^{1/2 } 2^n n!l_c } } { h_n [ ( y - y')/l_c]}\ ] ] is the one - particle harmonic oscillator function of frequency @xmath37 centered at @xmath38 , and @xmath39 is the landau - index state . in figure 1
we display the behavior of the energy of the eigenstates @xmath40 for @xmath41 .
the graphic shows the modification introduced by the parallel component of the magnetic field . as usual
, the landau - index gives the information about the orbital behavior of the system , while the spin ( which has its state controlled by the magnetic field ) is responsible for the shift of each level .
as can be seen in this figure , when the parallel component increases , at some point there is an inversion in the sequence of the energies of the states and higher landau levels become lower energy states . this behavior will give a unique characteristic to these systems since it will be responsible for an externally controlled modification of the fermi energy , as well as , the increasing of the degeneracy of the energy states with consequences on the transport properties of the system .
against the angle between the magnetic field and the @xmath4-direction .
the straight ( dashed ) lines are for @xmath42 ( @xmath43 ) , and n=1 is the line at the bottom.,title="fig:",width=245,height=188 ]
as mentioned before , a moderate or low electronic density allows us to use the one particle approximation to obtain information about the system , and then , a standard calculation can be used to obtain the fermi energy of the system .
the density of state is given by @xmath44 and we may assume a gaussian broadening of width @xmath45 to write @xmath46 as @xmath47 here @xmath48 ( @xmath49 is the length of the system along the @xmath50-direction ) , and we assume that the gaussian has the same width for all energy levels @xmath51 ( we use @xmath52 @xmath53 and @xmath54 @xmath55 in all numerical results presented here ) . since the density @xmath56 of particles ( electrons ) is given by @xmath57 with @xmath58^ { - 1}$ ] the fermi - dirac function , a straight calculation gives @xmath59 } = 0 \label{eqfermi}\ ] ] where @xmath60 is the error function .
for a given electronic density the fermi energy is the value of @xmath61 that satisfies the eq.([eqfermi ] ) . in figures 2a and 2b
we show the behavior of the fermi energy against the angle between the dc magnetic field and the @xmath4-direction for @xmath62 and @xmath63 respectively .
the oscillations observed are a consequence of inversions in the sequence of the energy states , since all of them converge to @xmath64 when @xmath65 is very small .
as the fermi energy is intimately related to the transport properties of the system , this behavior has a strong influence on this property . in figure 3 ,
the direction of the magnetic field is fixed and we let its intensity to vary . the jumps ( discontinuity ) observed are the result of the increasing of the splitting of the states energy . and the @xmath4-direction for two specific values of the electron density ; ( a)@xmath66 , ( b)@xmath67.,title="fig:",width=245,height=188 ] as a function of the intensity of a dc magnetic field applied parallel to the @xmath4-direction.,title="fig:",width=245,height=188 ]
when an electric field produces a carrier current in the presence of a non collinear magnetic field , there is always a transversal current as a result of the magnetic field .
this is the hall effect .
since the conductivity observed experimentally incorporates several properties of the sample , this effect is used as a tool to study several physical properties and to characterize several materials . here
, the hall conductivity is used to estimate the influence of a tilt dc magnetic field applied on the 2deg .
we use the results obtained in sections ii and iii to calculate the hall conductivity , which ( within the one particle approximation ) is given by ( see for example refs .
4 ) : @xmath68 \times \nonumber \\ & & \left[\frac{1-e^{(e_{\zeta}-e_{\zeta ' } ) /{k_bt}}}{(e_{\zeta}-e_{\zeta ' } ) ^2}\right]\langle\zeta |\widehat{v}_x | \zeta ' \rangle \langle \zeta'| \widehat{v}_y | \zeta\rangle \label{lab8}\end{aligned}\ ] ] with @xmath69 @xmath70 @xmath71 , where @xmath69 is the state @xmath72 . in the above equation ,
@xmath73 is the area of the system ( we use an unitary area in our numerical calculation ) , @xmath74 is the fermi - dirac function and @xmath75\times\nonumber\\ & & { \omega _ c l_c \cos { \theta } } \delta _ { k_x , k_x ' } \delta _ { \sigma , \sigma ' } \end{aligned}\ ] ] and @xmath76\times\nonumber\\ & & { \omega _ c l_c \cos { \theta}}\delta _ { k_x , k_x ' } \delta _ { \sigma , \sigma ' } \end{aligned}\ ] ] are the matrix element of the @xmath35 and @xmath77 components of the velocity operator . numerical results are depicted in figures 4 and 5 for @xmath78 . dc magnetic field and the @xmath4-direction ; a ) density @xmath66 ; b ) density @xmath67.,title="fig:",width=245,height=188 ] as a function of the intensity of a dc magnetic field applied parallel to the @xmath4-direction . , title="fig:",width=245,height=188 ]
in figure 2 we show the fermi energy as a function of the angle between the magnetic field and the @xmath4-direction .
the behavior of the fermi energy is a direct consequence of the behavior of the energy of the eigenstates depicted in figure 1 .
it is interesting to point out that when @xmath79 increases , the energies of all states go to @xmath80 and , as a consequence , there is an increasing of the number of states per unit of energy .
the increasing of the electronic density enhances this behavior because the electrons are accommodated " in a larger number of states and then there will be a bigger mix of populated states . also , for finite values of @xmath45 , there will be a mix of the gaussian associated with the neighbor energy levels . when the magnetic field increases the degeneracy of the states also increases , and there will be values of @xmath81 where these states are disconnected " resulting ( at these points ) in a singular behavior for the fermi energy
. it should be said that if @xmath45 is changed , the shape of the curves are different but the general behavior is preserved . for electronic density that has the fermi energy located at the landau energy larger than a given @xmath82 but much smaller than @xmath83
the increasing of its value with the external field has a smaller derivative than the case that the fermi energy is bellow @xmath82 , but much bigger than @xmath84 .
this is the main reason to the oscillatory behavior observed when the field is modified .
the hall conductivity also reproduce this result , and it has well defined terraces " with length related to the distance " of two consecutive minima ( or discontinuity ) of the fermi energy .
this behavior is better noticed when the density of electrons increases and stronger oscillations are observed for the fermi energy ( see figures 2 and 4 ) .
however , the length of the terrace decreases because the distance between the local minima is smaller .
our results show that the conductivity decreases when the magnetic field increases ( see figure 5 ) while the parallel component of the field provides a way to increase it . in both cases
we can see well defined plateaus with length related to the fermi energy behavior . in summary
, our calculation shows that the fermi energy is strongly dependent on the spin states , which is controlled by the external dc field .
moreover , any property that depends on the fermi - dirac distribution , ( which is dependent on the fermi energy ) will have the influence of these spin states .
we use the hall conductivity to show this fact .
we are now working on a more complex system to demonstrate that the spin states may be used to control the transport properties of selected materials .
this might be the reason for the behavior of the resistivity recently observed by k. vakili et al.@xcite .
f. f. fang and p. j. stiles , phys .
review , 174 , 823 , ( 1968 ) .
f. m. peeters and p. vasilopoulos , phys .
b , 46 , 4667 , ( 1992 ) .
x. f. wang and p. vasilopoulos , phys .
b , 67 , 085313 , ( 2003 ) .
x. f. wang , p. vasilopoulos and f. m. peeters , phys .
b , 71 , 125301 , ( 2005 ) . s. dattas and b. das , appl .
lett . , 56 , 665 ( 1990 ) ; x. f. wang , p. vasilopoulos and f. m. peeters , phys
b , 65 , 165217 ( 2002 ) . b. spivak and s.a .
kivelson , annals of physics 321 , 2071 ( 2006 ) and references therein .
p. vasilopoulos , phys .
review b , 32 , 771 ( 1985 ) and x. f. wang , p. vasilopoulos and f. m. peeters , phys .
b , 65 , 165217 , ( 2002 ) .
h. a. engel , b. i. halperin , and e. i. rashba , phys .
95 , 166605 ( 2005 ) ; d. d. awschalom and j. m. kikkawa , phys . today , 52(6 ) , 33 ( 1999 ) .
e. tutuc , e. p. de poortere , s. j. papadakis , and m. shayegan , phys .
lett . , 86 , 2858 ( 2001 ) .
j. sinova , d. culcer , q. niu , n. a. sinitsyn , t. jungwirth and a. h. macdonald , phys .
lett , 92 , 126603 ( 2004 ) .
k. vakili , y. p. shkolnikov , e. tutuc , n. c. bishop , e. p. de poortere , and m. shayegan , phys .
, 94 , 176402 ( 2005 ) .
s. k. bhattacharrya , phys . rev .
b , 25,3756 , ( 1982 ) ; e. j. pakulis , 31 , 6807 ( 1985 ) , and references therein . k. vakili , y. p. shkolnikov , e. tutuc , e. p.
de poortere , m. padmanabhan , and m. shayegan , appl .
89 ( 17 ) , 172118 ( 2006 ) . | the effect of a tilted dc magnetic field on the transport properties of a two dimensional electron gas ( 2deg ) is studied .
the influence of the component of magnetic field parallel to the surface is analyzed and the dependence of the fermi energy on the intensity and direction of the external field is discussed .
numerical results are obtained for the hall conductivity considering values of electronic densities and strength of the dc magnetic field that are currently obtained in many laboratories . |
multi core chip architectures are emerging as feasible solution to effectively utilizing the ever growing number of chip .
multi - core chip depends on success in system software technology ( compiler and runtime system ) , in order to have thread level parallelism and utilizing on - chip concurrency . with multi - core processors ,
additional speedups can be achieved by the use of parallelism in data - independent tasks .
there is a gradual shift towards making current algorithms and design into parallel algorithms .
it is rather difficult to achieve lock free and low cache contention parallel algorithms .
+ in 70 s papers appeared ideas on parallel compilation of programming languages and parallel execution of programs were expected . in those papers discussions on parallel lexical analysis , syntactic analysis and code generation were done . with vlsi applications ,
prominent increase in research on parallel compilation is observed . +
a compiler contains different phases : lexical analyzer , syntactic analyzer , semantic analyzer and code generator .
parsing or syntax analysis is the phase of compiler which analyses the program code according to the language . after analysis , it converts the code into another formal representation which will act as input for succeeding phases of compiler .
+ complexity of software source code is increasing . an effort to compile large code base
is very time consumable .
@xcite describes two types of parsers : top down and bottom up parsers .
top down parsers have less power as compared to bottom up parser . lr ( k ) ,
slr ( k ) and lalr ( 1 ) are types of bottom up parsers . with more power bottom up parsers also requires more space and more time in parsing a string as compared to top down parsers .
most of the compiler compilers like yacc @xcite and bison @xcite creates lr ( 1 ) parsers and compilers like clang @xcite , mono c # compiler @xcite etc . uses lr(1 ) parsers .
so , it is evident that programming languages can be represented easily by lr ( 1 ) languages .
+ parsing is very time consuming phase of compiler . parsing different files in parallel
is not enough . as programming languages like c and c++ can includes different files
( using # include ) in a single file which results in generation of very long file . if we can parallel the parsing phase of single file , it will give performance benefits in compiling the source code .
many significant techniques are already proposed for making parallel parsers ( @xcite , @xcite , @xcite , @xcite , @xcite ) .
a parallel parsing for programming language is given by @xcite .
+ a block is a section of code which is grouped together . in a language , a block may contain class definition , member or method declaration .
another block could be a block of statements also called compound statement .
this block is usually associated with a function code or if statement or loop .
programming languages such as c , c++ , java , python use the concept of blocks heavily .
one of most important property of parsing blocks is that they all are independent of each other i.e. each block can be parsed independently of other block .
so , we could parse many blocks in a parallel fashion . in this paper , we will propose a technique to parse various blocks of the code in parallel .
it can also work as a block parallel incremental parser .
our parser is termed as block parallelized parser ( bpp , for short ) .
+ our technique of parallel parsing is based on incremental parsing .
an incremental parser is the one that parse only those portions of a program that have been modified .
whereas an ordinary parser must process the entire program when it is modified .
an incremental parser takes only the known set of changes done in a source file and updates its internal representation of source file which may be an abstract syntax tree . by building upon the previously parsed files , the incremental
parser avoids the wasteful re - parsing of entire source file where most of the cod remains unchanged .
+ bpp is based on the properties that an incremental parser can parse any part of a source code without the need of parsing the whole source code and different blocks in a source code can be parsed independently of other blocks . in bpp
these parts are blocks in a source code . using the property of incremental parser , bpp parse each of the blocks independently of other blocks .
each of these blocks are parsed in their own thread .
it can be easily seen that bpp follows a divide and conquer approach .
it divides the source into different blocks , parse each of them in parallel and at the end conquer these blocks . in our scheme
the conquer step does nothing except waiting for all the bpp threads to complete their operations .
+ there have been many works on incremental parsing [ incremental parsing references ] .
we choose to extend on the works of incremental jump shift reduce parser of @xcite .
bpp is derived from incremental jump shift reduce parser . in @xcite , authors defined incremental jump shift reduce parser for slr ( 1 ) languages only .
however , we decided to extend this parser to accept lr(1 ) language because lr(1 ) languages has more power than slr ( 1 ) and nearly all programming languages can be defined in the form of lr(1 ) grammars .
we define the incremental categories to be a statement containing a block like class definition or function definition or if statement or for loop statement .
then , we give a notion of first non - terminal symbols of a non - terminal symbol .
we used this notion to create partitions of a grammar such that a partition includes an incremental category and its first non - terminals .
we observed that this scheme gives us a very interesting property in incremental jump shift reduce parser .
we used this property to create our block parallelized parser .
+ whenever a start of the block symbol is encountered the parser will first check whether the current block is related to any incremental category and can it be parsed independently .
if bpp is able to find the incremental category for it , bpp will start parsing the block in parallel . in this paper we developed this bpp for lr(1 ) languages but we believe it can be easily extended to lr(k ) or can be easily converted to lalr ( 1 ) or slr ( 1 ) grammars .
we also see that no major changes were done to the current lr(1 ) parsing algorithm and hence , it should be easy to create an abstract syntax tree . this parser can also work as an incremental parallel parser which can parse different blocks in parallel .
moreover , it could be seen that there is no requirement of any thread synchronization to communicate between different threads of bpp each of which is parsing a block in parallel .
this is because no two blocks are related in any way for the purpose of parsing .
+ we compiled c # implementation using mono c # compiler 3.12.1 and executed the implementation using mono jit compiler 3.12.1 on machine running fedora 21 with linux kernel 3.19.3 with 6 gb ram and intel core i7 - 3610 cpu with hyperthreading enabled .
we found out that our technique showed 28% performance improvement in the case of including header files and 52% performance improvement in the case of excluding header files . + the following paper is designed as follows .
section 2 shows some previous work done in parallel parsing .
section 3 and 4 provides the terminology we will use and the background required to understand our technique . in section 5 we will extend incremental jump shift reduce parser to accept lr(1 ) grammars . in section 6
, we introduced the concept of first non terminals of a non terminal . in section 7
we will use this concept to create partitions of the grammar .
we also showed that by creating partitions using this concept we get a very interesting property .
this property would be used by bpp .
we have generalized this property in a theorem and also provided a proof for it . in section 8
we presents our block parallelized parser and its parsing algorithm . in section 9
we will compare our algorithm with previous work .
section 10 shows our evaluation and results .
in section 11 and section 12 we complete our document with conclusion and related work .
a lot of previous work has been done in parallel parsing of lr ( 1 ) and context free languages .
the most recent work done by @xcite in parallel parsing of lr(1 ) is an extension of an lr substring parser for bounded context languages ( developed by cormack ) for parallel environment .
@xcite provided a substring parser for bounded context - lr grammars and simple bounded context - lr grammars .
@xcite distributes the work of parsing the substrings of a language over different processors .
the work was extended to different processors in a balanced binary tree fashion and achieved o(log n ) time complexity of parsing . but constructing a bounded context lr grammar for a programming language is also difficult .
c++ is one of the programming languages which can not be parsed by lr ( 1 ) parsing @xcite so creating a bounded context grammar is out of question here .
+ parallel and distributed compilation schemes can be divided into two broad categories , functional decomposition and data decomposition . @xcite and @xcite talks about distributed compilation using a scheme based on functional decomposition .
functional decomposition scheme divides different phases of compiler : lexer , parser , semantic analyzer into functional component and running each of them on separate processors like an instruction pipeline fashion .
the data decomposition scheme divide the input into sections of equal length and parse them in parallel .
bpp is data decomposition scheme which parallel the parser by divide and conquer approach .
the data decomposition scheme was developed by @xcite , @xcite , @xcite , @xcite .
these schemes are parsing lr ( k ) in parallel .
they divide the input into sections of equal length and then parse them in parallel .
@xcite , @xcite , @xcite describes asynchronous algorithms while @xcite develops a synchronous algorithm . @xcite
develops a parallel lr parser algorithm using the error recovery algorithm of @xcite .
+ @xcite has developed an incremental parallel compiler which could be used in interactive programming environment and he developed an incremental parallel parser also .
@xcite improves upon the divide and conquer parsing technique developed by @xcite .
they show that while the conquer step of algorithm in @xcite is @xmath0 but under certain conditions it improves to @xmath1 .
+ @xcite describes a grammar partitioning scheme which would help in parsing the language in parallel . in @xcite
a type of statement level parallelism has been developed .
the grammar is divided into n different sub - grammars corresponding to n subsets of the language which will be handled by each sub - compiler . for each
n sub grammars required to generate parse tables ( using parser generator ) along with driver routine constitute a parser for sub - compiler . for each sub - compiler ,
a requirement of modified scanner is there which recognizes subset of the language .
the technique described in @xcite requires a lot of modification to the lexical analyzer .
a separate lexical analyzer has to be developed for one type of language .
the parser of @xcite requires automatic tools for its implementation .
+ in all the above described techniques constructing abstract syntax tree for a block structured language is difficult . as blocks in a block
are independent on each other , so they can be parsed independently .
moreover this scheme would not involve inter - thread communication before a thread completes its parsing of blocks .
hence , no shared memory synchronization methods are required to coordinate between different threads .
it could be easily seen that the creation of an abstract syntax tree is also very easy . with all these required things in mind
we have developed block parallelized parser for lr ( 1 ) languages .
we assume the notation for context free grammar is represented by @xmath2 where _ n _ is set of non - terminals , _
t _ is set of terminals , _ s _ is start symbol and _ p _ is set of productions of the grammar .
the language generated by _
g _ is given as @xmath3 we will use the following conventions .
+ @xmath4 @xmath5 @xmath6 @xmath7 @xmath8 given a grammar _
g _ , we represent its augmented grammar as @xmath9 , where + @xmath10 @xmath11 @xmath12 here @xmath13 is called the augmented start symbol of @xmath14 and $ is the end of string marker .
we denote a set of end of string markers as eos . + in our paper we will represent block parallelized parser as bpp , jump shift reduce parser as jsr and incremental jump shift reduce parser as i_jsr .
an lr(1 ) item is represented as @xmath15 $ ] , where _ a _ is the lookahead symbol .
+ in a programming language , a block represents a section of code grouped together .
this section of code can be a group of statements , or a group of declaration statements .
for example in java , a block corresponding to class definition contains declaration statements for fields and methods .
block corresponding to function definition can contain declaration statement for local variables or expression statements or control flow statements .
a top - level block is the starting block of a program which contains definition for classes , functions , import / include statements etc .
child blocks are contained in either top - level block or another child block .
as we proceed further , block will be in reference to child block .
1 , shows an example of top - level and child blocks of a java program .
a start block symbol could be `` \ { '' in c style languages , `` begin '' in pascal style languages is represented as terminal @xmath16 . an end block symbol which could be `` } '' in c style languages or `` end '' in pascal style languages
is represented as @xmath17 .
we now survey lr ( 1 ) parsers and their generation algorithm as given by @xcite .
lr ( 1 ) parsers are table driven shift reduce parsers . in lr ( 1 ) , `` l '' denotes left - to - right scanning of input symbols and `` r '' denotes constructing right most derivation in reverse .
some extra information is indicated with each item : the set of possible terminals which could follow the item s lhs .
this set of items is called the lookahead set for the item . here
, `` 1 '' signifies that number of lookahead symbols required are 1 .
+ lr ( 1 ) parser consists of an input , an output , a stack , a driver routine .
a driver routine runs its parsing algorithm which interacts with two tables action and goto . any entry in action and goto tables
are indexed by a symbol which belongs to @xmath18 and the current state .
an entry in both the tables can be any one of the following : * if action [ j , a ] = @xmath19s , q@xmath20 then a shift action must be taken . *
if action [ j , a ] = @xmath19r , @xmath21 then reduce symbols to a production . * if action [ j , a ] = accept then grammar is accepted . * if action [ j , a ] = error then error has occurred . * if goto [ j , a ] = q then go to state q. an lr ( 1 ) item is of the form @xmath22 $ ] , where _ a _ is a lookahead symbol .
construction of lr ( 1 ) items requires two procedures closure and goto .
closure takes a set of items as its argument .
goto takes a set of items and a symbol as arguments .
both are defined as follows : + @xmath23~|~[a \rightarrow \alpha .
b\beta,~a ] \in i$ ] @xmath24 @xmath25~|~[a \rightarrow \alpha .
x\beta , a ] \in i\})$ ] collection of set of lr ( 1 ) items is created using closure and goto functions .
items function creates the collection of set of lr ( 1 ) items .
+ @xmath26\ } ) \cup \{goto ( i , x)~|~i
\in c~and~x \in p'\}$ ] action and goto tables are created using this collection .
following is the procedure to create these tables : 1 .
create collection of set of lr(1 ) items .
let this collection be @xmath27 2 .
let _ i _ be the state of a parser constructed from @xmath28 .
entries in action table are computed as follows : 1 .
action [ _ i _ , _ a _ ] = shift _ j _ , if @xmath29 \in i_i $ ] and @xmath30 2 .
action [ _ i _ , _ a _ ] = reduce @xmath31 , if @xmath32 \in i_i $ ] and @xmath33 3 .
action [ _ i _ , $ ] = accept , if @xmath34 \in i_i $ ] 4 .
goto [ _ i _ , _ a _ ] = _ j _ , if goto ( @xmath35 , a ) = @xmath36 3 .
all other entries not defined by ( b ) and ( c ) are set to error .
the initial state is the one containing the item @xmath37 $ ] .
most of the programming languages could be constructed from lr ( 1 ) grammar .
hence , lr ( 1 ) is the most widely used parser .
many parser generators like yacc and gnu bison generates an lr ( 1 ) parser . + a jump shift reduce @xcite ( jsr in short ) parser is an extension of lr ( 1 ) parser .
lr ( 1 ) parser generates action and goto table for the augmented grammar @xmath38 .
jsr parser first partition the grammar @xmath38 into several sub grammars and creates parsing sub- table of every sub grammar .
hence , the action and goto tables of lr ( 1 ) are split into several action and goto tables in jsr parser .
jsr parser is equivalent to the lr ( 1 ) parser that is it will only accept languages generated by lr ( 1 ) grammar @xcite .
+ let @xmath39 be the augmented grammar of grammar @xmath40 . let us partition the grammar g on the basis of non terminals .
let @xmath41 denotes a partition of the grammar _
g _ , such that we have + @xmath42 + where , @xmath43 @xmath44 @xmath45 @xmath46 @xmath47 + therefore , we have + @xmath48 for every subgrammar @xmath41 , a parsing subtable named tab(@xmath49 ) is built .
each subtable contains action and goto subtables which are represented by tab(@xmath49).action and tab(@xmath49).goto .
in addition to the shift , reduce and accept action there is an additional jump action .
jump action is associated with a sub - table .
whenever a jump action is encountered then the parsing algorithm jumps to a sub - table and parse the partition related to that sub - table .
+ we will now investigate few points about the incremental jump shift reduce parser @xcite .
incremental jump shift reduce parser ( i_jsr ) @xcite is based upon the jsr parser @xcite . a set of incremental categories will be defined which could be incrementally parsed by i_jsr parser .
given a grammar @xmath40 a set of incremental categories has to be defined @xmath50 and the incremental language of g is + @xmath51 where @xmath52 and @xmath53 + for every incremental category _ a _ , add a production @xmath54 , where @xmath55 is an end - of - string for _
a_. for a given grammar @xmath40 with a set of incremental categories @xmath56 an incremental grammar is defined as + @xmath57 where , eos is the set of end of string markers = @xmath58 + @xmath59 + a major change by this extension is that now the strings may contain incremental symbols which are also non - terminal symbols .
the difference between action and goto tables disappears , as an incremental category can also occur in the input string and can be shifted on the stack .
hence , we would have only action table and no need for goto table .
as we have also introduced eos set , the action table can now be indexed with symbols belonging to @xmath60 .
every incremental category will have its own start state and accept action .
we will represent the subtable as tab(@xmath61 ) .
entries of table will be as follows : * if tab(@xmath61 ) [ _ j _ , _ x _ ] = @xmath19s , q@xmath20 then a shift action must be taken . * if tab(@xmath61 ) [ _ j _ , _ x _ ] = @xmath19r , @xmath62 then a reduce action must be taken . *
if tab(@xmath61 ) [ _ j _ , _ $ _ ] = accept then input is accepted . *
if tab(@xmath61 ) [ _ j _ , @xmath63 = accept then input for incremental category @xmath64 will be accepted . * if tab(@xmath61 ) [ _ j _ , _ a _ ] = @xmath19j , k@xmath20 then jump to a subtable tab(@xmath65 ) . * if tab(@xmath61 ) [ _ j _ , _ a _ ] = error then error occurred .
as jsr generation algorithm was already developed for lr ( 0 ) items and incremental jsr generation algorithm was developed for slr ( 0 ) items @xcite . in this section , we will first extend the jsr parser generation algorithm to accept lr ( 1 ) grammar and then we will extend i_jsr parser to accept lr(1 ) grammar .
+ generation of subtables first requires the generation of canonical collection of sets of augmented items .
an augmented item is a triplet represented as @xmath19i , ff , tf@xmath20 , where i is an lr item , ff called from - field and tf called to - field are the names of parsing sub - tables
. from - field represents the sub - table which contains the last action performed by parser and to - field represent the sub - table which contains the next action to be performed .
although we focus only lr ( 1 ) items but the procedure for lr ( k ) items is very similar .
+ to - field of an augmented item of a state is determined using to function .
let us define the to function .
@xmath66 @xmath67 @xmath68 @xmath69 function to calls function choose_next .
this function selects a parsing table out of those in which the parsing of the remaining string could continue .
let @xmath70 be a total ordering relation over the set @xmath71 of the names of parsing sub - tables , such that @xmath72 + @xmath73 @xmath74 from - field of an augmented item is enriched using from function .
from takes two arguments @xmath75 , which is a set of items enriched with to - field and @xmath76 , whose items we have to enrich with from - field .
+ @xmath77 @xmath78 @xmath79,~tf >
\in i_t''$ ] @xmath80,~ff > \in i_j'$ ] @xmath81 states procedure is used to generate the collection of set of jsr items .
states algorithm first generates the collection of sets of lr(1 ) items using items procedure which was discussed previously .
afterwards , it calls to and from functions to generate set of augmented items from the corresponding set of lr(1 ) items .
+ @xmath82 @xmath83 @xmath84 @xmath85 @xmath86 @xmath87 @xmath88 we will now extend i_jsr parser to accept lr ( 1 ) languages . this extended parser is based on the previous jsr parsing algorithm .
the first function used to compute the set of first symbols related to a non - terminal has to be modified to include the incremental categories also .
the reason being an incremental category can also occur in the input string , can be shifted on the stack while parsing and can reduce to a production .
moreover , first should now also include the eos markers .
hence , the new first becomes + @xmath89 for an incremental category _
a _ , there will be a set of items containing item @xmath90 $ ] and items @xmath91 $ ] .
then the state corresponding to this set of items will be the start state of the incremental grammar corresponding to a. correspondingly , there will a single set of items that contains the item @xmath92 $ ] .
the state belonging to this set of item is the accepting state of the incremental grammar corresponding to a. the i_jsr parser has initial and final states for every incremental category .
+ now , we can extend the i_jsr parser for accepting languages generated by lr(1 ) grammars .
the procedure i_tabs given below is used to construct the i_jsr parsing table .
@xmath82 @xmath93 = < s , q>~where~goto ( i_j , x ) = i_q$ ] @xmath94=<j , k>$ ] @xmath93 = <
r , a \rightarrow \alpha>$ ] @xmath95 = < j , k>$ ] @xmath96=accept$ ] @xmath97 = accept$ ]
in this section we will define the concept of first non terminals .
+ we define a set of first non terminals for a non terminal _ a _ as a set of non - terminals that appear at the beginning of any sentential form derived from _ a _
i.e. a set of non terminals _ b _ such that there exists a derivation of the form @xmath98 .
@xmath99 represents the set of first non terminals for _ a _ and can be represented in set notations as : + @xmath100 to compute @xmath99 for any non - terminal a , apply the following rules until no more terminals can be added to the @xmath99 set . 1 .
if _ a _ is a terminal , then @xmath101 2 .
if _ a _ is a non - terminal and @xmath102 is a production for some @xmath103 , then place @xmath104 and everything in @xmath105 in @xmath106 if @xmath107 3 . if @xmath108 is a production , then @xmath101 example 1 : if we have following productions : * s d a b * s c * a a b * b b * c c * d d then find @xmath109 ?
solution 1 : + due to first two productions of s we have , + @xmath110 @xmath111 @xmath112
in this section we will use the concept of first non terminals to create partitions of grammar .
i_jsr parser will use these partitions to develop its tables .
we will see that this kind of partitioning leads to a very interesting property in i_jsr parsing algorithm .
we will generalize this property in a theorem and will also prove it . + we will partition the grammar in such way that : * every incremental category will have its own partition .
* the partition of incremental category will contain first non terminals of that incremental category also .
* intersection of set of first non - terminals of any two incremental categories must be empty .
* all other remaining non - terminals including the augmented and start symbol are included in the first partition . given a grammar @xmath113 with a set of incremental category ,
@xmath114 we define partitions of non - terminals as @xmath115 such that : + @xmath116 and + @xmath117 and first partition , + @xmath118 example 2 : partition the grammar given in example 1 using first non terminals as given above and create i_jsr parsing table .
then parse the string `` _ _ dabb _ _ '' and parse string `` _ _ ab _ _ '' incrementally for incremental category _
solution 2 : first we have to create an augmented grammar of the given grammar by adding production @xmath119 .
next we can create two partitions of grammar as given by the following partitions of non terminals .
+ @xmath120 @xmath121 as the incremental category we want is only _
hence , there will be two partitions , @xmath122 containing _ a _ and its first non terminals and @xmath123 will contain the remaining non terminals . moreover , we would also have to add an eos marker for _ a _ , let us say it is @xmath124 .
we can generate the i_jsr table using _
i_tabs _ procedure described in section 5 .
+ let us parse the string `` _ _ dabb _ _ '' .
table 1 shows the series of actions taken while parsing `` _ _ dabb _ _ '' . in this case
the start state will be the start state of table @xmath125 i.e. 0 . table 2 shows the series of actions taken when `` _ _ ab _ _ '' is parsed incrementally with the incremental category _
a_. in this case the start state will be the start state of table tab ( a ) i.e. 2 .
+ .parsing of `` _ _ ab _ _ '' [ cols="<,>,<",options="header " , ] in table 1 , when stack state reaches 0d2 there is a jump to the table tab ( a ) . from this point until when stack state changes to 0d2a6 , the actions taken are same as the actions of table 2 and in the same table i.e. tab ( a ) .
moreover , in between these stack states in table 1 `` _ _ ab _ _ '' is parsed to _ a_. +
we can generalize this example in the sense that same series of actions are taken when parsing a string and when parsing its substring incrementally for its incremental category . in the current example all the same actions happens in the same table because we created the partitions in such a way that all the first non terminals are in that partition .
if the partitions were not created in the way described , it could have happened that these actions would happen in the different sub tables .
+ this technique is utilized by our bpp and it is generalized and proved in the theorem below . given an incremental grammar @xmath126 with set of incremental categories @xmath127 such that the non terminal partition , @xmath128 related to incremental category @xmath129contains only @xmath129 and @xmath130 .
for an incremental category @xmath129 and any terminal @xmath131 , if @xmath132 and during parsing of the word @xmath133 the parser reaches at a state _ q _ in the subtable of @xmath129 after performing the shift action on _ b _ then during the incremental parsing of the word @xmath134 for the incremental category @xmath129 the parser will also reach the state _ q _ after performing the shift action on _ b _ in the sub table of @xmath129 .
* outline * : we will divide the proof in 4 cases . for each case
, we will first find two sets of jsr items reached after performing shift on _ b _ one during the parsing of word @xmath135 and other during the incremental parsing of word @xmath136 related to incremental category @xmath129 .
we will then show that both of these sets contains same jsr items which implies the above theorem . + it is given that @xmath129 is an incremental category and @xmath137 and let @xmath128 be the partition containing only @xmath129 and @xmath130 .
let @xmath138 be the sub - table related to @xmath129 .
for any non - terminal @xmath139 and @xmath140 , we must have @xmath141 .
+ during incremental parsing of the word @xmath142 for the incremental category @xmath129 , the state before performing actions related to
_ b _ will be the start state of the sub table @xmath143 .
let that start state be _
m_. + we will have four cases on the basis of whether the grammar has productions of the form , @xmath144 and @xmath145 + * case 1 * : if @xmath146 and @xmath147 + let the set of lr(1 ) items related to the start state of @xmath143 i.e. state _ m _
be @xmath148 and _ x _ lies in some partition other than @xmath128 i.e. @xmath149 and @xmath150 .
as noted by @xcite the start state of @xmath143 must contain the item @xmath151 $ ] , where _ d _ is a lookahead symbol .
so we have @xmath151 \in i_m$ ] .
it is evident that the item @xmath151 $ ] should be result of a closure of another item .
the only such item we can see is @xmath152 $ ] , where _ j _ is some lookahead symbol .
so , we must have @xmath152 \in i_m$ ] . as discussed in section 2 , to get a set of lr(1 ) items we have to apply closure ( @xmath152 $ ] ) .
hence , we have + @xmath153 \in i_m$ ] + @xmath154 \in i_m$ ] + @xmath155 \in i_m$ ] + @xmath156 \in i_m$ ] + where _ e _ and _ f _ are some lookahead symbols .
+ let @xmath157 be the set of lr(1 ) items such that @xmath158 .
so , @xmath152 \in i_o$ ] .
let @xmath159 be a set of jsr items enriched with to fields corresponding to all lr(1 ) items in @xmath157 .
after applying to procedure over @xmath157 we would get the to field for item @xmath152 $ ] as @xmath160 because @xmath149 .
+ to get to fields for all jsr items corresponding to lr ( 1 ) items of the set @xmath148 we have to apply to procedure over @xmath148 .
we could see that the to fields for items @xmath152 $ ] , @xmath154 $ ] , @xmath161 $ ] , @xmath156 $ ] will be @xmath138 .
+ to enrich jsr items for all the lr(1 ) items in @xmath148 with from field , we would apply from procedure as @xmath162 .
now we could see that the from field of jsr item of lr(1 ) item @xmath161 $ ] will be equal to the from field of @xmath152 $ ] which in turn is equal to the to field of @xmath153 $ ] , which is @xmath160 .
so , the set of jsr items ( @xmath163 ) corresponding to @xmath148 contains the following items : + @xmath164 \in i_m^a$ ] + @xmath165 \in i_m^a$ ] + @xmath166 \in i_m^a$ ] + @xmath167 \in i_m^a$ ] + let after performing shift operation in the state _
m _ over the symbol _
b _ , parser reaches state _
n_. so , we must have + @xmath168~|~\forall~g \in first ( \beta f)\})$ ] + moreover , the to and from fields of all items in the above state will be @xmath138 .
we have obtained the set of jsr items reached after performing shift over the symbol _
b _ during incremental parsing of word @xmath136 for incremental category @xmath129 . also , the subtable at this state is @xmath169 .
+ we will now obtain the set of jsr items reached after performing shift over the symbol _
b _ during the parsing of word @xmath135 for incremental category @xmath129 .
let us consider the derivation @xmath170 such that , _
b _ does nt derive @xmath171 . in the above derivation only one production out of @xmath144 and @xmath145 will be used .
hence , we have two cases of the basis of which production is used .
+ + _ case 1.1 _ : if production @xmath144 is used , then the set of items related to state reached just after performing shift on _ c _ in the above derivation say @xmath172 will contain the lr(1 ) item @xmath153 $ ] besides other items of the form @xmath173 $ ] , where @xmath174 .
after applying to procedure on @xmath172 , we should get the to field for lr(1 ) item @xmath153 $ ] as @xmath160 .
let us represent these set of items enriched with to field as @xmath175 .
+ after parsing @xmath176 next state will be obtained by performing goto over @xmath172 with symbol _
x_. let that state be @xmath177 .
now we must have , + @xmath178 .
after performing goto we get + @xmath152 \in i_y$ ] @xmath154 \in i_y$ ] @xmath161 \in i_y$ ] @xmath156 \in i_y$ ] after applying to procedure over @xmath177 , we get the to fields of items @xmath152 $ ] @xmath154 $ ] @xmath161 $ ] @xmath156 $ ] as @xmath138 . to enrich jsr items of lr(1 )
items in @xmath177 we would apply from procedure as @xmath179 .
now , we could see that the from field of jsr item for @xmath161 $ ] will be equal to the from field of jsr item for @xmath152 $ ] which in turn is equal to the to field of @xmath153 $ ] , which is @xmath160 .
so , the set of jsr items ( @xmath180 ) corresponding to @xmath177 contains the following items : + @xmath164 \in i_y^a$ ] @xmath165 \in i_y^a$ ] @xmath166 \in i_y^a$ ] @xmath167 \in i_y^a$ ] let the state reached after performing shift of _ b _ from the state @xmath177 be @xmath181 .
then , + @xmath182~|~\forall~g \in first ( \beta f)\right \rbrace)$ ] moreover , the to and from fields of all jsr items of @xmath181 will be @xmath138 .
+ so , we have @xmath183 .
this shows that the states _ z _ and _ n _ are same .
let us name these states as _
q_. also , the to fields for the items of @xmath181 and @xmath184 are same i.e. @xmath138 .
hence , in this case shift on terminal _
b _ in states _ m _ and _ x _ results only in one state _
q _ and in the sub - table for incremental category @xmath129 .
+ theorem is * proved * in this case .
+ _ case 1.2 _ : if @xmath185 is used .
x _ is the state reached before performing the shift on _
b_. then , the set of items , say @xmath172 related to state _ x _ will contain following items : + @xmath186 \in i_y^a$ ] @xmath154 \in i_y^a$ ] @xmath161 \in i_y^a$ ] @xmath156 \in i_y^a$ ] + after applying to operation to @xmath172 , the to fields of all jsr items for above lr(1 ) items will be @xmath138 .
+ let @xmath177 be the state reached after performing shift on _ b _ in the state @xmath172 .
so , @xmath187~|~\forall~g \in first ( \beta f)~\})$ ] also , the to and from fields of all the jsr items for the above set of lr ( 1 ) items will be @xmath138 .
+ hence , @xmath188 .
this shows that the states _
y _ and _ n _ are same .
let us name the state as _
q_. moreover , the to fields of lr(1 ) items of @xmath177 and @xmath184 are @xmath138 . hence , in this case shift on terminal _
b _ in the states _ m _ and _ x _ results only in one state _
q _ in the subtable of @xmath129 .
+ theorem is proved in this case .
+ + as theorem has been proved in _ case 1.1 _ and _ case 1.2_. so , for * case 1 * also the theorem has been proved .
+ we have proved for the case containing both productions .
two other cases are when only one of these productions is present .
the proof of both of these cases are very similar to the * case 1*. + please note that with the given set of conditions in the theorem , we could nt have the case in which none of the productions belong to this set .
theorem 1 is crucial to bpp . in the succeeding sections we will use this theorem to create our parallel parsing algorithm .
in this section we will present our block parallelized parser for lr ( 1 ) grammars
. we will first give the intuition and working of bpp .
then we will present our algorithm and give a proof that our parser can accept all the lr(k ) and lalr ( k ) languages which can be accepted by a shift reduce lr(k ) and lalr ( k ) parser .
+ let @xmath189 be the augmented grammar .
the incremental categories are the non terminals associated with the blocks to be parsed in parallel .
for example , if we want to parse class definitions in parallel then we can define _ class - definition _ as the incremental category .
other examples can be _ function - definition _ , _ if - statement _ , _ for - loop _ , _ while - loop _ if they can be parsed in parallel . for most of the programming languages
including c , c++ , java , c # above blocks can be parsed in parallel . in general
we can define an incremental category to be the non terminals which derive a string containing the start of the block symbol , @xmath190 and ending with the end of the block symbol @xmath191 .
in mathematical terms , for bpp the set of incremental category _ ic _
is defined as : @xmath192 + in the c programming language , a _ function - definition _ can be represented by the following context free grammar productions : + @xmath193 + @xmath194 + @xmath195 + @xmath196 + @xmath197 + @xmath198 + @xmath199 + @xmath200 + @xmath201 + according to the definition of incremental categories above , we can see that _ function - definition _ , _ if - stmt _ , _ while - loop _
, _ for - loop _ follows the condition for an incremental category .
+ in a programming language there could be many types of blocks like in java , a block related to the class definition may contain only method definitions and declaration of member variables while a block related to the method definition would contain statements including expression statements , variable declarations , if statement or loop etc .
this means in a programming language not all blocks contains same type of statements .
hence , encountering the start symbol of block does nt give us enough information about what kind of statements the block would contain . to overcome this problem
we will modify the productions of incremental category such that a reduce action will happen when the start symbol is encountered .
modify the productions of @xmath202 in such a way that every production @xmath203 is split into two productions : + @xmath204 @xmath205 if the productions of incremental category @xmath129 is structured as above then during the parsing of a word related to this incremental category there will be reduce action to reduce the current symbols to production @xmath205 when @xmath190 becomes the look ahead symbol . as each @xmath206
is related to only one @xmath129 and vice - verse , we can easily determine which incremental category has to be parsed next .
+ now we are in a stage to define block parallelized grammar . given a grammar @xmath207 and a set of incremental categories + @xmath208 we define block parallelized grammar @xmath209 such that + @xmath210 @xmath211 @xmath212 and @xmath213 is partitioned using firstnt as given in section 7 .
+ now we can use theorem 1 to create bpp .
let us have a string @xmath214 where @xmath215 and @xmath216 . during the parsing _
, when @xmath190 is encountered we should have a reduce action to @xmath206 , based on the production @xmath205 .
now , we can get @xmath129 associated with @xmath206 . according to theorem 1 , during parsing of the word _ w _ if the parser reaches at state _
q _ in the sub table of @xmath129 after performing shift action on _ a _ then during the incremental parsing of the word @xmath217 for the incremental category @xmath129 the parser will also reach the state _
q _ in the sub table of @xmath129 after performing the shift action on _
a_. this means , we can replace the state reached just before performing shift action on _ a _ with the start state of subtable of @xmath129 and @xmath136 can now be parsed incrementally .
+ it is now evident that why the partitions of non terminals should be created as described in section 7 .
if the partitions are not created in this way , then it may happen after a shift on _ a _ during the parsing of _ w _ and incremental parsing of @xmath136 may reach the same state but not in the same sub - table . by creating partitions as described in section 8 ,
we make sure that when @xmath190 is encountered by bpp then the newly created parallel parser knows in which sub - table it has to continue the parsing in . on the other hand
if partitions are not created as described in section 8 , then newly created parallel parser would nt know in which sub - table it has to continue parsing in . + this property is used by bpp to parse the block incrementally .
algorithm 1 is the bpp parsing algorithm of incremental table @xmath138 . if _
= 1 , then the algorithm is for the top level block .
otherwise it is for other incremental category .
+ @xmath218 @xmath219 @xmath220 @xmath221 @xmath222 @xmath223 @xmath224 @xmath225 @xmath226 @xmath227 @xmath228 @xmath229 @xmath230 @xmath231 @xmath232 @xmath233 @xmath234 @xmath235 @xmath236 @xmath237 @xmath238 this parsing algorithm is for any block be it top level block or child block .
lines 1 - 4 initializes different local variables .
lines 5 - 32 is the main loop of algorithm which does the whole work . line 6 , gets the top state on the stack .
lines 7 - 9 pushes the next state on the stack if there is a shift operation .
similarly , lines 10 - 11 changes current table if there is a jump operation .
line 12 - 27 are executed if there is reduce action which reduces according to the production @xmath239 .
line 14 checks if current input symbol is a start of the block symbol and if reduce action reduces @xmath205 for an incremental category @xmath129 .
if yes then lines 15 - 22 gets @xmath129 related to @xmath206 , pops @xmath240 states from stack and pushes these states and start state to a new stack , creates and starts a new bpp for @xmath129 and shifts to the end of block . in this case
next symbol will become @xmath129 .
if check of line 14 fails then it means this is a regular reduce action not associated with any block .
lines 24 - 27 , pops @xmath241 states from stack and shifts to a new state .
line 28 returns if there is an accept action .
accept action can be for both top level block and child block .
line 30 reports error if none of the above cases are satisfied .
2 shows an example of how bpp works .
it will start parsing the block of function _ f_.
when it will encounter an _ if _ block a new bpp in another thread will be created which will parse _ if _ block .
parent bpp will move ahead to the end of _ if _ block and will also create another thread to parse _ else _ block . in this way input is divided into different threads parsing each block .
+ in this algorithm we have tried to minimize the amount of serial work to be done to get to the end of the block for the new block parser .
one bpp does nt have to do any communication with other bpps . also , there are no side effects of the above algorithm .
all the variables which are being modified are local variables .
hence , there is no need of synchronization .
this also reduces any amount of cache contention between different processors .
generation of abstract syntax tree or parsing tree is easy using above algorithm and it requires very little change in the above algorithm .
+ it may be argued that the step `` go to the end of this block '' is a serial bottleneck for the parallel algorithm .
@xcite describes an algorithm to perform lexical analysis of string in _
o(log n ) _ time using _
o(n ) _ processors in a parallel fashion .
when performing lexical analysis in parallel as described in @xcite , lexer could store the start symbols of a block with its corresponding end symbol for that block .
now determining the end of block is just a matter of searching through the data structure .
many ways exist to make this searching as fast as possible like using a binary search tree or a hash table .
in this section we will show how our technique is better than other techniques .
@xcite developed a technique which divides whole grammar into n sub- grammars which are individually handled by n sub - compilers .
each sub - compiler needs its own scanner which can scan a sub - grammar .
it requires an automatic tool to generate sub - compiler .
this technique requires significant changes in not only in parser and grammar but also in lexical analyzer phase .
contrast to this our block parallelized parser is easy to generate as our technique does not require any change in the grammar and lexical analyzer and it is pretty easy to modify current yacc and bison parser generator tools to support the generation of our parser .
+ lr substring parsing technique described in @xcite is specifically for bounded context ( 1 , 1 ) grammars .
there are no limitations like this to block parallelized parser .
although in this paper we have shown how we can create an lr ( 1 ) block parallelized parser but we believe it can be extended to lr ( k ) class of languages and also could be used by lalr ( 1 ) parser .
hence , our technique accepts a larger class of grammars .
+ @xcite , @xcite , @xcite , @xcite all develops algorithm for parsing lr ( k ) class of languages in parallel . these techniques and in all other techniques the creation of abstract syntax tree is not as easy as itis in our technique .
moreover our technique is simpler than all others .
+ hence , we could see that block parallelized parser is easy to construct , accepts wider class of languages and supports an easy construction of abstract syntax tree .
we implemented lexer , block parallelized parser and shift reduce lr ( 1 ) parser for c programming language supporting a few gnu c extensions required for our tests .
implementation was done in c # programming language . to simplify our implementation
we only included function - definition as the incremental category for bpp .
moreover , function - definition would still give us a sufficient amount of parallelism as we would see in the evaluation .
we modified the lexer phase so that it will keep track of the position of s b and its corresponding e b .
this information was stored in the form of a c # dictionary ( which is implemented as a hash table ) with the position of s b as the key and position of e b as the value .
as , thread creation has significant overhead so we used c # taskparallel library which is thread pool implementation in c#. our implementation does nt have a c preprocessor implementation .
so , we first used gcc to perform preprocessing and the preprocessed file is used as input to our implementation .
+ we evaluated the performance of bpp with shift reduce lr ( 1 ) parser by parsing 10 random files from the linux kernel source code .
we compiled c # implementation using mono c # compiler 3.12.1 and executed the implementation using mono jit compiler 3.12.1 on machine running fedora 21 with linux kernel 3.19.3 with 6 gb ram and intel core i7 - 3610 cpu with hyperthreading enabled .
+ in c programming language , preprocessing ` # include ` could actually generate very long files .
normally , the header files contains declarations not function definitions .
so , this leads to less amount of parallelism being available .
hence we decided to evaluate with including header files and excluding header files .
3 shows the performance improvement with respect to shift reduce lr(1 ) parser of 10 random linux kernel files .
3 shows performance improvement for both cases including the header files and excluding the header files .
as expected we could see that the performance improvement with excluding the header files is more than the performance improvement including the header files .
+ the performance improvement in the case of excluding header files matters most for the programming languages like java , python , c # where most of the program is organized into blocks the results because in these programs the amount of parallelism available is high .
+ the average performance improvement in the case of excluding header files is 52% and including header files is 28% .
in this document we present our block parallelized parser technique which could parse the source code in a parallel fashion .
our approach is a divide and conquer approach in which , the divide step divides the source code into different blocks and parse them in parallel whereas the conquer step only waits for all the parallel parsers to complete their operation .
it is based on the incremental jump shift reduce parser technique developed by @xcite .
our technique does nt require any communication in between different threads and does nt modify any global data .
hence , this technique is free of thread synchronization .
we develop this technique for lr ( 1 ) languages and we believe that it can be extended to accept lr ( k ) languages and could be converted to an lalr ( 1 ) parser easily .
our technique does nt do any major changes in the parsing algorithm of a shift reduce parser hence the abstract syntax tree can be created in the same way as it has been creating in shift reduce parser .
moreover , our parser can also work as an incremental block parallelized parser .
we implemented block parallelized parser and shift reduce lr ( 1 ) parser for c programming language in c#. the performance evaluation of bpp with shift reduce lr ( 1 ) parser was done by parsing 10 random files from the linux kernel source code .
we compiled c # implementation using mono c # compiler 3.12.1 and executed the implementation using mono jit compiler 3.12.1 on machine running fedora 21 with linux kernel 3.19.3 with 6 gb ram and intel core i7 - 3610 cpu with hyperthreading enabled .
we found out that our technique showed 28% performance improvement in the case of including header files and 52% performance improvement in the case of excluding header files .
our parser accepts lr ( 1 ) languages we would like to extend it to accept lr ( k ) languages . in our technique , the parser determines when to create a new parallel parser thread .
if the responsibility of this decision can be given to the lexical analysis phase then the lexical analysis can actually start the parsers in parallel .
this will lead to significant performance advantage .
moreover , our technique has been applied to languages which does nt have indentation in their syntax like the way python has .
@xcite shows an efficient way to parse the language which has indentation as a mean to determine blocks .
our parser can be extended to accept those languages also .
we are working towards developing a block parallelized compiler which could compile different blocks of a language in parallel .
block parallelized parser is one of the components of a block parallelized compiler .
semantic analysis phase also share the same properties as the syntax analysis phase . in programming languages , an entity like variable or type
is declared before using it .
so , in this case also a lot can be done to actually parallelize the semantic analysis phase .
neal m. gafter and thomas j. leblanc .
parallel incremental compilation .
department of computer science , university of rochester : ph .
d. thesis .
113 gwen clarke and david t. barnard .
an lr substring parser applied in a parallel environment , journal of parallel and distributed computing . g. v. cormack .
an lr substring parser for noncorrecting syntax error recovery , acm sigplan notices alfred v. aho , monica s. lam , ravi sethi and jeffery d. ullman .
2007 . compilers principle tools and techniques second edition , prentice hall floyd , r. w. bounded context synctactic analysis .
acm 7 , 21 february 1961 j. h. williams , 1975 . bounded context parsable grammars .
information and control 28 , 314 - 334 r. m. schell , 1979 .
methods for constructing parallel compilers for use in a multiprocessor environment .
d. thesis , univ . of illinois at urbana - champaign j. cohen and s. kolodner .
1985 . estimating the speedup in parallel parsing , ieee trans .
software engrg .
se-11 c. n. fischer , 1975 . on parsing context free languages in parallel environments .
d. thesis , cornell univ .
m. d. mickunas and r. m. schell .
parallel compilation in a multi - processor environment ( extended abstract ) .
comput d. ligett , g. mccluskey , and w. m. mckeeman , parallel lr parsing tech . rep . , wang institute of graduate studies , july 1982 edward d. willink , meta - compilation for c++ , university of surrey , june 2001 pierpaolo degano , stefano mannucci and bruno mojana , efficient incremental lr parsing for syntax - directed editors , acm transactions on programming languages and systems , vol . 10 , no .
3 , july 1988 , pages 345 - 373 .
sanjay khanna , arif ghafoor and amrit goel , a parallel compilation technique based on grammar partitioning , 1990 acm 089791 - 348 - 5/90/0002/0385 w. daniel hillis and guy l. steele , data parallel algorithms , communications of the acm , 1986 bison , http://www.gnu.org/software/bison yacc , http://dinosaur.compilertools.net/yacc clang , http://clang.llvm.org mono , http//mono - project.com el - essouki , w. huen , and m. evans , `` towards a partitioning compiler for distributed computing system '' , ieee first international conference on distributed computing systems , october 1979 .
thomas j. pennello and frank deremer , `` a forward move algorithm for lr error recovery '' , fifth annual acm symposium on principles of programming language .
jean - philipe bernardy and koen claessen , efficient divide - and - conquer parsing of practical context - free languages .
l. valiant .
general context - free recognition in less than cubic time .
j. of computer and system sciences , 10(2):308 - 314 , 1975 .
micheal d. adams .
`` principled parsing for indentation - sensitive languages '' , acm symposium on principles of programming languages 2013 .
january 23 - 25 . | software s source code is becoming large and complex . compilation of large base code is a time consuming process .
parallel compilation of code will help in reducing the time complexity .
parsing is one of the phases in compiler in which significant amount of time of compilation is spent .
techniques have already been developed to extract the parallelism available in parser .
current lr(k ) parallel parsing techniques either face difficulty in creating abstract syntax tree or requires modification in the grammar or are specific to less expressive grammars .
most of the programming languages like c , algol are block - structured , and in most language s grammars the grammar of different blocks is independent , allowing different blocks to be parsed in parallel .
we are proposing a block level parallel parser derived from incremental jump shift reduce parser by @xcite .
block parallelized parser ( bpp ) can even work as a block parallel incremental parser .
we define a set of incremental categories and create the partitions of a grammar based on a rule . when parser reaches the start of the block symbol it will check whether the current block is related to any incremental category . if block parallel parser find the incremental category for it , parser will parse the block in parallel .
block parallel parser is developed for lr(1 ) grammar . without making major changes in shift reduce ( sr )
lr(1 ) parsing algorithm , block parallel parser can create an abstract syntax tree easily .
we believe this parser can be easily extended to lr ( k ) grammars and also be converted to an lalr ( 1 ) parser .
we implemented bpp and sr lr(1 ) parsing algorithm for c programming language .
we evaluated performance of both techniques by parsing 10 random files from linux kernel source .
bpp showed 28% and 52% improvement in the case of including header files and excluding header files respectively . |
single electrons are captured by confining a two - dimensional electron gas ( 2deg ) formed in an algaas / gaas / algaas double heterostructure quantum well with a gaas width of 7.3 nm for small electron @xmath25-factor .
the 2deg has a carrier density of 2.1 @xmath87 10@xmath88 @xmath89 , and a mobility of 1.0 @xmath87 10@xmath90 @xmath91/vs .
the upper ( lower ) barrier is 95 nm ( 100 nm ) al@xmath92ga@xmath93as .
the top 65 nm in the upper barrier is doped by si .
the device is placed in a dilution refrigerator at a base temperature of 25 mk .
the double dot charge configuration is monitored by measuring a charge sensor current using a room temperature current amplifier with a rise - time of 100 @xmath36s . in fig .
4 of the main text , the dependence of @xmath2 on the magnetic field magnitude @xmath72 shows three crossovers in the trend .
the values of @xmath72 for these crossovers differ for in - plane and perpendicular orientation of the field due to the @xmath25-factor anisotropy , since the zeeman energy , which determines the energy scale of the crossover fields , is proportional to the @xmath25-factor .
the g - factor is reduced compared to that of the bulk due to the small thickness of the gaas layer and al content of the barriers in our heterostructure @xcite .
the g - factor of the 2deg used in this work , @xmath94 , was measured in a perpendicular magnetic field using an electron spin resonance technique @xcite .
the ratio @xmath95 extracted from the fitted data of fig .
4 results in @xmath96 , which is consistent with the @xmath25-factor anisotropy found in ref .
@xcite and with the fact that the @xmath25-factor of a 2deg and a dot are comparable , concluded in ref .
. the electron spin relaxation rate due to spontaneous phonon emission is known to grow with the magnetic field magnitude .
we do not see such an effect in our data , observing only an increase of @xmath29 with the field magnitude .
also , as we analyze in detail in sec .
vi , such an inelastic transition within the ( 11 ) subspace [ for example @xmath97 is expected to result in a histogram of tunneling times qualitatively different from our observations ( being non - linear at long times , see fig .
s3 , versus linear at long times as seen in fig . 3 of the main text ) .
the absence of this effect is most probably due to the corresponding relaxation time being too long .
this , in turn , is due to the small @xmath25-factor in our sample .
based on the previous works @xcite , we estimate the spin inelastic relaxation time for the highest fields we used to be of the order of seconds .
this is much smaller than the longest inter - dot tunneling times that we observe ( and model as elastic tunnelings ) , of the order of 100 ms .
one expects that the effects of inelastic transitions would become appreciable in @xmath29 making the @xmath25-factor larger , and/or the inter - dot tunneling smaller .
here we show the charge diagram demonstrating the pauli spin blockade in a weak tunnel coupling regime . to obtain it , we tuned both interdot and dot - lead tunneling rates to the order of khz , applied a voltage of 1.4 mv across the dot , and swept through the displayed range in gate voltages @xmath98 and @xmath99 . at zero magnetic field
the charge stability diagram shows typical finite bias triangles , fig .
[ fig_s1](a ) , where electrons sequentially tunnel through the dqd @xcite .
the number of tunneling events per pixel is of the order of one , since the gate voltage sweep rate is comparable to the tunneling rates . under such conditions ,
charge tunneling is observed as a noisy signal in the diagram . at finite magnetic field , a trapezoidal region of blocked current
is developed at the base of the triangles , fig .
[ fig_s1](b ) , a hallmark of pauli spin blockade @xcite .
it reflects the trapping of the system in a polarized triplet state , where the electron can not transit across the double dot without a spin flip , corresponding to blocked tunneling sketched on fig .
1(b ) .
= 1 t ) where pauli spin blockade is observed as a suppressed signal inside the highlighted trapezoidal region .
[ fig_s1],scaledwidth=45.0% ]
we extract the two ( blocked and direct ) tunneling times by fitting to a bi - exponential distribution in the following way . for each value of the magnetic field ( one point in fig .
4 ) we collect a set of @xmath83 , typically @xmath100 to @xmath101 , individual ( 11 ) to ( 02 ) tunneling times ( such a set is plotted as a histogram in fig . 3 ) .
we order the tunneling times in the set in ascending order , labeling them by integer index @xmath102 , so that @xmath103 .
we define the cumulative probability function @xmath104 , which describes the probability that no tunneling occurred until time @xmath105 as @xmath106 . in the model where the tunneling from ( 11 ) to ( 02 ) proceeds with a rate dependent on the ( 11 ) state ( blocked or non - blocked , with initial occupation probabilities @xmath107 and @xmath108 , respectively ) we have @xmath109 a sum of two exponential decays weighted by their probabilities . assuming the system spends an equal amount of time in the blocked and unblocked configurations ,
we would have @xmath110 , and @xmath111 . however , to allow for possible imbalance and offset , we keep all four parameters @xmath112 , and @xmath0 as fitting parameters .
the fit means we find the values of these parameters which minimize the following sum @xmath113 ^ 2(t_i - t_{i-1})^n,\ ] ] where @xmath114 is typically set to [email protected] .
the value @xmath116 corresponds to the standard least square fitting of the discrete set of @xmath117 , whereas we found it useful to introduce different values of @xmath114 to weight tunneling times according to their density .
this improves the fit at longer times where we have fewer tunneling events , without the necessity for fitting from a data histogram , which we found less robust to extract the slower tunneling rate .
the fitting error of the extracted times @xmath118 , defined by the standard confidence interval @xmath119 , was below 1 % therefore , to make the error bars visible on fig .
4 we multiplied them by a factor of 50 .
this means that the error from the fitting is negligible compared to other errors ( such as the potential drift due to the 1/f noise , the measurement errors due to noise in the charge sensor , etc . ) .
the extracted individual tunneling times , @xmath1 and @xmath0 , have less systematic dependence ( figs . [ fig_s2](a ) and
[ fig_s2](b ) taken for in- and out - of - plane @xmath72 , respectively ) , compared to fig . 4 where their ratios are plotted .
the fluctuations of individual tunneling times are caused by fluctuations of the inter - dot tunnel coupling , in turn due to the influence of implementing changes in @xmath72 , from deliberately changing gate voltages for other purposes in a related experiment,@xcite from the limited precision in setting the detuning @xmath12 to zero , and probably the most importantly from the drift of local voltages due to the 1/f noise ( it takes approximately 10 minutes to gather data for a single point in fig .
except for @xmath12 , the above mentioned noise effects do not appreciably influence the ratio @xmath29 , since there the inter - dot tunnel coupling cancels out .
the detuning is important if it , accidentally , matches the zeeman energy , at which value the @xmath5 anti - crosses with @xmath120 .
this lifts the psb of this spin - polarized state , and results in a relatively sharp dip in @xmath29 .
we took this effect into account by averaging over a distribution of @xmath12 , as described in the main text
. a different probability distribution of @xmath12 , or its changes in the measurement course , might be the reason for the discrepancy with the fit , and also for the remaining fluctuations in @xmath29 , which are much larger than the error bars obtained from the histogram fitting .
particularly , an accidental degeneracy of @xmath5 and @xmath120 might be the reason for the dip and large fluctuations of @xmath29 around @xmath121 t.
, and spin - flip - assisted , @xmath1 , tunneling .
values are extracted from bi - exponential curve fitting for ( a ) in - plane and ( b ) out - of - plane magnetic field .
taking the ratio @xmath122 results in data plotted in fig . 4 of the main text .
[ fig_s2],scaledwidth=45.0% ] here we give the explicit form of the hamiltonian of eq .
( 1 ) . denoting the @xmath123-th electron position within the 2deg plane as @xmath124 , and its momentum operator as @xmath125 , the quantum dot hamiltonian is @xmath126 with @xmath127 the effective mass , @xmath128 the elementary charge , @xmath129 the vacuum permittivity , @xmath130 the relative permittivity , and @xmath131 the vector potential of the out - of - plane field .
the confinement potential of a weakly coupled dot might be modeled , e.g. , by the bi - quadratic form @xmath132 with @xmath133 the orbital energy scale , and the vector @xmath134 parametrizing the double dot geometry , with @xmath135 the distance between the two potential minima ( the inter - dot distance ) , and @xmath136 the angle between the dot main axis and the [ 100 ] crystallographic direction ( close to either 45 or -45 degrees ) .
the remaining terms of the total hamiltonian , using the single electron notation , are the following .
the zeeman term is @xmath137 with @xmath25 the @xmath25-factor , @xmath138 the bohr magneton , @xmath21 the external magnetic field , and the electron spin operator @xmath139 is the vector of sigma matrices .
the spin - orbit interaction is @xmath140 with @xmath141 , and @xmath142 the spin - orbit lengths corresponding to the dresselhaus , and the bychkov - rashba interactions , respectively . finally , the fermi contact interaction is @xmath143 with @xmath144 the electron wave function in the perpendicular direction ( the lowest subband of the 2deg ) , and the index @xmath114 labels nuclear spins @xmath145 at positions @xmath146 , which have volume density @xmath147 .
the material parameters of gaas are @xmath148 @xmath149 , @xmath150 , @xmath141 and @xmath142 typically of order @xmath36 m , @xmath151ev , @xmath82 , and @xmath152 , with the lattice constant @xmath153 nm . for a weakly
coupled double dot , the low energy subspace is well approximated using a basis built from the lowest fock - darwin solutions localized in the two potential minima , @xmath154 with the magnetic field renormalized confinement length @xmath155^{-1/4}$ ] .
these can be orthonormalized to @xmath156 where we took the leading order in the left / right wave functions overlap , @xmath157 which parametrizes the inter - dot coupling .
for a weakly coupled dot , the overlap is much smaller than 1 .
we now specify the two - electron states that constitute the basis of the restricted hilbert space of the two electron hamiltonian . using the single electron orbitals @xmath158 , and similarly for @xmath159 , we construct the following five states @xmath160 [ eqs:2e basis ] where we separated the orbital and spin parts of the wave function , for both of which we use the notation where the order of single particle wave functions denotes the state of the first and second electron , respectively . in the spin part , @xmath161 and @xmath162 denote a spinor oriented parallel and antiparallel to the external field @xmath21 .
we can rewrite eq . in the form of the zeeman term ,
eq . , by introducing @xmath163 where the single electron wavefunctions @xmath164 are given in eq .
neglecting the small overlap @xmath165 , the two fields can be considered independent .
using this notation , @xmath39 enters through the average nuclear field @xmath166 as a contribution to the external magnetic field , generating the following matrix elements among the triplet states @xmath167 [ eqs : hnuc m.e .
1 ] while the field gradient @xmath168 generates matrix elements between the triplets and the singlet @xmath8 , @xmath169 [ eqs : hnuc m.e .
2 ] the singlet @xmath5 is coupled to other states by matrix elements which are , compared to those given , suppressed by a factor @xmath165 and can be neglected . to obtain the spin - orbit interaction matrix elements
, we first unitarily transform the total hamiltonian @xcite @xmath170 with a unitary representing a position dependent spin rotation @xmath171 defined by the vector @xmath172 expressed through the crystallographic coordinates @xmath173 , @xmath174 , and the spin - orbit lengths @xmath175 . in the leading order of the small quantity @xmath176 ,
the only effect of the transformation is to express the spin - orbit interaction in terms of the position , rather than momentum , operator .
that means , the transformed hamiltonian has the same form as the one in eq .
( 1 ) , if the spin - orbit integration is replaced by @xmath177 this form is better suited for a perturbative treatment of the spin - orbit effects , as was demonstrated in refs .
we note that the transformation leaves the statistics of the random nuclear fields unchanged to all orders of @xmath176 , as a local rotation of a quantization axis is irrelevant for unpolarized spins ( only the correlators of distant spins are different in the original and transformed frame @xcite ) . to calculate the matrix elements of the effective form of the spin - orbit interaction , eq .
, we will use the inversion symmetries of the functions in eq . .
namely , in the coordinate system where the @xmath178 axis is aligned with @xmath179 , and @xmath180 is perpendicular to it , these functions transform @xmath181 [ eqs : inversion symmetries ] where @xmath182 is an inversion of the coordinates @xmath183 and analogously for @xmath180 , @xmath184 is a complex conjugation , and the dot coordinate system is defined by the angle @xmath136 of the vector @xmath179 and the crystallographic @xmath173 axis by @xmath185 [ eqs : dot coordinates ] using eq .
it is simple to show that the only non - zero dipole matrix elements within the single electron basis functions we consider are @xmath186 so that the orbital effects of the magnetic field are absent , apart from a small renormalization of the inter - dot overlap @xmath165 , which cancels in @xmath29 anyway . using this result
gives that the only non - zero matrix element of the effective spin - orbit interaction , eq .
, within the subspace of states in eq .
is ( up to an overall phase coming from the phase convention of the spinors @xmath161 , @xmath162 ) @xmath187 this gives eq .
( 2 ) upon identifying @xmath188 where we used eqs . , and denoted the angle of the magnetic field with the @xmath173 axis as @xmath189 .
this completes the derivation of eq .
as described in the main text , since the form of the superoperator @xmath190 in eq .
( 3 ) is invariant to basis changes within the ( 11 ) sector , we choose a basis that diagonalizes the hamiltonian in this sector .
this diagonalization results in a set of states @xmath47 , that are defined by @xmath191 , with @xmath192 the projector on the ( 11 ) subspace .
it also results in off - diagonal elements given in eq .
( 4 ) , which we treat perturbatively . assuming that initially the system is in such a state @xmath47 , we can neglect all other @xmath193 states in the lowest order of the time dependent perturbation theory , since the system can not make a direct tunneling @xmath194 , as there is no matrix element connecting these states .
the only possible tunneling is to @xmath5 , the rate of which is described by a 2 by 2 time dependent hamiltonian @xmath195 in this notation , we define @xmath196 as the time average of the energy difference , and @xmath197 its temporal fluctuation , with spectral density defined analogously to eq .
( 6 ) , @xmath198 assuming that the system is initially in state @xmath47 , and treating the off - diagonal terms in eq . as a perturbation , in the leading order of a time dependent perturbation
theory we get @xmath199 where we assumed @xmath200 ( that is , neglecting the initial oscillations which do not grow with time for long times ) , and @xmath201 with @xmath202 standing for principal value of the integral , and the approximation valid if the function @xmath203 is smooth on the scale @xmath204 .
similarly @xmath205 if we assume that the principal value term is negligible , the transition rate from state @xmath47 to @xmath48 given in eq .
( 5 ) , follows as @xmath206 from eq . , with the charge decoherence rate given as the spectral density of energy fluctuations at zero frequency , eq . .
we also considered alternative explanations for the behavior of the observed rates .
first of all , one might imagine that once the blocked tunneling becomes too slow , some other mechanism , unrelated to hyperfine or spin - orbit interactions and thus to the magnetic field , takes over in the electron spin relaxation .
however , cotunneling , as the only realistic candidate for such a process we are aware of , is expected to result in rates well below the dot - lead rate of 1 hz which we measured in our dot .
further , the data of fig .
[ fig_s2](b ) show that the value of the blocked tunneling time is not correlated at all with the appearance and disappearance of the plateau in the ratio of the rates .
second , we can extend our model to incorporate inelastic tunnelings by adding a noise term , e.g. , as a time dependent @xmath207 , into eq .
it results in a contribution to the last term in eq .
( 5 ) , through the spectral density @xmath208 , which is in general of a different functional form than the considered @xmath209 . before doing any quantitative estimates , we note that to explain the plateau in the ratio of the rates , the spectral density @xmath208 would have to be quadratic ( assuming the decoherence is low enough such that the lorentzian denominators cancel ) .
we do not deem this to be a probable explanation , since neither of the electric noise sources which are considered relevant for spin relaxation has such spectral density .
namely , the spectral density of phonons is cubic ( piezoelectric ) or @xmath210 ( deformation potential ) , the 1/f noise is inversely proportional to @xmath211 and the shot and thermal noise are frequency independent .
in addition , the dipolar moment of the fluctuating electric field does not couple s(11 ) and s(02 ) .
it only contributes to the decoherence , which our model already takes into account .
we expect the sub - dominant ( beyond the dipole ) moments of the noise fields to be very small .
third , we examined the possibility that such additional noise induces inelastic transitions within the ( 11 ) sector .
a blocked triplet @xmath6 in such a scenario first decays to the s(11 ) and only then tunnels to s(02 ) , instead of making a spin - flip - assisted tunneling directly to s(02 ) .
however , as derived in detail in the following subsection , we find that this leads to a decay described by @xmath212 , which we could clearly distinguish from an exponential decay @xmath213 which we observe for blocked tunnelings . to examine the effects of inelastic transitions
, we consider a rate equations model for three states , a polarized triplet denoted as @xmath47 , and the two singlets , @xmath8 and @xmath5 .
the corresponding occupation probabilities are denoted as @xmath215 , @xmath216 , and @xmath217 , respectively .
they evolve according to our minimal model therefore includes a decay of the triplet state into @xmath8 with relaxation rate @xmath219 and an unbiased equilibration of the two singlets with the rate @xmath220 , corresponding to the direct inter - dot tunneling @xmath221 .
let us assume that at time @xmath105 the @xmath5 was not occupied , @xmath222 . for time interval @xmath18 short enough , @xmath223 , eqs .
are solved by now we supplement this evolution by an assumption that the detector measures the system repeatedly with a rate ( bandwidth ) @xmath226 , projecting it between @xmath16 and @xmath15 subspaces . assuming the system started in ( 11 ) subspace , the probability that no tunneling is observed up to time @xmath105 , @xmath227 , is governed by the equation
@xmath228 \delta t,\ ] ] which can be solved , assuming again @xmath229 , and the initial condition being the state @xmath47 occupied , @xmath230 , to give @xmath231 the condition to observe the system in a blocked configuration is @xmath232 , otherwise the blocked " triplet would decay to @xmath8 before it could be observed as being blocked . for times
@xmath233 we then get from the previous equations the result stated in the main text @xmath234 the histograms of ( 11 ) residing times on fig . 3 correspond to the probability that the tunneling happens within some interval . according to the previous analysis the number of observed tunnelings within @xmath235 )
is given by @xmath236 the normalization constant @xmath237 is the number of blocked tunnelings , equal to the total time the system spends in a blocked ( 11 ) state @xmath238 divided by @xmath239 an average time to tunnel out from such a state .
this is to be compared with the initial state of the system in an unblocked ( 11 ) state .
this situation is described by eqs . with the state @xmath47
never occupied and thus disregarded from considerations .
an analogous calculation gives @xmath240 with @xmath241 .
assuming the system spends an equal amount of time in a blocked as in an unblocked state , @xmath242 , we can simulate a histogram plot assuming , for illustration , @xmath214 , a typical ratio we observe on the plateau .
we get the data plotted on fig .
[ fig_s3 ] . comparing with fig .
3 , it demonstrates that we can exclude the possibility of inelastic transitions being relevant in our experiment .
13 ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1 '' '' @noop [ 0]secondoftwo sanitize@url [ 0 ] + 12$12 & 12#1212_12%12 @startlink[1 ] @endlink[0 ] @bib@innerbibempty link:\doibase 10.1088/0268 - 1242/12/4/006 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevlett.100.046803 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevlett.110.266803 [ * * , ( ) ] link:\doibase 10.1103/physrevb.90.085306 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) | we detect in real time inter - dot tunneling events in a weakly coupled two electron double quantum dot in gaas . at finite magnetic fields ,
we observe two characteristic tunneling times , @xmath0 and @xmath1 , belonging to , respectively , a direct and a blocked ( spin - flip - assisted ) tunneling .
the latter corresponds to lifting of a pauli spin blockade and the tunneling times ratio @xmath2 characterizes the blockade efficiency .
we find pronounced changes in the we explain this behavior as due to the crossover of the dominant blockade lifting mechanism from the hyperfine to spin - orbit interactions and due to a change in the contribution of the charge decoherence .
electron spins in semiconductor quantum dots are promising resources for quantum information processing @xcite .
laterally gated dots @xcite are especially attractive due to the flexibility and scalability @xcite of their design , and the possibility to electrically initialize @xcite , manipulate @xcite , and measure @xcite the @xcite spin states .
( psb ) @xcite plays a crucial role in electrical manipulations .
psb is established when the conservation of spin blocks a transition from an excited state , where two electrons in two dots have parallel spins , to the ground state , where they form a singlet in one dot .
the spin can thus be detected by a local charge sensor as the presence or absence of a charge transition @xcite .
the blockade is lifted by , limiting the readout fidelities @xcite , as well as manipulations and preparations of quantum states.@xcite in gaas quantum dots , there are two important sources of spin flips : the spin - orbit coupling , and the hyperfine interaction with spins of atomic nuclei . respectively
, they dominate the spin relaxation time @xmath3@xcite and decoherence time @xmath4 @xcite . apart from ,
both of these can be utilized in quantum state manipulation as a means of coupling of the electrical control fields to spins @xcite .
it is known that the relative importance of these two effects changes with magnetic field strength and orientation @xcite . by resolving the direct and spin - flip - assisted inter - dot tunneling , here we investigate experimentally the limit these factors impose on the effectiveness of psb .
we find a crossover in their dominance upon changing the magnetic field strength , which is fully consistent with our theory .
our results give guidance on how to increase the psb effectiveness with importance for spin readout applications . , scaledwidth=45.0% ] our device is a gate defined lateral double quantum dot ( dqd ) , fig .
[ fig_1](a ) , weakly tunnel - coupled and isolated from reservoirs , with lead - dot tunneling rates of order hz , and the inter - dot tunneling rate of order khz . in this regime , where tunnel coupling energies are much smaller than orbital or charging energies , the two electron configurations span a basis of five states @xcite : one ( 02 ) charge state , the singlet @xmath5 , and four ( 11 ) charge states , the two spin polarized triplets @xmath6 , the unpolarized triplet @xmath7 , and the singlet @xmath8 . here by @xmath9 we denote the left and right dot occupancies as @xmath10 and @xmath11 , respectively . since the exchange energy splitting among the ( 11 ) charge states is negligible , the four ( 11 ) states are degenerate and , in general , energetically separated from the @xmath5 state by the detuning energy @xmath12 .
the nearby charge sensor can discriminate different charge states @xcite .
using gates l and r , we tune the dot close to the ( 11)-(02 ) degeneracy , @xmath13 , by balancing the time averaged occupations of the two charge configurations , and measure the sensor current @xmath14 . with the inter - dot tunneling time set above the time resolution of the sensor ,
we monitor in this way the dot charge configuration in real time .
, scaledwidth=45.0% ] figure 2(a ) shows random ( thermally excited ) switching of charge configurations @xcite at zero magnetic field .
the histogram of @xmath15 to @xmath16 tunneling times plotted on the upper left panel of fig .
3 shows that the tunneling is described by a single time constant @xmath17 , with the probability that no tunneling occurs for time @xmath18 being @xmath19 . despite different spin configurations of the ( 11 ) states
, a single tunneling rate into the ( 02 ) singlet is expected due to the hyperfine field of nuclear spins . indeed , if described as a zeeman term of a slowly fluctuating classical magnetic field located in the left(right ) dot @xmath20 @xcite , these quasi - static random fields in general couple all five states .
though the couplings between the ( 11 ) and ( 02 ) states are negligible , they are appreciable among the ( 11 ) states ( see below ) . as a consequence , no matter in which ( 11 ) spin state the system starts , within a few nanoseconds it contains the @xmath8 state with amplitude of order 1 from where it can tunnel to @xmath5 . as a result , within our time resolution , all ( 11 ) states tunnel out with the same rate and there is no psb .
the charge switching behavior is different at a large enough external magnetic field @xmath21 , see fig .
2(b ) for @xmath22 t. in addition to the fast switching fig .
2(a ) , there are long intervals where the system remains in a ( 11 ) state .
this is the pauli spin blockade : once @xmath23 , the zeeman energy offset of the polarized triplets suppresses their hyperfine induced admixture with @xmath8 and by that their tunneling to @xmath5 . since @xmath7 still mixes fast with @xmath8 , we expect to see two tunneling times , @xmath1 for spin - flip - assisted tunneling of spin polarized states , and @xmath0 for direct tunneling of spin unpolarized states .
the two processes are sketched on fig [ fig_1](b ) , and the histograms plotted in fig .
3 indeed show bi - exponential distributions for @xmath24 mt . , and long , @xmath1 , times for finite fields ( see sm for details of the fitting procedure ) .
[ fig_3],scaledwidth=45.0% ] we have investigated this psb manifestation as a function of the magnetic field strength . in search for generic features , we measured for various sample cooldowns , which can change the shape of the dot , and for both in - plane and perpendicular magnetic fields , by which we isolate the strong orbital effects of the latter .
due to the influence of the algaas barriers @xcite , the @xmath25-factor is small , @xmath26 for an out - of - plane magnetic field , and , as we find from the analysis below , about five times smaller for in - plane fields .
( these small values make it easier to analyze the behavior of the rates at small zeeman energies , which are pushed to higher magnetic fields by the small @xmath25-factors . ) because measurement conditions , there is little systematic dependence of the tunneling times taken individually ( see fig .
s2 in ref .
@xcite ) .
this is mostly due to the exponential sensitivity of the @xmath27 tunneling matrix element @xmath28 , which is hard to keep constant during the re - alignement of the states energies required in the measurement course .
however , in plotting the ratio @xmath2 as in fig .
[ fig_4 ] , @xmath28 drops out and a clear trend emerges . from zero to moderate external fields
, @xmath29 behaves as expected : initially equal to one , it offsets once the external field becomes larger than the nuclear field . here
it grows as @xmath30 , as is well known @xcite and can be understood from a simple perturbation theory ( see below ) . at higher fields , however , we find that the growth stops and @xmath29 saturates .
this is not completely unnatural , as it suggests that the psb effectiveness is limited by some process , expected to be eventually the case .
however , moving to even higher fields , @xmath29 increases again .- factor , and the corresponding zeeman energy , is five times larger than for the in - plane field . ]
this is , however , completely surprising , as it implies that the limitation disappears .
it is also at odds with the general behavior of the spin ( inelastic ) relaxation time @xmath3 between zeeman split states , which is known to decrease with the magnetic field ( as @xmath31 ) as was predicted in theory and confirmed experimentally @xcite .
to understand this peculiar behavior , it is necessary to consider several ingredients . to this end
, we consider a hamiltonian comprising the part defining the double dot @xmath32 , and the interactions with the external , hyperfine , and spin - orbit fields @xcite : @xmath33 with details in ref . , @xmath32 consists of the electron kinetic energy , confinement potential , and coulomb interaction and defines the hilbert space as described above , with the ( 11 ) states detuned from @xmath5 by @xmath12 and the two singlets tunnel coupled by @xmath34 .
the spin - polarized triplets are offset by the zeeman energy @xmath35 , which for our @xmath25-factors corresponds to 7 @xmath36ev for @xmath37 t and 0.7 @xmath36ev for @xmath38 t. a typical matrix element of @xmath39 within the ( 11 ) subspace is of order 0.1 @xmath36ev .
finally , assuming @xmath40 contains the linear - in - momentum dresselhaus and rashba terms , the only non - zero matrix elements are @xmath41 here , @xmath42 is an effective spin - orbit length , a combination of the dresselhaus and rashba coefficients , and @xmath43 is half of the interdot distance , which we estimate to be 130 nm from typical values of @xmath0 @xcite . ,
scaledwidth=45.0% ] the system dynamics is given by the equation for its density matrix @xmath44 , @xmath45 + l [ \rho ] .
\label{eq : rho}\ ] ] the last term is due to charge noise , usually dominated by fluctuating electric fields of phonons , gate potentials , impurities , and the charge sensor current .
it leads to a fast decay of charge superpositions , with the decoherence rate @xmath46 , typically several ghz @xcite . denoting a ( 11 )
state as @xmath47 , and @xmath5 as @xmath48 , the charge decoherence is described by @xmath49)_{xs } = -\gamma \rho_{xs}$ ] , a form independent of the ( 11 ) subspace basis .
we use this invariance to simplify eq . by choosing basis states @xmath50
in which @xmath51 is diagonal within the ( 11 ) subspace . because of the hyperfine and spin - orbit couplings , these eigenstates are in general superpositions of all four ( 11 ) states . in this basis ,
the remaining off - diagonal matrix elements are the tunneling terms @xmath52 which are much smaller than the states energy differences and can be treated perturbatively . in the leading order , the dynamics given by eq .
reduces to transitions between @xmath47 and @xmath48 with the rate @xmath53 for further convenience , we introduce the spectral density of the transition matrix element @xmath54 for a time independent hamiltonian , which we consider at the moment , @xmath55 . inserting this into eq .
gives fermi s golden rule formula with the initial and final state difference @xmath56 , lorentzianbroadened by the decoherence .
we fit this model to the data on fig .
[ fig_4 ] by averaging the rates given by eq . over hyperfine fields @xmath57 , assuming the latter having a gaussian probability distribution with zero mean value and dispersion @xmath58 .
the typical hyperfine field @xmath59 , the charge decoherence rate @xmath46 , and the spin - orbit length @xmath42 are fitting parameters . to take into account the limited accuracy of the detuning and the voltage jitter present in real experiments , we average over @xmath60ev , a range corresponding to the electron temperature.ev ) , and device stability . a small offset @xmath61 is expected because of the fourfold difference in the spin degeneracy of the ( 11 ) and ( 20 ) charge states .
] however , the influence of @xmath12 on the data fit in fig .
4 is relatively minor . on the contrary , @xmath59 , @xmath46 , and
@xmath42 have a profound influence and the energy scales connected to these key quantum dot spin qubit parameters can be directly read off from the magnetic field dependence of @xmath29 , as we now explain .
let us first assume the detuning is zero , and the hyperfine fields are fixed . using eqs .
we get the ratio of tunneling times of two ( 11 ) states @xmath62 , @xmath63 as @xmath64 take first @xmath65 .
as already explained , the hyperfine fields fully mix the ( 11 ) subspace , so that each eigenstate typically contains the same amount of admixture of @xmath8 . because in addition the zeeman energy of hyperfine fields is negligible compared to @xmath66 , the ratio in eq .
is 1 .
once @xmath67 , the singlet admixture into the zeeman split triplets is small , @xmath68 .
the remaining two states , typically equally mixed @xmath8 and @xmath7 , have @xmath69 .
this gives two tunneling times with the ratio proportional to @xmath70 .
the of @xmath29 at higher fields ( around 0.1 t for @xmath71 ) can be understood as the spin - orbit taking over the hyperfine field in the matrix element in eq . .
indeed , whereas the latter is independent of @xmath72 , the former grows linearly , see eq . .
this equation also gives the spin - orbit length as @xmath73 with @xmath29 the ratio on the plateau .
increasing the magnetic field further ( beyond 1 t for @xmath71 ) , @xmath29 starts to grow again in fig .
4 . this can be still reconciled with eq .
, as due to the first fraction on its right hand side .
namely , once the zeeman energy becomes larger than the decoherence , the spectral overlap of spin - polarized @xmath15 states and @xmath5 drops compared to spin - unpolarized @xmath15 states .
the zeeman energy where @xmath29 starts increasing for the second time gives therefore the charge decoherence rate @xmath66 .
the three energy scales extracted visually as just described from the slope changes of @xmath29 give the values of parameters @xmath59 , @xmath42 , and @xmath46 within a factor of order one .
we found that the best way to nail down these factors quantitatively is straightforward numerics .
namely , for given values of hyperfine fields , we diagonalize the @xmath74 hamiltonian in the ( 11 ) subspace numerically , and calculate the rates according to eq . .
we average these over typically 10@xmath75 hyperfine fields random configurations .
because we can not distinguish experimentally all four rates , we define in our numerics the `` blocked(direct ) '' rate as the average of the first(last ) two rates ordered by their magnitudes . in this way
, we obtain the solid lines in fig .
[ fig_4 ] using @xmath76ev , @xmath77 ghz , and @xmath78 m for the out - of - plane field , and @xmath79 m for the in - plane field . from the value of @xmath59
, we can infer the number of nuclei within the dot volume @xcite , @xmath80 , using @xmath81 @xmath36ev , and @xmath82 .
all extracted values are typical for gated dots in gaas , that is @xmath83 @xcite , the charge decoherence rate @xcite , and spin - orbit lengths @xcite .
we note that the different values of the effective spin - orbit length fitted for in - plane and out - of - plane magnetic fields are consistent with directional anisotropies of @xmath42 @xcite , observed in dot spectra @xcite , and spin relaxation @xcite .
we also considered alternative explanations , examining inelastic ( 11 ) to ( 02 ) transitions due to a non - dipolar electric noise , inelastic ( @xmath3 ) transitions within the ( 11 ) subspace , and lifting the spin - blockade by cotunneling .
as none of these can be naturally reconciled with the data , we give these details only in ref . @xcite .
we conclude by suggesting how to increase the psb effectiveness .
the spin - orbit effects should be minimized , what can be achieved by orienting the magnetic field along certain in - plane directions @xcite , specified by setting @xmath84 in eq .
( s23 ) in ref .
@xcite .
we predict that the quadratic growth @xmath85 will then extend to much higher fields and increase to @xmath86 once the zeeman energy becomes larger than the charge decoherence rate .
finally , these properties are to a large extent independent of the value of the interdot tunneling , which should therefore boost both direct and blocked rates while preserving their ratio .
this work was supported by grant - in - aid for scientific research s ( 26220710 ) and a ( 25246005 ) , innovative area ( 26103004 ) , crest , jst , impact program , the iarpa project `` multi - qubit coherent operations '' through copenhagen university , mext project for developing innovation systems and qpec , the university of tokyo .
t. f. & h. k. are supported by jsps research fellowships for young scientists .
p.
s. acknowledges the support of apvv-0808 - 12(qimabos ) .
a.l . and a.d.w .
acknowledge gratefully support of dfg - trr160 and bmbf-q.com-h 16kis0109 .
49 ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1 '' '' @noop [ 0]secondoftwo sanitize@url [ 0 ] + 12$12 & 12#1212_12%12 @startlink[1 ] @endlink[0 ] @bib@innerbibempty @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevlett.53.493 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1109/tnano.2003.820516 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) http://link.aps.org/supplemental/10.1103/physrevlett.xxx.yyyyyy[]@noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop ( ) * supplemental information * here we give details on device characteristics , additional measurements , and the derivation of formulae . in sec .
i we give additional information on the device , comment on the @xmath25-factor anisotropy , and spontaneous ( inelastic ) spin flip by phonon emission . in sec .
ii we show measurement data demonstrating pauli spin blockade in a configuration where there is a voltage bias across the dot . in sec .
iii we describe the fitting procedure used to extract the two tunneling times @xmath1 and @xmath0 , resulting in the bi - exponential fits shown in fig . 3 of the main text .
the rest contains details on the theory . in sec . iv we give the explicit form of the two electron double dot hamiltonian , the restricted hilbert space , and the corresponding matrix elements of the hyperfine and spin - orbit interactions . in sec .
v we derive eq .
( 5 ) . in sec .
vi we comment on alternative explanations of the data we considered . |
the physics of graphene monolayers continues to attract a lot of attention and to provide a rich source of interesting phenomena.@xcite by studying the effects of the spin - orbit interaction ( soi ) in a graphene layer , where symmetry allows for an `` intrinsic '' ( @xmath0 ) and a `` rashba '' ( @xmath1 ) term in the soi , kane and mele@xcite made a remarkable discovery that sparked the exciting field of topological insulators:@xcite for @xmath2 , there is a bulk gap with topologically protected edge states near the boundary of the sample .
this is similar to the quantum hall ( qh ) effect but happens in a time - reversal invariant system .
the resulting `` quantum spin hall '' ( qsh ) edge states form a one - dimensional ( 1d ) helical liquid , where right- and left - movers have opposite spin polarization and spin - independent impurity backscattering is strongly suppressed .
the qsh state has been observed in hgte quantum wells,@xcite but several works@xcite showed that @xmath0 is probably too small to allow for the experimental verification of this novel phase of matter in pristine graphene .
consequently , other material classes have been employed to demonstrate that topologically insulating behavior is indeed possible.@xcite however , graphene experiments@xcite have also demonstrated that the rashba coupling @xmath1 can be increased significantly by depositing graphene on ni surfaces .
moreover , very recent theoretical predictions@xcite suggest that already moderate indium or thallium adatom deposition will dramatically enhance @xmath0 by several orders of magnitude . by using suitable adatoms ,
it is expected that in the near future both soi parameters @xmath0 and @xmath1 can be varied over a wide range in experimentally accessible setups . in view of these developments , in this paper
we study the electronic properties of a graphene monolayer with artificially enhanced soi . besides the soi
, we include piecewise constant electrostatic potentials , orbital and zeeman magnetic fields , and strain - induced vector potentials .
the latter cause pseudo - magnetic fields but do not violate time reversal invariance ; for a review , see ref . .
while the interplay of the rashba term @xmath1 with ( pseudo-)magnetic fields in graphene has been studied in several theory works before,@xcite the intrinsic soi @xmath0 did not receive much attention so far .
however , the transmission properties of graphene s dirac - weyl ( dw ) quasiparticles through barriers with arbitrary soi have been studied recently@xcite in the absence of ( pseudo-)magnetic fields .
the structure of this article is as follows . in sec .
[ sec2 ] we formulate the model and construct the general solution for piecewise constant fields . on top of the orbital magnetic field , we allow for arbitrary soi parameters @xmath0 and @xmath1 , zeeman energy @xmath3 , and we also take into account aspects of strain - induced fields .
the homogeneous case is addressed in sec .
[ sec3 ] , where we determine the landau level states for this problem in closed and explicit form .
in particular , the fate of the zero modes residing at the dirac point ( energy @xmath4 ) will be discussed in the presence of the soi .
our results also apply to the case of a strain - induced homogeneous pseudo - magnetic field.@xcite next , in sec .
[ sec4 ] we study edge states near the boundary of a semi - infinite sample for vanishing rashba coupling , @xmath5 . for weak magnetic fields , one then expects to have helical ( spin - filtered ) qsh edge states .
interestingly , at the dirac point , upon increasing the magnetic field , we find that a quantum phase transition takes place between the qsh phase and a second qsh - like phase with spin - filtered edge states , considered previously by abanin _
et al._,@xcite where the spin current direction is reversed .
this spin current reversal should allow for an experimental detection of this quantum phase transition , on top of the obvious consequences for qh quantization rules.@xcite in sec .
[ sec5 ] , we turn to a mesoscopic waveguide geometry , where a suitable inhomogeneous magnetic field ( or exchange field produced by lithographically deposited ferromagnetic films ) defines the waveguide.@xcite we show that the soi parameters @xmath0 and @xmath1 give rise to interesting spin texture of the resulting propagating chiral states in such a waveguide . finally , we conclude in sec .
unless many - body effects are of crucial importance , the low - energy electronic properties of a graphene monolayer are well captured by two copies of a dw hamiltonian supplemented with various terms describing soi , ( pseudo-)magnetic fields , and electrostatic potentials.@xcite the wavefunction corresponds to a spinor comprising eight components , @xmath6 the pauli matrices @xmath7 below act in sublattice space corresponding to the two carbon atoms ( @xmath8 ) in the basis of the honeycomb lattice , while pauli matrices @xmath9 act in physical spin ( @xmath10 ) space .
finally , the valley degree of freedom ( @xmath11 ) corresponds to the two @xmath12 points@xcite and pauli matrices @xmath13 refer to that space .
specifically , we here consider models where the mentioned extra terms in the hamiltonian are piecewise constant along the @xmath14-direction and homogeneous along the @xmath15-axis . consequently , the momentum @xmath16 is conserved , and we have an effectively 1d problem in terms of the four - spinors @xmath17 . the orbital magnetic field @xmath18 ( with @xmath19 and @xmath20 ) is expressed in terms of the vector potential @xmath21 , where we choose the gauge @xmath22 inclusion of the constant @xmath23 is necessary when connecting regions with different magnetic fields in order to make @xmath24 continuous . assuming that the magnetic field is perpendicular to the graphene sheet , the zeeman field couples to @xmath25 and determines the coupling constant @xmath26 , where @xmath27 is the land factor and @xmath28 denotes the bohr magneton .
the full hamiltonian then reads@xcite ( @xmath29 ) @xmath30 \\ \label{ham } & + & v+ \epsilon b s_z + \frac{\lambda}{2 } ( \sigma_x s_y \tau_z -\sigma_y s_x)+ \delta \sigma_z s_z \tau_z.\end{aligned}\ ] ] in eq .
( [ ham ] ) @xmath31 is the conserved momentum in the @xmath15-direction , while @xmath32 is still an operator .
the constant @xmath23 in eq . can be included by shifting @xmath16 , and we suppose that this shift has been carried out in the remainder of this section .
the fermi velocity is @xmath33 m@xmath34s , while the soi couplings @xmath0 and @xmath1 ( both are assumed non - negative ) correspond to the intrinsic and rashba terms , respectively . in wrinkled graphene sheets
the coupling @xmath1 also captures curvature effects.@xcite a constant electrostatic potential , @xmath35 , has been included in eq . .
strain - induced forces@xcite lead to a renormalization of @xmath35 as well as to the appearance of an effective vector potential , @xmath36 expressed in terms of the in - plane strain tensor @xmath37 , see ref .
constant @xmath38 can be found in refs . .
as discussed by fogler _
et al._,@xcite in many cases it is sufficient to consider a piecewise constant strain configuration . assuming that the @xmath15-axis is oriented along the zig - zag direction , strain causes only a finite but constant @xmath39 while @xmath40
this can be taken into account by simply shifting @xmath16 in this region .
below we suppose that also this shift has already been done .
estimates for @xmath39 in terms of physical quantities can be found in refs . .
the resulting pseudo - magnetic field then consists of @xmath41-barriers at the interfaces between regions of different strain .
an alternative situation captured by our model is given by a constant pseudo - magnetic field , whose practical realization has been described recently.@xcite in that case , @xmath39 is formally identical to @xmath24 in eq . .
unless specified explicitly , we consider the case of constant @xmath39 below .
let us briefly comment on the symmetries of this hamiltonian . in position representation ,
the time reversal transformation is effected by the antiunitary operator@xcite @xmath42 with complex conjugation operator @xmath43 and implies the relation @xmath44 for @xmath45 in eq . with @xmath46 .
since @xmath45 is diagonal in valley space , eq .
( [ trs1 ] ) implies that the hamiltonian @xmath47 near the @xmath48 point is related to @xmath49 by the relation @xmath50^ * s_y.\ ] ] by solving the eigenvalue problem at the @xmath12 point , we could thus obtain the eigenstates at @xmath48 via eq . .
a simpler way to achieve this goal is sketched at the end of this subsection .
from now on we switch to dimensionless quantities by measuring all energies in units of the cyclotron energy @xmath51 , where we define @xmath52 .
the magnetic length @xmath53 sets the unit of length .
a field of 1 tesla corresponds to @xmath54 mev and @xmath55 nm . measuring @xmath56 in units of tesla , we get for the zeeman coupling @xmath57}$ ] . with the dimensionless coordinate @xmath58 and the auxiliary quantities @xmath59 we find the representation @xmath60 here we introduced the standard ladder operators @xmath61 with @xmath62=1 $ ] .
according to the above discussion , eigenstates at the @xmath48 point for @xmath63 could be obtained from the corresponding solutions at the @xmath12 point with @xmath64 .
alternatively , there is a simpler way to obtain the @xmath48 states as follows .
the 1d hamiltonians @xmath65 ( for given @xmath66 ) can be written in dimensionless notation as @xmath67 both hamiltonians are therefore related by the transformation @xmath68 without the need to invert the real magnetic field since this is not a time reversal transformation . as a consequence ,
the 1d eigenstates @xmath69 follow from the solutions at the @xmath12 point by multiplying with @xmath70 and inverting the sign of @xmath71 , @xmath72 we now determine the spinors @xmath73 solving the dw equation for energy @xmath74 , @xmath75 with @xmath76 in eq . .
we construct the solution to eq . within a spatial region where all parameters ( magnetic fields , strain , soi , etc . )
are constant but arbitrary .
this general solution will be employed in later sections , where specific geometries are considered by matching wavefunctions in adjacent parts . now
eq . is a system of four coupled linear differential equations that admits precisely four linearly independent solutions derived in app .
[ appa ] . in order to solve eq .
, it is instructive to realize that the parabolic cylinder functions,@xcite @xmath77 , obey the recurrence relations @xmath78 with the ladder operators @xmath79 in eq . .
similar relations for @xmath80 or @xmath81 are given in app .
[ appa ] . for given energy @xmath74 ,
the order @xmath82 can only take one of the two values @xmath83,\ ] ] where we define [ cf . eq . ]
@xmath84 for each of the two possible values for @xmath82 , we then have two basis states , @xmath85 and @xmath86 , which results in four linearly independent solutions .
we show in app .
[ appa ] that the ( unnormalized ) solution @xmath85 can be chosen as @xmath87 while @xmath86 is taken in the form @xmath88 next , we analyze the spatially homogeneous case .
in this section we study an unstrained infinitely extended graphene monolayer where the magnetic field @xmath89 ( we assume @xmath90 ) and the soi parameters @xmath0 and @xmath1 are constant everywhere .
( the electrostatic potential @xmath35 just shifts all states and is set to zero here . )
we are thus concerned with the relativistic landau level structure for graphene in the presence of arbitrary soi parameters , including also the zeeman field @xmath3 .
this problem was solved for the special case @xmath91 by rashba,@xcite see also ref . , and
below we reproduce and generalize this solution .
we focus on the @xmath12 point only , since the spectrum and the eigenstates at the @xmath48 point follow from eqs . and .
we also allow for a constant pseudo - magnetic field .
when only an orbital or a strain - induced pseudo - magnetic field is present but not both , each energy level below has an additional twofold valley degeneracy . in the homogeneous case , normalizability of the spinors @xmath85 [ eq . ] can only be satisfied if the order @xmath82 is constrained to integer values @xmath92 , while the @xmath86 [ eq . ] are not normalizable .
solutions for the homogeneous problem thus have to be constructed using @xmath85 only . expressing the energy @xmath74 ( we remind the reader that here all energy scales are measured in units of @xmath93 ) in terms of @xmath82 [ eq . ] , the sought ( valley - degenerate ) landau levels follow as the roots of the quartic equation @xmath94 \left[(e - b)^2-(p+\delta^2)\right]= \\ & & = \label{quartic } \lambda^2 \left [ ( e-\delta)^2-b^2\right].\end{aligned}\ ] ] for @xmath95 this recovers the standard relativistic spin - degenerate landau levels,@xcite @xmath96 for @xmath97 ( with @xmath98 for spin up and @xmath99 for spin down states ) , plus a spin - degenerate zero mode @xmath100 ( for @xmath101 ) .
we notice from eq .
that for @xmath102 , the combination of @xmath0 and @xmath1 breaks particle - hole symmetry , while the two couplings individually keep it .
furthermore , zero - energy solutions are generally not possible except for special fine - tuned parameters .
eq . also predicts that if @xmath74 is a solution for the parameter set @xmath103 then @xmath104 is a solution for the set @xmath105 .
the @xmath106 thus represent landau level states in the presence of soi and zeeman coupling .
the normalization constant @xmath107 , entering as a prefactor in eq .
, can be computed analytically since @xmath77 can be expressed in terms of hermite functions for integer @xmath82.@xcite for @xmath108 , we find @xmath109 . \end{aligned}\ ] ] remarkably , for @xmath110 , we find the _ exact _ normalized state for arbitrary system parameters , @xmath111 with the eigenvalue @xmath112 this unique admissible eigenstate for @xmath110 is endowed with full spin polarization in the @xmath113 direction . for @xmath114 ,
the secular equation ( [ quartic ] ) becomes effectively a cubic equation : the solution @xmath115 ( i.e. , @xmath116 ) does not correspond to any admissible eigenstate . the three allowed states are described by @xmath117.\end{aligned}\ ] ] this includes a `` zero - mode '' partner of the @xmath110 state , plus a pair of states obtained by mixing the spin - up @xmath118 and spin - down @xmath119 landau orbitals via the rashba soi . for @xmath91 but allowing for a finite rashba soi parameter @xmath1 , eq .
admits a simple solution , previously given in ref . and briefly summarized here for completeness .
for @xmath110 we have the solution ( [ p-1 ] ) , which now is a zero mode , while for @xmath120 , the eigenenergies are given by @xmath121^{1/2},\end{aligned}\ ] ] with @xmath122 . according to our discussion above , here @xmath123
should be counted only once , with eigenstate @xmath124 , while @xmath125 correspond to a particle / hole pair of first landau levels modified by the rashba soi , with eigenstates @xmath126 .
we thus get precisely two zero - energy states . for small @xmath1
, we find the expansion @xmath127 which shows that the states @xmath128 and @xmath129 , which form a degenerate landau level for @xmath5 , are split by a finite @xmath1 .
let us next consider the case @xmath5 , where one has a qsh phase@xcite for @xmath130 and @xmath131 .
now the hamiltonian is block diagonal in spin space and the eigenstates become quite simple even for finite zeeman coupling , since we can effectively work with the bi - spinors @xmath132 for spin @xmath133 .
we easily obtain the ( unnormalized ) eigenstates with @xmath134 in the form@xcite @xmath135 where the eigenenergies follow from eq .
, @xmath136 we employ the notation @xmath137 for @xmath114 , the second index in @xmath138 and @xmath139 should be replaced by @xmath140 , i.e. , there is only one solution for given spin ( and valley ) .
note that @xmath141 in the present notation corresponds@xcite to the solution .
when @xmath102 , interestingly enough , @xmath0 does _ not _ lift the spin degeneracy of the landau levels except for the zero mode ( @xmath114).@xcite a zeeman term with @xmath142 restores a true doubly - degenerate zero - energy state for @xmath114 again . in sec .
[ sec4 ] we show that this implies a quantum phase transition .
although the quartic equation ( [ quartic ] ) can be solved analytically when both soi couplings are finite , the resulting expressions are not illuminating and too lengthy to be quoted here . only the @xmath110 state in eq
. remains exact for arbitrary parameters .
we here specify the leading perturbative corrections around the special cases above , and then show the generic behavior in two figures . expanding around the rashba limit of sec .
[ sec3r ] , which is justified for @xmath143 , we get the lowest - order perturbative correction to the finite - energy ( i.e. , @xmath144 ) landau levels ( [ rashba ] ) in the form @xmath145 expanding instead around the intrinsic soi limit of sec . [ sec3i ] , we find the following small-@xmath1 corrections to the landau levels in eq . :
@xcite for @xmath114 , the state @xmath141 corresponding to the exact solution ( [ p-1 ] ) is not changed by @xmath1 to any order , while @xmath146 obtains the lowest - order correction @xmath147 the corresponding eigenstate is , however , not a spin-@xmath148 state anymore . for @xmath149 ,
the eigenenergy @xmath139 [ eq . ] acquires the perturbative correction @xmath150 we now consider two different soi parameter sets consistent with the estimates in ref . , and show the complete evolution of the landau levels from the weak- to the strong - field limit . in fig .
[ fig1 ] , numerical results for the few lowest - energy landau levels are depicted for @xmath2 , corresponding to a qsh phase for @xmath130 .
the ( valley - degenerate ) spin - split levels corresponding to the @xmath151 zero mode exhibit a zero - energy crossing at @xmath152 t for the chosen soi parameters .
this crossing signals a quantum phase transition from the qsh phase , which survives for sufficiently small @xmath56 and @xmath2 , to a peculiar qh phase for large @xmath56 . as we discuss in sec .
[ sec4 ] , one then again has helical edge states@xcite but with reversed spin current .
similar crossings can occur for higher landau states as well , as is shown in fig .
[ fig2 ] for a parameter set with @xmath153 where no qsh physics is expected . for even larger
@xmath56 not displayed in fig .
[ fig2 ] , we find an @xmath4 crossing where the rashba - dominated small-@xmath56 phase turns into the helical qh phase . given the landau level eigenstates , it is straightforward to compute the spin - polarization densities @xmath154 ( @xmath155 ) .
we find @xmath156 , while @xmath157 \nonumber,\end{aligned}\ ] ] where @xmath158 . in the absence of the rashba term ( @xmath159 , the in - plane component @xmath160 vanishes identically , since then the eigenstates are simultaneously eigenstates of @xmath25 . for finite @xmath1 , integration over @xmath14
yields a vanishing expectation value for the overall in - plane polarization , but the rashba coupling still induces _ local _ in - plane spin polarization .
the case @xmath91 has been discussed in detail by rashba.@xcite
in this section , we consider the edge states corresponding to the relativistic landau level problem in sec .
[ sec3 ] when a boundary at @xmath161 is present .
we focus on the case of purely intrinsic soi , @xmath5 , but the physics should be qualitatively unchanged for @xmath162 . in the region
@xmath163 we then have a homogeneous magnetic field @xmath164 , i.e , @xmath90 .
( for a pseudo - magnetic field , this holds at the @xmath12 point while at the @xmath48 point , @xmath165 . ) since the problem of edge states in graphene has been studied extensively before , some remarks are in order at this point .
in fact , putting @xmath166 , our results are consistent with those of refs . reporting chiral qh edge states in graphene . on the other hand ,
the @xmath130 model is equivalent to the continuum limit of the kane - mele model@xcite and thus exhibits helical qsh edge states.@xcite ( the helical state has a pair of counterpropagating 1d modes with opposite spin polarization . )
the kane - mele model with @xmath167 but without orbital magnetic field has recently been studied,@xcite and a quantum phase transition from a ( generalized ) qsh phase for @xmath168 to a quantum anomalous hall ( qah ) phase for @xmath169 has been predicted . it is worthwhile to stress that the qsh effect survives even when time - reversal symmetry is broken . in the qah phase , one has chiral edge states moving in the same direction for both spin polarizations.@xcite the valley analogue of this quantum phase transition has also been studied.@xcite furthermore , for the 2d topological insulator realized in hgte quantum well structures , a related transition has been predicted@xcite by including the orbital field but omitting the zeeman term .
however , the zeeman term is crucial in graphene near the dirac point : for @xmath170 and @xmath171 , spin - filtered helical edge states ( similar to the qsh case ) emerge again.@xcite our results below show that this qsh - like phase is separated from the `` true '' qsh phase by a _ quantum phase transition _ at @xmath142 . albeit both phases have spin - filtered edge states , they differ in the _ direction _ of the spin current
. this feature should allow to experimentally distinguish both phases and to identify the quantum phase transition separating them . in practice
, one may reach this transition simply by changing the magnetic field .
normalizability of the wavefunctions for @xmath172 implies@xcite that the only allowed solutions follow from the @xmath85 spinors in eq . , while the @xmath86 solutions [ eq . ] have to be discarded .
since we do not have to impose normalizability at @xmath173 , the order @xmath82 is not constrained to integer values and can now take any real value consistent with suitable boundary conditions at @xmath161 . for given conserved momentum @xmath174 and spin @xmath175 , the solutions for @xmath82 yield the edge state spectrum , @xmath176 .
note that for finite magnetic field and @xmath177 , the distance from the boundary is set by @xmath178 .
putting @xmath5 , possible solutions @xmath179 must be of the form in eq . , with energy @xmath139 given by eq . .
while @xmath134 in sec .
[ sec3i ] , we now consider arbitrary real @xmath82 . to make progress
, we have to specify boundary conditions at @xmath161 .
we investigate two widely used boundary conditions , namely the zig - zag edge and the armchair edge.@xcite for a zig - zag edge with the last row of carbon atoms residing on , say , sublattice @xmath180 , the microscopic wavefunction must vanish on the next row outside the sample , belonging to sublattice @xmath56 . in the continuum limit ,
since the @xmath15-axis here points in the zig - zag direction , the lower component of the spinor @xmath181 [ eq . ] has to vanish at @xmath161.@xcite for both spin directions @xmath182 , this yields the condition @xmath183 which has to be solved for the energy , expressed in terms of @xmath82 as @xmath184 . at the other dirac point ,
the lower component of the spinor @xmath185 should vanish at @xmath161 , where eq .
implies the condition @xmath186 with @xmath187 in eq . .
it is not possible to find simultaneous solutions to both eqs . and .
possible states are thus confined to a single valley : the boundary condition does not mix the valleys but lifts the @xmath188 degeneracy .
remarkably , for @xmath182 and arbitrary @xmath174 , eq . is satisfied by the @xmath48 solution for @xmath114 in sec .
[ sec3i ] , with @xmath189 , i.e. , we find a pair of `` flat '' states .
for all other states , eq . simplifies to condition ( [ c1 ] ) with @xmath190 ( and @xmath191 ) .
we mention in passing that for @xmath170 this condition reduces to eq . ( 9 ) in ref . .
equation ( [ c1 ] ) can be solved in closed form for @xmath192 using asymptotic properties of the parabolic cylinder function .
to exponential accuracy , with @xmath193 we find @xmath194 numerical analysis of the above equations recovers the expected spin - filtered helical edge states@xcite for @xmath169 , but the continuum approach used in this paper fails to give clear evidence for the helical qsh edge states for @xmath168 . as pointed out in ref .
, under the zig - zag boundary condition one needs a more microscopic description in order to capture these states .
the `` flat '' states above are remnants of the sought qsh edge states , but the continuum model is not sufficient to describe their proper dispersion relation .
we therefore turn to the armchair boundary condition . under the armchair boundary condition ,
we instead impose @xmath195 and @xmath196 at the boundary , with @xmath197 in eq . .
this boundary condition mixes the valleys and involves both sublattices . since in our coordinate system
the @xmath15-axis is parallel to the zig - zag direction , we first rotate the system by @xmath198 and then impose the boundary condition at @xmath161 . written in the original coordinates , we find ( for each spin direction @xmath175 )
@xmath199 we note that the relative phase between the @xmath12 and @xmath48 components is not fixed by the dirac equation , which is diagonal in valley space
. however , the only relative phase compatible with the boundary condition imposed simultaneously on both sublattices is @xmath200 .
each of the two conditions in eq . may thus be imposed separately .
we have checked that the numerical solution of eq . for @xmath170 recovers the known results for the qh edge state spectrum.@xcite in addition , for @xmath130 , the armchair edge
is known @xcite to yield qsh edge states .
our numerical results for the dispersion relation @xmath201 for the armchair edge are shown in fig .
[ fig3 ] , where @xmath202 corresponds to the symmetric or antisymmetric linear combination in eq . and
the magnetic field is @xmath203 t. the main panel shows results for @xmath204 mev .
then @xmath205 , and we have the ( generalized ) qsh phase .
indeed , for @xmath4 we find the helical edge state , where the right- ( left-)mover has spin @xmath206 @xmath207 .
the inset of fig .
[ fig3 ] is for @xmath208 mev , where @xmath209 and the spin - filtered helical qh phase@xcite is found .
here we have spin @xmath210 ( @xmath206 ) for the right- ( left-)mover .
hence the _ spin current
_ differs in sign for @xmath205 and @xmath209 , with a quantum phase transition at @xmath211 separating both phases .
this feature should allow for an experimentally observable signature of the transition .
( color online ) dispersion relation @xmath201 of a semi - infinite graphene sheet with an armchair edge at @xmath161 , obtained numerically from eq . .
we use @xmath5 , @xmath203 t , @xmath204 mev , and the @xmath212 ( @xmath213 ) sign is for the symmetric ( antisymmetric ) valley combination in eq . .
inset : same for @xmath208 mev .
, scaledwidth=45.0% ]
in this section , a spatially inhomogeneous situation is considered , where a magnetic waveguide@xcite along the @xmath15-direction can be realized . since the problem remains homogeneous along the @xmath15-direction , @xmath31 is still conserved . for the physics described below
, the zeeman coupling @xmath3 gives only tiny corrections@xcite and will be neglected .
moreover , there are no valley - mixing terms such that we can focus on a single valley .
we distinguish a central strip of width @xmath214 ( the waveguide " ) , @xmath215 , and two outer regions @xmath216 and @xmath217 . in the central strip
, we shall allow for arbitrary soi parameters @xmath0 and @xmath1 .
in addition , strain may cause a constant contribution to the vector potential , @xmath39 , and a scalar potential , @xmath35 . the magnetic field in the central strip
is denoted by @xmath218 . for @xmath219 , we assume that all strain- or soi - related effects can be neglected , @xmath220 . in principle , by lithographic deposition of adatoms , one may realize this configuration experimentally . for @xmath216 , the magnetic field is @xmath221 , while for @xmath217 , we set @xmath18 , where @xmath222 ( @xmath223 ) corresponds to the parallel ( antiparallel ) field orientation on both sides . for @xmath223 , we take @xmath224 , while for @xmath90 , we set @xmath225
. the setup with @xmath223 could be realized by using a `` folded '' geometry,@xcite cf .
recent experimental studies.@xcite note that when the magnetic field changes sign , one encounters `` snake orbits , '' which have been experimentally observed in graphene @xmath226 junctions.@xcite for the @xmath223 configuration , we have uni - directional snake orbits mainly localized along the waveguide , while for @xmath90 , we get two counterpropagating snake states centered near @xmath227 . for @xmath220 , both cases ( @xmath63 )
have been studied in detail in ref . .
technically , one determines the eigenstates and the spectrum , @xmath228 , by matching the wavefunctions in the three different regions , which results in an energy quantization condition .
this method can be straightforwardly extended to the more complex situation studied here by employing the general solution in sec .
[ sec2 ] for the central strip . before turning to results
, we briefly summarize the parameter values chosen in numerical calculations .
we take a magnetic field value @xmath229 t , and the waveguide width is @xmath230 nm .
the strain - induced parameters in the central strip are taken as @xmath231m@xmath232 and @xmath233 mev .
these values have been estimated for a folded setup,@xcite where @xmath35 comes from the deformation potential .
we consider two different parameter choices for the soi couplings : set ( a ) has @xmath234 mev and @xmath235 mev , corresponding to the qsh phase . for set ( b ) , we exchange both values , i.e. , @xmath236 mev and @xmath237 mev .
let us first discuss the @xmath223 configuration , where the magnetic field @xmath238 differs in sign in the regions @xmath216 and @xmath217 .
the dispersion relation of typical low - energy 1d waveguide modes is shown in fig .
for @xmath192 the centers of the quantum states are located deep in the left and right magnetic regions , far from the waveguide .
thus one has doubly - degenerate dispersionless `` bulk '' landau states . with increasing @xmath174
these states are seen to split up .
the dominant splitting , which is already present for @xmath239 , comes from the splitting of symmetric and anti - symmetric linear combinations of the landau states for @xmath216 and @xmath217 with increasing overlap in the waveguide region.@xcite asymptotically , the dispersion relation of all positive - energy snake states is @xmath240.@xcite for intermediate @xmath174 and @xmath241 , however , we get _ spin - split snake states _ out of the previously spin - degenerate states .
the spin splitting is mainly caused by the rashba coupling @xmath1 and disappears for @xmath242 , cf .
the inset of fig
. [ fig4 ] .
the zero - energy bulk landau state ( for @xmath192 ) shows rich and interesting behavior in this setup . while for @xmath243
, we expect one pair of snake states with positive slope and one pair with negative slope , for the studied parameter set and range of @xmath244 , there is just one state with negative slope while three branches first move down and then have a positive slope .
accordingly , at the dirac point ( @xmath4 ) , fig . [ fig4 ]
shows that there are three right - movers with different fermi momenta and different spin texture .
two of those states are indicated by stars ( * ) in the main panel of fig .
[ fig4 ] and their local spin texture is shown in fig .
[ fig5 ] . evidently , they are mainly localized inside the waveguide and have antiparallel spin polarization .
we find spin densities with @xmath245 for both states .
for the rashba - dominated situation in fig .
[ fig5 ] , spin is polarized perpendicular to the current direction and has a rather complex spatial profile . dispersion relation of the lowest few energy branches for a strained magnetic waveguide with @xmath223 and soi in the central strip of width @xmath214 .
energies are given in units of @xmath246 .
the main panel is for parameter set ( b ) .
the stars refer to the states further studied in fig .
inset : same for set ( a ) .
( see main text for details . ) , scaledwidth=45.0% ] ( color online ) spin density profile @xmath247 ( in arbitrary units ) vs @xmath248 for the two @xmath4 right - moving states indicated by stars in the main panel of fig .
the left star corresponds to @xmath249 , the right star to @xmath250 .
inset : particle density , @xmath251 , and current density , @xmath252 ( which is the only non - vanishing component ) , in arbitrary units vs @xmath248 .
we show the result only for @xmath253 , since @xmath250 yields practically the same .
, scaledwidth=45.0% ] next we come to the @xmath90 configuration , where the magnetic field is @xmath254 for @xmath219 and @xmath255 for @xmath256 .
one therefore expects two counterpropagating snake states in the @xmath15-direction localized around @xmath227 . the corresponding spectrum is shown in fig .
we focus on parameter set ( b ) , since for set ( a ) , the spin splitting is minimal and less interesting .
the spectrum consists of two qualitatively different states , namely states of bulk landau character for large @xmath257 , and a set of propagating waveguide modes.@xcite the spectral asymmetry seen in fig .
[ fig6 ] for all propagating modes , @xmath258 , is caused by the strain ( @xmath39)-induced shift of @xmath174 .
such a spectral asymmetry may give rise to interesting chirality and magnetoasymmetry effects.@xcite the spin texture is shown in fig .
[ fig7 ] for a pair of right- and left - moving states with @xmath259 , cf . the stars in fig .
we observe from the main panel in fig .
[ fig7 ] that the spin polarization of both states is approximately antiparallel . because of their spatial separation and the opposite spin direction , elastic disorder backscattering between these counterpropagating snake modes should be very strongly suppressed .
the inset of fig .
[ fig7 ] shows the current density profile across the waveguide . although the profile is quite complex , we observe that the current has opposite sign for both modes .
same as fig .
[ fig4 ] but for the setup with @xmath90 and parameter set ( b ) .
solid and dashed curves are for better visibility only .
the two states indicated by stars are studied in fig .
[ fig7 ] . , scaledwidth=45.0% ] ( color online ) spin density @xmath247 ( in arbitrary units ) vs @xmath248 for the two states indicated by stars in fig .
the left ( right ) star corresponds to a left- ( right-)mover with @xmath260 ( @xmath261 ) .
note that the spin polarizations of both states are approximately antiparallel .
inset : particle current profile @xmath252 ( in arbitrary units ) vs @xmath248 for both states .
black solid curve : @xmath262 .
dashed red curve : @xmath260 .
, scaledwidth=45.0% ]
in this work , we have studied the magnetoelectronic properties of monolayer graphene in the presence of strong intrinsic and rashba - type spin - orbit couplings . according to a recent proposal,@xcite
large intrinsic couplings may be realized by suitable adatom deposition on graphene .
we have presented an exact solution for the landau level states for arbitrary soi parameters .
when the intrinsic soi dominates , by increasing the magnetic field , we predict a quantum phase transition from the quantum spin hall phase to a helical quantum hall phase at the dirac point . in both phases , one has spin - filtered edge states but with opposite spin current direction .
thus the transition could be detected by measuring the spin current either in a transport experiment ( e.g. , along the lines of ref . ) or via a magneto - optical experiment . in inhomogeneous magnetic fields , especially when
also strain - induced pseudo - magnetic fields are present , interesting waveguides can be envisioned .
such setups allow for snake states , where spin - orbit couplings result in a spin splitting . in a double - snake setup
, there is a pair of counterpropagating snake states that carry ( approximately ) opposite spin polarization .
this implies that scattering by elastic impurities is drastically suppressed .
the resulting spin textures can in principle be detected by spin resolved arpes ( see , e.g. , refs . and ) or spin - polarized stm measurements .
we hope that our predictions can soon be tested experimentally .
we acknowledge financial support by the dfg programs spp 1459 and sfb tr 12 .
here we provide some details concerning the derivation of eq .
( [ sol1 ] ) ; the notation below is explained in sec .
first , additional relations like eq . can be stated , @xmath263 we wish to construct the solution @xmath264 satisfying eq .
( [ schr ] ) , @xmath265 we here show only the case @xmath90 near the @xmath12 point ; all other cases follow analogously .
solving the first and last equations for @xmath266 and @xmath267 , respectively , we find @xmath268 for @xmath269 , one has the solution ( [ p-1 ] ) instead , while for @xmath116 there are no solutions .
the second and third equations then yield two coupled second - order ordinary differential equations for @xmath270 and @xmath271 , @xmath272 solving for @xmath271 yields @xmath273 and we thus arrive at the equation @xmath274\phi_2=0.\ ] ] since the operator @xmath275 commutes with the `` number operator '' @xmath276 , the sought solutions for @xmath270 span the kernel of @xmath275 where the @xmath270 are eigenstates of @xmath276 , @xmath277 this leads to an algebraic equation for the eigenvalue @xmath82 , @xmath278 which implies the two solutions in eq . .
with @xmath279 , the eigenvalue equation for @xmath270 is just the differential equation of the parabolic cylinder functions,@xcite @xmath280 which has the four ( linearly dependent ) solutions @xmath281 .
given the solution for @xmath270 , all other components in @xmath73 follow by using the recurrence relations of the @xmath282 functions , see , e.g. , eq . .
after straightforward but lengthy algebra , we obtain the four solutions ( also quoted for @xmath223 ) for a given energy @xmath74 , eq .
admits precisely four linearly independent solutions for @xmath73 .
eq . implies two possible values for @xmath82 , i.e. , we have the freedom to choose just two out of the four quoted eigenstates ( for given @xmath66 ) and then allow both values of @xmath82 in eq . .
our conventions for these two basis states are specified in eqs . and in the main text .
thereby we have obtained all possible solutions to eq . .
we note that we made a wrong statement in ref . in that direction :
for the case @xmath286 on page 3 therein , we stated that for @xmath287 there are two normalizable states with @xmath288 which for @xmath289 coalesce into a single zero - energy landau level .
however , only the state @xmath290 is allowed , and the other one is not normalizable .
z. qiao , s.a .
yang , w. feng , w.k .
tse , j. ding , y. yao , j. wang , and q. niu , phys .
b * 82 * , 161414(r ) ( 2010 ) ; w.k .
tse , z. qiao , y. yao , a.h .
macdonald , and q. niu , phys .
b * 83 * , 155447 ( 2011 ) . | the electronic properties of a graphene monolayer in a magnetic and a strain - induced pseudo - magnetic field are studied in the presence of spin - orbit interactions ( soi ) that are artificially enhanced , e.g. , by suitable adatom deposition . for the homogeneous case , we provide analytical results for the landau level eigenstates for arbitrary intrinsic and rashba soi ,
including also the zeeman field .
the edge states in a semi - infinite geometry are studied in the absence of the rashba term . for a critical value of the magnetic field
, we find a quantum phase transition separating two phases with spin - filtered helical edge states at the dirac point .
these phases have opposite spin current direction .
we also discuss strained magnetic waveguides with inhomogeneous field profiles that allow for chiral snake orbits .
such waveguides are practically immune to disorder - induced backscattering , and the soi provides non - trivial spin texture to these modes . |
the rapid progress of miniaturization of electronic devices has led to chip features smaller than @xmath2 , for which standard semiconductor technology reaches its limit .
one proposed solution is a transistor consisting of a single molecule . in recent years transport through single molecules has been studied quite extensively@xcite for example , in break junctions.@xcite _ inelastic _ transport occurs due to the interaction of electrons with internal vibrational or magnetic degrees of freedom of the molecules .
transport through magnetic molecules@xcite is particularly interesting also from the point of view of _ spintronics_i.e .
, the idea of exploiting the electron spin in electronic devices . while most molecules are normally nonmagnetic , there are exceptions such as endohedral @xmath0i.e . , a nitrogen atom encapsulated in a @xmath1 cage.@xcite it is known that the encapsulated atom retains its @xmath3 electrons,@xcite leading to a localized spin @xmath4
. there are fascinating ideas of employing this spin in a quantum computer.@xcite in this paper we propose to measure the current through a single @xmath0 molecule in a break junction and we calculate the current - voltage ( _ i - v _ ) characteristics and the differential conductance @xmath5 .
since transport through a single @xmath1 molecule has been demonstrated@xcite and the synthesis of endohedral fullerenes is also feasible,@xcite such an experiment is possible with present - day apparatus . besides the typical coulomb - blockade behavior
we predict a characteristic fine structure of the coulomb - blockade peaks in @xmath5 due to the exchange coupling of the @xmath1 spin to the spin @xmath6 of the encapsulated nitrogen atom .
it should be mentioned that the discussion of transport through @xmath7 proceeds quite analogously and yields qualitatively identical results .
in our model the @xmath0 molecule is treated as a quantum dot and the leads , labeled as l ( left ) and r ( right ) , as macroscopic charge reservoirs .
relaxation in the leads is assumed to be sufficiently fast so that the electron distributions in the leads can be described by fermi functions . as @xmath1 generally prefers to be
singly or doubly negatively charged,@xcite we assume that electronic transport through the molecule involves only the threefold - degenerate lumo ( lowest unoccupied molecular orbital).@xcite since we concentrate on the fine structure of the differential conductance close to degeneracy points , we assume that the fivefold - degenerate homo ( highest occupied molecular orbital ) , which lies about @xmath8 below the lumo,@xcite remains fully occupied , whereas the threefold - degenerate lumo+1 , about @xmath9 above the lumo,@xcite remains empty.@xcite charge transfer from the nitrogen atom to the fullerene cage is assumed to be negligible . when the lumo is partially occupied , the net spin of the electrons in the lumo , @xmath10 , couples to the spin @xmath6 of the nitrogen atom , @xmath11 , and the total spin of the molecule is @xmath12 . the exchange interaction can be written in the simple form @xmath13 due to hund s first rule : note that the three lumo s and the three nitrogen @xmath3 orbitals both have odd parity with a single nodal plane each , which can be chosen as the @xmath14 , @xmath15 , or @xmath16 plane .
consequently , there is only a significant exchange interaction between any lumo and the @xmath3 orbital of the same symmetry .
the exchange interaction can thus be written as a sum of terms for the three pairs of orbitals . however , due to the strong hunds - rule coupling , the @xmath3 orbital spins combine to @xmath4 ; i.e. , they are all parallel .
then the spin of each @xmath3 orbital is @xmath17 , as can be proved by projecting the spins onto the @xmath4 subspace .
thus the exchange terms can be combined to the simple scalar product .
the full hamiltonian of the system then reads @xmath18 , where @xmath19 represents the molecular quantum dot , @xmath20 represents the leads , and @xmath21 describes the tunneling between the dot and the leads , which is assumed to be weak compared to typical excitation energies of the molecule .
here , the operator @xmath22 creates an electron with spin @xmath23 in the molecular orbital @xmath24 , while @xmath25 creates an electron in lead @xmath26 with spin @xmath23 , momentum @xmath27 , and energy @xmath28 relative to the fermi energy . @xmath29 and @xmath30 are the number and spin operators of electrons on the dot , respectively .
electron - electron interaction is taken into account by the local coulomb repulsion @xmath31 and the exchange interaction with the nitrogen spin by the exchange coupling @xmath32 .
the values of @xmath33 , @xmath31 , and @xmath32 are not well known at present .
_ ab initio _
calculations@xcite indicate that @xmath34 is the ground state , whereas @xmath35 has a slightly higher energy .
this is in agreement with the experimental observation of a very - long - lived metastable @xmath35 ( ref . ) .
however , there are recent contradicting _ ab initio
_ results predicting @xmath35 to be slightly bound relative to @xmath34 ( ref . ) . for our numerical calculations we use @xmath36 and @xmath37 in accordance with ref . .
@xmath32 appears to be _
we take @xmath38 from _ ab initio _ calculations of udvardi.@xcite this relatively strong exchange coupling is consistent with the absence of electron - paramagnetic - resonance ( epr ) signals for @xmath39 anions with charges @xmath40 through @xmath41 , while the signal is present for the neutral molecule and the hexa - anion.@xcite note that the exchange coupling is significantly smaller than the energy of relevant vibrational modes .
the oscillations of the molecule as a whole have an energy of the order of @xmath42.@xcite the oscillations of the nitrogen atom within the @xmath1 have an energy of @xmath43,@xcite whereas the modes of the @xmath1 cage lie at much higher energies .
we next derive rate equations for this model starting from the equation of motion for the density matrix @xmath44,@xcite @xmath45 $ ] . here
, the index @xmath46 denotes the interaction representation with respect to @xmath47 .
integration and iteration gives@xcite @xmath48 \\ & { } - \int_{0}^{t } dt'\ , \left[h_{\mathrm{t } i}(t ) , \left[h_{\mathrm{t } i}(t'),\rho_{i}(t')\right]\right ] .
\end{split}\ ] ] assuming that the leads are weakly affected by the quantum dot and neglecting correlations between the two , @xmath49 can be replaced by the direct product of the _ reduced density matrix _ of the dot , @xmath50 , and the density matrix @xmath51 of the leads , @xmath52 .
we then obtain @xmath53\right].\ ] ] returning to the schrdinger representation and using the markov approximation@xcite @xmath54 , we find @xmath55 - \text{tr}_{\text{leads } } \int_{0}^{\infty } \!\ !
dt'\ , \big[h_{\text{t } } , \\ & \quad\left [ e^{-i(h_{\text{d } } + h_{\text{leads}})t ' } h_{\text{t } } e^{i(h_{\text{d } } + h_{\text{leads}})t ' } , \rho_{\text{d}}(t ) \otimes \rho_{\text{leads } } \right ] \big ] \end{split}\ ] ] as the equation of motion for @xmath56 .
this expression shows that the tunneling hamiltonian @xmath57 is treated in second - order perturbation theory .
taking the trace over the degrees of freedom of the leads produces fermi functions according to @xmath58 where @xmath59 denotes the chemical potential of lead @xmath26 due to the applied source - drain voltage @xmath60 . expanding the nested commutators in eq .
( [ neumann4 ] ) and inserting eq .
( [ tunneling ] ) gives eight terms : @xmath61 e^{-i \epsilon_{\alpha \mathbf{k } } t ' } c_{n \sigma}^{\dagger } e^{-i h_{\text{d } } t ' } c_{n ' \sigma } e^{i h_{\text{d } } t'}\ !
\rho_{\text{d}}(t ) \\ &
\;{}- \left [ 1-f(\epsilon_{\alpha \mathbf{k}}-\mu_{\alpha } ) \right ] e^{i \epsilon_{\alpha \mathbf{k } } t ' } c_{n \sigma } \rho_{\text{d}}(t ) e^{-i h_{\text{d } } t ' } c_{n ' \sigma}^{\dagger } e^{i h_{\text{d } } t ' } \\ & \;{}- f(\epsilon_{\alpha \mathbf{k}}-\mu_{\alpha } ) e^{-i\epsilon_{\alpha \mathbf{k } } t ' } c_{n \sigma}^{\dagger } \rho_{\text{d}}(t ) e^{-i h_{\text{d } } t ' } c_{n ' \sigma } e^{i h_{\text{d } } t ' } \\ & \;{}- \left [ 1-f(\epsilon_{\alpha \mathbf{k}}-\mu_{\alpha } ) \right ] e^{-i\epsilon_{\alpha \mathbf{k } } t ' } e^{-i h_{\text{d } } t ' } c_{n \sigma } e^{i h_{\text{d } } t'}\ !
\rho_{\text{d}}(t ) c_{n ' \sigma}^{\dagger } \\ & \;{}- f(\epsilon_{\alpha \mathbf{k}}-\mu_{\alpha } ) e^{i \epsilon_{\alpha \mathbf{k } } t ' } e^{-i h_{\text{d } } t ' } c_{n \sigma}^{\dagger } e^{i h_{\text{d } } t ' } \rho_{\text{d}}(t ) c_{n ' \sigma } \\
& \;{}+ f(\epsilon_{\alpha \mathbf{k}}-\mu_{\alpha } ) e^{-i\epsilon_{\alpha \mathbf{k } } t ' } \rho_{\text{d}}(t ) e^{-i h_{\text{d } } t ' } c_{n \sigma } e^{i h_{\text{d } } t ' } c_{n ' \sigma}^{\dagger } \\ & \;{}+ \left [ 1-f(\epsilon_{\alpha \mathbf{k}}-\mu_{\alpha } ) \right ] e^{i \epsilon_{\alpha \mathbf{k } } t ' } \rho_{\text{d}}(t ) e^{-i h_{\text{d } } t ' } c_{n \sigma}^{\dagger } e^{i h_{\text{d } } t'}\ ! c_{n ' \sigma } \big\}. \end{split}\ ] ]
the probability of the dot being in the many - particle state @xmath62 is @xmath63 . introducing the overlap matrix elements @xmath64 and @xmath65 functions we can write eq .
( [ neumann5 ] ) as a set of rate equations @xmath66 with transition rates @xmath67 here , @xmath68 is the energy of the many - particle state @xmath62 of the dot and @xmath69 denotes the density of states per spin species in lead @xmath26 , which we take to be constant and equal for both leads .
the matrix elements @xmath70 ( @xmath71 ) can only be finite if the electron number of state @xmath72 is larger ( smaller ) by 1 than the electron number of state @xmath73 .
we are interested in the _ stationary state _ , which corresponds to @xmath74 for all states @xmath62 . in deriving eq .
( [ rateeq ] ) we have assumed that the density matrix @xmath56 is _ completely diagonal_. this assumption requires some thought since many of the eigenstates of our molecular quantum dot are degenerate so that one might expect finite off - diagonal components even in the stationary state .
however , this is not the case : let @xmath31 be a unitary matrix that leaves the dot hamiltonian @xmath75 invariant . with any stationary density matrix @xmath56 , @xmath76 is another solution .
now suppose that there exists a stationary solution @xmath56 that is _ not _ diagonal within a block of degenerate states .
then one can choose @xmath31 so that @xmath76 is diagonal since the nonzero off - diagonal components have been assumed to connect degenerate states ( we exclude the case of accidental degeneracy ) .
but then @xmath76 has _ unequal _
diagonal components
probabilities @xmath77for symmetry - related states .
this is clearly unphysical .
on the other hand , if @xmath56 is already diagonal with degenerate dot states having equal diagonal components , any allowed transformation @xmath31 leaves @xmath56 invariant . the current operator for lead @xmath78 reads@xcite @xmath79 = -i \sum_{n \mathbf{k } \sigma } ( t_{\alpha } c_{n \sigma}^{\dagger } a_{\alpha \mathbf{k } \sigma } - t_{\alpha}^{\ast } a_{\alpha \mathbf{k } \sigma}^{\dagger } c_{n \sigma } ) , \ ] ] where @xmath80 is the number operator for electrons in lead @xmath26 .
tracing out the leads we arrive at an expression for the expectation value of the current : @xmath81 |c_{lm}^{\sigma}|^2 \big ) p^m .
\\ \end{split}\ ] ] we here consider the symmetric case @xmath82 and @xmath83 . as there are @xmath84 possible ways of filling the threefold - degenerate @xmath1 lumo with @xmath85 electrons according to the pauli principle and as the nitrogen atom possesses a spin @xmath6 , solving the rate equations and calculating the current reduces to an eigenvalue problem of dimension @xmath86 .
the _ i - v _ characteristics plotted in fig .
[ fig_current ] show a conductance gap for @xmath87 due to coulomb blockade .
our numerical results show that the current @xmath46 is symmetric with respect to the applied source - drain voltage @xmath60 in accordance with the high symmetry of the fullerene molecule .
each step in the main _ i - v _ curve corresponds to the opening of additional current channels .
simultaneously , the average occupation @xmath88 of the dot changes . for the parameters chosen above , the @xmath34 state is the ground state.@xcite at the first step , the potential drop becomes large enough to allow transitions between the nearly degenerate charge states @xmath40 and @xmath89 , as the chemical potential @xmath90 reaches the value assumed for the ionization energy @xmath91 . at the second step ,
transitions between the charge states @xmath40 and @xmath92 become possible , etc .
our results for the occupation probabilities reveal that detailed balance is satisfied for the broad plateaus in fig .
[ fig_current]i.e .
, @xmath93 . as a consequence ,
the dot occupation probabilities @xmath77 for all _ occupied _ states are identical in the limit @xmath94 , as the transition rates @xmath95 are symmetric for each pair
@xmath24 , @xmath96 of occupied states .
this also accounts for the fact that the average occupation @xmath97 is exactly unity for @xmath98 , increases to @xmath99 at the first step , when the molecule is in one of 24 singly charged or 60 doubly charged states with equal probability , and decreases to @xmath100 at the second step , when 4 additional neutral states become available , etc .
furthermore , we find that each coulomb - blockade step shows a characteristic fine structure , which we discuss below . [ ] [ ] [ 1][0]@xmath60 ( v ) [ ] [ ] [ 1][0 ] [ ] [ ] [ 1][0]@xmath46 ( @xmath101 ) [ ] [ ] [ 1][0]@xmath88 and average occupation @xmath88 of the @xmath1 lumo as a function of the source - drain voltage @xmath102 for @xmath36 , @xmath37 , @xmath103 , @xmath104 , and @xmath105 .
the inset shows the fine structure of one particular coulomb - blockade step.,title="fig:",width=307 ] the calculation of the differential conductance @xmath5 as a function of source - drain voltage @xmath60 and gate voltage @xmath106 shows the usual _ coulomb diamonds _ ; see fig .
[ fig_diamonds ] .
close to the degeneracy points between different charge states we observe a relatively complex fine structure , corresponding to the steps in the inset of fig .
[ fig_current ] .
we assume very low temperatures , @xmath107 , to exhibit the structure more clearly . at higher temperatures
the peaks in @xmath5 are thermally broadened . in the following ,
we briefly explain the physics behind the fine structure , taking fig .
[ fig_diamonds](a ) as an example . [ ] [ ] [ 1][0]@xmath106 ( mv ) [ t][][1][0]@xmath60 ( mv ) [ b][][1][0 ] [ b][][1][0 ] [ b][][1][0 ] [ b][][1][0 ] [ ] [ ] [ 1][0 ] [ ] [ ] [ 1][0 ] [ ] [ ] [ 1][0 ] [ ] [ ] [ 1][0 ] [ ] [ ] [ 1][0 ] [ ] [ ] [ 1][0 ] [ ] [ ] [ 1][0 ] [ ] [ ] [ 1][0 ] [ ] [ ] [ 1][0 ] [ ] [ ] [ 1][0 ] [ ] [ ] [ 1][0 ] as a function of source - drain voltage @xmath60 and gate voltage @xmath106 for @xmath108 . shown are two particular ranges of gate voltages close to the degeneracy points between charge states @xmath40 and @xmath89 ( a),(c ) and between @xmath89 and @xmath109 ( b),(d ) .
( a ) and ( b ) show results for vanishing magnetic field and ( c ) and ( d ) for @xmath110.,title="fig:",width=327 ] since the @xmath1 spin @xmath111 , the spin of the nitrogen atom , @xmath112 , and the total spin @xmath113 ( where @xmath114 ) are good quantum numbers , the exchange energy is @xmath115,\ ] ] which leads to the level splitting illustrated in fig .
[ fig_levels ] .
the structure in fig . [ fig_diamonds](a ) arises from transitions between charge states @xmath40 and @xmath89 , taking spin excitations into account . in equilibrium ( @xmath98 )
only the ground state of the dot is occupied , which is the @xmath34 state with @xmath116 and @xmath117 for @xmath106 smaller than the degeneracy point @xmath118 and the @xmath35 state with @xmath119 and @xmath120 for @xmath121 ; cf .
[ fig_levels](a ) .
for @xmath122 the _ first _ peak in @xmath5 at nonzero @xmath60 originates from the transition with @xmath123 and @xmath124 , corresponding to a gain of exchange energy of @xmath125 .
second _ peak results from a transition with @xmath126 , @xmath127 , and @xmath128 .
simultaneously , the transitions with @xmath129 , @xmath130 , and @xmath131 and @xmath132 , @xmath133 , and @xmath134 are enabled [ dashed lines in fig .
[ fig_levels](a ) ] .
although energetically possible , these transitions are not excited at lower source - drain voltages , because the lower levels are unoccupied .
the last two peaks belong to transitions with @xmath132 , @xmath135 , and @xmath136 and @xmath137 , @xmath127 , and @xmath138 .
note that the values of @xmath139 account for the level spacing .
the situation is different for @xmath140 significantly larger than @xmath118 , where we observe only two peaks ; cf .
[ fig_diamonds](a ) .
as soon as the transition from the @xmath35 ground state into the lowest @xmath34 state with @xmath141 , @xmath142 , and @xmath143 becomes possible , the transitions corresponding to @xmath125 , @xmath144 , @xmath145 , and @xmath146 [ dashed lines in fig .
[ fig_levels](b ) ] are also enabled .
again the latter four would be energetically possible at lower @xmath60 , but do not appear as peaks of @xmath5 , since the corresponding lower levels are unoccupied . in the vicinity of @xmath118
we find that the slope of several lines abruptly changes sign .
this corresponds to the situation where two levels connected in fig .
[ fig_levels ] by a transition cross as @xmath106 is varied . the fine structure in fig . [ fig_diamonds](b ) can be discussed analogously .
the structure is different for all degeneracy points and can thus serve as a _ fingerprint _ of the particular charge transition .
this should be useful since the zero of the @xmath106 axis is often shifted significantly from one experiment to the next .
[ b][][1][0 ] [ b][][1][0 ] ) and two ( @xmath147 ) electrons , taking into account spin excitations .
( a ) situation with the @xmath148 multiplet lower in energy than the @xmath147 multiplet .
( b ) reverse case.,title="fig:",width=249,height=144 ] + ) and two ( @xmath147 ) electrons , taking into account spin excitations .
( a ) situation with the @xmath148 multiplet lower in energy than the @xmath147 multiplet .
( b ) reverse case.,title="fig:",width=249,height=144 ] selection rules for single - electron tunneling require that @xmath149 and @xmath150 . the different brightness of the peaks in fig .
[ fig_diamonds ] is correlated with the number of transitions that are possible at a given source - drain voltage .
each allowed transition may be thought of as one current channel .
experimentally , the _ magnetic _ origin of the fine structure is most conclusively tested by observing the behavior in a magnetic field . for ionized @xmath1 in lattices and in solution
, the orbital moment is quenched.@xcite we assume that the fields generated by the electrodes in a break junction are also sufficiently strong to quench the orbital moment
. then the molecule couples to a magnetic induction @xmath151 only through the _ spin _ moments , described by the new hamiltonian @xmath152 here , @xmath153 is the bohr magneton and @xmath154 is the @xmath154 factor , which is @xmath155 for both the nitrogen spin @xmath11 and the @xmath1 spin .
we choose a many - particle basis of simultaneous eigenstates of @xmath156 , @xmath157 , @xmath113 , and @xmath158 .
then the only difference is that additional zeeman energies appear in our expression for the transition rates , eq .
( [ rates ] ) . in figs .
[ fig_diamonds](c ) and [ fig_diamonds](d ) we show @xmath5 for the same parameters as in figs . [ fig_diamonds](a ) and [ fig_diamonds](b ) but with @xmath110 .
as expected , the peaks split , but in addition several peaks are absent since they are not allowed by the selection rules .
for example , for @xmath121 the first peak is due to a transition with @xmath159 , @xmath160 , and @xmath161 .
since the initial state has all spins aligned in parallel , one electron tunneling out of the dot can only reduce @xmath158 so that there is only a _
single _ peak in @xmath5 . to summarize
, we have presented a theory for transport through a single @xmath0 molecule weakly coupled to metallic electrodes .
our results for the differential conductance @xmath5 as a function of the source - drain and gate voltages show coulomb - blockade and exhibit a characteristic fine structure of the coulomb - blockade peaks due to the coupling of the @xmath1 spin to the spin of the encapsulated nitrogen atom .
h. b. weber , j. reichert , f. weigend , r. ochs , d. beckmann , m. mayor , r. ahlrichs , and h. v. lhneysen , chem .
phys . * 281 * , 113 ( 2002 ) ; j. reichert , h. b. weber , m. mayor , and h. v. lhneysen , appl.phys .
lett . * 82 * , 4137 ( 2003 ) .
l. udvardi , in _ the exchange coupling between the n atom and the fullerene cage in @xmath0 _ , in electronic properties of novel materials molecular nanostructures , edited by h. kuzmany _
et al_. , aip conf .
* 544 * ( aip , melville , ny 2000 ) , p. 187 .
p. jakes , b. goedde , m. waiblinger , n. weiden , k .-
dinse , and a. weidinger , in _ synthesis and epr studies of @xmath0 and @xmath162 radical anions _ , in electronic properties of novel materials molecular nanostructures , edited by h. kuzmany , j. fink , m. mehring , and s. roth , aip conf .
* 544 * ( aip , melville , ny 2000 ) , p. 174 .
f. uhlk , z. slanina , and e. sawa , in _ thermodynamic properties of @xmath0 _ , in electronic properties of novel materials
molecular nanostructures , edited by h. kuzmany _ et al_. , aip conf .
* 544 * ( aip , melville , ny 2000 ) , p. 183 .
t. kato , t. kodama , m. oyama , s. okazaki , t. shida , t. nakagawa , y. matsui , s. suzuki , h. shiromaru , k. yamauchi , and y. achiba , chem .
* 186 * , 35 ( 1991 ) ; t. kato , t. kodama , and t. shida , _ ibid . _
* 205 * , 405 ( 1993 ) . | we consider transport through a single @xmath0 molecule , weakly coupled to metallic leads . employing a density - matrix formalism
we derive rate equations for the occupation probabilities of many - particle states of the molecule .
we calculate the current - voltage characteristics and the differential conductance for @xmath0 in a break junction .
our results reveal coulomb - blockade behavior as well as a fine structure of the coulomb - blockade peaks due to the exchange coupling of the @xmath1 spin to the spin of the encapsulated nitrogen atom . |
the fact that three active neutrinos possess non - degenerate but tiny masses is a striking signature of new physics beyond the standard model @xcite .
perhaps the most popular approach towards understanding the small neutrino mass scale ( @xmath2 ev ) is the seesaw mechanism @xcite , which contains three right - handed neutrinos and retains @xmath3 gauge symmetry .
although this mechanism can naturally work at a superhigh - energy scale ( @xmath4 gev ) to generate tiny majorana neutrino masses , it loses direct testability on the experimental side and causes a hierarchy problem on the theoretical side @xcite .
a straightforward way out is to lower the seesaw scale down to the tev scale , an energy frontier to be soon explored by the large hadron collider ( lhc ) .
but such a tev - scale seesaw scenario inevitably suffers from a terrible fine - tuning of cancellations between the yukawa coupling texture and the heavy majorana mass matrix @xcite . to resolve this unnaturalness problem built in the canonical (
type - i ) seesaw mechanism at low energies , some new interest has recently been paid to a relatively old idea the inverse seesaw mechanism @xcite .
the inverse seesaw mechanism , which can be regarded as the simplest multiple seesaw picture @xcite , is an extension of the canonical seesaw mechanism by introducing three additional gauge - singlet neutrinos together with one gauge - singlet scalar .
its typical result for the effective mass matrix of three active neutrinos is @xmath5 in the leading - order approximation , where the scales of three mass matrices may naturally satisfy @xmath6 .
the smallness of @xmath7 can be attributed to both the smallness of @xmath8 and the smallness of @xmath9 at the tev scale ( i.e. , @xmath10 tev ) .
it is therefore possible to get a balance between theoretical naturalness and experimental testability of the inverse seesaw scheme .
nevertheless , the inverse seesaw mechanism itself is impossible to interpret the observed pattern of neutrino mixing , which is composed of two large angles ( @xmath11 and @xmath12 ) and one small angle ( @xmath13 ) @xcite , because the flavor structures of @xmath14 , @xmath15 and @xmath8 are entirely unspecified .
the latter can be determined by imposing certain flavor symmetries , but such flavor symmetries usually need to be broken in order to give rise to the correct neutrino mass spectrum and neutrino mixing pattern .
for example , a proper combination of the @xmath16 flavor symmetry and the inverse seesaw mechanism at the tev scale @xcite may successfully predict the tri - bimaximal neutrino mixing pattern @xcite ( with @xmath17 , @xmath18 and @xmath19 ) , which is definitely consistent with current experimental data on solar , atmospheric , reactor and accelerator neutrino oscillations .
the present work aims to combine the inverse seesaw mechanism with a newly - proposed flavor symmetry the friedberg - lee ( fl ) symmetry @xcite , so as to fix the flavor textures of @xmath20 , @xmath15 and @xmath8 and thus predict the mass spectrum and mixing pattern of three active neutrinos at the tev scale .
as the fl symmetry requires a fermion mass term to be invariant under a space - time - independent translation of the relevant fermion fields , it can more reasonably be applied to the gauge - singlet neutrino sector instead of the active neutrino sector .
we show that the fl symmetry forces one pair of the gauge - singlet neutrinos to be massless , leading to a simplified but viable version of the inverse seesaw mechanism which has two pairs of massive gauge - singlet neutrinos and allows one active neutrino to be massless . with the help of @xmath0-@xmath1 permutation symmetry
, we consider a very simple fl symmetry breaking ansatz in the gauge - singlet neutrino sector and obtain the tri - bimaximal neutrino mixing pattern by means of the inverse seesaw relation .
we find that non - unitary corrections to this special mixing pattern are possible to reach the percent level if @xmath21 holds at the tev scale .
we also demonstrate that a soft breaking of @xmath0-@xmath1 symmetry can easily accommodate cp violation in such an inverse seesaw scenario .
let us work in the basis where the flavor eigenstates of three charged leptons are identified with their mass eigenstates throughout this paper .
different from the canonical seesaw mechanism with three right - handed neutrinos @xmath22 ( for @xmath23 ) , the inverse seesaw scheme contains three additional gauge - singlet neutrinos @xmath24 ( for @xmath23 ) together with one gauge - singlet scalar @xmath25 . allowing for lepton number violation to a certain extent
, one can write out the following gauge - invariant neutrino mass terms in the inverse seesaw mechanism : @xmath26 where @xmath27 and @xmath28 stand respectively for the @xmath29 lepton and higgs doublets , @xmath30 and @xmath31 are the @xmath32 yukawa coupling matrices , and @xmath8 is a symmetric majorana mass matrix . after spontaneous gauge symmetry breaking , eq .
( 1 ) becomes @xmath33 where @xmath34 and @xmath35 are the @xmath32 mass matrices .
then we arrive at the overall @xmath36 neutrino mass matrix @xmath37 in the flavor basis defined by @xmath38 and their charge - conjugate states : @xmath39 note that the interesting nearest - neighbor - interaction pattern of @xmath37 is guaranteed by an implementation of the global @xmath40 symmetry with a proper charge assignment @xcite .
note also that the mass scales of three sub - matrices in @xmath37 may naturally have a hierarchy @xmath41 , because the second mass term in @xmath42 is not subject to the @xmath29 gauge symmetry breaking scale and the third mass term in @xmath43 violates the lepton number conservation @xcite . in the leading - order approximation , one obtains the inverse seesaw relation for the effective mass matrix of three active neutrinos : @xmath44 it becomes obvious that @xmath45 holds in the limit @xmath46 .
for instance , @xmath47 ev can easily be achieved from @xmath48 and @xmath49 kev . without loss of generality ,
let us work in a basis where @xmath31 ( or equivalently , @xmath14 ) is diagonal ; i.e. , @xmath50 with @xmath51 and @xmath52 ( for @xmath23 ) being real and positive .
now we impose the fl translation @xcite on both @xmath53 and @xmath54 fields : @xmath55 where @xmath56 ( for @xmath23 ) are in general complex numbers , and @xmath57 is a space - time - independent grassmann parameter ( i.e. , @xmath57 is anti - commuting and thus @xmath58 holds ) .
note that the gauge - singlet neutrinos @xmath22 and @xmath24 do not interact with the gauge bosons of the standard model .
note also that the kinetic terms of @xmath59 and @xmath24 change under the above translation , but the resulting action is invariant just because @xmath57 is independent of space and time @xcite .
hence the whole lagrangian of an inverse seesaw model will be invariant under the fl translation , if and only if we require three neutrino mass terms in @xmath60 to be invariant under the fl translation .
this requirement simply implies @xmath61 for @xmath62 . in other words ,
each mass matrix must have a zero eigenvalue . because @xmath31 is diagonal
, we may simply choose @xmath63 or equivalently @xmath64 such that @xmath65 and @xmath66 must hold to satisfy the second and third conditions given in eq .
this in turn implies that the elements in the third column of @xmath30 and those in the third row and the third column of @xmath8 must be vanishing , so as to satisfy the first , fourth and fifth conditions shown in eq .
the textures of @xmath30 and @xmath8 are therefore given as @xmath67 in this case , the symmetric @xmath68 neutrino mass matrix @xmath69 in eq .
( 3 ) can be simplified to an effective @xmath70 neutrino mass matrix @xmath71 where
@xmath72 " symbolically denotes an arbitrary ( non - zero ) matrix element . because @xmath73 holds , one active neutrino must be massless in this minimal inverse seesaw scenario .
the phenomenology of such a simple scenario has partly been explored in ref .
@xcite . here
we show that the fl symmetry applied to the gauge - singlet neutrino sector naturally explains why a pair of gauge - singlet neutrinos can be decoupled from the usual inverse seesaw mechanism .
it is worth remarking that the above conclusion does not depend on the chosen basis of @xmath31 ( i.e. , @xmath31 is diagonal ) , since a unitary transformation of either @xmath53 or @xmath54 does not change any physical content of the inverse seesaw mechanism . to derive a phenomenologically - favored neutrino mixing pattern from this interesting scheme , however , it is more convenient to introduce a less stringent fl symmetry into the gauge - singlet neutrino sector . here
we consider the case of @xmath74 , which is equivalent to the original fl translation ( imposed on @xmath75 @xcite ) . then
three mass terms in @xmath43 can keep invariant under the translations @xmath76 and @xmath77 if they take the following forms : @xmath78 where the greek index runs over @xmath79 and the latin index runs over @xmath80 .
the explicit expressions of @xmath15 , @xmath14 and @xmath8 turn out to be @xmath81 in which all the matrix elements are in general complex .
it is easy to check that @xmath82 holds , and thus each mass matrix has one zero eigenvalue .
we see that the above fl symmetry must be broken , at least for the second mass term of @xmath83 , such that @xmath84 holds to make the inverse seesaw formula in eq .
( 4 ) applicable .
there are certainly a variety of possibilities of breaking the fl symmetry . here
we follow the spirit of ref .
@xcite and consider a very simple symmetry breaking ansatz for the second and third mass terms in eq .
( 10 ) : @xmath85 where @xmath86 and @xmath87 are real and positive . to minimize the number of free parameters , we impose the @xmath0-@xmath1 permutation symmetry on these two mass terms ( i.e. , @xmath88 and @xmath89 ) and assume all of their parameters to be real .
the resultant mass matrices read @xmath90 \ ; , \nonumber \\ m^{}_\mu & = & \mu^{}_0 \left [ \left ( \matrix { 1 & 0 & 0 \cr 0 & 1 & 0 \cr 0 & 0 & 1 \cr } \right ) + \left ( \matrix { 2b^\prime & -b^\prime & -b^\prime \cr -b^\prime & a^\prime + b^\prime & -a^\prime \cr -b^\prime & -a^\prime & a^\prime + b^\prime \cr } \right ) \right ] \ ; , % ( 13)\end{aligned}\ ] ] in which @xmath91 ( or @xmath92 ) and @xmath93 ( or @xmath94 ) are defined to be two dimensionless parameters .
the diagonalization of @xmath14 or @xmath8 is rather straightforward : @xmath95 and @xmath96 , where @xmath97 is simply the tri - bimaximal mixing pattern @xcite @xmath98 then we obtain three mass eigenvalues @xmath99 , @xmath100 , @xmath101 ; and similarly @xmath102 , @xmath103 , @xmath104 . in terms of mass eigenvalues , @xmath14 and @xmath8
can be reexpressed as @xmath105 where @xmath106 note that @xmath107 , @xmath108 , @xmath109 and @xmath110 hold .
note also that the inverse matrix of @xmath14 takes the same form as @xmath14 itself : @xmath111 now let us make a purely phenomenological assumption : @xmath112 with @xmath113 gev being the electroweak scale and @xmath114 being the identity matrix , just for the sake of simplicity . then the inverse seesaw formula in eq .
( 4 ) allows us to arrive at @xmath115 it is straightforward to show that this effective mass matrix can also be diagonalized by the unitary transformation @xmath116 , where @xmath97 has been given in eq . ( 14 ) and three neutrino masses @xmath117 ( for @xmath23 ) directly reflect the salient feature of the inverse seesaw mechanism . in other words ,
the smallness of @xmath118 is ascribed to both the smallness of @xmath119 and that of @xmath120 .
current neutrino oscillation data only provide us with @xmath121 and @xmath122 @xcite , and thus these two neutrino mass - squared differences can easily be reproduced from our result for @xmath118 by adjusting six free parameters @xmath119 and @xmath52 ( for @xmath23 ) .
note that the neutrino mixing matrix @xmath123 appearing in the charged - current interactions of three active neutrinos is not exactly @xmath97 used to diagonalize @xmath7 in eq .
( 18 ) , just because of slight mixing between light and heavy neutrinos in the inverse seesaw mechanism .
as shown in appendix a , the charged - current interactions of three light majorana neutrinos read @xmath124 in which @xmath79 and @xmath125 are the mass eigenstates of charged leptons and active neutrinos , respectively .
the relationship between @xmath123 and @xmath97 is given by @xmath126 , where @xmath127 signifies the slight deviation of @xmath123 from @xmath97 and its approximate expression can be found in eq .
for the ansatz under consideration , we explicitly obtain @xmath128 \ ; .
% ( 20)\end{aligned}\ ] ] current experimental constraints on the matrix elements of @xmath129 are @xmath130 at the @xmath131 confidence level @xcite . combining eqs . ( 16 ) , ( 20 ) and ( 21 ) , we arrive at @xmath132 together with @xmath133 and @xmath134 due to the @xmath0-@xmath1 symmetry of hermitian @xmath129 . if @xmath135 tev , then @xmath136 is expected from the stringent constraint on @xmath137 .
this result accordingly implies @xmath138 , from which @xmath139 tev can be extracted .
it is in general difficult to get a limit on @xmath140 .
but if @xmath141 tev is taken , for example , then @xmath142 leads us to @xmath143 tev . with the help of eqs .
( 14 ) and ( 20 ) , it is straightforward to obtain an interesting expression of @xmath144 in this inverse seesaw scenario : @xmath145 this simple but instructive result clearly shows the deviation of @xmath123 from @xmath97 .
in particular , @xmath146 holds ; i.e. , the non - unitary effect does not contribute to this smallest neutrino mixing matrix element .
we remark that the above fl symmetry breaking ansatz is viable and suggestive for building a realistic inverse seesaw model at the tev scale . to generate non - vanishing @xmath147 and cp violation
, however , one has to invoke a different fl symmetry breaking pattern with one or more non - trivial @xmath148-violating phases .
we proceed with the fl symmetry breaking ansatz in eq .
( 12 ) but allow soft @xmath0-@xmath1 symmetry breaking for @xmath14 and @xmath8 ( i.e. , @xmath149 and @xmath150 with @xmath151 and @xmath152 being complex ) : @xmath153 \ ; , \nonumber \\ m^{}_\mu & = & \mu^{}_0 \left [ \left ( \matrix { 1 & 0 & 0 \cr 0 & 1 & 0 \cr 0 & 0 & 1 \cr } \right ) + \left ( \matrix { 2{\rm re}b^\prime & -b^\prime & -{b^\prime}^ * \cr -b^\prime & a^\prime + b^\prime & -a^\prime \cr -{b^\prime}^ * & -a^\prime & a^\prime + { b^\prime}^ * \cr } \right ) \right ] \ ; , % ( 24)\end{aligned}\ ] ] where @xmath86 ( or @xmath87 ) and @xmath154 ( or @xmath155 ) are real . as shown in appendix b , the inverse matrix of @xmath14 takes the same texture as @xmath14 itself : @xmath156 \ ; , % ( 25)\end{aligned}\ ] ] where @xmath157 which can easily be read off from eq .
( b3 ) in appendix b. furthermore , we show that @xmath158 is also of the same texture : @xmath159 \ ; , % ( 27)\end{aligned}\ ] ] where @xmath160 \nonumber \\ & & - { b^{\prime\prime}}^ * \left [ \left ( 1 + a^{\prime\prime } + b^{\prime\prime } \right ) b^\prime + b^{\prime\prime } \left ( 1 + 2{\rm re } b^\prime \right ) - a^{\prime\prime } { b^\prime}^ * \right ] \nonumber \\ & & + \left ( 1 + a^{\prime\prime } + { b^{\prime\prime}}^ * \right ) \left [ \left ( 1 + a^{\prime\prime } + b^{\prime\prime } \right ) a^\prime + a^{\prime\prime } \left ( 1 + a^\prime + { b^\prime}^ * \right ) - b^{\prime\prime } { b^\prime}^ * \right ] \ ; , \nonumber \\ \widehat{b } & = & -a^{\prime\prime } \left [ \left ( 1 + 2{\rm re } b^{\prime\prime}\right ) { b^\prime}^ * + { b^{\prime\prime}}^ * \left ( 1 + a^\prime + { b^\prime}^ * \right ) - b^{\prime\prime } a^\prime \right ] \nonumber \\ & & + b^{\prime\prime } \left [ \left(1 + 2{\rm re}b^{\prime\prime}\right ) \left(1 + 2{\rm re}b^\prime \right ) + b^{\prime\prime } b^\prime + { b^{\prime\prime}}^ * { b^\prime}^ * \right ] \nonumber \\ & & + \left ( 1 + a^{\prime\prime } + b^{\prime\prime } \right ) \left [ \left ( 1 + 2{\rm re}b^{\prime\prime } \right ) b^\prime + b^{\prime\prime } \left ( 1 + a^\prime + b^\prime \right ) - { b^{\prime\prime}}^ * a^\prime \right ] \ ; , % ( 28)\end{aligned}\ ] ] which can directly be read off from eq .
note that @xmath161 is real and @xmath162 is complex . making the same phenomenological assumption for @xmath15 as in section iii ( i.e. , @xmath163 )
, we simply obtain the effective mass matrix of three active neutrinos from eqs .
( 4 ) and ( 27 ) in this inverse seesaw scenario : @xmath164 % ( 29)\end{aligned}\ ] ] with @xmath165 .
the small mass eigenvalues of @xmath7 are therefore attributed to small @xmath87 as well as small @xmath166 at the scale of @xmath167 tev .
the diagonalization of @xmath7 in eq .
( 29 ) can be done by using the unitary transformation @xmath168 , where @xmath169 and @xmath170 arises from the soft @xmath0-@xmath1 symmetry breaking term of @xmath7 ( i.e. , @xmath171 ) . comparing this result with the standard parametrization of @xmath172 in terms of three mixing angles @xmath173 and three cp - violating phases @xmath174 @xcite
, we immediately find @xmath175 together with @xmath19 , @xmath176 and @xmath177 . on the other hand ,
three mass eigenvalues of @xmath7 are given by @xmath178 \ ; , \nonumber \\ m^{}_2 & = & m^{}_0 \ ; , \nonumber \\ m^{}_3 & = & m^{}_0 \left[\sqrt{\left ( 1 + \widehat{a } + 2{\rm re}\widehat{b}\right)^2 + 3 \left({\rm im}\widehat{b}\right)^2 } ~ + \widehat{a } - { \rm re}\widehat{b } \right ] \ ; .
% ( 32)\end{aligned}\ ] ] by adjusting four real parameters in eq .
( 32 ) , we can easily fit two observed neutrino mass - squared differences @xmath179 and @xmath180 .
for example , @xmath181 ev , @xmath182 , @xmath183 and @xmath184 lead to a hierarchical neutrino mass spectrum : @xmath185 ev , @xmath186 ev and @xmath187 ev , consistent with current neutrino oscillation data @xcite . in this illustrative case ,
we can also obtain @xmath188 , which in turn predicts @xmath189 and @xmath190 .
it is worth reiterating that the imaginary parts of @xmath191 , @xmath192 and @xmath193 represent the @xmath0-@xmath1 symmetry breaking effects of @xmath14 , @xmath8 and @xmath7 , respectively , in this phenomenological ansatz .
they have nothing to do with the @xmath0-@xmath1 symmetry breaking effect in the charged - lepton sector , because the latter is characterized by the mass ratio @xmath194 in the chosen basis where the charge - lepton mass matrix is diagonal and real .
now we calculate the non - unitary corrections to @xmath172 in order to figure out the neutrino mixing matrix @xmath195 , which is certainly more non - trivial in this cp - violating inverse seesaw scenario . by using the formula of @xmath129 given in eq .
( a7 ) and the expression of @xmath196 shown in eq .
( 25 ) , we obtain @xmath197 \ ; , \nonumber \\ \eta^{}_{e\mu } & \approx & \frac{v^2}{2m^2_0 } \left [ a^{\prime\prime } { b^{\prime\prime}}^ * - \left ( 1 + 2 { \rm re } b^{\prime\prime } \right ) { b^{\prime\prime}}^ * - \left ( 1 + a^{\prime\prime } + { b^{\prime\prime}}^ * \right ) b^{\prime\prime } \right ] \ ; , \nonumber \\ \eta^{}_{\mu\mu } & \approx & \frac{v^2}{2m^2_0 } \left [ { a^{\prime\prime}}^2 + |b^{\prime\prime}|^2 + \left| 1 + a^{\prime\prime } + b^{\prime\prime } \right|^2 \right ] \ ; , \nonumber \\ \eta^{}_{\mu\tau } & \approx & \frac{v^2}{2m^2_0 } \left [ { b^{\prime\prime}}^2 - 2 a^{\prime\prime } \left ( 1 + a^{\prime\prime } + b^{\prime\prime } \right ) \right ] \ ; , % ( 33)\end{aligned}\ ] ] together with @xmath198 and @xmath199 for hermitian @xmath129 . here
we have more free parameters to adjust , such that the values of @xmath200 ( for @xmath201 ) can fit their experimental upper bounds given in eq .
taking account of @xmath202 , we may simply choose @xmath203 in the calculation of @xmath123 .
in addition , the smallness of both @xmath204 in @xmath172 and that of @xmath205 , @xmath206 and @xmath207 allow us to obtain an approximate expression of @xmath123 as follows : @xmath208 where @xmath209 is defined , @xmath210 and @xmath211 are taken , and the terms of @xmath212 are omitted ( for @xmath213 ) .
it becomes quite obvious that the non - unitary corrections to @xmath172 contain a new cp - violating phase @xmath214 , which is possible to give rise to an observable effect of cp violation in @xmath215 and @xmath216 oscillations . to see the above point more clearly ,
let us calculate the following jarlskog invariants @xcite of leptonic cp violation : @xmath217 it is the sum @xmath218 that appears in the probabilities of @xmath219 and @xmath220 oscillations @xcite , and thus we have @xmath221 where @xmath222 with @xmath223 being the neutrino beam energy and @xmath224 being the baseline length . in obtaining eq .
( 36 ) , we have neglected those small non - unitary but cp - conserving effects , including the zero - distance " effect @xcite . considering @xmath225 and @xmath216 oscillations ( see ref .
@xcite for more detailed and model - independent discussions ) .
the origin of small masses of three active neutrinos , together with their unexpectedly large mixing angles , is a big puzzle in particle physics .
although a lot of attempts have been made in the past decade to solve this flavor problem , new ideas are eagerly wanted in the upcoming lhc era to achieve a balance between theoretical naturalness and experimental testability of the mechanisms of neutrino mass generation and flavor mixing . in the present work
we have combined the inverse seesaw mechanism with the fl symmetry to fix the flavor textures of neutrino mass matrices at the tev scale , such that both the neutrino mass spectrum and the neutrino mixing pattern can be calculated . to be explicit ,
we have applied the fl symmetry to the gauge - singlet neutrino sector and shown that it forces one pair of the gauge - singlet neutrinos to be massless , leading to a simplified but viable version of the inverse seesaw mechanism which has two pairs of massive gauge - singlet neutrinos and allows one active neutrino to be massless .
taking account of the exact @xmath0-@xmath1 permutation symmetry , we have proposed a very simple fl symmetry breaking ansatz in the gauge - singlet neutrino sector and then obtained the tri - bimaximal neutrino mixing pattern by means of the inverse seesaw relation .
we find that non - unitary corrections to this interesting neutrino mixing matrix are possible to reach the percent level at the tev scale .
we have also demonstrated that a soft breaking of @xmath0-@xmath1 symmetry can easily accommodate leptonic cp violation in such an inverse seesaw scenario .
we remark that it is technically more natural to make the inverse seesaw mechanism work at the tev scale , although it contains much more degrees of freedom than the canonical ( type - i ) seesaw mechanism .
we also remark that the fl symmetry discussed in this work , like some other flavor symmetries discussed in the literature , may serve as a phenomenological organizing principle for studying the flavor textures of neutrino mass matrices .
how to break this symmetry is certainly an open question and involves a lot of arbitrariness , but our ansatz shows that there exist one or more simple symmetry breaking patterns which allow us to predict ( or at least to understand ) small neutrino masses and large neutrino mixing angles . as for the tev - scale inverse seesaw picture under consideration , it is even possible to have observable effects , induced by the non - unitarity of the effective mixing matrix of three active neutrinos , in a delicate medium- or long - baseline neutrino oscillation experiment .
it is finally worth emphasizing that testing the unitarity of the light majorana neutrino mixing matrix in neutrino oscillations and searching for the signatures of heavy majorana ( or pseudo - dirac ) neutrinos at the lhc can be complementary to each other , both qualitatively and quantitatively , in order to deeply understand the intrinsic properties of majorana particles @xcite . in spite of many challenges on the road ahead
, we optimistically expect that some experimental breakthrough in this aspect will pave the way towards the true theory of neutrino mass generation , flavor mixing and cp violation .
one of us ( z.z.x . ) is deeply indebted to k.k .
phua for his warm hospitality at the ias of ntu .
he is also grateful to s. luo and s. zhou for some useful discussions .
this work is supported in part by the national university of singapore ( academic research grant no .
wbs : r-144 - 000 - 178 - 112 ) and by the national natural science foundation of china ( under grant no .
10425522 and no .
10875131 ) .
the standard charged - current interactions between three active neutrinos and three charged leptons are given by @xmath226 in the basis of their flavor eigenstates . without loss of generality ,
we choose the basis in which the flavor eigenstates of three charged leptons are identified with their mass eigenstates .
we proceed to diagonalize the @xmath68 neutrino mass matrix @xmath37 in eq .
( 3 ) so as to reexpress @xmath227 in terms of the mass eigenstates of both charged leptons and neutrinos .
because @xmath37 is symmetric , it can be diagonalized by the following unitary transformation : @xmath228 in which @xmath229 and @xmath230 denote the diagonal mass matrices of light and heavy neutrinos , respectively , in the inverse seesaw mechanism . after this diagonalization ,
the neutrino flavor eigenstates @xmath231 ( for @xmath232 ) can be expressed in terms of the neutrino mass eigenstates @xmath233 , @xmath234 and @xmath235 ( for @xmath23 ) as @xmath236 substituting this equation into eq .
( a1 ) , we immediately arrive at @xmath237 w^-_\mu + { \rm h.c . } \ ; % ( a4)\end{aligned}\ ] ] in the basis of mass eigenstates . as @xmath123 and @xmath238 belong to the same unitary transformation done in eq .
( a2 ) , they satisfy the normalization condition @xmath239 and the exact seesaw relation @xmath240 .
hence @xmath123 itself must be non - unitary , and its deviation from unitarity is just measured by @xmath238 , which actually determines the collider signatures of heavy majorana neutrinos at the lhc @xcite .
a global analysis of current experimental data shows that the strength of unitarity violation of @xmath123 can at most reach the percent level @xcite .
let us denote the slight deviation of @xmath123 from a unitary matrix @xmath97 as follows : @xmath241 with @xmath242 being a hermitian matrix
. then one may simply follow ref .
@xcite to derive the approximate inverse seesaw formula given in eq .
considering @xmath243 and neglecting small non - unitary effects ( i.e. , @xmath244 ) , we obtain @xmath245 in this excellent approximation , @xmath97 is actually defined to be the unitary transformation used to diagonalize the effective @xmath32 mass matrix of three light majorana neutrinos ( i.e. , @xmath7 ) . because of @xmath246 where @xmath247 is approximately unitary to the same degree of accuracy which assures eq .
( a5 ) to hold , we finally arrive at @xmath248 by taking account of @xmath6 .
this result allows us to estimate the non - unitary effects in @xmath123 , once the textures of @xmath15 and @xmath14 are specified in an inverse seesaw model .
note again that we have taken the natural hierarchy @xmath249 in deriving @xmath7 in eq .
( a5 ) and calculating @xmath129 in eq .
in fact , it is mathematically unnecessary to assume @xmath250 because the validity of eq .
( a5 ) is independent of the relative magnitude of @xmath14 and @xmath8 .
in other words , eq .
( a5 ) is also valid if @xmath251 holds , no matter whether @xmath14 is much larger or smaller than @xmath8 .
but we reiterate that @xmath6 is a physical condition of the inverse seesaw mechanism , in which the smallness of @xmath7 is naturally attributed to both the smallness of @xmath8 and the smallness of @xmath9 .
we refer to the following form of a mass matrix as the friedberg - lee ( fl ) texture : @xmath252 \ ; , % ( b1)\end{aligned}\ ] ] where @xmath253 , @xmath254 , @xmath255 and @xmath256 are in general complex parameters . given @xmath257 ,
it is easy to show that the inverse matrix of @xmath258 takes the same fl texture : @xmath259 \ ; , % ( b2)\end{aligned}\ ] ] where @xmath260 now let us consider another mass matrix of the fl texture : @xmath261 \ ; , % ( b4)\end{aligned}\ ] ] where @xmath262 , @xmath263 , @xmath264 and @xmath265 are in general complex parameters .
a lengthy but straightforward calculation shows that the seesaw - like mass matrix @xmath266 takes the same texture as @xmath267 and @xmath268 do : @xmath269 \ ; , % ( b5)\end{aligned}\ ] ] where @xmath270 - \gamma \left [ \left ( 1 + \alpha + \beta \right ) y^\prime + \beta \left ( 1 + y^\prime + z^\prime \right ) - \alpha z^\prime \right ] \nonumber \\ & & + \left ( 1 + \alpha + \gamma \right ) \left [ \left ( 1 + \alpha + \beta \right ) x^\prime + \alpha \left ( 1 + x^\prime + z^\prime \right ) - \beta z^\prime \right ] \ ; , \nonumber \\ \widehat\beta & = & -\alpha \left [ \left ( 1 + \beta + \gamma \right ) z^\prime + \gamma \left ( 1 + x^\prime + z^\prime \right ) - \beta x^\prime \right ] + \beta \left [ \left(1 + \beta + \gamma\right ) \left(1 + y^\prime + z^\prime\right ) + \beta y^\prime + \gamma z^\prime \right ] \nonumber \\ & & + \left ( 1 + \alpha + \beta \right ) \left [ \left ( 1 + \beta + \gamma \right ) y^\prime + \beta \left ( 1 + x^\prime + y^\prime \right ) - \gamma x^\prime \right ] \ ; , \nonumber \\ \widehat\gamma & = & -\alpha \left [ \left ( 1 + \beta + \gamma \right ) y^\prime + \beta \left ( 1 + x^\prime + y^\prime \right ) - \gamma x^\prime \right ] + \gamma \left [ \left ( 1 + \beta + \gamma \right ) \left ( 1 + y^\prime + z^\prime \right ) + \beta y^\prime + \gamma z^\prime \right ] \nonumber \\ & & + \left ( 1 + \alpha + \gamma \right ) \left [ \left ( 1 + \beta + \gamma \right ) z^\prime + \gamma \left ( 1 + x^\prime + z^\prime \right ) - \beta x^\prime \right ] \ ; .
% ( b6)\end{aligned}\ ] ] this interesting seesaw - invariant property is a salient feature of the fl texture .
h. fritzsch , m. gell - mann , and p. minkowski , phys .
b * 59 * , 256 ( 1975 ) ; t.p . cheng , phys .
d * 14 * , 1367 ( 1976 ) ; p. minkowski , phys .
b * 67 * , 421 ( 1977 ) ; t. yanagida , in _ proceedings of the workshop on unified theory and the baryon number of the universe _ , edited by o. sawada and a. sugamoto ( kek , tsukuba , 1979 ) , p. 95 ; m. gell - mann , p. ramond , and r. slansky , in _ supergravity _ , edited by p. van nieuwenhuizen and d. freedman ( north holland , amsterdam , 1979 ) , p. 315 ; s.l .
glashow , in _ quarks and leptons _
, edited by m. l@xmath271vy _ et al . _
( plenum , new york , 1980 ) , p. 707 ; r.n .
mohapatra and g. senjanovic , phys .
lett . * 44 * , 912 ( 1980 ) .
harrison , d.h .
perkins , and w.g .
scott , phys .
b * 530 * , 167 ( 2002 ) ; z.z .
xing , phys .
b * 533 * , 85 ( 2002 ) ; p.f .
harrison and w.g .
scott , phys .
b * 535 * , 163 ( 2002 ) ; x.g .
he and a. zee , phys .
b * 560 * , 87 ( 2003 ) .
s. luo and z.z .
xing , phys .
b * 646 * , 242 ( 2007 ) ; z.z .
xing , int .
e * 16 * , 1361 ( 2007 ) ; w. chao , s. luo , and z.z .
xing , phys .
b * 659 * , 281 ( 2008 ) ; c.s .
huang , t.j .
li , w. liao , and s.h .
zhu , phys .
d * 78 * , 013005 ( 2008 ) ; t. araki and r. takahashi , arxiv:0811.0905 ; t. araki and c.q .
geng , arxiv:0906.1903 .
e. fernandez - martinez , m.b .
gavela , j. l@xmath273pez - pav@xmath273n , and o. yasuda , phys .
b * 649 * , 427 ( 2007 ) ; z.z .
xing , phys .
b * 660 * , 515 ( 2008 ) ; s. luo , phys .
d * 78 * , 016006 ( 2008 ) ; s. goswami and t. ota , phys .
d * 78 * , 033012 ( 2008 ) ; g. altarelli and d. meloni , nucl .
b * 809 * , 158 ( 2009 ) ; s. antusch , m. blennow , e. fernandez - martinez , and j. l@xmath273pez - pav@xmath273n , phys . rev .
d * 80 * , 033002 ( 2009 ) . | the inverse seesaw mechanism with three pairs of gauge - singlet neutrinos offers a natural interpretation of the tiny masses of three active neutrinos at the tev scale .
we combine this picture with the newly - proposed friedberg - lee ( fl ) symmetry in order to understand the observed pattern of neutrino mixing .
we show that the fl symmetry requires only two pairs of the gauge - singlet neutrinos to be massive , implying that one active neutrino must be massless .
we propose a phenomenological ansatz with broken fl symmetry and exact @xmath0-@xmath1 symmetry in the gauge - singlet neutrino sector and obtain the tri - bimaximal neutrino mixing pattern by means of the inverse seesaw relation .
we demonstrate that non - unitary corrections to this result are possible to reach the percent level and a soft breaking of @xmath0-@xmath1 symmetry can give rise to cp violation in such a tev - scale seesaw scenario . |
quarkonium production at large transverse quarkonium momentum @xmath2 proceeds at leading power ( lp ) in @xmath2 through processes in which a high - energy collision produces a single parton , which subsequently fragments into a quarkonium @xcite .
fragmentation functions for a parton to fragment into a quarkonium play a central role in calculations of such processes . in this paper , we compute the fragmentation functions for a quark @xmath3 to fragment into a quarkonium through heavy quark - antiquark ( @xmath1 ) pair channels in which the @xmath1 pair is in a spin - triplet @xmath0-wave state and a color - singlet or a color - octet state .
we calculate fragmentation functions for the case in which the flavors of @xmath3 and @xmath4 are the same , as well as for the case in which the flavors are different .
we carry out these calculations at the leading nontrivial order in the strong coupling @xmath5 and at order @xmath6 , where @xmath7 is the relative velocity of the @xmath4 and the @xmath8 in the quarkonium rest frame .
previous calculations have given the fragmentation functions for a quark to fragment into an @xmath0-wave , spin - triplet @xmath1 pair for the case in which a sum over the @xmath1 spin polarizations has been taken .
the case in which the initial quark and final quark have different flavors and are in a color - octet state is discussed in refs .
the cases in which the initial quark and final quark have the same flavor and are in a color - octet or a color - singlet state is discussed in ref .
our calculations confirm all of these results .
we also verify a previous calculation of the fragmentation function for a quark to fragment into an @xmath0-wave , spin - triplet , color - singlet @xmath1 pair in which a sum over the @xmath1 spin polarizations is taken @xcite .
we have extended all of these spin - summed calculations to the cases in which the @xmath1 pair is in a longitudinally polarized state .
after our calculation was completed , we learned that these longitudinal - polarization fragmentation functions had been calculated in ref .
our calculation agrees with the results in ref .
@xcite for the color - octet and color - singlet same - flavor cases and corrects an error in ref .
@xcite for the color - octet different - flavor case .
the remainder of this paper is organized as follows . in sec .
[ sec : notation ] , we introduce our notation and the kinematics that we use in the calculation and present projectors for the @xmath9 spin and color . in sec .
[ sec : collins - soper ] , we present the collins - soper fragmentation function for an initial quark @xcite and give the feynman rules for its computation .
sections [ sec : unequal - flavors ] and [ sec : equal - flavors ] contain , respectively , the calculations of the color - octet fragmentation functions for the case in which the initial and final quarks have different flavors and the case in which the initial and final quarks have the same flavor . in sec .
[ sec : color - singlet ] , we present the fragmentation functions for the color - singlet case . section [ sec : summary ] contains a summary and discussion of our results .
in this paper , we use the following light - cone coordinates for a four - vector @xmath10 in the @xmath11 space - time dimensions : @xmath12 where we call @xmath13 the longitudinal component of the @xmath14-dimensional spatial vector @xmath15 , and @xmath16 is the @xmath17-dimensional component of @xmath15 that is transverse to @xmath13 . in this coordinate system ,
the scalar product of two four - vectors @xmath10 and @xmath18 is given by @xmath19 .
at the leading nontrivial order in @xmath5 , a quark fragments into the @xmath1 pair that forms the quarkonium plus an additional final - state quark .
we denote the momentum and mass of the fragmenting quark by @xmath20 and @xmath21 , respectively .
we denote the momentum of the final - state quark by @xmath22 , and we denote the momentum of the @xmath1 pair by @xmath23 .
we work at order @xmath6 , and so we take the @xmath4 and the @xmath8 to have identical momenta @xmath24 , with @xmath25 , where @xmath26 is the mass of the @xmath4 or @xmath8 . the mass of the @xmath1 state is then given by @xmath27 .
we work in the frame in which the transverse momentum of the @xmath28 pair vanishes . in this frame , the momenta are @xmath29 where @xmath30 and @xmath31 are the longitudinal momentum fractions of the @xmath32 pair and the final - state quark , respectively : @xmath33 we note that @xmath34 we wish to project the @xmath1 pair onto spin - triplet states .
the required spin - triplet projectors in order @xmath6 are @xcite @xmath35 where @xmath36 is the polarization vector for the spin state @xmath37 .
these projectors correspond to nonrelativistic normalization of the heavy - quark spinors .
the absolute squares of @xmath36 for various polarization states can be written in covariant forms .
the result for the sum over all @xmath37 is [ polarization - sums ] @xmath38 the result for the sum over transverse polarizations is @xcite @xmath39 where @xmath40 then , for the longitudinal polarization , we have @xmath41 the color - singlet and color - octet projection operators for the @xmath1 pair are @xmath42 where @xmath43 is a unit @xmath44-color matrix , @xmath45 is a generator of the fundamental representation of @xmath44 , @xmath46 , and @xmath47 .
collins and soper have given the following gauge - invariant definition of the quark fragmentation function in @xmath48 dimensions @xcite : @xmath49 , \label{collins - soper - frag } % ---------------\ ] ] where @xmath50 is the field of the initial quark and @xmath51 is the gauge link ( eikonal line ) @xmath52 .
% ---------------\ ] ] here , @xmath53 indicates path ordering , and @xmath54 is the qcd coupling constant .
the spinor field @xmath55 is an @xmath44-color column vector in the fundamental representation , and the gluon field @xmath56 is an @xmath44-color matrix in the fundamental representation .
the trace is over the color and dirac indices .
the factors @xmath57 and @xmath58 in the denominator of eq .
( [ collins - soper - frag ] ) arise from the average over the color and dirac indices of the initial - state quark , respectively .
@xmath59 is a projector onto states that include a hadron @xmath60 with momentum @xmath23 and polarization @xmath37 : @xmath61 where the summation is over all possible degrees of freedom .
the feynman rules for the fragmentation function are the standard ones for qcd , with the following exceptions .
first , there is an overall factor @xmath62 which arises from the definition of the fragmentation function .
second , there are additional feynman rules for the eikonal lines .
we state the rules for the part of the feynman diagram that lies to the left of the final - state cut .
the rules for the part of the diagram that lies to the right of the final - state cut can be obtained by complex conjugation .
each eikonal - line propagator that carries momentum @xmath63 , flowing from the cut side to the operator side , contributes a factor @xmath64 , where @xmath65 and @xmath66 are color indices .
each eikonal - line gluon vertex contributes a factor @xmath67 , where @xmath68 is the four - vector index of the gluon .
the final - state cut in an eikonal line carrying momentum @xmath63 contributes a factor @xmath69 . in general , the final - state phase space for a fragmentation function for @xmath70 unobserved particles in the final state is given by @xmath71 where @xmath0 is the statistical factor for identical final - state particles , @xmath72 and @xmath23 are the mass and momentum of the observed particle , respectively , and @xmath73 is the momentum of the @xmath65th unobserved particle .
we have included a factor @xmath74 in the phase space in order to compensate for the fact that we use nonrelativistic normalization for the heavy - quark spinors .
we associate the standard modified - minimal - subtraction ( @xmath75 ) scale factor @xmath76^\epsilon$ ] with each dimensionally regulated integration in @xmath11 space - time dimensions . here
, @xmath68 is the dimensional - regularization scale and @xmath77 is the euler - mascheroni constant . for our specific kinematics , with one unobserved particle in the final state
, the phase space reduces to @xmath78
in this section , we compute the fragmentation function for quark fragmentation into a color - octet @xmath1 pair for the case in which the initial quark @xmath3 and the quark @xmath4 that is a constituent of the quarkonium have different flavors .
the feynman diagrams for this calculation are shown in fig .
[ fig : unequal ] . in this calculation , and throughout the remainder of this paper , we work in the feynman gauge . in each of the contributions from the diagrams in fig .
[ fig : unequal ] , there is a common factor that arises from the annihilation of a virtual gluon into a color - octet spin - triplet @xmath0-wave pair @xmath79}_1 $ ] ) .
the contribution to this factor from the left side of the final - state cut can be written as @xmath80 .
% ---------------\ ] ] a straightforward calculation gives @xmath81 multiplying by the complex conjugate and summing over the final - state color index , we obtain @xmath82 then , the diagrams of fig .
[ fig : unequal ] give the following contributions to the fragmentation function : [ unequal - frag ] @xmath83 x_{\mu \nu}^{ab } d \phi , \hspace{5ex } \\
d_2(z ) & = & c_{\rm qfrag } { \rm tr } \bigg [ ( k \sl_1 + m_q ) \frac{i}{(k - k_1 ) \cdot n+i \varepsilon } ( i g_s n^\mu t^a ) n \sl \nonumber \\ & & \hspace{20ex } \times \frac{-i}{k \sl_1+p\sl - m_q - i \varepsilon } ( + i g_s \gamma^\nu t^b ) \bigg ] x_{\mu \nu}^{ab } d \phi , \\
d_3(z ) & = & c_{\rm qfrag } { \rm tr } \bigg [ ( k \sl_1+m_q ) ( -i g_s \gamma^\mu t^a ) \frac{i}{k \sl_1+p\sl - m_q+i \varepsilon } n \sl \nonumber \\ & & \hspace{20ex } \times ( -i g_s n^\nu t^b ) \frac{-i}{(k - k_1 ) \cdot n - i \varepsilon } \bigg ] x_{\mu \nu}^{ab } d \phi , \\
d_4(z ) & = & c_{\rm qfrag } { \rm tr } \bigg [ ( k \sl_1 + m_q ) \frac{i}{(k - k_1 ) \cdot n+i \varepsilon } (
+ i g_s n^\mu t^a ) n \sl \nonumber \\ & & \hspace{20ex } \times ( -i g_s n^\nu t^b ) \frac{-i}{(k - k_1 ) \cdot n - i \varepsilon } \bigg ] x_{\mu \nu}^{ab } d \phi .
% ---------------\end{aligned}\ ] ] in each contribution in eq .
( [ unequal - frag ] ) , the overall color factor , including the color factor in @xmath84 , is @xmath85 where @xmath86 .
the dependence on @xmath87 in these expressions comes from the quark - propagator denominators , which contribute factors @xmath88 .
thus , the integrals of these expressions over @xmath87 can be written in terms of the scalar integrals @xmath89 , where @xmath90 the only integral that diverges as @xmath91 is @xmath92 : @xmath93 for @xmath94 , @xmath95 , we can write @xmath96 summing over the four contributions in eq .
( [ unequal - frag ] ) , using the expressions for the absolute squares of the polarizations in eq .
( [ polarization - sums ] ) , multiplying by the factor in eq .
( [ norm - factor ] ) , and carrying out the phase - space integration in eq .
( [ 2p - ps ] ) , we obtain the following results for the fragmentation functions : @xmath97})(\lambda)}^{(1 - 4)}(z ) = \frac{g_s^4 c_f}{\pi m^3 z^{1 + 2 \epsilon } } \ { [ 1 + ( 1-z)^2-\epsilon z^2 ] j_1 ( m_q^2 + \tfrac{1-z}{z^2 } m^2 ) \nonumber \\ & & \hspace{32ex } - [ ( 1-\epsilon ) m^2 + 2 m_q^2 ] ( 1-z ) j_2 ( m_q^2 + \tfrac{1-z}{z^2 } m^2)\},\phantom{xxx } \\ & & d_{q\to q \bar q({}^3s_1^{[8]})(\lambda = 0)}^{(1 - 4)}(z ) = \frac{2 g_s^4 c_f}{\pi m } \frac{(1-z)^2}{z^{3 + 2 \epsilon } } j_2(m_q^2 + \tfrac{1-z}{z^2 } m^2 ) .
% ---------------\end{aligned}\ ] ] here , we retain the full @xmath98 dependence , as it may be useful for calculations of fragmentation functions at higher orders in @xmath99 .
the expression for @xmath100})(\lambda)}$ ] contains a pole in @xmath98 .
we renormalize this expression using the @xmath75 procedure @xcite : @xmath101 where the @xmath102 and the @xmath103 are the bare quark and gluon fragmentation functions and @xmath104 is the dokshitzer - gribov - lipatov - altarelli - parisi splitting function , which is given at lowest order in @xmath5 by @xmath105 the bare gluon fragmentation functions for the unpolarized and longitudinally polarized states are given at leading order in @xmath5 by @xcite [ bare - gluon - frag ] @xmath106})(\lambda ) } ( z ) = \frac{\pi \alpha_s}{m_q^3 } \delta(1-z ) , \\ & & d_{g\to q \bar q({}^3s_1^{[8]})(\lambda = 0 ) } ( z ) = 0 .
% ---------------\end{aligned}\ ] ] at this order in @xmath5 , the bare gluon fragmentation functions do not depend on @xmath98 .
dependence in eq .
( [ bare - gluon - frag ] ) is different from that in the corresponding expression in ref .
@xcite . in ref .
@xcite , a factor @xmath107^\epsilon$ ] was associated with each factor @xmath108 . in the present paper ,
we associate a factor @xmath107^\epsilon$ ] with each dimensionally regulated integral .
this difference between these conventions does not affect the finite result .
] carrying out the renormalization and dropping contributions of order @xmath98 and higher , we obtain [ qqbar - unequal ] @xmath97})(\lambda)}^{\overline{\rm ms } } ( z , \mu ) = \frac{\alpha_s^2 c_f}{2 m_q^3 } \bigg\ { \frac{z^2 - 2z+2}{z } \bigg [ \log \frac{\mu^2}{4 m_q^2 } - \log ( 1-z+r z^2 ) \bigg ] \nonumber \\ & & \hspace{35ex } - z - \frac { z ( 1-z ) ( 1 + 2 r ) } { 1-z + r z^2 } \bigg\ } \frac { \langle o^{q \bar q({}^3s_1^{[8]})}({}^3s_1^{[8 ] } ) \rangle } { 3 ( n_c^2 - 1 ) } , \\ & & d_{q\to q \bar q({}^3s_1^{[8]})(\lambda = 0)}(z ) = \frac{\alpha_s^2 c_f}{2 m_q^3 } \frac{2(1-z)}{z } \frac{1-z}{1-z + r z^2 } \frac{\langle o^{q \bar q({}^3s_1^{[8]})}({}^3s_1^{[8 ] } ) \rangle } { 3 ( n_c^2 - 1 ) } , % ---------------\end{aligned}\ ] ] where @xmath109 . in eq .
( [ qqbar - unequal ] ) , we have written the perturbative fragmentation function in the factorized form of a short - distance coefficient times a nonrelativistic - qcd ( nrqcd ) long - distance matrix element ( ldme ) @xcite by making use of the fact pair , even in the case of fragmentation into a single ( longitudinally polarized ) spin state . in that case
, we use the fact that the ldmes for different spin states are identical , up to corrections of relative order @xmath110 . note that our ldmes for the longitudinally polarized case are larger than those in ref .
@xcite by a factor of three . ] that , at order @xmath111 , @xmath112})}({}^3s_1^{[8 ] } ) \rangle = ( n_c^2 - 1)(d-1).\ ] ] the nrqcd ldme @xmath113})}({}^3s_1^{[8 ] } ) \rangle$ ] is defined by @xmath114})}({}^3s_1^{[8 ] } ) \rangle=\langle 0|
\chi^\dagger\sigma^it^a\psi \sum_{\lambda}{\cal p}_{q \bar q({}^3s_1^{[8]})(p,\lambda ) } \psi^\dagger\sigma^it^a\chi|0\rangle.\ ] ] here , @xmath115 is the two - component ( pauli ) spinor field operator that annihilates a heavy quark and @xmath116 is the two - component ( pauli ) spinor field operator that annihilates a heavy antiquark . the projection operator @xmath117})(p,\lambda)}$ ] is the free @xmath32 analogue of @xmath118 , except that , because we are considering an nrqcd ldme , the intermediate state can contain only light degrees of freedom in addition to the explicit @xmath1 pair . we can now obtain the quarkonium fragmentation functions for the unpolarized and the longitudinally polarized states by replacing the free @xmath1 ldmes in eq .
( [ qqbar - unequal ] ) with the quarkonium ldmes : [ ms - bar - frag - unequal ] @xmath119 \nonumber \\ & & \hspace{35ex } - z - \frac { z ( 1-z ) ( 1 + 2 r ) } { 1-z + r z^2 } \bigg\ } \frac{\langle o^h({}^3s_1^{[8 ] } ) \rangle } { 3 ( n_c^2 - 1 ) } , \\ \label{ms - bar - frag - unequal - long } & & d_{q\to h(\lambda=0)}(z ) = \frac{\alpha_s^2 c_f}{2 m_q^3 } \frac{2(1-z)}{z } \frac{1-z}{1-z + r z^2 } \frac{\langle o^h({}^3s_1^{[8 ] } ) \rangle } { 3 ( n_c^2 - 1 ) } , % ---------------\end{aligned}\ ] ] where the nrqcd ldme @xmath120 } ) \rangle$ ] is defined by @xmath121 } ) \rangle=\langle 0| \chi^\dagger\sigma^it^a\psi\sum_{\lambda}{\cal p}_{h(p,\lambda ) } \psi^\dagger\sigma^it^a\chi|0\rangle.\ ] ] identical expressions hold for the case of an initial antiquark . the result for the polarization - summed fragmentation function in eq .
( [ ms - bar - frag - unequal - sum ] ) confirms the result in eq .
( 4.2 ) of ref .
@xcite and the result in eq .
( c37 ) of ref .
the result for the longitudinal - polarization fragmentation function in eq .
( [ ms - bar - frag - unequal - long ] ) disagrees with the result in eq .
( c.10 ) of ref .
@xcite . , rather than @xmath122 , where @xmath123 .
we also find that the result in eq .
( c.10 ) of ref .
@xcite should be multiplied by an overall factor of three . here , we have taken into account the fact that the ldme in eq .
( [ ms - bar - frag - unequal - long ] ) is a factor of three larger than the corresponding ldme in ref . @xcite . ]
the author of ref .
@xcite has confirmed that the result in eq .
( [ ms - bar - frag - unequal - long ] ) is correct . in the case of light initial quarks , it is useful to take the limit @xmath124 , which gives [ unequal - zero - mass ] @xmath125 \frac{\langle o^h({}^3s_1^{[8 ] } ) \rangle}{3 ( n_c^2 - 1 ) } , \label{unequal - zero - mass - sum - pol}\\ d_{q\to h(\lambda=0)}(z ) & = & \frac{\alpha_s^2 c_f}{2 m_q^3 } \frac { 2 ( 1-z ) } { z } \frac{\langle o^h({}^3s_1^{[8 ] } ) \rangle}{3 ( n_c^2 - 1)}.\label{unequal - zero - mass - long - pol } % ---------------\end{aligned}\ ] ] we have compared our result in eq .
( [ unequal - zero - mass - sum - pol ] ) with the result in eq .
( 20 ) of ref .
@xcite for quark fragmentation into lepton pairs , taking into account differences in the color and phase - space factors , and have found that they are consistent with each other .
now let us consider the fragmentation function for the case in which the initial quark @xmath3 and the quark @xmath4 that is a constituent of the quarkonium have the same flavor .
in this case , there are contributions from the diagrams that are shown in fig . [
fig : unequal ] .
these are given by the expressions in eq .
( [ unequal - frag ] ) , but with @xmath21 set equal to @xmath26 .
in addition , there are contributions from the diagrams that are shown in fig .
[ fig : exchange ] and in fig .
[ fig : interference ] . the diagrams in fig .
[ fig : exchange ] differ from those in fig .
[ fig : unequal ] in that the identical quarks have been interchanged in the amplitudes on both the left and right sides of the final - state cut .
the diagrams in fig .
[ fig : interference ] differ from those in fig .
[ fig : unequal ] in that the identical quarks have been interchanged in an amplitude on only one side of the cut .
the contributions of the diagrams in fig .
[ fig : exchange ] are [ exchange - frag ] @xmath126 \nonumber \\ & & \hspace { 15ex } \times \frac{-i}{(k_1+p/2)^2+i \varepsilon } \frac{+i}{(k_1+p/2)^2-i \varepsilon } d \phi , \\ d_6 ( z,\lambda ) & = & c_{\rm qfrag } { \rm tr } \bigg [ \pi_3 ( p , p,\lambda ) \lambda_8^c \frac{i}{(k - p/2)\cdot n+i\varepsilon } ( i g_s n^\mu t^a ) n \sl \frac{-i}{k \sl_1 + p \sl - m_q - i\varepsilon } \nonumber \\ & & \hspace { 15ex } \times ( + i g_s \gamma^\nu t^b ) \bar{\pi}_3 ( p , p,\lambda ) \lambda_8^c ( + i g_s \gamma_\nu t^b ) ( k \sl_1 + m_q ) ( -i g_s \gamma_\mu t^a ) \bigg ] \nonumber \\ & & \hspace { 15ex } \times \frac{-i}{(k_1+p/2)^2+i \varepsilon } \frac{+i}{(k_1+p/2)^2-i \varepsilon } d \phi , \\ d_7 ( z,\lambda ) & = & c_{\rm qfrag } { \rm tr } \bigg [ \pi_3 ( p , p,\lambda ) \lambda_8^c ( -i g_s \gamma^\mu t^a ) \frac{i}{k \sl_1 + p
\sl - m_q+i\varepsilon } n \sl ( -i g_s n^\nu t^b ) \nonumber \\ & & \hspace { 15ex } \times \frac{-i}{(k - p/2 ) \cdot n - i\varepsilon } \bar{\pi}_3 ( p , p,\lambda ) \lambda_8^c ( + i g_s \gamma_\nu t^b ) ( k \sl_1 + m_q ) ( -i g_s \gamma_\mu t^a ) \bigg ] \nonumber \\ & & \hspace { 15ex } \times \frac{-i}{(k_1+p/2)^2+i \varepsilon } \frac{+i}{(k_1+p/2)^2-i \varepsilon } d \phi , \\
d_8 ( z,\lambda ) & = & c_{\rm qfrag } { \rm tr } \bigg [ \pi_3 ( p , p,\lambda ) \lambda_8^c \frac{i}{(k - p/2)\cdot n+i\varepsilon } ( i g_s n^\mu t^a ) n \sl ( -i g_s n^\nu t^b ) \nonumber \\ & & \hspace { 15ex } \times \frac{-i}{(k - p/2 ) \cdot n - i\varepsilon } \bar{\pi}_3 ( p , p,\lambda ) \lambda_8^c ( + i g_s \gamma_\nu t^b ) ( k \sl_1 + m_q ) ( -i g_s \gamma_\mu t^a ) \bigg ] \nonumber \\ & & \hspace { 15ex } \times \frac{-i}{(k_1+p/2)^2+i \varepsilon } \frac{+i}{(k_1+p/2)^2-i \varepsilon } d \phi .
% ---------------\end{aligned}\ ] ] in each contribution in eq .
( [ exchange - frag ] ) , the overall color factor is @xmath127 the contributions from the diagrams in fig .
[ fig : interference ] are [ interference - frag ] @xmath128 \times j_\nu^{bc}{}^ * \nonumber \\ & & \hspace { 15ex } \times \frac{-i}{(k_1+p/2)^2+i \varepsilon } d \phi+{\rm c.c . } , \\
d_{10}(z , \lambda ) & = & - c_{\rm qfrag } { \rm tr } \bigg [ \pi_3 ( p , p,\lambda ) \lambda_8^c \frac{i}{(k - p/2)\cdot n+i\varepsilon } ( i g_s n^\mu t^a ) n \sl \nonumber \\ & & \hspace { 15ex } \times \frac{-i}{k \sl_1 + p \sl - m_q - i\varepsilon } ( + i g_s \gamma^\nu t^b ) ( k \sl_1 + m_q ) ( -i g_s \gamma_\mu t^a ) \bigg ] \times j_\nu^{bc}{}^ * \nonumber \\ & & \hspace { 15ex } \times \frac{-i}{(k_1+p/2)^2+i \varepsilon } d \phi+{\rm c.c . } , \\ d_{11}(z ,
\lambda ) & = & -c_{\rm qfrag } { \rm tr } \bigg [ \pi_3 ( p , p,\lambda ) \lambda_8^c ( -i g_s \gamma^\mu t^a ) \frac{i}{k \sl_1 + p
\sl - m_q+i\varepsilon } n \sl \nonumber \\ & & \hspace { 15ex } \times ( -ig_sn^\nu t^b ) \frac{-i}{(k - k_1 )
\cdot n -i \varepsilon } ( k \sl_1 + m_q ) ( -i g_s \gamma_\mu t^a ) \bigg ] \times j_\nu^{bc}{}^ * \nonumber \\ & & \hspace { 15ex } \times \frac{-i}{(k_1+p/2)^2+i \varepsilon } d \phi+{\rm c.c .
d_{12}(z , \lambda ) & = & - c_{\rm qfrag } { \rm tr } \bigg [ \pi_3 ( p , p,\lambda ) \lambda_8^c \frac{i}{(k - p/2)\cdot n+i\varepsilon } ( i g_s n^\mu t^a ) n
\sl \nonumber \\ & & \hspace { 15ex } \times ( -ig_sn^\nu t^b ) \frac{-i}{(k - k_1 ) \cdot n -i \varepsilon } ( k \sl_1 + m_q ) ( -i g_s \gamma_\mu t^a ) \bigg ] \times j_\nu^{bc}{}^ * \nonumber \\ & & \hspace { 15ex } \times \frac{-i}{(k_1+p/2)^2+i \varepsilon } d \phi+{\rm c.c .
} , % ---------------\end{aligned}\ ] ] where c.c
. stands for complex conjugate . in each contribution in eq .
( [ interference - frag ] ) , the overall color factor , including the color factor from @xmath129 , is @xmath130 we note that the contributions in eq .
( [ interference - frag ] ) contain a phase @xmath131 relative to the contributions in eqs .
( [ unequal - frag ] ) and ( [ exchange - frag ] ) that arises because of the interchange of the identical quarks in an amplitude on only one side of the final - state cut .
we also note that the quantities @xmath132 [ eq . ( [ unequal - frag ] ) ] , @xmath133 [ eq . ( [ exchange - frag ] ) ] , and @xmath134 [ eq . ( [ interference - frag ] ) ]
are separately gauge invariant . the dependence on @xmath87 in the expressions in eqs .
( [ exchange - frag ] ) and ( [ interference - frag ] ) comes from the quark- and gluon - propagator denominators , which contribute factors @xmath135^{-1}$ ] .
thus , the integrals of these expressions over @xmath87 can be expressed in terms of the scalar integrals @xmath136 $ ] [ eq .
( [ jn ] ) ] .
summing over the contributions in eqs .
( [ exchange - frag ] ) and ( [ interference - frag ] ) , using the expressions for the absolute squares of the polarizations in eq .
( [ polarization - sums ] ) , multiplying by the factor in eq .
( [ norm - factor ] ) , and carrying out the phase - space integration in the phase space is unity .
this follows from the fact that there is no integration over the momentum of the @xmath1 pair or from the fact that the two final - state particles , namely , the @xmath1 pair and the single @xmath4 , are distinct . ] in eq .
( [ 2p - ps ] ) , we obtain the following contributions to the fragmentation functions : @xmath137 } j_2 \bigg
[ \bigg ( \frac{2-z}{2z } m \bigg)^2 \bigg ] \nonumber \\ & & \hspace{4ex } + m^2 a_3^{[8 ] } j_3 \bigg [ \bigg ( \frac{2-z}{2z } m \bigg)^2 \bigg ] + m^4 a_4^{[8 ] } j_4 \bigg [ \bigg ( \frac{2-z}{2z } m \bigg)^2 \bigg ] \bigg\ } , \\ { } \sum_{i=5}^{12}d_{i}(z,\lambda=0 ) & = & \frac{g_s^4 c_f ( 1-z)^2}{2 \pi n_c^2 z^{5 + 2 \epsilon } ( 2-z)^2 m } \bigg\{l_2^{[8 ] } j_2 \bigg [ \bigg ( \frac{2-z}{2z } m \bigg)^2 \bigg ] \nonumber \\ & & \hspace{4ex } + m^2 l_3^{[8 ] }
j_3 \bigg [ \bigg ( \frac{2-z}{2z } m \bigg)^2 \bigg ] + m^4 l_4^{[8 ] } j_4 \bigg [ \bigg ( \frac{2-z}{2z } m \bigg)^2 \bigg]\bigg\ } , % ---------------\end{aligned}\ ] ] where the dimensionless coefficients @xmath138}$ ] and @xmath139}$ ] are given in terms of @xmath30 and @xmath98 by @xmath140}&= & z^2 ( z-1 ) [ -(9z^2 + 4z+4 ) + 2 \epsilon ( 3 z^2 + 4 ) + \epsilon^2 ( 5 z^2 - 4 z-12 ) + 2 \epsilon^3 ( 4 - 4 z+z^2 ) ] \nonumber \\ & & -2 n_c z ( z-2 ) [ -2 ( 5 z^2 - 5 z+2 ) + \epsilon ( -z^3 + 8 z^2 - 6 z+4 ) + \epsilon^2 ( z^3 - 2 z^2 ) ] , \\ { } a_3^{[8 ] } & = & - z ( z-1 ) ( z-2 ) [ -2 ( z^2 + 6 z-4 ) + \epsilon z ( z+6 ) + 2 \epsilon^2 z ( z-2 ) ] \nonumber \\ & & - 2 n_c z ( z-1 ) ( z-2)^2 ( 3 - 2 \epsilon ) , \\ { } a_4^{[8 ] } & = & -(3 - 2 \epsilon ) ( z - 1)^2 ( z - 2)^2 , \\ { } l_2^{[8 ] } & = & z^4 [ z+2+\epsilon ( z-2)]^2 - 4 n_c z^3 ( z-2 ) [ z+2+\epsilon(z-2 ) ] , \\ { } l_3^{[8 ] } & = & -z^2 ( z-2 ) [ 4 ( z-1 ) ( z+2 ) + \epsilon ( z-2 ) ( z^2 + 4 z-8 ) ] \nonumber \\ & & + 2 n_c z ( z-2)^2 [ 4 ( z-1 ) + \epsilon ( z-2)^2 ] , \\ { } l_4^{[8 ] } & = & 2 [ 2 ( z - 1 ) + \epsilon ( z - 2)^2 ] ( z - 1 ) ( z - 2)^2
. % ---------------\end{aligned}\ ] ] here , the terms that are proportional to @xmath57 come from the contributions in eq . ( [ interference - frag ] ) .
such terms do not appear in @xmath141}$ ] and @xmath142}$ ] .
again , we retain the full @xmath98 dependence , as it may be useful for calculations of fragmentation functions at higher orders in @xmath5 . taking the limit @xmath91 ,
writing the result in the nrqcd - factorized form , and replacing free @xmath1 ldmes with quarkonium ldmes , we obtain [ equal - frag - eps-5 - 12 ] @xmath143\times \frac{\langle o^h({}^3s_1^{[8 ] } ) \rangle}{3 ( n_c^2 - 1 ) } , \label{equal - frag - eps-0-pol - sum}\\ d_{q\to h(\lambda=0)}^{(\textrm{5 - -12 } ) } & = & \frac{\alpha_s^2 c_f z ( 1-z)^2 } { 3 n_c^2 ( 2-z)^6 m_q^3 } \big [ 3 z^4 - 24 z^3 + 64 z^2 - 32 z+16 \nonumber\\ & & + 12 n_c ( 2-z)^2 ( 4-z ) ] \frac{\langle o^h({}^3s_1^{[8 ] } ) \rangle}{3 ( n_c^2 - 1)}. % ---------------\end{aligned}\ ] ] one obtains the complete fragmentation functions for the case in which the initial quark @xmath3 and the quark @xmath4 that is a constituent of the quarkonium have the same flavor by adding the contributions in eq .
( [ equal - frag - eps-5 - 12 ] ) to the contributions in eq .
( [ ms - bar - frag - unequal ] ) with @xmath144 set equal to @xmath145 .
the results are @xmath146 \frac{\langle o^h({}^3s_1^{[8 ] } ) \rangle } { 3 ( n_c^2 - 1 ) } , \label{equal - frag - eps-0 } \\ & & d_{q\to h(\lambda=0 ) } ( z ) = \frac{\alpha_s^2 c_f(1-z)^2}{3 n_c^2 z ( 2-z)^6 m_q^3 } \bigg [ 12 n_c^2 ( 2-z)^4 + 12 n_c z^2(2-z)^2 ( 4-z ) \nonumber \\ & & \hspace{20ex } + z^2(3 z^4 - 24 z^3 + 64 z^2 - 32 z+16 ) \bigg ] \frac{\langle o^h({}^3s_1^{[8 ] } ) \rangle}{3 ( n_c^2 - 1)}. \label{equal - frag - eps-0-long - pol } % ---------------\end{aligned}\ ] ] identical expressions hold for the case of an initial antiquark .
our polarization - summed result in eq . ( [ equal - frag - eps-0 ] ) differs from the result in eq .
( 111 ) of ref .
@xcite , which was duplicated in eq .
( a3 ) of ref .
@xcite . in ref .
@xcite , the color - octet fragmentation function was obtained by multiplying the color - singlet fragmentation function by a color factor . consequently , the contributions of the diagrams of figs .
[ fig : unequal ] and [ fig : interference ] were omitted .
our result in eq .
( [ equal - frag - eps-0 ] ) agrees with that in eq .
( c29 ) of ref .
@xcite . our longitudinal - polarization result in eq .
( [ equal - frag - eps-0-long - pol ] ) agrees with the result in eq .
( c.16 ) of ref .
@xcite , once one takes into account the fact that the ldme in eq .
( [ equal - frag - eps-0-long - pol ] ) is a factor of three larger than the corresponding ldme in ref .
in this section , we compute the fragmentation function for a quark to fragment through a spin - triplet color - singlet @xmath0-wave pair @xmath147})$ ] .
this process proceeds at leading order in @xmath5 only if the fragmenting quark has the same flavor as the quark in the @xmath1 pair .
the feynman diagrams that contribute to this process at leading order in @xmath5 are those in fig .
[ fig : exchange ] .
the corresponding contributions to the fragmentation function for a quark to fragment into a @xmath147})$ ] pair are identical to the contributions @xmath148@xmath149 in eq .
( [ exchange - frag ] ) , except that the color projectors @xmath150 are replaced with color projectors @xmath151 . in each contribution , the overall color factor is now @xmath152 instead of @xmath153 [ eq .
( [ exchange - cf ] ) ] . the dependence on @xmath87 in the expressions in eq .
( [ exchange - frag ] ) comes from the quark- and gluon - propagator denominators , which contribute factors @xmath154^{-1}$ ] .
thus , the integrals of these expressions over @xmath87 can be expressed in terms of the scalar integrals @xmath136 $ ] [ eq .
( [ jn ] ) ] . summing over the contributions in eq .
( [ exchange - frag ] ) , but taking the color factor in eq .
( [ singlet - cf ] ) , using the expressions for the absolute squares of the polarizations in eq .
( [ polarization - sums ] ) , multiplying by the factor in eq .
( [ norm - factor ] ) , and carrying out the integration over the phase space in eq .
( [ 2p - ps ] ) , we obtain the following contributions to the fragmentation functions : @xmath155})(\lambda)}^{(\textrm{5 - -8 } ) } & = & \frac { g_s^4 c_f^2(1-z)^2}{3 \pi z^{3 + 2 \epsilon } ( 2-z)^2 m } \biggl\{a_2^{[1 ] } j_2\left[\left(\tfrac{2-z}{2z } m\right)^2 \right ] + m^2 a_3^{[1 ] } j_3\left[\left(\tfrac{2-z}{2z } m\right)^2 \right]\nonumber\\ & & + m^4 a_4^{[1 ] } j_4\left[\left(\tfrac{2-z}{2z } m\right)^2 \right ] \biggr\ } , \\ d_{q\to q\bar q(^3s_1^{[1]})(\lambda=0)}^{(\textrm{5 - -8 } ) } & = & \frac { g_s^4 c_f^2(1-z)^2}{3 \pi z^{5 + 2 \epsilon } ( 2-z)^2 m } \biggl\{l_2^{[1 ] } j_2\left[\left(\tfrac{2-z}{2z } m\right)^2 \right ] + m^2 l_3^{[1 ] } j_3\left[\left(\tfrac{2-z}{2z } m\right)^2 \right]\nonumber\\ & & + m^4 l_4^{[1 ] } j_4\left[\left(\tfrac{2-z}{2z } m\right)^2 \right ] \biggr\ } , % ---------------\end{aligned}\ ] ] where the dimensionless coefficients @xmath156}$ ] and @xmath157}$ ] are given in terms of @xmath30 and @xmath98 by @xmath158 } & = & z^2 [ ( 9 z^2 + 4 z + 4 ) -\epsilon ( 6 z^2 + 8 ) - \epsilon^2 ( 5 z^2 - 4 z-12 ) - 2 \epsilon^3 ( z^2 - 4 z + 4 ) ] , \\ { } a_3^{[1 ] } & = & z ( z - 2 ) [ - 2 ( z^2 + 6 z - 4 ) + \epsilon z ( 2 \epsilon z - 4 \epsilon + z + 6 ) ] , \\ { } a_4^{[1 ] } & = & ( 3 - 2 \epsilon ) ( z - 1 ) ( z - 2)^2 , \\ { } l_2^{[1 ] } & = & [ z + 2 + \epsilon ( z - 2 ) ] ^2 z^4 , \\ { } l_3^{[1 ] } & = & - ( z-2 ) z^2 [ 4 ( z + 2 ) ( z - 1 ) + \epsilon ( z^2 + 4 z - 8) ( z - 2 ) ] , \\ { } l_4^{[1 ] } & = & 2 ( z - 1 ) ( z - 2)^2 [ 2 ( z - 1 ) + \epsilon(z - 2)^2 ] . % ---------------\end{aligned}\ ] ] taking the limit @xmath159 , writing the result in the nrqcd - factorized form , and replacing free @xmath1 ldmes with quarkonium ldmes , we obtain @xmath160 } ) \rangle,\label{sing - frag - sum}\\ d_{q\to h(\lambda=0)}&= & \frac{\alpha_s^2 c_f^2 z ( 1-z)^2 ( 3 z^4 - 24 z^3 + 64 z^2 - 32 z+16 ) } { 27 n_c ( 2-z)^6 m_q^3 } \langle o^h({}^3s_1^{[1 ] } ) \rangle,\label{sing - frag - long } % ---------------\end{aligned}\ ] ] where the nrqcd ldme is defined by @xmath161 } ) \rangle= \langle 0| \chi^\dagger\sigma^i \psi \sum_{\lambda}{\cal p}_{h(p,\lambda ) } \psi^\dagger\sigma^i\chi|0\rangle,\ ] ] and , in writing the result in the nrqcd - factorized form , we have used the fact that @xmath162})}({}^3s_1^{[1 ] } ) \rangle = 2n_c(d-1).\ ] ] the result in eq .
( [ sing - frag - sum ] ) agrees with the result in eq .
( 16 ) of ref .
@xcite and with the result in eq .
( c24 ) of ref .
@xcite , once one takes into account the fact that the ldme in eq .
( [ sing - frag - sum ] ) is a factor of @xmath163 larger than the corresponding ldme in ref .
our result in eq .
( [ sing - frag - long ] ) agrees with the result in eq .
( c.11 ) of ref .
@xcite , once one takes into account the fact that the ldme in eq .
( [ sing - frag - long ] ) is a factor of @xmath164 larger than the ldme in ref . @xcite .
in this paper we computed fragmentation functions for a quark @xmath3 to fragment into a heavy quarkonium through a heavy @xmath1 channel in which the @xmath1 pair is in a spin - triplet , @xmath0-wave state .
our computations are at the leading nontrivial order in @xmath5 and at leading order in the heavy - quark velocity @xmath7 .
we have considered the following cases : ( i ) the @xmath3 and @xmath4 have different flavors , in which case the @xmath1 pair must be in a color - octet state ; ( ii ) the @xmath3 and @xmath4 have the same flavor and the @xmath1 pair is in a color - octet state ; ( iii ) the @xmath3 and @xmath4 have the same flavor and the @xmath1 pair is in a color - singlet state . in each case , we have computed both the fragmentation function summed over all spin polarizations of the @xmath9 pair and the fragmentation function for longitudinal polarization of the @xmath1 pair . we have also presented expressions in @xmath11 dimensions , which may be useful in carrying out calculations of fragmentation functions at higher orders in @xmath5 .
our results for case ( i ) are given for finite @xmath21 in eq .
( [ ms - bar - frag - unequal ] ) and for @xmath165 in eq .
( [ unequal - zero - mass ] ) .
the result in eq .
( [ ms - bar - frag - unequal - sum ] ) for the sum over @xmath1 polarizations agrees with previous calculations in refs .
@xcite , and the result in eq .
( [ unequal - zero - mass - sum - pol ] ) for the sum over @xmath1 polarizations agrees with a calculation of quark fragmentation into lepton pairs in ref .
@xcite , once differences in the phase - space and color factors have been taken into account .
the result in eq .
( [ ms - bar - frag - unequal - long ] ) corrects the result in ref .
this correction has been confirmed by the author of ref .
@xcite .
our result for case ( ii ) for the fragmentation function summed over @xmath1 polarizations is given in eq .
( [ equal - frag - eps-0 ] ) and agrees with the result in ref .
@xcite , but disagrees with the result in ref .
our result for case ( ii ) for the longitudinal - polarization fragmentation function is given in eq .
( [ equal - frag - eps-0-long - pol ] ) and agrees with the result in ref .
@xcite .
our result for case ( iii ) for the fragmentation function summed over @xmath1 polarizations is given in eq .
( [ sing - frag - sum ] ) and agrees with the results in refs .
our result for case ( iii ) for the longitudinal - polarization fragmentation function is given in eq .
( [ sing - frag - long ] ) and agrees with the result in ref .
@xcite . the new results for longitudinally polarized fragmentation functions that we have obtained in this paper will make it possible to compute quark - initiated leading - power fragmentation contributions to the production of polarized @xmath0-wave spin - triplet quarkonia .
while gluon - initiated fragmentation dominates quark - initiated fragmentation in quarkonium hadroproduction , quark - initiated fragmentation may be important for other quarkonium production processes , such as production in @xmath166 annihilation .
as we have mentioned , our calculations are at the leading nontrivial order in @xmath5 and @xmath7 .
a complete calculation at order @xmath167 of the lp contributions to quarkonium production in the color - octet channels requires corrections to the fragmentation functions for the @xmath168 and @xmath169 channels at next - to - leading order ( nlo ) in @xmath5 and for the @xmath170 channel through next - to - next - to - leading order in @xmath5 .
corrections of nlo in @xmath5 to the fragmentation function for a gluon to fragment into a quarkonium through the @xmath170 color - octet channel have already been computed @xcite and give a contribution that is numerically large in comparison with the leading - order ( lo ) contribution @xcite .
corrections of higher order in @xmath7 are also known to be large , relative to the lo contribution , for the fragmentation functions for gluons to fragment into quarkonia through the @xmath170 color - singlet and color - octet channels @xcite .
these large corrections of higher order in @xmath5 and @xmath7 suggest that it may be important to compute higher - order corrections for additional quarkonium production channels and for quark - initiated , as well as gluon - initiated , fragmentation processes .
we thank jianwei qiu for pointing out ref .
@xcite to us .
we are grateful to hong zhang for helpful discussions regarding comparisons between our work and the results in refs .
the work of g.t.b . and h.s.c .
is supported by the u.s .
department of energy , division of high energy physics , under contract no .
de - ac02 - 06ch11357 . the work of u - r.k .
is supported by the national research foundation of korea under contract no .
the submitted manuscript has been created in part by uchicago argonne , llc , operator of argonne national laboratory .
argonne , a u.s .
department of energy office of science laboratory , is operated under contract no.de-ac02-06ch11357 .
collins and d.e .
soper , nucl .
b * 194 * , 445 ( 1982 ) .
ma , phys . rev .
d * 53 * , 1185 ( 1996 ) [ hep - ph/9504263 ] .
y. q. ma , j. w. qiu and h. zhang , phys . rev .
d * 89 * , 094029 ( 2014 ) [ arxiv:1311.7078 [ hep - ph ] ] .
e. braaten , k. cheung , and t.c .
yuan , phys .
d * 48 * , 4230 ( 1993 ) [ hep - ph/9302307 ] .
h. zhang , `` qcd factorization for heavy quarkonium production and fragmentation functions , '' a dissertation for the degree of doctor of philosophy in physics , stony brook university , august , 2014 + ( ` http://graduate.physics.sunysb.edu/announ/theses/zhang-hong-august-2014.pdf ` ) .
r. barbieri , r. gatto , and r. kogerler , phys .
b * 60 * , 183 ( 1976 ) .
r. barbieri , r. gatto , and e. remiddi , phys .
b * 61 * , 465 ( 1976 ) .
chang , nucl .
b * 172 * , 425 ( 1980 ) .
b. guberina , j.h .
khn , r.d .
peccei , and r. rckl , nucl .
b * 174 * , 317 ( 1980 ) .
e. l. berger and d. l. jones , phys . rev .
d * 23 * , 1521 ( 1981 ) .
bodwin and a. petrelli , phys . rev .
d * 66 * , 094011 ( 2002 ) .
p. cho , m.b .
wise , and s.p .
trivedi , phys .
d * 51 * , r2039 ( 1995 ) [ hep - ph/9408352 ] .
bodwin , u - r .
kim , and j. lee , jhep * 1211 * , 020 ( 2012 ) [ arxiv:1208.5301 [ hep - ph ] ]
bodwin , e. braaten , and g.p .
lepage , phys .
d * 51 * , 1125 ( 1995 ) [ erratum - ibid .
d * 55 * , 5853 ( 1997 ) ] [ hep - ph/9407339 ] .
e. braaten and j. lee , phys .
d * 65 * , 034005 ( 2002 ) [ hep - ph/0102130 ] .
yuan , phys .
d * 50 * , 5664 ( 1994 ) [ hep - ph/9405348 ] .
kniehl and g. kramer , phys .
d * 56 * , 5820 ( 1997 ) [ hep - ph/9706369 ] .
e. braaten and j. lee , nucl .
b * 586 * , 427 ( 2000 ) [ hep - ph/0004228 ] .
g. t. bodwin , h. s. chung , u. r. kim and j. lee , phys .
* 113 * , no . 2 , 022001 ( 2014 ) [ arxiv:1403.3612 [ hep - ph ] ] . g. t. bodwin and j. lee , phys .
d * 69 * , 054003 ( 2004 ) [ hep - ph/0308016 ] .
y. q. ma , j. w. qiu and h. zhang , arxiv:1501.04556 [ hep - ph ] . | we compute fragmentation functions for a quark to fragment to a quarkonium through an @xmath0-wave spin - triplet heavy quark - antiquark pair .
we consider both color - singlet and color - octet heavy quark - antiquark ( @xmath1 ) pairs .
we give results for the case in which the fragmenting quark and the quark that is a constituent of the quarkonium have different flavors and for the case in which these quarks have the same flavors .
our results for the sum over all spin polarizations of the @xmath1 pairs confirm previous results .
our results for longitudinally polarized @xmath1 pairs agree with previous calculations for the same flavor cases and correct an error in a previous calculation for the different - flavor case . |
as a chinese national research facility , the guoshoujing telescope ( also named the lamost and wang - su reflecting schmidt telescope , cui et al .
2012 ) is an innovative quasi - meridian reflecting schmidt telescope capable of simultaneously recording spectra of up to 4,000 objects in a large field of view ( fov ) of 5 in diameter .
the telescope is equipped with 16 low - resolution spectrographs , 32 ccds and 4,000 fibres .
each spectrograph is fed with the light from 250 fibres and covering a wavelength range 3700
9100 at a resolving power @xmath10 ( with slit masks of width of two thirds the fibre diameter , i.e. 2.2arcsec ; zhu et al .
2006 ; zhu et al .
2010 ) . with an effective aperture ranging from 3.6
4.9 m in diameter , depending on the pointing direction , the lamost can reach a faint limiting magnitude of @xmath11mag in a total integration time of 90 minutes under favourable observing conditions .
the telescope is located in xinglong station and operated by the national astronomical observatories , chinese academy of sciences ( naoc ) . following a two - year period of commissioning initiated in september 2009 that solved the problem of auto - positioning of the 4,000 fibres and a year long pilot surveys ( luo et al .
2012 ) from october 2011 june 2012 to test and verify the designs and scientific goals of the various components of the surveys , the lamost regular surveys began in october 2012 .
the lamost regular surveys consist of two parts ( zhao et al .
2012 ) : the lamost extragalactic survey ( legas ) and the lamost experiment for galactic understanding and exploration ( legue ; deng et al .
the legue has three components : the spheroid , the anti - centre and the disc surveys , each focusing on a distinct component of the galaxy and having different survey footprints and target selection algorithms .
the spheroid survey plans to observe at least 2.5 million stars in the galactic caps ( @xmath12 ) , down to a limiting magnitude of @xmath13 mag ( to 18.5 mag for limited fields ) , with targets selected from the fourth united states naval observatory ccd astrograph catalog ( ucac4 ; zacharias et al .
2013 ) and the panstarrs 1 ( ps1 ; kaiser 2004 ) catalogues using a preferential target selection algorithm ( carlin et al .
2012 ; zhang et al . 2012 ; yang et al .
the disc survey ( chen et al . 2012 ) plans to cover as much area of low galactic latitudes ( @xmath14 ) as visible from xinglong station and allowed by the available observing time , focusing on open clusters and selected star - forming regions in the galactic ( thin ) disk .
the targets are selected from a combination of the ucac4 , ps1 and the two micro all sky survey ( 2mass ; skrutskie et al .
2006 ) catalogues , down to a limiting magnitude of @xmath13 mag ( to 18.5 mag for limited fields ) . as a major component of the legue , the lamost spectroscopic survey of the galactic anti - centre ( lss - gac ; liu et al .
2014 ) aims to collect spectra for a statistically complete sample of @xmath5 million stars of all colours down to a limiting magnitude of @xmath13 mag ( to 18.5 mag for limited fields ) , distributed in a large and continuous sky area of over 3,400 deg@xmath0 centred on the galactic anti - centre ( @xmath1 , @xmath15 ) and covering a significant volume of the galactic thin / thick discs and halo . of the legue survey footprints ,
the area of galactic anti - centre is particularly interesting for several reasons : 1 ) disk is the defining component of a giant spiral galaxy like our own milky way in terms of stellar mass , angular momentum and star formation activities .
it is also the most challenging component of a galaxy to study .
the galactic disc has already been known to exhibit rich yet poorly - understood ( sub-)structures ( e.g. , minniti et al .
2011 ; bovy et al .
2012 ; gmez et al .
2012 ; li et al . 2012 ; liu et al . 2012 ; widrow et al . 2012 ; carlin et al . 2013 ; williams et al . 2013 ) . restricted by the observing capability , previous spectroscopic surveys including the sloan extension for galactic understanding and exploration ( segue ; yanny et al .
2009 ) , the apache point observatory galactic evolution experiment ( apogee ; eisenstein et al .
2011 ) and the radial velocity experiment ( rave ; steinmetz et al .
2006 ) suffer from various limitations , including complicated and hard to characterize target selection criteria , small or sporadic sky coverage , low sampling density or shallow survey depth , making it difficult to use them to establish a global census of the structure , stellar populations , kinematics and chemistry of the galactic disks . with an unprecedented number of 4,000 fibres afforded by the lamost , building a deep
, systematic spectroscopic survey targeting millions of stars drawn from a significant and contiguous volume of the galactic discs with simple yet nontrivial algorithms has become , for the first time , feasible ; 2 ) the lss - gac is ideally matched to the seasonal variation of weather conditions of the site , which has most observable nights spreading from september to march the following year
( yao et al . 2012 ) ; 3 ) the stellar number density in the gac direction is high enough to ensure full and efficient use of the unique multiplexing capability of lamost , i.e. 4,000 fibres ; 4 ) the interstellar extinction in the direction of gac ( schlegel et al . 1998 ; hereafter sfd ) is moderate even in the disk . within the footprint of lss - gac ,
about 70 per cent of the region has a colour excess @xmath16 less than 0.5 mag , and only about 10 per cent has @xmath16 larger than 1.0mag . being the largest and most luminous galaxy of the local group , m31 hosts a variety of interesting targets including planetary nebulae ( pne ) , hii regions , supergiants and globular clusters that are easily detectable with the lamost ( yuan et al .
2010 ; huo et al .
2010 , 2013 ) .
kinematic and elemental abundance studies of pne in m31 provide important information of the chemical composition , kinematics and structure of m31 as well as the surrounding , extremely extended and complex stellar streams revealed by the recent deep imaging surveys ( e.g. , ibata et al .
2001 , 2007 ; mcconnachie et al . 2009 ) , and are thus of great interest for the understanding of the assemblage history of the local group .
background quasars in the vicinity fields of m31 and m33 can potentially provide an excellent astrometric reference frame to measure the minute proper motions of m31 and m33 and their associated substructures .
bright quasars are also excellent tracers to probe via absorption line spectroscopy the kinematics and chemistry of the interstellar / intergalactic medium of the local group .
the direction towards m31 and m33 also harbours rich foreground galactic substructures ( bonaca et al . 2012 ; chou et al . 2011 ; majewski et al . 2004 ; martin et al .
2007 , 2014 ; rocha - pinto et al . 2004 ) . for the above reasons , m31 , m33 and the vicinity fields
have been observed with the lamost since the commissioning phase as parts of the lss - gac program , targeting foreground galactic stars , accessible objects of special interests in m31 and m33 as well as background quasars . to make full use of the telescope , randomly selected very bright ( vb ) stars of 9 @xmath17mag in sky areas accessible to the lamost ( @xmath18 dec @xmath19 )
are targeted using observing time of non - ideal conditions , including bright / grey lunar nights .
the vb survey is supplementary to the lss - gac main survey in the sense that it has a much larger sky coverage but brighter magnitude limit and uses a simpler target selection algorithm .
the vb survey thus yields an excellent sample of the local stars to probe the solar neighbourhood . a concise description of the scientific motivations , survey design and target selection , observations and data reduction and preliminary scientific results of lss - gac has been previously presented in liu et al .
( 2014 ) .
the lss - gac will deliver classifications , radial velocities and stellar atmospheric parameters ( , log @xmath20 , [ fe / h ] , @xmath21-element to iron abundance ratio [ @xmath21/fe ] and carbon to iron abundance ratio [ c / fe ] ) for a magnitude - limited , spatially continuous and statistically complete sample of about 3 million stars in the galactic thin / thick discs and halo .
the recently successfully launched astrometric satellite gaia ( perryman et al . 2001
; katz et al . 2004 ) will yield accurate proper motions and parallaxes for one billion galactic stars to @xmath22mag , radial velocities of 60 100 million stars to @xmath23 15 16mag , and atmospheric parameters of @xmath7 5 million stars to @xmath23 13mag .
the lss - gac forms excellent synergy with gaia . together with measurements of distances and tangential velocities provided by gaia
, the lss - gac will yield a unique dataset to : \a ) characterize the stellar populations , chemical composition , kinematics and structure of the thin / thick discs and their interface with the halo ; \b ) identify tidal streams and debris of disrupted dwarf galaxies and star clusters ; \c ) understand how resilient galaxy discs are to gravitational interactions ; \d ) study the temporal and secular evolution of the disks ; \e ) probe the gravitational potential and dark matter distribution ; \f ) map the interstellar extinction as a function of distance ; \g ) search for rare objects ( e.g. stars of peculiar chemical composition or hyper - velocities ) ; \h ) study variable stars and binaries with multi - epoch spectroscopy ; \i ) and ultimately advance our understanding of the assemblage of galaxies and the origin of their regularity and diversity .
following year - long pilot surveys , the lamost regular surveys were initiated in the fall of 2012 .
the surveys are expected to complete in june 2017 . by june 2013 ,
about 1 million spectra of @xmath24 per pixel have been collected , including 0.44 , 0.12 and 0.48 million from the main , m31/m33 and vb surveys , respectively .
the lamost early data release ( luo et al .
2012 ) , containing 319,000 spectra taken in the pilot surveys and a catalogue of those objects , was publicly released to the international community in august 2012 .
the first lamost regular data release ( dr1 ; bai et al .
2014 ) , available to the chinese astronomical community and the international partners from september 2013 and scheduled to be released to the whole international community in december 2014 according to the lamost data policy , contains 1.8 million calibrated stellar spectra collected during the pilot and the first year regular surveys and radial velocities derived from them .
about 21.0 , 3.7 and 28.4 per cent of the released spectra of lamost dr1 are from the lss - gac main , m31/m33 and vb surveys , respectively .
the lamost dr1 also includes a main catalogue providing stellar atmospheric parameters , log @xmath20 and [ fe / h ] of 1,085,404 a- , f- , g- and k - type stars , a supplementary catalogue giving spectral subtypes , luminosity classes and magnetic activities of 122,678 m - type stars ( yi et al . 2014 ) and another one listing spectral subtypes and luminosity classes of 101,513 a - type stars . in the main catalog , about 13.9 , 2.3 and 29.5 per cent stars are from the lss - gac main , m31/m33 and vb surveys , respectively . in this paper , we present a detailed description of the lss - gac , including the target selection , survey design , observations and data reduction . the first release of lss - gac value - added catalogues , complementary to the lamost dr1 and publicly available to the world - wide community ,
is described .
stellar parameters in the main catalogue of lamost dr1 are determined with the lamost stellar parameter pipeline ( lasp ; wu et al .
2014 ) by template matching with the elodie spectral library ( prugniel & soubiran 2001 ; prugniel et al .
the value - added catalogues presented in the current work contain values of @xmath25 , , log @xmath20 and [ fe / h ] derived from lamost spectra with the lamost stellar parameter pipeline at peking university ( lsp3 hereafter ; xiang et al .
2014a , b ) , values of the interstellar extinction and stellar distance determined by a variety of methods , multi - band photometry collected from the galaxy evolution explorer ( galex ; martin et al . 2005 ) in the ultraviolet , the xuyi schmidt telescope photometric survey of the galactic anti - center ( xstps - gac ; liu et al .
2014 ; zhang et al .
2014 ) in the optical and the 2mass and the wide - field infrared survey explorer ( wise ; wright et al . 2010 ) in the near infrared , proper motions compiled from various sources and corrected for systematics , as well as estimated orbital parameters for about 0.67 million stars surveyed under the umbrella of lss - gac for the survey period covered by the lamost official dr1 release .
the properties of lss - gac samples in this data release are discussed .
the paper is organized as follows .
section 2 presents a detailed description of the target selection algorithm and survey design of lss - gac .
survey progress and data reduction are described in section 3 .
the lsp3 stellar parameters are discussed in section 4 .
the methods and results of extinction and distance determinations are given in sections 5 and 6 , respectively .
compilations and comparisons of proper motions and estimates of stellar orbital parameters are described in section 7 .
the properties of lss - gac sample stars accumulated hitherto are discussed and the value - added catalogues described in section 8 .
we close with a summary of the main results in section 9 .
but for values of @xmath16 of the interstellar extinction .
the map is constructed from the data of schlegel et al .
( 1998 ) . , width=340 ] the lss - gac aims to deliver spectral classifications , radial velocities and atmospheric parameters ( , log @xmath20 , [ fe / h ] , [ @xmath21/fe ] and [ c / fe ] ) for a statistically complete sample of about 3 million stars of all colours down to a limiting magnitude of @xmath26 17.8mag ( 18.5mag for selected fields ) , distributed in a contiguous area of 3,438deg@xmath0 ( @xmath27 ) , sampling a significant volume of the galactic thin / thick discs and halo .
there are several constraints needed to be considered when designing the lss - gac survey : 1 ) for the desired limiting magnitude , there are over 10,000 stars per deg@xmath0 on the disc and about 3,000 stars per deg@xmath0 at 30@xmath28 above / below the disc within the lss - gac footprint ( figs.[footprint ] and [ r18.5 ] ) , too many to survey for a reasonable survey lifetime , say 5 years , even with 4,000 fibres offered by the lamost , six times more than available to the sloan digital sky survey ( sdss ; york et al .
2000 ) ; 2 ) the spatial distribution of stars is highly non - uniform , and this is further compounded by the relatively high interstellar extinction in this region ; 3 ) as a fully developed grand - design spiral , the galaxy , in particular its disk , consists of extremely complex and varying stellar populations ; 4 ) the fov of lamost plates must be centred on stars brighter than @xmath29mag for the active optics to work so as to bring the individual segments of the primary and corrector mirrors into focus .
stars of such brightness are relatively rare and not uniformly distributed on the sky ; 5 ) the lamost has a circular fov , 5deg in diameter , so field overlapping can not be avoided in order to make a contiguous sky coverage ; and 6 ) except for a few holes occupied by the four guiding cameras and the central shack - hartmann sensor , the 4,000 fibres of lamost , each controlled by two motors and patrolling a sky area of @xmath30arcmin in diameter around its parking position , are more or less uniformly distributed on the focal plane . given these constraints , a simple yet non - trivial target selection algorithm and sophisticated survey strategy
must be developed in order to achieve the defining scientific goals of lss - gac .
lss - gac targets are selected from the xstps - gac photometric catalogues ( liu et al .
2014 ; zhang et al .
2013 , 2014 ; yuan et al . in prep .
the basic idea of lss - gac target selection is to uniformly and randomly select stars from the colour - magnitude diagrams using a monte carlo method .
it has the advantages that : 1 ) it has a well defined selection function , such that whatever class of objects ( e.g. white dwarfs , white - dwarf - main - sequence binaries , extremely metal - poor or hyper - velocity stars ) are revealed by spectroscopic observations , they can be studied in a statistically meaningful way in terms of the underlying stellar populations for the survey volume , after various selection effects have been properly taken into account ; 2 ) stars of all colours ( spectral types ) and magnitudes ( distances ) are selected in large numbers as much and as evenly as possible ; 3 ) rare objects of extreme colours are first selected and targeted with high priorities , thus vastly increasing the discovery space of the survey .
based on the actual measured performance of lamost , the scientific goals and data quality requirements of lss - gac , the bright and faint limiting magnitudes of the main survey of lss - gac have been set at 14.0 and @xmath7 17.8mag in @xmath6-band , respectively . for a small number of selected fields , the limiting magnitude are set at 18.5mag . to make efficient use of observing time of different qualities and avoid fibre cross - talking , three categories of spectroscopic plates are designated , bright ( b ) , medium - bright ( m ) and faint ( f ) , targeting sources of brightness 14.0mag@xmath31 , @xmath32 and @xmath33mag , respectively . here
@xmath34 and @xmath35 are the border magnitudes separating b , m and f plates ( discussed later ) . above the galactic plane ( @xmath36 @xmath9 3.5@xmath37 )
, @xmath38 stars will be surveyed per deg@xmath0 , i.e. about five plates on average ( 2b , 2 m and 1f ) will be allocated for a given patch of sky . on the galactic plane ( @xmath36 @xmath39 3.5@xmath37 ) ,
the sampling is doubled and about ten plates ( 4b , 4 m , 2f ) will be allocated for a given patch of sky . given the limited amount of observing time of exceptional quality that is needed to achieve the depth of f plates , only a finite number of designed f plates are expected to get observed during the five - year survey lifetime .
targets of all designed plates are selected , assigned and checked in advance .
based on the above principles , there are five steps in the lss - gac survey design and target selection .
firstly , a clean sample of targets is generated from the xstps - gac catalogues .
secondly , targets are assigned with different priorities and fed to the lamost survey strategy system ( sss ; yuan et al .
2008 ) , which allocates fibres to targets for each designed plate scheduled for observation .
thirdly , field centres of all designed plates , which have to be on a star brighter than @xmath40mag , are carefully selected and adjusted to maximize the spatial uniformity of sampling over the whole survey area .
fourthly , for each chosen centre of fov , the sss is run to assign targets ( along with sky fibres ) for all designed plates .
the designed plates are then ready for observational scheduling .
finally , to account for the field rotation of lamost , prior to the actually execution of scheduled observations ( typically in a week ) , the sss has to be rerun and fibres re - assigned for targets of each designed plate .
a small fraction of targets , less than @xmath41 ( excluding sky fibres ) , fail to get reallocated a fibre and thus will not get observed .
the xstps - gac ( liu et al . 2014 ) surveys in sdss @xmath20 , @xmath6 and @xmath42 bands an area of approximately 5,400deg@xmath0 centred on the gac , from @xmath43 to 9@xmath44 and @xmath45 to + 60@xmath46 , plus an extension of @xmath47deg@xmath0 to the m31 , m33 region , with the xuyi 1.04/1.20 m schmidt telescope equipped with a 4k@xmath484k ccd .
the resulting catalogues archive about a hundred millions stars down to a limiting magnitude of @xmath49mag ( @xmath50 ) .
the catalogues are astrometrically calibrated using the ppmxl ( roeser et al .
2010 ) as reference and globally flux - calibrated ( using the ubercal algorithm , padmanabhan et al .
2008 ) against the sdss dr8 ( aihara et al .
2011 ) using the overlapping fields .
the calibration has achieved an astrometric accuracy of 0.1arcsec ( zhang et al .
2014 ) and a global photometric accuracy of 2 per cent ( yuan et al . , in prep . ) .
the catalogues are used to generate a clean sample of targets for the lss - gac , with the requirements that : 1 ) the stars are detected in at least two bands , including @xmath6 band , and have magnitudes @xmath51mag ; 2 ) positions of stars measured in different bands agree within 0.5 arcsec ; 3 ) they are not flagged as galaxies nor star pairs in either @xmath6 or @xmath42 band ; 4 ) they have no neighbors within 5 arcsec that are brighter than ( @xmath52 + 1 ) mag .
, where @xmath52 is magnitude of the star concerned in each of the three bands ; and 5 ) they have local sky background no more than three times brighter than the sky background of the whole frame , i.e. , they are not around extremely bright stars .
fig.[r18.5 ] shows the distribution of stellar number densities per square arcmin of the thus constructed clean photometric sample within the lss - gac footprint . for comparison ,
a reddening map for the same region that shows values of @xmath16 of the interstellar extinction from sfd is shown in fig.[sfd ] .
the dusty galactic disc is clearly visible , and exhibits number densities over 10,000 stars perdeg@xmath0 after imposing the above cuts .
the stellar number densities decrease rapidly toward high galactic latitudes , down to about 3,000 per deg@xmath0 at @xmath53 . patchy or filamentary dark lanes or bands caused by obscuration of dark clouds are abundant . in the direction of gac ,
more dark clouds are found in the southern galactic hemisphere than in the northern hemisphere , as can be clearly seen in fig.[sfd ] .
there are a few roughly rectangle regions in fig.[r18.5 ] around longitude @xmath54 and latitude @xmath55 , with a total area of about 30deg@xmath0 , that have fewer stars than their surroundings due to the shallower limiting magnitudes caused by poor observing conditions when those regions were observed .
the black corner in the top right of fig.[r18.5 ] has declinations higher than @xmath56 , falling outside the xstps - gac footprint . before target priority assignment ,
we first need to estimate values of @xmath34 and @xmath35 , the border magnitudes separating b , m and f plates . due to the spatial variation of the stellar populations , the survey depths of xstps - gac , as well as the patchy distribution of the interstellar extinction , @xmath34 and @xmath35 vary slightly over the survey area , even within a single lamost field . fixing @xmath34 and @xmath35 to constant values for the whole survey area will lead to discontinuities of sample star number density in the colour - magnitude diagrams at the border magnitudes separating b , m and f plates . to avoid such undesired features , we have allowed values of @xmath34 and @xmath35 to float and determined their appropriate values for individual patches of sky of 1deg . by 1deg . in size .
the whole survey area is first divided into boxes of 1deg .
side in ra and dec . for stars in each box , ( @xmath6 , @xmath57 ) and ( @xmath6 , @xmath58 )
hess diagrams are constructed from the clean sample
. stars of extremely blue colours , @xmath59 or @xmath60 , and of extremely red colours , @xmath59 or @xmath61 , are first selected and removed from the diagrams .
stars in the remaining colour space are then selected using a monte carlo method .
first two uniformly distributed random numbers , @xmath6 in the range ( 14 , 18.5 ] and @xmath62 in ( @xmath63 , 5 ] , are generated . if @xmath64 , then the ( @xmath6 , @xmath57 ) hess diagram is used to select the next star , assuming @xmath65 .
otherwise the ( @xmath6 , @xmath58 ) hess diagram is used instead , assuming @xmath66 . for a given colour - magnitude set ( @xmath6 , @xmath57 ) or ( @xmath6 , @xmath58 ) , if there are stars in the appropriate hess diagram in a box centred on the simulated set of colour - magnitude and of length 0.2mag in magnitude and 0.3mag in colour , then the star of colour - magnitude values closest to the simulated set is chosen and removed from the pool .
if not , the process is repeated until a total of 1,000 stars , including those of extreme colours , per deg@xmath0 are selected deg@xmath0 , i.e. smaller than 1 deg@xmath0 for declination @xmath67deg . ] .
the stars are then sorted in magnitude from bright to faint .
then @xmath34 and @xmath35 are set to the faint end magnitudes of the first 40% and 80% sources , respectively .
the above procedure applies to the survey area of @xmath68 , where the lss - gac plans to sample 1,000 stars per deg@xmath0 . for @xmath69 , a similar procedure
is applied except that 2,000 stars per deg@xmath0are selected and the selections are carried out in ( @xmath70 , @xmath71 ) space instead of ( ra , dec ) .
fig.[ds9_m1 ] shows the spatial variations over the survey footprint of @xmath34 and @xmath35 .
@xmath34 has a typical value of 16.3mag and varies between 15.8
16.5mag for most regions , whereas @xmath35 has a typical value of 17.8mag and varies between 17.6 18.0mag for most regions .
both @xmath34 and @xmath35 decrease toward the galactic mid - plane . in regions where the survey depths of xstps - gac are shallower than the designed limiting magnitude of lss - gac , such as around ( @xmath72 ) = ( @xmath73 ) and ( @xmath74 ) , the values of @xmath34 and @xmath35 are smaller than those of the surrounding areas .
( top panel ) and @xmath35 ( bottom panel ) within the footprint of the lss - gac main survey .
, width=340 ] once the values of @xmath34 and @xmath35 are set , stars of b , m and f plates are selected and assigned priorities , in batches of 200 stars per deg@xmath0 .
for each category of plates , up to 10 batches are selected pending on the availability of stars .
the first batch of stars are assigned the highest priority , the second one priority lower , and so forth . when selecting each batch of stars , if a star of a given coordinate set , ( ra , dec ) or ( @xmath70 , @xmath71 ) , is selected , then all other stars within 2arcmin of that star are removed from the pool to avoid small scale clustering .
those removed stars are however put back to the pool when selecting the next batch of stars . as an example
, fig.[target_selection ] shows the spatial distributions in ( ra , dec ) and in the ( @xmath6 , @xmath75 ) and ( @xmath6 , @xmath76 ) hess diagrams for all stars and stars of different priorities in the clean sample , for a field of relatively high galactic latitude centred around @xmath77 and @xmath78 .
note for the 2nd 4th columns in the middle panels , only stars selected in the ( @xmath6 , @xmath75 ) hess diagram are plotted . while for the 2nd 4th columns in the bottom panels ,
only stars selected in the ( @xmath6 , @xmath76 ) hess diagram are plotted . as expected , for stars of a given priority ,
their spatial distribution is fairly uniform and clustering in the ( @xmath6 , @xmath75 ) and ( @xmath6 , @xmath76 ) diagrams is much reduced .
note also that for a given priority , given the broader colour range of @xmath57 , more targets are selected from the ( @xmath6 , @xmath75 ) diagram than that of ( @xmath6 , @xmath76 ) , especially for bright plates .
after priority assignment , we define the field centers , which , as stated earlier , must be centred on bright stars in order for the lamost active optics to operate .
we use the hipparcos and tycho catalogs ( perryman & esa 1997 ) , which contain 118,218 stars with accurate positions , proper motions and parallaxes and are complete to @xmath79band magnitudes between 7.3 and 9.0mag .
we require that the stars must be single and have magnitudes @xmath80mag in order to make the active optics work even under unfavourable conditions such as bright moon or partially cloudy nights . a couple of stars with a companion of separations less than 0.5 arcsec are however retained in places of few bright stars .
the lamost active optics works well under such small separations . to maximize the uniformity of spatial sampling over the whole lss - gac survey area and achieve the designed sampling density , three groups ( g1 , g2 and g3 hereafter ) of fields
are defined , with the first two groups covering the whole survey footprint and g3 for the galactic plane ( @xmath36 @xmath39 3.5@xmath37 ) only . for the g1 ,
g2 and g3 groups of fields , two ( 1 b and 1 m ) , three ( 1 b , 1 m and 1 f ) and five ( 2 b , 2 m and 1 f ) plates are planned , respectively .
we first partition the survey area in ( @xmath81 ) by equilateral triangles of length @xmath82 , assuming that the celestial sphere is flat in ( @xmath81 ) , not a bad assumption given that the survey extends to a latitude of only 30@xmath83 .
bright stars from the hipparcos and tycho catalogues that satisfy the aforementioned criteria and are closest to the centres of those equilateral triangles are selected as g1 field central stars .
central stars of g2 fields are then selected by shifting the equilateral triangles in order to fill up as much as possible the gaps that are not covered by the g1 fields .
central stars of g3 are selected in a similar way , except that the survey area ( @xmath84 , @xmath85 ) are partitioned by squares of length @xmath86 such that the central stars must have a latitude @xmath36 smaller than @xmath87 . in total , 216
, 216 and 34 central stars are defined as the g1 , g2 and g3 field centers , respectively , yielding a total of 1,250 plates ( 500 b , 500 m and 250 f ) .
their spatial distributions are displayed in fig.[central_stars ] .
the field central stars are not necessarily unique and two fields may share the same star .
there are 5 , 4 and 6 common stars shared by g1 and g2 , g1 and g3 , and g2 and g3 fields , respectively .
since the lamost fov is circular , field overlapping can not be avoided in order to achieve a contiguous sky coverage . in the ideal cases
where the centres of partition fall on qualified bright stars , the fraction of overlapping areas is about 20 per cent for g1 and g2 fields , and more for g3 fields .
there are two options to make use of the overlapping areas of adjacent fields .
one is to increase the sampling density in the overlapping regions by observing different targets in different plates .
the other is to make duplicate observations for some sources in the overlapping areas for time - domain spectroscopy .
the latter is preferred considering that a ) the sampling density is already high and the margin gained by increasing the sampling density is not huge ; b ) the sampling density will be more uniform over the whole survey footprint ; and most important of all , c ) it opens up the possibility of time - domain spectroscopy that may lead to new discoveries ( e.g. , yang et al . 2014 ) .
to implement such an approach , we require that there are no common targets amongst plates of different groups of fields , but plates belonging to the same group , target selections of overlapping plates are independent , i.e. fibres are assigned to stars regardless whether the stars have been assigned a fibre or not in the adjacent plates .
the procedure guarantees that adjacent plates belonging to the same group target , with few exceptions , the same sources in their overlapping areas .
the top panels of fig.[sampling_density ] display the spatial distributions of plate density of the lss - gac main survey for all , b ( or m ) and f plates .
the middle panels show the same after discounting overlapping regions of adjacent plates from the same field group .
the distributions become much more uniform than those displayed in the top panels .
detailed fractions of survey footprint covered by different plate densities as illustrated in the second row of fig.[sampling_density ] are listed in table[sampling_density_table ] .
about three quarters of the survey area are sampled by five plates ( 2 b , 2 m and 1 f ) for a given patch of the sky out of the galactic plane . for the highly sampled galactic plane ( @xmath84 ) , essentially all the region , which accounts for about 13.5 per cent of the whole survey footprint
is sampled by ten plates ( 4 b , 4 m and 2 f ) for a given patch of the sky .
about 10 per cent of the survey area is covered only by plates of fields of either g1 or g2 .
only 0.4 per cent of the survey footprint is left uncovered .
the bottom panels of fig.[sampling_density ] display the plate density distributions for overlapping areas of adjacent fields for time - domain spectroscopy .
about one quarter of all targets will be observed twice and about 2 per cent will be observed three times .
note that in the above calculations , the effects of the four holes in the lamost focal plane have not been taken into account .
lccll plate & area fraction & area & plate & plate + density & ( % ) & ( deg@xmath0 ) & type & group + 0 & 0.4 & 134 & & + 2 & 5.1 & 175 & b+m & g1 + 3 & 5.8 & 200 & b+m+f & g2 + 5 & 74.3 & 2554 & 2b+2m+f & g1 , g2 + 7 & 0.45 & 15 & 3b+3m+f & g1 , g3 + 8 & 0.43 & 15 & 3b+3m+2f & g2 , g3 + 10&13.5 & 464 & 4b+4m+2f & g1 , g2 , g3 + once the target priorities have been assigned , the plate centres and categories as well as field groups fixed , the next ( fourth ) step in the lss - gac survey design and target selection is to run the sss to allocate fibres to targets of individual plates .
usually the sss requires three input catalogues : targets , standard stars for flux calibration and positions of blank sky for sky subtraction .
given the unknown extinctions , which are likely to be significant , to individual disc stars targeted by the lss - gac , it is not feasible , as in the case of sdss which primarily targets sources at high galactic latitudes , to pre - select f - turnoff stars as flux calibration standards based on the measured photometric colours . on the other hand , as described earlier , our target selection algorithm is specifically designed to target stars of all colours ( thus spectral types ) , and as shall be shown in the next subsection , our targets include a large fraction of f stars . consequently , we have used a modified version of the sss for which no standard stars need to be supplied for its operation . a separate pipeline that differs from the lamost default 2d
pipeline has also been been developed at peking university to flux - calibrate plates collected for the lss - gac using an iterative approach ( xiang et al .
catalogs of positions of blank sky contain a sub - sample of positions of empty sky that have no 2mass , gsc2.3 ( lasker et al .
2008 ) and usnob1.0 ( monet et al . 2003 ) sources in a square box of size 36arcsec centred on the position .
this sub - sample of blank sky positions is constructed such that the positions are more or less uniformly distributed spatially and has a number density of about 100 per deg@xmath0 .
typically 320 sky fibres are allocated per plate for sky background measurements .
this gives 20 sky fibres per spectrograph , or about 16 sky fibres per deg@xmath0 , which is about 3.5 times denser than the sdss .
in addition to the input catalogues , to run the sss , an observing date has to be specified and the date when the field passes the prime meridian at midnight is chosen . in running the sss ,
four guiding stars are first selected from the gsc2.3 catalogues to allow for corrections of fibre positioning errors produced by the expansion / contraction ( due to changes of the focal length ) , shift and rotation of the focal plane .
if four guiding stars can not be located , the field centre is shifted to the nearest qualified bright star .
the sss assigns lss - gac targets according to their priorities .
small numbers of targets , such as emission line objects from the int / wfc photometric h@xmath21 survey ( iphas ; drew et al .
2005 ) of the northern galactic plane ( witham et al .
2008 ) , candidates of young stellar objects and white dwarfs can be added_ad hoc _ to the input target catalogues , and assigned the highest priorities .
this ensures that adding those special targets will not jeopardize the overall spatial uniformity of the lss - gac targets .
ideally , a lamost plate should make full use of all the 4,000 fibres , with 320 of them targeting blank sky and the remaining targeting science objects . in reality ,
only 3700 3930 fibres per plate are assigned a target ( including blank sky ) for most plates designed for the lss - gac main survey .
the unassigned fibres are either dead or have no targets left in their reachable areas .
finally the sss has to be rerun in order to update the fibre allocation prior to the actual observation taking place ( usually within a week ) , based on the output target list ( including blank sky ) generating by the first run of the sss .
this is because the difference between the date specified when the sss is first run and the actual observing date can lead to loss or change of guiding stars originally selected . in most cases ,
no loss or change of guiding stars occurs , and most targets assigned a fibre in the first round of runs of sss get re - assigned a fibre
. only a few , if not nil targets , fail to get reallocation of a fibre .
the occasional loss of a small number of targets is caused by the small differences in aberration corrections for fibre positions at different dates .
this ensures that all lss - gac plates can be designed , targets assigned and checked well in advance of actual observation . in some rare cases , one or two of the selected guiding stars get changed .
the change of one guiding star can lead to a loss of 200 300 targets originally assigned a fibre .
if two guiding stars are changed , about 500 600 original targets can get lost .
this problem has however been fully solved since november , 2013 by supplying the sss when it is rerun lists of guiding stars that contain only those selected in the first run of sss , instead of the whole gsc2.3 catalogues .
it is at this very last stage that add - on targets are assigned fibres if possible .
add - on targets usually have the highest priorities .
they are assigned to fibres in a way that is very similar to blank sky ( or standard stars in the case of high galactic latitude plates targeted by the spheroid survey for example ) .
no more than 3 add - on targets per clip of 25 fibres and no more than 20 per spectrograph are assigned to ensure that the overall homogeneity and completeness of the main sample are not affected .
lcccccccc & & & & + type & num . & completeness & num . & completeness & num . & completeness & num . & completeness + k-(sub)giant & 2,005 & 0.55 & 174 & 0.56 & 14 & 0.67 & 4,285 & 0.47 + g-(sub)giant & 677 & 0.37 & 129 & 0.24 & 62 & 0.48 & 2,252 & 0.27 + a - dwarf & 1,573 & 0.32 & 12 & 0.75 & 0 & - & 2,989 & 0.39 + f - dwarf & 1,644 & 0.07 & 1,180 & 0.20 & 778 & 0.57 & 13,393 & 0.14 + g - dwarf & 1,277 & 0.07 & 994 & 0.07 & 1,081 & 0.38 & 12,451 & 0.08 + k - dwarf & 1,346 & 0.15 & 1,557 & 0.14 & 1,692 & 0.46 & 18,251 & 0.19 +
m - dwarf & 917 & 0.60 & 810 & 0.53 & 740 & 0.63 & 9,328 & 0.55 + lccccc type & @xmath88(@xmath20 ) & @xmath88(@xmath6 ) & @xmath88(@xmath42 ) & dist .
( kpc ) & dist .
( kpc ) + & & & & ( @xmath89mag ) & ( @xmath90mag ) + a0v & 1.29 & 1.53 & 1.75 & 3.12 & 24.77 + f0v & 2.35 & 2.27 & 2.33 & 2.22 & 17.62 + g0v & 4.47 & 4.08 & 4.01 & 0.96 & 7.66 + k0v & 6.28 & 5.62 & 5.45 & 0.55 & 3.77 + m0v & 10.93 & 9.62 & 8.85 & 0.08 & 0.60 + k0iii & 1.28 & 0.41 & 0.15 & 5.23 & 41.50 + m0iii & @xmath910.22 & @xmath911.71 & @xmath912.59 & 13.9 & 110.15 + band magnitude for two sightlines of the lss - gac main survey toward ( l , b ) = ( @xmath92 ) ( left ) and ( l , b ) = ( @xmath93 ) ( right ) , simulated with the besanon model .
[ fraction ] , width=340 ] the above survey strategy and target selection algorithm are tested using mock catalogues generated from the besanon galactic model ( robin et al .
2003 ) . as an example , for a @xmath94 wide stripe of 56.7 deg@xmath0 centred at @xmath95 and stretching from @xmath71 = @xmath96 to + 30@xmath37 , the model yields 396,075 stars of @xmath97mag . amongst them , 0.018 , 2.3 and 2.1 per cent are respectively m , k and g subgiants / giants , and 1.9 , 24 , 39 , 24 and 4.3 per cent are respectively a , f , g , k and m dwarfs .
note that @xmath98 magnitudes delivered by the besanon model refer to the cfht / megacam filter system , and have been transformed to the sdss photometric system using the calibration of the snls grouppritchet / sn / calib / colourterms-2006jun19/index.html . ] for the @xmath99 and @xmath100 bands and that from the cfht website for the @xmath101 band , respectively .
photometric errors are also assigned to magnitudes yielded by the model according to their values to simulate the error - magnitude relations of the xstps - gac photometry . running this mock catalogue through the target selection procedure outlined above
, one finds that in total 64,029 of all stars , 99 , 47 and 27 per cent of all m , k and g subgiants / giants , 39 , 14 , 8 , 19 and 55 per cent of all a , f , g , k and m dwarfs , are selected and targeted by the lss - gac , respectively ( table[tabcompleteness ] ) .
table[tabcompleteness ] also shows that the completeness increases toward high galactic latitudes for all stellar types . at @xmath36 = 25@xmath37 30@xmath37 , about half stars of all types get targeted .
fig.[simulation ] shows the spatial distributions of various types of star targeted in the ( x , z ) plane of galactic radius and height .
the lss - gac targets range from m dwarfs to m giants , probing volumes of a wide range of depths .
the first three columns of table[tabvolume ] give respectively the absolute magnitudes in @xmath20 , @xmath6 and @xmath42 band of individual types of star , whereas the last 2 columns list the maximum distances probed by those types of star for a limiting magnitude of @xmath102 and 18.5mag , respectively . at @xmath90mag , m0v , k0v , g0v , f0v , a0v , k0iii and m0iii stars as far as 0.6 , 3.8 , 7.7 , 17.6 , 25 , 40 and 110 kpc are reached , respectively , assuming zero reddening .
the simulation confirms that our target selection algorithm is well designed to meet the scientific goals of lss - gac .
fig.[fraction ] shows the relative fractions of targets from the galactic thin disk , thick disk and halo for two sightlines of the lss - gac main survey , simulated with the besanon model . in the direction of galactic anti - center , over 90 per cent of the targets are from the galactic thin disk . in the direction of ( l , b ) = ( @xmath93 ) ,
as the targets get fainter , the fraction of thick disk stars increases from 17 to 35 per cent , whereas the fraction of halo stars increases from zero to 6 per cent .
the targets in the m31-m33 field include known and candidate pne , hii regions , supergiants and globular clusters of m31 and m33 , known and candidate background quasars , as well as foreground galactic stars from the xstps - gac photometric catalogues .
the stars are selected , assigned and observed in the same way as for the lss - gac main survey .
the selection criteria of various types of target and candidate of interest are presented elsewhere ( e.g. , yuan et al .
2010 for pne ; huo et al .
2010 , 2013 for quasars ) .
they are assigned higher priorities than ( foreground galactic ) stars .
the field central stars for the m31-m33 survey are selected similarly to the lss - gac main survey in the ( ra , dec ) plane .
two groups of field central stars are chosen to cover the m31-m33 area ( @xmath103 ra @xmath104 , @xmath105 dec @xmath106 ) independently .
each group contains 41 central stars .
their distributions are displayed in fig.[m31_central_stars ] . to make use of bright nights or nights of unfavourable conditions ( large seeing or low atmospheric transparency ) ,
very bright ( vb ) plates are designed to target stars brighter than @xmath89mag in all sky area accessible to the lamost ( @xmath18 dec @xmath19 ) . within the xstps - gac footprint
, all stars of @xmath107mag from the xstps - gac catalogues and stars of @xmath108mag from the 2mass catalogues are selected as potential targets . outside the xstps - gac footprint , all stars of @xmath109mag , or @xmath110mag , or @xmath111mag , or @xmath112mag , or @xmath113mag from the ppmxl catalogues ( roeser et al .
2010 ) and stars of @xmath108mag from the 2mass catalogues are selected as potential targets .
stars are targeted with equal priorities .
the field centres are selected similarly to the lss - gac main survey but in the ( ra , dec ) plane after dividing the sky area accessible to the lamost into three bins in dec ( @xmath9110@xmath37 30@xmath37 , 30@xmath37 50@xmath37 and 50@xmath37 60@xmath37 ) .
only one group of fields centred on a total of 966 bright stars are chosen .
for each of them , 1 to 9 plates are planned , depending on the number of stars available , yielding a total number of 2,594 vb plates .
typically 2,000 3,400 stars get assigned a fibre for a vb plate , depending on its location ( in particular its latitude ) .
about 10 per cent targets are duplicates targeted by adjacent plates . in the overlapping areas of adjacent plates ,
fibres are assigned to stars of a plate regardless whether they have been assigned a fibre or not in the adjacent plates .
no flux standard stars are allocated , as for the lss - gac main survey . due to relatively low densities of stars available compared to those of the main survey and the fact that vb plates are observed under unfavourable observing conditions ( bright moon , poor seeing etc . ) , more ( @xmath7 460 ) sky fibres are allocated for more efficient use of available fibres and better sky subtraction .
the vb survey is supplementary to the lss - gac main survey in terms of ( wider ) sky coverage , ( brighter ) magnitude limit and ( simpler ) target selection criteria , yielding an excellent sample of local stars to probe the solar neighbourhood .
being a meridian reflecting schmidt telescope , the lamost can only observe a given plate between 2h before and after the transit , thus putting a strong constraint on the range of ra of plates that can be targeted at a given time .
the focal plane has a fov of 5 in diameter , where the 4,000 auto - positioning fibres are nearly evenly distributed , feeding the light to 16 low - resolution spectrographs , 250 fibres each .
the fibres have a diameter of 3.3 arcsec projected on the sky .
slit marks of width 2/3 the fibre diameter , i.e. 2.2arcsec , are placed in front of the fibre clips to increase the spectral resolution to a resolving power of @xmath4 ( deng et al .
2012 ; liu et al .
2014 ) , similar to that of the sdss .
the spectra cover a wavelength range from 3700 9000 and are recorded in two arms , 3700
5900 in the blue and 5700 9000 in the red .
a shack - hartmann system is housed at the centre of the fov for the operation of the active optics .
it takes typically 30 minutes , depending on the pointing of the telescope and brightness of the field central star , to bring individual segments of the primary and corrector mirrors into focus .
four guiding ccd cameras are placed about halfway out from the field centre to monitor guide stars during the exposures for accurate fibre positioning .
the guide stars are also used to measure seeing during the observation . following a two - year commissioning phase ,
the lamost pilot surveys were initiated in october 2011 and completed in june 2012 .
the regular surveys were initiated in october 2012 . in each year , a sufficient number of plates of the lss - gac are planned in advance for observations . for the main survey , we adopt a strategy that we start with the stripe along dec = 29@xmath37 observed during the pilot surveys ( fig.[central_stars ] ) and then extend both ways to higher and lower declinations .
the main survey plates are observed in dark / grey nights .
typically 2 3 exposures are obtained for each plate , with typical integration time per exposure of 600 1200s , 1200 1800s , 1800
2400s for b , m and f plates , respectively , depending on the weather . some test nights reserved to monitor the telescope performance
are also used to observe the lss - gac plates . the seeing varies between 3 4 arcsec for most plates , with a typical value of about 3.5 arcsec . by june 2013 ,
142 b , 75 m and 20 f plates were targeted , yielding a total number of 727,478 spectra of 536,602 unique targets recorded in 237 plates .
about 73.4 , 20.2 , 4.5 , 1.3 , 0.4 , 0.1 per cent targets were observed 1 to 6 times , respectively .
the duplicate observations came from targets reobserved in the overlapping areas of adjacent plates of the same category and group of plates or from targets of observed plates that failed to meet the quality control and then got re - observed .
fig.[snr_gac ] shows histogram distributions of s / n(4650 ) and s / n(7450 ) for the observed targets , which roughly follow a power law distribution for s / n s higher than 10 .
a total of 448,224 spectra of 359,433 unique targets of s / n(4650 ) or s / n(7450 ) higher than 10 are obtained .
this includes 225,522 spectra of 189,042 unique targets of s / n(4650 ) higher than 10 , and 439,560 spectra of 352,775 unique targets of s / n(7450 ) higher than 10 .
their spatial distribution is shown in fig.[spatial_gac_bluered ] .
about two thirds of the spectra are from b plates , less than one third from m plates and about 5 per cent from f plates . more quality spectra were obtained in the first year regular surveys than in the pilot surveys due to more observing time available and better observing strategy and quality control .
the distributions in ( @xmath6 , @xmath75 ) and ( @xmath6 , @xmath76 ) hess diagrams of those stars are shown in fig.[gac_cmd_bluered ] .
for comparison , ( @xmath6 , @xmath75 ) and ( @xmath6 , @xmath76 ) hess diagrams of all targets observed in the main survey during the pilot and the first year regular surveys are shown in fig.[gac_cmd_observed ] .
note that a few targets brighter than @xmath89mag were observed during the early stage of the pilot surveys , october 2011 and november 2011 , when the bright and faint limiting magnitudes were set as @xmath114mag or @xmath115mag or @xmath116mag instead of @xmath117mag .
, width=340 ] plates of the m31-m33 survey were observed in a similar way as the main survey . by june 2013 ,
87 plates were targeted , yielding 265,745 spectra of 154,319 unique stars . about 64.0 , 17.8 , 9.4 , 3.9 , 2.4 , 1.3 ,
0.7 per cent stars were observed by 1 to 7 times , respectively .
fig.[snr_m31 ] shows histogram distributions of s / n(4650 ) and s / n(7450 ) for those stars .
again the distributions follow a power law for s / n s higher than 10 .
a total of 131,156 spectra of 84,084 unique targets of s / n(4650 ) or s / n(7450 ) higher than 10 are obtained , including 67,439 spectra of 46,501 unique targets having s / n(4650 ) higher than 10 , and 123,029 spectra of 79,733 unique targets having s / n(7450 ) higher than 10 .
their spatial distribution is shown in fig.[spatial_m31_bluered ] .
about 80 per cent of the spectra are from b plates .
the distributions in ( @xmath6 , @xmath75 ) and ( @xmath6 , @xmath76 ) hess diagrams are shown in fig.[m31_cmd_bluered ] .
for comparison , ( @xmath6 , @xmath75 ) and ( @xmath6 , @xmath76 ) hess diagrams of all stars observed in the m31-m33 fields during the pilot and the first year regular surveys are shown in fig.[m31_cmd_observed ] .
again a few stars brighter than @xmath89mag were observed during the early stage of the pilot surveys . ,
width=340 ] 10 or s / n(7450 ) @xmath9 10 in the lss - gac m31-m33 fields during the pilot and the first year regular surveys .
the positions of m31 and m33 are marked by red pluses . in total , 131,156 spectra have been collected . to avoid overcrowding , only one - in -
ten stars are shown .
[ spatial_m31_bluered],width=340 ] vb plates were observed in bright nights or nights of poor observing conditions .
the observations of vb plates began on january 5 , 2012 .
typically each plate was observed with 2 @xmath48 600s .
plates sharing the same central star are observed as many times as possible in one pointing in order to increase time - on - target efficiency of the telescope . by june 2013 , 259 plates were observed , yielding 791,530 spectra of 638,836 unique stars .
about 81.1 , 15.1 , 2.9 , 0.7 and 0.1 per cent of the targets were observed by 1 to 5 times , respectively .
fig.[snr_vb ] shows histogram distributions of s / n(4650 ) and s / n(7450 ) for the observed stars .
a total of 545,255 spectra of 452,758 unique targets having s / n(4650 ) or s / n(7450 ) higher than 10 are obtained , including 457,906 spectra of 385,672 unique targets having s / n(4650 ) higher than 10 , and 479,997 spectra of 398,381 unique targets having s / n(7450 ) higher than 10 .
their spatial distribution is shown in fig.[spatial_vb_bluered ] .
most vb stars are within the footprint of lss - gac main survey . the distribution of those stars in ( @xmath118 , @xmath119 ) hess diagram is shown in fig.[vb_cmd_bluered ] . for comparison ,
a ( @xmath118 , @xmath119 ) hess diagram of all stars observed in the vb plates during the pilot and the first year regular surveys is shown in fig.[vb_cmd_observed ] .
red bright giants stars are clearly visible . ,
width=340 ] versus @xmath118 plane for lss - gac vb targets observed during the pilot and the first year regular surveys that have spectral s / n(4650 ) @xmath9 10 or s / n(7450 ) @xmath9 10 .
the magnitudes are from the 2mass without corrections for the reddening .
the bin sizes are ( 0.05 , 0.05)mag .
duplicate targets are not accounted .
a colourbar is over - plotted by the side .
histograms of star counts in @xmath120 colour and in @xmath118 magnitude are also plotted .
[ vb_cmd_bluered],width=340 ] for all lss - gac vb targets observed during the pilot and the first year regular surveys , including those of s / n s below 10 .
[ vb_cmd_observed],width=340 ] the instrumental throughputs differ spectrograph by spectrograph and fibre by fibre .
fig.[snr_focal_gac ] shows distributions of spectra that have s / n(4650 ) @xmath9 10 or s / n(7450 ) @xmath9 10 in the focal plane for the lss - gac main survey targets observed during the pilot and the first year regular surveys .
large fibre - by - fibre variations are seen clearly .
spectrographs close to the centre have much higher success rates than those near the edge , for both the blue and red arms . clearly , when defining the selection effects of the samples , one needs to keep such variations in mind . within a given spectrograph ,
a small fraction of fibres have very low success rates , probably due to their relatively larger fibre positioning errors .
similar but not exactly the same behaviors are found for the m31-m33 fields ( fig.[snr_focal_m31 ] ) and for vb plates ( fig.[snr_focal_vb ] ) .
the raw data were reduced with the lamost 2d pipeline ( version 2.6 ) ( luo , zhang & zhao 2004 ; bai et al .
2014 ) , including steps of bias subtraction , cosmic - ray removal , 1d spectral extraction , flat - fielding , wavelength calibration , and sky subtraction .
as described earlier , the lamost spectra are recorded in two arms , 3700
5900 in the blue and 5700 9000 in the red .
the blue- and red - arm spectra are processed separately in the 2d pipeline and joined together after flux calibration .
no scaling or shifting is performed in cases where the blue- and red - arm spectra are not at the same flux level in the overlapping region , as it is unclear whether the misalignment is caused by poor flat - fielding or sky subtraction , or both .
flux calibration for the lss - gac observations is not straight - forward , as flux standard stars can not be selected based on ( optical ) photometric colours alone due to the unknown extinction towards individual stars in the galactic disk .
however , the target selection algorithm of lss - gac ensures that there are plenty of f - type stars targeted by each spectrograph . to flux - calibrated the lss - gac spectra , xiang et al .
( 2014a ) have developed an iterative algorithm . for a given spectrograph ,
the spectra are first flux - calibrated using the nominal spectral response curve ( src ) in order to derive the initial stellar atmospheric parameters with the lsp3 ( xiang et al .
2014b ) . then based on the initial estimates of stellar parameters ,
f - type stars targeted in that spectrograph are selected as flux standards to deduce an updated src after taking into account the interstellar reddening .
the reddening is estimated by comparing the observed and synthetic photometric colours .
the new src is then used to re - calibrate the spectra and revise the stellar parameters .
the above process is repeated until a convergence is achieved .
note that the flux calibration is performed for each spectrograph independently .
comparisons of stellar colours deduced by convolving the flux - calibrated lss - gac spectra with the transmission curves of photometric bands and actual measurements from photometric surveys , as well as comparisons of multi - epoch observations of duplicate targets suggest that an accuracy of about 10 per cent has been achieved for the whole wavelength ranges of lss - gac spectra .
note that in the current implementation of flux - calibration of xiang et al .
( 2014a ) , the telluric absorptions , including the prominent fraunhofer a band at 7590 and b band at 6867 have not been removed . when combining spectra from the individual exposures , the standard lamost 2d pipeline , following the sdss approach , first sorts all flux densities by wavelength and then performs a high - order spline fitting to obtain the final , combined flux densities .
while the approach is intended to preserve the spectral resolution as much as possible , it is prone to large , unpredictable errors .
this problem has now been fixed ( bai et al . 2014 ) . considering that lamost spectra
are grossly over - sampled , xiang et al .
( 2014a ) has adopted linear rebinning when combining spectra from the individual exposures with strict conservation of flux , and the results are therefore immune from the problem .
radial velocities and basic stellar atmospheric parameters ( , log @xmath20 and [ fe / h ] ) released in the lamost dr1 are determined by template matching with the elodie spectral library ( prugniel & soubiran 2001 ; prugniel et al .
2007 ) using the lasp ( wu et al . 2014 ) .
considering that the miles library ( snchez - blzquez et al .
2006 ; falcn - barroso et al .
2011 ) is particularly suitable for the determinations of stellar atmospheric parameters from lamost spectra , a separate pipeline , the lsp3 , has been developed at peking university . for a given target spectrum , the lsp3 determines the radial velocity by cross - correlating the continuum normalized spectrum with the normalized elodie template spectra and values of , log @xmath20 and [ fe / h ] by template matching with the miles library , using both a weighted mean algorithm and a simplex downhill algorithm .
results from the simplex downhill method are only used to cross - check and assign warning flags to parameters obtained from the weighted mean method .
the lsp3 is thoroughly tested and applied to the lss - gac spectra .
radial velocities and atmospheric parameters yielded by the lsp3 serve as the core data of the add - on catalogues presented in the current work . a full description of the lsp3 is presented in a companion paper by xiang et al .
( 2014b ) . the lsp3 attempts to determine radial velocities and atmospheric parameters for all spectra of a s / n(4650 ) @xmath9 2.76 .
stars of all spectral types , from the cool m to the hot o , are dealt with .
however , parameters derived for the hot oba - type and very cool m - type stars should be treated with caution , mainly due to the limitation of parameter space coverage of the available template spectra .
a variety of techniques are used to estimate the errors of parameters delivered by the lsp3 , including both the systematic and random errors , as a function of the s / n , , log @xmath20 and [ fe / h ] . for fgk stars and a s / n(4650 ) = 10 , it is estimated that the lsp3 has achieved an accuracy of 5kms@xmath121 , 150k , 0.25dex and 0.15dex for @xmath25 , @xmath122 , @xmath123 and [ fe / h ] determinations , respectively .
a systematic offset of about @xmath1243.5 km s@xmath121 is found for the lsp3 velocity measurements by comparing with external databases . a constant of + 3.5 km s@xmath121 has been added to all radial velocities yielded by the lsp3 .
systematic offsets in ranging from 100 300k are also found for stars hotter than 7000k or cooler than 3700k by comparing the lsp3 estimates with those predicted by photometric colours .
those offsets have been corrected for using a third - order polynomial .
the lsp3 parameters for @xmath25 and refer to those corrected values hereafter unless otherwise specified .
stars whose parameters may have suffered from significant boundary effects are marked by flags assigned by the lsp3 .
fig.[spectra1 ] displays six example lamost spectra of main sequence stars that range from early the b to the late m. examples of a metal - poor star , a rc star and a red giant are given in the top three panels of fig.[spectra2 ] , respectively . for each star ,
stellar parameters delivered by the lsp3 , s / n(4650 ) and the unique lamost spectral i d are labelled .
stellar parameters of those normal " dwarf and giant stars are well determined by the lsp3 .
since the lss - gac adopts a simple yet non - trivial target selection algorithm that targets stars of all colours , it also includes many targets of special interests such as white dwarfs ( wds ) , emission line stars and even quasars .
inclusion of those targets significantly increases the discovery space of the lss - gac survey . as described above ,
18 spectra of da type wds , carbon stars and late - m / l stars retrieved from the sdss database are added to the miles spectral template library for the purpose of classifications of such objects in the lsp3 .
templates of targets of special spectral characteristics are far from complete in the current version of lsp3 . as a result , many sources targeted by the lss - gac remain to be properly treated .
the bottom three panels of fig.[spectra2 ] illustrate example lamost spectra of a cataclysmic variable star , a dc type wd and a white - dwarf - main - sequence binary , respectively . at the moment
, their stellar parameters can not be reliably determined yet .
the interstellar reddening is a key parameter that its accurate determination is vital for reliable derivation of basic stellar parameters , including effective temperature and distance .
the 2d sfd extinction map provides a robust estimate of the ( total ) line - of - sight galactic extinction at high latitude regions of low and moderate reddening , although there is some evidence that it has over - estimated the true values of @xmath16 by about 14 per cent ( schlafly & finkbeiner 2011 ; yuan , liu & xiang 2013 ) .
however , the sfd map has the following limitations : a ) the map gives the total amount of extinction , integrated along the line - of - sight to infinite , and consequently the value is an upper limit of the real one for a local disc star ; b ) the map fails at low galactic latitudes ( @xmath125 ) ; and c ) the map has a limited spatial resolution about 6 arcmin , whereas the extinction may vary at smaller scales . given that most of the lss - gac targets locate at low galactic latitudes and in the disk , the sfd map is of limited use .
determining extinction for individual stars becomes an urgent task for the lss - gac .
the sdss dr9 contains about 0.7 million low - resolution spectra of galactic stars ( ahn et al .
2012 ) . the on - going lamost galactic surveys ( deng et al . 2012
; liu et al . 2014 ) release over 1 million stellar spectra in its first data release ( bai et al .
2014 ) and will take over 5 million spectra when the surveys complete . with millions of stellar spectra available , observations of stars of essentially identical stellar atmospheric parameters in different environments can be easily paired and compared , enabling a number of studies with the standard pair technique ( e.g. , yuan et al .
2014 ) . using this technique of pairing a star with its twins that suffer from almost nil extinction but otherwise have almost identical atmospheric parameters , yuan , liu & xiang ( 2013 ) have determined the dust reddening in a number of photometric bands for thousands of galactic stars by combining photometric measurements from the far ultra - violet ( uv ) to the mid - infrared ( mid - ir ) as provided by the galex , sdss , 2mass and wise surveys .
they further derive the empirical , model - free reddening coefficients for those colours .
their approach has the advantage that the method is straight - forward , model - free and applicable to stars of almost all spectral types .
the same technique has been applied to stars targeted by the lss - gac to derive their multi - band reddening values .
the star pair method requires a control sample consisting of stars of nil or extremely low reddening in order to estimate the intrinsic colours of target stars . in order to estimate values of reddening in different colours ranging from the uv to the mid - ir , 5 control samples ( i.e. ,
r 1 5 ) are constructed .
r1 , r2 and r5 are used to determine reddening values of colours @xmath126 , @xmath127 and @xmath128 , respectively .
r3 is used to determine reddening values of colours @xmath75 , @xmath76 , @xmath129 , @xmath130 and @xmath131 , and r4 for those of colours @xmath132 , @xmath133 and @xmath134 .
the criteria of selecting stars of the control samples are listed in table[control ] .
the distributions of control sample stars in galactic coordinates , in the log @xmath20 and [ fe / h ] planes are shown in fig.[ref ] .
note that within the footprint of lss - gac , stars in the southern galactic hemisphere overall suffer from higher extinction than those in the north ( chen et al .
2014 ) . as a consequence ,
most control sample stars are from the northern hemisphere .
all stars in the control samples are reddening corrected using the sfd extinction map and the reddening coefficients derived by yuan , liu & xiang ( 2013 ) .
the reddening coefficients for colours @xmath134 and @xmath128 are not available in yuan , liu & xiang ( 2013 ) , which are adopted to be zero here when dereddening @xmath134 and @xmath128 colours . because the reddening values of the control samples are small , the errors caused by reddening corrections are ignorable .
lccl sample & colours & num . of stars & criteria + r1 & @xmath135 & 3,106 & @xmath16 @xmath39 0.1mag , s / n@xmath136 @xmath9 20 , flags[1]@xmath137 @xmath39 1 , @xmath138 @xmath39 20mag , err(@xmath139 ) @xmath39 0.1mag , + & & & @xmath140(galex - xuyi@xmath141 ) @xmath39 2 , @xmath140(2mass - xuyi ) @xmath39 1 , q@xmath142(2mass ) = ` aaa ' + & & & + r2 & @xmath127 & 10,330 & @xmath16 @xmath39 0.05mag , s / n@xmath136 @xmath9 20 , flags[1]@xmath137 @xmath39 1 , err(@xmath139 ) @xmath39 0.1mag , + & & & err(@xmath20 ) @xmath39 0.02mag , @xmath143mag , @xmath140(galex - xuyi ) @xmath39 2 + & & & + r3 & @xmath75 , @xmath76 , @xmath129 & 27,146 & @xmath16 @xmath39 0.05mag , s / n@xmath136 @xmath9 20 , flags[1]@xmath137 @xmath39 1 , err(@xmath99 ) @xmath39 0.02mag , + & @xmath130 , @xmath131 & & @xmath144mag , q@xmath142(2mass ) = ` aaa ' , @xmath140(2mass - xuyi ) @xmath39 1 + & & & + r4 & @xmath132 , @xmath133 & 16,985 & @xmath16 @xmath39 0.05mag , s / n@xmath136 @xmath9 20 , flags[1]@xmath137 @xmath39 1 , q@xmath142(2mass ) = ` aaa ' , + & @xmath134 & & ( @xmath145 ) @xmath9 8mag , err(@xmath146 ) @xmath39 0.03mag , err(@xmath147 ) @xmath39 0.2mag , + & & & ext_flag@xmath148(wise ) = ` 0 ' , cc_flags@xmath148(@xmath149 ) = ` 000 ' , var_flags@xmath148(@xmath145 ) @xmath39 6 + & & & ph_qual@xmath148(@xmath149 ) = ` a / b / c ' , @xmath140(wise - xuyi ) @xmath39 1 + & & & + r5 & @xmath128 & 452 & @xmath16 @xmath39 0.1mag , s / n@xmath136 @xmath9 20 , flags[1]@xmath137 @xmath39 1 , 4 @xmath39 @xmath147 @xmath39 10mag , + & & & err(@xmath147 ) @xmath39 0.2mag , err(@xmath150 ) @xmath39 0.4mag , @xmath150 @xmath397.7mag , @xmath128 @xmath39 0.3mag , + & & & ext_flag@xmath148(wise ) = ` 0 ' , cc_flags@xmath148 = ` 0000 ' , var_flags@xmath148(@xmath151 ) @xmath39 6 , + & & & ph_qual@xmath148(@xmath152 )
= ` a / b / c ' , @xmath140(wise - xuyi ) @xmath39 1 + @xmath136 : : s / n(4650 ) per pixel .
@xmath137 : : second flag assigned to the final lsp3 parameters . see table 2 of xiang et al .
( 2014b ) .
@xmath153 : : angular distance between the source positions as measured by the two surveys in parentheses .
@xmath141 : : stands for the xstps - gac survey .
@xmath142 : : quality flags of the 2mass photometry in @xmath118 , @xmath154 and @xmath155 bands .
@xmath148 : : flags from the wise photometric catalogues . for a given star targeted by the lss - gac ,
its control stars are selected from the control samples as those having values of , log @xmath20 and [ fe / h ] that differ from the target values by less than 5 per cent , 0.5 dex and 0.3 dex , respectively . for a given colour ,
if the number of selected control stars are no less than four , the colour excess is then calculated as the difference between the observed colour of the target and the intrinsic colour of a pseudo star of atmospheric parameters identical to those of the target .
the intrinsic colour of the pseudo star is derived from the observed colours of the selected control sample stars assuming that the colour varies linearly with , log @xmath20 and [ fe / h ] , which is likely to be valid given the narrow ranges of values of , log @xmath20 and [ fe / h ] being considered .
if the number of selected control stars is less than four , the colour excess of the target is not calculated .
to estimate the errors of reddening values derived from the star pair method , we select a sample of stars from the lss - gac main survey with the following criteria : a ) s / n(4650 ) @xmath9 10 ; 2 ) @xmath156 15 ; 3 ) the @xmath157 values from the sfd map , @xmath158 mag ; and 4 ) the stars are well detected in @xmath20 , @xmath6 and @xmath42 bands by the xstps - gac and in @xmath118 , @xmath154 and @xmath155 bands by the 2mass . in total , about 90,000 stars are selected .
we then convert the colour excess @xmath159 of colour @xmath160 derived from the pair method for band @xmath161 into @xmath16 using the empirical reddening coefficients of yuan , liu & xiang ( 2013 ) and compare the result with @xmath162 .
the results are displayed in fig.[check_with_sfd ] , with the mean difference and standard deviation labelled at the top of each panel .
the mean differences are small except for colour @xmath119 , suggesting that the empirical reddening coefficients derived by yuan , liu & xiang ( 2013 ) are applicable to the current sample .
for @xmath119 , the coefficient deduced by yuan , liu & xiang ( 2013 ) may have been underestimated for the current sample of stars , due to , say the spatial variations of the extinction law .
the standard deviations for colours @xmath75 , @xmath76 , @xmath129 , @xmath130 and @xmath131 are 0.044 , 0.054 , 0.050 , 0.161 and 0.259 mag , respectively .
note that the deviations include contributions from the photometric errors of the xstps - gac and 2mass measurements , the random errors of the lsp3 stellar parameters and possible spatial variations of the extinction law .
the dispersions have only a weak dependence on s
/ n for s / n(4650 ) @xmath9 10 , indicating that the random errors of lsp3 parameters are likely to be small even at s / n(4650 ) = 10 .
the large values of dispersion for the @xmath130 and @xmath131 colours are simply because the two colours are not as sensitive to extinction as the @xmath75 , @xmath76 and @xmath129 colours of shorter wavelengths .
to reduce errors of reddening derived from the star pair method , results from different colours are combined . if a target has been detected in @xmath20 , @xmath6 and @xmath42 bands by the xstps - gac and in @xmath118 , @xmath154 and @xmath155 bands by the 2mass , the adopted @xmath16 is the weighted mean of values yielded by colours @xmath75 , @xmath76 and @xmath129 weighted by the corresponding reddening coefficients .
values yielded by colours @xmath130 and @xmath131 are excluded given their low sensitivity to extinction . if a star is only detected in the optical , then the weighted mean of values yielded by colours @xmath75 and @xmath76 is adopted .
similarly , if a star is only detected in the near ir by the 2mass , the weighted mean of values given by colours @xmath130 and @xmath131 is used . here
all reddening coefficients are again taken from yuan , liu & xiang ( 2013 ) .
comparisons between the mean values of extinction derived respectively from the optical plus the near - ir data , from the optical or from the near - ir data only , and those given by the sfd map are also displayed in fig.[check_with_sfd ] .
the dispersions are 0.035 , 0.036 and 0.116 mag , respectively . clearly , optical photometry is crucial for accurate determinations of the reddening .
the accuracy of extinction derived from the star pair method is further examined by applying the method to member stars of open cluster m67 targeted by the lamost .
the list of member stars is from xiang et al .
( 2014b ) .
the upper left panel of fig.[m67 ] shows that a mean value of @xmath16 of 0.013 mag is obtained , is consistent within the errors with the value of 0.041mag given by the sfd map and the value from the literature , 0.04 @xmath163 0.02mag ( pancino et al .
the small dispersion of values yielded from individual stars , 0.017mag , suggests that the star pair method is capable of determining reddening to a high accuracy .
a weak dependence of the extinction derived on is observed ( cf .
the lower left panel of fig.[m67 ] ) .
this is probably caused by the low s / n s of spectra of those cool late type stars ( cf .
the upper right panel of fig.[m67 ] ) .
reddening can also be determined by comparing the observed and synthetic colours from stellar model atmospheres . to calculate the model predicted colours ,
a grid of synthetic colours is constructed by convolving the synthetic spectra ( castelli & kurucz 2004 ) with the transmission curves of the sdss and 2mass photometric systems . for a star of given atmospheric parameters , ,
log @xmath20 and [ fe / h ] , the synthetic colours are derived by linearly interpolating the grid
. values of extinction derived from individual colours are then averaged as in the case of star pair method .
given that the stellar colour loci for main sequence stars are well described by a single parameter , i.e. , the effective temperature , reddening can also be determined from multi - band photometry alone by fitting the stellar sed from the optical to the near - ir ( e.g. , berry et al . 2012 ) .
using this technique and combining the photometry of xstps - gac in the optical , 2mass and wise in the near to mid - ir , chen et al .
( 2014 ) has derived values of extinction and distances for about 15 million stars surveyed by the xstps - gac , including most stars targeted by the lss - gac .
errors of reddening derived from the observed and synthetic multi - band colours are estimated by comparing the results with those from the sfd map for the same selected sample of lss - gac targets used above to estimate the errors of extinction derived from the star pair method .
the results are shown in the top panels of fig.[compare_ebv ] .
the @xmath16 values derived from the star pair method , @xmath164 , are also compared with those derived by comparing the observed and synthetic colours , @xmath165 , and those derived by fitting multi - band photometry to the empirical stellar loci ( chen et al .
2014 ) , @xmath166 , in the middle and bottom panels of fig.[compare_ebv ] .
values of @xmath167 and @xmath165 have comparable errors , and are about two times smaller than those of @xmath166 . a tight correlation between @xmath167 and @xmath165
is expected given that both methods use the same set of stellar parameters yielded by the lsp3 .
the small systematic discrepancy between @xmath167 and @xmath165 suggests that there are some systematics between the model predicted colours and the observed ones .
values of @xmath166 are systematically smaller than those of @xmath167 by about 0.06 mag .
the differences are possibly caused by the spatial variations of the extinction law as well as by the uncertainties of the photometric method which tends to underestimate the true values of reddening ( chen et al .
2014 ) .
( left ) and @xmath166 ( right ) with those from the sfd map for a selected sample of stars of high galactic latitudes ( @xmath156 15 ) and low extinction [ @xmath158mag ] .
middle : comparisons between values of @xmath167 and @xmath165 ( left ) and @xmath166 ( right ) .
bottom : differences @xmath167 @xmath91 @xmath165 ( left ) and @xmath167 @xmath91 @xmath166 ( right ) plotted as a function of . to avoid overcrowding , only 10 per cent of targets , randomly selected ,
are shown in the middle and bottom panels .
, width=340 ]
studying the six - dimensional phase - space distribution of a large sample of galactic stars plays an essential role in understanding the true nature of the galaxy and galaxies in general .
it requires reliable estimates of distance to individual sample stars .
distance determinations are coupled with reddening corrections .
corresponding to the three methods of reddening determinations described above , we have implemented three sets of distance estimates for the lss - gac sample stars . for stars with spectroscopically measured atmospheric parameters , their distances
are commonly estimated with the aid of stellar evolutionary models .
the evolution and properties , including luminosity , of a ( single ) star are fully determined by its initial mass , chemical composition and age . for
a given set of initial mass and chemical composition , the stellar evolutionary models of structure and atmosphere yield , log @xmath20 , [ fe / h ] and absolute magnitudes of specified photometric bands as a function of age .
the absolute magnitudes have a tight relation with , log @xmath20 and [ fe / h ] .
therefore , it is possible to estimate absolute magnitudes , thus distances , from the values of , log @xmath20 and [ fe / h ] yielded by spectroscopy when combined with reddening corrected photometry of apparent magnitudes .
note that [ fe / h ] is often used as a proxy of the chemical composition , the effects of possible variations in individual elemental abundances , such as the @xmath21-element enhancement have been neglected . for stars in the miles library ,
the absolute magnitudes can be directly calculated using the hipparcos parallaxes ( anderson & francis 2012 ) .
together with the atmospheric parameters , the stars thus provide an excellent sample to establish empirical relations between the absolute magnitudes and , log @xmath20 and [ fe / h ] , enabling us to determine distances from , log @xmath20 and [ fe / h ] directly .
compared to the theoretical relations from the stellar evolutionary models , the approach has the advantage of being model - independent and straight - forward to use .
considering that the atmospheric parameters of lss - gac targets are derived by template matching with the miles library , it is quite natural to apply the empirical relations of absolute magnitudes as a function of stellar atmospheric parameters as derived from the miles stars to the lss - gac targets .
we use polynomials to construct the empirical relations between the absolute magnitudes in @xmath168 and @xmath155 bands and , log @xmath20 and [ fe / h ] for the miles stars .
the apparent magnitudes in @xmath20 and @xmath6 bands of the stars are transformed from the observed , dereddened @xmath169 and @xmath170 magnitudes using relations , @xmath171mag and @xmath172mag ( jester et al . 2005 ) .
errors of the derived absolute magnitudes are estimated by combining the photometric and distance ( parallax ) errors . when fitting the data , we only include those miles stars whose estimated errors of absolute magnitudes are smaller than 0.4mag , and weight the remaining data points by the inverse square of the absolute magnitude errors .
a @xmath173 clipping is used to exclude outliers .
the miles library includes stars of a wide range of spectral types and luminosity classes , from the o to the m and from dwarfs to giants .
fitting all the stars with a single polynomial will require a very high order polynomial .
the result is also likely to be unstable and liable to large errors near the periphery of parameter space coverage . to solve the problem
, we divide the stars into four groups of spectral type and luminosity class 1 ) the hot oba stars ; 2 ) fgk dwarfs ; 3 ) km dwarfs and 4 ) gkm giants .
each group is fitted independently .
the groups are defined in atmospheric parameter space with some overlapping .
the definitions are : oba stars : : : @xmath174 k ; fgk dwarfs : : : @xmath175 k , log @xmath20 @xmath176 3.2 ( cms@xmath177 ) ; km dwarfs : : : @xmath178 k , log @xmath20 @xmath176 3.2 ( cms@xmath177 ) ; gkm giants : : : @xmath179 k , log @xmath20 @xmath39 3.5 ( cms@xmath177 ) . a three - order polynomial of the following form
is then used to fit the absolute magnitude of a given band : @xmath180 } ) = \nonumber\\ & a_0 + a_1 \times t_{\rm eff } + a_2 \times \log g + a_3 \times { \rm [ fe / h ] } + \nonumber\\ & a_4 \times ( t_{\rm eff})^2 + a_5 \times ( \log g)^2 + a_6 \times ( [ fe / h])^2 + \nonumber\\ & a_7 \times t_{\rm eff } \times \log g + a_8 \times t_{\rm eff } \times { \rm [ fe / h ] } + \nonumber\\ & a_9 \times \log
g \times { \rm [ fe / h ] } + \nonumber\\ & a_{10 } \times ( t_{\rm eff})^3 + a_{11 } \times ( \log g)^3 + a_{12 } \times ( { \rm [ fe / h]})^3 + \nonumber\\ & a_{13 } \times ( t_{\rm eff})^2 \times \log g + a_{14 } \times ( t_{\rm eff})^2 \times \log g + \nonumber\\ & a_{15 } \times ( \log g)^2 \times t_{\rm eff } + a_{16 } \times ( \log g)^2 \times { \rm [ fe / h ] } + \nonumber\\ & a_{17 } \times ( { \rm [ fe / h]})^2 \times t_{\rm eff } + a_{18 } \times ( { \rm [ fe / h]})^2 \times \log g + \nonumber\\ & a_{19 } \times t_{\rm eff } \times \log g \times { \rm [ fe / h]}. \label{eqn : dev}\end{aligned}\ ] ] the fit coefficients and residuals are listed in table[param ] . the residuals are also plotted in fig.[check_miles ] and have dispersion on average of about 0.35 , 0.32 , 0.22 and 0.44 mag , for the four groups of oba stars , fgk dwarfs , km dwarfs and gkm giants , respectively , corresponding to a fractional distance error of 17 , 16 , 11 and 22 per cent .
the residuals are partly contributed by errors of the absolute magnitudes as calculated from the apparent magnitudes and hipparcos parallaxes . as a robustness check of the relations derived above ,
we apply them to selected stars in the pastel database to derive the absolute magnitudes and compare the results with those calculated from the hipparcos distances .
the results are given in table[param ] and plotted in fig.[check_pastel ] .
the differences between the two sets of estimates have an average dispersion of about 0.34 , 0.25 and 0.25 mag for the groups of oba stars , fgk dwarfs and km dwarfs , respectively , comparable to the fit residuals of the miles stars .
the results suggest that the fits are robust and applicable to most types of stars , especially the dwarfs .
there are , however , some small systematic differences between the two sets of estimates of absolute magnitudes for the selected pastel stars , on the level of 0.09 , 0.12 and 0.20 mag for the groups of oba stars , fgk dwarfs and km dwarfs , respectively . for gkm giants ,
the differences have a dispersion of 0.70 mag , significantly larger than the fit residuals .
in addition , the average absolute magnitudes calculated from the fitted relations are systematically larger by 0.48 mag than the ones calculated from the measured parallaxes .
the results suggest that there are some systematic discrepancies between stellar parameters given by miles library and the pastel database , in particular of log @xmath20 for giants .
lrrrrrrrrrrr & @xmath181 & @xmath182 & @xmath183 & @xmath184 & @xmath185 & & @xmath181 & @xmath182 & @xmath183 & @xmath184 & @xmath185 + & + a@xmath186 & 1.90e+02 & 1.71e+02 & 1.65e+02 & 8.95e+01 & 1.98e+02 & & @xmath914.11e+01 & @xmath911.24e+01 & @xmath917.58e+01 & @xmath918.22e+01 & @xmath918.27e+01 + a@xmath187 & 7.34e@xmath9104 & 8.34e@xmath9104 & @xmath912.43e@xmath9103 & @xmath911.24e@xmath9102 & @xmath912.33e@xmath9103 & & 3.90e@xmath9104 & 3.12e@xmath9102 & 5.93e@xmath9103 & 5.98e@xmath9103 & 8.53e@xmath9103 + a@xmath188 & @xmath911.47e+02 & @xmath911.33e+02 & @xmath911.22e+02 & @xmath914.40e+01 & @xmath911.47e+02 & & 2.89e+01 & @xmath913.82e+01 & 4.41e+01 & 4.78e+01 & 4.48e+01 + a@xmath189 & 6.70e+00 & 6.83e+00 & 4.28e+00 & 4.94e+01 & @xmath911.15e+01 & & 2.32e+01 & 1.56e+01 & @xmath914.67e+01 & @xmath915.83e+01 & @xmath914.24e+01 + a@xmath190 & @xmath915.80e@xmath9108 & @xmath911.20e@xmath9107 & @xmath913.73e@xmath9108 & 4.38e@xmath9107 & @xmath917.41e@xmath9108 & & @xmath916.79e@xmath9107 & @xmath914.03e@xmath9106 & @xmath911.44e@xmath9106 & @xmath911.28e@xmath9106 & @xmath911.37e@xmath9106 + a@xmath191 & 3.82e+01 & 3.40e+01 & 2.96e+01 & 7.18e+00 & 3.56e+01 & & @xmath915.01e+00 & 1.42e+01 & @xmath919.23e+00 & @xmath919.76e+00 & @xmath918.38e+00 + a@xmath192 & @xmath913.49e+00 & @xmath914.06e+00 & @xmath916.82e+00 & @xmath912.06e+00 & @xmath916.30e+00 & & 1.60e+00 & 2.14e+00 & @xmath913.41e@xmath9101 & @xmath915.93e@xmath9101 & @xmath911.02e@xmath9101 + a@xmath193 & @xmath914.28e@xmath9104 & @xmath912.07e@xmath9105 & 1.43e@xmath9103 & 4.21e@xmath9103 & 1.57e@xmath9103 & & 1.62e@xmath9104 & @xmath914.32e@xmath9103 & 3.64e@xmath9104 & @xmath914.64e@xmath9105 & @xmath911.07e@xmath9103 + a@xmath194 & 3.36e@xmath9103 & 3.56e@xmath9103 & 3.80e@xmath9103 & 1.64e@xmath9103 & 4.63e@xmath9103 & & @xmath915.50e@xmath9103 & @xmath911.46e@xmath9103 & 4.94e@xmath9103 & 6.02e@xmath9103 & 3.51e@xmath9103 + a@xmath195 & @xmath911.14e+01 & @xmath911.21e+01 & @xmath911.15e+01 & @xmath912.96e+01 & @xmath915.61e+00 & & @xmath912.64e+00 & @xmath914.65e+00 & 1.59e+01 & 2.00e+01 & 1.59e+01 + a@xmath196 & @xmath913.96e@xmath9112 & @xmath912.46e@xmath9112 & @xmath911.26e@xmath9112 & @xmath913.74e@xmath9112 & @xmath911.02e@xmath9112 & & 7.62e@xmath9111 & 1.55e@xmath9110 & 1.00e@xmath9110 & 9.84e@xmath9111 & 9.80e@xmath9111 + a@xmath197 & @xmath913.12e+00 & @xmath912.72e+00 & @xmath912.26e+00 & @xmath913.35e@xmath9101 & @xmath912.73e+00 & & 2.12e@xmath9101 & @xmath911.27e+00 & 6.23e@xmath9101 & 6.12e@xmath9101 & 4.59e@xmath9101 + a@xmath198 & 3.58e@xmath9101 & 3.59e@xmath9101 & 3.38e@xmath9101 & 3.57e@xmath9101 & 3.27e@xmath9101 & & 2.91e@xmath9101 & 3.42e@xmath9101 & 3.78e@xmath9101 & 4.27e@xmath9101 & 4.34e@xmath9101 + a@xmath199 & 5.85e@xmath9108 & 5.82e@xmath9108 & 2.22e@xmath9108 & @xmath917.88e@xmath9108 & 2.88e@xmath9108 & & @xmath911.04e@xmath9107 & 3.28e@xmath9107 & @xmath915.66e@xmath9108 & @xmath918.94e@xmath9108 & @xmath916.48e@xmath9108 + a@xmath200 & @xmath912.21e@xmath9108 & @xmath912.73e@xmath9108 & @xmath911.20e@xmath9108 & @xmath913.75e@xmath9108 & @xmath912.82e@xmath9108 & & 1.31e@xmath9107 & @xmath915.58e@xmath9108 & @xmath912.07e@xmath9107 & @xmath912.41e@xmath9107 & @xmath911.67e@xmath9107 + a@xmath201 & @xmath911.05e@xmath9104 & @xmath911.59e@xmath9104 & @xmath912.58e@xmath9104 & @xmath913.72e@xmath9104 & @xmath912.90e@xmath9104 & & 1.30e@xmath9104 & 3.54e@xmath9105 & 3.88e@xmath9105 & 1.39e@xmath9104 & 2.34e@xmath9104 + a@xmath202 & 2.27e+00 & 2.41e+00 & 2.52e+00 & 4.05e+00 & 1.93e+00 & & @xmath913.71e@xmath9101 & 1.59e@xmath9101 & @xmath911.53e+00 & @xmath911.89e+00 & @xmath911.67e+00 + a@xmath203 & @xmath911.81e@xmath9104 & @xmath911.52e@xmath9104 & @xmath911.26e@xmath9104 & @xmath918.28e@xmath9106 & @xmath917.03e@xmath9105 & & 1.16e@xmath9104 & 9.20e@xmath9105 & 4.09e@xmath9104 & 4.49e@xmath9104 & 4.00e@xmath9104 + a@xmath204 & 1.44e+00 & 1.52e+00 & 2.20e+00 & 7.19e@xmath9101 & 1.92e+00 & & @xmath913.61e@xmath9101 & @xmath914.21e@xmath9101 & @xmath912.88e@xmath9101 & @xmath912.54e@xmath9101 & @xmath913.03e@xmath9101 + a@xmath205 & @xmath917.24e@xmath9104 & @xmath917.46e@xmath9104 & @xmath918.94e@xmath9104 & @xmath912.16e@xmath9104 & @xmath911.01e@xmath9103 & & 9.03e@xmath9104 & 4.74e@xmath9104 & @xmath916.06e@xmath9104 & @xmath917.74e@xmath9104 & @xmath913.85e@xmath9104 + @xmath206@xmath136 & @xmath910.01 & @xmath910.01 & 0.06 & @xmath910.06 & @xmath910.01 & & @xmath910.03 & @xmath910.03 & @xmath910.06 & @xmath910.06 & @xmath910.06 + @xmath207@xmath137 & 0.35 & 0.36 & 0.33 & 0.29 & 0.32 & & 0.32 & 0.30 & 0.31 & 0.34 & 0.32 + @xmath207@xmath208@xmath153 & 0.19 & 0.19 & 0.28 & 0.24 & 0.21 & & 0.13 & 0.13 & 0.24 & 0.22 & 0.18 + @xmath206@xmath141 & @xmath910.09 & @xmath910.09 & @xmath910.10 & @xmath910.08 & @xmath910.08 & & @xmath910.15 & @xmath910.13 & @xmath910.10 & @xmath910.10 & @xmath910.10 + @xmath207@xmath142 & 0.32 & 0.33 & 0.34 & 0.37 & 0.32 & & 0.26 & 0.25 & 0.24 & 0.23 & 0.23 + @xmath207@xmath208@xmath148 & 0.16 & 0.16 & 0.27 & 0.24 & 0.20 & & 0.17 & 0.17 & 0.17 & 0.17 & 0.15 + & + a@xmath186 & @xmath911.91e+01 & @xmath914.71e+01 & 1.28e+01 & @xmath919.41e+01 & @xmath912.68e+01 & & 3.69e+01 & 4.85e+00 & 5.28e+01 & 3.39e+01 & 2.79e+01 + a@xmath187 & 1.38e@xmath9101 & 1.33e@xmath9101 & @xmath911.58e@xmath9101 & 3.96e@xmath9102 & @xmath911.57e@xmath9102 & & 1.55e@xmath9102 & 4.02e@xmath9102 & @xmath912.56e@xmath9102 & @xmath911.19e@xmath9102 & @xmath918.33e@xmath9103 + a@xmath188 & @xmath917.73e+01 & @xmath915.82e+01 & 1.13e+02 & @xmath914.13e+01 & 1.22e+01 & & @xmath911.07e+02 & @xmath911.19e+02 & @xmath916.51e+01 & @xmath917.11e+01 & @xmath916.80e+01 + a@xmath189 & @xmath918.53e+02 & @xmath918.01e+02 & 3.03e+02 & 5.94e+02 & 2.99e+02 & & @xmath915.41e+01 & @xmath916.60e+01 & @xmath911.63e+02 & @xmath911.58e+02 & @xmath911.04e+02 + a@xmath190 & @xmath916.08e@xmath9105 & @xmath915.57e@xmath9105 & 1.93e@xmath9105 & @xmath914.64e@xmath9106 & @xmath915.09e@xmath9106 & & @xmath911.58e@xmath9105 & @xmath912.21e@xmath9105 & @xmath911.02e@xmath9109 & @xmath913.34e@xmath9106 & @xmath913.90e@xmath9106 + a@xmath191 & @xmath912.40e+01 & @xmath912.37e+01 & @xmath913.78e+01 & 2.05e+01 & @xmath911.01e+01 & & @xmath911.92e+01 & @xmath911.94e+01 & @xmath911.90e+01 & @xmath911.86e+01 & @xmath911.66e+01 + a@xmath192 & @xmath911.21e+01 & @xmath911.31e+01 & @xmath919.90e+00 & 3.33e+01 & @xmath911.33e+00 & & 1.71e+01 & 2.01e+01 & 3.21e+01 & 3.35e+01 & 3.07e+01 + a@xmath193 & 6.25e@xmath9102 & 5.49e@xmath9102 & 3.69e@xmath9102 & @xmath911.19e@xmath9103 & 2.09e@xmath9102 & & 6.39e@xmath9102 & 6.91e@xmath9102 & 4.43e@xmath9102 & 4.66e@xmath9102 & 4.38e@xmath9102 + a@xmath194 & 1.72e@xmath9101 & 1.63e@xmath9101 & @xmath914.87e@xmath9102 & @xmath911.07e@xmath9101 & @xmath914.86e@xmath9102 & & 2.27e@xmath9102 & 2.86e@xmath9102 & 9.49e@xmath9102 & 9.20e@xmath9102 & 5.99e@xmath9102 + a@xmath195 & 2.08e+02 & 1.93e+02 & @xmath918.40e+01 & @xmath911.49e+02 & @xmath918.12e+01 & & 5.95e+00 & 4.79e+00 & @xmath913.75e+01 & @xmath913.60e+01 & @xmath911.92e+01 + a@xmath196 & 2.44e@xmath9109 & 2.12e@xmath9109 & 9.84e@xmath9110 & 1.30e@xmath9109 & 1.16e@xmath9109 & & 2.31e@xmath9109 & 2.84e@xmath9109 & 6.85e@xmath9110 & 9.54e@xmath9110 & 9.62e@xmath9110 + a@xmath197 & 6.72e+00 & 6.31e+00 & 1.92e+00 & @xmath912.71e+00 & 8.13e@xmath9101 & & @xmath916.67e@xmath9101 & @xmath916.56e@xmath9101 & @xmath919.12e@xmath9101 & @xmath919.35e@xmath9101 & @xmath917.95e@xmath9101 + a@xmath198 & 2.12e+00 & 1.93e+00 & 4.33e+00 & 2.97e+00 & 4.30e+00 & & 2.10e@xmath9102 & @xmath913.71e@xmath9101 & @xmath911.66e@xmath9101 & @xmath912.76e@xmath9101 & @xmath911.22e@xmath9101 + a@xmath199 & 5.79e@xmath9106 & 5.61e@xmath9106 & @xmath917.09e@xmath9106 & @xmath912.85e@xmath9106 & @xmath912.40e@xmath9106 & & @xmath919.09e@xmath9106 & @xmath919.63e@xmath9106 & @xmath917.03e@xmath9106 & @xmath917.26e@xmath9106 & @xmath916.74e@xmath9106 + a@xmath200 & @xmath918.49e@xmath9106 & @xmath918.17e@xmath9106 & @xmath917.22e@xmath9107 & 2.30e@xmath9106 & @xmath915.14e@xmath9107 & & @xmath912.25e@xmath9106 & @xmath912.84e@xmath9106 & @xmath911.40e@xmath9105 & @xmath911.37e@xmath9105 & @xmath919.11e@xmath9106 + a@xmath201 & @xmath911.27e@xmath9102 & @xmath911.17e@xmath9102 & 2.25e@xmath9103 & 2.16e@xmath9103 & @xmath913.53e@xmath9104 & & 5.06e@xmath9103 & 5.06e@xmath9103 & 5.41e@xmath9103 & 5.37e@xmath9103 & 4.76e@xmath9103 + a@xmath202 & @xmath911.22e+01 & @xmath911.12e+01 & 3.06e+00 & 6.26e+00 & 3.00e+00 & & 1.00e+00 & 1.34e+00 & @xmath912.53e+00 & @xmath913.03e+00 & @xmath912.38e+00 + a@xmath203 & @xmath914.05e@xmath9104 & @xmath912.96e@xmath9104 & 1.62e@xmath9103 & 2.18e@xmath9104 & 8.66e@xmath9104 & & @xmath913.25e@xmath9103 & @xmath913.90e@xmath9103 & @xmath918.65e@xmath9103 & @xmath918.66e@xmath9103 & @xmath917.69e@xmath9103 + a@xmath204 & 3.53e+00 & 3.58e+00 & 1.03e+00 & @xmath917.23e+00 & 5.10e@xmath9102 & & @xmath912.01e@xmath9101 & @xmath913.67e@xmath9101 & 3.99e+00 & 3.42e+00 & 2.75e+00 + a@xmath205 & @xmath912.16e@xmath9102 & @xmath912.02e@xmath9102 & 1.21e@xmath9102 & 1.89e@xmath9102 & 1.16e@xmath9102 & & @xmath911.99e@xmath9103 & @xmath912.15e@xmath9103 & 1.17e@xmath9102 & 1.18e@xmath9102 & 7.50e@xmath9103 + @xmath206@xmath136 & @xmath910.08 & @xmath910.04 & 0.00 & 0.01 & 0.02 & & @xmath910.04 & @xmath910.02 & @xmath910.03 & @xmath910.03 & @xmath910.02 + @xmath207@xmath137 & 0.25 & 0.13 & 0.20 & 0.24 & 0.21 & & 0.42 & 0.40 & 0.47 & 0.46 & 0.44 + @xmath207@xmath208@xmath153 & 0.11 & 0.11 & 0.16 & 0.18 & 0.17 & & 0.12 & 0.12 & 0.32 & 0.27 & 0.30 + @xmath206@xmath141 & @xmath910.38 & @xmath910.28 & @xmath910.17 & @xmath910.14 & @xmath910.15 & & 0.43 & 0.50 & 0.46 & 0.49 & 0.47 + @xmath207@xmath142 & 0.29 & 0.25 & 0.23 & 0.19 & 0.16 & & 0.72 & 0.70 & 0.68 & 0.65 & 0.66 + @xmath207@xmath208@xmath148 & 0.16 & 0.16 & 0.13 & 0.15 & 0.13 & & 0.14 & 0.14 & 0.34 & 0.30 & 0.33 + @xmath136 : : average of fit residuals for the selected miles stars .
@xmath137 : : dispersion of fit residuals for the selected miles stars .
@xmath153 : : median error of the absolute magnitudes calculated from the photometry and hipparcos parallaxes for the selected miles stars . @xmath141 : : average difference of absolute magnitudes derived from the fitted empirical relations and those deduced from the hipparcos parallaxes for the selected pastel stars .
@xmath142 : : dispersion of differences of absolute magnitudes derived from the fitted empirical relations and those deduced from the hipparcos parallaxes for the selected pastel stars . @xmath148 : : median error of the absolute magnitudes calculated from the photometry and hipparcos parallaxes for the selected pastel stars .
once the empirical relations between the absolute magnitudes and , log @xmath20 and [ fe / h ] have been determined , the following formula is used to calculate distances of the lss - gac targets from the photometry of a given band @xmath209 ( @xmath210 or @xmath155 ) : @xmath211 } ) + 5\ } \times 0.2 } , \label{eqn : dist}\end{aligned}\ ] ] where @xmath212 is the absolute magnitude of photometric band @xmath209 , @xmath213 the observed apparent magnitude in band @xmath209 , @xmath167 the interstellar reddening derived using the star - pair method and @xmath214 the extinction coefficient from yuan , liu & xiang ( 2013 ) .
@xmath212 is calculated from , log @xmath20 and [ fe / h ] as derived from the lss - gac spectra with the lsp3 .
note that the values used here are those before corrected for the systematics as described in section4 . for stars of different groups ,
the appropriate relations between the absolute magnitudes and stellar atmospheric parameters are used . to avoid ambiguities ,
the defining boundaries between the groups of stars are adjusted slightly from those adopted when fitting the relations , and are : oba stars : : : @xmath215 k ; fgk dwarfs : : : @xmath216 k , log @xmath20 @xmath1763.5(cms@xmath177 ) ; km dwarfs : : : @xmath217k , log @xmath20 @xmath1763.5(cms@xmath177 ) ; gkm giants : : : @xmath218k , log @xmath20 @xmath393.5(cms@xmath177 ) .
distances derived from individual bands agree typically with each other within 2 3 per cent for fgk dwarfs , and 5 10 per cent for giants , as shown in fig.[compare_dist_1 ] .
however , the discrepancies become quite large for stars near the boundaries of the four groups of stars such as subgiants .
the final distances are adopted to be the mean values derived from all available bands . and @xmath6 band magnitudes ( top ) , those from @xmath118 and @xmath154 bands ( middle ) and those from @xmath20 and @xmath118 bands ( bottom ) as a function of ( left ) and log @xmath20 ( right ) . , width=340 ]
rcs constitute a significant fraction of giant stars in the galactic disc targeted by the lss - gac .
rcs have a very narrow luminosity range .
their absolute magnitudes depend very weakly on age and metallicity .
consequently , they have been widely used as standard candles to measure distances . in the current work ,
we have selected stars of 4,500 @xmath39 @xmath39 5,200k and 2.0 @xmath39 log @xmath20 @xmath39 3.0dex as rcs .
their distances are determined assuming a 2mass @xmath155-band absolute magnitude of @xmath219mag ( grocholski & sarajedini , 2002 ) and an sdss @xmath42-band absolute magnitude of [email protected] , derived from the xstps - gac data using a technique similar to that of grocholski & sarajedini ( 2002 ) .
note that the above simple cuts on and log @xmath20 used to select rcs may lead to some contamination from red giant stars .
the accuracy of distances derived from the empirical relations is examined using the lamost observations of member stars of open cluster m67 .
the upper middle panel of fig.[m67 ] shows that the method yields a mean distance of 880 pc for m67 , consistent with the value from the literature ( 860 @xmath163 45 pc ; pancino et al .
the distances for individual stars have a dispersion of 7 per cent , consistent with the dispersion of residuals of the fits to the miles stars , as well as with the results of applying the relations to the selected pastel stars as discussed above . as we have made use of the synthetic colours predicted by the stellar models to determine the extinction of lss - gac targets , we have also used the absolute magnitudes predicted by the theoretical stellar isochrones to derive the distances .
once the extinction of a given star has been determined , its absolute magnitude is estimated using the isochrones of dartmouth stellar evolution database ( dotter et al .
the isochrones are first linearly interpolated to a dense grid in [ fe / h ] and [ @xmath21/fe ] , with steps of 0.1 and 0.2dex , respectively . since the current implementation of lsp3 does not provide estimates of [ @xmath21/fe ]
yet , we simply assign values of [ @xmath21/fe ] to stars according to their [ fe / h ] values .
we assign [ @xmath21/fe ] values of 0 , 0.2 , 0.4dex for stars of [ fe / h ] @xmath220 , @xmath221 [ fe / h ] @xmath222 and [ fe / h ] @xmath223 , respectively to mimic the observed variations of [ @xmath21/fe ] as a function of [ fe / h ] for galactic stars ( lee et al . 2011 ) . for a star of given , log @xmath20 , [ fe / h ] and [ @xmath21/fe ] ,
its absolute magnitudes are approximated as those of the nearest model of the isochrones .
since the grid of isochrones is dense enough , the maximum offset in parameter due to the simplification is less than 50k in @xmath224 , 0.05dex in [ fe / h ] , 0.1dex in [ @xmath21/fe ] , and 0.05dex in log@xmath20 , much smaller than the estimated uncertainties of the corresponding parameter .
a comparison of distances estimated using the theoretical absolute magnitudes predicted by the stellar isochrones and using those predicted by the empirical relations is given in the upper panels of fig.[compare_dist ] .
for dwarf stars of 4,000 @xmath225 7,500k and log @xmath20 @xmath9 3.8dex , the two estimates agree well , although the distances estimated using absolute magnitudes predicted by the theoretical stellar isochrones are about 5 per cent systematically lower than those deduced from the empirical relations .
the differences depend on and log @xmath20 .
both the above two approaches to estimate distances rely on stellar atmospheric parameters yielded by spectroscopic observations , thus are limited to stars having high - quality spectra .
alternatively , distances can also be estimated using photometric measurements alone .
estimates of photometric distances require accurate measurements of photometric colours and reddening . compared to spectroscopic distances
, photometric distances can be constructed for very large samples of stars of different stellar populations . using the reddening values derived from sed fitting by combining the xstps - gac , 2mass and wise photometry , chen et al .
( 2014 ) have derived photometric distances of 15 million stars in the footprint of xstps - gac , using the colour - magnitude relation of ivezi et al ( 2008 ) for an assumed [ fe / h ] = @xmath910.2 dex . a comparison of photometric distances obtained by chen et al . and those deduced above using the empirical relations of stellar absolution magnitudes as a function of spectroscopically derived atmospheric parameters is shown in the lower panels of fig.[compare_dist ] . for dwarf stars of 4,000 @xmath225 5,400k and log @xmath20
@xmath9 4.4 dex , the two estimates agree well at a level of about 10 per cent .
however , for stars hotter than 5,400k , or of surface gravities lower than 4.4 dex , the agreement deteriorates , and the discrepancies increase with increasing @xmath122 or decreasing @xmath123 .
the discrepancies at low surface gravities are possibly caused by the facts that 1 ) the colour - magnitude relation of ivezi et al .
is not applicable to turn - off stars in the galactic thin disk ; and 2 ) the lsp3 may have underestimated log @xmath20 for some fg dwarfs .
for convenience , we also provide proper motions of the lss - gac stars collected from various sources .
we have cross - matched the lss - gac targets with the ucac4 ( zacharias et al .
2013 ) and ppmxl ( roeser et al . 2010 ) catalogues . a match radius of 3 arcsec is used .
the ucac4 is an all - sky , astrometric catalogue that contains 113,780,093 objects complete to @xmath226mag . among them over 105 million stars have proper motions , with typical random errors of 4 mas yr@xmath121 .
the ppmxl is compiled based on the usno - b1.0 , 2mass , and its previous version ppmx ( roeser et al . 2008 ) .
it contains over 900 million sources down to @xmath227mag , with proper motions for almost all of them .
the errors of proper motion range from 4 to over 10mas yr@xmath121 , depending on the observational history of the sources .
wu et al . (
2011 ) investigate the systematic and random errors of proper motions provided by the ppmxl using a sample of over ten thousand quasars , which , being distant and point - like objects , should have zero proper motions .
they find systematic errors on the level of @xmath1242.0 and @xmath1242.1 masyr@xmath121 in @xmath228 and @xmath229 , respectively .
the errors correlate with source positions but not with magnitudes or colours for objects of @xmath230mag and @xmath231mag .
carlin et al . (
2013 ) find similar results and fit the systematic errors as a function of source position using a two - dimensional polynomial .
we have adopted the fit of carlin et al .
( 2013 ) to correct for systematic errors of the proper motions of ppmxl .
the corrections range from @xmath912.7 to 0.7 mas yr@xmath121 in @xmath228 and @xmath913.9 to 1.1 masyr@xmath121 in @xmath229 for the lss - gac targets .
for those sources , the random errors vary from 4.0 to 8.0 mas yr@xmath121 , depending on the magnitudes .
the systematic errors of proper motions of the ucac4 catalogue are examined by huang et al .
( 2014 ) , using about 1,700 quasars from the 2nd release of the large quasar astrometric catalog ( lqac , souchay et al .
2012 ) that have listed proper motions in the ucac4 . as in the case of ppmxl , no obvious dependence of the errors on magnitude or colour
is found for objects of @xmath232mag and @xmath233mag .
similar to carlin et al . ( 2013 ) , huang et al .
fit the systematic errors as a function of source position using a two - dimensional polynomial .
their results are used to correct the proper motions of ucac4 . the corrections range between @xmath911.1 and 1.0 mas yr@xmath121 in @xmath228 and @xmath911.8 to @xmath911.1 mas yr@xmath121 in @xmath229 for the lss - gac targets .
essentially all targets in the lss - gac are selected from the xstps - gac ( xuyi hereafter ) survey , which provides accurate positions ( about 0.1arcsec ) at a typical epoch of 2010 ( zhang et al .
2014 ) . by combining the measurements of xstps - gac and those of 2mass observed at a typical epoch of 2000 , one can obtain independent measurements of the proper motions .
the xuyi-2mass proper motions are especially useful to reject proper motions of large errors from the ucac4/ppmxl catalogues and to search for high proper motion targets , such as nearby hyper - velocity stars .
to compare proper motions from the ucac4 , ppmxl and xuyi-2mass , we select lss - gac targets which have lamost spectral s / n(4650 ) higher than 10 and proper motions from all the three sources .
the results are shown in fig.[compare_pm ] .
compared to the ppmxl , proper motions from the ucac4 are systematically lower in @xmath228 by @xmath7 0.97masyr@xmath121 and slightly higher in @xmath229 by @xmath7 0.22masyr@xmath121 .
note that the systematic errors in the ppmxl and ucac4 proper motions have already been corrected for as described above .
compared to the values of ucac4 and ppmxl , the proper motions given by the xuyi-2mass measurements are systematically lower in @xmath228 by respectively 0.33 and 1.30 mas yr@xmath121 , and in @xmath229 by respectively 2.41 and 2.18 mas yr@xmath121 .
if one ignores the small systematic differences , the proper motions from the ucac4 and ppmxl catalogues agree very well , with typical differences of about 4.4 mas yr@xmath121 in both @xmath228 and @xmath229 .
the comparisons suggest that the random errors of the ucac4 and ppmxl proper motions for the selected lss - gac targets are typically 3.0 mas yr@xmath121 , assuming that they contribute equally to the observed differences .
the proper motions of xuyi-2mass have larger random errors given the relatively short time span between the 2mass and xstps - gac observational epochs , and are about 6.0 mas yr@xmath121 as inferred from the differences between the xuyi-2mass and ucac4/ppmxl proper motions . the above selected sample of lss - gac targets are divided into bins in colour @xmath75 and in magnitude @xmath6 , ranging from 0 2 and 12 17mag , respectively .
the bin size is 0.1 mag in both quantities . for bins that have more than 10 stars , the median differences of proper motions given by the ucac4 and ppmxl are calculated and the results are displayed in fig.[compare_pm_cmd ] .
the differences in @xmath228 are found to be independent of @xmath75 and @xmath6 . however , the differences in @xmath229 show a strong dependence on @xmath75 and @xmath6 , in the sense that the differences are larger for redder and fainter stars .
the result is inconsistent with the findings of analyses based on the proper motions of quasars .
one possible explanation is that most quasars discovered so far are from the high galactic latitude regions , the findings from those quasars may not be applicable to the low galactic latitude regions .
identifying more quasars at low galactic latitudes will help clarify the issue .
the lss - gac targets are also divided into bins of ra and dec in steps of 1 degree in order to examine possible spatial variations of the differences of proper motions given by the ucac4 and ppmxl .
the results are shown in fig.[compare_pm_radec ] .
no obvious patterns are found .
in addition , the discrepancies in @xmath228 and @xmath229 are found to be independent .
however , for some randomly distributed patches of sky of 2 5deg in size , the differences are more than twice the typical values .
colour and @xmath6 magnitude .
the colourbars are in units of mas yr@xmath121 . ,
width=340 ] .
, width=340 ] for each lss - gac target , given its celestial position , distance , radial velocity and proper motions , we calculate its orbit in the galactic potential of gardner & flynn ( 2010 ) . in the model of gardner & flynn ,
the galaxy consists of a disc ( miyamoto & nagai 1975 ) , a plummer bulge ( plummer 1911 ) with an inner core , and a spherical logarithmic dark matter halo .
the characteristic parameters can be found in their table 1 . in calculating the orbit ,
we take the local standard of rest ( lsr ) values from huang et al . ( 2014 ) , i.e. , ( u , v , w ) = ( 7.01 , 10.13 , 4.95 ) km s@xmath121 and the sun is located at ( x , y , z ) = ( @xmath918.0 , 0.0 , 0.0)kpc for a galactocentric cartesian coordinate system where the x - axis oriented along a line connecting the sun and the galactic centre and points towards the galactic center , the y - axis is in the direction of disc rotation and the z - axis points towards the north galactic pole . the eccentricity of the orbit is defined as : @xmath234 , where @xmath235 and @xmath236 are respectively the maximum and minimum galactic radii reached by the orbit . the guiding radius @xmath237 is also computed using the z - component of the angular momentum and the rotation curve . for low eccentricities ,
@xmath237 is very close to the mean of @xmath235 and @xmath236 .
we define three lss - gac samples : the main , the m31-m33 and the vb samples , corresponding to the lss - gac main , the lss - gac m31-m33 and the lss - gac vb surveys , respectively . in this section
, the properties of those samples are discussed below . by june 2013 , 142 b ,
75 m and 20 f plates of the lss - gac main survey have been observed , yielding a total of 727,478 spectra of 536,602 unique targets recorded in 237 plates . amongst them , 225,522 spectra of 189,042 unique targets have s / n(4650 ) higher than 10 , and they are defined the lss - gac main sample . the spatial distribution of the main sample is shown in fig.[spatial_gac_b ] .
most are from the b plates , with the remaining 12 per cent from the m plates and only a few hundred spectra from the f plates .
about 84.5 , 12.6 , 2.2 , 0.5 and 0.1 per cent targets were observed by 1 to 5 times , respectively . as shown in fig.[multi_epoch_spatial_gac_b ] , the duplicate observations come from targets in the overlapping areas of adjacent plates of the same category and group as well as from disqualified plates that get re - observed .
the distributions of the 189,042 unique targets in ( @xmath6 , @xmath75 ) and ( @xmath6 , @xmath76 ) hess diagrams are shown in fig.[gac_cmd_blue ] .
compared to the roughly flat but slightly rising distribution in @xmath6 magnitude of the observed sample of each plate category ( fig.[gac_cmd_observed ] ) , the distribution in @xmath6 of the main sample is roughly flat but decreases slowly between @xmath238mag , and then drops rapidly at fainter magnitude . compared to a nearly flat distribution in @xmath75 colour between 0.4 @xmath239 1.5mag for the observed sample ( fig.[gac_cmd_observed ] ) ,
the distribution of the main sample keeps roughly flat between @xmath240mag , but then drops at redder colours . compared to the observed sample
, the main sample is biased to blue and bright stars .
most stars in the main sample are brighter than @xmath241mag and bluer than @xmath75 = 1.2mag and @xmath76 = 0.5mag .
the m - dwarf sequence is almost invisible in fig.[gac_cmd_blue ] because m - dwarfs are red and faint , and thus their spectra do not have good snrs in the blue in general .
note the small number of targets brighter than @xmath89mag or fainter than @xmath90mag were observed in the early stage of the pilot surveys between october 2011
november 2011 , when the bright and faint limits were set 14 @xmath39 @xmath20 , @xmath6 or @xmath42 @xmath39 19mag instead of @xmath117mag .
fig.[gac_hr ] shows contour distributions of the main sample in the log @xmath20 , [ fe / h ] and @xmath25 [ fe / h ] planes .
the stellar parameters are from the lsp3 .
about 15 per cent of the sample are giants [ log @xmath20 @xmath39 3.5(cms@xmath177 ) ] , about two thirds of which are possibly rc stars . among the dwarfs [ log @xmath20 @xmath176 3.5(cms@xmath177 ) ] , 16.2 , 29.5 , 37.9 , 13.4 and 3.0 per cent of the stars fall in the range of @xmath176 7,000k , 6,000 7,000k , 5,000 6,000k , 4,000 5,000k and @xmath242 4,000k , respectively .
most of the giants and stars hotter than 7,500k in the main sample are from the galactic disc and intrinsically luminous , and their fractions increase at low galactic latitudes .
most stars are metal - rich , only 1.5 per cent of them have [ fe / h ] values lower than @xmath911.0 dex and 11.7 per cent lower than @xmath910.5 dex , suggesting that the sample consists mainly of thin disc stars , plus a small fraction of thick disc stars and only few halo stars .
this result is consistent with the small dispersions of radial velocities of the sample .
there are very few metal - poor ( [ fe / h ] @xmath242 @xmath910.5 dex ) a- and m - type main sequence stars in the sample .
one reason is that the miles library contains few such stars , and as a consequence , the lsp3 is unable to determine the atmospheric parameters for them . but more importantly , one does not expect many metal - poor main sequence a stars given their young ages , nor a lot of metal - poor mdwarfs given that they can only be observed from a very small local volume .
fig.[gac_ebv_dist ] shows histogram distributions of values of @xmath16 and distances for the main sample .
the values of @xmath16 are those given by the star pair method and the distances are deduced using the empirical relations between absolute magnitudes and stellar atmospheric parameters .
about 44.6 , 25.4 and 2.3 per cent stars have @xmath16 values higher than 0.3 , 0.5 and 1.0mag , respectively . because most targets in the current lss - gac main sample are brighter than @xmath243mag , the volume that they probe is smaller than designed .
the median distance of the sample is about 1.5 kpc . only 38.9 , 14.8 and 2.7 per cent sample stars have distances larger than 2.0 , 4.0 and 8.0 kpc , respectively .
most distant targets are giants . the spatial distributions of the sample in the galactic ( x , y ) , ( x , z ) and ( y , z ) planes are shown in fig.[gac_xyz ] , assuming that the sun is located at ( x , y , z ) = ( @xmath918.0 , 0.0 , 0.0)kpc .
most stars are from the disk , consistent with the kinematics and chemistry as shown in fig.[gac_hr ] .
fig.[eccentricity_gac ] shows histogram distributions of eccentricities for the main sample .
the values calculated using proper motions from the ppmxl and ucac4 catalogues agree with each other .
the distributions peak at @xmath244 and have median values of 0.15 .
note that a small fraction of targets have eccentricities close to unity and caused by large uncertainties in proper motion measurements or distance estimates .
( left ) and distances ( right ) for the lss - gac main sample .
the bin size is 0.01mag in the left panel and 0.1kpc in the right panel .
duplicate targets have been excluded .
[ gac_ebv_dist],width=340 ] , width=340 ] by june 2013 , 87 plates have been observed in the m31-m33 fields as parts of the lss - gac , yielding a total of 265,745 spectra of 154,319 unique stars . amongst them , 67,439 spectra of 46,501 unique targets have s / n(4650 ) higher than 10 and they are defined as the m31-m33 sample .
their spatial distribution is shown in fig.[spatial_m31_blue ] . about 71.0 , 18.9 , 6.5 , 2.4 , 0.8 , 0.3 per cent of them
are observed 1 6 times , respectively .
note that compared to the lss - gac main sample , the fraction of duplicate stars in the m31-m33 sample are about two times higher .
the distributions in ( @xmath6 , @xmath75 ) and ( @xmath6 , @xmath76 ) hess diagrams are shown in fig.[m31_cmd_blue ] .
as in the case of the main sample , compared to the targeted sample ( fig.[m31_cmd_observed ] ) , the current m31-m33 sample that meet the s / n requirement in blue , is biased to bluer and brighter stars . as in the case of the main sample , the m - dwarf sequence is hardly seen in fig.[m31_cmd_blue ] either .
as for the lss - gac main sample , the sample includes a few stars brighter than @xmath89mag or fainter than @xmath90mag observed during the early stage of the pilot surveys .
but for the m31-m33 sample .
[ spatial_m31_blue],width=340 ] the distributions of the m31-m33 sample in the log @xmath20 , [ fe / h ] and @xmath25 [ fe / h ] planes are shown in fig.[m31_hr ] . only 8.5 per of the stars are giants . among the dwarfs ,
1.4 , 18.2 , 56.0 , 20.8 and 3.6 per cent fall the temperature range of @xmath176 7,000k , 6,000 7,000k , 5,000 6,000k , 4,000 5,000k and @xmath242 4,000k , respectively . due to the relatively high galactic latitudes of the m31-m33 fields , compared to the lss - gac main sample , the m31-m33 sample contains a smaller fraction of giants and hot stars but a larger fraction of fgk dwarfs .
the sample is also metal - rich , with only 2.8 per cent of the stars having a [ fe / h ] value lower than @xmath911.0 dex and 14.6 per cent lower than @xmath910.5 dex .
this is consistent with the small dispersions of radial velocities of the sample .
note that due to the galactic rotation , most of the m31-m33 sample stars have a blue - shifted radial velocity .
fig.[m31_ebv_dist ] shows distributions of values of @xmath16 and distances for the m31-m33 sample .
the values adopted are as for the lss - gac main sample .
most of the stars have @xmath16 values lower than 0.3 mag .
the median value is 0.093 mag , consistent with the sfd extinction map .
the median distance of the m31-m33 sample is about 1.2 kpc , slightly smaller than the main sample . only 23.7 , 8.9 and 3.1 per cent targets have distances larger than 2.0 , 4.0 and 8.0 kpc , respectively . spatial distributions of the m31-m33 sample in the ( x , y ) , ( x , z ) and ( y , z ) planes are displayed in fig.[gac_xyz ] .
similar to the main sample , most stars in the m31-m33 sample are from the galactic disk .
fig.[eccentricity_m31 ] shows histogram distributions of eccentricities for the m31-m33 sample .
the distributions peak at @xmath245 and have median values of 0.16 . again , a small fraction of targets have false values close to unity .
but for the m31-m33 sample .
[ m31_ebv_dist],width=340 ] but for the m31-m33 sample .
[ eccentricity_m31],width=340 ] observations of vb plates began on january 5 , 2012 . by june 2013 , 259 vb plates
have been observed , yielding a total number of 791,530 spectra of 638,836 unique stars . amongst them , 457,906 spectra of 385,672 unique targets have s / n(4650 ) higher than 10 , and they consist the current vb sample .
their spatial distribution is shown in fig.[spatial_vb_blue ] .
most of them are in the gac direction .
about 84.5 , 12.8 , 2.1 , 0.4 and 0.1 per cent targets are observed by 1 to 5 times , respectively .
unlike those duplicate targets in the main and m31-m33 samples , most duplicate observations of the vb sample are taken in the same night . the distribution of the vb sample in ( @xmath118 , @xmath119 ) hess diagram is shown in fig.[vb_cmd_blue ] .
compared to the sample targeted by the vb survey ( fig.[vb_cmd_observed ] ) , the actual sample that meets the s / n requirement slightly biases to blue and bright stars .
the distributions in @xmath119 and @xmath118 reflect well what have been designed except that the numbers of stars are less by 40 per cent .
note the branch of red giants is clearly visible . for the vb sample .
[ vb_cmd_blue],width=340 ] fig.[vb_hr ] shows the distributions of the vb sample in the log @xmath20 , [ fe / h ] and @xmath25 [ fe / h ] planes .
given the brightness of the vb sample , a large 27.1 per cent of the sample are giants , twice that of the main sample .
about 70 per cent of the giants are possibly rc stars .
due to the same reason , the vb sample contains more hot dwarfs .
about 17.8 , 39.0 , 38.0 , 4.5 and 0.7 per cent dwarfs fall in the range of @xmath176 7,000k , 6,000 7,000k , 5,000 6,000k , 4,000 5,000k and @xmath242 4,000k , respectively .
the sample is also metal - rich .
about 1.3 per cent of the stars have [ fe / h ] values lower than @xmath911.0 dex and 6.7 per cent lower than @xmath910.5 dex .
fig.[vb_ebv_dist ] shows histogram distributions of @xmath16 and distances for the vb sample .
the median value of @xmath16 is 0.15 mag .
about 28.1 , 13.1 and 1.4 per cent stars have @xmath16 values higher than 0.3 , 0.5 and 1.0 mag , respectively .
the median distance of the vb sample is about 900 pc .
only 20.6 , 4.1 and 0.3 per cent targets have distances larger than 2.0 , 4.0 and 8.0 kpc , respectively .
spatial distributions of the vb sample in the ( x , y ) , ( x , z ) and ( y , z ) planes are displayed in fig.[vb_xyz ] .
most stars are from the galactic thin disc ( @xmath246 0.5 kpc ) , consistent with their kinematics and chemistry as shown in fig.[vb_hr ] .
fig.[eccentricity_vb ] shows histogram distributions of eccentricities for the vb sample .
the distributions peak at @xmath247 and have median values of 0.12 , smaller that those for the main sample .
again , a small fraction of targets show flase values close to unity .
but for the vb sample .
[ vb_ebv_dist],width=340 ] but for the vb sample .
[ eccentricity_vb],width=340 ] the first release of lss - gac value - added catalogues consist of three binary tables in fits format , i.e. , lss - gac_dr1_main.fits , lss - gac_dr1_m31.fits and lss - gac_dr1_vb.fits , corresponding respectively to the main , the m31-m33 and the vb samples described above .
the catalogues include basic information of the lamost observations , stellar atmospheric parameters and corresponding errors and flags yielded by the lsp3 , multi - band photometric data from the far - uv to the mid - ir collected from the galex , xstps - gac , 2mass and wise surveys , values of the interstellar reddening and distances determined with the methods described in sections 5 and 6 , proper motions collected from the ppmxl and ucac4 catalogues after correcting for the systematic errors , proper motions newly derived by comparing the xstps - gac and 2mass position measurements , as well as orbital parameters of sample stars .
the catalogues are accessible from http://162.105.156.249/site/lss-gac-dr1/ along with a descriptive readme.txt file .
table[value - added ] gives a detailed description of the content of catalogues .
an example entry from the catalogue of the lss - gac main sample is shown in table[value - added3 ] .
lll col . & name & description + 1 & spec_id & lamost unique spectral i d , in format of date - plateid - spectrographid - fibreid + 2 & date & date of observation + 3 & plate & lamost plate i d , not necessarily unique + 4 & sp_id & lamost spectrograph i d , ranging from 1 to 16 + 5 & fibre_id & lamost fibre i d for a given spectrograph , ranging from 1 to 250 + 6 & platetype & plate type ( ` b ' , ` m ' , ` f ' and ` v ' ) + 7 & objid & object i d in the input catalogues + 8 & objtype & initial object type from the survey input catalogues , could be stars ( ` star ' and ` star ' ) , + & & emission line objects ( ` emos ' ) and young stellar objects ( ` yso ' ) + 9 & snr_b & s / n(4650 ) per pixel + 10 & snr_r & s / n(7450 ) per pixel + 11 & ra & right ascension of j2000.0 ( deg ) + 12 & dec & declination of j2000.0 ( deg ) + 13 & l & galactic longitude ( deg ) + 14 & b & galactic latitude ( deg ) + 15 & uqflag & uniqueness flag . if the target has been observed @xmath248 times , then uqflag runs from 1 to @xmath248 , + & & with 1 denoting the spectrum of the highest snr_b + 16 & vr & radial velocity yielded by the lsp3 , after corrected for a systematic offset of @xmath913.5 km s@xmath121 ( km s@xmath121 ) + 17 & teff1 & yielded by the lsp3 from the weighted mean algorithm ( k ) + 18 & logg1 & log @xmath20 yielded by the lsp3 from the weighted mean algorithm ( cm s@xmath177 ) + 19 & feh1 & [ fe / h ] yielded by the lsp3 from the weighted mean algorithm ( dex ) + 20 & teff2 & yielded by the lsp3 from the @xmath249 minimization algorithm ( k ) + 21 & logg2 & log @xmath20 yielded by the lsp3 from the @xmath249 minimization algorithm ( cm s@xmath177 ) + 22 & feh2 & [ fe / h ] yielded by the lsp3 from the @xmath249 minimization algorithm ( dex ) + 23 & teff & final yielded by the lsp3 from the weighted mean algorithm , calibrated to photometric temperatures ( k ) + 24 & logg & final log @xmath20 yielded by the lsp3 from the weighted mean algorithm , same to logg1 in the current version ( cm s@xmath177 ) + 25 & feh & final [ fe / h ] yielded by the lsp3 from the weighted mean algorithm , same to feh1 in the current version ( dex ) + 26 & vr_err & error of radial velocity yielded by the lsp3 ( km s@xmath121 ) + 27 & teff_err & error of the final yielded by the lsp3 ( k ) + 28 & logg_err & error of the final log @xmath20 yielded by the lsp3 ( cm s@xmath177 ) + 29 & feh_err & error of the final [ fe / h ] yielded by the lsp3 ( dex ) + 30 & flags[9 ] & flags assigned to the final lsp3 parameters@xmath136 + 31 & fuv & galex @xmath138 band magnitude ( mag ) + 32 & nuv & galex @xmath139 band magnitude ( mag ) + 33 & g & xstps - gac @xmath20 band magnitude ( mag ) + 34 & r & xstps - gac @xmath6 band magnitude ( mag ) + 35 & i & xstps - gac @xmath42 band magnitude ( mag ) + 36 & j & 2mass @xmath118 band magnitude ( mag ) + 37 & h & 2mass @xmath154 band magnitude ( mag ) + 38 & k & 2mass @xmath155 band magnitude ( mag ) + 39 & w1 & wise @xmath250 band magnitude ( mag ) + 40 & w2 & wise @xmath251 band magnitude ( mag ) + 41 & w3 & wise @xmath147 band magnitude ( mag ) + 42 & w4 & wise @xmath150 band magnitude ( mag ) + 43 & err_fuv & error of the galex @xmath138 magnitude ( mag ) + 44 & err_nuv & error of the galex @xmath139 magnitude ( mag ) + 45 & err_g & error of the xstps - gac @xmath20 band magnitude ( mag ) + 46 & err_r & error of the xstps - gac @xmath6 band magnitude ( mag ) + 47 & err_i & error of the xstps - gac @xmath42 band magnitude ( mag ) + 48 & err_j & error of the 2mass @xmath118 band magnitude ( mag ) + 49 & err_h & error of the 2mass @xmath154 band magnitude ( mag ) + 50 & err_k & error of the 2mass @xmath155 band magnitude ( mag ) + 51 & err_w1 & error of the wise @xmath250 band magnitude ( mag ) + 52 & err_w2 & error of the wise @xmath251 band magnitude ( mag ) + 53 & err_w3 & error of the wise @xmath147 band magnitude ( mag ) + 54 & err_w4 & error of the wise @xmath150 band magnitude ( mag ) + 55 & ph_qual_2mass & 2mass photometric quality flag@xmath137 + 56 & ph_qual_wise & wise photometric quality flag@xmath153 + 57 & var_flg_wise & wise variability flag@xmath153 + 58 & ext_flg_wise & wise extended source flag@xmath153 + 59 & cc_flg_wise & wise contamination and confusion flag@xmath153 + @xmath252 see table2 of xiang et al .
( 2014b ) .
+ @xmath253 see http://www.ipac.caltech.edu/2mass/releases/allsky/doc/sec1_6b.html#origphot .
+ @xmath254 see http://wise2.ipac.caltech.edu/docs/release/allsky/expsup/sec2_2a.html#ext_flg .
lll col . & name & description + 60 & ebv_sfd & @xmath16 from the sfd extinction map ( mag ) + 61 & ebv_sp & @xmath16 derived from the star pair method ( mag ) + 62 & ebv_mod & @xmath16 derived by comparing the observed and synthetic model atmosphere colours ( mag ) + 63 & ebv_phot & @xmath16 derived by fitting multi - band photometry to the empirical stellar loci ( chen et al .
2014 ) ( mag ) + 64 & dist_em & distance derived using the empirical relations of absolute magnitudes as a function of stellar atmospheric + & & parameters as yielded by the miles stars except for rcs .
distances of rcs are derived assuming they are + & & standard candles ( kpc ) + 65 & dist_mod & distance derived from the theoretical isochrones of stellar models ( kpc ) + 66 & dist_phot & distance derived from the photometric parallax method ( chen et al .
2014 ) ( kpc ) + 67 & pmra_ppmxl & proper motion @xmath255 in ra from the ppmxl catalogues ( mas yr@xmath121 ) + 68 & pmdec_ppmxl & proper motion @xmath256 in dec from the ppmxl catalogues ( mas yr@xmath121 ) + 69 & epmra_ppmxl & proper motion error in ra from the ppmxl catalogues ( mas yr@xmath121 ) + 70 & epmdec_ppmxl & proper motion error in dec from the ppmxl catalogues ( mas yr@xmath121 ) + 71 & pmra_ppmxl_corr & proper motion @xmath255 in ra from the ppmxl catalogues after corrected for systematics + & & using the formula of carlin et al .
( 2013 ) ( mas yr@xmath121 ) + 72 & pmdec_ppmxl_corr & proper motion @xmath256 in dec from the ppmxl catalogues after corrected for systematics + & & using the formula of carlin et al .
( 2013 ) ( mas yr@xmath121 ) + 73 & pmra_ucac4 & proper motion @xmath255 in ra from the ucac4 catalogues ( mas yr@xmath121 ) + 74 & pmdec_ucac4 & proper motion @xmath256 in dec from the ucac4 catalogues ( mas yr@xmath121 ) + 75 & epmra_ucac4 & proper motion error in ra from the ucac4 catalogues ( mas yr@xmath121 ) + 76 & epmdec_ucac4 & proper motion error in dec from the ucac4 catalogues ( mas yr@xmath121 ) + 77 & pmra_ucac4_corr & proper motion @xmath255 in ra from the ucac4 catalogues after corrected for systematics + & & using the formula of huang et al .
( 2014 ) ( mas yr@xmath121 ) + 78 & pmdec_ucac4_corr & proper motion in @xmath256 in dec from the ucac4 catalogues after corrected for systematics + & & using the formula of huang et al .
( 2014 ) ( mas yr@xmath121 ) + 79 & pmra_xuyi2mass & proper motion @xmath255 in ra derived by comparing the xstps - gac and 2mass positions ( mas yr@xmath121 ) + 80 & pmdec_xuyi2mass & proper motion @xmath256 in dec derived by comparing the xstps - gac and 2mass positions ( mas yr@xmath121 ) + 81 & x & x coordinate in a galactocentric cartesian reference system , positive towards the galactic centre ( kpc ) + 82 & y & y coordinate in a galactocentric cartesian reference system , positive in the direction of disc rotation ( kpc ) + 83 & z & z coordinate in a galactocentric cartesian reference system , positive towards the north galactic pole ( kpc ) + 84 & u[2 ] & galactic space velocities in the x direction computed from dist_em and the corrected ppmxl and ucac4 + & & proper motions , respectively , positive towards the galactic centre ( km s@xmath121 ) + 85 & v[2 ] & galactic space velocities in the y direction computed from dist_em and the corrected ppmxl and ucac4 + & & proper motions , respectively , positive in the direction of galactic rotation ( km s@xmath121 ) + 86 & w[2 ] & galactic space velocities in the z direction computed from dist_em and the corrected ppmxl and ucac4 + & & proper motions , respectively , positive towards the north galactic pole ( km s@xmath121 ) + 87 & v_r[2 ] & galactic space velocities in the @xmath6 direction in a galactocentric cylindrical polar coordinate system , + & & computed from dist_em and the corrected ppmxl and ucac4 proper motions , respectively ( km s@xmath121 ) + 88 & v_phi[2 ] & galactic space velocities in the @xmath257 direction in a galactocentric cylindrical polar coordinate system , + & & computed from dist_em and the corrected ppmxl and ucac4 proper motions , respectively , + & & positive in the direction of counter disc rotation ( km s@xmath121 ) + 89 & v_z[2 ] & galactic space velocities in the @xmath100 direction in a galactocentric cylindrical polar coordinate system , + & & computed from dist_em and the corrected ppmxl and ucac4 proper motions , respectively , + & & positive towards the north galactic pole ( km s@xmath121 ) + 90 & e[2 ] & orbital eccentricities computed from dist_em and the corrected ppmxl and ucac4 proper motions , respectively + 91 & rapo[2 ] & maximum galactic radii reached by the orbits from dist_em and the corrected ppmxl and ucac4 + & & proper motions , respectively ( kpc ) + 92 & rperi[2 ] & minimum galactic radii reached by the orbits from dist_em and the corrected ppmxl and ucac4 + & & proper motions , respectively ( kpc ) + 93 & rg[2 ] & galactic guiding radii from dist_em and the corrected ppmxl and ucac4 proper motions , respectively ( kpc ) + [ cols= " < , < , < , < , < " , ]
taking advantage of the 4,000 fibres provided by the lamost and tailored to the geographical location of the telescope and the seasonal variations of the site weather , the lss - gac main survey aims to survey a significant volume of the galactic thin and thick discs and the halo in a large , contiguous sky area centred on the gac ( @xmath1 , @xmath2 ) , and obtain @xmath33700 9000 low resolution ( @xmath4 ) spectra , spectral classifications , radial velocities and atmospheric parameters for a statistically complete sample of @xmath7 3 million stars of all colours down to a limiting magnitude of @xmath13mag ( 18.5mag for selected fields ) that are uniformly and randomly selected from the ( @xmath6 , @xmath57 ) and ( @xmath6 , @xmath58 ) hess diagrams . together with distances , proper motions and radial velocities provided by gaia
, the lss - gac will yield a unique and promising dataset to advance our understanding of the structure and assemblage of the galaxy , particularly the galactic disk(s ) .
in addition to the main survey , the lss - gac also targets hundreds of thousands of objects in the vicinity fields of m31 and m33 , the other two spirals in the local group .
the targets include foreground galactic stars , background quasars and various objects of special interests in m31 and m33 , such as pne , hii regions , supergiants and globular clusters .
supplementary to the main survey , the lss - gac vb survey covers a significant fraction of randomly selected very bright stars ( @xmath8 mag ) in the whole northern celestial hemisphere , yielding an excellent sample of local stars of a wide sky coverage , selected with a simpler algorithm , to probe the solar neighbourhood . in this paper , we present a detailed description of the survey design , target selection , observation and data reduction of the lss - gac .
properties of data collected so far are also presented and discussed . by june 2013 ,
end of the first year regular surveys , a total number of 1,040,649 [ 672,721 ] spectra of s / n(7450 ) @xmath9 10 [ s / n(4650 ) @xmath9 10 ] have been collected , including 441,495 [ 219,045 ] , 123,916 [ 60,869 ] and 475,238 [ 392,807 ] from the main , the m31-m33 and the vb surveys , respectively .
about 15 25 per cent spectra are from duplicate targets , opening up the possibility of time - domain spectroscopy . for stars of spectral s / n(4650 ) @xmath9 10 , the pipelines for spectral flux - calibration and for stellar parameter determinations that have been developed at peking university ( cf . the two companion papers by xiang et al . 2014a ,
b ) have been able to deliver accurately flux - calibrated spectra and robust stellar parameters , including radial velocities @xmath25 , and atmospheric parameters , log @xmath20 and [ fe / h ] , with an accuracy of 5kms@xmath121 , 150k , 0.25dex and 0.15dex , respectively .
three samples consisting of stars of spectral s / n(4650 ) @xmath258 are defined , corresponding to , respectively , the footprints of the lss - gac main , m31-m33 and vb surveys .
spectral classifications and stellar parameters yielded by the lsp3 are presented in publically accessible online value - added catalogues , supplementary to the official lamost dr1 .
the value - added catalogues also contain multi - band photometric data from the far - uv to the mid - ir collected from the galex , xstps - gac , 2mass and wise surveys .
the values of the interstellar extinction @xmath16 , as well as distances of the sample stars have been determined .
an accuracy of 0.04 mag in @xmath16 is achieved .
proper motions collected from the ppmxl and ucac4 catalogues , after corrected for systematics using quasars , are also provided .
the values are typically accurate to 5 mas yr@xmath121
. new determinations of proper motions are deduced by combining measurements of the xstps - gac and 2mass surveys and are accurate to about 6 mas yr@xmath121 . finally , from the distances , proper motions and radial velocites , the orbital parameters ( eccentricity @xmath259 , guiding radius @xmath237 ) are calculated and included in the value - added catalogues for all sample stars .
samples released in this first release of value - added catalogues are biased to brighter and bluer stars than those observed with the lamost hitherto , in particular for stars targeted with m plates .
this is partly caused by the fact that some of the observations fail to reach the designed depth of the plates . with better observing strategy and more strict quality control ,
the situation is much improved in the second year regular surveys from september 2013 to may 2014 . restricted by the lack of suitable spectral templates in the red wavelength range ( beyond 7400 ) ,
the current implementation of the lsp is also unable to deliver robust stellar atmospheric parameters for many red stars targeted by the surveys . as such ,
a significant chunk of ( almost half ) stars observed hitherto that have s / n(4650 ) @xmath260 but s / n(7450 ) @xmath261 ( many of them are cool dwarfs or giants ) are missing in the current release of value - added catalogues .
we plan to improve the situation in the next data release .
* acknowledgments * we would like to thank the referee for his / her helpful comments .
this work is supported by national key basic research program of china 2014cb845700 .
guoshoujing telescope ( the large sky area multi - object fibre spectroscopic telescope , lamost ) is a national major scientific project built by the chinese academy of sciences .
funding for the project has been provided by the national development and reform commission .
lamost is operated and managed by the national astronomical observatories , chinese academy of sciences .
this work has made use of data products from the sloan digital sky survey ( sdss ) , galaxy evolution explorer ( galex ) , two micron all sky survey ( 2mass ) , wide - field infrared survey explorer ( wise ) , nasa / ipac infrared science archive and the cross - match service provided by cds , strabourg .
2012 , _ apjs _ , 203 , 21 aihara , h. , allende prieto , c. , an , d. , et al .
2011 , apjs , 193 , 29 anderson , e. , & francis , c. 2012 , astronomy letters , 38 , 331 bai , z .- r .
2014 , to be submitted berry , m. , ivezi , . ,
sesar , b. , et al .
2012 , apj , 757 , 166 bovy , j. , rix , h .- w . , & hogg , d. w. 2012 , apj , 751 , 131 bonaca , a. , geha , m. , & kallivayalil , n. 2012 , apjl , 760 , l6 carlin , j. l. , delaunay , j. , newberg , h. j. , et al .
2013 , apjl , 777 , castelli , f. , & kurucz , r. l. 2004 , arxiv : astro - ph/0405087 carlin , j. l. , lpine , s. , newberg , h. j. , et al . 2012 , raa , 12 , 775 chen , l. , hou , j. , et al .
2012 , raa , 12 , 805 chen , b .- q . ,
liu , x .- w . , yuan , h .- b . , et al .
2014 , mnras , 443 , 1192 chou , m .- y . ,
majewski , s. r. , cunha , k. , et al .
2011 , apjl , 731 , l30 2012 , _ raa _ , 12 , 1197 deng , l. , newberg , h. , liu , c. , et al .
2012 , raa , 12 , 735 2008 , _ apjs _ , 178 , 89 drew , j. e. , greimel , r. , irwin , m. j. , et al .
2005 , mnras , 362 , 753 2011 , _ aj _ , 142 , 72 falcn - barroso , j. , snchez - blzquez , p. , vazdekis , a. , et al .
2011 , a&a , 532 , a95 gmez , f. a. , minchev , i. , oshea , b. w. , et al .
2012 , mnras , 423 , 3727l5 grocholski , a. j. , & sarajedini , a. 2002 , aj , 123 , 1603 huang , y. , liu , x .- w . , yuan , h .- b . , et al . 2014 , mnras , submitted huo , z .- y . , liu , x .- w . , yuan , h .- b . , et al .
2010 , research in astronomy and astrophysics , 10 , 612 huo , z .- y . , liu , x .- w . ,
xiang , m .- s .
, et al . 2013 , aj , 145 , 159 ibata , r. , irwin , m. , lewis , g. , ferguson , a. m. n. , & tanvir , n. 2001 , nature , 412 , 49 ibata , r. , martin , n. f. , irwin , m. , chapman , s. , ferguson , a. m. n. , lewis , g. f. , & mcconnachie , a. w. 2007 , apj , 671 , 1591 ivezi , . ,
sesar , b. , juri , m. , et al .
2008 , apj , 684 , 287 jester , s. , schneider , d. p. , richards , g. t. , et al .
2005 , aj , 130 , 873 katz , d. , munari , u. , cropper , m. , et al .
2004 , mnras , 354 , 1223 kaiser , n. 2004 , proc .
spie , 5489 , 11 lasker , b. m. , lattanzi , m. g. , mclean , b. j. , et al .
2008 , aj , 136 , 735 lee , y. s. , beers , t. c. , an , d. , et al .
2011 , apj , 738 , 187 li , j. , newberg , h. j. , carlin , j. l. , et al .
2012 , apj , 757 , 151 liu , c. , xue , x. , fang , m. , et al .
2012 , apjl , 753 , l24 liu , x .- w . , yuan , h .-
b , huo , z .- y . , et al . , 2014 , in feltzing s. , zhao g. , walton n. , whitelock p. , eds , proc .
298 , setting the scene for gaia and lamost , cambridge university press , pp .
310 - 321 , preprint ( arxiv : 1306.5376 ) luo , a .-
zhang , h .-
t . , zhao , y .- h . ,
2012 , research in astronomy and astrophysics , 12 , 1243 luo , a .-
l . , zhang , y .- x . ,
& zhao , y .- h . 2004 ,
spie , 5496 , 756 majewski , s. r. , ostheimer , j. c. , rocha - pinto , h. j. , et al .
2004 , apj , 615 , 738 martin , d. c. , fanson , j. , schiminovich , d. , et al .
2005 , apjl , 619 , l1 martin , n. f. , ibata , r. a. , & irwin , m. 2007 , apjl , 668 , l123 martin , n. f. , ibata , r. a. , rich , r. m. , et al .
2014 , apj , 787 , 19 mcconnachie , a. w. , irwin , m. j. , ibata , r. a. , et al .
2009 , nature , 461 , 66 minniti , d. , saito , r. k. , alonso - garca , j. , lucas , p. w. , & hempel , m. 2011 , apjl , 733 , l43 monet , d. g. , levine , s. e. , canzian , b. , et al .
2003 , aj , 125 , 984 padmanabhan , n. , schlegel , d. j. , finkbeiner , d. p. , et al .
2008 , apj , 674 , 1217 pancino , e. , carrera , r. , rossetti , e. , & gallart , c. 2010 , a&a , 511 , a56 perryman , m. a. c. , de boer , k. s. , gilmore , g. , et al .
2001 , a&a , 369 , 339 perryman , m. a. c. , & esa 1997 , esa special publication , 1200 plummer , h. c. 1911 , mnras , 71 , 460 prugniel , p. , & soubiran , c. 2001 , a&a , 369 , 1048 prugniel , p. , soubiran , c. , koleva , m. , & le borgne , d. 2007 , arxiv : astro - ph/0703658 robin , a. c. , reyl , c. , derrire , s. , & picaud , s. 2003 , a&a , 409 , 523 rocha - pinto , h. j. , majewski , s. r. , skrutskie , m. f. , crane , j. d. , & patterson , r. j. 2004 , apj , 615 , 732 roeser , s. , demleitner , m. , & schilbach , e. 2010 , aj , 139 , 2440 roeser , s. , schilbach , e. , schwan , h. , et al .
2008 , vizier online data catalog , 1312 , 0 schlafly , e. f. , & finkbeiner , d. p. 2011 , apj , 737 , 103 schlegel , d. j. , finkbeiner , d. p. , & davis , m. 1998 , apj , 500 , 525 skrutskie , m. f. , cutri , r. m. , stiening , r. , et al .
2006 , aj , 131 , 1163 snchez - blzquez , p. , peletier , r. f. , jimnez - vicente , j. , et al .
2006 , mnras , 371 , 703 soubiran , c. , le campion , j .- f .
, cayrel de strobel , g. , & caillo , a. 2010 , a&a , 515 , a111 souchay , j. , andrei , a. h. , barache , c. , et al .
2012 , a&a , 537 , a99 steinmetz , m. , zwitter , t. , siebert , a. , et al .
2006 , aj , 132 , 1645 widrow , l. m. , gardner , s. , yanny , b. , dodelson , s. , & chen , h .- y .
2012 , apjl , 750 , l41 williams , m. e. k. , steinmetz , m. , binney , j. , et al .
2013 , mnras , 436 , 101 witham , a. r. , knigge , c. , drew , j. e. , et al .
2008 , mnras , 384 , 1277 wright , e. l. , eisenhardt , p. r. m. , mainzer , a. k. , et al .
2010 , aj , 140 , 1868 wu , y. , luo , a .-
l . , du , b. , et al .
2014 , submitted to the proceedings of iau symposium no.306 wu , z .- y .
, ma , j. , & zhou , x. 2011 , pasp , 123 , 1313 xiang , m .- s . , liu , x .- w . , yuan , h .- b . , et al.,2014a , mnras , accepted xiang , m .- s . , liu , x .- w . , yuan , h .- b . , et al.,2014b , mnras , accepted yang , f. , carlin , j. l. , liu , c. , et al .
2012 , raa , 12 , 781 yang , f. , deng , l. , liu , c. , et al . 2014 , new astronomy , 26 , 72 yanny , b. , rockosi , c. , newberg , h. j. , et al . 2009 ,
aj , 137 , 4377 yao , s. , liu , c. , zhang , h .-
, et al . 2012 , research in astronomy and astrophysics , 12 , 772 yi , z. , luo , a. , song , y. , et al .
2014 , aj , 147 , 33 york , d. g. , adelman , j. , anderson , j. e. , jr . , et al .
2000 , aj , 120 , 1579 yuan , h .-
l . , ren , j. , wang , j. , jin , g. , & zhang , h .-
2008 , proc .
spie , 7016 yuan , h. b. & liu , x. w. 2012 , mnras , 425 , 1763 yuan , h .- b . , liu , x .- w . , huo , z .- y . , et al .
2010 , research in astronomy and astrophysics , 10 , 599 yuan , h .- b . , liu , x .- w . , huo , z .- y . , et al .
2014 , in feltzing s. , zhao g. , walton n. , whitelock p. , eds , proc .
298 , setting the scene for gaia and lamost , cambridge university press , pp .
240 - 245 , preprint ( arxiv : 1306.5614 ) yuan , h. b. , liu , x. w. , & xiang , m. s. 2013 , mnras , 430 , 2188 zacharias , n. , finch , c. t. , girard , t. m. , et al . 2013 , aj , 145 , 44 zhang , y .- y . , carlin , j. l. , yang , f. , et al .
2012 , raa , 12 , 792 zhang , h .- h . ,
liu , x .- w . , yuan , h .- b . , et al . , research in astronomy and astrophysics , 2013 , 13 , 490 zhang , h .- h . ,
liu , x .- w . , yuan , h .- b . , et al .
2014 , research in astronomy and astrophysics , 14 , 456 zhao , g. , zhao , y .- h . , chu , y .- q . , et al .
2012 , research in astronomy and astrophysics , 12 , 723 zhu , y. , hu , z. , zhang , q. , wang , l. , & wang , j. 2006 , proc .
spie , 6269 , 20 zhu , y. , hu , z. , wang , l. , et al .
2010 , scientia sinica phys , mech & astron , 40 , 1 | as a major component of the lamost galactic surveys , the lamost spectroscopic survey of the galactic anti - centre ( lss - gac ) aims to survey a significant volume of the galactic thin / thick discs and halo for a contiguous sky area of over 3,400 deg@xmath0 centred on the galactic anti - centre ( @xmath1 , @xmath2 ) , and obtain @xmath33700 9000 low resolution ( @xmath4 ) spectra for a statistically complete sample of @xmath5 m stars of all colours down to a limiting magnitued of @xmath6 @xmath7 17.8mag ( to 18.5mag for limited fields ) . together with gaia , the lss - gac will yield a unique dataset to advance our understanding of the structure and assemblage history of the galaxy , in particular its disk(s ) .
in addition to the main survey , the lss - gac will also target hundreds of thousands objects in the vicinity fields of m31 and m33 and survey a significant fraction ( over a million ) of randomly selected very bright stars ( vb ; @xmath8 mag ) in the northern hemisphere . during the pilot and the first year regular surveys of lamost ,
a total of 1,042,586 [ 750,867 ] spectra of a signal to noise ratio s / n(7450 ) @xmath9 10 [ s / n(4650 ) @xmath9 10 ] have been collected . in this paper , we present a detailed description of the target selection algorithm , survey design , observations and the first data release of value - added catalogues ( including radial velocities , effective temperatures , surface gravities , metallicities , values of interstellar extinction , distances , proper motions and orbital parameters ) of the lss - gac .
[ firstpage ] galaxy : disc galaxy : structure galaxy : stellar content stars : distances dust : extinction surveys . |
stars can be accompanied by circumstellar matter throughout their life . in the early phase of their evolution ,
they are surrounded by massive gas - rich primordial disks that form during the collapse of the molecular cloud core as a consequence of angular momentum conservation ( e.g. , * ? ? ?
according to the current paradigm , most of these disks dissipate within a few million years @xcite . during this evolution disk material accretes onto the star ,
forms planets , or is removed from the system through outflows and photoevaporation ( * ? ? ? * and references therein ) .
what left is a tenuous debris disk which differs from their predecessors in many ways .
dust grains in these disks are second generation : collisional erosion and evaporation of previously formed planetesimals provides a continuous replenishment of dust particles removed by the stellar radiation or wind on a significantly shorter timescale than the age of the star @xcite .
debris disks are thought to be gas - poor with a significantly lower gas - to - dust ratio than that of primordial disks ( e.g. * ? ? ?
* ) , because processes like collisions , evaporation , and photodesorption from icy grains only produce a small amount of secondary gas @xcite .
one way to study the dust component in debris disks is to observe its thermal emission at infrared ( ir ) and millimeter wavelengths .
thanks to a series infrared space missions now we know hundreds of debris disks . the spectral energy distribution ( sed ) of the disks excess emission
is routinely used to infer their fundamental properties such as the characteristic dust temperature and the amount of dust grains , while resolved images give information about their spatial extent and even the grain size distribution ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
due to the intimate link between the grains and larger parent planetesimals the study of dust disk properties can also give insight into the characteristics and evolution of parent planetesimal belt(s ) @xcite .
observing the signatures of presumably very small amounts of gas is a more challenging task . despite extensive surveys @xcite , the number of debris disks with known gas component is still very limited . in the edge - on disks around @xmath1pic and hd32297
the gas was firstly detected via absorption lines @xcite . at 49 cet and
hd21997 a substantial amount of co gas has been observed @xcite , while the gaseous disks around hd172555 , hd181296 and aumic were identified based on , and fluorescent h@xmath3 line emissions , respectively @xcite .
recently , @xcite identified hi ( 76 ) lines in the mid - infrared spitzer / irs spectra of 8 young ( 520myr old ) f7-k0 type stars that hosted debris disks .
it is yet an open question whether these emission lines are linked to low level accretional processes and thereby indicate the presence of long lived gaseous circumstellar disks or they have pure chromospheric origin @xcite .
while debris dust disks have been found around stars with a variety of ages the current sample of gaseous disks is mostly limited to young ( typically 1040myr old ) stars .
the origin of gas in these systems is not unambiguous yet . because of their relatively young age we can not exclude that some of them harbor hybrid disks where the dust has secondary origin and produced from planetesimals , while the gas is predominantly primordial , the remnant of the original disk
indeed , as one of the early discoveries of alma , our group identified such a hybrid disk around the @xmath030-myr - old hd21997 @xcite .
this discovery challenges the current paradigm on the timescale of primordial disk evolution .
nevertheless , based on this small and , in many ways , limited sample we can not draw general and comprehensive conclusions about the production mechanism and long term evolution of the gas component in debris disks . motivated to find additional examples of gaseous debris disks and scrutinize the fundamental properties of their gas and dust contents , we used the atacama pathfinder experiment and the iram30 m radio telescopes to search for molecular gas in 20 debris disks in the rotational lines of co. this work is a continuation of our similar survey of 20 objects @xcite . in this paper
we review the results of the new survey and report on the discovery of a gaseous debris disk around hd131835 , a @xmath016myr old member of the upper centaurus lupus association . for this object
we present additional photometric and spectroscopic observations with the _ herschel space observatory _ @xcite .
in our previous survey we searched for gas in 20 young ( @xmath450myr ) debris disks with high fractional luminosity ( @xmath5 ) around a- and f - type stars @xcite .
the current project focuses on 20 additional debris disks .
we included seven disks where the selection criteria were similar to those applied in the first survey , except that we allowed lower fractional luminosities down to @xmath6 .
we added four targets ( hd10939 , hd17848 , hd161868 and hd182681 ) which are similar to the gaseous 49cet and hd21997 systems in that they harbor cold , very extended debris disks . by including six debris disks around g k - type star we also extended the sample to sun - like stars .
recently , @xcite and @xcite reported nightly variability of the caii k absorption line in the spectra of some a - type debris disk host stars , indicating the presence of circumstellar gas .
note that among the known gaseous disks , @xmath1pic , 49ceti , and hd172555 also show similar variable ca absorption @xcite .
thus we selected three nearby ( @xmath475pc ) targets , hd110411 , hd182919 , and hd183324 , from this sample .
basic properties of the selected targets are summarized in table [ basictable ] .
[ [ observations - with - apex . ] ] observations with apex .
+ + + + + + + + + + + + + + + + + + + + + + + the majority of our observations were carried out with the 12 m atacama pathfinder experiment radio telescope ( apex , * ? ? ?
* ) in the framework of three different programmes .
all of our targets were observed in the @xmath7=3@xmath82 transition of @xmath9co at a rest frequency of @xmath10 = 345.796ghz . in program m-087.f-0001 - 2011
the observations were performed with the swedish heterodyne facility instrument / apex2 ( shefi , * ? ? ?
* ) receiver , while in m-092.f-0012 - 2013 and m-093.f-0010 - 2014 the first light apex submillimeter heterodyne receiver ( flash+ , * ? ? ?
* ) was used .
the latter is a dual - frequency receiver that operates simultaneously in the 345ghz and 461ghz atmospheric windows , therefore we obtained @xmath9co @xmath7=4@xmath83 line ( @xmath10 = 461.041ghz ) observations as well . for hd131835 , co ( 21 ) line observations
were also conducted using the shefi / apex1 instrument . for shefi observations acquired before june 2011 we utilized the fast fourier transform spectrometer ( ffts )
backend with a spectral resolution of 488khz ( 0.42kms@xmath11 at @xmath7=32 transition ) . for later shefi measurements
the extended bandwidth fast fourier transform spectrometer ( xffts , * ? ? ?
* ) with 32768 channels providing a spectral resolution of 77khz ( 0.066kms@xmath11 at @xmath7=32 transition ) was connected to the receiver .
flash+ was always connected to the xffts backend .
beam sizes of apex are @xmath027 , @xmath018 and @xmath014 at 230 , 345 and 460ghz , respectively .
all of our apex observations were performed in on - off observing mode .
the co ( 32 ) spectral line observations of hd131835 were further supplemented by yet unpublished measurements obtained in our previous apex programme e-083.c-0303 ( for details see , * ? ? ?
[ [ observations - with - the - iram-30m - telescope . ] ] observations with the iram 30 m telescope .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the three targets selected because of variable caii k absorption line were observed with the iram 30 m telescope using the multi - band heterodyne eight mixer receiver ( emir , * ? ? ?
* ) as part of the iram programme no . 172 - 13 .
we searched for gas in our targets at 230.538ghz , the @xmath12 transition of the @xmath9co .
the observations were conducted in wobbler - switching on - off mode with a wobbler throw of 60 . for the backend
, we used the new fast fourier transform spectrometer ( fts ) with a frequency resolution of 200khz providing a velocity resolution of @xmath00.25kms@xmath11 in the @xmath7=2@xmath81 transition .
table [ obslog ] summarizes the main characteristics of the observing programmes .
[ [ data - reduction . ] ] data reduction .
+ + + + + + + + + + + + + + + both the apex and iram spectra have been processed using the gildas / class package .
for the final average spectrum , we discarded noisy scans and a baseline was subtracted from each individual scan .
the baseline was typically linear , except for a few cases where we used second order polynomials .
the final spectrum was derived as an average of the individual spectra weighted by the inverse square of their rms noise . as a final step
the obtained antenna temperatures were converted to line flux densities . for the apex data we used kelvin - to - jansky conversion factors of 39 , 41 , and 48jy k@xmath11 for co ( 21 ) , co ( 32 ) , and co ( 43 ) transitions , respectively . for the iram co ( 21 )
observations the corresponding number was 7.8jy k@xmath11 .
these conversion factors were taken from the relevant apex and iram home pages .
[ [ outcome ] ] outcome of the survey .
+ + + + + + + + + + + + + + + + + + + + + + co emission was evident only for hd131835 , where the integrated line flux of the 32 transition was successfully detected at a 5.0@xmath13 level .
figure [ co32 ] shows the baseline subtracted co ( 32 ) profile of this target .
the measured line has a width of roughly 12 km s@xmath11 and is centred at a velocity of 7.2 km s@xmath11 with respect to the local standard of rest ( lsr ) , in very good agreement with the systemic lsr velocity of the star ( [email protected] km s@xmath11 , computed from the radial velocity derived in sect .
[ ferosanalysis ] ) .
the measured line tentatively shows a double peak profile in accordance what we expect for a gas disk in keplerian rotation around a star .
the peak flux is @xmath0370mjy , the integrated line flux
obtained by integrating the line over a 12 km s@xmath11 interval
is [email protected] km s@xmath11 . by inspecting the same velocity interval in the @xmath7=21 and 43 spectra ( fig .
[ co32 ] ) we found no statistically significant evidence for lines , the obtained integrated line fluxes are [email protected] and [email protected] km s@xmath11 , respectively .
we note that hd 131835 was previously observed in the co 21 transition by @xcite resulting in a non - detection .
this observation was about two times less sensitive than ours . assuming a linewidth of 5kms@xmath11
, @xcite deduced a 5@xmath13 line flux upper limit of 5.1jy km s@xmath11 from this measurement , that is compatible with our result . ) .
the best fit model spectra ( see sect .
[ gasparams ] ) are plotted by red lines . _
[ co32 ] ] none of our other targets were detected at any of the co transitions .
for these sources , upper limits were computed as @xmath15 where @xmath16 is the 1@xmath13 measured noise , @xmath17 is the velocity channel width , and @xmath18 is the number of velocity channels over an interval of 10kms@xmath11 . the obtained line fluxes and upper limits
are listed in table [ basictable ] . because of the detection of co ( sect .
[ outcome ] ) , for hd131835 we performed several additional observations . in order to better characterize its spectral energy distribution at far infrared / submillimeter wavelengths and to search for and emission from the disk we obtained photometric and spectroscopic measurements using the photodetector array camera and spectrometer ( pacs ,
* ) and the spectral and photometric imaging receiver ( spire , * ? ? ?
* ) onboard the _ herschel space observatory _
( program i d ot2_amoor_3 ) .
these data were complemented by mid - infrared spectra obtained with the _
spitzer space telescope _
@xcite . to investigate the stellar properties a high - resolution ground - based optical spectrum was also taken . [
[ pacs - maps . ] ] pacs maps .
+ + + + + + + + + + far - infrared maps were obtained with the pacs detector on 2012 september 9 and 11 in mini scan - map mode ( pacs observer s manual v2.5 ) using a scan speed of 20s@xmath11 with 10 scan - legs of 3 length separated by 4 .
we made two scans with scan angles of 70@xmath19 and 110@xmath19 both at 70 ( obsid : 1342250784 , 1342250785 ) and at 100 ( obsid : 1342250868 , 1342250869 ) . since the pacs photometer observed in two bands simultaneously ( at 160 in addition to 70 or 100 ) these measurements provided four 160 scans as well .
data processing was done with the herschel interactive processing environment ( hipe , * ? ? ?
* ) version 13 using pacs calibration tree no.65 and the standard hipe script optimized for the reduction of bright sources .
additionally we used the recently developed `` gyro '' correction to reduce the pointing jitter .
we selected only those data frames from the timeline where the actual scan speed of the spacecraft was between 15 and 25 s@xmath11 . to eliminate the marked low - frequency ( 1/f ) noise we applied highpass filtering with filter width parameters of 15 , 20 , and 35 for the 70 , 100 , and 160 data , respectively . in order to avoid flux loss the 25 radius vicinity of our targets
was excluded from the filtering .
for glitch removal we used the second - level deglitching algorithm .
as a final step , in each band we applied the photproject task to combine all frames into a map using the default pixel fraction ( 1.0 ) and pixel sizes of 11 , 14 , and 21 at 70 , 100 , and 160 , respectively .
hd131835 was clearly detected in all pacs bands .
we performed aperture photometry on the source using a radius of 18 , while the sky background was estimated in an annulus between 40 and 50 .
the aperture was placed at the source s centroid position .
the offsets between the derived centroids and the targets optical position ( corrected for the proper motion using the epochs of pacs observations ) were @xmath41 in all bands . for sky noise determination , we distributed sixteen apertures with radii of 18 ( identical to the the source aperture ) randomly along the background annulus . by performing aperture photometry without background subtraction in each aperture , we computed the sky noise as the standard deviation of these background flux values .
we applied aperture correction in each band , by using correction factors taken from the calibration file attached to the measurement . to derive the final uncertainty of our photometry we added the measurement errors and the calibration uncertainty ( 7% , * ? ? ?
* ) quadratically .
the obtained flux density values and their uncertainties are listed in table [ phottable ] .
[ [ pacs - spectroscopy . ] ] pacs spectroscopy .
+ + + + + + + + + + + + + + + + + + we carried out pacs spectroscopic observations centred on the 63 ( obsid : 1342248686 ) and the 158 ( obsid : 1342248687 ) lines .
these fine structure lines if arising from circumstellar disks are unresolved in _ herschel _ pacs observations .
observations were performed on 2012 july 29 using the line spectroscopy observing template that enables the coverage of a small wavelength interval around these lines . to efficiently eliminate telescope and sky background ,
the measurements were obtained in chop - nod mode .
the pacs spectrometer array consists of 5@xmath205 spectral pixels ( spaxel ) , each having a size of 94 .
we made small 3@xmath203 and 2@xmath202 raster maps with raster point / line step sizes of 3 and 2 for the and observations , respectively .
the maps were centred on the target , and they were repeated two and four times in the case of and observations , respectively .
raster map observing mode has been selected to mitigate the risk of eventual pointing offsets of the telescope significant larger than 2 .
while the 47@xmath2047 spatial footprint of the integral field unit is sufficiently large to almost entirely cover the spectrometer spatial beam , a single spaxel of [email protected] size is strongly undersampling the beam especially at the short wavelength of the line . as a consequence of undersampling , in the baseline pointed observing scheme the spectrum of a point source
is extracted from the central spaxel of the 5@xmath205 unit applying a correction for fluxes due to small pointing offsets from the centre of the central spaxel and point jitter . however ,
if the source flux is lower than about a few jy per spaxel then correction factors estimated from the source continuum sampled by the eight neighbors of the central spaxel may become unreliable and the spectrum extracted from the central spaxel may have to be left uncorrected for mispointing . to overcome this problem
, we apply mapping observation of the same field of view with sub - spaxel raster step size . when data from raster positions is combined the sampling of the beam
is highly improved even in case of an eventual pointing offset of the raster central position .
this technique directly provides a high resolution image of the source where the spectrum can be extracted within a synthetic aperture centered on the measured peak position .
we apply the telescope normalization scheme for spectro - photometric flux calibration , where the herschel telescope background calibrated on neptune is used as an absolute radiometric reference .
the method efficiently eliminates any drifts with time in the system response at the frequency of individual chopper cycles resulting the offset signals in nod a and nod b positions perfectly cancel each other .
this method may improve signal - to - noise - especially for faint sources in the sub - jansky per spaxel regime - comparing to the baseline flux calibration method which relies on response estimates from calibration blocks executed at the beginning of the observation and propagated to the covered wavelength range through the relative spectral response function .
flux calibrated ifu cubes obtained at subsequent raster positions are combined into a single spectral cube using a 3d drizzling algorithm @xcite available in the pacs interactive pipeline in hipe .
the strength of the drizzling algorithm is that the convolution with the detector footprint is minimized especially when we apply the `` gyro '' improved high spatial resolution reconstruction of instantaneous pointing jitter .
the projected pixel size is adjusted to nyquist sample the beam at the line central wavelength . in the last data processing step , spectra from the drizzled spectral cube
are combined within an extraction aperture of 6.3 radius centered on the source position .
figure [ herschellines ] shows the obtained and spectra .
though in the spectrum there is a peak at 157.7 its significance is only 2.3@xmath13 , thus neither the or line was detected towards our target . in order to derive upper limits for the line fluxes we used the continuum subtracted spectra and determined the rms noise ( @xmath21 ) in a 20 pixels wide window centred on the expected line wavelength .
the upper limits were computed as @xmath22 where @xmath23 is the width of one pixel in hz , while @xmath24 is the width of an unresolved emission line in pixels .
values of @xmath24 were calculated as the ratio of the full width half maximum ( fwhm ) of unresolved lines ( taken from pacs observer s manual v2.5 , 0.020 and 0.126 for and spectra , respectively ) to the appropriate pixel widths ( in ) . the obtained upper limits for the and lines
are 1.5@xmath2010@xmath25wm@xmath26 and 5.3@xmath2010@xmath27wm@xmath26 , respectively . ]
[ [ spire - maps . ] ] spire maps .
+ + + + + + + + + + + we observed hd131835 with the spire photometer on 2012 august 9 , in small scan map mode ( spire handbook v2.5 ) at 250 , 350 , and 500 simultaneously .
data were processed with hipe 13 following the standard pipeline processing steps with default values for all applied tasks .
the final maps were produced using the naivemapper task .
the beam sizes are 176 , 239 , and 352 at 250 , 350 , and 500 , respectively , and the maps were resampled to pixel sizes of 6 , 10 , and 14 at these wavelengths .
our target is detected in all bands .
we used point - spread function ( psf ) photometry to extract the source fluxes , the corresponding spire beam profiles were taken from the calibration context .
this method yielded flux densities of [email protected] , [email protected] , and [email protected] at 250 , 350 , and 500 , respectively . in order to validate our flux values the photometry
was also performed using the spire timeline fitter task in hipe . instead of using the final maps , this task fits two dimensional elliptical or circular gaussian functions to the baseline - subtracted timeline data of the photometer at the coordinates of the source @xcite .
the radius of the region that includes the peak of the source was set to 22 , 30 , 42 for 250 , 350 , and 500 maps .
the background was estimated from an annulus between 300 and 350 centred on the source . at 250 and 350
we used elliptical gaussian during the fitting , at 500 , because of the source s lower signal - to - noise ratio , we applied a circular gaussian .
using the timeline fitter we obtained flux densities of [email protected] , [email protected] , and [email protected] at 250 , 350 , and 500 , respectively , in good accordance with the ones derived from psf photometry .
in the following analysis , we will use the psf photometry .
the final uncertainties were derived as the quadratic sum of the measurement errors and the overall calibration uncertainty of the spire photometer ( 5.5% * ? ? ? * ) .
the derived results are quoted in table [ phottable ] .
hd131835 was observed with the infrared spectrograph ( irs , * ? ? ?
* ) twice , first on 2007 august 30 ( pid40651 ) , then on 2008 april 1 ( pid40235 ) .
both spectra were obtained in staring mode using the low - resolution irs modules ( sl and ll ) , covering the @xmath28 m wavelength range with a spectral resolution of @xmath29 .
we retrieved the processed spectra from the cassis database @xcite .
as post - processing , some outlying data points were discarded .
we fitted polynomials to the data of individual irs modules using a robust method and then searched for data points outlying by more than 4@xmath13 from these fits .
we found that there is a mismatch between the sl1 and ll2 modules in both spectra .
since the shortest wavelength parts of the spectra match perfectly with the predicted photospheric fluxes ( see sect .
[ stellarprops ] ) , the modules were stitched together by scaling the ll modules to the sl ones .
the multiplicative scaling factors were estimated from the overlapping spectral regions of the sl1 and ll2 modules .
the spectra of hd131835 do not show any prominent silicate features .
following @xcite and @xcite , we also looked for atomic metal and hydrogen lines , and rotational lines of molecular hydrogen in the spectrum .
however , none of the lines are detected .
we compared the two irs spectra of hd131835 obtained at different epochs .
the flux differences were typically in the order of 5% over the whole spectral range , i.e. the continuum level was unchanged .
for the further analysis the two spectra were combined using a simple weighted average . for the sed modelling process
the combined irs spectrum was split into 15 adjacent wavelength bins between 6 and 35 .
the derived flux density values are listed in table [ phottable ] .
their uncertainties were computed by adding an 5% absolute calibration uncertainty quadratically to the measured ones .
high - resolution optical spectroscopy of hd131835 was performed using the fiber - fed extended range optical spectrograph ( feros , * ? ? ?
* ) installed on the 2.2 m mpg / eso telescope in la silla , on 2011 april 17 using the `` object - sky '' configuration , with one fiber positioned at the target , and the other one on the sky . the integration time was 180 s. feros has a mean resolving power of r @xmath0 48000 and covers the wavelength range between 3500 and 9200 in 39 echelle orders .
data reduction , including bias subtraction , flat - field correction , background subtraction , the definition and extraction of orders , and wavelength calibration , was performed using the feros data reduction system pipeline at the telescope .
hd131835 is an a2-type star @xcite located at 122.7@xmath30pc away @xcite .
based on astrometric data made with esa s _ hipparcos _ satellite , @xcite proposed the star to be a member of the @xmath016myr old @xcite upper centaurus lupus ( ucl ) subgroup of the sco - cen association with a membership probability of 95% . according to our current knowledge hd131835 has no stellar companion @xcite .
we used spectroscopic and photometric data to model the stellar photosphere and to estimate the fundamental stellar properties of hd131835 .
moreover , supplementing the _ hipparcos _ astrometric data by our ground - based radial velocity measurement , its membership in ucl was also reconsidered .
[ [ ferosanalysis ] ] analysis of the feros spectrum .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in order to estimate the radial velocity and different stellar parameters of our target the measured feros spectrum was compared with templates taken from the spectral library of @xcite . in this spectral library the continuum normalization
was performed by dividing the absolute flux spectrum by its calculated continuum .
because of the applied method , in some spectral regions the continuum of the model spectra can significantly differ from a continuum that would be obtained using a regular normalization scheme of observational spectroscopy .
therefore first we used high order polynomials to flatten the model spectra , and then the same method was applied in the continuum normalization of our observed spectrum as well .
we used an iterative method to estimate the stellar parameters .
we first applied the cross - correlation technique to derive the radial velocity of the star by convolving the measured spectrum with a template taken from the spectral library .
we selected a template spectrum with @xmath31 = 0.0 $ ] , @xmath32 , and @xmath33k ( corresponding to the a2 spectral type of the target quoted by the simbad ) .
the cross correlation function was calculated by the fxcor task in iraf . after transforming the measured spectrum to the laboratory system we compared it with a grid of munari synthetic spectra , using the 40006200 wavelength range , excluding the @xmath34 region and na d lines .
the grid was compiled by varying the effective temperature , surface gravity , and projected rotational velocity of the model spectra , while the metallicity was fixed to @xmath31 = 0.0 $ ] . by finding the global minimum we used the derived parameters to repeat the radial velocity determination and
then recompute the stellar parameters .
this process yielded the following best - fit parameters : @xmath35k , @xmath36 , @xmath37kms@xmath11 .
based on dwarf spectral type vs. effective temperature scale presented in @xcite the derived @xmath38 rather corresponds to a spectral type of a4 .
the radial velocity was found to be [email protected] km s@xmath11 , consistently with our previous measurement of [email protected] km s@xmath11 @xcite .
taking the weighted average of the two radial velocity values we obtained @xmath39 km s@xmath11 .
[ [ analysis - of - photometric - and - near - ir - data . ] ] analysis of photometric and near - ir data .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + to provide an additional independent estimate for fundamental stellar properties and model the stellar photosphere we fitted an atlas9 atmosphere model @xcite to the optical and near - ir observations of the target .
photometric data were taken from the _ tycho 2 _ @xcite , _ hipparcos _
@xcite , and two micron all sky survey catalogues ( 2mass , * ? ? ?
these data were further supplemented by wide - field infrared survey explorer ( _ wise _ ) @xmath40 band photometry at 3.4 from the _ wise _ all - sky database @xcite and johnson b , v photometry from the catalogue of @xcite . with a distance of @xmath0123pc , hd131835 could be outside the local bubble , thus its reddening might not be negligible . in order to provide a rough estimate of the reddening at the distance of hd131835
, we collected those stars in its vicinity ( with separations @xmath43 ) that have both _ hipparcos_-based trigonometric parallax and measured strmgren colour indices and h@xmath1 index in the photometric catalogue of @xcite . for these stars we derived @xmath41 colour excesses ( as @xmath42 ) from the strmgren data by using the appropriate calibration processes @xcite .
figure [ ebv ] shows the derived @xmath41 values as a function of trigonometric parallaxes . at the distance of hd131835 ( marked by a vertical line )
the reddening is between @xmath00.015 and @xmath00.075mag .
we note that consistent with our results , the extinction map of @xcite shows a reddening of @xmath00.083mag in the direction of our target .
this value gives the total reddening within the milky way for the line of sight , thus it can be considered as an upper limit . therefore in the photosphere modelling we fitted both the effective temperature and the reddening ,
the latter was limited between 0.0 and 0.08mag ) . by adopting solar metallicity and
@xmath43 our @xmath44 minimization yielded @xmath38 = 8250@xmath45k and @xmath41 = 0.025@xmath46 . considering these data and the @xmath47-based trigonometric distance we derived a luminosity of @xmath48 = [email protected]@xmath49 for hd131835 . ]
[ [ ucl - membership . ] ] ucl membership .
+ + + + + + + + + + + + + + + using the new combined radial velocity and _ hipparcos _ astrometric data of hd131835 we computed a galactic space motion of [email protected]@xmath141.3kms@xmath11 , [email protected]@xmath142.1kms@xmath11 , and [email protected]@xmath140.8kms@xmath11 with respect to the sun .
this space velocity is in perfect agreement with the charateristic space motion of ucl ( [email protected]@xmath140.6kms@xmath11 , [email protected]@xmath140.4kms@xmath11 , [email protected]@xmath140.3kms@xmath11 , * ? ? ?
the position of hd131835 in the color - magnitude diagram ( fig .
[ cmd ] ) matches well the locus defined by ucl members , indicating that its isochrone age is consistent with that of the cluster .
our results confirm the previously proposed membership of hd131835 in ucl , therefore we adopted the age of the cluster , @xmath016myr , for our star . with this age and our previous effective temperature and stellar luminosity estimates
we derived a stellar mass of [email protected]@xmath49 by using solar metallicity isochrones from @xcite .
i@xmath50 color indices for known members of ucl ( the member list was taken from * ? ? ?
* ) located within 150pc and having parallax error @xmath410% ( gray dots ) and for hd131835 ( black square ) .
photometric data were taken from the _ hipparcos _ catalog and dereddened using extinction values from @xcite and , for hd131835 , based on our result . _ [ cmd ] ] [ [ sed - modelling . ] ] sed modelling .
+ + + + + + + + + + + + + + in order to construct the spectral energy distribution of hd131835 , the previously mentioned pacs , spire , and irs data were further supplemented by infrared and submillimeter photometry from the literature ( table [ phottable ] ) .
the compiled sed is plotted in figure [ sed ] .
hd131835 exhibits strong excess at far - ir wavelengths , but as the irs spectrum shows , the sed of the source already starts to departing from photospheric emission at @xmath07 , indicating the presence of warmer circumstellar dust as well .
we used the atlas9 atmosphere model of hd131835 ( sect .
[ stellarprops ] ) to predict photospheric flux contribution at different mid- and far - ir wavelengths .
the average accuracy of the predicted photospheric fluxes is around 5% .
the excess emission of debris disks is generally well fitted by a single ( modified ) blackbody or a combination of two different temperature blackbody components @xcite .
this simple model can provide estimates of fundamental disk properties such as the characteristic dust temperature(s ) and the fractional luminosity ( @xmath51 is the ratio of the luminosity of the dust emission to the bolometric luminosity of the host star ) .
we fitted the excess emission of hd131835 by single- and two - temperature models .
for the single - temperature fitting we used a modified blackbody : @xmath52 where @xmath53 is the measured excess emission at @xmath10 , @xmath54 is the planck function , @xmath55 is the dust temperature , @xmath56 is a scaling factor that is proportional to the solid angle of the emitting region , while @xmath57 if @xmath58 and @xmath59 if @xmath60 . in the two - temperature model we used a combination of a warmer simple blackbody ( there is no sense to use a modified blackbody since the values of @xmath61 and @xmath1 can not be constrained well from the sed ) and a modified blackbody .
we applied a levenberg - marquardt algorithm @xcite to find the best - fitting model .
an iterative way was used to compute and apply color corrections for the photometric data during the fitting process ( see e.g. * ? ? ?
we found that only the two - temperature model ( plotted in figure [ sed ] ) could reasonably fit both the mid- and far - ir / submm excess emission .
this model provides dust temperatures of @xmath62k and @xmath63k for the warm and cold dust components , respectively .
we obtained @xmath64 and @xmath65 for the cold belt .
the warm component has a fractional luminosity of ( [email protected])@xmath2010@xmath66 , while the cold component has a fractional luminosity of ( [email protected])@xmath2010@xmath67 . ]
we estimated the dust mass in the disk based on the submillimeter brightness using the standard formula : @xmath68 where @xmath69 is the measured excess at 500 ( we used this data point because the longest wavelength laboca measurement has very low signal - to - noise ratio ) , @xmath70 is the distance to the source , @xmath71 is the mass absorption coefficient , and @xmath54 is the planck function . by adopting @xmath722@xmath73g@xmath11 at @xmath74 ghz ( e.g. *
* ) , and taking @xmath1 and @xmath75 values from the previous sed modelling , we derived a dust mass of [email protected]@xmath76 for the disk .
[ [ spatial - extent . ] ] spatial extent .
+ + + + + + + + + + + + + + + in order to evaluate whether the disk is spatially extended its profiles measured at 70 , 100 , and 160 were compared with appropriate psfs of the pacs instrument .
the psf profiles were compiled using mini - scan map observations of four stars ( @xmath77 boo , @xmath77 tau , @xmath77 cet , @xmath1 and ) that do not exhibit infrared excesses and served as fiducial standards in the calibration of pacs photometry @xcite . for the data processing of these observations we used identical reduction steps as in the case of hd131835 , and finally the obtained psfs were rotated to match the roll angle of the telescope at the time of observing our target . by fitting a two dimensional gaussian function to the psf profiles we derive averaged fwhms of 570@xmath14002@xmath20559@xmath14003 and 686@xmath14004@xmath20676@xmath14003 at 70 and 100 , respectively , while for the target we obtained 649@xmath14020@xmath20587@xmath14019 and 774@xmath14034@xmath20719@xmath14032 , implying that the disk is slightly elongated and marginally extended along its major axis at these wavelengths . at the longest wavelength the target s profile was consistent with that of the psf measurements . in order to derive the characteristic size , inclination , and position angle of the disk ,
the 70 and 100 pacs images were fitted using a simple , non - physical disk model grid in the same way as described in @xcite . in this model
we assumed that the dust emitting at these wavelengths is located in a narrow outer ring around the central star .
the model has three free parameters , the average radius ( @xmath78 ) , the position angle ( @xmath79 ) and the inclination ( @xmath80 ) of the disk , while the width of the disk was fixed to 0.1@xmath78 following @xcite .
we used a bayesian analysis in the selection of the best fitting model .
our best solution has @xmath81au , @xmath828 , and @xmath83 at 70 and @xmath84au , @xmath8512 , and @xmath8624 at 100 .
the final disk parameters computed as a weighted average of those derived in the two bands are @xmath87au , @xmath887 , and @xmath89 .
it is worth noting that the measured profiles can also be fitted using a more extended disk model with a smaller inner radius and a larger outer radius .
such a model , would provide very similar @xmath79 and inclination parameters as above , and the average disk radius would also not change significantly . the disk around hd131835 has also been successfully resolved at 11.7 and 18.3 using the gemini south telescope @xcite deriving a @xmath79 of @xmath061 and an inclination of @xmath074 .
these parameters are broadly consistent with our results . by modeling the sed and the mid - ir images simultaneously , @xcite proposed a combination of an extended continuous power - law disk between 35 and 310au and two narrow rings at 105@xmath145 and 220@xmath1440au stellocentric distances as the best - fitting solution .
the three disk components are made of three different grain populations .
all of the emitting grains are proposed to be hotter than blackbodies .
the radius of the outer narrow ring ( 220@xmath1440au ) , that is predominantly responsible for the disk emission at wavelengths @xmath9050 , is broadly consistent with the characteristic disk radius derived from the marginally resolved pacs images .
observations of a single co isotopologue ( @xmath9co in our case ) can not provide information on the optical depth of the radiation .
assuming that the measured co emission of hd131835 is optically thin , the mass of co gas in the disk can be estimated as @xmath91 where @xmath92 is the mass of the co molecule , @xmath70 is the distance of the object , @xmath93 is the planck constant , @xmath94 and @xmath95 are the rest frequency and the einstein coefficient for the given transition between the @xmath96 upper and @xmath97 lower levels , @xmath98 is the observed integrated line flux , while @xmath99 is the fractional population of the upper level .
apart from @xmath99 all of the parameters from the right side of the equation are known . in case of local thermodynamic equilibrium ( lte ) , the level populations are thermalised and governed by the boltzmann equation , allowing the computation of @xmath99 if we know the gas temperature . in optically thin disks
the gas kinetic temperature can significantly differ from the temperature of dust grains ( see e.g. * ? ? ?
* ) thus the dust emission does not constrain the gas temperature . since in the 21 and 43 transitions we have upper limits for the co emission ( fig .
[ co32 ] ) , from the line ratios we can provide only a poor constrains for the excitation temperature , yielding @xmath1008k .
taking @xmath101k , the value measured in @xmath1pic @xcite we derived a total co gas mass of [email protected]@xmath2010@xmath66m@xmath76 from our @xmath7=32 line flux .
in lte @xmath102 and @xmath103 are equal . as fig .
[ gasmasses ] demonstrates the derived co mass is not very sensitive to the excitation temperature . for a range between 8 and 250k
, @xmath104 varies between 3.7@xmath2010@xmath66 and 2.3@xmath2010@xmath67m@xmath76 . in principle using similar assumptions we can derive upper limits both for and masses , however for these constituents especially for the estimates have stronger dependency on the excitation temperature ( see fig . [ gasmasses ] ) .
additionally , lte excitation of 63 line requires a dense medium , e.g. considering h@xmath3 molecules as collisional partners the critical density is typically @xmath05@xmath2010@xmath105cm@xmath67 @xcite . in a lower density environment
this line becomes subthermal with excitation temperatures far below what would be expected from lte . since these issues would make the mass estimate of unreliable we focused on , where the critical densities are lower ( e.g. @xmath03@xmath2010@xmath107cm@xmath67 , assuming hydrogen molecules as collisional partners , * ? ? ?
* ) . for @xmath1008k
, we obtained @xmath108m@xmath76 as an upper mass limit for .
] it is important to note , however , that our basic assumptions concerning the optically thin gas emission and the lte could be questioned .
for instance , in the disk around hd21997 not only the @xmath9co but even the @xmath109co lines turned out to be optically thick @xcite , while in debris disks with low amount of gas , radiative excitation can dominate over collisional processes leading to subthermal non - lte level populations @xcite . to partly consider these caveats and to further constrain the gas disk fundamental properties the measured line profile
was modeled with a simple disk geometry using the lime radiation transfer code @xcite .
we adopted a prescription for the structure of the disk which is frequently used to model passively irradiated protoplanetary disks .
the radial surface density distribution was taken to be a powerlaw .
we assumed that the disk is vertically isothermal and in hydrostatic equilibrium , resulting in a gaussian vertical density distribution , with radially dependent pressure scale height . assuming keplerian rotation around a 1.77m@xmath49 star and a radial temperature profile of r@xmath110
, the ratio of the pressure scale height to the radius has a profile of r@xmath111 . using these approximations
the disk can be described as : @xmath112},\ ] ] @xmath113 @xmath114 where @xmath115 is the outer disk radius , @xmath116 is the gas surface density at the outer disk radius in the midplane and @xmath117 is the scale height .
further parameters are the inner radius ( @xmath118 ) , the inclination ( @xmath80 ) of the disk , and the gas temperature @xmath119 . in the modelling we always assumed that the gas disk is
coplanar with the dust disk and therefore its inclination was fixed to 74 ( sect .
[ dustprops ] ) .
the origin of gas in this system is not clarified yet , it could either be primordial , or secondary . in the modelling
therefore we used two different scenarios corresponding to these alternative hypotheses . in our first model
we assumed a primordial gas disk where the co gas is accompanied with so much molecular hydrogen that the whole disk is in lte , i.e. the density of h@xmath3 exceeds the critical density everywhere and the rotational excitation is dominated by collisions . concerning the radial distribution first we examined whether a very narrow gas ring co - located with the inner or the outer narrow dust rings proposed by @xcite can reproduce the observed co ( 32 ) spectrum .
we found that none of these solutions are feasible because the obtained model spectra are significantly narrower than the measured one .
then we introduced a more extended gas disk model whose inner radius was fixed to 35au ( corresponding to the inner radius of the proposed continuous dust disk component in the model of * ? ? ?
* ) while the outer radius was varied between 80 and 310au .
the radial distribution of gas temperature was described as @xmath120 the temperature at the disk s inner edge , @xmath121 , was also a free parameter and varied between 10 and 100k .
we found that only models with @xmath115 between 80 and 150au provide spectral profiles consistent with the measured 32 line spectrum .
the co ( 32 ) line emission turned out to be at least partly optically thick in all models .
the gas temperature can not be strongly constrained , only @xmath122 could be excluded , since with this low temperature the obtained model spectra always found to be too faint . the lowest co mass , @xmath12310@xmath66m@xmath76 was obtained with @xmath124au , @xmath125cm@xmath26 and @xmath127k .
co spectra in the relevant transitions belong to this model were plotted in figure [ co32 ] . by adopting a canonical co / h@xmath3 abundance ratio of 10@xmath66 and using this minimum mass model we could check our original assumption on lte a posteriori .
we found that an overwhelming part ( more than 95% ) of co gas is located in disk regions where the density of h@xmath3 exceeds the critical density of co j=32 transition .
actually , by relaxing our assumption of lte and repeating the lime computations with the canonical co / h@xmath3 abundance ratio for this disk model we obtained co spectra consistent with the ones derived in the lte assumption .
these results support that lte is a reasonable assumption in our models . as an alternative scenario
, we also examined a disk whose gas material is produced from icy grains and planetesimals . in this case , the gas would mainly contain @xmath128 and co , similarly to comets in our solar system @xcite .
since the low dust content of the disk does not provide effective shielding against stellar and interstellar uv photons , the released molecules are quickly photodissociated .
photodissociation of water molecules most commonly produce oh radicals and h atoms , from which the former are then photodissociated into h and o atoms . by comparing the number of dissociating photons from the stellar photosphere model ( sect . [ stellarprops ] ) and from the interstellar radiation field @xcite , we concluded that the radiation field at the relevant wavelengths of @xmath41900 is dominated by stellar uv photons everywhere in the disk . using photodissociation cross sections from @xcite we found
that water molecules are very rapidly photodissociated , their lifetime is lower than 3.4days even at the outer edge of the disk ( @xmath0120au ) .
photodissociation of co molecules requires more energetic uv photons ( @xmath41118 , * ? ? ?
* ) . within @xmath045au ,
stellar radiation is the dominant source of these photons , out of this region the contribution of the interstellar radiation field is more important .
@xmath9co molecules can survive longer , their lifetime is estimated to be @xmath12940yr in the disk ( see sect .
[ gasprops ] ) .
the photodissociation of co results in o and c atoms .
a fraction of carbon atoms then could be ionized by the stellar or interstellar uv photons , yielding carbon ions and electrons . following the above mentioned considerations , we constructed a disk model containing well mixed co and gas . from the potential collisional partners we only took into account the electrons .
collisional rate coefficients for co@xmath130 were derived based on @xcite , while -@xmath130 coefficients were taken from @xcite .
no coefficients are available for co o , o or co , collisions . based on coefficients for hydrogen atoms from @xcite and @xcite
, we found that for @xmath131 200 the collisions with electrons dominate the excitation of both co and .
we adopted the same co gas density distribution as in the primordial model , @xmath118 , @xmath115 and @xmath80 were fixed to 35au , 120au , and 74 .
@xmath121 and the abundance ratio of ions to co molecules ( @xmath132 ) were free parameters .
since h and o are not ionized in regions not subject to extreme uv photons , and thereby carbon can be considered as the dominant source of free electrons , we adopted @xmath133 everywhere throughout the disk ( see also * ? ? ?
* ; * ? ? ?
the stellar parameters and the radial extent of the disk in the hd131835 system resemble those of @xmath1pic ( see sect .
[ gaseousatype ] ) .
therefore in our modelling we used the parameters of the @xmath1pic system as a benchmark .
based on recent observations with _ herschel _ and alma , the disk around @xmath1pic contains 5.5@xmath2010@xmath67m@xmath76 of , 7.1@xmath2010@xmath67m@xmath76 of @xcite and 2.85@xmath2010@xmath134m@xmath76 of co gas @xcite .
this corresponds to a to co abundance ratio of @xmath0450 , and even if we take into account the uncertainites of the mass estimates , the ratio is above 150 .
therefore we varied @xmath135 between 1 and 450 in our models .
we performed non - lte radiative transfer modelling .
in the course of modelling first we determined the co density that reproduces the observed co ( 32 ) line for a certain gas temperature and to co abundance ratio , and then we computed the model flux .
the resulting electron densities in the disk always exceeded the critical electron density for , leading to lte level populations . for abundance ratios of @xmath136 , only models with @xmath137k
were found to be consistent with the measured upper limit of .
because of the low gas temperatures , these models would require the presence of at least 5@xmath2010@xmath67m@xmath76 ( @xmath138cm@xmath26 ) of co gas for the reproduction of the observed co ( 32 ) line .
taking into account the ions and atoms , the total gas mass in these models is higher than 0.7m@xmath76 .
the energy of the upper level for the line is @xmath139k , thus , for a given amount of gas , the line emission becomes brighter in higher temperature models .
we found that in models with inner gas temperatures higher than 30k , the to co abundance ratio must be @xmath140 for producing a line fainter than our upper limit .
co gas is excited subthermally in most of the disk in these cases , therefore the reproduction of the observed co ( 32 ) line requires at least a co mass of 1@xmath2010@xmath67m@xmath76 ( @xmath141cm@xmath26 ) . in these models
the total gas mass of co , , and together is at the same level , or only slightly larger , than in the disk of @xmath1pic .
we investigated whether hd131835 shows any signatures of active accretion .
a possible excess in the balmer discontinuity can be used to estimate the rate of accretion in disks around herbig ae stars @xcite .
the excess parameter @xmath142 was calculated as @xmath143 where @xmath144 the intrinsic color was derived from the kurucz photospheric model of the source using synthetic photometry , while for computation of @xmath145 , the dereddened measured color index , we used u , b photometry from the catalogue of @xcite .
we obtained a @xmath146mag , in the computation of uncertainty we took into account both the measurement errors and the uncertainties in the kurucz photospheric model ( the uncertainties in the stellar parameters ) .
this calculation clearly shows that there is no excess in the balmer discontinuity . by calculating a 3-@xmath13 upper limit of @xmath00.1 for @xmath142 and following the outline described in @xcite we derived an upper limit of @xmath02@xmath2010@xmath147m@xmath49yr@xmath11 for the accretion rate . by inspecting the high - resolution optical spectra
we found that the possible accretional indicator lines of h@xmath148 , h@xmath149 , h@xmath150 , and hei 5876 are in absorption and are consistent with our fitted spectral model .
for the h@xmath148 line luminosity we derived an upper limit of 2@xmath2010@xmath66l@xmath49 . based on the calibration obtained for haebe stars by @xcite this upper limit corresponds to an accretion luminosity of @xmath151 .
considering that @xmath152 , from the absence of h@xmath148 line excess we obtained an upper limit of @xmath153m@xmath49 yr@xmath11 for the mass accretion .
for the non - detected sources co mass upper limits were estimated assuming optically thin emission and local thermodynamic equilibrium ( lte ) .
the excitation temperature was assumed to be 20k , the line flux upper limits were taken from table [ basictable ] . in those cases
where both co ( 32 ) and ( 43 ) line observations were available we used the lower transition in the calculations .
we note again that our basic assumptions might not be fulfilled for all of the studied systems . with the achieved sensitivity we can not exclude that some of our targets harbor optically thick gas disks . in very tenuous gas disks
the excitation could be subthermal even leading to a very low excitation temperature of @xmath410k ( e.g. * ? ? ?
in both of these cases our upper limits would underestimate the possible total co mass in the disk .
with the discovery of co gas in hd131835 the number of known gaseous debris disks around a - type stars is increased to seven .
tables [ gaseousdisks1 ] and [ gaseousdisks2 ] present the main stellar and disk properties .
all of these systems are likely younger than 50myr , thus they represent the very early phase of debris disk evolution .
hd131835 may be one of the youngest objects in this sample and based on their nearly identical stellar properties , it can be considered as a comparably young sibling of @xmath1pic .
figure [ sco100 ] shows the integrated co(32 ) fluxes and upper limits for these seven sources , normalized to 100pc , and plotted against fractional luminosities .
partly based on our two surveys , additional debris disks around main - sequence stars located within 125pc , as well as hd141569 , a nearby transitional disk that may represent a very final phase of protoplanetary disk evolution @xcite , are also displayed . although there is no clear trend with the fractional luminosity , apart from @xmath154tel all known gaseous debris disks have a fractional luminosity @xmath905@xmath2010@xmath66 . based on debris disk catalogues of @xcite , @xcite , and @xcite we identified eleven a - type stars within 125pc that harbor disks with @xmath155 / @xmath48 above 5@xmath2010@xmath66 .
all of these systems are younger than 50myr and nearly all of them have already been observed in co rotational transitions in different surveys ( the two exceptions are hd98363 and hd143675 ) . among the observed nine disks , four harbor co gas ( 49ceti , hd 21997 , @xmath1pic and hd131835 ) , while in the case of hd32297 and hd172555 atomic gas tracers were detected .
the gas component of hd95086 , hr4796 , and hd110058 have remained undetected so far .
it suggests that the most massive debris disks around young a - type stars quite commonly possess detectable amount of gas .
it is important to note that among these disks only hd21997 and @xmath1pic were observed with alma and the sensitivity of most of the other observations did not allow the detection of even a @xmath1pic - like gas disk . from the four co bearing debris disks
hd131835 is the brightest , its normalized line flux is @xmath017 times higher than that of the disk around @xmath1pic , but falls below the normalized line flux of hd141569 .
@xcite , @xcite ; hd32297
@xcite , @xcite ; hd172555
@xcite , @xcite ; hr4796
@xcite , @xcite ; hd110058
@xcite , @xcite ; hd141569
@xcite , @xcite ; hd131835 , @xmath154tel , hd95086 table [ basictable ] . for the rest of the objects the data are from @xcite .
disks with known gas component are denoted by larger symbols . _ [ sco100 ] ] within 125pc there are many f - type stars with age of @xmath450myr that possess debris disks with a fractional luminosity of @xmath905@xmath2010@xmath66 ( fig . [ sco100 ] ) , i.e. having similar properties than that of the abovementioned a - type sample .
we observed 15 such systems in the framework of our two surveys but none of them were detected , thus based on the current data sets the presence of co molecules may be the characteristic of debris disks around young a - type stars .
@xcite detected hi lines in mid - infrared spectrum of eight young debris disks around f - k type stars .
the presence of these lines can be explained either by low rate gas accretion onto these stars or chromospheric activity . in the case of former scenario
these findings would indicate that some disks around f - k type stars can also retain primordial gas for their debris phase .
based on its infrared and submillimeter continuum data the dust disk properties of hd131835 are similar to those of the other young debris disks .
although the disk has a high total fractional luminosity of @xmath03@xmath2010@xmath67 this value is still less than the usually invoked threshold value of 0.01 , that divides protoplanetary and debris disks .
recently , @xcite introduced a new classification scheme to differentiate between herbig ae disks and debris disks with a - type host stars .
this method is based on the ratios of the observed to the stellar photospheric fluxes at 12 and 70 ( @xmath156 and @xmath157 ) . for debris disks @xmath156 and @xmath157 have to be lower than 3 and 2000 , respectively . based on color - corrected wise 12 and pacs 70 data and predicted photospheric fluxes we derived @xmath158 and @xmath159 for hd131835 , i.e. it can be classified as a debris disk .
the majority of herbig ae systems show pah emission .
the absence of pah features in the irs spectrum of hd131835 is also in accordance with its classification as a debris disk .
these arguments suggest that the observed excess emission is due to dust grains that are probably second generation , produced via erosion of larger unseen bodies .
according to the current paradigm the gas content of debris disks may also be second generation , derived from previously formed planetesimals .
however , in young disks we can not exclude the possibility that the evolution of dust and gas were not parallel and the observed gas is predominantly composed of residual primordial material ( of course we never exclude that a part of the gas is produced from already emerged icy bodies ) .
indeed , though most known gaseous debris disks are proposed to be rather secondary , one of the oldest among them , hd21997 may likely harbor a hybrid disk with a significant amount of primordial gas @xcite . in the following ,
we investigate the origin of gas in the hd131835 system . in this analysis we considered the @xmath1pic and hd21997 systems as references :
the former may be the best established representantive of young debris disks with secondary gas @xcite , while the latter is the sole known example for a hybrid disk .
furthermore , both systems have already been observed with alma @xcite providing very detailed information on the spatial distribution of co gas and dust , and their host stars particularly @xmath1pic is quite similar to hd131835 . in the absence of significant h@xmath3 and dust
content in a debris disk with secondary gas , only self - shielding can protect co molecules againt the uv photons of stellar and interstellar radiation field . to judge the efficiency of self - shielding we took our secondary gas disk model with a co mass of 1.0@xmath2010@xmath67m@xmath76 ( i.e. the model with the minimum co mass ) and computed the shielding factors using the photodissociation model of @xcite .
the stellar uv flux found to be dominant at @xmath445au , were determined from the fitted atlas model ( sect .
[ stellarprops ] ) .
the contribution of interstellar radiation field were computed from @xcite .
our calculations yielded a co lifetime of 40yr at the directly illuminated inner edge of the disk and of 500yr in the most efficiently self - shielded mid - plane regions . assuming that the gas is continuously replenished and taking the co mass of 1.0@xmath2010@xmath67m@xmath76 we obtained a co production rate of @xmath129 2@xmath2010@xmath160m@xmath76yr@xmath11 or 1.2@xmath2010@xmath161 kg yr@xmath11 .
this is at least eight times higher than the gas production rate obtained for the @xmath1pic disk from its alma observation @xcite .
assuming a co mass abundance of 10% in planetesimals , it would require destruction at least of 2@xmath2010@xmath134m@xmath76yr@xmath11 of icy bodies .
the co gas can be released from icy grains / planetesimals in different processes .
the temperature significantly exceeds the value needed for the sublimation of pure co ice ( @xmath020k ) everywhere in the disk .
in such environment co ice is thought to be mainly present in deeper layers of icy planetesimals and as admixture in the amorphous water ice on the surface of grains and larger bodies .
thus photodesorption or , in warmer regions ( @xmath90 110k ) , sublimation of surface water ices can produce co gas as well .
co entrapped in water ice matrix can also be released via collisions between grains and planetesimals , moreover fragmentation of larger bodies can lead to the excavation of co ices persisted in deeper layers .
figure [ mcomdust ] shows the ratio of co mass to the dust mass as a function of age for debris disks with known gaseous component and for some selected disks ( additional debris disks around a - type stars with @xmath162 ; fomalhaut and hd107146 , two debris systems recently observed by alma ; and hd141569 ) .
based on alma observations of @xcite the @xmath163 ratio measured for @xmath1pic disk is 3.6@xmath164 ( note that they used an assumption of lte , and if it is not fulfilled the co mass could be higher , see also in * ? ? ?
the spatially resolved images also revealed a clump at @xmath085au from the star , in which an enhanced co - to - dust mass ratio was measured .
sensitive alma observations are available only for a few debris disks .
upper limits for @xmath163 at fomalhaut @xcite and at hd107146 @xcite , however , hint that the ratio measured at @xmath1pic is already quite high .
interestingly the other three debris disks with co gas show even higher @xmath163 ratios .
for hd21997 we measured a @xmath01000 times higher @xmath163 ratio ( note that here we used the optically thin @xmath9c@xmath165o line to estimate the co mass ) than for the @xmath1 pic .
this object clearly differs from the others and likely harbors a significant fraction of primordial gas .
the lower limits of co - to - dust mass ratios for hd131835 ( in calculation of its @xmath163 we used the minimum co mass of 5.2@xmath2010@xmath66m@xmath76 that we obtained in our models ) and 49ceti are @xmath1292.5@xmath20 higher than that of @xmath1pic . in a secondary disk
the @xmath163 ratio depends both on the dust and gas production rates and on the removal time of the two elements .
the observed higher co - to - dust mass ratios thus can be explained e.g. by a more effective self - shielding due to larger co density which would result in higher co lifetime or by higher fraction of volatile material in icy planetesimals / grains . and [ gaseousdisks2 ] .
hd32297 was placed at its lower age limit proposed by @xcite . for fomalhaut and hd107146
we used data from the literature ( * ? ? ? * ; * ? ? ? * and references therein ) , in the latter case the co upper mass estimate derived from the secondary disk scenario were taken into account . in the case of hd141569 and hr4796
the co mass and upper limit were computed by adopting an excitation temperature of 20k and using the co ( 32 ) line fluxes from @xcite .
their dust masses were computed from their submillimeter fluxes @xcite using the standard way .
age estimate of hd141569 was taken from @xcite . as a member of the twhya association , for hr4796 we applied the age of the group @xcite .
the data of hd95086 are from table [ basictable ] and @xcite .
the typical @xmath166 ratio for protoplanetary disks ( horizontal line ) was calculated by assuming a total gas to dust mass ratio of 100 and taking a canonical co to h@xmath3 abundance ratio of 10@xmath66 . _
[ mcomdust ] ] stellar and interstellar uv photons can permeate debris disks without any hindrance and ionize carbon atoms formed via photodissociation of co molecules .
the fraction of neutral and ionized carbon depends on the strength of the local radiation field . due to their longer lifetime in such environment ,
the amount of and can significantly exceed that of co molecules , e.g. in the disk of @xmath1pic there is approximately 200 times more gas than co gas in terms of mass @xcite .
figure [ cii ] displays the @xmath167/@xmath168 line flux ratios for debris disks where cii and/or co gas has been detected .
the ratio of @xmath1pic is about 30 times higher than the upper limit we obtained for hd131835 .
this may reflect a real difference in the mass ratio , or as our modelling showed ( sect .
[ gasparams ] ) may be a consequence of the low gas temperature in hd 131835 . in the latter models the total gas mass and
the gas - to - dust mass ratio would be rather high .
/s@xmath169 ratio as a function of total co mass for gaseous debris disks ( apart from hd172555 where neither of co , nor were detected ) and hd141569 .
line flux data for hd131835 and hd21997 were taken from sect [ herschelobs ] and from juhsz et al .
( in prep . ) for the other targets s@xmath170 values were from 49ceti @xcite ;
@xmath1pic @xcite ; hd32297 @xcite ; @xmath154tel @xcite ; hd141569 @xcite .
references for the s@xmath169 data are summarized in fig .
[ sco100 ] caption .
_ [ cii ] ] in the secondary gas scenario , gas and dust are released at the same location from planetesimals , thus the two components are expected to be co - located , just as we see at @xmath1pic . based on our current data and modelling ( sect .
[ gasparams ] ) we can not determine the accurate location of gas , we just can say that the results do no exclude that the two components are co - located .
it is worth noting , however that our gas disk model coincides with that part of the dust disk model of @xcite , which is composed by very hot grains .
these grains give only a low fraction of the dust disk mass and it is questionable whether they are accompanied with large amount planetesimals that could be the source of observed co gas .
all in all , in the secondary scenario , hd131835 could be considered as of comparable age and in terms of its disk more massive analogue of the @xmath1pic system . however
, none of our current data exclude that hd131835 in fact possess a hybrid disk .
our modelling in sect .
[ gasparams ] shows that the co ( 32 ) emission may be optically thick , thus we can not exclude that the co mass is significantly higher reaching even the same level as in the case of hd21997 . the low @xmath167/@xmath168 flux ratio can be explained in the secondary scenario , but as figure [ cii ] shows , among the currently known gaseous debris disks only hd21997 , the sole hybrid disk , exhibits similarly low ratio . for a better clarification of the origin of gas additional observations
are necessary . by measuring the emission of less abundant @xmath109co and c@xmath165o isotopologues from the disk
a more reliable co gas mass estimate could be obtained . moreover , by mapping and comparing the spatial distribution of gas and dust allow us to investigate whether the two components are co - located , which is a pre - requisite in a secondary gas scenario .
these observations requires better sensitivity and spatial resolution than that of a single dish radio telescope but could definitely be carried out using the atacama large millimeter / submillimeter array .
by searching for co gas in 20 debris disks using the apex and iram30 m radiotelescopes we identified a new gas rich system around the @xmath016myr old ucl member , hd131835 , where the 32 transition of co was successfully detected .
based on spectroscopic and photometric data the stellar properties of hd131835 resembles well those of @xmath1pic . by observing hd131835 with the _ herschel space observatory
_ we found that the disk is spatially resolved both at 70 and 100 and the disk characteristic radius is @xmath0170au .
thanks to our observations at 250 , 350 , and 500 the submillimeter sed of the source is now significantly better characterized .
based on its infrared and submillimeter continuum data the dust disk properties of hd131835 are similar to those of the most massive young debris disks . with the detection of gas in hd131835
the number of of known debris disks with co component is increased to four ( 49ceti , hd 21997 , @xmath1pic and hd131835 ) .
all of these disks encircle young ( @xmath240myr ) a - type stars . within 125pc we know 11 a - type stars whose debris disks fractional luminosity exceed @xmath17110@xmath66 . among these disks , 9 have already observed in co rotational transitions .
four of them harbor co gas , while in two other objects atomic gas tracers were detected .
this detection rate of 4/9 ( or 6/9 if atomic gas detection was also taken into account ) suggests that the most massive debris disks around young a - type stars commonly possess detectable amount of gas . based on our current data
we can not draw a secure conclusion on the origin of gas in hd131835 .
if the gas is secondary then hd131835 could be considered as a comparably young and in terms of its disk more massive analogue of the @xmath1pic system .
however , we can not exclude that this system similarly to hd21997 possess a hybrid disk , where the gas material is predominantly primordial , while the dust grains are mostly derived from planetesimals .
we thank the anonymous referee for useful comments that helped us to improve the manuscript .
this work was supported by the momentum grant of the mta csfk lendlet disk research group , the pecs-98073 program of the european space agency ( esa ) and the hungarian research fund otka grants k101393 and k104607 .
a.m. acknowledges support from the bolyai research fellowship of the hungarian academy of sciences .
acknowledges support under the nasa origins of solar system program on nng13pb64p .
a.j . acknowledges the support of the discsim project , grant agreement 341137 funded by the european research council under erc-2013-adg .
this publication makes use of data products from the wide - field infrared survey explorer , which is a joint project of the university of california , los angeles , and the jet propulsion laboratory / california institute of technology , funded by the national aeronautics and space administration .
the publication also 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 .
lequeux , j. 2005 , the interstellar medium , translation from the french language edition of : le milieu interstellaire by james lequeux , edp sciences , 2003 edited by j. lequeux .
astronomy and astrophysics library , berlin : springer , 2005 , m. a. c. 1997 , in esa special publication , vol .
402 , hipparcos - venice 97 , ed .
r. m. bonnet , e. hg , p. l. bernacca , l. emiliani , a. blaauw , c. turon , j. kovalevsky , l. lindegren , h. hassan , m. bouffard , b. strim , d. heger , m. a. c. perryman , & l. woltjer , 14 lcccccccccc + hd 16743 & f0/f2iii & 58.9 & 30 ( 2 ) & - & 5.1@xmath2010@xmath66 ( 3 ) & 2 & & @xmath41.24 & & @xmath44.6@xmath172 + hd 38206 & a0v & 75.1 & 30 & col & 1.4@xmath2010@xmath66 ( 1 ) & 2 & & @xmath42.40 & & @xmath41.4@xmath164 + hd 95086 & a8iii & 90.4 & 17 & lcc & 1.7@xmath2010@xmath67 ( 3 ) & 2 & &
@xmath41.95 & & @xmath41.7@xmath164 + hd 121191 & a5iv / v & 130.0 & 17 & lcc & 4.9@xmath2010@xmath67 ( 2 ) & 4 & & @xmath41.47 & & @xmath42.6@xmath164 + hd 131488 & a1v & 150.0 & 16 & ucl & 6.0@xmath2010@xmath67 ( 2 ) & 4 & & @xmath41.39 & & @xmath43.3@xmath164 + hd 131835 & a2iv & 122.7 & 16 & ucl & 3.0@xmath2010@xmath67 ( 5 ) & 1,2,3 & [email protected] & [email protected] & [email protected] & @xmath905.2@xmath164 + hd 181296 & a0v & 48.2 & 23 & bpmg & 2.4@xmath2010@xmath66 ( 1 ) & 4 & & @xmath41.38 & @xmath45.36 & @xmath43.4@xmath172 + + hd 10939 & a1v & 62.0 & 346 ( 4 ) & - & 1.2@xmath2010@xmath66 ( 3 ) & 2 & & @xmath41.37 & & @xmath45.6@xmath172 + hd 17848 & a2v & 50.5 & 372 ( 4 ) & - & 7.6@xmath2010@xmath134 ( 3 ) & 2 &
& @xmath41.42 & & @xmath43.8@xmath172 + hd 161868 & a0v & 31.5 & 450 ( 3 ) & - & 1.3@xmath2010@xmath66 ( 3 ) & 2 & & @xmath41.82 & & @xmath41.9@xmath172 + hd 182681 & b8/b9v & 69.9 & 144 ( 4 ) & - & 2.8@xmath2010@xmath66 ( 3 ) & 2 & & @xmath42.15 & & @xmath41.1@xmath164 + + hd 105 & g0v & 39.4 & 30 & tha & 1.9@xmath2010@xmath66 ( 1 ) & 3 &
& @xmath41.42 & @xmath46.08 & @xmath42.4@xmath172 + hd 377 & g2v & 39.1 & 183 ( 5 ) & - & 2.6@xmath2010@xmath66(1 ) & 3 &
& @xmath41.03 & @xmath42.41 & @xmath41.7@xmath172 + hd 61005 & g3/g5v & 35.3 & 40 & arg & 1.6@xmath2010@xmath67 ( 1 ) & 3 & & @xmath41.27 & @xmath44.37 & @xmath41.7@xmath172 + hd 92945 & k1v & 21.4 & 211 ( 5 ) & - & 4.9@xmath2010@xmath66(1 ) & 3 & & @xmath41.33 & @xmath43.86 & @xmath40.7@xmath172 + hd 160305 & f8/g0v & 72.5 & 23 & bpmg & 8.6@xmath2010@xmath134 ( 4 ) & 3 & & @xmath41.61 & @xmath45.54 & @xmath49.0@xmath172 + hd 202917 & g5v & 43.0 & 30 & tha & 2.8@xmath2010@xmath66 ( 1 ) & 3 & & @xmath41.46 & @xmath47.22 & @xmath42.9@xmath172 + + hd 110411 & a0v & 36.3 & 90 ( 4 ) & - & 5.9@xmath2010@xmath134 ( 1 ) & 5 & @xmath40.20 & & & @xmath40.6@xmath172 + hd 182919 & a0v & 72.9 & 198 ( 1 ) & - & 3.4@xmath2010@xmath134 ( 1 ) & 5 & @xmath40.16 & & & @xmath42.0@xmath172 + hd 183324 & a0v & 61.2 & 140 ( 1 ) & - & 2.2@xmath2010@xmath134 ( 1 ) & 5 & @xmath40.16 & & & @xmath41.4@xmath172 + lcccc e-083.c-0303 & apex & shefi / apex2 & ffts & 2009 - 05 - 18 + m-087.f-0001 - 2011 & apex & shefi / apex2 & ffts , xffts & 2011 - 04 - 142011 - 12 - 16 + & apex & shefi / apex1 & xffts & + m-092.f-0012 - 2013 & apex & flash+ & xffts & 2013 - 09 - 292013 - 10 - 17 + m-093.f-0010 - 2014 & apex & flash+ & xffts & 2014 - 04 - 232014 - 04 - 27 + iram-172 - 13 & iram & emir & fts & 2014 - 01 - 08 + ccccc 3.35 & [email protected] & wise & 302.9 & @xcite + 4.60 & [email protected] & wise & 168.1 & @xcite + 6.00 & [email protected] & irs & 101.6 & this work + 7.00 & [email protected] & irs & 75.5 & this work + 7.97 & [email protected] & irs & 58.7 & this work + 9.00 & [email protected] & irs & 46.4 & this work + 9.00 & [email protected] & irc & 46.4 & @xcite + 10.00 & [email protected] & irs & 37.8 & this work + 10.99 & [email protected] & irs & 31.3 & this work + 11.56 & [email protected] & wise & 28.4 & @xcite + 13.05 & [email protected] & irs & 22.3 & this work + 14.83 & [email protected] & irs & 17.3 & this work + 16.99 & [email protected] & irs & 13.3 & this work + 19.02 & [email protected] & irs & 10.6 & this work + 21.28 & [email protected] & irs & 8.5 & this work + 22.09 & [email protected] & wise & 7.9 & @xcite + 23.67 & [email protected] & mips & 6.8 & @xcite + 24.48 & [email protected] & irs & 6.4 & this work + 25.00 & [email protected] & iras & 6.1 & @xcite + 27.45 & [email protected] & irs & 5.1 & this work + 30.50 & [email protected] & irs & 4.1 & this work + 33.55 & [email protected] & irs & 3.4 & this work + 60.00 & [email protected] & iras & 1.1 & @xcite + 70.00@xmath173 & [email protected] & pacs & 0.8 & this work + 71.42 & [email protected] & mips & 0.7 & @xcite + 100.00@xmath173 & [email protected] & pacs & 0.37 & this work + 160.00 & [email protected] & pacs & 0.14 & this work + 250.00 & [email protected] & spire & 0.06 & this work + 350.00 & [email protected] & spire & 0.03 & this work + 500.00 & [email protected] & spire & 0.01 & this work + 870.00 & [email protected] & laboca & 0.005 & @xcite + lcccccc 49ceti & 59.4 & 8900 ( 5 ) & 16.4 ( 5 ) & 2.00 ( 5 ) & 40@xmath174 ( 3,11 ) & arg + hd21997 & 71.9 & 8300 ( 5 ) & 11.2 ( 5 ) & 1.85 ( 5 ) & 30@xmath174 ( 3,11 ) & col + @xmath1pic&19.44 & 8200 ( 2 ) & 8.7 ( 2 ) & 1.75 ( 2 ) & 23@xmath175 ( 4 ) & bpmg + hd131835 & 122.7 & 8250 ( 10 ) & 9.2 ( 10 ) & 1.77 ( 10 ) & 16@xmath176 ( 7 ) & ucl + hd32297 & 112.4 & 8000 ( 9 ) & 6.2 ( 9 ) & 1.65 ( 9 ) & 15500 ( 9 ) & - + hd172555 & 28.5 & 7800 ( 7 ) & 7.8 ( 7 ) & 1.68 ( 7 ) & 23@xmath175 ( 4 ) & bpmg + @xmath154tel & 48.2 & 9500 ( 1 ) & 20.9 ( 1 ) & 2.20 ( 1 ) & 23@xmath175 ( 4 ) & bpmg + lcccccc 49ceti & 1.1@xmath2010@xmath67 ( 14 ) & [email protected]@xmath2010@xmath11 ( 14 ) & y / y / y / y ( 22,19,20 ) & @xmath902.5@xmath2010@xmath66 ( 4 ) & y ( 12 ) & secondary ( 19 ) + hd21997 & 5.7@xmath2010@xmath66 ( 14 ) & [email protected]@xmath2010@xmath11 ( 14 ) & y / n / n / n ( 13 ) & 48@xmath2010@xmath26 ( 10 ) & n & primordial ( 10 ) + @xmath1pic & 2.6@xmath2010@xmath67 ( 15 ) & [email protected]@xmath2010@xmath26 ( 5 ) & y / y / y / y ( 5,18,2 ) & 2.9@xmath2010@xmath134 ( 5 ) & y ( 1 ) & secondary ( 7,5 ) + hd131835 & 3.0@xmath2010@xmath67 ( 21 ) & [email protected]@xmath2010@xmath11 ( 21 ) & y / n / n / n ( 21 ) & @xmath905.2@xmath2010@xmath66 ( 21 ) & n & ? + hd32297 & 4.4@xmath2010@xmath67 ( 3 ) & [email protected]@xmath2010@xmath11 ( 11 ) & n / n / y / n ( 6 ) & @xmath42.2@xmath2010@xmath66 ( 13 ) & n & secondary ( 6 ) + hd172555 & 7.3@xmath2010@xmath66 ( 16 ) & [email protected]@xmath2010@xmath66 ( 16 ) & n / n / n / y ( 16 ) & @xmath42.4@xmath2010@xmath134 ( 13 ) & y ( 9 ) & secondary ( 8,16 ) + @xmath154tel & 2.4@xmath2010@xmath66 ( 3 ) & [email protected]@xmath2010@xmath26 ( 17 ) & n / n / y
/ n ( 17 ) & @xmath46.8@xmath2010@xmath134 ( 21 ) & n & secondary ( 17 ) + | debris disks are considered to be gas - poor , but recent observations revealed molecular or atomic gas in several 1040myr old systems .
we used the apex and iram30 m radiotelescopes to search for co gas in 20 bright debris disks . in one case , around the 16myr old a - type star hd131835
, we discovered a new gas - bearing debris disk , where the co 32 transition was successfully detected .
no other individual system exhibited a measurable co signal .
our _ herschel space observatory _ far - infrared images of hd131835 marginally resolved the disk both at 70 and 100 , with a characteristic radius of @xmath0170au .
while in stellar properties hd131835 resembles @xmath1pic , its dust disk properties are similar to those of the most massive young debris disks . with the detection of gas in hd131835
the number of known debris disks with co content has increased to four , all of them encircling young ( @xmath240 myr ) a - type stars . based on statistics within 125pc
, we suggest that the presence of detectable amount of gas in the most massive debris disks around young a - type stars is a common phenomenon .
our current data can not conclude on the origin of gas in hd131835 .
if the gas is secondary , arising from the disruption of planetesimals , then hd131835 is a comparably young and in terms of its disk more massive analogue of the @xmath1pic system .
however , it is also possible that this system similarly to hd21997 possesses a hybrid disk , where the gas material is predominantly primordial , while the dust grains are mostly derived from planetesimals . |
thermal inflation @xcite provides the most compelling solution to the moduli ( polonyi ) problem @xcite
. however , for a theory of the early universe to be viable it must be capable of producing a baryon asymmetry @xcite @xmath6 by the time of nucleosynthesis .
thermal inflation probably dilutes any pre - existing baryon asymmetry to negligible amounts , and the final reheat temperature after thermal inflation ( @xmath7 few gev ) is probably too low even for electroweak baryogenesis .
thus if thermal inflation really is the solution of the moduli problem , then it is likely also to be responsible for baryogenesis . in section [ ti ]
we explain why the flaton that gives rise to thermal inflation probably also generates the masses of right - handed neutrinos as well as the @xmath1-term of the minimal supersymmetric standard model ( mssm ) .
we also note the various ways in which a potential domain wall problem can be avoided . in section [ ad ]
we describe how a somewhat complicated affleck - dine type mechanism can naturally generate the required baryon asymmetry after thermal inflation . in section [ con ]
we give our conclusions .
the superpotential of the mssm is @xcite @xmath8 thermal inflation @xcite requires that there is in addition at least one flaton @xmath9 with vacuum expectation value @xmath10 in the range @xmath11 , the lower bound coming from the requirement that thermal inflation sufficiently dilutes the moduli , and the upper bound from the requirement that the final reheat temperature after thermal inflation @xmath12 be high enough to thermalise the lightest supersymmetric particles ( lsp s ) produced in the flaton s decay and so avoid an excess of lsp s . also , in order for @xmath9 to be held sufficiently strongly in eq .
( 35 ) of ref .
@xcite . ] at @xmath13 by the finite temperature during thermal inflation , @xmath9 must have unsuppressed couplings to at least one other field , say @xmath14 , that is light when @xmath13 .
we therefore either require a term @xmath15 with @xmath16 in the superpotential , or require @xmath9 to spontaneously break a continuous gauge symmetry with gauge coupling @xmath17 ( with @xmath14 being the gauge field in this case ) .
one reason to prefer the yukawa coupling over the gauge coupling is that the renormalisation group @xcite effect of the yukawa coupling would be to drive the soft supersymmetry breaking mass squared of @xmath9 negative at low energies as is required for a flaton , while the gauge coupling would have the opposite effect .
another reason is that the gauge symmetry has no independent motivation , while , as we shall see , the yukawa coupling is very well motivated .
we will therefore focus on the case of the yukawa coupling .
after @xmath9 acquires its vacuum expectation value @xmath18 , @xmath14 will acquire a mass @xmath19 to @xmath20 and so is not a mssm field . in order for @xmath9 to be coupled to the thermal bath , which is presumably composed of mssm fields , we therefore require @xmath14 to couple to the mssm .
we therefore is not charged with respect to the mssm continuous gauge symmetries as this would , in general , destroy susy gut gauge coupling unification .
however , if @xmath14 is a complete @xmath21 multiplet the unification of the gauge couplings will be unaffected .
for example , one could have @xmath22 .
alternatively , appropriate choices of representations could shift the unification scale to the string scale @xcite . ]
require at least one s , one charged and one neutral under r - parity .
] of the terms @xmath23 or @xmath24 in the superpotential , as these are the only possible renormalisable couplings of a singlet to the mssm .
the former is the standard coupling @xmath25 of a right - handed neutrino @xmath26 to the mssm , and furthermore @xmath27 automatically acquires a mass @xmath28 to @xmath20 in the right range for the seesaw mechanism @xcite to generate a left - handed tau neutrino mass @xmath29 suitable for the mixed dark matter scenario for the formation of the large scale structure of the universe @xcite .
the latter coupling @xmath24 does not have any good independent motivation , apart from the fact that it is possible .
we will therefore focus on the former case .
later though , we shall make use of this possibility for coupling to the mssm in a different context . now that we have a more precise picture of the couplings of the flaton , @xmath30 we can hope to make a more precise estimate of the decay rate of the flaton and so the final reheat temperature after thermal inflation @xmath12 .
we first note that the effective superpotential coupling @xmath31 obtained by integrating out @xmath27 , eliminating @xmath27 via the constraint @xmath32 , will give a decay rate of order @xmath33 which is negligible .
the @xmath9 dependence of the low energy renormalised coupling constants will give larger decay rates . to estimate these we first need to know the contributions of the right - handed neutrinos to the renormalisation group equations .
they are @xcite @xmath34 where @xmath35 is the logarithm of the renormalistion scale , the @xmath36 s are the soft supersymmetry breaking masses , the @xmath37 s are the soft supersymmetry breaking parameters in the terms in the scalar potential of the form @xmath38 , the @xmath36 s and the magnitudes of the @xmath37 s are of the order of the soft supersymmetry breaking scale @xmath39 to @xmath40 , and the stand for other terms independent of the right - handed neutrinos .
@xmath41 sets the threshold for the right - handed neutrinos , and so writing @xmath42 we get the effective couplings @xmath43 and @xmath44 from these couplings we estimate the total decay rate to be @xmath45 this would give a final reheat temperature after thermal inflation of @xmath46 where the first bracket is constrained to be of order one , or perhaps less , by eq .
( [ mnu ] )
. however , in order not to over - produce lsp s we require @xcite @xmath47 thus our model seems to be in trouble unless we can add some extra coupling that gives a stronger decay rate .
the only possibility is to couple @xmath9 to @xmath48 in the superpotential .
a term @xmath49 would require a very small coupling constant to avoid generating too large a @xmath1-term , but a term would be an alternative if @xmath50 .
we assume the displayed case for simplicity . ]
@xmath51 is not only allowed but could naturally generate a @xmath1-term @xmath52 of the required size @xcite .
for example , for @xmath53 and @xmath54 we get @xmath55 . from now on we will assume that the @xmath1-term is generated in this way so that @xmath56 , or at least @xmath57 , and @xmath58 . writing @xmath59 , where @xmath60 and @xmath61 , we get the relatively unsuppressed couplings @xmath62 where a tilde denotes the fermionic component of the superfield , a bar denotes the hermitian conjugate , and @xmath63 is the soft supersymmetry breaking parameter in @xmath64 and so @xmath65 . the decay rate is then estimated to be @xmath66 where we have roughly assumed @xmath67 .
we therefore get a reheat temperature @xmath68 which is sufficiently high .
we thus expect the flaton to have the following couplings to the mssm is a possible alternative to @xmath69 . ]
@xmath70 in this section we consider the self - couplings of the flaton .
the flaton ( or , in the case of a multi - component flaton , at least one component of the flaton ) should have a negative soft supersymmetry breaking mass squared @xmath71 to drive it away from @xmath13 after thermal inflation .
it will also need a term to stabilise its potential at its vacuum expectation value @xmath72 to @xmath73 .
the simplest possibility is would be an alternative if @xmath50 .
again we assume the displayed case for simplicity .
one might even be able to use the renormalisation group running of @xmath74 to stabilise the potential but then one would not automatically get a value for @xmath18 in the correct range . ]
@xmath75 and this is what we shall assume . in the case of a multi - component flaton this could be interpreted as for example @xmath76 , @xmath77 or a sum of such terms .
see ref .
@xcite for an explicit multi - component example . here , for simplicity
, we will focus on the case of a single component flaton , though it should be born in mind that a multi - component flaton might be preferable from the model building point of view .
we then get the following scalar potential @xmath78 where @xmath79 .
this potential has four degenerate minima with @xmath10 , where @xmath80 for example , for @xmath81 and @xmath82 , we get @xmath83 .
the eigenvalues of the mass squared matrix at the minima are @xmath84 and @xmath85\ ] ] and the flatino mass squared is @xmath86 requiring zero cosmological constant at the minima gives @xmath87\ ] ] with four degenerate minima we clearly have to worry about a potential domain wall problem . the simplest way to eliminate the domain walls is to add a small term which breaks the degeneracy of the vacua , the difference in pressure exerted on the walls causing the domains with greater vacuum energy to collapse .
they collapse before the domain walls come to dominate the energy density if the difference in vacuum energies @xmath88 satisfies @xmath89 where @xmath90 is the energy per unit area of the domain walls @xcite . for flaton domain walls @xmath91 and
so we require @xmath92 therefore a term in the superpotential of the form @xmath93 with @xmath94 odd would be sufficient to eliminate the domain walls
. a term with @xmath95 , for example @xmath96 , would reduce the @xmath97 domain walls to @xmath98 domain walls .
note that @xmath99 can be extremely small , and hence have a negligible effect on the dynamics to be discussed in the next section , but still solve the domain wall problem .
another way to avoid a domain wall problem is to gauge the discrete symmetry so that there is really only one vacuum . however , non - trivial anomaly cancellation conditions must be satisfied @xcite . in the case of a single - component flaton with the superpotential of eq .
( [ wti ] ) , and no extra light su(3 ) multiplets , the mixed discrete - su(3 ) anomaly cancellation condition requires @xmath100 to be neutral under any unbroken by hidden sector supersymmetry breaking . ]
anomaly free discrete gauge symmetry .
we therefore can not use a discrete gauge symmetry to remove the @xmath97 domain walls .
however , the anomaly free @xmath97 subgroup of @xmath101 , under which @xmath9 has charge 2 , can be used to gauge away the @xmath98 domain walls left by @xmath96 above .
furthermore , this symmetry is broken down to the standard matter parity of the mssm ( which is equivalent to the r - parity of the mssm ) by the vacuum expectation value of @xmath9 .
in the case of a multi - component flaton the discrete symmetry can be extended to a continuous symmetry which may or may not be gauged . in the case of a continuous global symmetry , for example the peccei - quinn symmetry of ref .
@xcite , the goldstone bosons may prove troublesome @xcite .
one also has more freedom to satisfy the anomaly cancellation conditions in the case of a multi - component flaton .
an extension of the mssm that gives rise to viable thermal inflation , and so does not suffer from a moduli problem , should have the following terms in its superpotential @xmath102
to orientate the reader we will first sketch the basic idea we have in mind before plunging into the details . the @xmath5-flat direction parametrised by @xmath4 provides an ideal affleck - dine field @xcite . in order for it to behave as an affleck - dine field
we must first get it away from zero .
we therefore require its mass squared at the end of thermal inflation , @xmath103 , to be negative .
it is simplest to assume that it rolls away from zero before @xmath9 does .
however , when @xmath13 the right - handed neutrinos are light , and so @xmath4 is not @xmath104-flat ( it has a quartic term in its potential coming from the superpotential ) .
@xmath4 will thus be stabilised at a modest value .
next @xmath9 will roll away from zero .
the right - handed neutrinos become heavy and so can be integrated out leaving the effective seesaw coupling @xmath105 given in eq .
( [ wseesaw ] ) .
this is now the term that stabilises the @xmath4 direction and we see that it gets smaller , and so the @xmath4 direction gets flatter , as @xmath9 gets larger .
thus , as @xmath9 rolls away from zero , @xmath4 will roll further away from zero .
furthermore , the soft supersymmetry breaking term derived from @xmath105 will correlate the phases of @xmath9 and @xmath4 .
when @xmath9 becomes sufficiently large , it will start to feel the basin of attraction of one of the minima of its potential , and so will start curving in towards that minimum , its phase will be roughly determined modulo @xmath106
. the phase of @xmath4 will then be roughly determined modulo @xmath107 by the soft supersymmetry breaking term derived from @xmath105 .
when @xmath41 becomes of order @xmath18 , a cross term from the supersymmetric part of the potential becomes significant and changes the correlation between the phases of @xmath9 and @xmath4 , and so gives the phase of @xmath4 a kick .
the direction of the kick is determined by the parameters in the lagrangian ( this is our @xmath108 violation ) and so gives a non - zero net contribution when averaged over different spatial locations , unlike the rest of the angular momentum that is flying around .
furthermore , as this is happening @xmath109 ( see eq .
( [ wdecay ] ) ) starts to give a significant contribution to the mass of @xmath110 and hence @xmath4 .
for @xmath111 this gives the @xmath4 direction an overall positive mass squared ( as it must because @xmath4 has a positive mass squared in the true vacuum ) , and so sends @xmath4 spiralling back in towards zero .
the effective friction on the motion of @xmath9 and @xmath4 coming from the hubble expansion is negligible .
however , the effective mass squareds of both @xmath9 and @xmath4 have been changing sign during the above dynamics and so one would expect them both to decay via broad parametric resonance @xcite .
this will lead to approximately critical damping , and so it seems reasonable to expect that both @xmath9 and @xmath4 will be trapped near their vacuum expectation values essentially immediately after the dynamics described above has occurred .
once they are trapped , parametric resonance becomes less efficient because the mass squareds are now always positive .
@xmath4 s potential near @xmath112 conserves angular momentum , or in other words lepton number , and so @xmath4 s newly acquired lepton number is conserved .
the dynamics outlined above is illustrated in figure 1 .
the decay of the @xmath4 affleck - dine condensate will generate enough partial reheating to restore the electroweak symmetry , and so its lepton number can be converted to baryon number by the usual electroweak effects @xcite .
note that the energy density is still dominated by the flaton and the reheating in the affleck - dine sector has a negligible effect on the now decoupled flaton .
finally , after the temperature has dropped to a few gev , the flaton decay will complete , releasing substantial entropy .
our basic model is @xmath113 the squark fields have no linear terms in their potential and have positive mass squareds .
they will therefore be held at zero apart from thermal fluctuations , and so can be ignored apart from their contribution to the finite temperature effective potential .
the zero temperature potential for the other fields is @xmath114 where the @xmath36 s and the magnitudes of the @xmath37 s are of order @xmath115 .
we assume @xmath116 so that the @xmath5-flat direction parametrised by @xmath4 is also unstable , in addition to the flaton @xmath9 .
note that after @xmath9 acquires its vacuum expectation value @xmath117 , it will give an extra contribution @xmath118 to @xmath110 s mass squared .
this will be of order @xmath119 if @xmath120 .
we assume @xmath121 so that the @xmath4 direction is stable in the true vacuum .
a rigorous study of the dynamics of this model is beyond the scope of this paper .
instead we will make some simplifying assumptions in order to illustrate how the affleck - dine mechanism might be implemented after thermal inflation and to crudely estimate the resultant baryon asymmetry .
we assume that all fields are initially held at zero by the finite temperature during thermal inflation .
we assume that the @xmath4 direction rolls away from zero first .
it will be quickly stabilised by the term @xmath122 at a value @xmath123 .
the term @xmath124 then causes @xmath27 to roll away from zero .
then the term @xmath125 causes @xmath9 to roll away from zero , instead of @xmath4 .
however , unlike @xmath4 , if it was unstable it would roll away from zero at some early time because all its couplings would be small .
the term @xmath126 would then cause @xmath9 to roll away from zero causing a premature end to thermal inflation . ] in the direction @xmath127\ ] ] for simplicity we will assume that @xmath128 and @xmath129 remain at zero , or at least that any expectation values they acquire can be neglected . with this assumption ,
once @xmath9 and @xmath4 escape beyond the reach of the temperature , their dynamics will be governed by the zero temperature potential @xmath130 as @xmath41 increases , @xmath27 will quickly acquire a large mass @xmath131 , and so will be constrained to the minimum of its potential @xmath132 the effective potential then becomes @xmath133 \nonumber \\ & & \mbox { } + m_{l}^2 \left| l \right|^2 - m_{h_u}^2 \left| h_u \right|^2 - m_{\phi}^2 \left| \phi \right|^2\end{aligned}\ ] ] we assume the @xmath5-terms constrain @xmath134 and @xmath110 to the @xmath5-flat direction @xmath4 . then writing @xmath135 we get @xmath136\end{aligned}\ ] ] where @xmath137 . to make this potential more transparent
, we make the following change of variables @xmath138 @xmath139 where the @xmath140 s and the magnitudes of the @xmath141 s and @xmath142 s are of order one and we assume @xmath143 . we then get @xmath144 \label{v}\end{aligned}\ ] ] when @xmath145 , @xmath146 s potential is stabilised at @xmath147 while its phase is coupled to that of @xmath148 by the term ) and so the phase of @xmath14 will get a kick in the direction @xmath149 while the phase of @xmath9 will get a kick in the opposite direction .
this may contribute to the net lepton number generated , in addition to the similar effect to be described below . ]
@xmath150 the second term in the brackets being negligible at this stage .
when @xmath151 , the potential for the phase of @xmath148 will be dominated by the term @xmath152 and so in some sense we can regard the phase of @xmath148 as being determined modulo @xmath106 .
put in a different way , @xmath148 will be pulled towards one of the minima of its potential and so its phase will be strongly biased towards @xmath153 the phase of @xmath146 is then determined modulo @xmath154 by the term in eq .
( [ phase ] ) .
when @xmath148 becomes of order one , two things happen .
first , the second term in the brackets in eq .
( [ phase ] ) becomes of order one and gives the phase of @xmath146 a kick in the direction @xmath155 .
note that even before this @xmath146 will have had some angular momentum about @xmath156 , but it averages out to zero in the universe as a whole , as is shown in figure 2(a ) .
this new contribution has a direction determined by the parameters of the lagrangian and so will give a non - zero net contribution , as is shown in figure 2(b ) .
put another way , the difference in phase between @xmath157 and @xmath158 is our source of @xmath108 violation .
secondly , the last term in the brackets in @xmath159 becomes of order one giving @xmath146 a net positive mass squared and so causing it to spiral back in to @xmath156 .
assuming the expected broad parametric resonance @xcite provides enough damping , both @xmath9 and @xmath14 should then become trapped near their vacuum expectation values , after which the parametric resonance becomes less efficient
. @xmath14 s potential near @xmath160 conserves angular momentum , or in other words lepton number , and so @xmath14 s lepton number is conserved .
the affleck - dine condensate @xmath14 will decay well before the hubble expansion reduces its amplitude to the electroweak scale , and so will release enough thermal energy to restore the electroweak symmetry .
the lepton asymmetry will then be converted into a baryon asymmetry @xmath161 by the usual electroweak effects @xcite . finally ,
after the temperature has dropped to a few gev , the flaton decay will complete , releasing substantial entropy . the baryon asymmetry generated in this way
is roughly estimated to be @xmath162 where we have used eqs .
( [ npsi ] ) and ( [ seesaw ] ) .
@xmath163 is defined by this equation and will depend on the phase difference between @xmath157 and @xmath158 as well as the detailed dynamics . as discussed in section [ ti ]
, we expect @xmath164 to @xmath165 and @xmath166 to @xmath167 .
neutrino phenomenology @xcite suggests @xmath168 , @xmath169 to @xmath170 and @xmath171 .
therefore , in order for eq .
( [ nb ] ) to give the baryon asymmetry of eq .
( [ baryon ] ) , we require @xmath163 to be roughly corresponds to the scenario being unviable . ]
@xmath172 @xmath173 depending on which generations make up the affleck - dine @xmath4 direction .
the eventual measurement of the higgs and slepton masses should help to determine which of these ranges is the appropriate one ( or rule out the whole scenario ) , and a measurement of @xmath174 would narrow the uncertainty in @xmath175 .
right - handed neutrinos should acquire their masses due to the vacuum expectation value of the flaton that gives rise to thermal inflation , not some composite gut operator .
this will have important implications for gut model building .
in particular , @xmath176 gut s are strongly disfavoured because the flaton would have to be in a * 126 * representation which is difficult to derive from superstrings and one would have a flaton-125 splitting problem in addition to the usual doublet - triplet splitting problem .
the @xmath1-term of the mssm should also be generated by the vev of the flaton .
our affleck - dine type mechanism generates a baryon asymmetry which is roughly estimated to be @xmath177 where @xmath163 is the lepton asymmetry per affleck - dine particle .
@xmath163 depends on the difference in phase between the soft supersymmetry breaking parameters of @xmath105 and @xmath178 ( eqs .
( [ wseesaw ] ) and ( [ wvev ] ) ) , as well as the detailed dynamics
. we also make the following prediction @xmath179 modulo renormalisation effects , where @xmath180 is the soft supersymmetry breaking mass squared of @xmath110 , and @xmath181 is the soft supersymmetry breaking mass squared of a lepton doublet .
eds thanks b. de carlos and d. h. lyth for helpful discussions .
eds was supported by a royal society fellowship at lancaster university during the early stages of this work and is now supported by a jsps fellowship at resceu . the work of eds is supported by monbusho grant - in - aid for jsps fellows no .
95209 . 99 d. h. lyth and e. d. stewart , hep - ph/9510204 , phys .
rev . d*53 * , 1784 ( 1996 ) .
l. kofman , a. linde and a. a. starobinsky , hep - th/9510119 , phys .
* 76 * , 1011 ( 1996 ) ; t. barreiro , e. j. copeland , d. h. lyth and t. prokopec , hep - ph/9602263 .
for earlier related work see ref .
@xcite and p. bintruy and m. k. gaillard , phys .
rev . d*34 * , 3069 ( 1986 ) ; g. lazarides and q. shafi , nucl .
b392 * , 61 ( 1993 ) .
g. d. coughlan et al .
, phys . lett . *
131b * , 59 ( 1983 ) ; j. ellis , d. v. nanopoulos and m. quiros , phys . lett .
* b174 * , 176 ( 1986 ) ; t. banks , d. b. kaplan and a. e. nelson , hep - ph/9308292 , phys .
d*49 * , 779 ( 1994 ) ; b. de carlos , j. a. casas , f. quevedo and e. roulet , hep - ph/9308325 , phys .
b318 * , 447 ( 1993 ) ; l. randall and s. thomas , hep - ph/9407248 , nucl . phys . *
b449 * , 229 ( 1995 ) ; t. banks , m. berkooz and p. j. steinhardt , hep - th/9501053 , phys .
rev . d*52 * , 705 ( 1995 ) .
m. dine , l. randall and s. thomas , hep - ph/9503303 , phys .
lett . * 75 * , 398 ( 1995 ) ; hep - ph/9507453 , nucl . phys . *
b458 * , 291 ( 1996 ) .
a. linde , hep - th/9601083 , phys .
d*53 * , 4129 ( 1996 ) ; t. banks , hep - th/9601151 .
b. d. fields , k. kainulainen , k. a. olive and d. thomas , astro - ph/9603009 . for reviews of supersymmetry , see h. p. nilles , phys .
* 110 * , 1 ( 1984 ) and d. bailin and a. love , _ supersymmetric gauge field theory and string theory _ , iop , bristol ( 1994 ) .
c. bachas , c. fabre and t. yanagida , hep - th/9510094 , phys .
b370 * , 49 ( 1996 ) ; k. r. dienes , hep - th/9602045 .
t. yanagida , in proceedings of the workshop on the unified theory and the baryon number in the universe , edited by o. sawada and a. sugamoto ( kek , tsukuba , 1979 ) ; m. gell - mann , p. ramond and r. slansky , in supergravity , edited by p. van nieuwenhuizen and d. z. freedman ( north - holland , amsterdam , 1979 ) . k. s. babu , r. k. schaefer and q. shafi , astro - ph/9507006 , phys . rev . d*53 * , 606 ( 1996 ) .
k. inoue , a. kakuto , h. komatsu and s. takeshita , prog .
68 * , 927 ( 1982 ) ; s. p. martin and m. t. vaughn , hep - ph/9311340 , phys .
rev . d*50 * , 2282 ( 1994 ) ; b. de carlos and p. l. white , hep - ph/9602381 .
t. moroi , m. yamaguchi and t. yanagida , hep - ph/9409367 , phys .
b342 * , 105 ( 1995 ) ; m. kawasaki , t. moroi and t. yanagida , hep - ph/9509399 , phys . lett . * b370 * , 52 ( 1996 ) .
j. e. kim and h. p. nilles , phys . lett . *
138b * , 150 ( 1984 ) . h. murayama , h. suzuki and t. yanagida , phys . lett . * b291 * , 418 ( 1992 ) .
a. vilenkin , phys .
rev . d*23 * , 852 ( 1981 ) .
l. e. ibez and g. g. ross , phys .
b260 * , 291 ( 1991 ) ; nucl .
* b368 * , 3 ( 1992 ) ; t. banks and m. dine , phys . rev . d*45 * , 1424 ( 1992 ) ; l. e. ibez , hep - ph/9210211 , nucl . phys . *
b398 * , 301 ( 1993 ) .
d. h. lyth , in preparation .
i. affleck and m. dine , nucl .
* b249 * , 361 ( 1985 ) . h. murayama and t. yanagida , hep - ph/9310297 , phys .
b322 * , 349 ( 1994 ) .
l. kofman , a. linde and a. a. starobinsky , hep - th/9405187 , phys .
lett . * 73 * , 3195 ( 1994 ) .
m. fukugita and t. yanagida , phys . lett .
* b174 * , 45 ( 1986 ) . k. a. olive , hep - ph/9404352 .
j. w. f. valle , hep - ph/9602369 .
numerical simulation to illustrate the dynamics of @xmath9 and @xmath182 after thermal inflation . the potential of eq .
( [ v ] ) was used with the parameters @xmath183 , @xmath184 , @xmath185 and @xmath186 .
the initial conditions were @xmath187 , @xmath188 and @xmath189 .
a friction term @xmath190 with @xmath191 was added to the equation of motion of @xmath9 to crudely simulate the effects of parametric resonance .
a friction term was not added for @xmath14 because it would obscure the total lepton number generated which in reality is contained in both the homogeneous @xmath14 field and its decay products ( such as the inhomogeneous @xmath14 modes produced by parametric resonance ) . numerical simulation to show the non - zero net lepton number generated .
the same parameters as in figure 1 were used except for the following .
motivated by eq .
( [ init ] ) , the initial phase of @xmath14 was taken to be random while the initial phase of @xmath9 was taken to be given by @xmath192 .
the lepton number produced , as measured by @xmath193 , is plotted against the initial phase of @xmath14 .
the dotted line gives the average value .
the plots correspond to the following values of the parameters ( a ) @xmath194 ( which is @xmath108 conserving ) and @xmath195 , ( b ) @xmath186 and @xmath195 . | we argue that an extension of the minimal supersymmetric standard model that gives rise to viable thermal inflation , and so does not suffer from a polonyi / moduli problem , should contain right - handed neutrinos which acquire their masses due to the vacuum expectation value of the flaton that drives thermal inflation .
this strongly disfavours @xmath0 grand unified theories .
the @xmath1-term of the mssm should also arise due to the vev of the flaton . with the extra assumption that @xmath2 , but of course @xmath3 ,
we show that a complicated affleck - dine type of baryogenesis employing an @xmath4 @xmath5-flat direction can naturally generate the baryon asymmetry of the universe .
+ + + |
suppose @xmath3 is a projective plane of order @xmath1 , where @xmath4 is the set of points and @xmath5 is the set of lines in @xmath6 . for @xmath7 and @xmath8
, we say that @xmath9 is a _ nonincident _ set of points and lines if @xmath10 for every @xmath11 and every @xmath12 .
define @xmath13 to be the maximum integer @xmath0 such that there exists a nonincident set of @xmath0 points and @xmath0 lines in @xmath6 .
equivalently , @xmath13 is the size of the largest square submatrix of zeroes in the incidence matrix of @xmath6 .
we use a simple combinatorial argument to prove the upper bound @xmath14 , which holds for any projective plane @xmath6 of order @xmath1 .
we also show that this bound is tight in certain cases , namely , for the desarguesian plane pg@xmath15 when @xmath1 is an even power of two .
this is done by utilising maximal arcs .
[ t1 ] for any set @xmath16 of @xmath0 points in a projective plane of order @xmath1 , the number of lines disjoint from @xmath16 is at most @xmath17 suppose that @xmath18 is a projective plane of order @xmath1 . for a subset @xmath19 of @xmath0 points , define @xmath20 and define @xmath21 .
furthermore , for every @xmath22 , define @xmath23 , and then define @xmath24 .
observe that @xmath25 consists of the nonempty intersections of the lines in @xmath5 with the set @xmath16 .
denote @xmath26 .
we will study the set system @xmath27 .
we have the following equations : @xmath28 from the above equations , it follows that @xmath29 the @xmath30 blocks in @xmath25 through any point @xmath11 contain the @xmath31 points in @xmath32 once each , as well as @xmath30 occurrences of @xmath33 .
so the average size of a block in @xmath25 that contains any given point @xmath34 is @xmath35 .
therefore we define @xmath36 and compute as follows : @xmath37 from which it follows that @xmath38 therefore , @xmath39 _ remark . _ the inequality @xmath40 that we proved above
is in fact a well - known result that has been proven in many different guises over the years .
for example , mullin and vanstone @xcite proved that @xmath41 in any @xmath42 design on @xmath43 points .
if we let @xmath44 , @xmath45 and @xmath46 , then we obtain @xmath40 . if there exists a nonincident set of @xmath0 points and @xmath47 lines in a projective plane of order @xmath1 , then @xmath48 before proving our next general result , we look at a small example . in figure
[ plot1.fig ] , we graph the functions @xmath49 and @xmath0 for @xmath50 and @xmath51 . the point of intersection is @xmath52 and it is then easy to see that @xmath53 for any projective plane @xmath6 of order @xmath54 .
, width=340 ] in general , it is easy to compute the point of intersection of these two functions as follows : @xmath55 since @xmath56 , the point of intersection occurs when @xmath57 the following result is now straightforward . [ main.thm ] for any projective plane @xmath6 of order @xmath1 , it holds that @xmath58 .
suppose there is a nonincident set of @xmath0 points and @xmath0 lines in a projective plane of order @xmath1 .
theorem [ t1 ] implies that @xmath59 .
however , for @xmath60 , we have that @xmath61 , just as in the example considered above .
it follows that @xmath62 .
next , we examine the case of equality in theorem [ t1 ] .
this will involve maximal arcs , which we now define .
maximal @xmath63-arc _ in a projective plane of order @xmath1 is a set @xmath16 of @xmath0 points such that every line meets @xmath16 in @xmath64 or @xmath65 points .
it is well - known that a maximal @xmath63-arc has @xmath66 and the number of lines that intersect the maximal arc is precisely @xmath67 . for additional information on maximal arcs , see @xcite .
[ c3 ] suppose we have a set @xmath16 of @xmath0 points in a projective plane of order @xmath1 such that the number of lines disjoint from @xmath16 is equal to @xmath17 then @xmath16 is a maximal @xmath63-arc , where @xmath68 .
conversely , if @xmath16 is a maximal @xmath63-arc in a projective plane of order @xmath1 , then number of lines disjoint from @xmath16 is equal to @xmath49 . from the proof of theorem [ t1 ] , it is easy to see that equality holds if and only if every line in @xmath69 meets @xmath16 in exactly @xmath65 points , where @xmath36
. it immediately follows that every line in the plane meets @xmath16 in @xmath64 or @xmath65 points , and therefore @xmath16 is a maximal arc .
the converse follows from the basic properties of maximal arcs mentioned above .
[ even.thm ] when @xmath1 is an even power of @xmath70 , there exist nonincident sets of @xmath0 points and @xmath0 lines in pg@xmath15 , where @xmath71 .
denniston @xcite proved that there is a maximal @xmath72-arc in pg@xmath73 whenever @xmath74 .
suppose @xmath43 is even and we take @xmath75 .
therefore we have a maximal @xmath63-arc , where @xmath76 , @xmath77 and @xmath71 .
suppose we take @xmath16 to be the @xmath0 points in the arc and we apply corollary [ c3 ] . since @xmath66 , there are exactly @xmath0 lines in @xmath78
. therefore we have a nonincident set of @xmath0 points and @xmath0 lines in pg@xmath15 . [ even.cor ] if @xmath1 is an even power of @xmath70 , then @xmath79 . from theorem [ even.thm ]
, we have @xmath80 .
however , @xmath81 from theorem [ main.thm ] .
it follows that @xmath79 . in the case where @xmath1 is odd ,
it was shown by ball , blokhuis and mazzocca @xcite that there is no nontrivial maximal arc in the desarguesian plane pg@xmath15 .
the existence of maximal arcs for nondesarguesian projective planes of odd order is unresolved at the present time .
thanks to douglas west for posing the problem studied in this paper , and for making useful comments on earlier drafts of this paper . | in this paper , we study the problem of finding the largest possible set of @xmath0 points and @xmath0 lines in a projective plane of order @xmath1 , such that that none of the @xmath0 points lie on any of the @xmath0 lines .
we prove that @xmath2 .
we also show that equality can be attained in this bound whenever @xmath1 is an even power of two . |
the `` curvaton '' is an alternative formulation for generating curvature perturbation that does not require fluctuation of the inflaton field@xcite .
the curvaton model has several characteristic features , some of which distinguish curvaton - induced fluctuation from inflaton fluctuation . in this section we briefly review the calculations related to the lower bound of the inflation scale as derived by lyth in ref.@xcite . in this paper
we consider a significant relaxation of the bound .
we initially assume that a curvaton ( @xmath4 ) is frozen at @xmath5 during and after inflation . at the beginning of curvaton oscillation ,
the curvaton density is given by @xmath6 . denoting the total density of all other fields by @xmath7 ,
the ratio @xmath8 at the beginning of curvaton oscillation is @xmath9 where @xmath10 denotes the hubble parameter at the beginning of the oscillation . during the radiation - dominated epoch
, the ratio @xmath8 grows and reaches @xmath11 provided the curvaton potential is dominated by a positive quadratic term .
here , @xmath12 is the temperature for curvaton decay .
the basic concept of the curvaton scenario is that the curvature perturbation @xmath13 is generated by curvaton decay if the curvaton dominates the energy density of the universe .
the spectrum of the curvature perturbation related to the curvaton fluctuation is given by @xmath14 where @xmath15 denotes the hubble parameter during inflation .
since the observational data requires @xmath16 , the curvaton scenario leads to the condition @xmath17 using ( [ rfirst ] ) and ( [ pert1 ] ) , the condition for the cosmological parameters related to the curvaton scenario is given by @xmath18 here , the obvious bound from nucleosynthesis is @xmath19 mev , which gives a bound @xmath20 alternatively , we may use a lower bound for the curvaton decay rate ( @xmath21 ) that leads to the condition @xmath22 .
thus , we obtain an alternative condition @xmath23 the four parameters ( @xmath24 ) that appear in the above calculations give a cosmological boundary condition that characterizes the model . considering both ( [ pert1 ] ) and @xmath25 , we find @xmath26 and then from eq.([ineq1 ] ) and ( [ constraint1 ] ) we find @xmath27 alternatively , eq.([ineq1 ] ) and ( [ constraint2 ] ) give the condition @xmath28 in order to determine a clear bound for the inflation scale ( @xmath15 ) , we need to fix the boundary condition .
we assume that the curvaton mass ( @xmath29 ) is constant during the period that we are interested in .
then , the oscillation of the curvaton field starts when the hubble parameter falls below the curvaton mass ( i.e. , when the hubble parameter reaches @xmath30 ) , which leads to the boundary condition @xmath31 since @xmath32 always holds , we introduce a parameter @xmath33 that measures the ratio between @xmath15 and @xmath10 , defined as @xmath34 . using this new parameter we can replace @xmath10 and @xmath29 in eq.([constraint1a ] ) with @xmath15 and @xmath33 , giving @xmath35 from eq.([constraint2a ] ) we also find @xmath36 note that the bound in eq.([constraint2b ] ) is effective only when @xmath37 , which suggests that a tiny @xmath38 does not simply relax the bound .
therefore , the above conditions lead to a lower bound for the inflation scale ; @xmath39 this result is true for the traditional curvaton that satisfies the above boundary conditions . to summarize , an alternative ( maybe the curvaton )
is needed for low - scale inflation to avoid problems in the generation of the curvature perturbation , but the bound in eq.([bound - original ] ) suggests that the curvaton paradigm does not accommodate a low inflation scale . for the simplest curvaton
, it is impossible to support low - scale inflation@xcite , even if careful fine - tuning is introduced to the theory .
this is a critical problem for low - scale gravity signals that might be observed in the large hadron collider(lhc ) .
the curvaton could be identified from the brane distance , bulk field or the flat direction of supersymmetric gauge theory , though we will not specify the origin of the curvaton in this paper .
there are many model - dependent discussions for the curvaton hypothesis , but they are beyond the scope of this paper . our goal in this paper is to relax the energy condition of eq.([bound - original ] ) starting with the curvaton hypothesis so that the curvaton treatment can support low - scale inflation .
there are many alternatives generation@xcite or hybrids of these approaches@xcite .
the bound for the curvaton can not be applied to these alternatives . ] that can be applied to this problem .
we believe that it is worthwhile finding the simplest solution to this problem without introducing significant modifications to the traditional curvaton hypothesis .
arguments for the validity of the curvaton hypothesis are highly model - dependent . in this section
we examine the bound given in eq.([bound - original ] ) if @xmath40 and @xmath4 evolve after inflation but before oscillation begins . following the argument given in ref.@xcite ,
the ratio @xmath41 is almost constant after inflation even if @xmath42 and @xmath4 evolve after inflation .
therefore , for a concave potential @xmath43 , we introduce a parameter @xmath44 defined by @xmath45 where @xmath46 and @xmath47 denote the time when curvaton oscillation starts and the time when inflation ends , respectively .
is invariant during this evolution , and that the same result applies for the tachyonic evolution ( @xmath48 ) on a convex potential . ]
we can use these relations to evaluate a new energy bound for @xmath49 . using @xmath44 , we find @xmath50 rather than the expression given in ( [ constraint1a ] ) .
we also find @xmath51 rather than ( [ constraint2a ] ) . from these equations and the definition of @xmath33
, we find @xmath52 which is not relaxed by curvaton evolution after inflation if @xmath53 .
therefore , for a concave potential , the bound of the energy scale is not relaxed by curvaton evolution .
this result supports the condition @xmath54 gev given in ref.@xcite .
it was suggested in ref.@xcite that this condition can be relaxed by considering a `` heavy curvaton''@xcite that gives the non - standard boundary condition @xmath55 which is due to a phase transition at the end of inflation .
this new boundary condition significantly alters the bound and gives a milder condition @xmath56 however , this is not sufficient to support the inflation energy scale as low as @xmath57tev .
as we have presented in ref.@xcite , if the phase transition that gives a heavy curvaton occurs later than the primordial inflation , it may lead to another boundary condition that may remove the bound for @xmath15 .
other solutions for this problem have been extensively studied@xcite using other alternative formulations . in this paper
we consider a generic case in which both the coefficient @xmath58 and the sign of the quadratic term depend on the specific conditions of the system .
we show that hilltop curvatons can significantly relax the condition for the inflation scale without introducing additional components or symmetries to the traditional curvaton hypothesis
. evolution of a curvaton on a convex potential that leads to tachyonic enhancement of the fluctuation has already been discussed for a specific model of peccei - quinn field curvatons@xcite . in this model , amplification occurs when the curvaton is stabilized for a period near a maximum of the potential .
although tachyonic amplification with static potential has already been discussed for some specific models , the generality of tachyonic enhancement is unclear for a non - static potential .
we show that a hilltop curvaton , whose mass is not protected from modest @xmath0 corrections , can achieve significant enhancement of the curvaton fluctuation . ) and ( [ non - static ] ) for the difference between tachyonic amplification on static and non - static potentials . ]
( 550,160)(0,0 ) ( 550,160)(0,0 ) we now discuss the enhancement of curvaton fluctuation with a convex potential without introducing additional fine - tuning to the traditional curvaton scenario .
the evolution of the curvaton ( @xmath4 ) and its fluctuation ( @xmath42 ) is given in terms of the effective potential ( @xmath59 ) by the equation @xmath60 where @xmath61 is the @xmath4-derivative of the effective potential , and by the equation @xmath62 here the dot denotes differentiation with respect to the cosmic time @xmath63 . following the original curvaton hypothesis , we consider an effective potential that represents the conventional curvaton potential except for the explicit @xmath0-correction ; @xmath64\sigma^2 + \frac{g}{n!}\frac{\sigma^n}{m_*^{n-4}},\ ] ] where @xmath65 is an integer ( @xmath66 ) , and the cut - off scale @xmath67 is supposed to be @xmath68
. can be much smaller than the planck scale . in that case symmetries or mechanisms are required to suppress the coefficients of higher dimensional interactions so that the curvaton potential remains flat .
we follow the original curvaton hypothesis and do not discuss symmetries and mechanisms in this paper since they are highly model - dependent . ]
following the curvaton hypothesis , quartic term of the curvaton potential must be negligible .
the evolution of the curvaton after @xmath69 is precisely the same as the conventional scenario .
evolution before the curvaton oscillation is determined by the @xmath0-correction term ; in our model it leads to tachyonic enhancement .
the sign of the hubble - induced mass term depends on the conditions and must be negative during tachyonic enhancement . for simplicity
, we fix the sign of the hubble - induced mass term to be negative , as this assumption does not reduce the generality of the following argument .
we assume that the expectation value of the curvaton @xmath70 is always smaller than the effective minimum until @xmath29 dominates the curvaton mass .
note that if oscillation occurs about the effective minimum , the curvaton fluctuation is reduced .
then , the evolution of the curvaton during inflation is given by @xmath71 where @xmath72 this is the so - called fast - roll evolution@xcite .
comparing ( [ sigma - eq ] ) and ( [ delsig - eq ] ) , it is straightforward to derive the equation for @xmath40 as @xmath73 since the spectrum index of the curvaton perturbation is given by @xmath74 for models with negligible slow - roll parameter @xmath75 , we take the condition - dependent coefficient to be @xmath76 during inflation .
is responsible for the current observational value of the spectral index @xmath65 , the value of @xmath77 during inflation is not determined by @xmath65 . in this case
, there is an upper bound @xmath78 if there is no fine - tuning between @xmath75 and @xmath79 . ] here the definition of the slow - roll parameter @xmath79 is @xmath80 therefore , the exponential factors in the above equations are at most @xmath81 .
thus , we conclude that the evolution of the curvaton and its fluctuation during inflation is not significant .
the curvaton dynamics after inflation are different from those of the inflationary epoch .
we introduce the barotropic parameter ( @xmath82 ) of the dominant fluid , which does not relate to the curvaton , but instead relates to the background radiation ( @xmath83 for radiation ) or the inflaton oscillation ( for inflaton oscillation @xmath82 depends on the order of the inflaton potential@xcite ) . as we are considering the evolution of the curvaton before the beginning of curvaton oscillation at @xmath84 ,
the dominant part of the effective potential is @xmath85 here , both @xmath1 and @xmath4 depend on the cosmic time @xmath63 , while @xmath77 is a constant determined by the dominant component of the universe .
the value of @xmath77 during inflaton oscillation is believed to be the same as that of the inflationary epoch , while @xmath77 may be different during the radiation - dominated epoch . substituting the @xmath63-derivative by @xmath86 ,
we find@xcite @xmath87 where the prime denotes differentiation with respect to @xmath88 . solving the above equation for @xmath88 and then substituting @xmath63 , we find@xcite @xmath89 where @xmath90 is defined as @xmath91 here , @xmath92 and @xmath93 are the time at the beginning and end of the evolution , respectively . the evolution of the curvaton and its fluctuation is negligible for @xmath94 , while it is significant for @xmath95 .
since the potential becomes concave near the minimum of the effective potential , we consider the evolution within @xmath96 , where @xmath97 denotes the effective minimum for @xmath98 .
as the time passes , the bare mass ( @xmath29 ) starts to dominate the potential and tachyonic amplification stops at @xmath99 , @xmath100 .
then , the curvaton oscillation starts at @xmath99 , as for a traditional curvaton .
we now consider a concrete example .
we consider a case in which the coefficient of the induced mass is as large as @xmath101 after reheating .
reheating occurs at @xmath102 , then the curvaton oscillation starts at @xmath103 . with this simple and generic situation
we examine the energy condition for the hilltop curvaton .
we introduce a new parameter @xmath104 , which gives the reheating temperature @xmath105 using this parameter , the ratio @xmath106 is given by @xmath107 . considering eq.([enhg1 ] ) and ( [ enhg2 ] ) , we find the enhancement factor @xmath108 for @xmath109 , the enhancement factor @xmath44 is given by @xmath110 .
note that the condition @xmath111 applies because of the relation @xmath112 , where @xmath113 denotes the time when inflation ends .
since ( [ bound - alt ] ) is true for @xmath114 , we find @xmath115 for @xmath101 , @xmath109 and @xmath114 . in this case
, we find @xmath116gev and @xmath117gev .
the required condition for the effective potential is that @xmath118 is smaller than @xmath97 during the tachyonic amplification .
this significant reduction of the inflation energy is a generic consequence of the enhancement of the curvaton fluctuation induced by the @xmath0 mass .
the required condition for this amplification are ( 1 ) if the coefficient of the hubble - induced mass ( @xmath77 ) is somewhat large during the radiation - dominated epoch and ( 2 ) the sign of the mass term is negative , then the curvaton perturbation grows significantly on a convex potential during the radiation - dominated epoch .
this significant enhancement gives a significant relaxation of the energy condition for the primordial inflation .
we have studied a scenario for hilltop curvatons in which a curvaton evolves on a convex potential .
we assumed that the effective mass of a curvaton is expressed as @xmath2 , where the coefficient varies within @xmath3 depending on the dominant component of the universe .
a negative mass term leads to enhancement of the curvaton fluctuation . in our model , both @xmath40 and the ratio @xmath8 at the beginning of the oscillation are enhanced during this epoch , which has a significant impact on the energy bound for low - scale inflation . following the curvaton hypothesis ,
we have demonstrated the generality of this enhancement .
the hubble - induced mass is necessary for the significant enhancement of the curvaton perturbation in our model .
if there is no @xmath0 correction to the curvaton potential , our scenario requires two different kinds of phase transitions that change the effective mass of the curvaton at least twice .
these phase transitions must occur ( 1 ) between inflation and tachyonic enhancement and ( 2 ) between tachyonic enhancement and curvaton oscillation . moreover ,
tachyonic enhancement with a constant mass is very dangerous for our scenario because a curvaton will soon start to oscillate about the effective minimum . in order to avoid the reduction of @xmath40 during this oscillation
, the second phase transition must take place soon after the first .
the important idea of our scenario is that for a @xmath119 potential , the tachyonic amplification lasts long enough that the conventional curvaton oscillation may start before the beginning of the oscillation about the effective minimum .
in addition to the condition required for the tachyonic enhancement , we must consider the origin of the spectral index @xmath120 .
for the curvaton , this index is given by @xmath121 where @xmath122 is the slow - roll parameter .
in many models of inflation in which @xmath75 is very small , @xmath79 must be responsible for @xmath120 .
however , if there is no mechanism that forces @xmath123 ( i.e. , @xmath124 ) during inflation , it is very difficult to attain the present observational value of the spectral index without introducing fine - tuning .
this may reduce the generality of the curvaton hypothesis , narrowing the range of possible inflation models in which the curvaton can explain the curvature perturbation . in this sense ,
the @xmath0 correction is important for the generality of the curvatons , even when tachyonic enhancement does not play an important role .
it should be noted again that the required condition for the curvature perturbation is very severe for low - scale inflation@xcite if it is generated by the inflaton fluctuation .
we wish to thank k.shima for encouragement , and our colleagues at tokyo university for their kind hospitality .
the purpose of this appendix is to show how @xmath0 corrections to light scalar fields appear in the radiation - dominated epoch .
we will mainly follow the original argument given by dine et.al , in ref.@xcite .
the supergravity scalar potential is given by @xmath125 + v_d,\ ] ] where @xmath126 and @xmath90 are the superpotential and khler potential , and the derivatives are defined as @xmath127 and @xmath128 .
looking at this scalar potential , there is a @xmath0-correction appears for the soft mass of a light field @xmath4 when the energy density is dominated by @xmath129-terms .
this is the most discussed origin of the @xmath0-corrections .
however in the original argument the origin of such terms is not confined to the scalar potential .
following ref.@xcite , we consider @xmath0 corrections that may appear from nonrenormalizable interactions such as @xmath130 where the scalar component of the superfield @xmath131 is the light field @xmath4 , and @xmath132 is a field that dominates the energy density of the universe .
then , the energy density of the universe is given by @xmath133 .
note that in the thermal phase , the expectation value arises from kinetic terms , while the usual @xmath134 correction is significant in a @xmath135-dominated universe . following this argument
, a @xmath0 correction may result even if d - term dominates the potential . from the above argument it is clear that there is no reason to assume @xmath77 in @xmath136 takes exactly the same value in both the inflation stage and the thermal phase .
this is the main reason why we considered @xmath137 variation to @xmath77 . 1 a. d. linde and v. f. mukhanov , _ nongaussian isocurvature perturbations from inflation ,
phys.rev.d56(197)535_ [ astro - ph/9610219 ] ; d. h. lyth , d. wands , _ generating the curvature perturbation without an inflaton , phys.lett.b524(2002)5_ [ hep - ph/0110002 ] ; t. moroi , t. takahashi , _ effects of cosmological moduli fields on cosmic microwave background , phys.lett.b522(2001)215_ [ hep - ph/0110096 ] .
d. h. lyth , _ can the curvaton paradigm accommodate a low inflation scale ? , phys.lett.b579(2004)239-244,2004_[hep-th/0308110 ] .
d. h. lyth , _
constraints on tev scale hybrid inflation and comments on nonhybrid alternatives , phys.lett.b466(1999)85_ [ hep - ph/9908219 ] ; a. m. green and a. mazumdar , _ dynamics of a large extra dimension inspired hybrid inflation model , phys.rev.d65(2002)105022_ [ hep - ph/0201209 ] ; r. n. mohapatra , a. perez - lorenzana , and c. a. de s. pires , _ inflation in models with large extra dimensions driven by a bulk scalar field , phys.rev.d62(2000)105030_ [ hep - ph/0003089 ] ; t. matsuda , _ kaluza - klein modes in hybrid inflation , phys.rev.d66(2002)107301_ [ hep - ph/0209214 ] ; t. matsuda , _ nontachyonic brane inflation , phys.rev.d67(2003)083519_ [ hep - ph/0302035 ] ; a. mazumdar , _ new developments on embedding inflation in gauge theory and particle physics _ [ arxiv:0707.3350 ] . g. dvali , a. gruzinov and m. zaldarriaga , _ a new mechanism for generating density perturbations from inflation , phys.rev.d69(2004)023505_ [ astro - ph/0303591 ] ; f. vernizzi , _ cosmological perturbations from varying masses and couplings , phys.rev.d69(2004)083526_ [ astro - ph/0311167 ] .
d. h. lyth , _ generating the curvature perturbation at the end of inflation , jcap 0511:006,2005 _ [ astro - ph/0510443 ] ; f. bernardeau , l. kofman and j.p .
uzan , _ modulated fluctuations from hybrid inflation , phys.rev.d70(2004)083004_ [ astro - ph/0403315 ] ; t. matsuda , _
elliptic inflation : generating the curvature perturbation without slow - roll , jcap 0609:003,2006 _ [ hep - ph/0606137 ] ; t. matsuda , _
brane inflation without slow - roll , jhep 0703:096,2007 _ [ astro - ph/0610402 ] .
e. w. kolb , a. riotto and a. vallinotto , _ curvature perturbations from broken symmetries , phys.rev.d71(2005)043513_[astro-ph/0410546 ] ; t. matsuda , _ generating the curvature perturbation with instant preheating , jcap 0703:003,2007 _ [ hep - th/0610232 ] ; t. matsuda , _ generating curvature perturbations with mssm flat directions , jcap 0706:029,2007_[hep - ph/0701024 ] .
t. matsuda , _ topological curvatons , phys.rev.d72(2005)123508_ [ hep - ph/0509063 ] .
t. matsuda , _ no curvatons or hybrid quintessential inflation , jcap 0708:003,2007 _ [ arxiv:0707.1948 ] ; t. matsuda , _ cosmological perturbations from inhomogeneous preheating and multi - field trapping , jhep 0707:035,2007 _ [ arxiv:0707.0543 ] ; t. matsuda , _ hybrid curvatons from broken symmetry , jhep 0709:027,2007 _ [ arxiv:0708.4098 ] .
k. dimopoulos , g. lazarides , d. lyth and r. r. de austri , _ curvaton dynamics , phys.rev.d68(2003)123515_ [ hep - ph/0308015 ] . t. matsuda , _ curvaton paradigm can accommodate multiple low inflation scales , class.quant.grav.21(2004)l11_ [ hep - ph/0312058 ] ; k. dimopoulos , d. h. lyth and y. rodriguez , _ low scale inflation and the curvaton mechanism , jhep 0502:055,2005 _ [ hep - ph/0411119 ] ; y. rodriguez , _ low scale inflation and the immediate heavy curvaton decay , mod.phys.lett.a20:2057,2005_ [ hep - ph/0411120 ] .
k. dimopoulos , g. lazarides , d. lyth and r. r. de austri , _ the peccei - quinn field as curvaton , jhep 0305:057,2003 _ [ hep - ph/0303154 ] .
k. dimopoulos , d. h. lyth , a. notari and a. riotto , _ the curvaton as a pseudonambu - goldstone boson , jhep 0307:053,2003 _ [ hep - ph/0304050 ] .
a. linde , _ fast roll inflation , jhep 0111:052,2001 _ [ hep - th/0110195 ] .
m. s. turner , _ coherent scalar field oscillations in an expanding universe , phys.rev.d28(1983)1243_. l. boubekeur and d. h. lyth , _ hilltop inflation , jcap 0507:010,2005_[hep - ph/0502047 ] .
m. dine , l. randall , s. d. thomas , _
supersymmetry breaking in the early universe , phys.rev.lett.75(1995)398_ [ hep - ph/9503303 ] . | we study `` hilltop '' curvatons that evolve on a convex potential .
hilltop curvatons evolving on the hubble - induced potential are generic if supergravity is assumed in the theory .
we do not consider curvatons whose potential is protected from @xmath0 corrections , where @xmath1 is the hubble parameter .
we assume that the effective mass of a curvaton is expressed as @xmath2 , where the coefficient varies within @xmath3 depending on the circumstances .
a negative mass term may lead to enhancement of curvaton fluctuation , which has a significant impact on the energy bound for low - scale inflation . using a simple curvaton model and following the conventional curvaton hypothesis
, we demonstrate the generality of this enhancement .
sit - hep / tm-41 1.5 truecm .75 truecm * tomohiro matsuda * .4 truecm _ laboratory of physics , saitama institute of technology , _
_ fusaiji , okabe - machi , saitama 369 - 0293 , japan _
\1 .
truecm addtoresetequationsection 1 . truecm |
iras 17423@xmath11755 ( hen 3@xmath11475 ) was first suggested by @xcite as a possible member of the transition phase from the asymptotic giant branch ( agb ) to the planetary nebula ( pn ) stage due to its unusual iras colors .
the high values of the [ nii]/h@xmath5 ratios in the outflowing material detected by @xcite and the low luminosity deduced for the central star allowed them to confirm the classification of this object as an evolved star . _ hubble space telescope _
( _ hst _ ) and _ very large array _ ( _ vla _ )
observations by @xcite showed the presence of both oh maser emission and highly collimated ionized outflows like those detected in oh / ir stars or very young pn .
_ hst _ images revealed a rich and complex morphological structure in the circumstellar material ( see section 4 for details ) .
the outflow is collimated in bipolar jets along several condensations of shock - excited gas that extend about 11 arcsec .
the lobes show expansion velocities of about 425 km s@xmath6 and a high velocity jet ( @xmath0900 km
s@xmath6 ) in the inner part of the lobes @xcite . the nebula displays a remarkable point symmetry that has been interpreted as due to the precession of a central binary system that undergoes episodic events of mass loss .
@xcite proposed that the expanding shell has a torus - like structure where the oh emission originates in a high density region , where h@xmath3o is dissociated and further collimated in the observed jets . additionally , @xcite found evidences of ultrafast winds ( up to 2300 km s@xmath6 ) highly collimated and located close to the central star which could be a relatively young post - agb outflow not strongly altered by interaction with the agb .
stars at the end of the agb phase are characterized by severe mass loss ( @xmath7 to @xmath8 m@xmath9 yr@xmath6 ) , which results in the formation of circumstellar envelopes @xcite .
the spherical symmetry of the envelopes of agb stars is translated into a variety of shapes in the pn phase by a mechanism or mechanisms not as yet well understood .
there is increasing evidence that at least in some instances the shaping starts at the end of the agb phase @xcite .
iras 17423@xmath11755 is a spectacular example that may represent a link between oh / ir stars with extreme outflows and highly bipolar pn .
the spectral energy distribution ( sed ) of extreme ( e.g. , highly embedded ) oh / ir agb stars is characterized by the presence of strong and broad amorphous silicate absorption features at 9.7 and 18 @xmath2 m together with crystalline silicate absorption / emission features from 10 to 45 @xmath2 m @xcite . at the end of the agb phase , the crystalline silicate features become dominant and can be observed in more evolved o - rich pn @xcite .
comparison of the _ infrared space observatory _
( _ iso _ ) observations of o - rich dust shells surrounding evolved stars with laboratory data suggested the presence of several families of crystalline silicates , such as olivines and pyroxenes , and marked the beginning of an emerging discipline : the mineralogy of stellar and other astronomical ( i.e. , cometary ) dust shells .
water ice features at 3.1 , 43 , and 62 @xmath2 m have been additionally observed in heavily obscured and extremely bipolar sources such as the post - agb star iras 19343@xmath42926 ( or m1@xmath192 , see e.g. dijkstra et al .
2006 and references therein ) .
@xcite studied the _ iso _
spectra of seven hot post - agb stars including iras 17423@xmath11755 .
dusty models @xcite were fitted to optical , near- and far - infrared ( iras and _ iso _ ) photometry in order to reconstruct the seds and to derive physical parameters such as dust temperatures , mass loss rates , angular radii and the inner boundary of the dust envelopes . for the particular case of iras 17423@xmath11755
they considered a combination of silicates and carbon in the circumstellar environment .
they reported the presence of a broad absorption feature at 3.1 @xmath2 m that they identified as due to the presence of c@xmath3h@xmath3 and/or hcn in the circumstellar envelope .
this identification led these authors to infer a c - rich chemistry for the shell .
more recently , @xcite presented observations of a sample of 26 hot post - agb stars with the infrared array camera and the infrared spectrograph ( irs ) on board the _ spitzer space telescope_. these observations were analyzed together with two micron all sky survey , iras and radio centimeter data in order to model the seds of the targets .
@xcite classified iras 17423@xmath11755 as a c - rich star on the basis of the @xcite report of the c@xmath3h@xmath3 feature at 3.1 @xmath2 m and in the absence of a strong 9.7 @xmath2 m amorphous silicate absorption / emission feature in their low - resolution ( r@xmath064@xmath1128 ) _ spitzer _ spectrum . however , they pointed out that the expected polycyclic aromatic hydrocarbon ( pah ) features in the 512 @xmath2 m region are not detected .
it is to be noted here that weak and narrow molecular absorptions from c - based molecules such as 13.7 @xmath2 m c@xmath3h@xmath3 , 14.0 @xmath2 m hcn , etc .
, are difficult to detect at the low resolution of their _ spitzer _ spectrum .
the detection of these c - rich molecular absorptions - typical of c - rich agb / post - agb stars - requires in most of the cases higher resolution observations such as those provided by the high - resolution modes of _ spitzer _ ( _ r_@xmath0600 ) and _ iso _ ( _ r_@xmath01000 ) ( see e.g. , @xcite ) . the controversial origin of the mid- to far - ir features in iras 17423@xmath11755 merits a re - analysis of the dust features observed in the high - resolution and higher quality _ spitzer _ spectrum . in section 2 we present the new _
observations together with the construction of the overall sed of the nebula as observed by both _
spitzer _ and _ iso _ , while in section 3 the evidence for an o - rich chemistry is analyzed and discussed .
the evolutionary stage of iras 17423@xmath11755 is discussed , including our new results , in section 4 while a summary of our main conclusions is presented in section 5 .
high - resolution ( r@xmath0600 ) _
spitzer_/irs spectra of iras 17423@xmath11755 are now available in the _ spitzer _ public database . short - high ( sh : 9.9 - 19.6 @xmath2 m ) and long - high ( lh : 18.7 - 37.2 @xmath2 m ) observations were obtained on 2009 april 21 under the general observer program @xmath1050777 ( p.i .
: b. mccollum ) . a typical signal - to - noise ratio ( s / n ) higher than 50 was easily reached by using just three cycles of 6 s for both sh and lh modules .
however , the s / n is much lower for wavelengths longer than 34 @xmath2 m - the red end of the lh module that is affected by a strong noise level .
the post - bcd products ( one spectrum for each nod position ) automatically reduced by the irs custom extractor ( spice ) were retrieved from the _ spitzer _ database .
the automatic data reduction includes the extraction from the 2-dimensional images as well as the wavelength and flux calibration .
spitzer_-contributed software smart @xcite was used for cleaning residual bad pixels , spurious jumps and glitches and for smoothing and merging .
the short - low ( sl : 5.2 - 14.5 @xmath2 m ) _ spitzer _ spectrum reported by @xcite was also retrieved from the _ spitzer _ database .
we found a good match ( @xmath115% ) between the sl and sh module spectra .
however , the absolute flux level of the lh module spectra was found to be @xmath08 @xmath12 higher than the sh module spectra .
thus , we scaled the lh observations to the sh ones in order to obtain the final high - resolution _ spitzer _
spectrum of iras 17423@xmath11755 .
the good match between the _
spitzer_/irs sl and the sh spectra is illustrated in figure 1 .
for comparison , we also retrieved from the _ spitzer _ database the high - resolution _
observations of the similar post - agb stars iras 19343@xmath42926 and iras 17393@xmath12727 ( an oh / ir massive post - agb already reported by @xcite ) . the full range _ iso _ sws+lws spectrum of iras 17423@xmath11755 was obtained in 1997 march as part of the open - time program pgarcia.pn on spectroscopy of proto - pn candidates .
both sws and lws spectra were taken in the full scan aot1 mode @xcite at speed 1 .
the data were processed using the standard interactive analysis software , version 7.0 of the sws and lws off - line processing system at the max planck institute for extraterrestrial astronomy ( garching , germany ) ; see @xcite for more details on the reduction procedure .
all detector signals were inspected for spurious features , which were removed .
no fringes were seen in the _ iso _ spectrum of this source . there is good agreement between our sws spectrum and iras low resolution overlap region , and a good match between the sws and lws spectra .
the 3.1 @xmath2 m absorption band can be clearly observed in sws aot band 1d .
no firm conclusions can be drawn concerning the presence of any other solid - state dust feature because of the strong noise level in the _ iso _ spectrum .
the contribution from galactic cirrus emission was estimated by irsky at the galactic coordinates of the nebula .
it was fitted by polynomials and subtracted from the resulting _ iso _ spectra .
the final spectrum was transformed to standard fits format and the subsequent processing and analysis was performed using iraf routines .
another post - agb nebula sharing some observational properties with iras 17423@xmath11755 is iras 19343@xmath42926 ( m1@xmath192 ) , the minkowski footprint @xcite .
both nebulae display similar morphological properties ( bipolarity ) , a comparable effective temperature for the central ( b - type ) star , host an oh maser and also present analogous infrared spectral distributions ( see also section 4 ) . in order to compare their infrared properties ,
we have extracted the _ iso _ spectra for this source from the archive and performed a similar sws+lws reduction procedure as in the case of iras 17423@xmath11755 .
the _ iso _ spectrum of iras 19343@xmath42926 is of much higher quality than that of iras 17423@xmath11755 .
indeed , @xcite have previously classified the _ iso _ spectra of iras 17423@xmath11755 and iras 19343@xmath42926 with the classes 5.f : and 5.sa , respectively .
this means that both spectra have red continua with the _ iso _ spectrum of iras 17423@xmath11755 displaying no strong features superimposed on the sed , and the _ iso _ spectrum of iras 19343@xmath42926 displaying a clear 10 @xmath2 m silicate absorption feature and classified as an o - rich source .
objects in class 5.sa often also possess water ice , co and co@xmath3 features .
additionally , the irs spectrum of the post - agb star iras 17393@xmath12727 @xcite , will also be considered for comparison with iras 17423@xmath11755 .
we have constructed the seds of iras 17423@xmath11755 and iras 19343@xmath42926 in the spectral region from 2 to @xmath0180 @xmath2 m by using the irs _ spitzer _ data in the interval from 5 to 34 @xmath2 m and _ iso _ sws and lws in the remaining range .
the _ iso _ data were smoothed to the resolution of _ spitzer _
( r@xmath0600 ) and scaled to the _ spitzer _ fluxes by using maximum scaling factors of @xmath020 @xmath12 .
the resulting seds for both objects are shown in figure 2 .
the overall seds of iras 17423@xmath11755 and iras 19343@xmath42926 peak between 40 and 60 @xmath2 m .
notably , the absorption feature in the 2.5 - 3.5 @xmath2 m region appears strong in both objects , together with [ cii ] in emission at 158 @xmath2 m ; however , this latter feature might be galactic residual background emission .
the @xmath02@xmath1180 @xmath2 m seds in figure 2 display a very cold bimodal continuum that can be interpreted in terms of thermal emission from the dust .
obviously , it would correspond to several wavelength - dependent dust emissivities with a certain range of dust temperatures , but we found that two main components of the nebular dust , cold ( strong ) and hot ( weaker ) , could roughly reproduce the general trend of the observed overall energy distribution .
we performed a multiple blackbody fitting using the iraf routine nlfit on the f@xmath13 over @xmath14 spectrum and obtained blackbody temperatures of 120 k and 965 k for iras 17423@xmath11755 ( in agreement with the values reported in @xcite ) and temperatures of 120 k and 965 k in the case of iras 17423@xmath11755 .
the fitted blackbody curves are shown in the f@xmath15 over @xmath14 spectra of figure 2 , where we also detail the fitting in the region of the 3.1 @xmath2 m absorption feature . @xcite
have recently reported difficulties in modeling the seds of heavily obscured o - rich post - agb stars by using a more sophisticated radiation transfer code for dusty environments such as dusty .
they needed to add amorphous carbon dust in their dusty models in order to reproduce both the observed red continua and the apparently weak 10 @xmath2 m amorphous silicate absorption.1755 .
this fact probably led these authors to identify the strong 3.1 @xmath2 m absorption band as due to c@xmath3h@xmath3 .
] it is to be noted here that the inclusion of amorphous carbon grains in the dusty models of o - rich post - agb stars does not necessarily mean that these sources are c - rich ; their infrared spectra show o - rich dust features only .
other more unusual o - rich dust species with optical properties similar to those of the amorphous carbon grains could be present in the circumstellar shells of heavily obscured o - rich post - agb stars ( see section 4 ) , also giving a good fit to the observed seds of these stars ( r. szczerba 2009 , private communication ) .
these difficulties prevent us from proceeding further with this modeling , as the thermal continuum around 3 @xmath2 m was found to provide an adequate reference for the optical depth of the observed feature and we are mainly interested in the analysis of the dust features present in the new high - resolution _ spitzer _ spectrum .
the most interesting and well - defined feature in the _ iso _ spectrum of iras 17423@xmath11755 is a clear absorption band at 3.1 @xmath2 m , which we identify for the first time as water ice ( see below ) present in the dust grains of the circumstellar material .
similar detections are not very numerous , the comparison object being iras 19343@xmath42926 , a bipolar post - agb star with an o - rich composition , another case that has also been observed with _ iso_. the identification of this 3.1 @xmath2 m absorption feature as c@xmath3h@xmath3 by @xcite is erroneous , as the new high - resolution _
spitzer_/irs spectrum confirms .
figure 3 displays the sed of iras 17423@xmath11755 from 2 to @xmath016 @xmath2 m together with the comparison o - rich post - agb stars iras 19343@xmath42926 and iras 17393@xmath12727 .
no other feature characteristic of a c - rich chemistry ( e.g. , sic , pahs , c@xmath3h@xmath3 , hcn ) in the circumstellar envelope can be found in the observed _ spitzer _ spectrum . in particular , there is no indication of the presence of c@xmath3h@xmath3 absorption at 13.7 @xmath2 m and the key point here is that the c@xmath3h@xmath3 ro - vibrational lines at @xmath03 and 14 @xmath2 m are of similar strength ( @xmath020 @xmath12 from the continuum ; see e.g. , cernicharo et al .
( 1999 ) ) .
thus , if the 3.1 @xmath2 m absorption feature is due to c@xmath3h@xmath3 , then the c@xmath3h@xmath3 13.7 @xmath2 m feature should be detected in the high - resolution _
spectrum presented here .
the non - detection of this feature , or even the other small hydrocarbons detected in c - rich post - agb stars @xcite supports our identification of the 3.1 @xmath2 m absorption band with water ice . _
spitzer_/irs sl and sh spectra of iras 17423@xmath11755 reveal for the first time a mid - ir feature that was not evident in the low s / n _ iso _ spectrum @xcite and that was not recognized by @xcite in their irs / sl spectrum .
a weak and broad 9.7 @xmath2 m absorption from amorphous silicates ( o - rich ) is present .
this is shown in figure 3 , in comparison with the other well known o - rich post - agb stars iras 17393@xmath12727 and iras 19343@xmath42926 .
@xcite discussed the presence of crystalline silicates in the _
spitzer_/irs spectrum of iras 17393@xmath12727 .
they reported the presence of several weak crystalline silicate emission features in the 27 - 31 @xmath2 m region , and crystalline silicate absorption features at shorter wavelengths .
in particular , an absorption feature at 15.4 @xmath2 m can be clearly observed in the irs spectrum of this star presented in figure 3 .
such crystalline silicate absorption / emission features can not be claimed for definitely in either iras 17423@xmath11755 or iras 19343@xmath42926 at the s / n level of our _ spitzer_/irs spectra .
it is worth noticing here that these weak crystalline silicate features have been observed in the circumstellar envelopes of o - rich evolved stars with a variety of strengths and at slightly different wavelengths from source to source , depending on the specific chemical composition and the size and density of the dust grains in the circumstellar envelope @xcite . our identification at 3.1 @xmath2 m of water ice and the weak 9.7 @xmath2
m amorphous silicates absorption permit us to infer for the first time an o - rich chemistry for the circumstellar envelope around iras 17423@xmath11755 .
figure 3 shows that a notable similarity exists between the infrared spectra of the o - rich post - agb star iras 19343@xmath42926 and that of iras 17423@xmath11755 .
in addition , [ neii ] nebular emission at 12.8 @xmath2 m is also detected for the three objects shown in this figure , suggesting that the ionization of the circumstellar material may have already started in these evolved stars .
the conditions for condensation and properties of amorphous and crystalline h@xmath3o ice in astrophysical environments have been extensively reviewed by several authors @xcite .
different forms of water ice can be observed , depending mainly on temperature , pressure and the mechanisms of deposition . for typical low pressure conditions present in the circumstellar envelopes ,
temperatures higher than 150170 k produce hexagonal ice , which remains stable during further cooling . if deposition takes place at lower temperatures ( between 110 and 130 k ) cubic ice is formed , while at temperatures lower than 100130 k the resulting ice is amorphous @xcite . in the circumstellar envelopes of evolved stars
water ice can condense forming icy mantles on dust grains which have previously condensed while the gas cooled down in the expanding envelopes . in o - rich envelopes all the carbon atoms
are thought to be locked in co molecules and the remaining o atoms are supposed to form h@xmath3o aggregates . determining the occurrence and characterization of water ice features is interesting because ice in these mantles can provide important diagnostics of the physical conditions in circumstellar envelopes .
water ice has been observed in a significant sample of agb stars - in particular in oh / ir stars - some post - agb stars and a few pn @xcite .
the specific conditions for the formation of water ice in circumstellar envelopes around evolved stars have been discussed and modeled by dijkstra et al .
( 2003 , 2006 ) , who also studied the trends that can be observed in the 3 , 43 and 62 @xmath2 m water ice features during the stellar evolution from the agb to the pn phase .
according to @xcite it continues to be a puzzle why some stars form water ice and not others .
one clue these authors found might be the fact that the strength of the 3 , 43 and 62 @xmath2 m water ice features increases with the increasing initial mass of the star .
their model calculations suggest that water ice features will be too weak to be detectable for stars with zero age main sequence ( zams ) masses lower than 5 m@xmath9 ; the water ice features completely disappear for initial masses lower than @xmath0 3 m@xmath9 .
a high mass loss rate also favors the detectability of the water ice features , as well as large values of the mass loss rate to luminosity ratio .
the absence of a strong interstellar uv radiation field also preserves ices , as well as the presence of a high density region that can provide shielding from an energetic radiation field .
both crystalline and amorphous water ice can form in circumstellar envelopes .
the models developed by @xcite show that the shape of the spectral features is very sensitive to the type of water ice aggregate formed .
the 3.1 @xmath2 m crystalline water ice absorption feature is characterized by a sharp core with two unequally strong shoulders . on the other hand , the amorphous water ice absorption feature is broader , with no substructures , and displaced to shorter wavelengths . in the 30100 @xmath2 m region
crystalline water ice shows two prominent and very broad emission features near 43 and 62 @xmath2 m . in the case of amorphous water ice ,
these latter features are undetectable because they are broad and show almost no contrast with the dust continuum .
figure 4 details the 3.1 @xmath2 m water ice band in iras 17423@xmath11755 and in iras 19343@xmath42926 . in both cases
the band presents a sharp profile and shows the characteristic shoulders on both sides of the central core , indicating that most of the water ice formed must be crystalline .
this is demonstrated by the good match that can be appreciated among the observed profiles and the model displayed in the figure , calculated for crystalline water ice deposited at a dust excitation temperature of 150 k by @xcite .
additionally , the _ iso _ spectrum of iras 19343@xmath42926 shows the presence of the water ice bands at 62 and 43 @xmath2 m @xcite , confirming the crystalline nature of the water ice in this star .
the presence of these latter features can not be excluded in the case of iras 17423@xmath11755 because of the poor quality of the _ iso _ lws spectrum ( see figure 2 ) .
a much higher s / n spectrum would be needed in order to confirm / discard the detection of the crystalline water ice features at the longer wavelengths .
the water ice features are providing us with valuable information for interpreting the precise evolutionary stage of these nebulae .
this point will be addressed in section 4 together with the analysis of other observational properties such as the nebular morphology and the presence of amorphous silicates .
the available _ hst _ optical images for iras 17423@xmath11755 ( bobrowsky et al . 1995 ) and iras 19343@xmath42926 @xcite were retrieved from the _ hst _ archive in order to compare the morphological properties of both objects .
these optical images are presented in figure 5 , together with the _ hst _ image of iras 17393@xmath12727 ( unpublished ) , showing that strong and highly collimated bipolar outflows are observed in the three objects .
note that bipolar morphologies are mainly found among type i pn , which are expected to be the descendants of the evolution of the more massive agb / post - agb stars @xcite .
the models by @xcite discussed in section 3.2 have shown that the conditions for the formation and prevalence of crystalline water ices , like those observed in iras 17423@xmath11755 and iras 19343@xmath42926 , would imply high initial masses ( at least higher than @xmath03 m@xmath9 ) for the central star and the probable presence of high density structures such as a dusty disk or torus , which could completely obscure the central star in the optical / near - ir wavelength range .
the existence of such dusty disk / torus structures is confirmed in the optical images of both objects . @xcite
have proposed that heavily obscured high - mass precursors of pn like iras 17393@xmath12727 may be already developing strong bipolar outflows , and it is likely that a thick circumstellar disk / torus - where the crystallization of water ice could take place - is surrounding the central post - agb star .
the presence of a highly collimated bipolar outflow as well as a thick circumstellar disk / torus that completely obscures the central source in iras 17393@xmath12727 is now confirmed by the available _ hst _ images shown in figure 5 .
@xcite used the _ iso _ spectra of a sample of oh / ir stars to analyze the ir characteristics of their circumstellar dust .
they found that the optical depth of the @xmath010 and 18 @xmath2 m amorphous silicate features increases with increasing mass loss rate during the agb .. the same results were found by @xcite , who modeled the evolution of the 2200 @xmath2 m sed for a 5 m@xmath9 ( zams ) star evolving from the early agb until the early pn stage . at the end of the agb ,
the circumstellar envelope detaches from the star , becoming optically thin and causing the amorphous silicate features to disappear . this infrared evolutionary sequence from the agb to the pn stage has been observationally confirmed by @xcite using _ spitzer _ spectra of massive o - rich agb / post - agb stars .
regarding the evolutionary state of iras 17423@xmath11755 , it is likely that this star is more evolved than iras 19343@xmath42926 because the amorphous silicate absorption and the water ice band are weaker @xcite .
iras 17393@xmath12727 could be the least evolved object of the three because it has no optical / near - ir counterpart ( the central star is not detected even in the k - band ) .
however , in the case of iras 19343@xmath42926 and iras 17423@xmath11755 , the central star has already reappeared in the optical / near - ir range .
the presence of [ neii ] nebular emission could provide evidence on the onset of the circumstellar envelope ionization , although shock excitation in the high velocity outflows can not be excluded . the sed of iras 17423@xmath11755 is similar to those of extreme oh / ir agb stars , which exhibit absorption features from amorphous silicates together with crystalline silicate features that alternate between emission and absorption depending on the specific physical and chemical properties of the circumstellar dust grains @xcite .
we think that the presence of crystalline silicates in the 13 - 35 @xmath2 m region in iras 17423@xmath11755 and iras 19343@xmath42926 can not be excluded but that higher s / n spectra would be needed in order to confirm or discard such a presence . in summary , the bipolar morphology ,
the detection of oh maser emission and crystalline water ice as well as the 9.7 @xmath2 m absorption from amorphous silicates present in the _ spitzer _ spectrum indicate that iras 17423@xmath11755 is a massive o - rich post - agb star similar to iras 19343@xmath42926 and iras 17393@xmath12727 .
the three sources represent a link between massive oh / ir agb stars and bipolar type i pn .
an o - rich chemistry for the circumstellar envelope around the post - agb object iras 17423@xmath11755 is here confirmed , despite a previous classification as c - rich .
this result is based on the detection of a weak and broad 9.7 @xmath2 m amorphous silicate absorption in the high - resolution _ spitzer_/irs spectrum .
the complete lack of c - rich mid - ir features ( in particular c@xmath3h@xmath3 at 13.7 @xmath2 m ) supports our identification of the strong 3.1 @xmath2 m absorption band seen in the _ iso _ spectrum as due to water ice as well as our o - rich classification for iras 17423@xmath11755 .
iras 17423@xmath11755 , iras 19343@xmath42926 and iras 17393@xmath12727 present clear evidences of the presence of a circumstellar disk or torus , where the conditions would be very similar to those found in less evolved and more embedded oh / ir stars . a recent strong mass - loss event has been reported in the case of iras 19343@xmath42926 @xcite , which would favor this scenario .
water ice and crystalline silicates would preferentially form in the outer region of the inner torus , where low temperature conditions and shielding from the central star would allow a favorable rate of crystallization to take place .
both in iras 17423@xmath11755 and iras 19343@xmath42926 , the ice band at 3.1 @xmath2 m is sharp and presents substructures , and comparison with models of water ice growth around evolved stars @xcite allows us to confirm that the ice is mostly in a crystalline state . the morphological properties , detection of oh maser emission , and the _ spitzer_/irs spectra observed in iras 17423@xmath11755 are similar to those of the o - rich post - agb stars iras 19343@xmath42926 and iras 17393@xmath12727 , allowing us to interpret the evolutionary stage of iras 17423@xmath11755 as belonging to an intermediate stage between those oh / ir stars with extreme outflows and highly bipolar type i pn .
m. m. and a. u. acknowledge financial support from the spanish ministry of science and innovation ( micinn ) through grant aya 2009 - 14648 - 02 and from the xunta de galicia through grants incite09 e1r312096es and incite09 312191pr , all of them partially supported by e.u .
feder funds .
d. a. g. h. and a. m. acknowledge support for this work provided by the micinn under the 2008 juan de la cierva program and under grant aya-2007 - 64748 .
this work is based on spectra obtained with _ iso _ , an esa project with instruments funded by esa member states ( especially the pi countries : france , germany , the netherlands and the uk ) and with participation of isas and nasa ; and _ spitzer _ , a nasa s great observatories program .
alcolea , j. , neri , r. , & bujarrabal , v. 2007 , a&a , 468 , l41-l44 baratta , g. a. , spinella , f. , leto , g. , strazzulla , g. , & foti , g. 1991 , a&a , 252 , 421 bobrowsky , m. , zijlstra , a. a. , grebel , e. k. , tinney , c. g. , te lintel hekkert , p. , van de steene , g. c. , likkel , l. , & bedding , t. r. 1995 , apj , 446 , l89 bunzel , f. , garca -hernndez , d. a. , engels , d. , perea - caldern , j. v. , & garca - lario , p. 2009 , in asp conf .
418 , akari , a light to illuminate the misty universe , ed .
t. osaka et al .
( san francisco , ca : asp ) , 431 cernicharo , j. , yamamura , i. , gonzlez - alfonso , e. , de jong , t. , heras , a. , escribano , r. , & ortigoso , j. 1999 , apj , 526 , l41 cernicharo , j. , heras , a. m. , tielens , a. g. g. m. , pardo , j. r. , herpin , f. , gulin , m. , & waters , l. b. f. m. apj , 546 , l123 cerrigone , l. , hora , j. l. , umana , g. , & trigilio , c. 2009 , apj , 703 , 385 corradi , r. l. m. , & schwarz , h. e. 1995 , a&a , 293 , 871 de graauw , t. , haser , l. n. , beintema , d. a. , roelfsema , p. r. , van agthoven , h. , barl , l. , bauer , o. h. , bekenkamp , h. e. g. et al .
1996 , a&a , 315 , l49 dijkstra , c. , dominik , c. , hoogzaad , s. n. , de koter , a. , & min , m. 2003 , a&a , 401 , 599 dijkstra , c. , dominic , c. , bouwman , j. , de koter , a. 2006 , a&a , 449 , 1101 eiroa , c. , hefele , h. , & qian , z .- y .
1983 , a&as , 54 , 309 eiroa , c. & hodapp , k .- w .
1989 , a&a , 223 , 271 gauba , g. & parthasarathy , m. 2004 , a&a , 417 , 201 garca - hernndez , d. a. , perea - caldern , j.v . , bobrowsky , m. , & garca - lario , p. 2007 ,
apj , 666 , l33 garca - hernndez , d. a. , bunzel , f. , engels , d. , perea - caldern , j. v. , & garca - lario , p. 2009
, in asp conf .
418 , akari , a light to illuminate the misty universe , ed . t. osaka et al .
( san francisco , ca : asp ) , 435 garca - lario , p. , manchado , a. , ulla , a. , manteiga , m. 1999 , apj , 513 , 941 herwig , f. 2005 , ara&a , 43 , 435 higdon , s. j. u. , devost , d. , higdon , j. l. , brandl , b. r. , houck , j. r. , hall , p. , barry , d. , charmandaris , v. et al . 2004 ,
pasp , 116 , 975 hoogzaad , s. n. , molster , f. j. , dominik , c. , waters , l. b. f. m. , barlow , m. j. , & de koter , a. 2002 , a&a , 547 ivezi , z. , & elitzur , m. 1997 , mnras , 287 , 799 kouchi , a. , yamamoto , t. , kozasa , t. , kuroda , t. , & greenberg , j. m. 1994 , a&a , 290 , 1009 kraemer , k. e. , sloan , g. c. , price , s. d. , walker , h. j. 2002 , apjs , 140 , 389 leger , a. , klein , j. , de cheveigne , s. , guinet , c. , defourneau , d. , & belin , m. 1979 , a&a , 79 , 256 leger , a. , gauthier , s. , defourneau , d. , & rouan , d. 1983 , a&a , 117 , 164 molster , f. j. , lim , t. l. , sylvester , r. j. , waters , l. b. f. m. , barlow , m. j. , beintema , d. a. , cohen , m. , cox , p. et al . , 2001 , a&a , 372 , 165 omont , a. , forveille , t. , moseley , s. h. , glaccum , w. j. , harvey , p. m. , likkel , l. , loewenstein , r. f. , & lisse , c. m. 1990 , apj , 355 , l27 parthasarathy , m. , & pottasch , s. r. 1989 , a&a , 225 , 521 riera , a. , garca - lario , p. , manchado , a. , pottasch , s. r. , & raga , a. c. 1995 , a&a , 302 , 137 snchez - contreras , c. & sahai , r. 2001 , apj 553,l173 smith , r. g. , sellgren , k. , & tokunaga , a. t. 1989 , apj 344 , 413 smith , r. g. , robinson , g. , hyland , a. r. , & carpenter , g. l. 1994 , mnras , 271 , 481 sylvester , r. j. , kemper , f. , barlow , m. j. , de jong , t. , waters , l. b. f. m. , tielens , a. g. g. m. , & omont , a. 1999 , a&a , 352 , 587 trammell , s. r. , & goodrich , r. w. 1996 , apj , 468 , l107 van der veen , w. e. c. j. , & habing , h. j. 1988 , a&a , 195 , 125 van winckel , h. 2003 , ara&a , 41 , 391 2 to 180 @xmath2 m for iras 17423@xmath11755 and iras 19343@xmath42926 .
the seds were constructed by combining _
spitzer _ and _ iso _ data ( see the text for details ) .
the blackbody curves fitted to the spectra of both stars are also shown , corresponding to dust temperatures of 120 k and 965 k for iras 17423@xmath11755 , and 107 k and 800 k for iras 19343@xmath42926 .
the spectral region around 3 @xmath2 m is detailed in the small window.,scaledwidth=95.0% ] | the high - resolution ( _ r_@xmath0600 ) _ spitzer_/irs spectrum of the bipolar proto planetary nebula ( pn ) iras 17423@xmath11755 is presented in order to clarify the dominant chemistry ( c - rich versus o - rich ) of its circumstellar envelope as well as to constrain its evolutionary stage . the high quality _
spitzer_/irs spectrum shows weak 9.7 @xmath2 m absorption from amorphous silicates .
this confirms for the first time the o - rich nature of iras 17423@xmath11755 in contradiction to a previous c - rich classification , which was based on the wrong identification of the strong 3.1 @xmath2 m absorption feature seen in the _ infrared space observatory _ ( _ iso _ ) spectrum as due to acetylene ( c@xmath3h@xmath3 ) . the high - resolution _
spitzer_/irs spectrum displays a complete lack of c - rich mid - ir features such as molecular absorption features ( e.g. , 13.7 @xmath2 m c@xmath3h@xmath3 , 14.0 @xmath2 m hcn , etc . ) or the classical polycyclic aromatic hydrocarbon infrared emission bands .
thus , the strong 3.1 @xmath2 m absorption band toward iras 17423@xmath11755 has to be identified as water ice .
in addition , an [ neii ] nebular emission line at 12.8 @xmath2 m is clearly detected , indicating that the ionization of its central region may be already started .
the spectral energy distribution in the infrared ( @xmath02@xmath1200 @xmath2 m ) and other observational properties of iras 17423@xmath11755 are discussed in comparison with the similar post - asymptotic giant branch ( agb ) objects iras 19343@xmath42926 and iras 17393@xmath12727 .
we conclude that iras 17423@xmath11755 is an o - rich high - mass post - agb object that represents a link between oh / ir stars with extreme outflows and highly bipolar pn . |
in the last 3 decades , knot theory has found many important applications in biology @xcite .
more often than not , in such applications , a knot can no longer be treated as a volumeless simple closed curve in @xmath8 as in classical knot theory .
instead , it has to be treated as a rope like object that has a volume .
for example , it has been reported that various knots occur in circular dna extracted from bacteriophage heads with high concentration and it has been proposed that these ( physical ) knots can be used as a probe to investigate how dna is packed ( folded ) inside a cell @xcite .
such applications motivate the study of thick knots , namely knots realized as closed ( uniform ) ropes of unit thickness .
an essential issue here is to relate the length of a rope ( with unit thickness ) to those knots that can be tied with this rope . to define the ropelength of a knot
, one has to define the thickness of the knot first .
there are different ways to define the thickness of a knot , see for example @xcite . in this paper
, we use the so called _ disk thickness _ introduced in @xcite and described as follows .
let @xmath9 be a @xmath10 knot .
a number @xmath11 is said to be _ nice _ if for any distinct points @xmath12 , @xmath13 on @xmath9 , we have @xmath14 , where @xmath15 and @xmath16 are the discs of radius @xmath17 centered at @xmath12 and @xmath13 which are normal to @xmath9 . the _ disk thickness _ of @xmath9 is defined to be @xmath18 is nice@xmath19 .
it is shown in @xcite that the disk thickness definition can be extended to all @xmath20 curves .
therefore , we restrict our discussions to such curves in this paper .
however , the results obtained in this paper also hold for other thickness definitions with a suitable change in the constant coefficient . for any given knot type @xmath0 , a _ thick realization _
@xmath9 of @xmath0 is a knot @xmath9 of unit thickness which is of knot type @xmath0 .
the _ ropelength _
@xmath1 of @xmath0 is the infimum of the length of @xmath9 taken over all thick realizations of @xmath0 .
the existence of @xmath1 is shown in @xcite .
the main goal of this paper is to establish an upper bound on @xmath1 in terms of @xmath2 , the minimum crossing number of @xmath0 .
it is shown in @xcite that there is a constant @xmath3 such that for any @xmath0 , @xmath21 .
this lower bound is called the _ three - fourth power law_. this three - fourth power law is shown to be achievable for some knot families in @xcite .
that is , there exists a family of ( infinitely many ) knots @xmath22 and a constant @xmath23 such that @xmath24 as @xmath25 and @xmath26 . on the other hand
, it is known that the three - fourth power law does not hold as the upper bound of ropelengths .
in fact , it is shown in @xcite that there exists a family of infinitely many prime knots @xmath27 such that @xmath28 as @xmath25 and @xmath29 . that is , the general upper bound of @xmath1 in terms of @xmath2 is at least of the order @xmath30 .
in @xcite it is shown that the upper bound of @xmath1 is of the order @xmath30 for any conway algebraic knot .
the family of conway algebraic knots is a very large knot family that includes all 2-bridge knots and montesinos knots as well as many other knots .
the approach used in @xcite is in fact a simpler version of the divide - and - conquer techniques used in this article .
the best known general upper bound of @xmath1 until now is of order @xmath31 , which was obtained in @xcite .
it remains an open question whether @xmath30 is the general ropelength upper bound for any knot @xmath0 . in this paper , we prove that the general upper bound of @xmath1 is almost linear in terms of @xmath2 .
more specifically , it is established that there exists a positive constant @xmath32 such that @xmath33 for any knot @xmath0 .
this is accomplished by showing that a minimum projection of @xmath0 can be embedded in the cubic lattice as a planar graph in such a way that the total length of the embedding is of the order at most @xmath34 and that the original knot can be recovered by some local modifications to this embedding without significantly increasing the total length of this embedding .
the construction of the embedding heavily relies on a divide - and - conquer technique that is based on separator theorems for planar graphs @xcite , but many new concepts and results are also needed in this quest .
overall , this is a rather complicated ( at least technically ) task that requires attention to many technical details .
the rest of this paper is organized as follows . in section [ sec2 ]
we introduce the basic concepts of topological plane graphs , cycle cuts and vertex cuts of plane graphs , and some important graph theoretic results concerning these cuts ( separator theorems of planar graphs due to miller @xcite ) . in section [ sec3 ]
, we introduce the concept of a class of special plane graphs called brt - graphs .
these are the plane graphs we use as basic building blocks when we subdivide knot diagrams . in section [ sec5 ] , we apply the separator theorems to show the existence of subdivisions of brt - graphs .
it is a typical divide - and - conquer technique that such subdivisions must be balanced " , that is , each subdivsion produces two smaller brt - graphs of roughly equal size .
we then show that we can apply these concepts recursively subdividing a brt - graph into smaller and smaller graphs . in section [ sec6 ]
, we introduce the concepts of two special kinds of plane graph embeddings ( called standard 3d - embedding " and grid - like embedding " ) .
these are not lattice embeddings .
the grid - like embedding is almost on the lattice and is used as a basic building block to reconstruct the subdivided graphs .
the standard 3d - embedding is used as bench mark for verifying the topology preservation of the reconstructed graphs obtained using the grid - like embedding .
section [ sec77 ] is devoted to providing detailed descriptions on how to obtain a grid - like embedding of a plane graph either directly , or indirectly from reconnecting two grid - like embeddings of smaller brt - graphs obtained in the subdivision process .
then in section [ sec73 ] , we show that a grid - like embedding obtained from reconstruction using our algorithm preserves the topology of the original graph .
section [ sec9 ] establishes the upper bound on the length of the embedding generated by our embedding algorithm , from which our main theorem result follows trivially .
finally , we end the paper with some remarks and open questions in section [ sec91 ] .
throughout this paper , we use the concept of topologically equivalent graphs . in this case , the vertices are points in @xmath35 and the edges are space curves that can be assumed to be piecewise smooth . if two edges are incident at a vertex or two vertices , then they intersect each other at these vertices , but they do not intersect each other otherwise . a plane ambient isotopy is defined as a homeomorphism @xmath36\rightarrow \r^2 $ ] such that @xmath37 is a homeomorphism from @xmath38 to @xmath38 for each fixed @xmath39 with @xmath40 .
a plane graph @xmath41 refers to a particular drawing of a planar graph on the plane ( with the above mentioned conditions , of course ) .
two plane graphs @xmath42 and @xmath43 are said to be _ topologically equivalent _ if there exists a plane isotopy @xmath44 such that @xmath45 .
it is possible that there are plane graphs @xmath43 that are isomorphic to @xmath42 as graphs , but not topologically equivalent to @xmath42 .
since the plane graphs of interest in this paper arise as knot or link projections with the over / under strand information ignored at the crossings , only topologically equivalent plane graphs are considered . at each vertex @xmath46 of a plane graph @xmath41 , a small circle @xmath47 centered at @xmath46
is drawn such that each edge of @xmath41 incident to @xmath46 intersects @xmath47 once ( unless the edge is a loop edge in such case the edge will intersect @xmath47 twice ) . if @xmath47 is assigned the counterclockwise orientation , then the cyclic order of the intersection points of the edges incident to @xmath46 with @xmath47 following this orientation is called the _
cyclic edge - order _ at @xmath46 .
the following lemma assures that two isomorphic plane graphs are topologically equivalent if the cyclic edge - order is preserved by the graph isomorphism at every vertex .
this fact can be easily established by induction on the order of the graph .
we leave its proof to our reader as an exercise .
[ equivalence ] let @xmath42 , @xmath43 be two isomorphic plane graphs with @xmath48 being the isomorphism . if for each vertex @xmath46 of @xmath42 , the cyclic edge - order of all edges @xmath49 , @xmath50 , ... ,
@xmath51 ( that are incident to @xmath46 ) around @xmath46 is identical to the cyclic edge - order of @xmath52 , @xmath53 , ... , @xmath54 around @xmath55 , then there exists a plane isotopy @xmath36\longrightarrow \r^2 $ ] such that @xmath56 . in other words ,
@xmath42 and @xmath43 are topologically equivalent plane graphs .
frequently , a plane graph needs to be redrawn differently while keeping its topology .
these redrawn graphs occur in rectangular boxes and throughout the paper only rectangles and rectangular boxes whose sides are parallel to the coordinate axes are used .
it is understood from now on that whenever a rectangle or a rectangular box is mentioned , it is one with such a property .
the following simple lemma is also needed later .
it can be proven using induction and the proof is again left to the reader .
[ ncurvesinrec ] let @xmath57 be a rectangle and @xmath58 , @xmath59 , ... , @xmath60 , @xmath61 , @xmath62 , ... , @xmath63 be @xmath64 distinct points in @xmath57 , then there exist @xmath6 disjoint ( piecewise smooth ) curves @xmath65 , @xmath66 , ...
, @xmath67 such that @xmath68 starts at @xmath69 and ends at @xmath70 .
furthermore , for any given simply connected region @xmath71 in the interior of @xmath57 that does not contain any of the points @xmath58 , @xmath59 , ... ,
@xmath60 , @xmath61 , @xmath62 , ... ,
@xmath63 , the curves @xmath65 , @xmath66 , ... , @xmath67 can be chosen so that they do not intersect @xmath71 .
in fact , many such regions may exist , so long as they do not intersect each other . using this simple fact , it is possible to redraw any plane graph @xmath41 in any given rectangle such that the vertices of @xmath41 are moved to a set of pre - determined points in @xmath57 .
this is stated in the following lemma .
[ redrawginr ] let @xmath41 be a plane graph with vertices @xmath72 , @xmath73 , ... ,
@xmath74 and let @xmath57 be a rectangle disjoint from @xmath41 with @xmath6 distinct points @xmath61 , @xmath62 , ... , @xmath63 chosen .
then there exists a plane isotopy @xmath44 such that @xmath75 is contained in @xmath57 and @xmath76 .
furthermore , for any given simply connected region @xmath71 in the interior of @xmath57 that does not contain any of the points @xmath70 , @xmath44 can be chosen so that it keeps @xmath71 fixed .
we give a proof for the case that @xmath77 .
the case when @xmath78 is left to the reader .
a shrinking isotopy @xmath79 is used such that @xmath80 is contained in a small rectangle @xmath81 that is small enough to be contained in @xmath57 .
then @xmath81 is moved through a translation to within @xmath57 such that the vertices @xmath69 of the resulting graph do not overlap with the @xmath70 s . by lemma [ ncurvesinrec ]
, @xmath69 can be connected to @xmath70 with a curve @xmath68 such that @xmath65 , @xmath66 , ... , @xmath67 do not intersect each other .
@xmath44 can then be obtained by deforming the plane within @xmath57 by pushing @xmath69 to @xmath70 along @xmath68 while keeping the other @xmath82 fixed , one at a time .
the following lemma is similar to the above under a different and more restrictive setting .
again this can be proven easily by induction and the proof is left to the reader .
[ fixedcurvelemma ] let @xmath41 be a plane graph drawn in a rectangle @xmath57 .
suppose that @xmath57 contains @xmath6 disjoint curves @xmath83 , @xmath84 , ... ,
@xmath85 which are not closed and are without self intersections .
moreover , these curves do not intersect the vertices of @xmath41 .
let @xmath72 , @xmath73 , ... , @xmath86 be any @xmath87 vertices of @xmath41 ( @xmath88 ) and @xmath58 , @xmath59 , ... , @xmath89 be any @xmath87 distinct points in @xmath57 that are not contained in the curves @xmath90 , then there exists a plane isotopy that is identity outside a small neighborhood of @xmath57 as well as on the @xmath90 and that takes @xmath91 to @xmath69 .
the following is a 3d variation of lemma [ redrawginr ] which is needed later .
[ straighten ] let @xmath41 be a plane graph with vertices @xmath72 , @xmath73 , ... ,
@xmath74 contained in a rectangle @xmath92 in the plane @xmath93 .
let @xmath61 , @xmath62 , ... , @xmath63 be @xmath6 distinct points in the rectangle @xmath94 in the plane @xmath95 for some number @xmath96 .
assume that @xmath70 is connected to @xmath91 by a curve @xmath97 that is strictly increasing in the @xmath98-direction such that the curves @xmath97 are disjoint , then there exists a 3d isotopy @xmath44 such that @xmath44 is level preserving in the @xmath98-direction , the identity in the space @xmath99 and outside a small neighborhood of @xmath100 $ ] , and @xmath101 is a vertical line segment ending in @xmath70 for each @xmath102 .
this is obvious if there is only one curve @xmath103 .
assume that this is true for @xmath104 .
then for @xmath105 , apply such a level preserving isotopy @xmath79 to the first @xmath87 curves .
@xmath106 is still a strictly increasing simple curve from @xmath107 to @xmath108 .
modify @xmath106 in a level preserving fashion so that its projection to the plane @xmath93 is a simple curve without self intersection .
now a plane isotopy can be constructed by a push back along this curve from @xmath107 to the projection @xmath109 of @xmath108 to the plane @xmath93 .
this does not affect the projections @xmath110 of the points @xmath111 into the plane @xmath93 for @xmath112 .
this plane isotopy can be used to define the level preserving isotopy that works for @xmath105 curves .
although there are still some technical details in the argument , it is intuitively obvious at this point and the details are left to the interested reader .
[ re22 ] if we relax the condition that the simple curves @xmath97 s are strictly increasing in the @xmath98 direction to that they are non - decreasing in the @xmath98 direction , then one can show that the result of lemma [ straighten ] holds without the requirement that the isotopy is level preserving .
it is important to note that in this case @xmath113 is still the identity and @xmath75 is topologically equivalent to @xmath41 .
a 3-dimensional ambient isotopy can be similarly defined on @xmath114 $ ] .
however we have to be much more careful about the cyclic edge order at the vertices . for a plane graph the cyclic order of the edges at a vertex
is well defined however for a graph in 3 dimensional space it is not .
we will fix this problem by requiring that for a graph in 3 dimensional space each vertex @xmath46 is contained in a small 2 dimensional disk - neighborhood @xmath115 . all the edges that terminate at @xmath46 intersect @xmath115 in a short arc that terminates at @xmath46 .
the intersection points of these arcs on @xmath115 now define the cyclic edge order at @xmath46 .
we will require that all 3-dimensional isotopies preserve this structure , that is , a 3-dimensional isotopy can move the small disk @xmath115 around in 3-space , it can even deform it , however throughout the isotopy @xmath46 remains on @xmath115 and all edges terminating at @xmath46 keep their short arcs on @xmath115 . in this way
the cyclic edge order at @xmath46 remains invariant under the 3-dimensional isotopy .
as we will see in section [ sec6 ] , we use two types of such neighborhoods called blue triangles and red square .
since all 3d - isotopies used in this paper are ambient isotopies that preserve the neighborhood structure of a vertex , we will call them _
3d vnp - isotopies _
( where vnp stands for vertex neighborhood preserving " ) .
let @xmath41 be a plane graph .
@xmath41 is called a _ weighted graph _
if each vertex , edge , and face of @xmath41 is assigned a weight ( i.e. , a non - negative number ) and the sum of these weights is 1 .
let @xmath41 be a weighted plane graph .
cycle cut _ of @xmath41 is a cycle @xmath116 in @xmath41 such that deleting all vertices in @xmath116 ( and the edges connected to them ) divides @xmath41 into two subgraphs @xmath42 , @xmath43 , on opposite sides of @xmath116 , i.e. @xmath117 . moreover , if the weights in each @xmath118 sum to no more than @xmath119 for some real number @xmath119 , @xmath120 , then the cycle cut is called an @xmath119-cycle cut . the _ size _ of the cycle cut @xmath116 is the number of vertices in @xmath116 .
the following theorems are proved in @xcite and play vital roles in the proof of the main theorem of this paper .
[ cyclecut ] @xcite let @xmath41 be a 2-connected and weighted plane graph such that no face of @xmath41 has weight more than @xmath121 , then there exists a @xmath122-cycle cut of size at most @xmath123 , where @xmath124 is the maximal face size of @xmath41 .
if one drops the assumption that @xmath41 is 2-connected then theorem [ cyclecut ] may not hold since @xmath41 may be a tree . in this case
there is the following theorem .
[ generalcut ] @xcite let @xmath41 be a connected and weighted plane graph such that all faces have been assigned weight zero , then either there exists a @xmath122-cycle cut of size @xmath123 , or there exists a cut vertex @xmath46 of @xmath41 such that each connected component in @xmath125 has a total weight less or equal to @xmath121 .
we refer to the cycle @xmath116 that generates a cycle cut as defined by theorems [ cyclecut ] and [ generalcut ] as a _ cut - cycle_.
a main tool used to achieve the ropelength upper bound obtained in this paper is the divide - and - conquer " approach familiar to researchers in graph theory .
that is , a knot projection ( treated as a 4-regular plane graph ) is divided repeatedly using the theorems given in the last section .
several non trivial issues arise in this process .
first , the graphs obtained after repeated subdivisions may become highly disconnected .
second , one needs to keep track of the topology of the subdivided piece of the graph so that the pieces resulting from the subdivisions can later be assembled correctly and the original graph can be recovered .
this section describes in detail how we handle these problems .
the aim is to impose a special structure on the graphs that is preserved after each cut and that this structure allows us to reconstruct the original graph ( with the correct topology ) from the pieces generated in the divide - and - conquer process .
let @xmath41 be a connected plane graph without loops .
we say that @xmath41 admits a _ proper br - partition _ if the vertices of @xmath41 can be partitioned into two sets @xmath126 and @xmath127 ( called _ blue _ and _ red _ vertices , respectively ) such that there is no edge between any two blue vertices .
let @xmath41 be a plane graph that admits a proper br - partition with @xmath126 and @xmath127 being the set of blue and red vertices respectively .
@xmath127 induces a subgraph @xmath128 of @xmath41 that is itself a plane graph . a connected component of @xmath128
is called a _ red component_. for a red component @xmath129 , let @xmath130 be its vertex set and let @xmath131 be the set of blue vertices that are adjacent to some vertices in @xmath130 .
let @xmath132 and let @xmath133 be the ( plane ) subgraph of @xmath41 induced by @xmath134 .
@xmath133 is called a _
br - component _ of @xmath41 .
notice that under this definition , different br - components may share common blue vertices , but each edge @xmath135 of @xmath41 belongs to exactly one br - component .
a graph is said to be _ triangulated _ if each face of the graph is either a triangle or a digon .
[ cwdef ] let @xmath41 be a connected plane graph that admits a proper br - partition .
@xmath41 is called a _
brt - graph _ if each br - component @xmath133 of @xmath41 is triangulated
. a triangulated br - component @xmath133 is called a _
brt - component_. [ l3.3 ] let @xmath41 be a brt - graph , then ( i ) the boundary of any face of @xmath41 contains at most one blue vertex and ( ii ) any cut vertex of @xmath41 is a blue vertex .
notice that condition ( ii ) above is equivalent to the following statement : for any red component @xmath129 , the brt - component @xmath133 is 2-connected .
suppose that the boundary @xmath136 of a face @xmath137 in @xmath41 contains two different blue vertices @xmath46 and @xmath138 .
let @xmath139 be a path in @xmath136 that connects @xmath46 and @xmath138 which contains at least three vertices , since @xmath139 can not be a single blue - blue edge .
we assume that @xmath46 and @xmath138 and @xmath139 were chosen so that there are only red vertices on @xmath139 besides @xmath46 and @xmath138 . @xmath46 and @xmath138 must belong to @xmath133 where @xmath129 is the red component containing the red vertices on @xmath139 .
it is easy to see that even if @xmath139 contains only a single red vertex , @xmath136 must contain at least 4 edges since @xmath46 and @xmath138 can not be connected by a single edge .
this implies that @xmath133 is not triangulated , which contradicts the given condition that @xmath41 is a brt - graph .
this proves ( i ) .
let @xmath46 be a cut vertex of @xmath41 .
then there exists a face @xmath137 of @xmath41 such that @xmath46 appears on @xmath136 at least twice , i.e. , @xmath136 can be described by a walk @xmath140 where @xmath141 and @xmath142 are closed walks in @xmath41 starting and ending at @xmath46 .
@xmath141 can not contain only one edge since that would force a loop edge .
thus @xmath141 contains at least two edges .
similarly , @xmath142 also contains at least two edges . thus there
exist vertices @xmath143 and @xmath144 on @xmath141 and @xmath142 , respectively , that are adjacent to @xmath46 .
note that @xmath145 can not be an edge in @xmath41 since @xmath46 is a cut - vertex .
let us assume that @xmath46 is a red vertex , and let @xmath129 be the red component that contains @xmath46 .
then @xmath143 and @xmath144 must also be contained in @xmath133 and must be on the same face of @xmath133 .
however this contradicts the assumption that @xmath133 is triangulated .
thus @xmath46 must be blue .
this proves ( ii ) . for a brt - graph @xmath41 ,
let us define a graph @xmath146 that is associated with @xmath41 in the following way .
each blue vertex of @xmath41 corresponds to a vertex in @xmath146 ( which is still called a blue vertex ) and each red component of @xmath128 corresponds to a vertex in @xmath146 ( which is called a red vertex ) .
vertices of the same color in @xmath146 are never adjacent and a blue vertex @xmath12 and a red vertex @xmath13 in @xmath146 are connected by a single edge if and only if the blue vertex @xmath46 in @xmath41 corresponding to @xmath12 is contained in @xmath133 where @xmath129 is the red component corresponding to @xmath13 .
the following lemma asserts that the graph @xmath146 so constructed is a tree .
[ cycleinbrt ] the graph @xmath146 defined above is a tree .
equivalently , any cycle in @xmath41 is contained in a single brt - component . by construction
the graph @xmath146 is a simple bipartite graph since two vertices of different colors can be connected by at most one edge and vertices of the same color are never connected .
moreover there can be no digon in @xmath146 by construction .
it follows that a cycle @xmath47 in @xmath146 ( if it exists ) must contain at least 4 vertices and the colors of the vertices on the cycle must alternate if one travels along @xmath47 .
assume that there is a cycle @xmath47 in the graph @xmath146 which contains a path @xmath147 where the @xmath148 are red and the @xmath111 are blue .
let @xmath129 be the red component of @xmath133 which corresponds to @xmath58 .
let @xmath61 and @xmath62 be the two blue vertices in @xmath47 that are adjacent to @xmath58 .
let @xmath72 and @xmath73 be the two blue vertices in @xmath41 that correspond to @xmath61 and @xmath62 , respectively .
the fact that @xmath61 and @xmath62 are connected to @xmath58 in @xmath146 implies that @xmath72 and @xmath73 are adjacent to some vertices in @xmath129 .
the red component @xmath149 of @xmath128 that corresponds to @xmath58 divides the plane into one outer face @xmath137 and several ( possibly zero ) inner faces . in the interior of each face of @xmath149
there are blue vertices of @xmath133 ( if any ) or vertices of the other brt - components besides @xmath149 ( if any ) .
if @xmath72 and @xmath73 are both contained in the interior of @xmath137 , then there exists a path from @xmath72 to @xmath73 of the form @xmath150 where @xmath151 and the path @xmath152 is contained in @xmath136 . since the face in @xmath153 containing this path
can not be triangulated without using blue - blue edges , this is not possible .
thus @xmath72 and @xmath73 can not be both contained in the interior of @xmath137 .
similarly , on can show that @xmath72 and @xmath73 can not be both contained in the interior of the same face for any face of @xmath149 .
therefore , one of them , say @xmath72 is contained in the interior of an inner face @xmath154 of @xmath149 and the other @xmath73 is either contained in the interior of a face @xmath155 which is either the outer face @xmath137 or a different inner face .
since the cycle @xmath47 has at least length four , there exists a second red vertex @xmath59 in @xmath47 neighboring @xmath62 .
let @xmath156 be the red component of @xmath128 corresponding to @xmath59 .
this means that @xmath73 is connected by an edge to a red vertex in @xmath156 .
since @xmath73 is contained in the interior of @xmath155 and @xmath41 is a plane graph , this implies that @xmath156 is entirely contained in the interior of @xmath155 .
similarly , continuing to move along the cycle @xmath47 in the direction established by moving from @xmath62 to @xmath59 , leads to the conclusion that all blue vertices and red components ( of @xmath41 ) corresponding to the vertices on @xmath47 are contained in the interior of @xmath155 .
this contradicts the fact that @xmath72 is not contained in the interior of @xmath155 .
thus the cycle @xmath47 can not exist .
it is easy to see that any cycle passing through more than one brt - component gives rise to a cycle in @xmath146 .
thus any cycle in @xmath41 is contained in a single brt - component of @xmath41 .
let @xmath41 be a brt - graph .
let us consider two different ways of dividing @xmath41 into subgraphs .
first , consider the case that @xmath41 has a cut - vertex @xmath46 ( recall that @xmath46 must be a blue vertex by lemma [ l3.3 ] ) .
let @xmath42 , @xmath43 , ...
, @xmath157 be the connected components of @xmath158 with @xmath159 being the set of vertices of @xmath118 .
pick an arbitrary proper subset @xmath160 of @xmath161 and let @xmath162 .
let @xmath163 and @xmath164 .
let @xmath165 be the induced subgraph of @xmath41 consisting of the brt - components whose vertices are contained in @xmath166 and @xmath167 be the induced subgraph of @xmath41 consisting of the brt - components whose vertices are contained in @xmath168 .
note that @xmath165 and @xmath167 inherit the plane graph structure naturally from @xmath41 .
in particular , the cyclic ordering of the edges around the vertex @xmath46 in each subgraph @xmath165 , @xmath167 is naturally inherited from the cyclic edges ordering around @xmath46 in @xmath41 this describes the first kind of subdivision of @xmath41 which is formally defined below .
[ subdivisionvertex ] let @xmath41 be a brt - graph with a cut - vertex @xmath46 , then dividing @xmath41 into two subgraphs @xmath165 and @xmath167 as described above is called _ subdividing @xmath41 by a vertex - cut_. before describing the second subdivision which depends on a cycle of @xmath41 , we introduce the concept of a _
normal cycle_. let @xmath169 , @xmath46 , and , @xmath138 be three vertices of @xmath41 such that @xmath170 is a triangle ( namely a cycle with three edges ) in @xmath41 .
t he triangle @xmath170 is empty if it is the boundary of a face of @xmath41 .
a cycle @xmath116 in @xmath41 is said to be _ normal _ if it contains at least two vertices and no three consecutive red vertices on @xmath116 form an empty triangle in @xmath41 .
[ cutcycleexistance ]
let @xmath41 be a brt - graph .
if @xmath41 has a cut - cycle , then it has a normal cut - cycle .
assume that @xmath139 is a cut - cycle in @xmath41 with length @xmath171 .
then @xmath171 is at least @xmath172 , since @xmath41 does not contain any loops .
if @xmath139 contains three consecutive red vertices @xmath169 , @xmath46 , @xmath138 which form an empty triangle , then remove @xmath46 and replace @xmath173 by @xmath174 .
since this operation can never reduce the number of vertices below 2 , it always leads to a normal cut - cycle with length at least 2 .
now let us describe how to use a normal cut cycle @xmath116 to divide @xmath41 .
first , @xmath116 is pushed off the red vertices by a small distance , resulting in a simple closed curve @xmath175 .
the push off @xmath175 is required to have the following properties : ( 1 ) all intersections of @xmath175 with the edges of @xmath41 must happen transversely ( except possibly at their endpoints if they are connected to a blue vertex ) ; ( 2 ) if @xmath46 is a blue vertex on @xmath116 then @xmath46 stays on @xmath175 ; ( 3 ) if @xmath135 is an edge on @xmath116 that is connected to a blue vertex @xmath46 on @xmath116 , then @xmath175 does not intersect @xmath135 ; ( 4 ) if @xmath135 is an edge on @xmath116 that is connected to two red vertices , then @xmath175 may intersect @xmath135 at most once ; ( 5 ) for an edge @xmath135 that is not on @xmath116 but is connected to exactly one red vertex on @xmath116 , @xmath175 can intersect @xmath135 at most once ; ( 6 ) for an edge @xmath135 that is not on @xmath116 but is connected to two red vertices on @xmath116 , @xmath175 can intersect @xmath135 at most twice .
( the right of figure [ digonavoid ] shows a non - trivial example of this . )
it is easy to see that such a push off is always possible .
in addition , if @xmath175 intersects @xmath135 twice then no empty digons ( an empty digon is a digon whose interior or exterior does not intersect @xmath41 ) should be created .
the empty digons can always be avoided by routing @xmath175 around the digon as shown in figure [ digonavoid ] .
note that such digons be nested and @xmath175 may have to be pushed across several digons , see figure [ digonavoid ] on the right .
a simple closed curve @xmath175 obtained from a cycle @xmath116 with these properties is called a _
push - off _ of @xmath116 .
intersects an edge twice .
left : removing empty digons by re - routing @xmath175 .
middle : an edge intersected by @xmath175 ( dashed ) twice in a non - trivial way .
right : pushing @xmath175 across several digons .
, title="fig : " ] 0.3 in intersects an edge twice .
left : removing empty digons by re - routing @xmath175 .
middle : an edge intersected by @xmath175 ( dashed ) twice in a non - trivial way .
right : pushing @xmath175 across several digons .
, title="fig : " ] 0.3 in intersects an edge twice .
left : removing empty digons by re - routing @xmath175 .
middle : an edge intersected by @xmath175 ( dashed ) twice in a non - trivial way .
right : pushing @xmath175 across several digons .
, title="fig : " ] let @xmath116 be a normal cycle of @xmath41 and let @xmath175 be a push - off of @xmath116 .
let us ( temporarily ) insert into @xmath41 a white colored vertex at each intersection point of @xmath175 with the edges of @xmath41 and a new edge for every arc on @xmath175 connecting two such white vertices or connecting a white vertex and a blue vertex .
these white vertices are considered the vertices of @xmath175 and are denoted by @xmath176 .
the arcs of @xmath177 are considered as the edges of @xmath175 .
this graph obtained by adding the edges and vertices of @xmath175 is called @xmath178 .
let @xmath179 denote the plane graph obtained from @xmath178 by deleting all vertices @xmath176 and the edges connected to these vertices .
@xmath179 is separated into two disjoint subgraphs @xmath180 and @xmath181 with @xmath180 inside of @xmath175 and @xmath181 outside of @xmath175 .
let @xmath182 be the subgraph of @xmath178 induced by the vertices @xmath183 .
similarly , let @xmath184 be the subgraph of @xmath178 induced by the vertices @xmath185 .
finally , in @xmath182 we contract @xmath175 to a single vertex and mark it as a new blue vertex @xmath72 .
the resulting plane graph is denoted by @xmath186 .
similarly , in @xmath184 we contract @xmath175 to a single vertex and mark it as a new blue vertex @xmath73 and call the resulting plane graph @xmath187 . see figures [ gcyclecut ] , [ gsubdivided ] , and [ unicycleneighborhood ] for an illustration of this process .
shown by the dashed edges .
right : the modified normal @xmath116 ( dashed ) and its push - off @xmath175 ( dashed and grey ) .
blue vertices are marked by dark circles and red vertices are marked by while circles .
the edges cut by @xmath175 are labeled in cyclic order . ]
notice that the cyclic orders of the edges connected to the newly created blue vertices @xmath72 in @xmath186 and @xmath73 in @xmath187 are inherited from the cyclic order of the intersection points of @xmath41 with @xmath175 .
thus @xmath182 and @xmath184 can be recovered from @xmath186 and @xmath187 by expanding @xmath72 or @xmath73 back to @xmath175 .
@xmath41 can be recreated by gluing @xmath182 and @xmath184 along @xmath175 with the original cyclic order of the edges along @xmath175 preserved .
notice that @xmath186 and @xmath187 may not admit a proper br - partition for two reasons .
first , @xmath186 and @xmath187 may contain loop edges .
second , the newly created blue vertices @xmath72 and @xmath73 ( from the contraction of @xmath175 ) may be adjacent to some other blue vertices in @xmath186 or @xmath187 which already exist in @xmath41 . if this happens in @xmath118 , one of the edges connecting these blue vertices to @xmath188 is simply contracted . more precisely ,
assume that @xmath138 is a blue vertex in @xmath189 that is connected by several edges @xmath190 to @xmath188 .
one of these edges is picked , say @xmath49 and contracted .
this combines @xmath138 and @xmath188 into one blue vertex - still denoted @xmath188 , and generates @xmath191 loop edges at @xmath188 , see figure [ gsubdivided ] .
after all blue - blue edges ( that are not loop edges ) have been eliminated in this way , the resulting ( plane ) graphs are denoted by @xmath192 and @xmath193 , the @xmath194 indicating that the graphs may contain loop edges .
a loop edge , after it is created , is never cut again , nor does it influence any further subdivisions . thus there is no reason to keep these loop edges in the plane graphs for the future subdivision process . after deleting the loop edges from @xmath192 or @xmath193 ,
the resulting graphs are denoted by @xmath42 and @xmath43 .
information about @xmath175 , the contracted edges , and the deleted loops which are not included in @xmath42 or @xmath43 are kept as described later ( see definition [ glueinginstructions ] ) . only @xmath42 and @xmath43 are used in the subsequent subdivisions . by @xmath116 .
left : @xmath42 obtained from the subgraph inside @xmath175 .
middle : the graph @xmath187 obtained from the subgraph outside @xmath175 .
right : @xmath43 , obtained by contracting edge 4 in @xmath187 . the loop edge created by this contraction ( dashed )
is deleted from @xmath43 . ]
[ subdivisioncycle ] let @xmath41 be a brt - graph with a normal cycle @xmath116 , then dividing @xmath41 into two plane graphs @xmath42 and @xmath43 using a push - off of @xmath116 as described above is called _ subdividing @xmath41 by an edge - cut_. after a brt - graph @xmath41 is subdivided into two new plane graphs as described in the above two definitions , are the newly obtained graphs also brt - graphs ?
this is not obvious in the case of an edge - cut subdivision since some original brt - components may have been modified by the subdivision process . the answer to this is affirmative and established in lemma [ splittinglemma ] below .
[ splittinglemma ] let @xmath41 be a brt - graph .
if @xmath42 and @xmath43 are the two graphs obtained after a vertex - cut subdivision or an edge - cut subdivision is applied to @xmath41 , then @xmath42 and @xmath43 are also brt - graphs . in the case
that @xmath42 and @xmath43 are obtained by a vertex - cut subdivision the lemma is obvious .
thus we concentrate on the case of an edge - cut subdivision . by construction the new graphs @xmath42 and
@xmath43 do not have blue - blue edges .
furthermore , if @xmath41 is connected , then @xmath42 and @xmath43 are also connected .
let @xmath116 be the cycle used in this edge - cut subdivision .
by lemma [ cycleinbrt ] , @xmath116 is contained in a single brt - component @xmath133 for some red component @xmath129 in @xmath41 .
obviously , all other brt - components of @xmath41 remain unchanged in the subdivision process .
these brt - components remain as triangulated brt - components in either @xmath42 or @xmath43 .
let @xmath195 be a red component of @xmath42 that contains some vertices of @xmath129 . to show that @xmath196 is triangulated ,
it is first shown that all faces in @xmath186 or @xmath187 created in the process of contracting @xmath175 are either triangles or digons .
note that @xmath186 or @xmath187 may contain loop edges and those are addressed later . to show that after contracting @xmath175 the resulting graph is still triangulated
, the contraction of the edges on @xmath175 is considered one edge at a time .
let @xmath137 be a face of @xmath133 and let @xmath136 be its boundary . if the interior of @xmath137 does not intersect @xmath175 , then @xmath137 is not affected by the contraction process of @xmath175 and it remains a triangle or a digon after the subdivision .
if the interior of @xmath137 and @xmath175 intersect each other , then @xmath136 must contain at least one vertex on @xmath116 .
let @xmath135 be an edge of @xmath175 that intersects @xmath137 and assume that @xmath135 splits @xmath137 into two faces @xmath154 and @xmath155 in @xmath197 and @xmath198 , respectively .
@xmath135 can intersect @xmath136 in two white vertices ( i.e. in two different edges ) or @xmath135 can intersect @xmath136 in one white and one blue vertex ( i.e. in one edge and the blue vertex of @xmath137 ) .
it is easy to see that regardless of whether @xmath137 is a digon or a triangle and regardless of the particular location of these intersections the contraction process changes @xmath154 and @xmath155 into loops , digons , or triangles , see figure [ brtpreservation ] . on @xmath175 can intersect the interior of a triangle or a digon .
the three cases for a triangle are on the left , the two cases for a digon are on the right .
the resulting faces after @xmath175 ( marked by the dashed line ) is contracted to the new blue vertex are on the right side of the arrows . ] at this point we have established that after the contraction of @xmath175 all faces that are changed by the subdivision remain triangulated in @xmath186 and @xmath187 . to obtain @xmath192 and @xmath193 ,
it is necessary to contract some of the blue - blue edges that may have been created when the new blue vertices were introduced .
however , the contraction of an edge in a graph does not increase the size of any face , and thus the triangulation property of all affected faces is preserved .
the final step to obtain @xmath42 and @xmath43 from @xmath192 and @xmath193 is to delete all loop edges that may have been created in the contraction process of @xmath175 and some of the blue - blue edges .
let @xmath135 be a loop edge created by this contraction connected to a blue vertex @xmath46 .
let @xmath154 and @xmath155 be the two faces on the different sides of the loop edge @xmath135 .
if both @xmath199 and @xmath200 contain vertices different from @xmath46 then these vertices are red and belong to two different brt - components ( after the cut ) .
therefore the face @xmath201 created by deleting @xmath135 does not have to be triangulated and @xmath135 can be deleted .
if one , say @xmath199 contains no other vertices ( that is @xmath199 is consists of one or two edges from @xmath46 to @xmath46 ) then the deletion of @xmath135 causes the number of edges in @xmath200 to remain the same or to decrease by one and thus can not violate the triangulation property of @xmath155 .
therefore all brt - components in @xmath42 and @xmath43 remain triangulated . after a brt - graph @xmath41
is subdivided into brt - graphs @xmath42 and @xmath43 , it may no longer be possible to reconstruct @xmath41 from @xmath42 and @xmath43 unless information about the subdivision step is kept . for each blue vertex @xmath46 in @xmath41 which is not involved in the subdivision process ,
the order of its edges in @xmath41 is the order of the edges of the corresponding blue vertex in @xmath42 or @xmath43 and no additional information must be kept .
the information about the blue vertices involved in the subdivision process is captured in a small neighborhood @xmath195 .
@xmath195 includes information about the edges involved in the subdivision and their relative order .
in addition , for an edge - cut , the neighborhood contains information about the edges contracted ( if any ) and the loop edges ( if any ) temporarily created during the subdivision .
all loop edges are deleted in the final step of the process that changes @xmath202 to @xmath118 .
notice that all the edges in @xmath41 are labeled at the beginning and these labels do not change in the subdivision process .
thus the edges in @xmath42 , @xmath43 , and @xmath195 share the same label if and only if they are part of the same edge in @xmath41 , therefore the labeled graphs @xmath42 and @xmath43 together with @xmath195 contain all the information needed to reconstruct @xmath41 ( since lemma [ equivalence ] can then be applied to the reconstructed graph ) .
the detailed information stored in @xmath195 is different for a circular edge - cut and for a vertex cut .
\(i ) for a circular edge - cut , @xmath195 is an annulus which contains a small neighborhood of @xmath203 , where @xmath204 is the set of edges of @xmath178 that are incident to both a white and a blue vertex and that are contracted in the steps of the subdivision process which changes @xmath205 to @xmath206 , see figure [ unicycleneighborhood ] for an example .
the outside and inside boundaries of @xmath195 are used to keep track of the cyclic order of the edges of @xmath42 and @xmath43 ( as well as the deleted loop edges ) around their corresponding blue vertices created by the edge - cut . for reasons discussed in section [ sec6 ] and in subsection [ sec7 ]
we impose a _ linear order _ on the edges around a blue @xmath207 at the time when it is created in the subdivision process .
this is accomplished by identifying a path @xmath208 in @xmath195 that connects the inside and outside boundary of @xmath195 without intersecting any of the edges of @xmath41 .
cutting @xmath195 along @xmath208 results in the linear order of the edges for @xmath209 and @xmath210 inherited from the counterclockwise orientation on each of the boundary components of @xmath195 , see figure [ unicycleneighborhood ] . from now on ,
it is understood that the linear order at a blue vertex is so defined if the vertex is created by a circular edge - cut subdivision of a brt - graph .
a loop may be created and becomes part of @xmath202 in two situations .
first , if an edge @xmath135 not on @xmath116 is cut twice by @xmath175 , ( see the right diagram in figure [ digonavoid ] for an example ) a loop is created in the @xmath202 which does not contain the vertices incident to @xmath135 . in this case
the same edge label appears twice on each boundary component of @xmath195 .
second , if @xmath87 red - blue edges @xmath190 with @xmath211 which are incident to a blue vertex @xmath169 are cut by @xmath175 ( see figures [ gcyclecut ] and [ gsubdivided ] for an example ) , then @xmath169 is contracted into the @xmath207 associated with the @xmath189 which contains @xmath169 and @xmath191 loops are created .
the loops are in the @xmath202 associated with that @xmath207 . here
the same edge label appears three times in @xmath212 for each of these loops : twice on the boundary of the component which contains the loop edge before its deletion and once on the other .
loops are not included in @xmath118 .
however , to enable a correct reconstruction of @xmath41 the information about how the loop connections interleave with other edges cut by @xmath175 is kept by the edge labels on @xmath212 . of the edge - cut subdivision of figures [ gcyclecut ] and [ gsubdivided ]
as it arises in @xmath41 ; center : @xmath195 deformed into an annulus together with the path @xmath175 , the orientations on @xmath212 and the path @xmath208 used to establish the linear order ; right : @xmath195 ( cut open along @xmath208 ) deformed into a rectangle . ]
\(ii ) for a vertex - cut subdivision , @xmath195 is a disk which contains a small neighborhood of @xmath46 , the cut - vertex used for the subdivision .
once a single point @xmath208 that does not belong to any edge on the boundary of @xmath195 is chosen , the linear order of the edges around the new blue vertices @xmath213 and @xmath214 is inherited from the cyclic order of @xmath46 . from now on
, it is understood that the linear order is so defined at a blue vertex created by a vertex - cut subdivision of a brt - graph , see figure [ vertexcut ] .
note that we can indicate the linear order of the edges at a blue vertex @xmath46 in @xmath195 by a small circular arrow @xmath215 around a blue vertex .
the edge on which the tail of @xmath215 is placed indicates the first edge of the linear order and the arrow head points into the direction of that linear order . even though we could think of a brt - graph @xmath41 as a graph where every blue vertex has an arrow indicating a linear order we are only interested in assigning a linear order ( with an arrow ) resulting from subdivsions .
when we apply an embedding algorithm to a brt - graphs later in this paper , every blue vertex originated from a subdivision process and has an assigned linear order .
the linear order of the blue vertices in a neighborhood @xmath195 is included in @xmath195 .
for an example see the two small arrows around two blue vertices in figure [ unicycleneighborhood ] .
note that we did not include the small arrows at the blue vertices in figures [ gcyclecut ] and [ gsubdivided ] , since the linear orders were not relevant to our discussions at that time . with a vertex - cut at a blue vertex @xmath46 .
the gray shaded ellipses are the different brt - components that contain the vertex @xmath46 .
these brt - components can be of any size ; middle : the graphs @xmath42 and @xmath43 after the vertex - cut subdivision ; right : the neighborhood @xmath195 of the blue vertex @xmath46 containing the gluing instruction , the edges of @xmath42 are dashed .
the arrow on @xmath212 shows the orientation and the point @xmath208 is used to define the linear order of the edges . ]
[ glueinginstructions ] the small neighborhood @xmath195 addressed above , together with the labels of the edges of @xmath42 , @xmath43 , is called _ the gluing instruction _ of the corresponding subdivision of @xmath41 . the orientation of a component of @xmath212 is called the _ orientation vector _ of the boundary component . the information provided by @xmath42 , @xmath43 , and the gluing instruction suffices to reconstruct the graph @xmath41 uniquely ( up to a plane isotopy ) by observing that the position of the loop edges ( up to a plane isotopy ) in @xmath206 can be derived from the given information .
for each such loop edge the gluing instruction determines a unique vertex and a unique face that must contain the loop edge and therefore the position of the loop edge is unique ( up to a plane isotopy ) .
the reconstruction of @xmath41 is a reversed process of the subdividing and contracting ( used to obtain @xmath42 and @xmath43 ) : deleted loop edges are first glued back to @xmath42 and @xmath43 , the contracted edges ( if any ) between two blue vertices @xmath213 and @xmath214 are expanded back , then the blue vertices in @xmath42 and @xmath43 resulting from the contraction of @xmath175 are expanded back to a closed curve equivalent to @xmath175 and the edges cut by @xmath175 are glued back together in the last step .
this reversed process is made possible since all the information needed is stored in the gluing instruction .
we summarize this in the following lemma .
[ reconstructionlemma ] let @xmath41 be a brt - graph and let @xmath42 and @xmath43 be the two brt - graphs obtained by a vertex - cut subdivision or an edge - cut subdivision of @xmath41 .
then the planar embeddings @xmath42 and @xmath43 induced from @xmath41 together with the gluing instruction that arise from this subdivision allow a reconstruction of a graph that is plane isotopic to the original graph @xmath41 .
in the last section , it was shown that subdividing a brt - graph @xmath41 by a vertex - cut or by an edge - cut ( based on a normal cut - cycle ) results in two brt - graphs @xmath42 and @xmath43 .
the main task of this section is to show that it is possible to subdivide a brt - graph @xmath41 such that the sizes of @xmath42 and @xmath43 are balanced .
recall from the definition of the subdivision that a red vertex of @xmath41 remains a red vertex in one of @xmath42 and @xmath43 ( but not both ) and no new red vertices are created in the process . that is , if @xmath42 and @xmath43 are the graphs obtained from @xmath41 by subdivision with @xmath216 and @xmath217 being the sets of red vertices respectively , then @xmath127 is the disjoint union of @xmath216 and @xmath217 . for a brt - graph @xmath41 ,
let us define its _ standard weight _ @xmath218 as the number of its red vertices , i.e. , @xmath219 . if @xmath41 is subdivided into @xmath42 and @xmath43 by a vertex - cut or an edge - cut , then @xmath220 . in this section
non - standard weight systems are used which are denoted by a lower case @xmath138-function .
let @xmath41 be a brt - graph and let @xmath221 be a constant independent of @xmath41 .
a subdivision of @xmath41 into @xmath42 and @xmath43 ( by either a vertex - cut or an edge - cut ) is _ balanced _ if @xmath222 and in the case that the subdivision is an edge - cut subdivision , the length of the normal cycle ( i.e. , the number of red vertices in the cycle ) used for the edge - cut is at most @xmath223 . in a general brt - graph
, the number of blue vertices may not be bounded above by a function of the number of red vertices as shown in figure [ bluevertexbound ] . however ,
if the degrees of the red vertices of a brt - graph @xmath41 are bounded above by a constant @xmath224 , then the number of blue vertices in @xmath41 is related to the number of red vertices in @xmath41 as shown in lemma [ vbbound ] .
[ vbbound ] let @xmath41 be a brt - graph with @xmath225 and @xmath226 be an upper bound of the degrees of the red vertices , then @xmath227 .
consequently , @xmath228 .
it suffices to prove the inequality for a brt - component @xmath133 of @xmath41 . deleting all but one edge from each set of multiple edges connecting the same two vertices in @xmath133 results in a simple graph @xmath229 with the same number of blue and red vertices as that of @xmath133 .
( here we think of multiple edges as edges that create an empty digon and not just edges that have the same end vertices .
for example , the graph in figure [ bluevertexbound ] does not have a multiple edge . )
now @xmath229 has only triangular faces .
let @xmath230 ( @xmath231 ) be the number of red ( blue ) vertices in @xmath229 and let @xmath232 be the number of faces in @xmath229 .
the boundary of each triangular face contains at least 2 red vertices and any red vertex can be on the boundaries of at most @xmath226 different faces .
thus @xmath230 is bounded below by @xmath233 . on the other hand ,
each blue vertex is on the boundaries of at least two faces and the boundary of each face contains at most one blue vertex ( lemma [ l3.3 ] ) .
so @xmath234 .
[ balancelemma ] let @xmath224 be a given constant .
there exists a constant @xmath235 such that for any brt - graph @xmath41 with @xmath236 and the maximum degree of the red vertices in @xmath41 being @xmath237 , there exists a balanced subdivision of @xmath41 .
first consider the case that @xmath41 contains a brt - component @xmath133 such that @xmath238 .
( note that it is possible that @xmath239 . )
a non - standard weight system @xmath138 is assigned to @xmath133 as follows .
let @xmath240 be the number of red vertices in @xmath41 .
each red vertex in @xmath133 is assigned weight @xmath241 .
all blue vertices in @xmath133 are assigned weight zero . each face @xmath232 of @xmath133
is assigned a weight @xmath242 , where @xmath243 is the number of red vertices of @xmath244 that are contained in @xmath232 .
the total weight is equal to 1 since every red vertex of @xmath41 is either in @xmath129 or is contained in a face of @xmath133 . since @xmath245 , no face of @xmath133 has weight larger than @xmath246
under this non - standard weight assignment , theorem [ cyclecut ] implies that there exists a @xmath122-cycle cut that divides @xmath133 ( hence @xmath41 ) into two subgraphs .
moreover the length of the cycle @xmath116 used is at most @xmath247 where @xmath248 by lemma [ vbbound ] .
if @xmath116 is not normal , then it can be modified into a normal cut - cycle in @xmath133 by lemma [ cutcycleexistance ] . the normal cut - cycle @xmath83 so obtained is shorter than @xmath116 , and each of the two subgraphs separated by it has at most @xmath249 red vertices . now choose @xmath250 to be a constant large enough so that @xmath251 holds for every @xmath252 .
next consider the case that every brt - component @xmath133 has a standard weight @xmath253 ( where @xmath240 ) .
let @xmath146 be the tree defined in section [ sec3 ] ( before lemma [ cycleinbrt ] ) .
a red vertex @xmath254 of @xmath146 that corresponds to a red component @xmath129 is assigned the weight @xmath255 .
all blue vertices of @xmath146 are assigned weight zero .
notice that under this weight assignment , the total weight is 1 .
thus by theorem [ generalcut ] there exists a cut - vertex @xmath46 in @xmath146 such that each connected component in @xmath256 has a total weight less than or equal to @xmath121 .
if @xmath46 is a blue vertex in @xmath146 , then the blue vertex @xmath169 in @xmath41 corresponding to @xmath46 is a cut vertex and can apparently be used to obtain a balanced vertex - cut subdivision of @xmath41 .
( a vertex - cut using the cut vertex @xmath169 obtained in this manner in facts leads to connected components each of which has a weight of @xmath257 or less . ) on the other hand , if @xmath46 is a red vertex then it corresponds to a red component @xmath129 of @xmath128 .
assign @xmath133 the non - standard weight system @xmath143 as before : each red vertex in @xmath129 is assigned the weight @xmath241 , each blue vertex in @xmath133 is assigned weight zero , and each face @xmath232 of @xmath133 is assigned the weight @xmath242 , where @xmath243 is the number of red vertices of @xmath244 that are contained in @xmath232 .
again the total weight is 1 since every red vertex of @xmath41 is either in @xmath129 or is contained in a face of @xmath133 .
no face @xmath232 in @xmath133 has a weight @xmath258 since otherwise deleting the red vertex @xmath46 in @xmath146 corresponding to @xmath129 results in a connected component in @xmath146 with weight @xmath259 , contradicting the given property of @xmath46 .
thus by theorem [ cyclecut ] there exists a cycle @xmath116 in @xmath133 that yields a @xmath122-cycle cut of @xmath41 .
again modify @xmath116 as before to obtain a normal cycle @xmath84 and use @xmath84 to obtain an edge - cut subdivision of @xmath41 .
the only difference is that this time @xmath84 causes a smaller bound on @xmath260 since the weight of @xmath133 is less than @xmath261 , so the total weight of each of the two graphs obtained by the edge - cut using @xmath84 is bounded above by @xmath262 where @xmath263 is the number of vertices in @xmath133
. the definition of a ( balanced ) vertex - cut may allow many different choices for @xmath42 and @xmath43 by choosing different unions of brt - components . in order to allow a successful reconstruction ,
constraints are imposed on the selection of the brt - components for @xmath118 for a balanced vertex - cut .
lemma [ diskcomponent ] specifies these constraints and asserts that they can always be met .
assume that @xmath41 is a brt - graph and @xmath46 is a blue vertex in @xmath41 that can be used for a balanced vertex - cut .
let @xmath119 be an arc that starts and ends at @xmath46 and is otherwise disjoint from @xmath41 .
@xmath119 separates @xmath41 into two subgraphs @xmath42 and @xmath43 both containing @xmath46 and that are unions of complete brt - components .
we call @xmath42 ( and @xmath43 ) a _ disk - component _ of @xmath41 .
it is possible that one of the two graphs contains only the vertex @xmath46 .
however , if both @xmath42 and @xmath43 contain at least one vertex other than @xmath46 , then @xmath42 and @xmath43 are called _ proper disk - components _ of @xmath41 . a disk - component
@xmath229 is _ separable _ if there exist two proper disk - components @xmath264 and @xmath265 such that @xmath266 and @xmath264 lies in the outer face of @xmath265 .
@xmath229 is called _ inseparable _ if it is not separable .
@xmath267 is a _
maximal disk - component _ of @xmath229 if @xmath267 is a proper disk - component contained in @xmath229 and if for every disk - component @xmath268 in @xmath229 that contains @xmath267 either @xmath269 or @xmath270 .
see figure [ maxdisks ] . .
] [ diskcomponent ] let @xmath41 be a brt - graph that admits a balanced vertex cut using the cut - vertex @xmath169 ( in @xmath41 ) , where @xmath169 is a cut vertex obtained as in the proof of lemma [ balancelemma ] .
let @xmath42 and @xmath43 be the two subgraphs obtained by the vertex - cut .
then one of the subgraphs , say @xmath42 , can be chosen as one of the following : \(i ) @xmath42 is a disk - component of @xmath41 .
\(ii ) @xmath42 consists of a union of several maximal disk - components of a single inseparable disk - component of @xmath41 .
notice that figure [ vertexcut ] is an example of case ( ii ) .
if there exists a disk - component @xmath229 of @xmath41 with @xmath271 then let @xmath42 be @xmath229 ( case ( i ) )
. now assume that no disk - component @xmath229 exists in @xmath41 such that @xmath271 .
assume there exists a disk - component @xmath229 of @xmath41 such that @xmath272 .
moreover assume that among all disk - components @xmath229 with @xmath272 , @xmath229 is the smallest one .
claim 1 : @xmath229 must be inseparable .
otherwise , @xmath266 for two disjoint proper disk - components @xmath264 and @xmath265 .
@xmath264 or @xmath265 must have weight less than or equal to @xmath273 since @xmath229 is the smallest disk - component with weight more than @xmath273 .
but then it must be true that @xmath274 or @xmath275 since @xmath276 .
this is a contradiction since we assumed that there are no disk - components with weight between @xmath277 and @xmath273 .
claim 2 : @xmath229 must contain at least one proper disk - component @xmath267 with @xmath278 . if this is not the case , deleting @xmath169 from @xmath41 results in a connected component of weight more than @xmath273 , contradicting the fact that @xmath169 is a cut - vertex for a balanced vertex cut .
remember that the cut - vertex @xmath169 obtained in the the proof of lemma [ balancelemma ] leads to connected components with weights @xmath279 .
it follows that @xmath229 contains maximal proper disk - components .
let @xmath280 be the maximal proper disk - components of @xmath229 , then for each @xmath281 , @xmath282 by our assumptions .
let @xmath283 .
we must have @xmath284 , otherwise the graph @xmath285 has weight @xmath286 and remains connected after @xmath169 is deleted , contradicting the fact that @xmath169 is a cut - vertex for a @xmath121-balanced vertex cut .
thus @xmath42 can be chosen to be the union of some or all of the @xmath287 s ( case ( ii ) ) .
the last case we need to consider is that all disk - components @xmath229 of @xmath41 satisfy the condition @xmath288 .
however this is impossible since @xmath41 is a disk - component of itself . in order to apply the divide - and - conquer technique
, it is necessary for us to use repeated balanced subdivisions to a brt - graph @xmath41 .
[ recursivesubdivision ] a brt - graph @xmath41 ( and the resulting brt - subgraphs ) can be divided recursively using balanced subdivisions .
when the standard weight of a brt - graph obtained in this repeated subdivision process falls below a pre - determined threshold @xmath260 , the subdivision process stops on this brt - graph and it is called a _ terminal brt - graph_. the subdivision process has to terminate at the point when all the resulting brt - graphs are terminal brt - graphs .
the balanced subdivisions used to reach this stage are called a _ balanced recursive subdivision sequence _ of @xmath41 . to keep track of the brt - graphs obtained when
a balanced recursive subdivision sequence is applied to a brt - graph @xmath41 , the following notations are adopted .
when the first subdivision is applied , the two resulting brt - graphs are denoted by @xmath290 and @xmath291 , the gluing instruction of the subdivision process is denoted by @xmath292 , and the two new blue vertices created are denoted by @xmath293 and @xmath294 .
the two brt - graphs obtained from subdividing @xmath290 are denoted by @xmath295 and @xmath296 and the two brt - graphs obtained from subdividing @xmath291 are denoted by @xmath297 and @xmath298 , and so on . in general , the brt - graphs obtained from subdividing @xmath299 ( if @xmath300 ) are denoted by @xmath301 and @xmath302 , the newly created blue vertices are @xmath303 and @xmath304 , and the gluing instruction is @xmath305 . see figure [ subdivisiontree ] for an illustration of this relation .
notice that the lengths of the paths from the root ( @xmath306 ) of the tree to the leaves are not necessarily the same as shown in figure [ subdivisiontree ] , since some brt - graphs may terminate earlier than others due to size differences .
we say that a brt - graph @xmath229 is _ induced _ from the plane graph @xmath41 if @xmath229 is one of the @xmath299s described above .
if @xmath307 is a terminal brt - graph where @xmath308 is largest among all other terminal brt - graphs induced from @xmath306 ( from the same recursive subdivision sequence ) , then @xmath308 is called the _ depth _ of the corresponding recursive subdivision sequence . .
] lemmas [ splittinglemma ] and [ balancelemma ] lead to the following theorem .
[ trecexist ] there exists a balanced recursive subdivision sequence for each brt - graph @xmath41 .
furthermore , @xmath309 since the subdivisions are balanced .
it follows that there exists a constant @xmath310 ( @xmath311 depends only on @xmath260 and the maximal degree @xmath226 of all the red vertices in @xmath41 ) such that the depth of any balanced recursive subdivision sequence of @xmath41 is bounded above by @xmath312 .
[ depthbound ] let @xmath308 be the depth of a balanced recursive subdivision sequence of @xmath41 and @xmath299 be one of the brt - graphs induced from @xmath41 by this sequence .
if we apply theorem [ trecexist ] with @xmath299 playing the role of @xmath41 ( as the starting graph in the subdivision sequence ) , then the depth @xmath124 of the subdivision sequence leading @xmath299 to its terminal brt - graphs is at most @xmath313 and we have @xmath314 . to apply the recursive subdivision to a knot diagram , we start with a minimum knot diagram @xmath268 of the knot @xmath0 so that the number of crossings in @xmath268 is equal to @xmath315 . ignoring the over / under information of @xmath268 at its crossings ,
we treat @xmath268 as a 4-regular plane graph . in general
, @xmath268 is not a brt - graph since @xmath268 may contain faces of arbitrarily large size .
thus the previously established results can not be applied directly to @xmath268 . to remedy this problem ,
artificial edges are added to @xmath268 so that the resulting graph is a brt - graph .
these added edges may simply be removed from the embedding of the modified graph at the end of the process .
the following lemma asserts that @xmath268 can be modified into a brt - graph in such a way that the maximum degree of its vertices is bounded by a constant .
[ deg12 ] let @xmath268 be a minimum projection of @xmath0 .
if @xmath268 is treated as a plane graph so that crossings of @xmath268 are treated as vertices and strands connecting crossings are treated as edges , then by simply adding some new edges to @xmath268 , @xmath268 can be modified into a plane graph @xmath41 such that @xmath41 is triangulated and the maximum degree of the vertices of @xmath41 is bounded above by 12 .
each face @xmath137 of @xmath268 can be triangulated in a way as shown in figure [ facetriangulation ] . in doing so , at most two edges
are added to a vertex of @xmath137 .
since each vertex in @xmath268 belongs to at most 4 faces this results in at most 8 new edges being added to each vertex .
hence the maximum degree of the resulting graph is bounded above by 12 .
let @xmath41 be the triangulated graph obtained from @xmath268 as described in the proof of lemma [ deg12 ] . at this stage , every vertex in @xmath41
is considered to be a red vertex .
since @xmath41 contains no blue vertices and no loop edges , it admits a proper br - partition .
furthermore , it contains only one red component ( namely itself ) and this component is triangulated .
thus by definition , @xmath41 is a brt - graph ( without blue vertices ) . by theorem [ trecexist ]
, there exists a balanced recursive subdivision sequence for @xmath41 .
since the subdivision operations do not increase the degree of a red vertex , the maximum degree of red vertices in such a graph @xmath229 is still bounded above by 12 , see lemma [ deg12 ] .
in this section we introduce two special kinds of embeddings : the standard 3d - embedding and the grid - like embedding .
the purpose of introducing the standard 3d - embeddings of brt - graphs is to use these embeddings as benchmarks to verify that the topology of a graph is preserved when it is reconstructed from its two induced brt - graphs . on the other hand , the purpose of introducing the grid - like embeddings is to simplify the reconstruction process : if two induced brt - graphs are grid - like , then their grid - like structure will allow us to reconnect them in a way to preserve this grid - like structure so this reconnected graph can be used again in the next round of the reconstruction process of @xmath41 .
furthermore , a grid - like embedding is almost on the lattice and in the last step when @xmath316 itself is reconstructed ( from its two immediate induced brt - graphs @xmath290 and @xmath291 ) , it will be easily modified into a lattice embedding . in the following
we assume that all brt - graphs @xmath299 involved are induced from a plane graph @xmath41 and that the maximum degree of red vertices is bounded above by 12 .
below we are introducing some terminology that we will use for the graphs @xmath299 and their blue vertices throughout the next sections .
* rectangles . *
since @xmath229 = @xmath299 is a plane graph drawn in the plane @xmath93 , it can be embedded in the interior of a rectangle @xmath57 in the plane @xmath93 .
that is , there exists a plane isotopy @xmath36\longrightarrow \r^2 $ ] such that @xmath317 and @xmath318 .
we will assume that all graphs @xmath319 are contained in such a rectangle .
keep in mind that our the sides of our rectangles are parallel to either the @xmath12-axis or the @xmath13-axis . * blue squares . * for each blue vertex @xmath46 of @xmath229 create a small square @xmath320 in the plane @xmath93 with side length @xmath321 for some fixed small positive number @xmath322 such that @xmath46 is at the center of the square and @xmath323 .
@xmath320 is called a _ blue square_. see figure [ bluesq_route ] .
( marked by the thickened line segment ) .
the edge with label 1 is the first edge in the linear order assigned at the blue vertex . ] without loss of generality we can assume that the boundary of @xmath320 intersects each edge leading out of @xmath46 exactly once transversely .
notice that we can isotope the graph locally so that the edges within @xmath324 are single line segments as shown on the left of figure [ bluesq_route ] .
let @xmath325 be the square with side length @xmath171 and center @xmath46 and @xmath326 be a side of @xmath325 .
the purpose of @xmath325 and @xmath326 is for us to use a local ( vnp-)isotopy to re - route the edges connected to @xmath46 in such a way that they all enter @xmath325 from @xmath326 .
moreover within @xmath324 the edges use only segments parallel to the @xmath12- and @xmath13-axis .
the right side of figure [ bluesq_route ] then shows an example of how to re - route these edges within @xmath327 to achieve the desired result .
as the figure shows , this can be done for each edge in @xmath324 involved with at most six right angle turns in the @xmath328-plane .
let @xmath329 be the vector ( parallel to either the @xmath12- or the @xmath13-axis ) that points perpendicular from @xmath326 to @xmath46 and we call @xmath329 an _
extension vector_. * orientation assignment of @xmath330 . * without loss of generality we can assume that the boundary of @xmath320 intersects each edge leading out of @xmath46 exactly once transversely . recall that when a new blue vertex is created by an edge - cut subdivision , we oriented the components of @xmath212 counterclockwise and used this orientation to define the cyclic order of the edges intersecting @xmath212 . if a new blue vertex @xmath331 is in the brt - graph @xmath299 obtained by using the part of the original graph outside of @xmath175 , then one may treat @xmath330 as a deformation ( contraction ) of the outer component of @xmath212 . in this case
we give @xmath330 a counterclockwise orientation . on the other hand ,
if @xmath331 is in the brt - graph obtained using the part of the original graph inside of @xmath175 , then @xmath330 should be treated as a deformation of the inner component of @xmath212 , where one would have to flip the inner component of @xmath212 to realize the resulting graph on the plane without edge crossings .
thus in this case we will assign @xmath330 a clockwise orientation .
finally , in the case that @xmath46 is created by a vertex - cut subdivision , @xmath330 is always assigned the counterclockwise orientation . * the orientation vector . * the order of the intersection points on @xmath326 is inherited from the order of intersection points on the boundary of @xmath324 induced by the orientation of @xmath324 . to be more precise
, we can choose any edge on the boundary of @xmath324 and using any path to connect it to @xmath326 .
after that we can choose a second edge to go on either side of the first edge along @xmath326 .
after that the order of all other intersection points on @xmath326 is determined . in the example of figure [ bluesq_route ] we chose @xmath326 to be on the left side of @xmath325
once we fix the edge with label 1 anywhere on @xmath326 then there are only two choices for the other edges to follow : we can obtain edge order @xmath332 ascending along @xmath326 , or an edge order of @xmath333 ascending along @xmath326 .
we can think of both of these as the same orientation along @xmath326 with the difference that one starts with the edge labeled 1 and the other with the edge labeled 2 .
thus the counterclockwise or clockwise cyclic order of the edges around @xmath320 introduces a unique direction on @xmath326 .
we call the vector given by this orientation the _ orientation vector _ @xmath215 . furthermore
, we will choose the first edge to intersect @xmath326 so that the linear order so obtained on @xmath326 matches the linear order at @xmath334 , i.e. the linear order obtained from the counterclockwise orientation on @xmath212 together with the choice of the path @xmath208 in @xmath195 ( edge - cut ) or the point @xmath208 on @xmath212 ( vertex - cut ) as given by the gluing instructions .
notice that we use the same name and symbol for the small circular arrow around blue vertices in a neighborhood @xmath195 that contains the glueing instructions , see definition [ glueinginstructions ] .
we also use the word orientation vector of the orientation on @xmath212 .
the reason is that these orientation vectors in @xmath195 or on @xmath212 directly induce the orientation vector along the segments @xmath326 .
the vectors are equivalent ( they tell us the linear order ) and this justifies the identical names .
it follows that the extension vector @xmath329 can be obtained from the orientation vector @xmath215 by a 90 degree clockwise or counterclockwise rotation . in the example of figure [ bluesq_route ]
we need to rotate the the orientation vector @xmath215 90 degree clockwise to obtain the extension vector @xmath329 .
this tells us that in this case @xmath46 arose in an edge - cut and belongs to @xmath42 ( the graph obtained from the inside of @xmath175 ) .
if we need to rotate the orientation vector @xmath215 90 degree counter clockwise to obtain the extension vector @xmath329 then the vertex @xmath46 arose in a vertex - cut or @xmath46 arose in an edge - cut and belongs to @xmath43 ( the graph obtained from the outside of @xmath175 ) . finally , if in the formation of a blue vertex @xmath46 an edge between a newly temporarily created blue vertex and an older existing blue vertex is contracted , then the above orientation determination still applies : the linear order of the intersection points on @xmath212 does include the edges from the existing blue vertex and these existing orders do not change when the blue vertices are merged . * blue bands and blue triangles .
* let @xmath335 be the cross - section of @xmath325 that contains the blue vertex @xmath46 and is parallel to @xmath326 and let @xmath336 be the line segment obtained by moving @xmath335 up to the plane @xmath337 for some positive @xmath338 .
let @xmath339 be a small positive real number and let @xmath139 be the point directly over @xmath46 in the plane @xmath95 where @xmath340 .
the rectangle formed defined by @xmath335 and @xmath336 is called a _ blue band _ and
is denoted by @xmath341 .
the ( vertical ) triangle formed by @xmath336 and @xmath139 is called a _ blue triangle _ and is denoted by @xmath342 .
we will now redraw the graph locally within the boundary of @xmath343 as shown in figure [ bluetriangle ] . under this redrawing ,
the point @xmath139 becomes the blue vertex @xmath46 and each edge from the boundary of @xmath324 to @xmath334 is replaced a path consisting at most 9 straight line segments : at most 7 in the @xmath328-plane , one vertical ( from @xmath335 to @xmath341 ) and one slant ( from @xmath341 to @xmath139 ) .
this obviously does not change the topology of the graph , it simply creates a 3-d structure of the graph for us to work with .
we place the extension vector @xmath329 at the point @xmath139 for future references .
see figure [ bluetriangle ] for an illustration of this .
note that half of @xmath344 with @xmath345 as a side but opposite to @xmath326 is not occupied by any edge under this construction . with @xmath325 as its base .
the unused area in @xmath325 is lightly shaded . ]
assume that the above process is applied to every blue vertex in the graph , then we arrive at a new graph @xmath346 that is bounded in the rectangular box @xmath347 $ ] .
finally , we require that the projections of the blue squares to the @xmath12-axis and to the @xmath13-axis do not overlap each other .
this can be done since we can pre - determine the positions of the blue vertices in @xmath57 by lemma [ fixedcurvelemma ] and we can choose the side length of these squares arbitrarily small . [ 3ddrawing ] the embedding @xmath346 obtained in the above process from the brt - graph @xmath229 in the rectangular box @xmath348
is called a _
standard 3d - embedding _ .
notice that a standard 3d - embedding @xmath346 of @xmath229 is isotopic to the brt - graph @xmath229 by a vnp - isotopy that is the identity outside the space @xmath349 $ ] for an arbitrarily small positive constant @xmath350 .
let us first describe the part of the isotopy that involves the space @xmath351 .
here at each blue vertex @xmath46 we retract the blue band @xmath341 until the base of the blue triangle reaches the plane @xmath93 .
after that we fold the blue triangle rigidly into the half of @xmath344 that is not occupied by any edge .
this give us an embedding that is entirely contained in the plane @xmath93 .
the rest of the vnp - isotopy involves only moves within the plane ( which is a plane isotopy that is automatically vnp ) . in other words ,
the vnp - isotopy described here is just a trivial extension of a plane isotopy . from lemma [ redrawginr ] we know that we may pick the locations of the deformed blue vertices on the top of the box @xmath100 $ ] with almost total freedom
. therefore , we like to have standard 3d - embeddings of @xmath229 with the property defined in the following definition . [ predetermined3ddrawing ] let @xmath229 be a brt - graph induced from @xmath41 with blue vertices @xmath352 .
let @xmath57 be any given rectangle in @xmath93 whose sides are parallel to either the @xmath12- or @xmath13-axis .
let @xmath353 , @xmath354 , ... , @xmath355 be any @xmath87 distinct points in the interior of @xmath57 and let @xmath141 , @xmath142 , ... , @xmath356 be the corresponding points on the plane @xmath95 directly above the points @xmath353 , @xmath354 , ... , @xmath355 .
if there exists a standard drawing @xmath346 of @xmath229 in @xmath100 $ ] such that @xmath357 is the blue vertex in @xmath346 corresponding to @xmath91 for @xmath358 , @xmath172 , ... ,
@xmath87 , then @xmath346 is called pre - determined standard 3d - embedding of @xmath229 ( with @xmath141 , @xmath142 , ... , @xmath356 being the pre - determined blue vertices ) .
the following lemma asserts that it is indeed possible to create pre - determined standard 3d - embeddings of @xmath229 .
[ pre3dexist ] let @xmath229 be a brt - graph induced from @xmath41 with @xmath87 blue vertices and let @xmath57 be any given rectangle in @xmath93 whose sides are parallel to either the @xmath12- or @xmath13-axis .
let @xmath353 , @xmath354 , ... , @xmath355 be any @xmath87 distinct points in the interior of @xmath57 and let @xmath141 , @xmath142 , ...
, @xmath356 be the corresponding points on the plane @xmath95 directly above the points @xmath353 , @xmath354 , ... , @xmath355 .
then there exists a pre - determined standard 3d - embeddings of @xmath229 with @xmath141 , @xmath142 , ... , @xmath356 being the pre - determined blue vertices .
let @xmath229 be a brt - graph induced from @xmath41 with blue vertices @xmath352 .
let @xmath57 be any given rectangle in @xmath93 whose sides are parallel to the @xmath12- and @xmath13-axis .
let @xmath353 , @xmath354 , ... , @xmath355 be any @xmath87 distinct points in the interior of @xmath57 .
then by lemma [ redrawginr ] there exists a plane isotopy @xmath44 such that @xmath75 is contained in @xmath57 and @xmath359 .
we can then obtain the desired pre - determined standard 3d - embedding of @xmath75 with the @xmath357s being the blue vertices of the new graph by the previously outlined construction . for the purpose of embedding the graph @xmath41 into the cubic lattice ,
the structure offered by a standard 3d - embedding is not enough .
we need to use a structure that is almost like a lattice embedding for the graphs obtained in the subdivision process .
a graph embedding with such a structure will be called a _ grid - like _ embedding .
the detailed description of this embedding is given in this section .
assume that @xmath229 = @xmath299 for some valid @xmath360 from the recursive subdivision process and that the degree of any red vertex in @xmath41 is at most 12 .
[ gridlikedrawing ] we call an embedding @xmath361 of a brt - graph @xmath229 a grid - like embedding if it satisfies the following conditions : \(i ) all red vertices of @xmath361 are lattice points in the plane @xmath93 .
moreover , each red vertex @xmath46 is contained in the interior of a lattice rectangle @xmath362 of dimensions @xmath363 where @xmath364 ( called a _ red square _ ) that does not contain any other vertices of @xmath361 .
the edges of @xmath361 connected to @xmath46 must pass through ( different ) lattice points on the boundary of @xmath362 .
figure [ redsquare ] shows this for a vertex of degree 12 in the smallest possible lattice rectangle ( a @xmath365 square ) .
lattice square . ]
\(ii ) @xmath361 is contained in a rectangular box @xmath347 $ ] for some integer @xmath96 .
all blue vertices of @xmath361 are on the top of the box , i.e. , in @xmath94 . similar to the standard 3d - embedding , each blue vertex is the top vertex of a blue triangle that is on top of a blue band originated from a blue square .
the difference here is that the blue square ( from where the blue band originates ) may be in a horizontal plane @xmath366 for some integer @xmath367 ( it can also be in the plane @xmath93 ) , here @xmath339 is a small positive constant .
all the other requirements on the blue bands , triangles , and squares as given in the definition of a standard 3d - embedding must also be satisfied .
\(iii ) all edges outside the red squares and blue triangles are paths consisting of only line segments parallel to the @xmath12- , the @xmath13- , or the @xmath98-axis .
moreover all horizontal line segments must have integer @xmath98-coordinates .
\(iv ) all red - red edges ( edges connecting two red vertices ) are on the ( cubic ) lattice with the ( possible ) exception of the segments contained in red squares
. the red - blue edges do not have to be on the lattice .
\(v ) @xmath361 is isotopic to a standard 3d - embedding of @xmath229 by a vnp - isotopy @xmath368 that is identity outside the box @xmath369 $ ] .
furthermore the isotopy restricted to any red square must be a plane isotopy , that is throughout the isotopy a red square remains in the plane @xmath93 .
however it does not need to remain a lattice rectangle , the red square simply will play the role of a disk neighborhood of the red vertex .
the requirement that the vnp - isotopy is the identity outside the box @xmath369 $ ] enforces that the edge order of the blue vertices can not be changed since the blue triangles do not move at all .
the requirement on the red squares enforces that the edge order of the red vertices can not be changed either .
the reason for the extension @xmath370 of the space the isotopy can use will become clear later .
there are two approaches to obtain a grid - like embedding of a brt - graph @xmath229 .
the first is a direct construction from the graph @xmath229 and the second is a reconstruction using two grid - like embeddings of the two brt - graphs obtained from @xmath229 by a vertex - cut or an edge - cut subdivision .
for the terminal brt - graphs obtained from @xmath41 in the subdivision process , we will have to use the first approach to obtain their grid - like embeddings .
although we can use this first approach to get a grid - like embedding of @xmath41 itself as well , it will not achieve the desired efficiency in the embedding length .
for that we will then need the second approach to assemble these grid - like embeddings of the terminal brt - graphs into grid - like embeddings and ultimately obtain a grid - like embedding of @xmath41 ( which will then be modified into a lattice embedding of the knot diagram ) .
the following lemma assures that the first approach is always possible .
[ pregridexist ] a grid - like embedding of a brt - graph @xmath229 can be directly constructed from @xmath229 .
let @xmath264 be a standard 3d - embedding of @xmath229 that is guaranteed by lemma [ pre3dexist ] .
we stretch @xmath57 in the @xmath12- and @xmath13-direction by inserting additional lattice lines .
first we add lattice lines at the locations of the red vertices to put these on lattice .
next we add enough lattice lines to create required minimum size red rectangle . by induction on the number of edges in @xmath264
, we can prove that all edges , except the parts contained in the red rectangles @xmath371 s can be straightened by a vnp - isotopy so that they consist of only line segments parallel to a coordinate axis .
we now stretch @xmath57 in the @xmath12- and @xmath13-direction in a recursive manner .
each stretch keeps the line segments already on the square lattice on the lattice , but takes at least one line segment on a red - red edge path that is not on the square lattice to the lattice .
we need to be careful to not disturb the existing blue squares and red rectangles by these stretches .
if a blue square is intersected by a line @xmath39 that contains a segment of a red - red edge then we first move the blue square slightly to ensure that @xmath39 cuts edges in the blue square only transversely .
next the lattice line is inserted at @xmath39 which stretches one side of the blue square to a length of @xmath372 .
the blue square is reconstructed inside this rectangle with side length @xmath321 without intersecting @xmath39 .
see figure [ bluesquareexpand ] for an illustration of this .
if a red rectangle is intersected by a line @xmath39 , adding the lattice line results in a red rectangle which still satisfies the required size restrictions and no additional steps must be taken .
furthermore , the red rectangles remain disjoint and the blue triangles still share no common @xmath12- or @xmath13-coordinates after the stretches .
it is easy to see that this is always possible .
the resulting graph is denoted by @xmath373 .
this process does not change the structure of the blue bands , it only moved some blue squares with their corresponding blue bands and blue vertices by a rigid motion .
thus the graph @xmath373 satisfies all the conditions of a grid - like embedding . using a recursive subdivision process of @xmath41 to construct a grid - like embedding of @xmath41 on the lattice
requires a procedure to combine two grid - like embeddings of brt - graphs ( obtained from either a circular edge - cut or a vertex - cut subdivision ) into a new grid - like embedding .
more precisely , let @xmath374 , @xmath375 and @xmath376 be three brt - graphs obtained in the subdivision process of @xmath41 by a vertex - cut subdivision or an edge - cut subdivision as defined in the paragraph after definition [ recursivesubdivision ] .
we assume that for @xmath377 , @xmath118 has a grid - like embedding @xmath378 embedded in a box @xmath379 , which has the rectangle @xmath380 as its base in @xmath93 and the height @xmath381 . in addition , we assume the following convention : if @xmath374 was divided using a circular edge - cut , then @xmath42 refers to the brt - graph which is derived from the graph inside @xmath175 ; if @xmath299 was divided using a vertex - cut , then @xmath42 refers to the subgraph which contains the maximal disk - components mentioned in [ diskcomponent ] ( ii ) , or if the balanced vertex cut is chosen following [ diskcomponent ] ( i ) it refers arbitrarily to any one of the two subgraphs .
this section describes a procedure to obtain a grid - like embedding @xmath382 of @xmath299 using only information from the given embeddings @xmath383 and @xmath384 and from the gluing instruction @xmath305 . during the description ,
we refer to the blue vertices @xmath385 and @xmath304 created by the subdivision as @xmath209 and @xmath210 , respectively .
we split the construction into 7 steps .
steps ( 1 ) through ( 3 ) serve to prepare the graphs @xmath383 and @xmath384 for the connection process , while steps ( 4 ) through ( 7 ) make the actual connection of the edges .
\(1 ) align the boxes @xmath379 containing the @xmath378 properly in the @xmath12- and @xmath13-directions next to each other , with their base rectangles in the plane @xmath93 . without loss of generality we assume that the space between the two boxes is exactly one unit in the @xmath12-direction and that one side of the boxes coincides with the @xmath12-axis . furthermore all the red squares and all the red - red edges outside the red rectangles are still on the cubic lattice . in the gap between the boxes we put a _ connecting rectangle _ at height @xmath386 .
the connecting rectangle has dimensions @xmath387 where @xmath388 and @xmath111 is the @xmath13-dimension of the box @xmath379 in their new location , see figure [ bandex1 ] .
\(2 ) create a ( rectangular ) box @xmath389 which includes both boxes @xmath390 and @xmath391 and the connecting rectangle and is of height @xmath392 .
\(3 ) extend the blue vertices in @xmath383 and @xmath384 other than the @xmath207 to the top of the box @xmath389 by extending their blue bands vertically by one unit and also lifting the blue triangles vertically by one unit .
\(4 ) delete the blue triangles at the blue vertices @xmath209 and @xmath210 and extend the corresponding blue bands in the @xmath98-direction to the plane @xmath393 . then extend the blue bands horizontally in the plane @xmath394 to the connecting rectangle .
we refer to this horizontal extension of the blue band as the _ extension band_. the two extension bands and the connecting rectangle is referred to as the _ connecting strip_. the extension band consists of rectangles in the plane @xmath394 .
it starts in the direction of the extension vector in the case of circular edge - cut and in the opposite direction of the extension vector in the case of vertex - cut . with at most two right angle
turns within @xmath395 ( where @xmath396 is the blue square of the corresponding blue vertex ) , it can be made moving toward the connecting rectangle .
with two right angle turns in the connecting rectangle and a suitable bandwidth change , it is then connected to the extension band coming from the other blue vertex , see figure [ bandex1 ] ( which is a case of the circular edge - cut subdivision ) . since the turns only happen in the connecting rectangle and in the blue squares ( at the @xmath394 level )
, the projection of the extension band into the @xmath328-plane does not intersect any other blue square ( hence itself will not intersect any other blue band in the rectangular boxes @xmath379 ) .
this is true because of the properties of the blue squares and blue bands .
, @xmath390 , @xmath391 and the connecting rectangle ( shaded rectangle in the middle ) in the case that the blue vertices are created by a circular edge - cut subdivision .
several typical paths from the top of a blue band to the other blue band are shown as well , with the labels of the edges to show how the corresponding edges should line up on the opposite sides of the connecting rectangle .
only the center parts of the blue squares are shown and they are greatly enlarged to reveal the details . ]
\(5 ) this step applies when @xmath42 and @xmath43 are obtained after a circular edge - cut subdivision , so the extension band starts in the direction of the extension vector .
recall that we had assumed that in this case @xmath42 is the brt - graph inside @xmath175 .
so @xmath209 is clockwise and @xmath210 is counterclockwise , see the paragraph marked as orientation determination " before definition [ 3ddrawing ] .
cut the neighborhood @xmath305 open along @xmath208 and stretch it into a rectangle called @xmath397 as we did in figure [ unicycleneighborhood ] .
modify and re - scale @xmath397 so that ( a ) its side lengths are smaller than one third of the side lengths of either of the two blue squares and ( b ) each path connecting two opposite boundary points that correspond to the intersection points of the same edge with @xmath398 is just a single line segment ( parallel to either the @xmath12- or the @xmath13-axis ) .
this rectangle is then placed into the connecting rectangle and is denoted by @xmath399 . since the linear order of the edges along the extension band is the same as the linear order of @xmath213 ( and @xmath214 ) by the definition of grid - like embedding and the fact that @xmath42 and @xmath43 are grid - like embeddings .
thus edges with the same labels ( edges that are to be connected ) from each side ( namely either from the @xmath42 or @xmath43 side ) can be aligned perfectly with their counter parts of the edges of @xmath399 as shown in figure [ connectingstrip ] . and
@xmath43 are reconnected : the connecting rectangle is made much wider to show the details . in the middle of the connecting rectangle
is @xmath399 , which is a deformation of the neighborhood of figure [ unicycleneighborhood ] .
the curve that went under two edges represents a loop edge deleted when @xmath42 and @xmath43 were created . ]
this connects all the edges in the blue bands and their extension bands arising from @xmath383 and @xmath384 .
in particular , an edge from @xmath383 is connected to an edge @xmath384 only if both edges have the same label and no additional crossings are introduced . by our construction , these edges consist of only straight line segments . in particular any edge passing from the extension band of @xmath383 to the extension band of @xmath384
contains only two right angle turns on the connecting rectangle .
@xmath397 may contain one or more new blue vertices .
for each such blue vertex @xmath169 , extend each edge connected to @xmath169 from where it enters @xmath399 to the centerline of @xmath399 , then extend it up by one unit . create a small blue square in @xmath400 over @xmath399 for this blue vertex .
then a blue band , and a blue triangle with the new blue vertex on top of the blue triangle in the plane @xmath401 , following the same rules as before ( for grid - like embeddings ) .
see figure [ newbluesquares ] .
are grid - like and two new blue squares with blue bands and triangles have been created .
the figure is not to scale and the new blue squares appear to be much larger than they actually are .
the linear orders at the new blue vertices are recovered from their corresponding gluing instructions .
only the top view is shown so the new blue vertices and blue triangles are not visible in the figure . ]
since the size of the new blue squares can be arbitrarily small and each blue vertex in @xmath399 has certain free space to move ( without crossing the straight paths that have been placed ) , the projections of the blue squares to the @xmath12- and the @xmath13-axes can be adjusted so as not to overlap with each other or with any other existing blue squares . for each blue square
, the edges are combined into a vertical blue band as before and topped with a blue triangle at @xmath400 . at this point
all edges that are connected to @xmath209 or @xmath210 are accounted for .
note that we have not addressed the labels on @xmath305 that correspond to loop edges that were deleted from @xmath192 and @xmath193 when @xmath42 and @xmath43 were created ( although one such edge has been illustrated in figure [ newbluesquares ] ) .
we will address this in step ( 6 ) .
( 5@xmath402 ) this step applies when @xmath42 and @xmath43 are obtained after a vertex - cut subdivision . for the vertex - cut ,
@xmath403 and @xmath404 have the same orientations in the plane .
since the extension band in this case starts in the direction opposite to the orientation vector , the linear orders of the edges along the band sides will again align correctly .
see figures [ bandex3 ] and [ principalvertexcut ] for an illustration of this . .
in this case the original linear order of the edges at the blue vertex @xmath405 is @xmath406 .
the solid dots can be thought of as the point @xmath208 used to define the linear order . a simplified 3d view of this is shown in figure [ principalvertexcut ] .
] , @xmath407 , and @xmath408 ( in gray ) together with the three extension vectors @xmath409 , @xmath410 , and @xmath411 that arises in a vertex cut .
the connection is only shown schematically without horizontal turns and the width of the bands is not to scale .
only two edges ( black ) are shown to illustrate the layout . ]
the connecting rectangle contains @xmath305 with a single blue vertex @xmath405 that was split into the two blue vertices @xmath209 and @xmath210 . as in the case of an edge - cut ,
a blue square is created in @xmath400 first , then a blue band , and a blue triangle with the new blue vertex on top of the blue triangle in @xmath401 , satisfying all requirements of a grid - like embedding .
this connects all the edges in the extension bands originated in the blue bands @xmath412 and @xmath407 .
notice that the linear order of the edges at @xmath405 is restored when the edges from two sides meet the middle bar of the connecting rectangle .
\(6 ) this step deals with the loop edges deleted after a circular edge - cut subdivision ( such deletion can only happen in the case of a circular edge - cut subdivision ) . in the case that @xmath42 and @xmath43 are obtained after some loop edges are deleted , then @xmath397 contain labels not used to @xmath413 and @xmath414 .
the creation of temporary loop edges can happen in two ways .
the first case is when a red - red edge @xmath135 ( not on @xmath116 ) was cut twice by @xmath175 .
both red vertices incident to @xmath135 are contained in one of the graphs , say in @xmath42 ( the case if the vertices are contained in @xmath43 is identical ) and therefore in @xmath383 .
the middle arc of @xmath135 is contracted into a loop in @xmath187 and is eventually deleted .
this leads to 4 identical labels on @xmath398 , two each on each component of @xmath398 .
two labels in one boundary component of @xmath398 are accounted for by their corresponding edges on the extension band originated from @xmath415 .
the other two labels on the other boundary component of @xmath398 have no corresponding labels on @xmath210 .
the two edges coming up from @xmath383 just end on the opposite side of @xmath399 ( and there will be no other edges with the same label later on to connect them ) .
however this is no problem since the gluing instruction tells us that these two edge ends must be connected to each other at this stage .
usually , this can not be done in the plane @xmath393 without creating crossings .
the connection is made using five edge segments , three of which are in the plane @xmath416 .
both edges are extended to the end of the connecting rectangle , see figure [ newbluesquares ] for an example of such an edge .
a vertical segment of unit length is added at the end of each of the edges , connecting the plane at level @xmath393 with the plane at level @xmath417 .
a short segment parallel to the @xmath13-axis is added to the ends of both vertical segments and then one segment parallel to the @xmath12-axis connects the two end points .
this construction builds a small ` hook ' which hangs below the connecting rectangle .
several such loops may have been removed during a subdivision step . for each of them
a hook creates the correct connection between the edges without adding unwanted crossings .
loops may be nested , and the hooks can be nested too .
a higher nesting level of the loops leads to slightly longer pairs of parallel segments which are parallel to the @xmath13-axis and a longer horizontal segment parallel to the @xmath12-axis . in the second case , a red - blue edge ( not on @xmath116 ) connected to the same blue vertex @xmath46 not on @xmath116 are cut once by @xmath175 .
if there is just one such edge then this edge is contracted and does not lead to a loop .
however if there is more than one , only one is contracted and the others result in loops .
let @xmath135 be one such edge .
then the red vertex connected to @xmath135 is contained in one of the graphs , say in @xmath42 ( the case if the vertex is contained in @xmath43 is identical ) and therefore in @xmath383 .
the part of @xmath135 connected to the blue vertex is contracted into a loop in @xmath418 and was deleted .
this leads to 3 identical labels on @xmath398 , one on the boundary component belonging to @xmath42 and the other two on the boundary component belonging to @xmath43 .
the single label in one boundary component of @xmath398 is accounted for by by a corresponding label of edges on the extension band originating from @xmath415 .
the two labels on the other boundary component of @xmath398 are not accounted for by corresponding labels on edges connected to @xmath210 .
we now construct a ` small ' hook exactly as in the first case .
notice that the loops in @xmath419 and @xmath418 are positioned at different ends of the neighborhood rectangle , that is , the end that is closer to @xmath420 and @xmath421 , respectively . the loops on each side nest perfectly but the loops on both sides combined may not exhibit a nesting behavior .
after this step the connecting rectangle contains edges accounted for all labels on @xmath305 and in the same arrangement as is specified in @xmath305 .
\(7 ) this last step only applies to the case when @xmath42 and @xmath43 are obtained after a circular edge - cut subdivision .
the reason is in this case , we may have created red - red edges that are no longer on the lattice hence the reconnected graph described in the earlier steps is not grid - like yet .
we will remedy this problem by adding new gridlines in the @xmath12- and @xmath13-direction to put newly formed red - red edges on lattice . for each line segment ( on the re - connected red - red edge )
parallel to the @xmath13-axis , a new @xmath12-gridline is added ( which corresponds to a stretching isotopy ) . for each line segment parallel to the @xmath12-axis , a new @xmath13-gridline is added .
see figure [ gridlineaddition ] . in this process
we destroy the blue squares that formerly belong to the now vanished blue vertices @xmath209 and @xmath210 .
these blue squares are no longer needed .
all other blue squares must be preserved .
if one of these new gridlines hits the projection of a blue square @xmath320 then we need to adjust the blue square as follows , see figure [ bluesquareexpand ] .
first we slightly move the blue square to make sure that the new gridline hits edges in the blue square only transversely .
then we expand the blue square as is required by the insertion of the new grid line .
the expanded @xmath320 now becomes a rectangle with one of its sides having a length of more than one unit . in this rectangle
we put a copy of the original center square @xmath325 with the original small width @xmath171 by translation .
( if we inserted an @xmath12- or @xmath13- gridline then we translate @xmath325 in @xmath12- or @xmath13-direction , respectively . )
this can be connected up with exactly an many turns for the edges as before . around this newly positioned square @xmath325
we reposition an new blue square @xmath320 with the original size . by default
this new blue square has a projection that is disjoint in @xmath12- and @xmath13-coordinates from the projections of all the other blue squares .
if one of these new gridlines hits a red rectangle then we simply stretch the red rectangle into a larger rectangle .
this does not introduce any new turns and preserves all the required properties of a grid - like embedding .
-direction without creating any additional turns .
the original center square @xmath325 has been translated in @xmath13-direction and a new blue square of the same size as the original has been created with @xmath325 at its center .
the figure is not to scale and the new blue squares appear to be larger than they are . ]
this completes the description of how to combine the two grid - like embeddings of @xmath42 and @xmath43 into a new grid - like embedding of @xmath422 .
while a grid - like embedding of a brt - graph obtained as given in lemma [ pregridexist ] preserves its topology by its construction , it is far from obvious that the embedding obtained by reconnecting two grid - like embeddings as given in section [ sec72 ] preserves the topology of the original brt - graph from which the two grid - like embeddings are induced .
we will prove that this is indeed the case in this section .
[ extconstr ] let @xmath375 and @xmath376 be brt - graphs obtained from the brt - graph @xmath374 by a subdivision .
then the grid - like embedding @xmath382 of @xmath422 as described in section [ sec72 ] , confined in a rectangular box of the form @xmath423 $ ] , is isotopic to @xmath299 by a vnp - isotopy that is the identity outside a small neighborhood of the box @xmath424 $ ] .
therefore , @xmath382 is indeed a grid - like embedding of @xmath422 since it satisfies all other requirements of a grid - like embedding of @xmath422 . before stating the proof ,
let us recall that @xmath42 is either the interior graph in the case of a circular edge - cut subdivision , or the graph containing one disk - component or a union of some maximal disk - components in the case of a vertex - cut subdivision .
all the requirements for a grid - like embedding of @xmath299 as specified in definition [ gridlikedrawing ] are already satisfied by the construction process outlined in section [ sec72 ] , as one can check .
thus it suffices to show that the grid - like embedding obtained is isotopic to a standard 3d - embedding of @xmath422 . in the last step in the construction process described in section [ sec72 ] , the box @xmath389 ( and the boxes @xmath390 and @xmath391 )
are stretched to make room for new gridlines so that the newly created red - red edges can be put on the lattice .
let us call the stretched boxes @xmath425 , @xmath426 and @xmath427 .
case 1 : the case of a circular edge - cut subdivision .
the vnp - isotopy will be constructed by a sequence of isotopies using the following steps : \(1 ) remove some gridlines to get back to an almost grid - like embedding ; \(2 ) shrink @xmath426 and @xmath427 ( with the graphs contained in them ) back to @xmath390 , @xmath391 and deform the graphs contained in them to 3d standard embeddings ; \(3 ) deform the hooks ( representing the loop edges that were deleted during the subdivision ) from the connecting rectangle into the plane @xmath93 ; \(4 ) shrink @xmath42 and drag the shrank @xmath42 along the extension bands and the connecting rectangle and drop it into @xmath43 on @xmath93 .
\(5 ) straighten out the blue squares and make the blue - bands vertical as required by a standard 3d - embedding of @xmath422 using lemma [ straighten ] .
we now address each of these points in detail .
\(1 ) clearly such stretching isotopies are reversible .
however the definition of a grid - like embedding does not allow us to use an isotopy that changes the entire box @xmath389 .
thus we can only shrink the box @xmath425 ( together with @xmath426 and @xmath427 ) back to its original size under @xmath428 .
the first isotopy is defined by this shrinking isotopy @xmath429 for @xmath430 , the identity for @xmath431 and @xmath432 where @xmath339 is the number chosen so that all blue triangles have bases on @xmath433 .
let @xmath232 be a blue band at a blue vertex @xmath46 .
@xmath232 intersects the base of its corresponding blue triangle at @xmath434 and intersects the plane @xmath435 at @xmath436 .
let @xmath437 be the image of @xmath436 under @xmath429 .
then the isotopy for @xmath438 is chosen so that the part of @xmath232 between @xmath434 and @xmath436 ( which is a vertical band ) is mapped to the band joining @xmath434 and @xmath437 ( which may no longer be a vertical band ) .
this can be done since the shrinkage @xmath429 on @xmath439 ( @xmath440 is the base of @xmath389 ) will not cause these bands to intersect each other by the conditions on the positions of the blue squares .
this isotopy restores the original boxes @xmath390 and @xmath391 together with all the blue squares .
the blue triangles in @xmath425 remain the same , the vertical blue bands from @xmath441 to @xmath435 are no longer vertical ( however they remain disjoint from each other and are strictly increasing in the @xmath98-coordinates ) . the part of a blue band under @xmath435 remain vertical after this isotopy is applied .
\(2 ) after isotopy ( 1 ) the graphs @xmath42 and @xmath43 fit back into the original @xmath390 and @xmath391 .
the resulting embeddings almost " restore @xmath383 and @xmath384 .
the almost " stems from the exception that the blue triangles in each box that would be part of a grid - like embedding are distorted , however they are identical to @xmath383 and @xmath384 below the @xmath98-level where the bases of their blue triangles are . by a slight abuses of notation we call these almost " grid - like embeddings still @xmath413 and @xmath414 . by the definition of grid - like embedding
, @xmath413 ( @xmath414 ) is isotopic to a standard 3d - embedding @xmath442 ( @xmath443 ) by a vnp - isotopy that is identity outside the box @xmath444 $ ] ( @xmath445 $ ] ) , see definition [ gridlikedrawing ] ( v )
. we will now apply these two isotopies to @xmath413 and @xmath414 .
after this , an edge path from the base rectangle @xmath440 to a blue vertex already existed in @xmath383 and @xmath384 before the reconnection consists of four straight line segments : a single vertical line segment from @xmath440 to @xmath435 , then a line segment ( that is in a deformed blue band ) from @xmath435 to @xmath441 , then a vertical line segment from @xmath441 to the base of a blue triangle in @xmath446 , followed a line segment in a blue triangle leading to the blue vertex .
notice that the last two line segments are not changed by the isotopy applied so far .
note also that these paths do not intersect the connecting rectangle ( which is also not affected by the last two isotopies since it is in @xmath447 ) . see figure [ deformedband ] .
( or @xmath448 ) to a blue vertex after the isotopy in ( 2 ) is applied to the grid - like embedding @xmath449 constructed from @xmath413 and @xmath414 . ]
\(3 ) during an edge - cut it is possible that in the contracting process we created temporary loop edges that were deleted in order to form @xmath42 and @xmath43 .
these loop edges are realized by some small hooks " that are attached at the two ends of the neighborhood rectangle on the connecting rectangle ( @xmath399 ) below @xmath435 in the reconnecting process ( step ( 6 ) of section [ sec72 ] ) . at this stage
we must realize these loops in the plane @xmath93 as it is required in a standard 3d - embedding of @xmath422 .
a top view of two nested such hooks and their relative positions with the other edges involved in the same blue square ( on @xmath93 ) are illustrated in figure [ hookmove ] .
used in a blue square , the large outside dashed curve is a conceptual depiction of the deformed @xmath175 , the two small dashed curves represent the actual loops corresponding to the two hooks . ] we accomplish this one hook at a time starting with an innermost hook .
we slide a hook along the blue bands down into the plane @xmath93 , see figure [ loopisotopy ] .
once they are in the plane @xmath93 we fold them by a 90 degree turn into the unused space in the blue square so they look just as shown in figure [ hookmove ] .
it is clear from the figure that the hook can then be deformed to the dashed curves from under the plane @xmath93 by a vnp - isotopy that is identity below @xmath450 .
of course it needs to remain in the box @xmath379 to which it belongs .
see figure [ loopisotopy ] for an illustration of this process . .
the edge on the right is dashed because it might have to be stretched out to fit into the plane @xmath93 and can be quite long . ]
\(4 ) let us recall that at this point all red vertices are in the plane @xmath93 and all blue vertices are in the plane @xmath451 .
also , all the red - red edges are also in the plane @xmath93 except those going through the connecting strip ( created in the re - connecting process ) .
we can not simply project these into the plane @xmath93 since that will likely to create crossings so we will not be able to recover our original plane graph this way . since we know that the edges connected to @xmath214 in @xmath391 are in the outer face of @xmath42
, we will try to shrink the graph @xmath42 first and then move the whole graph along the connecting strip into the interior face @xmath137 of @xmath43 where it becomes clear that the original graph structure is recovered .
the blue bands remain connected to the top of the box @xmath389 . in this way
we avoid the creation of unwanted intersections . in the following
we describe these steps in more detail .
assume that @xmath143 , @xmath144 , ... , @xmath452 are the centers of the blue squares associated with the blue vertices in @xmath413 other than @xmath213 .
let @xmath453 be the half of the square @xmath454 in the center of the blue square @xmath455 between @xmath456 and @xmath457 as shown in figure [ bandex2 ] .
choose @xmath87 points @xmath61 , @xmath62 , ... ,
@xmath458 in the small rectangle @xmath453 such that the @xmath70 s do not share the same @xmath12-coordinates nor @xmath13-coordinates .
this small rectangle can be viewed from the top as the rectangle with dotted line boundary and the letter @xmath459 as marked in figure [ bandex2 ] . by lemma [ redrawginr ]
, there exists a plane isotopy @xmath460\rightarrow r_1 $ ] that is identity outside a small neighborhood of @xmath81 and the identity on @xmath461 ( the base of the blue band ) that takes @xmath462 to @xmath70 and moves the all points of the embedding @xmath463 in the plane @xmath93 into @xmath453 .
here we assume that @xmath464 is the identity on @xmath81 and @xmath465 has moved all the @xmath462 s to the @xmath70 s .
this plane isotopy is extended to a vnp - isotopy in the following way : ( a ) it is the identity outside of the box @xmath466 $ ] ; ( b ) its action on @xmath467 for each @xmath468 is the same as that of @xmath469 for @xmath470 $ ] on @xmath471 .
note that this extension keeps all blue bands disjoint from each other and each edge on a blue band is a path that is non - decreasing in the @xmath98 direction .
furthermore , the isotopy can be so chosen that the bases of the blue bands are mapped to bands perpendicular to the direction of the extension vector .
see figure [ shrink1 ] .
looks after it is shrank to fit in @xmath453 .
the dots are the blue vertices and the bars on them indicate the bases of the blue squares after the isotopy . ] we now define an isotopy that retracts the connecting strip while dragging the rectangle @xmath453 with the edges of @xmath42 along the connecting strip . at first , the rectangle @xmath453 containing @xmath42 can be lifted vertically to the level @xmath394 . in doing
so we retract the connecting strip at the same time .
all the blue bands that are connecting @xmath453 to the blue vertices at the top of the box @xmath389 are deformed as well to keep the bands nondecreasing in the @xmath98-coordinates .
an illustration of this process is shown in figures [ bandex2 ] and [ bandex4 ] .
this isotopy clearly preserves the neighborhood structures of the vertices . and
moving it to the unused space in the blue square at @xmath214 . ]
corresponds to a blue vertex recovered in the reconnection process . ]
the retraction continues along the horizontal parts of the extension bands and the connecting rectangle .
whenever the connecting strip turns horizontally we turn the whole rectangle @xmath453 accordingly . again
all the blue bands that are connecting @xmath453 to the blue vertices at the top of the box @xmath389 are dragged along and kept as increasing in their @xmath98-coordinates as shown in figure [ bandex4 ] . at the end of the extension band that leads into @xmath391
it is time to drop the small rectangle containing @xmath42 vertically down . by our assumptions on the extension vector at @xmath210
, there is free space in the small rectangle @xmath472 in the blue square @xmath473 to put the box @xmath453 into the plane @xmath93 .
once the small rectangle @xmath453 is in the plane @xmath93 the connecting strip has been completely eliminated .
we now have the graph @xmath374 embedded into the plane @xmath93 with the exception of the blue vertices of @xmath422 . for each blue vertex
there is a small blue square ( the blue squares have different sizes ) and a blue band that connects the blue square to a blue vertex on the top of @xmath389 along a path that is non - decreasing in the @xmath98 direction . since we can deform the blue vertices ( with the edges connected to them ) into the plane @xmath93 along the deformed bands one by one without any interference and without crossing the boundary of @xmath472 , it is clear that the resulting graph bounded within @xmath472 is topologically equivalent to @xmath42 and the resulting graph outside @xmath472 is topologically equivalent to @xmath43 .
these two graphs are connected along the boundary of @xmath472 following the gluing instruction .
thus the graph obtained after the reconnection is indeed isotopic to @xmath422 .
\(5 ) now let @xmath474 , @xmath475 , ... , @xmath476 be all the intersection points of the edges leading out from the blue vertices of @xmath382 with the bases of the blue triangles in @xmath389 ( they are all in the plane @xmath446 ) and let @xmath477 , @xmath478 , ... @xmath479 be the first intersection points of the corresponding edges with the plane @xmath93 . by our construction , @xmath480 and @xmath481 are connected by a path that is non - decreasing in the @xmath98 direction and these paths do not intersect each other . by lemma
[ straighten ] and remark [ re22 ] after it , there exists a vnp - isotopy @xmath44 such that @xmath44 is the identity in @xmath482 and outside a small neighborhood of @xmath389 , and @xmath44 deforms each path connecting @xmath480 to @xmath481 to a straight line segment and @xmath44 is a plane isotopy when restricted to @xmath93 .
the edges in @xmath93 can be further deformed to create spaces for the required blue squares in @xmath93 .
the result is the desired standard 3d - embedding of @xmath422 .
this finishes the proof of the first case .
case 2 : the case of a vertex - cut subdivision .
the situation in this case is slightly simpler because no temporary loop edges are created and the connecting rectangle contains exactly one blue square @xmath483 .
thus one of the steps in the prior isotopy is no longer needed .
the isotopy is again constructed by a sequence of isotopies using the following steps : \(1 ) remove some gridlines to get back to an almost grid - like embedding ; \(2 ) shrink @xmath426 and @xmath427 ( with the graphs contained in them ) back to @xmath390 , @xmath391 and deform the graphs contained in them to 3d standard embeddings ; \(3 ) shrink @xmath42 and drag the shrank @xmath42 along the extension bands and the connecting rectangle and drop it into @xmath43 on @xmath93 .
\(4 ) straighten out the blue squares and make the blue - bands vertical as required by a standard 3d - embedding of @xmath422 using lemma [ straighten ] .
note that it is essential that @xmath42 is used in step ( 4 ) .
for example , suppose we are in case ( ii ) of lemma [ diskcomponent ] , that is @xmath42 consists of a union of several maximal disk - components of a single inseparable disk - component .
then it is possible to drag @xmath42 into @xmath43 in step ( 3 ) above , but it might not be possible to drag @xmath43 into @xmath42 without getting hung up on some blue band that is connected to the top of the box @xmath389 .
some more argument is needed for the isotopy in step ( 4 ) . in the case for the edge - cut
, the shrunk graph @xmath42 is positioned in the unused rectangle in @xmath472 so placing @xmath42 into that space apparently will not cause of edge intersections . however , in this case the shrunk graph @xmath42 is positioned into the _ used _ half of @xmath472 in @xmath43 . since @xmath42 contains some ( or all ) maximal disk components of some inseparable disk component @xmath484 of @xmath422 , the edges from @xmath42 connected to the blue band @xmath415 fit precisely between the edges of the @xmath485 in @xmath43 already connecting to @xmath486 .
if one combines figures [ principalvertexcut ] and [ bandex4 ] ( but without the red - red edges shown ) , then this becomes clearer .
the details of the isotopies are otherwise similar to the edge - cut argument and are thus left to the reader .
this finishes the proof and also concludes this section .
one way to estimate the embedding length is to estimate the volume of the box @xmath389 in this construction . by our design of the construction ,
the height of @xmath389 is only two units larger than the maximum height of the two boxes @xmath390 and @xmath391 .
we can thus estimate the volume of @xmath389 by estimating the area of the base rectangle of @xmath389 .
one way to do this is to count the number of horizontal gridlines in the @xmath12- and @xmath13-direction we must create to accommodate a newly created red - red edge .
a gridline is consumed either by a segment of the red - red edge in the @xmath328-plane using that gridline or by the red - red edge moving through a vertical segment parallel to the @xmath98-axis .
more precisely we are interested in the following .
let @xmath135 be a newly created red - red edge and consider a vertical projection @xmath487 of @xmath135 into the @xmath328-plane .
we count the 90 degree turns of @xmath487 and add to it the number of segments parallel to the @xmath98-axis to obtain an estimate on the number of gridlines that need to be available in the base rectangle @xmath440 .
[ turningnumber ] let @xmath488 be a grid - like embedding of a brt - graph @xmath41 contained in a rectangular box @xmath348 .
let @xmath489 be the set of red - blue edges in @xmath41 .
for any edge @xmath490 ( @xmath135 can be a red - red or a blue - red edge ) denote with @xmath491 the number of different horizontal gridlines ( in the @xmath12- and @xmath13-direction ) needed and call this the _ turning number _ of the edge @xmath135 .
we define the _ turning number _
@xmath492 of @xmath488 as follows : @xmath493 note that in the above definition we ignore the turns which are not a 90 degree angle that may occur when an edge enters a red rectangle or a blue triangle .
counting the 90 degree turns in @xmath487 over counts the gridlines needed for the horizontal segments in the red - red edge , since several segments of @xmath487 may end up on the same gridline .
note also that @xmath487 may have 180 degree turns that arise when two consecutive turns in the @xmath494 or @xmath495 plane occur , as is the case for a hook . those are not counted , since the vertical segments are counted separately .
we are now able to show the following theorem .
[ sumextconstr ] let @xmath42 and @xmath43 be the brt - graphs obtained from the brt - graph @xmath422 by a single subdivision ( either a vertex - cut or an edge - cut ) .
let @xmath382 be the grid - like embedding of @xmath422 in a rectangular box @xmath496 $ ] , obtained by reconnecting the grid - like embeddings @xmath413 , @xmath414 of @xmath42 , @xmath43 as described in section [ sec72 ] .
if @xmath497 and @xmath498 , then the grid - like embedding @xmath382 of @xmath422 has the following properties : \(i ) for a newly constructed red - red edge @xmath135 of @xmath422 @xmath499 .
\(ii ) @xmath500 .
we first prove ( i ) .
note that a newly constructed red - red edge can only appear if @xmath42 and @xmath43 are created by an edge - cut in @xmath422 .
the path a newly created red - red edge @xmath135 travels is divided into three parts , where each part is a union of one or more line segments each of which is parallel to one of the coordinate axis .
there are two cases to consider : the curve @xmath175 ( the push - off of @xmath116 as defined in section [ sec3 ] ) intersects @xmath135 once or the curve @xmath175 intersects @xmath135 twice . in the first case when traveling along the edge
@xmath135 we encounter the following parts : the first part starts at a red vertex , say in @xmath413 , moves to the blue square @xmath501 , and continues through the blue square to the base of the blue band @xmath412 and then up along the blue band . at this point we know that this first part of @xmath135 has at most turning number @xmath502 .
the middle part of @xmath135 begins with a 90 degree turn in the @xmath494- or @xmath495-plane , from the vertical direction on the blue band @xmath412 to the extension band , moves through the connecting strip and ends with a 90 degree turn ( again in the @xmath494- or @xmath495-plane ) from the horizontal extension band to the vertical direction on the blue band @xmath407 .
it takes at most 2 turns to reach the connecting rectangle from each box @xmath379 and two turns to turn onto and off the connecting rectangle , giving us a turning number of at most 6 for the middle part .
the two 90 degree turns off and onto the vertical blue band @xmath503 are not counted , since the need for this gridline is already accounted for when the blue band is created .
thus @xmath504 in @xmath449 .
if we assume that an edge @xmath505 is intersected by @xmath175 twice then both vertices @xmath143 and @xmath144 are in the same graph @xmath118 .
the middle part of @xmath135 that does not contain any of the two vertices , is contracted into a loop edge and then deleted . in the reconstruction process a small hook is constructed whose projection looks like the letter @xmath229 .
the horizontal bar of the @xmath229 is the edge segment of the hook in the @xmath328-plane below the connecting rectangle which is not connected to a vertical segment and one gridline needs to be added for it .
one pair of ends of the two parallel lines in the @xmath229 is connected to @xmath143 and @xmath144 , the other pair of ends is connected to the vertical segments of the constructed hook . since both ends are on the same line ( in the @xmath12- or @xmath13-direction ) only one gridline needs to be added even though there are two vertical segments .
we account for the fact that both @xmath143 and @xmath144 are in the same graph by using a maximum . in this case
the turning number satisfies @xmath506 turns in @xmath449 . next we prove ( ii ) by proving that @xmath491 for a red - blue edge @xmath135 is @xmath507 .
we consider the case of an edge - cut first .
let @xmath135 be a red - blue edge and assume that the red vertex @xmath138 of @xmath135 is contained in @xmath383 ( the case of @xmath384 is accounted for by using a maximum as before ) and that @xmath12 is the blue vertex of @xmath135 .
we note that the curve @xmath175 can intersect @xmath135 at most once .
the result of the edge - cut is a red - blue edge in @xmath383 from @xmath138 to the blue vertex @xmath209 and potentially a deleted loop edge in @xmath43 that results in a hook in @xmath382 .
no loop ( and thus no hook ) is created if the blue vertex @xmath12 is on @xmath175 , or if @xmath12 is in @xmath43 and is not connected to any edge other than @xmath135 which is cut by @xmath175 .
first let us consider the case when there is no hook in @xmath382 . in this case ,
@xmath135 consists of two parts in @xmath383 .
as before , the first part starts at the red vertex @xmath138 in @xmath413 , moves to the blue square @xmath501 , and continues through the blue square to the base of the blue band @xmath412 and then up along the blue band .
this part contains at most @xmath502 turns .
the second part of @xmath135 begins with a 90 degree from the blue band @xmath412 to the extension band , moves through the extension band , onto the middle section of the connecting rectangle with at most 3 turns , then moves up to the new blue square with at most 4 more turns , once in the blue square , it may need up to another 6 turns to be re - routed to enter the middle square of the blue square in the correct order ( to achieve the desired linear order ) , then finally move up to the base of the new blue triangle with one more turn .
this part of the path will add at most 15 right angle turns for the second part ( and 14 total ) in this case . in case
there is a hook , more turns are needed for the second part . as before , the edge turns from the blue band onto the extension band , moves through the extension band , across the connecting rectangle , runs through the small hook , returns to the middle section of the connecting rectangle and then move up a level to a new blue square and finally to the base of a blue triangle . ends in one of the blue vertices @xmath508 of @xmath382 on the connecting rectangle .
the extra number of gridlines needed for the loop is 2 , as already established during our earlier discussion for the red - red edges .
thus @xmath509 for an edge - cut .
we now consider the case of a vertex - cut . here
the situation is easier since the red - blue edge @xmath135 is not cut into parts at all .
it already exists in one of the two graphs @xmath378 , @xmath510 , @xmath172 and it is simply extended from the blue band @xmath503 to the single blue vertex @xmath511 on the connecting rectangle .
it is easy to see that this extension does not exceed @xmath512 turns and thus the results from the edge - cut case suffices as an upper bound for @xmath491 .
we are now ready to prove the main theorem of this paper as stated below .
[ maint ] let @xmath41 be a 4-regular plane graph with @xmath6 vertices .
then there exists a realization of @xmath41 on the cubic lattice @xmath513 which is contained in a rectangular box whose volume is bounded above by @xmath514 .
consequently , the ropelength of any knot or link @xmath9 is at most of the order @xmath515 , where @xmath516 is the minimum crossing number of @xmath9 . to prove the theorem
, we will use the non - linear recurrence analysis approach as described in @xcite .
the result of theorem [ sumextconstr ] and the results from the earlier sections will be needed when we apply this analysis . for any rectangle @xmath57 under discussion
we assume that its length @xmath194 is greater or equal to its width @xmath138 .
the _ aspect ratio _
@xmath517 of the rectangle @xmath57 is defined by @xmath518 . from this
it follows that @xmath519 and @xmath520 where @xmath521 is the area of the rectangle @xmath57 .
the aspect ratio is important to us because throughout the algorithm we want to operate with boxes whose base rectangles are not too skinny , that is the rectangles have an aspect ratio that is bounded away from zero by a positive constant . in the previously described divide - and - conquer algorithm
we divide a given brt - graph at each step into subgraphs ( also brt - graphs ) each of which has a size of at least 1/6 of the previous graph , where the size is measured by the number of red vertices in a graph .
this allows us to operate with rectangles whose aspect ratios are at least 1/6 . before getting started on the details we need a preliminary lemma that asserts that we can divide rectangles while preserving the minimal aspect ratio of 1/6 .
[ aspectratio ] let @xmath57 be a rectangle with length @xmath194 and width @xmath138 and aspect ratio @xmath522 .
let @xmath523 be a real number and divide the rectangle @xmath57 by a line that is parallel to its width into two rectangles @xmath81 and @xmath448 of areas @xmath524 , @xmath525 and aspect ratios @xmath526 , @xmath527 respectively , then @xmath528 for each @xmath281 . it suffices to show this for one of the aspect ratios , say @xmath526 of the rectangle @xmath81 .
we can assume that the area of @xmath529 with dimensions @xmath530 and @xmath138 .
there are two cases to consider , either @xmath531 or @xmath532 . if @xmath531 then @xmath533 and @xmath534 .
if @xmath532 then @xmath535 and @xmath536 . from the previous sections we know that a brt - graph @xmath41 with @xmath6 red vertices has a balanced recursive subdivision sequence whose depth is bounded above by @xmath537 for some constant @xmath538 , see theorem [ trecexist ] . at the end of the subdivision process
we have terminal brt - graphs @xmath299 that satisfy @xmath539 .
that is , each such terminal brt - graph has less than @xmath260 red vertices .
by lemma [ vbbound ] , the number of blue vertices in @xmath299 is bounded above by @xmath540 ( since the @xmath226 used in lemma [ vbbound ] is equal to 12 ) .
thus all graphs @xmath299 have at most @xmath541 vertices ( blue and red combined ) .
since there are only finitely many plane brt - graphs with at most @xmath260 red vertices and whose maximum vertex degree is 12 or less , the following lemma holds for a grid - like embedding @xmath542 of a terminal brt - graph @xmath299 .
[ n0lemma ] there exists an integer @xmath543 such that if @xmath299 is a terminal brt - graph ( so it has less than @xmath260 red vertices ) and @xmath57 is any rectangle of area @xmath544 with an aspect ratio @xmath545 , then @xmath57 is large enough such that the rectangular box @xmath546 of height 1 with @xmath57 as its base rectangle can be used to hold a grid - like embedding @xmath542 of @xmath299
. moreover there exists a constant @xmath547 that only depends on @xmath260 such that for any such grid - like embedding @xmath542 in the box @xmath546 , the turning number @xmath548 .
let us now examine what happens in the reconstruction process .
let @xmath308 be the depth of the subdivision sequence .
recall that when two grid - like embeddings are reconnected , the turning number of a red - blue edge that remains a red - blue edge increases by at most @xmath549 for some constant @xmath549 ( @xmath549 can be chosen to be @xmath512 in fact , see theorem [ sumextconstr ] ( ii ) ) . therefore if a grid - like graph @xmath550 is reconstructed from the grid - like embeddings @xmath551 and @xmath552 in a box @xmath553 , the height of @xmath553 is @xmath554 and @xmath555 .
inductively , we can see that if a grid - like graph @xmath542 is reconstructed from the grid - like embeddings @xmath556 and @xmath557 in a box @xmath546 , then the height of @xmath546 is @xmath558 and @xmath559 for any @xmath560 . by remark [ depthbound ] , we can then claim the following .
\(i ) the height of the box @xmath546 is bounded above by @xmath561 ; \(ii ) @xmath562 is bounded above by @xmath563 . if a red - red edge @xmath135 is created in @xmath542 by connecting two red - blue edges ( one in @xmath551 and one in @xmath552 ) , then the turning number of the resulting edge @xmath135
is the sum of the two turning numbers of the two red - blue edges in @xmath551 and @xmath552 , as well as an addition of at most 6 , as shown in theorem [ sumextconstr ] ( i ) . by the inequality given in ( ii ) above , the turning number of the edge @xmath135
is then bounded above by @xmath564 .
however the turning number of @xmath135 no longer changes in the subsequent reconnection process .
we can summarize this as follows : \(iii ) let @xmath299 be a brt - graph that is subdivided into @xmath301 and @xmath302 .
then for any newly created red - red edge @xmath135 in the construction process of @xmath542 we have @xmath565 in the reconstruction of @xmath542 from @xmath566 and @xmath567 , for each new red - red edge @xmath135 created , we need to add up to @xmath491 many gridlines to place @xmath135 on the lattice .
the number of new red - red edges created at that stage is the same as the number of red - red edges cut by @xmath175 .
since the number of edges cut by @xmath175 is proportional to the square root of the total number of the vertices in the brt - graph @xmath299 , by lemma [ vbbound ] , the number of edges cut is proportional to the square root of the red vertices of @xmath299 .
thus the total number of gridlines which may have to be added to create the rectangle @xmath568 from the rectangles @xmath569 and @xmath570 is at most @xmath571 for some constant @xmath221 that only depends on the construction algorithm ( by the inequalities we obtained from ( i)(iii ) above ) .
let @xmath572 be the minimum over all positive integers @xmath573 with the property that if the area of a rectangle @xmath57 with aspect ratio @xmath574 is equal to or greater than @xmath573 , then any brt - graph @xmath41 with @xmath6 red vertices whose degrees are at most @xmath575 has a grid like embedding in a box @xmath576 $ ] where @xmath577 is a positive integer and @xmath578 . in the following theorems @xmath119 or @xmath579 is always a number in [ 1/6 , 5/6 ] .
however only those values in [ 1/6 , 5/6 ] whose product with @xmath6 is an integer make sense where @xmath6 is the number of red vertices of the graph that is subdivided . the reader should keep this in mind in the statement and proof of theorem [ upperboundan ] below .
[ upperboundan ] there exists a constant integer @xmath580 such that the function @xmath581 defined by @xmath582 bounds @xmath572 above , where @xmath436 is the same constant used in equation [ bound_const ] .
we will prove this theorem inductively .
we choose @xmath580 as the constant guaranteed by lemma [ n0lemma ] . by definition
, we have @xmath583 if @xmath584 .
now assume that @xmath585 is true for any @xmath586 for @xmath587 , we need to prove that @xmath588 .
let @xmath545 be any given aspect ratio and @xmath57 a rectangle with area @xmath581 and an aspect ratio @xmath517 .
let @xmath41 be a brt - graph with @xmath6 red vertices .
we know that there exists a subdivision of @xmath41 that divides @xmath41 into two brt - graphs @xmath42 and @xmath43 with @xmath589 and @xmath590 red vertices respectively for some @xmath591 $ ] .
let @xmath440 be a rectangle with aspect ratio @xmath517 and area @xmath592 .
by lemma [ aspectratio ] , @xmath440 can be divided into two rectangles @xmath81 and @xmath448 whose aspect ratios are in @xmath593 $ ] and whose areas are @xmath594 and @xmath595 respectively . thus by our induction hypothesis , there exist grid - like embeddings @xmath383 , @xmath384 of @xmath42 , @xmath43 respectively in the rectangular boxes @xmath390 , @xmath391 whose bases are @xmath81 , @xmath448 and whose heights are at most @xmath596 , @xmath597 respectively .
suppose that @xmath440 is placed in such a way that its longer side is parallel to the @xmath12-axis as shown in figure [ rectangle ] .
we will now extend @xmath440 to a new rectangle @xmath598 by adding a strip of width @xmath599 to its top and adding a strip of width @xmath600 to its right .
the strip added to the right of @xmath440 is thicker than the strip added to its top since @xmath601 .
observe that @xmath598 also has aspect ratio @xmath517 .
, @xmath81 , @xmath448 and @xmath598 .
the rectangles @xmath81 and @xmath448 are the base rectangles of the boxes @xmath390 and @xmath391 . ] we know that we can obtain a grid - like embedding of @xmath41 from @xmath413 ( which is bounded in @xmath390 with @xmath81 as its base ) and @xmath414 ( which is bounded in @xmath391 with @xmath448 as its base ) . at most @xmath602 new horizontal and @xmath599 new vertical grid lines
are needed to accommodate the red - red edges ( which need to be put on the lattice ) .
this means that the base of the new rectangular box @xmath348 ( that houses this new grid - like embedding of @xmath41 ) can be fit into @xmath598 .
furthermore , the height of @xmath348 is 2 plus the larger of the heights of @xmath390 and @xmath391 .
say the height of @xmath390 is larger .
then the height of @xmath348 is at most @xmath603 . since @xmath604 , @xmath605 .
thus the height of @xmath348 is at most @xmath606 . in other words
, @xmath41 has a grid - like embedding in the box @xmath607 with @xmath598 as its base and with a height at most @xmath608 .
let @xmath609 and @xmath610 be the width and length of @xmath440 respectively
. then @xmath611 and @xmath612 , so @xmath613 .
thus the area of @xmath598 is @xmath614 however since @xmath615 , @xmath616 , the area of @xmath598 is bounded above by @xmath617 since @xmath57 and @xmath598 have the same aspect ratio and @xmath57 has a larger area , @xmath598 can be fit into @xmath57 .
therefore , @xmath41 has a grid - like embedding in a box with base @xmath57 and height at most @xmath608 .
this proves that @xmath588 .
the next theorem gives the function @xmath581 ( hence @xmath572 ) an explicit bound .
there exists a constant @xmath618 such that @xmath619 it follows that @xmath620 as well .
following the proof given in @xcite we define a function @xmath621 as follows : for @xmath622 , @xmath623 and for @xmath624 @xmath625 we show that @xmath626 for all @xmath627 by induction .
clearly this is true for @xmath628 .
assume that @xmath629 is true for all values of @xmath87 such that @xmath630 .
for @xmath631 we have : @xmath632 the third line in the above inequalities can be explained as follows : if @xmath633 for the value of @xmath119 realizing the maximum then this is obvious . if @xmath634 then the same result follows where @xmath119 is replaced by @xmath635
. then a change of variable of @xmath635 for @xmath119 produces the same result .
now @xmath621 needs to be estimated . using equation ( [ eqn1 ] ) repeatedly for different values of @xmath119 results in : @xmath636 where @xmath637 is the depth of the recursion and each value
@xmath579 is the value of @xmath119 that realizes the maximum at each stage of the recursion .
since all @xmath638 we have for some constant @xmath437 @xmath639 from this it follows that there exists a constant @xmath618 such that @xmath640 .
[ graphlatticeembed ] let @xmath41 be a 4-regular plane graph with @xmath6 vertices .
then there exists a realization of @xmath41 on the cubic lattice @xmath513 which is contained in a rectangular box whose volume is bounded above by @xmath514 .
first , we change @xmath41 to a brt - graph @xmath641 by triangulation .
next we construct a grid - like embedding of @xmath641 in the lattice in a box @xmath642 by the previously described algorithm .
the number of vertices in @xmath41 is equal to the number of red vertices in @xmath641 .
the algorithm described in the paper generates a rectangular box with a height of at most @xmath643 for some constant @xmath644 . from theorem [ upperboundan ]
we know that the area of the rectangle @xmath57 is of the order @xmath645 and thus the volume of the box containing the grid - like embedding of @xmath41 is bounded above by @xmath646 .
the lattice embedding described in the prior section assumes that edge segments in the red rectangles are not on lattice . in order to obtain a complete lattice embedding of @xmath41 ,
the edges which are added to @xmath41 in the triangulation to obtain the brt - graph @xmath641 are removed from the embedding .
now each red rectangle only intersects 4 edges and these edges can be re - arranged in the rectangle such that they connect to the red vertex using only lattice connections .
thus the lattice embedding of @xmath41 fits into the same rectangular box as the grid - like embedding of @xmath41 .
let @xmath268 be a knot diagram of @xmath0 with @xmath6 crossings .
@xmath268 can be thought of as a 4-regular plane graph and by corollary [ graphlatticeembed ] there exists a lattice embedding of the 4-regular plane graph @xmath268 in a rectangular box with a volume of order @xmath646 . in order to change this lattice embedding of the graph @xmath268 to the embedding of the knot @xmath0 , we need to recover the crossings of the original knot diagram from the vertices of the embedded graph . in order to accomplish this ,
the edges at a vertex @xmath46 are locally modified as shown in figure [ crossingrecover ] .
one unit away from the vertex @xmath46 the edge is rerouted through the plane @xmath647 to recover the desired crossing .
the space between @xmath93 and @xmath647 and the space between @xmath647 and @xmath648 are then stretched to one unit thick each ( so that @xmath647 becomes the lattice plane @xmath648 afterwards ) .
the new rectangular box containing the recovered knot is identical to the previous one except that its height has increased by one .
the volume of the surrounding box is still bounded above by @xmath646 .
the length of the lattice embedding of a diagram @xmath268 is bounded above by the volume of the box surrounding the lattice embedding of @xmath268 .
furthermore , the ropelength of a knot @xmath0 is bounded above by twice the length of the lattice embedding .
the result of our main theorem ( theorem [ maint ] ) then follows .
theorem [ maint ] answers the following question raised in @xcite negatively . since we know that there exist knots @xmath0 with ropelength of order @xmath30 , the only possible further improvement on theorem [ maint ] is to reduce the power on the @xmath654 term . in @xcite ,
a numerical study suggests the possibility of a ropelength upper bound of the form @xmath655 .
we thus end this paper with the following question : j. arsuaga , m. vazquez , s. trigueros , d. w. sumners and j. roca , _ knotting probability of dna molecules confined in restricted volumes : dna knotting in phage capsids .
usa * 99 * ( 2002 ) , 53735377 . | for a knot or link @xmath0 , let @xmath1 be the ropelength of @xmath0 and @xmath2 be the crossing number of @xmath0 . in this paper , we show that there exists a constant @xmath3 such that @xmath4 for any @xmath0 .
this result shows that the upper bound of the ropelength of any knot is almost linear in terms of its minimum crossing number , and is a significant improvement over the best known upper bound established previously , which is of the form @xmath5 .
the approach used to establish this result is in fact more general .
in fact , we prove that any 4-regular plane graph of @xmath6 vertices can be embedded into the cubic lattice with an embedding length at most of the order @xmath7 , while preserving its topology . since a knot diagram can be treated as a 4-regular plane graph . more specifically , although the main idea in the proof uses a divide - and - conquer technique , the task is highly non - trivial because the topology of the knot ( or of the graph ) must be preserved by the embedding . |
the description of compton scattering by the nucleon via dispersion theory was developed at the beginning of the 1960s @xcite . the theory made use of relations which may be denoted as @xmath21-channel dispersion relations and @xmath22-channel dispersion relations .
the singularities entering into the @xmath21-channel dispersion relations may be taken from meson photoproduction experiments , whereas the singularities entering into the @xmath22-channel dispersion relations are related to @xmath23 pairs created in two - photon reactions in case of the scalar @xmath22-channel , and to the @xmath24 meson in case of the pseudoscalar @xmath22-channel .
the first application of the scalar @xmath22-channel to the polarizabilities of the nucleon came with the work of j. bernabeu , t.e.o ericson et al .
( beft ) in 1974 @xcite where it was shown that the largest part of the electric polarizability and the total diamagnetic polarizability are due to the @xmath22-channel .
the smaller part of the electric polarizability is due to the `` pion cloud '' showing up as a nonresonant meson photoproduction process .
the paramagnetic polarizability is mainly due to the photoabsorption cross section provided by the @xmath25 resonance .
a convenient version of a dispersion theory applicable in a wide angular interval and at energies up to 1 gev was developed by lvov et al .
this dispersion theory is of the fixed-@xmath22 variety where @xmath21-channel integrals are carried out along integration paths at constant @xmath22 , and the @xmath22-channel contributions are taken care of in the form of `` asymptotic '' contributions , being an equivalent of the @xmath22-channel contributions . in principle
this version of dispersion theory is completely equivalent to other versions as there are fixed-@xmath26 dispersions theories or hyperbolic dispersion theories . at first sight
there appears to be a disadvantage in case of fixed-@xmath22 dispersion theories because the integration paths are partly located in the unphysical range of the scattering plane .
however , it has been shown by lvov et al .
@xcite that this only leads to minor technical problems which can be solved without loss of precision .
the validity of this latter statement has been clearly demonstrated by experiments , showing that fixed-@xmath22 dispersion theories lead to a precise representation of the experimental differential cross sections in the whole angular interval and at energies up to 1 gev @xcite . in this respect
the nonsubtracted dispersion theory differs from the subtracted dispersion theory @xcite which looses validity already at the peak energy of the @xmath27 resonance . for the spin - independent @xmath22-channel contribution
the assumption was made that it is possible to represent it via a scalar @xmath19-meson pole - term , in analogy to the well - known pseudoscalar @xmath24 pole term as entering into spindependent amplitudes . with this representation it was possible to arrive at agreement with experimental data in the angular and energy ranges described above , whereas without this representation of the pole term there remained a large discrepancy between prediction and the experimental data .
this implies that at energies of the second resonance region of the nucleon and large scattering angles the @xmath19-meson pole makes a dominant contribution to the compton differential cross section .
furthermore , the @xmath19-meson mass was determined to be @xmath28 mev @xcite in agreement with other available data .
these first investigations published in 2001 @xcite remained preliminary because the @xmath19-meson pole was only an ansatz at that time and there was an uncertainty about its validity .
this uncertainty has been removed in later investigations where it was shown that the @xmath19-meson pole term has a very firm theoretical basis and that the quantitative predictions calculated from well known properties of the @xmath19-meson are precise ( see @xcite for a summary ) . as a conclusion we may state that the dispersion theory of lvov et al .
@xcite is precise and well tested .
therefore , there are good reasons to trust in the evaluations of electromagnetic polarizabilities from low - energy compton - scattering data where use is made of this type of dispersion theory .
this is the case for all experimental data which are summarized in @xcite leading to the _ recommended _ nucleon polarizabilities given in the abstract . for the neutron also electromagnetic scattering of slow neutrons
has been taken into account .
furthermore , these _ recommended _ values are in excellent agreement with independent predictions obtained from high - precision cgln amplitudes @xcite and well - investigated properties of the @xmath19 meson without making use of experimental compton differential cross sections @xcite .
in addition to dispersion theory @xmath29eft plays a prominent rle in current investigations of nucleon compton scattering and polarizabilities .
the present investigation is motivated by the fact that recently @xmath29eft has been used as a tool for analyses of experimental differential cross sections for compton scattering by the proton , and results have been obtained which are considerable different from the well - founded standard values @xmath0 and @xmath30 .
examples are the chpt - investigation of beane et al .
@xcite where the values @xmath31 , @xmath32 have been obtained , and mcgovern et al .
@xcite where @xmath33 , @xmath34 have been obtained .
these values have been included in the data listing of the particle data group @xcite though the magnetic polarizabilities @xmath35 of @xcite deviate from the respective _ recommended _ value @xcite by a factor @xmath36 and the electric polarizability @xmath37 of @xcite by @xmath36 standard deviations .
these new results are based on the same set of experimental data as the previous ones @xcite so that the deviations are solely due to differences in the methods of data analysis .
it is obvious that the new analyses are justified only if they are superior to or at least of the same quality as the previous ones .
the purpose of the present work is to explore whether or not this is the case .
in a recent article @xcite a complete description of the present status of dispersion theory of nucleon compton scattering and polarizabilities has been given .
the _ recommended _ experimental polarizabilities introduced in the 2005 summary @xcite have been confirmed in the 2013 summary @xcite except for a slight correction applied to @xmath38 .
the reason for this slight correction was due to the fact that new high - precision analyses of cgln amplitudes have become available @xcite which made this revision necessary .
these new high - precision analyses @xcite were also of great importance in connection with the prediction of the @xmath21-channel components of the nucleon polarizabilities for all resonant and nonresonant excitation processes of the nucleon . results of these analyses which are of interest here are summarized in table [ tab1 ] .
.@xmath21-channel electromagnetic polarizabilities compared with experimental data . [ cols="<,^,^,^,^",options="header " , ] [ lensky ] the variants considered in @xcite are as follows : + i : the results for compton scattering with nucleon and @xmath24 born graphs ( tree graphs ) , plus polarizabilities as given by the counterterm ( c.t . ) .
+ ii : tree graphs plus the effects of the ( dressed ) @xmath27 @xmath21- and @xmath39-channel pole graphs .
+ iii : tree graphs plus @xmath20n loops : the @xmath40 calculation in heavy - baryon @xmath29eft without an explicit @xmath27 degree of freedom .
+ iv : as ( iii ) but with relativistic nucleon propagator in the @xmath20n and @xmath41 loops .
+ v : the @xmath42 calculation in heavy - baryon @xmath29eft with explicit @xmath27 , including tree graphs , @xmath27 poles and hb @xmath20n and @xmath41 loops .
+ vi : as ( v ) , but with relativistic nucleon propagators in the @xmath20n and @xmath41 loops .
according to the authors @xcite all of these variants of @xmath29eft are based on the same low - energy symmetries of qcd and in this sense are equivalent .
but only the last two versions v and vi in table [ lensky ] are realistic calculations which can be compared with experimental compton differential cross sections in the resonance region and some way below .
a comparison with experimental data at angles from @xmath43 to @xmath44 shows that these two variants of @xmath29eft which both include the leading @xmath20n loop effects and an explicit @xmath27 are quite similar , provided the counterterms ( c.t . )
are included and adjusted to yield identical values for the scalar dipole polarizabilities .
however , the values found for the two sets of counterterms in the @xmath29eft variants considered here are rather different .
they are particularly large in the @xmath45 calculation ( column v ) , especially as compared to the relativistic @xmath41 one ( column vi ) .
the statements of the authors @xcite cited in the forgoing paragraph indicate that in case of @xmath29efts a major part of the interest is directed to the low - energy symmetries of qcd , whereas the precision of the numbers attributed to the different components of the polarizabilities is of secondary importance .
this attitude contrasts with the aim of the non - subtracted dispersion theory where methods have been developed which lead to a high precision of the different components of the polarizabilities .
for the data analysis certainly high precision is required .
\(i ) the analysis of compton scattering and polarizabilities in terms of nonsubtracted dispersion relations is precise and well tested experimentally in a large angular range and at energies up to 1 gev . in this respect
the nonsubtracted dispersion theory differs from the subtracted dispersion theory @xcite where the latter looses validity already at the peak energy of the @xmath27 resonance .
+ ( ii ) the polarizabilities of the nucleon are determined from experimental data in two ways , firstly by adjusting the predictions of the nonsubtracted dispersion theory to the experimental differential cross sections for compton scattering in the low - energy domain below the @xmath25 peak and secondly by calculating them from the @xmath21-channel and @xmath22-channel singularities as provided by high - precision analyses of cgln amplitudes and well known properties of the scalar @xmath22-channel . in the latter case
no use is made of compton scattering differential cross sections .
the results obtained by these two methods are in excellent agreement with each other .
the paper of beane et al .
@xcite cited in the introduction is essentially of the variety iii in table [ lensky ] .
the amplitude for unpolarized compton scattering is expanded in powers of the parameter @xmath46 used in efts , representing a typical external momentum . to @xmath47 the polarizabilities
are given by pion - loop effects and lead to the hbchpt prediction . at @xmath48
there are new long - range contributions to these polarizabilities .
four new parameters also appear which encode contributions of short - distance physics to the spin - independent polarizabilities .
thus one needs four pieces of experimental data to fix these four short - distance contributions .
after these adjustment to experimental data have been carried out good fits to experimental differential cross sections have been obtained .
the paper of mc govern et al .
@xcite cited in the introduction includes effects of the @xmath27-resonance and , therefore , essentially is of the variety v in table [ lensky ] . the foregoing analysis has shown that by `` integrating out '' @xmath18- and @xmath17- degrees of freedom and by using the derivative coupling alone , the @xmath49 component of the electric polarizability predicted by chpt is increased by a factor @xmath50 compared to the correct empirical value .
furthermore , by carrying out the kinematical modifications entering into hbchpt a further increase by a factor of 1.7 ( proton ) and 1.2 ( neutron ) is obtained .
the `` integrating out '' of the @xmath19-meson leads to the omission of the major part , @xmath51 , of the electric polarizability and to the omission of the diamagnetic polarizability @xmath52 .
the author is indebted to a.i .
lvov , v. pascalutsa and h.w .
griesshammer for valuable comments .
m. schumacher , eur .
j. a * 30 * ( 2006 ) 413 , erratum * 32 * ( 2007 ) 121 , [ hep - ph/0609040 ] ; m.i .
levchuk , a.i .
lvov , a.i .
milstein , m. schumacher , proceedings of the workshop on the physics of excited nucleons , world scientific , nstar 2005 ( 2005 ) 389 , [ hep - ph/0511193 ] . | based on compton scattering and meson photoproduction data the polarizabilities of the nucleon are precisely studied and well understood due to recent experimental and theoretical work using nonsubtracted dispersion relations .
the _ recommended _ experimental values are @xmath0 , @xmath1 , @xmath2 , @xmath3 , @xmath4 , @xmath5 , @xmath6 , @xmath7 in units of @xmath8@xmath9 and @xmath10 , @xmath11 , @xmath12 , @xmath13 , @xmath14 , @xmath15 in units of @xmath8@xmath16 @xcite .
the numbers given in parentheses are predicted values .
it is shown that all versions of chiral effective field theories applied in analyses of nucleon polarizabilities and compton scattering ignore essential effects of @xmath17 , @xmath18 and @xmath19 exchanges and of pseudoscalar @xmath20n coupling .
+ [ 1ex]martin schumacher + ii .
institut der universitt gttingen , friedrich - hund - platz 1 + d-37077 gttingen , germany |
the inner kiloparsecs of starburst and interacting galaxies harbor stunning amounts of molecular gas , @xmath15 ( e.g. , scoville 1991 ; bryant & scoville 1996 ) . in these environments ,
molecular clouds are subject to intense radiation fields , supernovae explosions , winds from newborn hot stars , strong tidal forces , and gas surface densities several order of magnitudes higher than in the milky way disk . these are also extremely active star formation sites .
knowledge of the physical conditions and structure of molecular gas in interacting systems is essential to understand the starburst activity and its role in galaxy evolution .
arp 299 is an ir - luminous ( @xmath16 l@xmath17 ) merger system of two galaxies , ic 694 and ngc 3690 .
strong emission has been detected from the nuclei of ic 694 and ngc 3690 and from the interface between the two galaxies ( solomon & sage 1988 ; casoli 1989 ; sargent 1987 ; sargent & scoville 1991 ) .
the two nuclei , as well as the western overlap region , currently harbor intense star formation activity ( c.f .
, gehrz 1983 ; baan & haschick 1990 ) .
furthermore , the nucleus of ic 694 is a flat - spectrum radio source , and may be an agn ( gehrz 1983 ) .
lower resolution ( single dish ) observations reveal an unusually large / 10 line intensity ratio , @xmath18 in arp 299 ( aalto 1991 ; casoli 1992 ) .
these observations left it unclear whether this is due to weak in the whole system or to a varying / 10 line ratio .
observations at 20@xmath19 and 11@xmath19 resolution by casoli ( 1992 ) suggest little variation in the / 10 line ratio and none in the / 21 line ratio .
in contrast , aalto ( 1995 ) note substantial variations at 28@xmath19 resolution in the / 21 ratio : the ratio is about 17 in ic 694 ; close to 9 in ngc 3690 ; and about 7 in the interface region between the two disks .
solomon ( 1992 ) detected bright hcn emission in 28@xmath19 maps of arp 299 .
they measured /hcn ratios of 11 in ic 694 and 13 in the interface region .
aperture synthesis , , and hcn 10 mapping of arp 299 was carried out with the owens valley radio observatory ( ovro ) millimeter array between march , 1995 , and february , 1996 .
sis receivers on the six 10.4 m telecopes provided typical system temperatures ( ssb ) of 600 k , 450 k , and 350 k for , , and hcn .
quasars 1150 + 497 and 0917 + 449 were used for phase calibration and uranus and neptune for absolute flux calibraton .
the synthesised beams are @xmath20 for ( uniform weighting ) , @xmath21 for ( natural weighting ) , and @xmath22 for hcn ( natural weighting ) . at 2.6 mm wavelength with @xmath0 resolution
, a brightness temperature of 1 k corresponds to 57 mjy beam@xmath23 .
the digital correlator , centered at @xmath24 , provided a total velocity coverage of 1123 for , 1175 for , and 1407 for hcn .
data were binned to 4 mhz resolution , corresponding to 10 for and and 13 for hcn . at 110 ghz
an unresolved continuum source of @xmath25 mjy was detected in the center of ic 694 .
continuum emission was also detected in ngc 3690 ( @xmath26 mjy ) and the overlap region ( @xmath27 mjy ) .
we subtracted this continuum emission from the line emission before maps were made .
the main structures found by sargent & scoville ( 1991 ) with the three telescopes array are recovered in our new map ( figure 1a ) , but our increased @xmath28 coverage enable improved deconvolution . lower surface brightness emission ( a2 ) ,
possibly a molecular disk or bar , extends 10 - 15@xmath19 southeast of the ic 694 nucleus ( a1 ) , coincident with the remnant optical disk .
the center of ngc 3690 ( b1 ) is also surrounded by weak , extended emission ( b2 and b3 ) with a somewhat s - shaped morphology .
where the two galaxies overlap , three distinctive clumps ( c1 , c2 , and c3 ) can now be discerned .
weaker , extended emission is also recovered better in our new map and clumpy structures can be distinguished at the center of the map ( f ) .
these structures appear to connect the major components a , b , and c. in addition , there is a clump ( d ) north of the main features with systemic velocity @xmath29 3280 , which is beyond the bandwidth of the earlier ovro data .
bright 6 cm radio continuum peaks at the nuclei of ic 694 and ngc 3690 ( gehrz 1983 ; condon 1991 ) coincide with the peaks to within the estimated positional uncertainty ( @xmath30 ) .
there is also reasonable positional agreement between the two brightest clumps in the overlap region ( c ) and two additional 6 cm radio continuum peaks : c1 and the western radio continuum peak are also coincident within @xmath30 , although the discrepancy between c2 and the eastern radio continuum peak is somewhat larger , @xmath31 .
weaker radio continuum emission at 18 cm and 6 cm is spatially coincident with the extended emission in regions a2 , b2 , b3 , and f ( baan & haschick 1990 ; gehrz 1983 ) .
the velocity field image ( figure 2 , plate 1 ) reveals a monotonic shift from the blueshifted emission from a2 in ic 694 to redshifted emission from the overlap region c and from region d. velocity gradients within c and d are small .
the velocity field of ngc 3690 is complicated by a double - peaked emission structure of b2 , and both b2 and b3 appear blueshifted relative to the center , b1 .
the nuclei of both ic 694 ( a1 ) and ngc 3690 ( b1 ) remain unresolved by our @xmath32 synthesised beam . a two - dimensional gaussian fit to the nuclear emission of ic 694 and ngc 3690 yields upper limits to the source diameters of @xmath33 and @xmath34 , corresponding to radii of 140 and 150 pc , respectively ( for @xmath35=42 mpc ) .
this implies a lower limit of 18 k to the brightness temperature in the ic 694 nucleus , and therefore , the cloud filling factor is quite high in the inner 200 pc . even if the intrinsic brightness temperature is as high as 100 k , the surface filling factor of clouds is still almost 20% .
the three features in the overlap region are resolved , with sizes of @xmath36 pc ( c1 ) , @xmath37 pc ( c2 ) , and @xmath38 pc ( c3 ) .
derived properties for all designated regions are presented in table 1 .
the total molecular mass of arp 299 , estimated from @xmath39 m@xmath17 ( @xmath35 is the distance in mpc ; @xmath40 is the integrated 10 line flux in jy k kms@xmath23 ) , is @xmath41 m@xmath17 , 87% of the value estimated by solomon and sage ( 1988 ) from their single dish map .
this formula corresponds to @xmath42(h@xmath43)/@xmath44()=@xmath45 @xmath46 ( k ) @xmath23 ( sargent & scoville 1991 ) , a standard galactic luminosity to h@xmath43 mass ratio .
the conversion factor may vary , however , across arp 299 , since the line ratios indicate different cloud properties .
the most striking feature of the 10 map ( figure 1b ) is the _ absence of strong emission at the nucleus of ic 694 ( a1)_. emission _ is _ detected in the a2 region of ic 694 , in the overlap region c , and in ngc 3690 .
unlike casoli ( 1992 ) we find significant variation in the /10 line ratio across arp 299 .
> from the very high value of @xmath4 at the ic 694 nucleus ( a1 ) the ratio drops to @xmath5 in the a2 region .
the global ratio for ngc 3690 is @xmath47 , but there is an indication that the ratio is higher in the nucleus , b1 , than in b2 or b3 .
the ratio for the east ( c3 ) and west ( c1 ) overlap region is @xmath9 and @xmath6 respectively .
the spatial correlation between the emission and radio continuum emission peaks is also poor . as for
, the hcn map ( figure 1c ) is dominated by a peak at the center of ic 694 ( a1 ) , where the /hcn line ratio is @xmath7 .
emission is also detected at the nucleus of ngc 3690 , where the ratio is @xmath8 , and in the western ( c1 ) overlap region , where /hcn= @xmath9 .
there is very little hcn in regions a2 and c3 .
the hcn and line emission peaks appear anticorrelated .
as a system , arp 299 is not deficient in . in all regions but
the core of ic 694 , we observe / line ratios typical of star forming regions in other galaxies ( e.g. , aalto 1995 , 1991 ; young & sanders 1986 ; rickard & blitz 1985 ) . relative to , 10 is brightest in quiescent regions of low surface brightness , and weak in starburst regions and galactic nuclei . in contrast , hcn 10 , like , is bright in the two galaxy centers and in regions of active star formation .
these line ratio variations are , most likely , caused by differences in line excitation .
our results support the suggestion ( aalto 1995 ) that _ unusually _ high ( i.e. @xmath48 ) / 10 line ratios tend to be associated with extremely compact molecular distributions centered on the nuclei of merging galaxies and are primarily due to a small or moderate optical depth , @xmath10 , in the 10 line .
high ambient pressures , strong tidal forces and ongoing starburst or agn activity lead to substantial changes in cloud structure and physical conditions .
bryant ( 1996 ) also finds that hcn 10 is bright in regions of high surface brightness in merging galaxies . the line ratio variation from the ic 694 disk , where / = 10 and /hcn @xmath49 25 , to its nucleus , where the corresponding values are 60 and 9 , reflects a dramatic change in cloud properties .
the bright hcn line accompanied by relatively weak emission ( hcn/ = @xmath50 ) implies a population of unusually dense and warm clouds .
the hcn 10 strength implies densities @xmath51 @xmath13 if the hcn excitation is dominated by collisions with h@xmath43 .
it is also likely that the density is @xmath52 , so that most of the hcn population will remain in the lower levels and @xmath53 . at these densities
, the and 10 transitions are thermalised .
if the kinetic temperature is also high , the lower levels may become significantly depopulated , effectively reducing the optical depth of the 10 line .
then the 21/10 line ratio should be @xmath11 1 . comparing the single dish 21 flux ( aalto 1995 ) with our 10 flux from the nucleus of ic 694 , we estimate 21/10 @xmath54 , implying that the gas temperature is high , @xmath11 50 k. although the single dish beam was large , 28@xmath19 , the bulk ( @xmath5565% ) of the 21 emission within the beam originates in the nucleus of ic 694 ( a1 ) .
since the lower transitions of and appear thermalised , lte can be assumed to estimate the column density , @xmath42 ( ) . we find it unlikely that the / abundance ratio is extremely high ( section 3.3 ) .
therefore , a high / 10 line ratio indicates a low to moderate optical depth ( @xmath10 ) in the 10 line .
the high intrinsic 10 brightness temperature makes @xmath56 unlikely , and we therefore assume @xmath57 .
the optical depth of the 10 transition can be expressed as @xmath58 .
for a temperature @xmath59=100 k , line width @xmath60 , and @xmath61=1 , the column density @xmath62 @xmath46 ( per cloud ) and the resulting brightness temperature @xmath63 k. for a density of @xmath64 @xmath13 and a abundance , [ /h@xmath43]=@xmath66 , the cloud radius @xmath67=0.7 pc not unlike cores or clumps within giant molecular clouds in the galaxy . at this gas density
, hcn is not thermalised so @xmath68 for the 10 line will be significantly lower than that for . if the abundance ratio [ /hcn ] @xmath69 , then @xmath61(hcn ) is an order of magnitude higher than @xmath61 ( ) .
the resulting @xmath70(hcn ) is sufficiently high to account for a /hcn line ratio of 9 .
above , we infer a clumpy molecular medium because we chose a @xmath71 = 5 , yielding small clouds with @xmath72 pc ( cf aalto 1991 ) .
we can not , however , exclude significantly larger @xmath71 which would indicate a continous , non - cloudy structure perhaps even a smooth , rotating disk .
a third alternative is a molecular ism consisting of dense clumps surrounded by diffuse , non - cloudy molecular gas ( e.g. aalto 1994,1995 ; dahmen 1996 ) .
molecular line ratio variations are also seen within the overlap region c , albeit smaller than those within ic 694 .
the weakest and strongest hcn 10 emission , relative to , is found in c1 , the location of the brightest h@xmath73 emission in arp 299 ( gehrz 1983 ) .
continuum emission at 3.4 @xmath74 m and 10@xmath74 m also peaks close to the c1 clump .
thus , c1 appears to be currently the most active starforming region within c. we suggest that the observed molecular line ratio gradients are the result of a temperature and/or density gradient across the overlap region .
the / 10 line ratio in c1 is considerably lower than in the nucleus of ic 694 , perhaps because c1 is an extranuclear starburst . unlike
, the emission peak is not connected with the nucleus of ngc 3690 ( b1 ) .
the radial excitation gradient is similar , therefore , to that of ic 694 , with the highest / line ratio in the central region .
the nucleus of ngc 3690 , with intense associated h@xmath73 emission , is a site of starburst activity ( gehrz 1983 ) .
it has been suggested that high / intensity ratios in mergers are caused by unusually high isotopic abundance ratios in molecular clouds with optically thick 10 lines .
an influx of very low metallicity gas from the outer disk of the galaxies is the proposed cause of such an extreme abundance ratio ( e.g. casoli 1992 ; henkel & mauersberger 1993 ) .
since the / intensity ratio is normal in the outskirts of ic 694 and ngc 3690 this scenario is unlikely to be the explanation for arp 299 .
instead , the measured line ratio variations most likely indicate differences in the line excitation and gas properties in different parts of the system .
the observed / 10 line ratio is , however , a lower limit to the /abundance ratio in the emitting region and for @xmath61 ( ) @xmath75 this implies an abundance ratio not much greater than 60 in the center of ic 694 .
this value is typical for gmcs in the galactic disk , but higher than in the inner region of our galaxy , where the ratio is @xmath76 ( langer & penzias 1990 ) .
perhaps the ism in the nucleus of ic 694 recently arrived from the disk of the galaxy . in this case , the difference in line ratio between a2 and a1 is solely caused by a dramatic change in mean optical depth of the line . on the other hand , selective photodissociation of by a starburst and/or an agn may change the isotopic abundance ratio . a young nuclear starburst may also produce extra @xmath77c and thus temporarily increase the @xmath77c/@xmath78c abundance ratio ( e.g. , henkel & mauersberger 1993 ) .
the / 10 line ratio varies dramatically within arp 299 , from @xmath4 at the nucleus of ic 694 to 5 - 10 in its disk and in the eastern and north interface regions ( c3 and d ) .
the 10 brightness , relative to , is high in quiescent regions of low surface brightness , and low in starburst regions and galactic nuclei .
in contrast , hcn 10 is bright in the two galaxy centers and in the active extranuclear star formation region .
10 is @xmath7 at the nucleus of ic 694 , @xmath8 for ngc 3690 and @xmath9 for the extranuclear starburst region c1 .
unusually high / line ratios ( @xmath11 20 ) appear to be associated with extremely compact molecular distributions in the nuclei of merging galaxies ( cf .
aalto 1995 ) . the large / 10 intensity ratio at the nucleus of ic 694 can be attributed to low to moderate optical depth ( @xmath10 ) in the 10 line , possibly combined with abundance effects . a medium consisting of dense ( @xmath12 @xmath13 ) , warm ( @xmath79 50 k )
gas is consistent with the observations .
we thank peter bryant for helpful discussions and suggestions .
the ovro mm array is supported in part by nsf grant ast 9314079 and the k.t . and e.l .
norris foundation .
the national radio astronomy observatory is a facility of the national science foundation operated under cooperative agreement by associated universities , inc .
lccccrr ic694 : nucleus ( a1 ) & @xmath80 & 97 & 122 & 350 & @xmath81 & @xmath82ic694 : disk ( a2 ) & @xmath83 & 12 & 120 & 65 & @xmath5 & @xmath84ngc3690 : nucleus ( b1 ) & @xmath85 & 29 & 36 & 260 & & ngc3690 : total & & & 53 & & @xmath86 & @xmath87overlap east ( c3 ) & @xmath88 & 9 & 15 & 60 & @xmath5 & @xmath89overlap west ( c2 ) & @xmath90 & 20 & 40 & 80 & & overlap west ( c1 ) & @xmath91 & 21 & 47 & 60 & @xmath6 & @xmath92region d & @xmath93 & @xmath93 & 4 & 80 & @xmath94 & @xmath95region f & @xmath93 & @xmath93 & 11 & 40 - 80 & & | high resolution observations of ( @xmath0 ) , ( @xmath1 ) , and hcn ( @xmath2 ) @xmath3=10 in the galaxy merger arp 299 ( ic 694 and ngc 3690 ) show the line ratios vary dramatically across the system .
the / ratio is unusually large , @xmath4 , at the ic 694 nucleus , where emission is very strong , and much smaller , @xmath5 , in the southern extended disk of that galaxy .
elsewhere , the / line ratio is 5 - 20 , typical of spiral galaxies .
the line ratio variation in the overlap between the two galaxies is smaller , ranging from @xmath5 in the east to @xmath6 in the west .
the /hcn line ratio also varies across arp 299 , although to a lesser degree .
hcn emission is bright towards each galaxy nucleus and in the extranuclear region of active star formation ; it was not detected in the ic 694 disk , or the eastern part of the overlap region , leading to lower limits of 25 and 20 respectively . by contrast , at the nuclei of ic 694 and ngc 3690 the ratios are @xmath7 and @xmath8 respectively . in the western part of the overlap region
it is @xmath9 .
the large / 10 intensity ratio at the nucleus of ic 694 can primarily be attributed to a low to moderate optical depth ( @xmath10 ) in the 10 line .
these data support the hypothesis that unusually high / line ratios ( @xmath11 20 ) are associated with extremely compact molecular distributions in the nuclei of merging galaxies .
relative to , 10 is brightest in quiescent regions of low surface brightness and weakest in starburst regions and the galactic nuclei . a medium consisting of dense ( @xmath12 @xmath13 ) and warm ( @xmath14 k )
gas will reproduce the extreme line ratios observed in the nucleus of ic 694 , where the area filling factor must be at least 20% .
-1s^-1 # 1*#1 * # 1#22.0pt # 1#2_#1^#2 # 1_-#1^+#1 # 1#2_-#1^+#1_-#2^+#2 # 1 ' '' '' # 1 ' '' '' # 1 # 1 |
in the past few years , intensive experimental efforts have taken place at major laboratories around the world to study the baryon spectra using the electromagnetic probes . on the theoretical side ,
since the baryon resonances are unstable and couple strongly to baryon - meson continuum states , coupled - channel models are needed to analyze the experimental data and to interpret the extracted resonance parameters . for our present study
, we employ the well - known jlich @xmath0 model @xcite that is based on time - ordered perturbation theory .
it is a coupled - channel meson - exchange model which includes the @xmath0 and @xmath1 stable channels as well as the @xmath2 , @xmath3 and @xmath4 effective channels accounting for the resonant part of the @xmath10 channel .
the interaction kernel corresponding to the @xmath11- and @xmath12-channel diagrams is constructed based on the chiral lagrangians of wess and zumino @xcite , supplemented by additional terms for the coupling of @xmath13 , @xmath14 , @xmath15 , @xmath16 and @xmath17 @xcite . for the @xmath18-channel diagrams ,
apart from the bare nucleon dressed by the coupling to the @xmath0 continuum state to reproduce the physical nucleon , the interaction kernel includes eight genuine bare resonances , namely @xmath19 , @xmath20 , @xmath21 , @xmath22 , @xmath23 , @xmath24 , @xmath25 and @xmath26 .
the bare genuine resonances get their dressed masses and widths from the re - scattering of the baryon - meson continuum states ; the @xmath27 ( roper ) resonance appears as a dynamically generated resonance due to the strong coupling of the @xmath0 and @xmath3 channels .
this hadronic model has been quite successful in reproducing the @xmath0 partial - wave amplitudes up to the center - of - mass energy of @xmath28 gev and for the total angular momentum states with @xmath29 and @xmath30 . in ref .
@xcite a dynamical coupled - channel model for pseudoscalar meson photoproduction based on a field - theoretical approach of haberzettl @xcite has been introduced in conjunction with the jlich hadronic coupled - channel model .
this photoproduction model is distinguished from the majority of the existing dynamical models by the fact that , in addition to satisfying unitarity and analyticity as a matter of course , it also satisfies the full gauge - invariance condition dictated by the generalized ward - takahashi identity .
by contrast , the vast majority of existing dynamical models satisfy only current conservation but not gauge invariance . in ref .
@xcite , a preliminary feasibility study of this field - theory approach was presented . in the present work ,
we carry out an extended and more quantitative calculation of the neutral and charged pion photoproduction reactions within this approach using the jlich @xmath0 model @xcite .
we calculate cross sections as well as the beam asymmetries up to @xmath31 gev .
the full single - meson photoproduction amplitude , @xmath32 , is obtained by attaching a photon everywhere into the dressed @xmath33 vertex , which gives @xcite @xmath34 where the first three terms describe the coupling of the photon to the external legs of the dressed @xmath33 vertex , with the subscripts @xmath18 , @xmath12 and @xmath11 referring to the appropriate mandelstam variables of their respective intermediate hadrons .
the last term @xmath35 is interaction current , where the photon couples inside the @xmath33 vertex . in our work
it is chosen as @xcite @xmath36 , \label{eq : mint2}\ ] ] where @xmath37 denotes the transverse part of @xmath38 ; @xmath39 stands for an undetermined transverse contact current taken to be zero in the present work for simplicity ; @xmath40 is the non - polar part of the hadronic scattering amplitude ; @xmath41 describes the free propagation of the intermediate meson - baryon two - body system .
@xmath42 is the generalized contact current as specified in ref .
@xcite ; it accounts for the complicated part of the interaction current which can not be treated explicitly . the amplitude @xmath43 in eq .
( 1 ) is given by @xmath44 where @xmath45 stands for the dressed @xmath46 vertex with @xmath47 denoting either the nucleon or baryon resonance , @xmath48 denotes the dressed baryon propagator .
the dressed hadronic vertices and propagators are determined by the hadron dynamics of the jlich @xmath0 coupled - channel model .
@xmath49 is the full dressed @xmath50 vertex as derived in ref .
differential cross sections for @xmath51 , @xmath52 and @xmath53 .
the solid curves show the results from the full calculation .
the dotted and the dashed curves are obtained by respectively switching off the loop integral or the contact current @xmath42 .
data are from ref . @xcite . ]
photon spin asymmetries for @xmath54 , @xmath52 and @xmath53 .
the solid curves show the results from the full calculation .
the dotted and the dashed curves are obtained by respectively switching off the loop integral or the contact current @xmath42 .
data are from ref . @xcite . ]
the parameters of the present pion photoproduction model are determined by fitting the differential cross section and the beam asymmetry data of the charged and neutral pion photoproduction up to @xmath31 gev .
they are listed in ref .
the calculated differential cross sections and beam asymmetries for @xmath55 , @xmath52 and @xmath56 are shown in figs .
[ fig : dsdo ] and [ fig : sa ] , respectively , together with the corresponding experimental data .
more results for differential cross sections and beam asymmetries at 13 additional energies from the @xmath0 threshold up to @xmath31 gev can be found in ref .
figure [ fig : sig ] shows the total cross sections for the three charged pion photoproduction channels .
note that the total cross - section data have not been included in our fit . in figs .
[ fig : dsdo]-[fig : sig ] , the solid curves correspond to the results of our full calculation , while the dotted and dashed curves are obtained by respectively switching off the loop integral and the contact current @xmath42 in the full amplitude .
one sees that our theoretical results for both the differential cross sections and beam asymmetries ( figs .
[ fig : dsdo ] and [ fig : sa ] ) as well as the total cross sections ( fig . [ fig : sig ] ) are in good agreement with the corresponding experimental data .
juli - daz _ et al .
_ @xcite have also performed an investigation of the single - pion photoproduction within the ebac dynamical coupled - channel model .
compared to our model , they included two additional resonances , namely @xmath57 and @xmath58 ; however , the @xmath59 channel was not analyzed in their work . compared with the results of ref .
@xcite , the agreement of our results with the data is a little bit better for both the @xmath60 and @xmath52 channels .
figures [ fig : dsdo]-[fig : sig ] also show clearly that the loop integral and the contact current @xmath42 are important for reproducing the data , which means that the coupled - channel effects _ and _ gauge invariance are essential for understanding the single - pion photoproduction reactions . while the importance of coupled - channel effects was pointed out already in ref .
@xcite , we emphasize here that the majority of existing dynamical models of photoproduction reactions are actually not gauge invariant since their amplitudes obey current conservation only .
the latter is shown in ref .
@xcite to be necessary but not sufficient for maintaining full gauge invariance across all levels of a microscopic description of the photoreaction .
the importance of fulfilling the gauge - invariance condition was also emphasized for the @xmath61 bremsstrahlung reaction @xcite .
total cross sections for @xmath62 , @xmath52 and @xmath53 .
the solid curves show the results from the full calculation . the dotted and
the dashed curves are obtained by respectively switching off the loop integral or the contact current @xmath42 .
data are taken from ref .
@xcite but not included in the fit.,scaledwidth=44.0% ] finally , we mention that the residues and pole positions of the full photoproduction amplitude are being evaluated in our model by making an analytic continuation of the amplitudes to complex energy plane .
the results will be published elsewhere .
this work is supported by the ffe grant no .
41788390 ( cosy-058 ) .
the authors acknowledge the research computing center at the university of georgia and the jlich supercomputing center at forschungszentrum jlich for providing computing resources that have contributed to the research results reported within this proceeding .
99 c. schtz , j. haidenbauer , j. speth , and j.w .
durso , _ phys .
c _ * 57 * , 1464 ( 1998 ) .
o. krehl , c. hanhart , s. krewald , and j. speth , _ phys .
c _ * 62 * , 025207 ( 2000 ) .
a.m. gasparyan , j. haidenbauer , c. hanhart , and j. speth , _ phys .
c _ * 68 * , 045207 ( 2003 ) .
j. wess and b. zumino , _ phys .
rev . _ * 163 * , 1727 ( 1967 ) .
h. haberzettl , k. nakayama , and s. krewald , _ phys .
c _ * 74 * , 045202 ( 2006 ) .
h. haberzettl , _ phys .
c _ * 56 * , 2041 ( 1997 ) .
f. huang , m. dring , h. haberzettl , s. krewald , and k. nakayama , in preparation .
cns data analysis center , the george washington university , http://gwdac.phys.gwu.edu .
b. juli - daz , t .- s.h .
lee , a. matsuyama , t. sato , and l.c .
smith , _ phys .
c _ * 77 * , 045205 ( 2008 ) .
k. nakayama and h. haberzettl , _ phys .
c _ * 80 * , 051001 ( 2009 ) . | pion photoproduction reactions are investigated in a dynamical coupled - channel approach based on the jlich @xmath0 model , which presently includes the hadronic @xmath0 and @xmath1 stable channels as well as the @xmath2 , @xmath3 and @xmath4 effective channels .
this model has been quite successful in the description of @xmath5 scattering for center - of - mass energies up to @xmath6 gev .
the full pion photoproduction amplitude is constructed to satisfy the generalized ward - takahashi identity and hence , it is fully gauge invariant .
the calculated differential cross sections and photon spin asymmetries up to 1.65 gev center - of - mass energy for the reactions @xmath7 , @xmath8 and @xmath9 are in good agreement with the experimental data .
address = department of physics and astronomy , university of georgia , athens , ga 30602 , usa address = institut fr kernphysik and jlich center for hadron physics , forschungszentrum jlich , 52425 jlich , germany address = center for nuclear studies , department of physics , the george washington university , washington , dc 20052 , usa , altaddress = institut fr kernphysik and jlich center for hadron physics , forschungszentrum jlich , 52425 jlich , germany address = institut fr kernphysik and jlich center for hadron physics , forschungszentrum jlich , 52425 jlich , germany , altaddress = institute for advanced simulations , forschungszentrum jlich , 52425 jlich , germany address = department of physics and astronomy , university of georgia , athens , ga 30602 , usa , altaddress = institut fr kernphysik and jlich center for hadron physics , forschungszentrum jlich , 52425 jlich , germany |
the reaction @xmath0 is an interesting one from the theoretical viewpoint . because the coupling of the higgs to electrons is proportional to the electron mass , the tree graphs are so small that the one loop contribution gives the dominant part of the crosssection .
this has been recognized in the past @xcite and it has been found that the crosssections are rather small . only for special higgs particles in for instance susy models can the crosssection be larger , mainly due to a bigger tree graph contribution . in principle
this would finish the subject if it were not for the fact that during a project in which we were testing methods we recalculated this reaction and found both papers to have a number of shortcomings
. correction of these altered the final numbers significantly .
of course this does nt mean that now this reaction will become important at a future ilc , but at least reports about the variety of reactions that may or may not be of interest can quote the correct numbers . and if there exist extra higgs particles the errors of the past can be avoided in the future .
the major problem with one loop calculations is that they involve much work .
hence we like to use symmetry arguments as much as possible to minimize the number of diagrams to compute .
a traditionally safe symmetry to use is either bose or fermi statistics when there are identical particles involved .
because there are two higgs particles involved , both previous papers have used this symmetry , but both got the relative sign of the diagrams wrong .
this is because they used an indirect argument that for identical particles the amplitude should be maximal at the region of overlap and the scattering at 90 degrees was seen as the region of overlap .
they adjusted the relative sign for this .
we will show in the next section that this is not the case and that actually at 90 degrees the amplitude is zero .
this diminishes the crosssection considerably .
we have verified this result with the use of the feynarts @xcite system and it agrees with our findings .
this letter is arranged as follows .
we start with explaining the relative sign of the graphs in the formulas , using the tree graphs as an example .
then we show the diagrams to compute and how the sign of the tree graphs transfers to the loop graphs .
because the calculation itself can be done in many ways and can even be done automatically with systems like feynarts , we do nt go into unnecessary details .
finally we present the new numbers and show a few distributions .
the lowest order of the reaction @xmath0 has in principle three diagrams , which we will label the t - channel , the u - channel and the s - channel graphs .
( 310,60)(0,0 ) ( 20,50)(50,50 ) ( 50,50)(50,10 ) ( 50,10)(20,10 ) ( 50,50)(80,50)3 ( 50,10)(80,10)3 ( 50,50)1.5 ( 50,10)1.5 ( 13,50)[]@xmath4 ( 13,10)[]@xmath5 ( 89,50)[]@xmath6 ( 89,10)[]@xmath7 ( 45,30)[r]@xmath8 ( 100,30)[]@xmath9 ( 120,50)(150,50 ) ( 150,50)(150,10 ) ( 150,10)(120,10 ) ( 150,50)(180,10)3 ( 150,10)(180,50)3 ( 150,50)1.5 ( 150,10)1.5 ( 113,50)[]@xmath4 ( 113,10)[]@xmath5 ( 189,50)[]@xmath6 ( 189,10)[]@xmath7 ( 145,30)[r]@xmath10 ( 200,30)[]@xmath9 ( 220,50)(240,30 ) ( 240,30)(220,10 ) ( 240,30)(270,30)3 ( 270,30)(290,10)3 ( 270,30)(290,50)3 ( 240,30)1.5 ( 270,30)1.5 ( 213,50)[]@xmath4 ( 213,10)[]@xmath5 ( 299,50)[]@xmath6 ( 299,10)[]@xmath7 ( 255,25)[t]@xmath11 + fig .
1 : the three tree graphs for the reaction @xmath12 because of the very small coupling of the higgs to the electron these diagrams are suppressed by powers of @xmath13 which is a number that is less than @xmath14 .
the t - channel and u - channel graphs have two of such powers and the s - channel graph has only one .
let us however , for study purposes , look at the t - channel and u - channel graphs assuming that the coupling is a fixed constant c that does nt depend on the electron mass . in that case
the amplitude for the t - channel graph becomes : @xmath15 using the dirac equation in the limit that the electron mass goes to zero this gives @xmath16 for the u - channel diagram we have to exchange @xmath6 and @xmath7 and hence also @xmath8 and @xmath10 which gives @xmath17 in which we replaced @xmath7 by @xmath18 and the @xmath4 and @xmath5 can be removed with the dirac equation .
it is this change in sign , which at first looks counter intuitive , that caused all the problems because this way the sum of the t - channel and u - channel graphs becomes proportional to @xmath19 : @xmath20 and the result is that at 90 degrees this part of the amplitude vanishes .
the complete matrix element is still invariant under the exchange of @xmath8 and @xmath10 as @xmath21 and @xmath22 are invariant under this exchange and all @xmath8 and @xmath10 dependence can be written in terms of these two variables .
their contribution to the full matrix element becomes : @xmath23 we leave out the s - channel graph because the equivalent one loop graphs are not important : they always vanish in the limit that the electron mass is put to zero .
let us see how this works out at the one loop level .
( 280,60)(0,0 ) ( 20,50)(50,50 ) ( 50,50)(50,10 ) ( 50,10)(20,10 ) ( 50,50)(90,50)34.5 ( 90,50)(90,10)34.5 ( 90,10)(50,10)34.5 ( 90,50)(120,50)3 ( 90,10)(120,10)3 ( 50,50)1.5 ( 50,10)1.5 ( 90,50)1.5 ( 90,10)1.5 ( 13,50)[]@xmath4 ( 13,10)[]@xmath5 ( 129,50)[]@xmath6 ( 129,10)[]@xmath7 ( 70,45)[t]@xmath24 ( 85,30)[r]@xmath24 ( 70,15)[b]@xmath24 ( 160,50)(190,50 ) ( 190,50)(190,10 ) ( 190,10)(160,10 ) ( 190,50)(230,50)34.5 ( 230,50)(230,10)3 ( 230,10)(190,10)34.5 ( 230,50)(260,50)3 ( 230,10)(260,10)3 ( 190,50)1.5 ( 190,10)1.5 ( 230,50)1.5 ( 230,10)1.5 ( 153,50)[]@xmath4 ( 153,10)[]@xmath5 ( 269,50)[]@xmath6 ( 269,10)[]@xmath7 ( 210,45)[t]@xmath24 ( 227,30)[r]@xmath25 ( 210,15)[b]@xmath24 + fig .
2 : some one loop graphs in the reaction @xmath12 there are two types of diagrams : the ones with the vector bosons and the ones with the ghosts . as it turns out , the ones with the vector bosons are the easier ones as they involve only one power of the loop momentum in the numerator . due to the dirac equation and the fact that the loop momentum gives us , after integration , only lorenz tensors that involve the external momenta only the box diagrams are nonzero in the limit that the electron mass becomes zero .
if the terms in a diagram have only an odd number of gamma matrices ( @xmath26 counts as an even number for these purposes ) , eventually the diagram becomes of the type @xmath27 and when exchanging @xmath28 and @xmath29 we get the same minus sign as before .
this is different for an even number of gamma matrices where we have a relative plus sign . in the case of the standard model
however we have only diagrams with an odd number of gamma matrices when we put the electron mass equal to zero . while doing the calculation we ran into another problem with reference @xcite .
their figures show a rather strange and unphysical behaviour when the higgs mass is varied .
we could track this down to the exchange of two indices in a tensor in their equation 8 .
once this is corrected we get a much more physical behaviour although we still do nt get exactly the same result .
there are however not enough details to investigate this further .
we do nt understand the statement in reference @xcite that they agree with the previous paper as they seem to have only a problem with the relative sign of the diagrams .
for the rest the calculation was rather standard .
we evaluated the standard model higgs reaction first using form @xcite to reduce to scalar loop integrals according to the methods in ref @xcite .
then we did the scalar loop integrals by standard techniques .
eventually we also used feynarts as an extra check as one should take disagreement with two papers not lightly .
we would also have used the grace @xcite system as a check , but unfortunately the version we had was nt ready for it yet and for the version that might have handled it we did nt have the proper supporting programs .
because in the old calculations the amplitude reached a maximum at 90 degrees and the new calculation gives zero there , it should come as no surprise that the total crosssection is now significantly smaller .
we give some numbers : in the table we show a number of crosssections for different energies and masses .
this can be used as a reference to test programs . for the coupling constants we used the running values .
this is not automatic in the feynarts system . in figure 4
we show the crosssections as a function of the cm energy for a 120 gev higgs mass .
we see a slowly declining crosssection when we are away from the threshold .
this must clearly be a loop effect , as the tree graphs would show a slow ( logarithmic ) rise . in figure 5
we see the differential distribution in terms of the cos of the cm azimuthal angle . as mentioned before
, this distribution should be zero at 90 degrees when @xmath30 .
the immediate consequence of this is that if the energy is much above threshold , one should look for these events in a different region of phase space than the one that the previous papers would indicate .
we have also redone some of the mssm calculations of ref @xcite . for this
we used solely the feynarts system @xcite . in this case
the situation is potentially more complicated , but the diagrams with even numbers of gamma matrices between the spinors are much smaller again and hence the same sign problem makes that the crosssections becomes correspondingly smaller than in the original calculation .
this means that the ratio between the signals from the standard model and the mssm model does nt change significantly .
the first author acknowledges financial support from the comunidad de madrid through a fpi contract and the helpful advice of t. hahn regarding the feynarts system .
the work of second author was supported by fom and the humboldt foundation .
we would like to thank desy zeuthen for its hospitality during part of this work .
gaemers and f. hoogeveen , z. phys .
c26 ( 1984 ) 249 a.djouadi , v. driesen and c. junger ; phys.rev d54 ( 1996 ) 759 .
t. hahn , comput .
140 ( 2001 ) 418 j.a.m .
vermaseren , math - ph/0010025 g.j .
van oldenborgh and j.a.m .
vermaseren , z. phys .
c46 ( 1990 ) 425 g. belanger et al .
, phys.rept .
430 ( 2006 ) 117 - 209 t. hahn and c. schappacher , comput .
commun . 143
( 2002 ) 54 | we notice that the existing literature about the reaction @xmath0 suffers from a mistake in the relative sign between the t - channel and u - channel graphs . correcting this
lowers the crosssections by about an order of magnitude . * the reaction @xmath1 recomputed . *
+ j.j .
lopez - villarejo@xmath2 , j.a.m .
vermaseren@xmath3 + [ 3 mm ] * departamento de fsica terica , c - xi , + universidad autnoma de madrid , + canto blanco , + e-28034 madrid , spain . *
nikhef , p.o .
box 41882 , + 1009 db , amsterdam , the netherlands |
the accretion which built up the supermassive black holes ( smbhs ) now hosted in many local galaxies is widely thought to be associated with a sequence of output episodes observed as active galactic nuclei ( agns ) . since only a minority ( @xmath12 , see richstone et al .
1998 ) of local galaxies host a currently active agn , the corresponding lifetimes are estimated to be close to @xmath13 yrs .
several observations indicate the accretion episodes to be fundamentally related to the galaxy growth ; one such indication is provided by the narrow scatter in the observed correlation of the smbh mass @xmath14 with its host galaxy mass @xmath15 , when the former is in the range @xmath16 ( ferrarese & merritt 2000 , gebhardt et al .
the emissions of agns thus may conceivably constitute a probe for the history of accretion and growth of smbhs , and for its interplay with the galaxy building process .
the co - evolution of galaxies and agns and their so called `` downsizing '' ( faster evolution for more luminous objects ) depends also on feedback between nuclear and other galactic activities .
in fact , the density of the high luminosity qsos is peaked at high redshift and declines strongly toward us ; similarly , massive galaxies are characterized by a star formation history peaked at high redshifts .
luminous agns are efficient in `` sterilizing '' their massive host galaxies by heating the interstellar matter through winds , shocks , and high energy radiation , see granato et al .
( 2004 ) ; murray , quataert , thompson ( 2005 ) ; hopkins et al .
( 2006 ) ; bower et al .
( 2006 ) ; menci et al .
intriguingly , the latter authors found that the bimodal color distribution of galaxies at z@xmath17 can only be explained if agn feedback is considered . in this picture
an agn phase precedes the phase when a galaxy is caught in a passive state with red optical - uv colors , most of the star - formation having been inhibited by the agn activity . indeed , pozzi et al .
( 2007 ) using spitzer photometry found that that a sample of optically obscured qso at z=12 are mainly hosted by red passive galaxies , suggesting a later stage in their evolution . on the other hand , at low redshift many weak agns have been found in star - forming galaxies ( salim et al .
2007 ) . in these cases
feedback from less powerful agns ( the so - called radio mode ) is probably acting to self - regulate accretion and star - formation , and cold gas is left available for both processes for a much longer time ( croton et al .
the same cold gas can intercept the line of sight to the nucleus .
indeed , compton - thin absorbers ( with column densities @xmath18 @xmath7 ) may well be located in the galactic disk ( malkan , gorjian , tam 1998 ; matt 2000 ; see also ballantyne , everett , murray 2006 ) .
therefore a natural expectation in this scenario is the fraction of obscured agns to be large at low agn luminosities .
it is well known since the pioneering work done with the _ einstein _ satellite ( lawrence & elvis 1982 ) and with optically and radio - selected agns ( lawrence 1991 ) that this fraction strongly decreases with increasing agn luminosity ( see ueda et al .
2003 ; la franca et al . 2005 ; gilli , comastri , hasinger 2007 ; triester , krolik , dullemond 2008 ; hasinger 2008 ) .
a widely shared view holds that the luminosity dependence of the obscured fraction is related to the energy fed back by the agns onto the surrounding gas that constitutes the interstellar medium ( ism ) .
given that the agn emission is proportional to the fraction of such a gas available for accretion , a _ positive _ correlation between luminosity and absorption would be expected instead in the absence of an energy feedback depleting the ism after the onset of the agn activity .
a mounting body of observations cogently indicates that strong nuclear feedback is present in galaxies hosting agns ( see for a review elvis 2006 and references therein ) . on small ( sub - pc ) scales ,
the observed x - ray absorption lines indicate the presence of outward winds with velocities up to some @xmath19 km / s ( weymann 1981 ; turnshek et al .
1988 ; creenshaw et al . 2003 ;
chartas et al . 2002 ; pounds et al .
2003 , 2006 ; risaliti et al .
these likely originate from the acceleration of disk outflows due to the agn radiation field ( proga 2007 and references therein ) . on larger scales , broad absorption lines in about 10% of optically luminous qsos indicate fast outflows ( up to 30,000 km / s ) .
massive ( 10 - 50 @xmath20 ) flows of neutral gas with speed @xmath21 km / s are observed through 21-cm absorption of radio - loud agns ( see morganti , tadhunter , oosterloo 2005 ) , indicating that agns have a major effect on the circumnuclear gas in the central kiloparsec region around agns . on even larger scales of some @xmath22 kpc ,
the presence of agn - induced outflows is revealed by x - ray observations of the intra - cluster medium ( see mcnamara & nulsen 2007 for a review ) showing cavities and expanding shocks with mach numbers ranging from @xmath23 to @xmath24
. how the outflows produced in the innermost regions of the active galaxies are transported outwards to affect such large scales is still matter of investigation ; buoyant bubbles ( see reynolds , heinz & begelman 2001 , churazov 2001 ) and expanding blast waves ( see cavaliere , lapi & menci 2002 ; lapi , cavaliere & menci 2005 ) constitute viable mechanisms for such a transport .
nuclear obscuration is directly linked to agn feedback , since the same gas ( and dust ) which feed the agn output may well be responsible for its obscuration .
therefore , modelling agn obscuration is essential to connect the observed agn properties to the accretion history of smbhs over cosmological time .
this is a significant theoretical challenge as it requires not only connecting the agn evolution to the galaxy formation and growth , but also implementing in a model a self - consistent description of the agn feedback on the galactic gas .
indeed , very few attempts have been carried out in this direction so far .
an attempt to include agn absorption into an _ ab initio _ galaxy formation model has been proposed by nulsen & fabian ( 2000 ) , by relating both the smbh fueling and the agn absorption to the cooling flows associated with the hot gas pervading the growing dark matter haloes , but this assumption did not lead to a full description of the statistical distribution of qso luminosity . here we develop our semi - analytic model of hierachical galaxy formation and agn evolution ( see menci 2006 ) to self - consistently include the absorption of agns , with the aim of investigating the cosmic evolution of the latter and its dependence on agn properties like luminosity and redshifts .
the model is suited to our scope as it includes a detailed treatment of the feedback on the interstellar gas .
our aim is to investigate the dependence of agn _ absorption _ on luminosity @xmath25 and redshift @xmath26 arising in hierarchical galaxy formation scenarios that include an effective description of the evolution of agns and of their feedback .
the plan of the paper is as follows : in sect .
2 we describe our model for galaxy formation and the associated agn evolution .
3 is focussed on describing our treatment of the agn feedback onto the interstellar gas and its effects on the agn absorption . in sect .
4 we test our feedback - inclusive model for agn evolution by comparing its outcomes with the redshift distribution of the number density of agns for different x - ray luminosities . our results on the luminosity and redshift dependence of the absorbed fraction of agns are shown and discussed in sect .
6 is devoted to summarize our conclusions .
the semi - analytic model we develop and use connects , within a cosmological framework , the accretion onto smbhs and the ensuing agn activities with the evolution of galaxies .
galaxy formation and evolution is driven by the collapse and growth of dark matter ( dm ) haloes , which originate by gravitational instability of overdense regions in the primordial dm density field .
this is taken to be a random , gaussian density field with cold dark matter ( cdm ) power spectrum within the concordance cosmology " ( spergel et al .
2006 ) for which we adopt round parameters @xmath27 , @xmath28 , baryonic density @xmath29 and hubble constant ( in units of 100 km / s / mpc ) @xmath30 .
the normalization of the spectrum is taken to be @xmath31 in terms of the variance of the field smoothed over regions of 8 @xmath32 mpc . as cosmic time increases , larger and larger regions of the density field collapse , and
eventually lead to the formation of groups and clusters of galaxies ; previously formed , galactic size condensations are enclosed . in closer detail ,
the process implies not only smooth mass inflow , but also merging and coalescence of smaller condensations .
the corresponding merging rates of the dm haloes are provided by the extended press & schechter formalism ( see bond et al . 1991 ; lacey & cole 1993 ) .
the clumps included into larger dm haloes may survive as satellites , or merge to form larger galaxies due to binary aggregations , or coalesce into the central dominant galaxy due to dynamical friction ; these processes take place over timescales that grow longer over cosmic time , so the number of satellite galaxies increases as the dm host haloes grow from groups to clusters .
all the above processes are implemented in our model as described in detail in menci et al .
( 2005 , 2006 ) , based on canonical prescriptions of semianalytic modeling .
the radiative gas cooling , the ensuing star formation and the supernova events with the associated feedback occurring in the growing dm haloes ( with mass @xmath33 and circular velocity @xmath34 ) are described in our previous papers ( e.g. , menci et al .
2005 ) . the cooled gas with mass @xmath35 settles into a rotationally supported disk with radius @xmath36 ( typically ranging from @xmath37 to @xmath38 kpc ) , rotation velocity @xmath39 and dynamical time @xmath40 .
the gas gradually condenses into stars at a rate @xmath41 ; the stellar ensuing feedback returns part of the cooled gas to the hot gas phase with mass @xmath42 at the virial temperature of the halo .
an additional channel for star formation implemented in the model is provided by interaction - driven starbursts , triggered not only by merging but also by fly - by events between galaxies ; such a star formation mode provides an important contribution to the early formation of stars in massive galaxies , as described in detail in menci et al .
( 2004 , 2005 ) .
the model also includes a treatment of smbhs growing at the centre of galaxies by interaction - triggered inflow of cold gas , following the physical model proposed by cavaliere & vittorini ( 2000 ) and implemented in menci et al .
the accretion of cold gas is triggered by galaxy encounters ( both of fly - by and of merging kind ) , which destabilizes part of the cold gas available by inducing loss af angular momentum .
in fact , small scale ( 0.1-a few @xmath43 ) regions are likely to have disk geometry if they are to efficiently remove angular momentum and convey to the ( unresolved ) pc scales the gas provided on larger scales ( these may be isotropized by head - on , major merging events ) .
the rate of such interactions is given by menci et al .
( 2003 ) in the form @xmath44 here @xmath45 is the number density of galaxies in the same halo and @xmath46 is their relative velocity .
encounters effective for angular momentum transfer require i ) the interaction time to be comparable with the internal dynamical times of the partner galaxies ( a resonance condition ) , ii ) the orbital specific energy of the partners not to exceed the sum of their specific internal gravitational energies .
the cross section @xmath47 for such encounters is given by saslaw ( 1985 ) in terms of the tidal radius @xmath48 associated to a galaxy with given circular velocity @xmath34 ( see menci et al . 2003 , 2004 ) .
the fraction of cold gas accreted by the bh in an interaction event is computed in terms the variation @xmath49 of the specific angular momentum @xmath50 of the gas to read ( menci et al .
2003 ) @xmath51 here @xmath52 is the impact parameter , evaluated as the average distance of the galaxies in the halo . also , @xmath53 is the mass of the partner galaxy in the interaction , and the average runs over the probability of finding such a galaxy in the same halo where the galaxy with mass @xmath33 is located .
the values of the quantities involved in the average are computed from the semi - analytic model recalled in sect .
2.1 , and yield values of @xmath54 . for minor merging events and for the encounters among galaxies with very unequal mass ratios
@xmath55 , dominating the statistics in all hierarchical models of galaxy formation , the accreted fraction takes values @xmath56 .
the average amount of cold gas accreted during an accretion episode is thus @xmath57 , and the duration of an accretion episode , i.e. , the timescale for the qso or agn to shine , is assumed to be the crossing time @xmath58 for the destabilized cold gas component .
the time - averaged bolometric luminosity so produced by a qso hosted in a given galaxy is then provided by @xmath59 we adopt an energy - conversion efficiency @xmath60 ( see yu & tremaine 2002 ) , and derive the x - ray luminosities @xmath61 in the 2 - 10 kev band from the bolometric corrections given in marconi et al .
the smbh mass @xmath62 grows mainly through accretion episodes as described above , besides coalescence with other smbhs during galaxy merging . as initial condition
, we assume small seed bhs of mass @xmath63 ( madau & rees 2001 ) to be initially present in all galaxy progenitors ; our results are insensitive to the specific value as long as it is smaller than some @xmath64 . in our monte carlo model , at each time step
we assign to a galaxy the interaction probability corresponding to the rate given in eq .
according to it , we assign to the galaxy an active smbh accretion event of duration @xmath65 .
then we compute the accreted cold gas and the associated agn emission through equations ( 2 ) and ( 3 ) .
the model for agn evolution presented here relative to previous implementation by menci et al .
( 2003 - 2004 ) is complemented with the feedback from agns onto the surrounding ism .
this affects : a ) the available cold gas left over in the galactic disk after each agn event , that determines the subsequent gas accretion history ; b ) the density distribution of the galactic gas during each agn event , that determines the agn absorption . as mentioned in the introduction , fast winds with velocity up to @xmath66
are observed in the central regions of agns ; they likely originate from the acceleration of disk outflows by the agn radiation field ( see begelman 2003 for a review ) , and affect the environment in the host galaxy and beyond , leaving imprints out to large scales of some @xmath67 kpc in the intra - cluster medium ( icm ) . a detailed model for the transport of energy from the inner , outflow region to the large scales has been developed by cavaliere , lapi & menci ( 2002 ) , and lapi , cavaliere & menci ( 2005 ) .
central , highly supersonic outflows compress the gas into a blast wave terminated by a leading shock front , which moves outwards with a lower but still supersonic speed and sweeps out the surrounding medium .
eventually , this is expelled from the galaxy ; in the case of powerful shocks , it is expelled even from a group or poor cluster hosting the galaxy .
the key quantity determining all shock properties is the total energy @xmath68 injected by agns into the surrounding gas .
this is computed as @xmath69 ( see sect .
2.2 ) for each smbh accretion episode in our monte carlo simulation ; the value of the energy feedback efficiency for coupling with the surrounding gas is taken as @xmath70 , consistent with the values required to match the x - ray properties of the icm in clusters of galaxies ( see cavaliere , lapi & menci 2002 ) .
this is also consistent with the observations of wind speeds up to @xmath71 in the central regions , that yield @xmath72 by momentum conservation between photons and particles ( see chartas 2002 , pounds 2003 ) ; this value has been also adopted in a number of simulations ( e.g. , di matteo , springel & hernquist 2005 ) and semi - analytic models of galaxy formation ( e.g. , menci et al .
2006 ) .
the blast expands into the ism or the icm as described by the hydrodynamical equations ; these are to include the effects not only of initial density gradient , but also those of upstream pressure and dm gravity , clearly important quantities in galaxies , as discussed by lapi , cavaliere , menci ( 2005 ) ; the solutions show in detail how the perturbed gas is confined to an expanding shell bounded by an outer shock at the radius @xmath73 which sweeps out the gas surrounding the agn .
the overall effects have been emulated in the simulations by di matteo et al .
( 2005 ) , performed for specific cases of major mergers .
our treatment covers also the less energetic but more frequent events , with accretion rates and agn energy inputs @xmath68 provided by eqs .
( 1 ) to ( 4 ) . on the other hand ,
our model assumes spherical symmetry ; note however that this approximatively holds in the central regions ( the ones mainly affecting our results , as we shall show below ) due the exponential decline of the gas density .
in addition , both the effects discussed in points 1 ) and 2 ) of the appendix ( see below sect .
3.2 ) make the average absorption even less dependent on the disk outside the central region . as for the outflow geometry
, our model also applies to segments of spherical shells expanding outwards , originated by winds driven by a spherically symmetric agn radiation field ; these segments may not cover uniformly the central source , as in the conical distribution of elvis et al .
( 2000 ) . while its is true that random line - of - sights may not intercept some of these segments , the random orientation of the internal bh accretion disk with respect to the outer , galactic disk ( gallimore , baum , odea 1997 ; nagar & wilson 1999 ; thean et al .
2001 ) makes our results realistic on average .
finally , we do not consider clumpy outflows ; however , these are not expected to affect the dependence of the column density distribution on the agn luminosity and redshift , though they could broaden the column density distributions that we derive from our model .
lapi et al .
( 2005 ) showed in detail how the hydrodynamical equations for the finite amplitude gas perturbation constituting the blast wave , supplemented with the rankine - hugoniot boundary conditions at the shock , still admit self - similar solutions ( see sedov 1959 ) when gravity , initial gradients and pressure are included .
these solutions at the distance @xmath74 from the center are expressed in terms of @xmath75 , where @xmath73 is the shock position after a time @xmath76 from an agn outburst ; this is given by the eq .
( c7 ) in lapi et al .
( 2005 ) , that we recast in the form @xmath77^{1/\omega } { \mathcal m}^{2/\omega}\big[{t\over t_d}\big]^{2/\omega},\ ] ] where @xmath78 is the mach number @xmath79 . here
the initial gas density has a radial profile given by a power law @xmath80 with the exponent in the realistic range @xmath81 .
while self - similar , analytical solutions for the shock expansion law can only be found for spherical ( or in general 1d ) symmetry and power law gradients , we shall apply eq .
( 5 ) also to blasts expanding in the inner regions of a galactic disk , where the actual density profile follows an exponential law @xmath82 ( @xmath83 is the disk thickness and @xmath84 is the central surface density ) , with a cutoff in the vertical direction corresponding to the thickness of the disk . to use the above self - similar blast we construct piecewise power - law approximations of the above exponential density by subdividing the radial disk coordinate into a sequence of shells ; within each shell the exponential decline is approximated with a power - law with a different exponent @xmath85 .
the expansion of the blast in each shell will be computed on using eq .
( 5 ) with the appropriate value of @xmath85 .
self - similarity imply a definite time behavior for the agn energy injection which is determined by the exponent @xmath85 defining the density decline , normalized to reach its total value @xmath68 by the end of the accretion episode . in turn
, this implies a well - defined form for the time decay of the agn luminosity , as shown in the appendix .
however , in the self - similar solutions the ratio @xmath86 between the energy injected up to a time @xmath76 and the energy of the ism within the shock radius @xmath73 is shown to be independent of time and position ( see appendix ) , and thus constitutes the basic quantity marking the strength of the shock .
in fact . the mach number is shown to be simply related to @xmath86 which we compute at the final time of the agn active phase . in the following
, we shall take @xmath87 as derived by lapi et al .
( 2005 ) . in terms of the rescaled variable @xmath75 ,
the previous authors derived the density distribution ; this is found to be confined to a shell close to @xmath88 between the outer shock and an inner `` piston '' at a position @xmath89 ( @xmath90 ) with a width @xmath91 weakly dependent on the mach number . inside the shell little gas
is left over to absorb the agn radiation .
the amount of unshocked gas absorbing the agn emission will be given , for a given line - of sight , by the time - dependent fraction still unperturbed outside the shock located at @xmath92 ( see fig .
1 ) . the faster the shock , the lower will be the fraction of still unperturbed gas outside @xmath73 after a time @xmath76 from an agn outburst . the dependence of @xmath73 in eq .
( 5 ) on the agn luminosity will then affect the absorption by the unperturbed gas ; in particular , for the case @xmath93 ( corresponding to the standard isothermal sphere ) the relation @xmath94 holds , yielding _ lower _ absorbing gas fraction outside @xmath73 for _ brighter _ agns at a fixed time @xmath76 . an additional contribution to the absorption will be given by the gas compressed by the shock in the shell of width @xmath91 ( see lapi , cavaliere & menci 2005 ) , where @xmath95 can be also expressed in terms of the relevant ratio @xmath86 ; however , we shall show this second contribution to depend only weakly on the agn luminosity . [
0.73 ] fig .
1 . - a schematic representation of the effect of the blast wave induce by agn feedback on the density distribution of the interstellar gas .
the shock radius @xmath73 expands outwards , compressing the swept gas into a thin shell ( represented in darker colour ) with width @xmath96 , and leaving a cavity inside .
such a density distribution @xmath97 is also plotted at the bottom .
a line - of - sight ( at an angle @xmath98 with respect to the plane of the galactic disk ) is also represented in the scheme , along with the length @xmath99 of the line of sight intercepting the unperturbed gas outward of the shock front .
we implement the above model for agn feedback and for absorption in our sam as follows .
0 ) for each galaxy in our monte carlo simulations , with disk exponential scale length @xmath36 and total disk mass @xmath35 computed as described in 2.1 , we assume for the density distribution of the unperturbed gas the simplified form @xmath100 ( where @xmath74 is the distance from the centre of the galaxy ) with a cutoff in the vertical direction ( perpendicular to the disk ) at @xmath101 corresponding to the disk thickness . the value of the vertical scale height @xmath83 for the _ gaseous _ disk is taken to be @xmath102 ( see narayan & jog 2002 ) , corresponding to @xmath103 pc for a typical @xmath104 galaxy .
with such a low value of @xmath83 , our assumed density for the gaseous disk closely follows a constant vertical density distribution with a radial exponential profile @xmath105 , @xmath106 being the distance from the centre in the plane of the disk ; the deviations are @xmath107 in the central region , and @xmath108 at @xmath109 .
these are both smaller than the present uncertainty on the vertical distribution of the density of the gaseous disks ; on the other hand , the above schematic form for the gas density distribution allows us to apply a simple anaytical description for the evolution of the blast .
the density distribution is normalized as to recover the total gas mass @xmath35 when integrated over the disk volume .
1 ) we construct a piecewise power - law approximation of the above exponential density distribution , by subdividing the radial coordinate into shells ; in each shell we approximate the exponential shape @xmath110 with a power - law @xmath111 where @xmath112 , where @xmath113 , @xmath114 , and @xmath85 are parameters that we adjust to optimize the fit in each radial shell .
we find that , on retaining @xmath115 at all radii , the following set of radial shells and parameters provides a good fit to the exponential law : @xmath116 , @xmath93 for @xmath117 , @xmath118 , @xmath119 for @xmath120 , @xmath116 , @xmath121 for @xmath122 ; here the minimal radius considered is @xmath123 pc . with the above choice , the r.m.s . deviation in the fits
are @xmath124 within the inner region ( @xmath125 , the dominant range for the agn abosrption ) , and @xmath126 for @xmath127 ( a region irrelevant to our computation of the agn absorbtion ) .
the temperature of the gas in the disk is assumed to be @xmath128 k. 2 ) when a smbh starts its accretion phase triggered by a close interaction of the host galaxy with a companion , we compute the corresponding agn total energy injection @xmath68 . for the sake of simplicity , we shall label each agn with a constant luminosity @xmath25 after eq .
( 3 ) , although a full consistency with the self - similar solutions adopted for the blastwave expansion would require a time - decay of the luminosity related to the exponent @xmath85 .
however , we show in the appendix that assuming a constant @xmath25 yields results not distinguishable from those obtained by assigning the agns the exact , time - dependent luminosity .
3 ) during the time interval @xmath65 we follow the expansion of the shock given by eq . ( 5 ) for each radial shell defined under 1 ) , on using the appropriate value of @xmath85 given by the fitting procedure described above .
we compute the amount of gas still unperturbed outwards of the shock position @xmath73 at a time @xmath76 within the interval @xmath65 ; the unshocked gas distribution is given by the density profile @xmath97 defined under ( 1 ) , with @xmath129 ( see fig .
4 ) we extract a random line - of - sight angle @xmath98 defining the disk inclination to the observer ; at a time @xmath76 within the interval @xmath65 corresponding to the active agn , we compute the column density corresponding to the gas outside the shock position along the selected line - of - sight as @xmath130 5 ) if for the chosen line - of - sight the shock position is within the disk , we add the absorption column density corresponding to the gas compressed by the shock ( see fig .
lapi et al .
( 2005 ) showed that the density of the gas swept by the blast and compressed to a shell of width @xmath91 rises toward the center as a function @xmath131 diverging weakly ( but still integrating to a limited mass ) at the piston position @xmath89 . on adopting the behaviour @xmath132^{(\omega-6)/3(7-\omega)}$ ] ( strictly valid on approaching the piston position ) , and integrating over @xmath74 within the shocked shell , the corresponding column density is @xmath133^{15 - 5\omega\over 3(7 - 2\omega)}\,{n_2\over n_1}\,n_1\,r_s(t)~.\ ] ] here @xmath134 is the density jump at the shock , and @xmath135 is the unperturbed gas density just outside the shock front .
both the density jump @xmath134 and the @xmath95 grow slowly with increasing mach number @xmath136 ( in turn a function of @xmath86 ) , as given in lapi et al .
( 2005 ) ; both quantities saturate to constant values ( @xmath137 and @xmath138 ) in the limit of very strong shocks ( @xmath139 ) . in sum
, the blastwave has two effects : on the one hand , it depletes the internal regions thus decreasing the column density of the absorber ( as computed at point 4 above ) ; on the other hand , it increases the density inside a thin layer close to the piston ( an effect computed at point 5 above ) .
the first effect however dominates , since the gas initially distributed in the denser central region is spread out by the blast into an expanding spherical shell ; thus , the integration along the line of sight intercepts only a tiny fraction of the gas initially concentrated in the core of the distribution .
in addition , the minor contribution to the column density from eq .
( 7 ) represents an upper limit , since the gas temperature in the blast wave is increased to values ( given in eq .
c8 of lapi , cavaliere , menci 2005 ) which generally exceed the ionization temperature of the gas and thus limit the absorption . at the end of the agn accretion episode ( i.e. , at @xmath140 ) only a fraction of the initial cold gas mass @xmath35 will be left in the galaxy , as discussed in detail in cavaliere et al .
( 2002 ) and lapi et al .
note that , the brighter is the agn , the larger is the injected energy @xmath68 , the faster is the shock espansion @xmath73 ( see eq .
5 ) , the lower will be the resulting column density at a given time after the start of the outburst ( see eq . 6 ) .
thus the evolution of the gas distribution and hence of the gas absorption during the agn event is best described in terms of the key ratio @xmath86 which determines all the relevant quantities of the blast : the expansion of the shock front @xmath73 , the thickness of the shocked shell @xmath141 , and the density jump at the shock .
such a ratio is computed in the model for each galaxy when an active accretion episode is triggered by a close encounter .
the effects of the above detailed model for the agn feedback on the evolution of the cold galactic gas and on the galactic absorption of the agn emission are given next .
here we present our results concerning the evolution of the agn number density and the obscuration by galactic interstellar gas . we do not provide a description of absorption due to the clouds within the central regions of agns , so we can not disentangle compton - thick sources ( where the obscuration is presumably provided by gas in regions @xmath142 @xmath143 ) from the absorbed population we obtain from our model .
indeed , the shock may originate at any point within our resolution scale ( 50 @xmath143 ) , and it may lie inside or outside the compton - thick region .
therefore we can not make any strong statement about the dependency of compton - thick absorption on the outflow and hence on the agn luminosity . before addressing the problem of the luminosity and redshift dependence of agn absorption , we test the basic predictions of our model with recent data . to this aim , we first compare in fig .
2 the predicted evolution of the cosmic density of agn with different _ intrinsic _ luminosity @xmath25 with data from la franca et al . ( 2005 ) . note that the model reproduces the marked _ downsizing _ of agns , i.e. , the faster redshift evolution of brighter agns compared to the fainter sources ( a behaviour already obtained in semi - analytic models , see menci et al .
2004 ; fontanot et al .
2006 ) . in the model ,
such an effect is related to the faster consumption of gas in the progenitors of massive galaxies at high redshifts @xmath144 ; in fact , these are formed in biased high regions of the cosmological density field , where the early cooling of galactic gas provides early reservoirs for bh accretion , and the galaxy merging and interactions are enhanced .
when such progenitors assemble to form a massive galaxy ( and correspondingly a bright agn when the bh is actively accretiing ) at @xmath145 , the available galactic gas has largely been converted into stars or accreted onto the bh at higher redshifts .
conversely , low - mass galaxies retain a large fraction of their gas down to low redshifts , and this results into prolonged star formation ( yielding blue colors as observed ) allowing for effective bh accretion down to @xmath146 . [ 0.5 ] fig .
2 . - the evolution of the cosmic density of agns with x - ray luminosities ( in the @xmath147 kev band ) in 5 different luminosity bins : @xmath148 log @xmath25/erg s@xmath149 ( top , blak line ) , @xmath150 log @xmath25/erg s@xmath151 ( red line ) , @xmath152 log @xmath25/erg s@xmath153 ( green line ) , @xmath154 log @xmath25/erg s@xmath155 ( blue line ) , log @xmath25/erg s@xmath156 ( bottom cyan line ) . note that the the plotted absorbed fraction includes the compton - thick sources .
note that at @xmath2 the model predictions lie above the observed points by a factor @xmath1 for low luminosity agns with @xmath157 erg / s .
this , at least in part , is because the model predictions in fig .
2 concern all agns , regardless of whether they are compton - thin or compton - thick , since we can not disentangle the two populations in our model , while the data we compare with include only compton - thin sources .
a complementary check would be constituted by the x - ray background at its 30 kev peak ( where observations are not affected by copton - thick obscuration ) .
however this does not actually constitute a strong constraint for the model , since it is mainly contributed by objects at @xmath158 ( gilli , comastri , hasinger 2007 ) whose observed density is well reproduced by the model .
we have repetedly checked since menci et al .
( 2003 ; 2004 ) that the basic features of our model are consistent with the constraint set by the observed x - ray background ( see barcons , mateos , ceballos , 2000 ) and that the predicted global history of bh assembly is consistent with the local observed smbh mass function ( marconi et al .
2004 ) .
thus the comparison between model predictions and observation strongly suggests the existence of a relevant fraction of compton - thick sources at @xmath159 ( as found by martinez - sansigre et al .
2005 , 2007 ; fiore et al .
2008a , b ) .
note however that a complementary possibility , which could partly account for the model overestimate of low - luminosity sources at @xmath2 , is that the bh growth in small - mass galaxy halos ( dm masses @xmath160 ) be inhibited by some process not included in the model ; we shall comment further this point in the conclusions .
finally , we stress that the model predicts at @xmath161 a density of very luminous - objects ( @xmath162 erg / s ) similar to that at @xmath163 .
these are the x - ray counterparts of the luminous @xmath164 optical qsos observed up to @xmath165 ( see fan et al .
2004 ; richards et al .
2006 ) and imply that fast building up of massive bh at early epochs is possible in hierarchical scenarios ( as long as a constant mass - energy conversion factor @xmath166 holds , see discussion in the final section ) . in turn
, this is a consequence of the speed up of structure growth ( enhanced merging rate ) for haloes collapsed in biased , high - density regions of the primordial density field , which constitutes a natural feature of hierarchical scenarios .
we now turn to investigating the predictions of our model concerning the luminosity and the redshift dependence of the absorption due to galactic gas . in fig .
3 we show the predicted distribution of the column densities of the absorbers for low redshift ( @xmath8 ) agns with different luminosity , and compare it with data from la franca et al .
( 2005 ) , and with expectations based on agn synthesis models for the cosmic x - ray background ( gilli et al . 2007 ) .
the rapid expansion of the blast wave produced by bright agns yields a low fraction of absorbed ( @xmath167 @xmath7 ) , bright ( @xmath168 erg / s ) agns , while at lower luminosity the @xmath5 distribution shifts to larger values .
note that the above relation between the luminosity and the obscuring column density holds only after averaging over _ time _ after the start of the agn active phase and over the _ orientations _ of the galactic disk , the two fundamental parameters that define the absorption properties of any single agn in our model .
the probability for a line - of - sight to intercept a given amount of galactic gas is independent of the agn luminosity ( although models predicting such a dependence have been proposed , see lamastra , perola & matt 2006 ) , while the probability to observe an agn at a time when the blast wave has depleted the gas content of the galaxy is larger for luminous agns , since these produce faster blast waves .
[ 0.6 ] fig .
3 . - left panel : the distribution of column densities @xmath5 at @xmath8 for agn with different luminosities : 42@xmath169 log @xmath61/erg s@xmath170 43 ( green dotted line ) , 43@xmath169 log @xmath61/erg s@xmath170 44 ( red dashed line ) , log @xmath61/erg s@xmath171 44 ( violet solid line ) .
the lowest @xmath5 bin actually includes the contribution from all absorbing column densities @xmath172 @xmath7 .
right panel : the predicted fraction of absorbed agns with @xmath173 @xmath7 is plotted as a function of their x - ray luminosity for @xmath174 ( solid line ) , and compared with data from la franca et al .
( 2005 , solid squares ) ; the latter have been corrected for the selection effects as described by the authors .
we plot also the expectations of the agn synthesis model for the cosmic x - ray background by gilli et al .
( 2007 , dots ) .
we also show as a dashed line the obscured fraction as a function of agn luminosity when no agn feedback is considered ; note that this has been computed with the same choice of disk parameters and therefore it is normalized as to yield the same fraction of obscured agns at low luminosities where the feedback is not effective .
we stress that the inverse correlation between the fraction of obscured objects and luminosity constitutes a key _
test _ for agn feedback as a source of the luminosity - dependence of agn obscuration ; indeed , when such a feedback is not included , an opposite correlation is found ( as shown in fig .
3 ) since more luminous agns result from a larger fraction of cold gas available for both accretion and obscuration . note that qualitatively similar results are obtained for a range of agn feedback efficiencies @xmath175 .
values outside this range would primarily violate constraints set by galaxy colors , the @xmath176 correlation , and the specific star formation rate of massive galaxies at high redshifts .
we expect the overall absorption to evolve strongly with redshift , as shown in fig . 4 where we plot as a function of @xmath26 the fraction of agn ( with any luminosity ) absorbed by gas with column density @xmath167 @xmath7 .
such a fraction of absorbed agns rises appreciably with redshift , in agreement with the observational findings by la franca et al .
although there is still some disagreement among observers on whether the agn obscuration is purely luminosity dependent or both luminosity and redshift dependent ( see gilli et al . 2007 and references therein ) ,
the latter case constitutes a generic expectation of models in which obscuration is generated on host galaxy scales .
this is due to the larger fraction of cool galactic gas available for both bh accretion and agn absorption at high redshifts ; in turn , this is due to the rapid gas cooling corresponding to the higher densities of the collapsed structures hosting the agns , and to the frequent galaxy merging events which allow to continuously replenish the cold gas to be turned into stars at high redshift .
note that in our model the fraction of obscured objects with @xmath6 @xmath7 plotted in fig .
4 includes compton - thick sources , since it does not resolve the inner regions of agns .
these are the regions where the compton - thick absorbers are presumably located according to several observational indications , from the local abundance of compton - thick sources ( favoring large covering factors and consequently compact sizes , see risaliti , maiolino , salvati 1999 ) , to the direct limits from chandra observations of the circinus galaxy ( see guainazzi et al .
2001 and references therein ) to the rapid oscillation timescale of the absorber in ngc1365 ( risaliti et al .
2005a ) . in the fig .
4 , we tentatively estimate the contribution of compton - thin sources ( dashed line ) to the obscured fraction by assuming that the difference between the predicted and the observed density of agns shown in fig . 2 is entirely due to compton - thick sources . finally , in fig .
4 we give our predictions concerning the redshift evolution of the absorbed fraction of agns with different luminosities , which are shown by the thiner lines .
note how for @xmath2 the luminosity dependence of the absorbed fractions weakens for faint agns with @xmath177 erg / s , while it exhibits a sharp transition to low values @xmath178 for brighter agns .
[ 0.45 ] fig .
4 . - the evolution of the predicted fraction of absorbed ( @xmath6 @xmath179 ) agns for all sources with 43@xmath169 log @xmath61/erg s@xmath170 46 ( solid heavy line ) . the data from la franca et al . ( 2005 ) have been corrected for the selection effects .
the heavy dotted line represents the contribution to the fraction of absorbed agns due to only compton - thick sources estimated assuming that the difference between the observed and the predicted agn density in fig .
2 is entirely due to the compton - thick agns .
the thin solid lines correspond to the absorbed fraction of agns within luminosity bins of width 0.25 dex centered on the following values : @xmath61/erg s@xmath180 43.25 ( green , dashed line ) , @xmath61/erg s@xmath180 44.25 ( blue , long dashed line ) , @xmath61/erg s@xmath180 44.7 ( cyan , dot dashed line ) .
such a _ combined _ luminosity and redshift dependence of agn absorption , which constitutes a _ specific prediction _ of our model , is shown in fig . 5 for a continuous , wide range of luminosities and redshift .
as redshift increases , the overall fraction of absorbed agns increases , but with a luminosity dependence much weaker than holding at low @xmath26 ( represented in fig .
2 ) for luminosities @xmath177 erg / s . in fact
, the quantity of absorbing gas is predicted to be large enough ( due to the processes discussed above ) as to make the expanding blast wave induced by the agn feedback ineffective to expell a major fraction of the galactic gas , which is continuously replenished in the galactic potential wells . for higher luminosities at @xmath2
a sharp transition occurs between a highly obscured population with @xmath181 erg / s and a nearly unobscured population with @xmath182 erg / s .
thus , a testable prediction of our model is that for intermediate luminosities ( @xmath183 ) the ratio of optical to x - ray luminosity functions at redshifts @xmath2 should be much _ lower _ than that at z=0 , while for higher luminosities it remains _ close _ to the local value . [ 0.8 ] fig .
- the combined luminosity and redshift dependence of the absorbed fraction of agns , represented by the color code shown on top .
we have computed the effects of the galactic absorption on agn emission in a cosmological context , by including a physical model for agn fueling and feedback into a semi - analytic model of galaxy formation in the concordance cosmology .
the model is based on galaxy interactions as triggers for agn accretion and on expanding blast waves as a mechanism to propagate outwards the agn energy injected into the interstellar medium at the center of galaxies .
we have shown that an _ inverse _ dependence of agn absorption on luminosity ( fig .
3 ) and a _ direct _ dependence on redshift ( fig .
4 ) is a natural outcome in such a context .
the former arises from the faster expansion of blast waves induced by feedback of energetic , luminous agns onto the galactic interstellar gas ; the rapid sweeping of gas in the inner regions , where the density was initially higher , results in the fast formation of a gas - depleted region with size larger for higher agn intrinsic luminosities .
qualitatively similar conclusions were reached by hopkins et al .
( 2005b ) on the basis of dedicated hydrodynamical n - body simulations .
on the other hand , the redshift dependence of the agn absorption is due to the larger amount of cold galactic gas available at high redshift , when the higher densities allowed for fast cooling occurring in galactic haloes .
the quantitative _ predictions _ of our model are consistent with existing observations concerning the fraction of absorbed ( @xmath6 @xmath7 ) agns as a function of their luminosity and redshift .
our model specifically predicts that for agns with @xmath11 erg / s the luminosity dependence of the absorbed fraction weakens with increasing redshift ( see fig .
5 ) , while for the brightest objects with @xmath184 the absorbed fraction quickly decreases with luminosity for @xmath185 ( see fig . 5 ) . note that , after averaging over the line of sight , unobscured or mildly obscured agns correspond to late stages of the feedback action ; in particular , for a given orientation of the line of sight , the observed column density depends on the time elapsed since the start of the blast wave expansion .
the faster expansion characterizing the blast wave of luminous agns thus corresponds to a _ larger probability _ to observe them when the blast has already swept out the central regions of the galaxy igm .
this picture constitutes an _ extension _ of the unified picture for agns ( see antonucci 1993 ) beyond the canonical scheme based on the single orientation parameter , since the absorption properties now depend on the combination of _ orientation _ and _ time _ needed to sweep the central regions of the galaxy disk .
note that our picture has also straightforward implications on the connection between star formation and obscuration properties ( proposed , since from our results we on average expect larger star formation in heavily obscured objects , while luminous , mildly absorbed agns should be generally associated to galaxies with lesser star formation ( or in transition to a passive state ) .
the relevance of evolutionary effects of the kind modeled here in determining the absorption properties of agns is supported by several observational works .
stevens et al .
( 2005 ) and page et al .
( 2004 ) find that x - ray obscured qsos have much higher submillimeter detection rates than x - ray unobscured qsos , suggesting strong star formation on - going in the host of obscured agn only .
sajina et al .
( 2007 ) and martinez - sansigre et al .
( 2008 ) report spitzer irs spectra dominated by agn continuum but showing pahs features in emission , typical of starforming galaxies , in samples of ulirgs and radio selected obscured qsos at z@xmath186 .
lacy et al .
( 2007 ) find evidence for dust - obscured star formation in type-2 qsos . from the analysis of the x - ray background , baalntyne ey al .
( 2006 ) argue that the agn obscuration is connected with the star formation in the host galaxy .
finally , martinez - sansigre et al .
( 2005 , 2008 ) found little or no lyman-@xmath187 emission in a sample of z@xmath188 obscured qsos , suggesting large scale ( kpc ) dust distribution .
all these findings are in general agreement with the evolutionary picture , although some disagreeing works argue for a dominance of geometrical effect ( see triester et al .
a geometrical effect may be constituted by the gravitational bending of the interstellar gas due to the bh , as described in lamastra et al .
( 2006 ) ; however , except for very massive bhs ( @xmath189 ) , this will affect mainly regions below our resolution scale of 50 @xmath143 .
another class of geometrical models , based on the luminosity dependence of the sublimation radius of the bh accretion disk , is that commonly referred to as the receeding torus picture ( lawrence 1991 ) .
these however provide an appreciable luminosity dependence only at very high agn luminosities @xmath190 erg / s and for dust located close to the sublimation radius ( @xmath191 pc ) .
dust located in galaxy disk can hardly be affected directly by the agn radiation . in our model we did not try to model the processes leading to compton - thick absorption with @xmath192 @xmath7 , generally thought to be caused by gas directly associated with the central regions of the agns . on the other hand ,
our model provides a hint concerning luminosity and redshift dependence to be expected as for the abundance of compton - thick sources .
inspection of fig .
2 , where we compare the predicted global ( including compton - thick sources ) density of agns with different _ intrinsic _ luminosities with data corrected for absorption ( but not including compton - thick sources ) , shows that our model predicts a number of low / intermediate - luminosity agns ( @xmath3 erg / s@xmath193 ) larger than the observed compton - thin sources by a factor around 2 at @xmath2 .
while it is possible that our model overestimates the agn fueling in this range of @xmath61 and @xmath26 ( e.g. , due to its specific modeling of interaction - driven destabilization for cold gas in galactic disks ) , the above excess could support the view that at such luminosities and redshifts a fraction around 1/2 of the total agns are compton - thick . a complementary process which may explain our overprediction of low - luminosity agns at @xmath2 is suppression of bh growth in small mass galactic haloes ( dm masses @xmath160 ) .
this may be provided by gravitational - rocket effect on the bhs due to the recoil following the emission of gravitational waves during the coalescence of bh binaries following galaxy mergers ( see madau & quataert 2004 ) .
such a recoil may produce bh velocities of order @xmath67 km / s , sufficient to unbind the hole from galaxies with dm velocity dispersion @xmath194 km / s . in this respect , the observational selection of compton - thick sources by combining mid - infrared to near - infrared and optical photometry of galaxies ( fiore et al .
2008a , b ) will provide crucial constrains on the relative role of obscuration and bh depletion in low - mass galactic halos .
our results are also relevant for constraining the physical mechanisms of agn feedback onto the interstellar gas , a key issue in recent developments of cosmological galaxy formation models . indeed , the recent realizations of such models include agn feedback as the key process to suppress gas cooling in massive galactic haloes .
actually , two kinds of feedback are at present implemented in this context : on the one hand , models based on galaxy interactions as triggers of agn activity and feedback relate the latter to the bright , accretion phase onto supermassive black holes , i.e. , to the same phase corresponding to agn activity ( see menci et al . 2003 , 2007 ; di matteo et al . 2005 ; hopkins et al .
2005a ) ; on the other hand , other authors associate the agn feedback only to a quiescent ( so called radio ) phase of accretion , characterized by very low accretion rates ( @xmath195/yr ) and not observable as radiative agns ( bower et al . 2006 ,
croton et al .
since the latter mode would result into a feedback activity continuing down to low redshift , we expect such models to provide a much milder dependence of agn absorption on redshift ; so observational results on the high - redshift absorption of agn emission will constitute an effective test for models of agn feedback . finally , we note that the feedback related to the active agn phase described here is effective to decrease the galactic gas and the associated absorption mainly at low redshift @xmath145 and for bright ( @xmath196 erg / s ) agns ( see , e.g. , cavaliere & menci 2007 ) .
this implies that at high redshifts the effective cooling and the continuous replenishing of galactic gas due to fequent merging events will override the depletion due to the agn feedback , so the latter can not suppress the early growth of supermassive bh .
indeed , the predicted number density of early ( @xmath197 ) , bright ( @xmath198 erg / s ) agns shown in fig . 2 is consistent with that observed for bright ( @xmath199 ) optical qsos up to @xmath200 ( hopkins et al .
note however that such a result does not include the bh spin - up which may occur during the growth due to accretion of gas endowed with angular momentum ( for different modeling of such a process see , e.g. , volonteri et al .
2005 ; king , pringle & hofman 2007 ) , which in turn may yield larger radiative efficiencies up to values @xmath201 for a dominant fraction of bhs ; as noted by the above authors , the lower mass accretion rate related to larger radiation efficiencies may delay the mass assembly of massive bh at early epochs when included in a cosmological model .
we shall investigate these issues in a following paper .
we thank our referee for helpful comments which contributed to improve the paper .
we acknowledge grants from asi .
the key quantity determining the blastwave evolution is energy deposition ratio @xmath86 of the energy injected up to the time @xmath76 over to the ism energy out to the shock radius at the same time .
this is constant in self - similar solutions , which imply for the energy injection the law @xmath202 ( sect . 5 and appendix c2 of lapi , cavaliere , menci 2005 ) , normalized to yield the total value @xmath203 ( eq .
4 ) at the end of the agn active phase .
its time behavior is thus related to the exponent @xmath85 defining the decline of the gas density distribution . since the supersonic wave affects the plasma only out to the shock distance @xmath73 where the unperturbed ( initial ) energy content is @xmath204 ( lapi et al .
2005 ) , the ratio @xmath86 is constant , and constitutes the basic quantity determining the strength of the shock ; in other words , @xmath86 remains constant while the wave expands , since both the energy injected up to the considered time and the ism energy content within the shock radius grows in time with the same law .
thus we choose to compute @xmath86 as the total energy injected by the agn divided into the total energy content of the ism within the whole volume of the disk .
the luminosity @xmath25 is the time derivative of @xmath68 , and thus in the self - similar model is to behave as @xmath205 ; so its decay is related to the exponent @xmath85 defining the density decline ( see lapi , cavaliere , menci 2005 , sect.5 and appendix c2 ) . for our piecewise approximation of the density profile ( point 1 in sect .
3.2 ) , @xmath25 is constant during the time when the wave sweeps the inner region @xmath206 , declines as @xmath207 when the blast has expanded into the region @xmath208 , and then drops to zero when @xmath209 , where @xmath68 remains constant at the value attained at the border @xmath210 , consistently with the index @xmath211 holding in that region . full consistency with
the self - similar blastwave model would require us to assigne each agn a time - dependent luminosity , decaying as described above .
however , the results are indistingishable from those obtained with the simple prescription @xmath25=constant ( provided both luminosities are normalized to the same final total injection @xmath68 ) for two reasons .
\1 ) as for the central @xmath206 region @xmath25 is indeed constant , as we showed above ; in the outer region the self - similar behavior @xmath207 results in a much slower decay of the luminosity compared to the decline of the ism density encountered by the blast @xmath212 .
in fact , using eq .
( 5 ) for the time expansion of @xmath213 , the density in front of the blast declines as @xmath214 . clearly , the regime when the blast expands into the outer region @xmath215 is not important , since here the luminosity is null and the object is not recognized as an active agn in computing the @xmath5 distributions shown in sect .
\2 ) most important , the scale height of the disk @xmath216 is much shorter than the radius @xmath217 enclosing the first radial shell we use to fit the exponential profile of the disk .
this in turn means that nearly all the random line - of - sights that we draw from the galaxy center only intercept the first radial shell , where the luminosity is effectively constant according to the self - similar construction of the model .
quantitavely , the probability for a line - of - sight to intercepts the outer ( @xmath218 ) disk is @xmath219/(\pi/2)\approx 8\,10^{-2}$ ] . beyond the details of our specific self - similar blastwave model ,
the physical point is that the geometry of the disk is such that the vast majority of the line - of - sights is only affected by the inner region ; the latter is completely swept out by the blast early on , when the luminosity decay has not yet occurred . | we compute the effect of the galactic absorption on agn emission in a cosmological context by including a physical model for agn feeding and feedback in a semi - analytic model of galaxy formation .
this is based on galaxy interactions as triggers for agn accretion , and on expanding blast waves as a mechanism to propagate outwards the agn energy injected into the interstellar medium at the center of galaxies .
we first test our model against the observed number density of agns with different intrinsic luminosity as a function of redshift .
the model yields a downsizing behavior in close agreement with the observed one for @xmath0 . at higher redshifts ,
the model predicts an overall abundance of agns ( including compton - thick sources ) larger than the observed compton - thin sources by a factor @xmath1 for @xmath2 and @xmath3 erg / s .
thus , we expect that at such luminosities and redshifts about @xmath4 of the total agn population is contributed by compton - thick sources .
we then investigate the dependence of the absorbing column density @xmath5 associated to cold galactic gas ( and responsible for the compton - thin component of the overall obscuration ) on the agn luminosity and redshift .
we find that the absorbed fraction of agns with @xmath6 @xmath7 decreases with luminosity for @xmath8 ; in addition , the total ( integrated over luminosity ) absorbed fraction increases with redshift up to @xmath9 , and saturates to the value @xmath10 at higher redshifts .
finally , we predict the luminosity dependence of the absorbed fraction of agns with @xmath11 erg / s to weaken with increasing redshift .
we compare our results with recent observations , and discuss their implications in the context of cosmological models of galaxy formation . |
in this paper we continue a study initiated in @xcite concerning self - avoiding walk on the infinite percolation cluster obtained after performing supercritical percolation on @xmath2 .
this problem has been studied extensively in the physics litterature , using non - rigorous methods ( see @xcite for the first paper on the subject , and @xcite for later contributions ) , but its understanding from a mathematical point view remains very weak , the main reason being that self - avoiding walk on @xmath2 is , in many respects already a big challenge for mathematician ( see @xcite for a review ) . our starting point is to assume that some of the properties of the walk , such as the replica overlap ( i.e. the mean proportion of edges shared by two independent trajectories ) and the end - to - end distance can be deduced from the asymptotic behavior of the partition function @xmath0 , which is simply the number of open self - avoiding path of length @xmath1 from the origin .
this belief is somehow ascertained by both the physics literature ( e.g. see @xcite ) and analogies with rigorous results obtained for directed polymers in a random environment @xcite , but has not yet been brought on rigorous ground ( and we will not do it in this paper )
. hence our main object of study will be the growth rate @xmath9 .
in @xcite we have aproved the existence of a quenched connective constant for the percolation cluster , or in other words , that the upper growth - rate of @xmath0 is not a random variable .
we also proved that in two dimension , this upper - growth rate is always strictly smaller than the growth rate of the expected value , underlining a localization phenomenon .
our aim in this paper is to show that there is a phase where localization occur also in high dimension : that when @xmath6 is large and @xmath10 is close to the percolation threshold , @xmath0 does not grow as fast as its mean value .
our result apparently contradicts some of the prediction made in the litterature .
for instance in @xcite , @xcite and many pait is claimed that disorder never modifies the behavior of the trajectories above the percolation threshold .
the reason for this contradiction might be that some author may have considered the `` annealed model '' ( where averaging with respect to the percolation cluster is performed ) which is mathematically trivial . on the other - hand our result agrees with the predictions present in @xcite . furthermore according to le doussal and machta @xcite
, this localization phenomenon does not occur for small dilution , i.e. when @xmath10 is close to one .
this means that there is a non - trivial transition in @xmath10 separating a phase where trajectories are delocalized from one where the are delocalized . on the mathematical level , this remains a challenging conjecture .
we consider @xmath11 the set of self - avoiding path of length @xmath1 starting from the origin in @xmath2 which is equipped with its usual lattice structure .
the notation @xmath12 denotes the graph distance in @xmath2 .
@xmath13}\ | \
s_0=0,\ \forall n\in [ 0,n-1 ] , |s_{n}-s_{n+1}|=1 , \\
\forall i\ne j \in[0,n ] , s_i\ne s_j\}.\end{gathered}\ ] ] let @xmath14 be a field of iid bernoulli random variable of parameter @xmath10 indexed by the edges of @xmath2 .
we denote by @xmath15 or @xmath16 the associated law .
when @xmath17 we say that the edge is open .
we say that a lattice path in @xmath2 is open if all the edges that composes it are open .
we are interested in the asymptotic behavior of the number of self - avoiding open paths of length @xmath1 starting from the origin @xmath18 we can define an analogous quantity @xmath19 by counting open self - avoiding path starting from @xmath20 instead of the origin . of course @xmath19 is equal to zero for large @xmath1 if @xmath21 lies in a finite cluster of open edges . for the reason
, we focus only on the supercritical regime of percolation @xmath22 ( @xmath23 denoting the percolation threshold ) where a.s .
there is a unique infinite connected component of open edges ( see e.g. the first chapters of @xcite for an introduction to percolation ) .
we call @xmath24 the unique infinite percolation cluster . in @xcite
, we introduced the notion of connective constant for the dilute lattice . recall that @xmath25 the connective constant of @xmath2 is defined by @xmath26 here @xmath12
denote the cardinal of a set and we will keep this notation throughout the paper as it brings no confusion .
[ taex ] for @xmath22 , for every @xmath27 , the limit @xmath28 does not depend on @xmath21 and is @xmath16-a.s .
we call this limit the _ quenched _ connective constant of the dilute lattice and denote it by @xmath29 .
it satisfies the inequality @xmath30 where @xmath31 is the connective constant of @xmath2 .
we call @xmath32^{\frac1n}= p\mu_d(1).\ ] ] the _ annealed _ connective constant
. moreover , the ratio between quenched and annealed connective constant @xmath33 is a non - decreasing function of @xmath10 on @xmath34 $ ] .
in particular there exists @xmath35 $ ] such that @xmath36 the notion of connective constant for non - transitive graphs has also recently been studied in a non - random setup ; in @xcite a universal lower - bound is given for the for the connective constant of @xmath6-regular graphs . whether the inequality is strict or not is related , at least on a heuristic level to trajectorial properties of the self - avoiding walk on @xmath37 .
the self - avoiding walk of length @xmath1 on @xmath37 is the process defined on @xmath11 by the probability law @xmath38 the general idea is that if @xmath0 behaves asymptotically like its expected value , it means that the environment is spatially averaging so that the self - avoiding walk in the inhomogeneous medium keeps the features of self - avoiding walk on @xmath2 ( see e.g. @xcite for conjectures and mathematically proved results ) . on the other hand if @xmath39 $ ] , and _ a fortiori _ if the two quantities have a different behavior on the exponential scale ( _ i.e. _ if @xmath40 ) one should observe localization of the trajectories under @xmath41 ( see @xcite and references therein for the case directed polymers , and @xcite for physicists prediction ) .
an important issue is then to decide whether a genuine phase transition occurs at @xmath42 _
i.e. _ if @xmath43 .
in @xcite , we proved that in dimension @xmath44 , the inequality is always strict for all values of @xmath45 so that there is no phase transition ( @xmath46 ) .
le doussal and machta @xcite believe that this is the case also for @xmath47 , and we agree with this conjecture although it should be difficult to prove with available tools ( see the introduction of @xcite for more discussions ) . on the contrary , when @xmath48 , the same authors conjectured that for small edge dilution , the disorder is irrelevant : in other words that for @xmath10 close to one @xmath49 or that @xmath50 .
note that the conjectures in @xcite are not formulated for the connective constant but in terms of trajectorial properties . in this paper
we focus on the question : is @xmath7 in general ? we give a positive answer to this question in high dimension .
the strategy we adopt to prove such a result presents some similarities with the one we used in @xcite to solve an analogous question for the problem of oriented percolation : we prove an asymptotic lower bound on @xmath42 which is larger than the asymptotic development in @xmath51 of @xmath23 . the asymptotic development for the percolation threshold up to the third order has been computed by hara and slade @xcite using the lace expansion .
they proved that @xmath52 see also @xcite for a related result and a bibliography on this topic .
let us present now our main result [ mainres ] given @xmath53 ,
when @xmath6 is large enough and @xmath54 then @xmath55 as a consequence when @xmath6 is large enough , @xmath56 our method to obtain a lower bound on @xmath42 relies on three ideas : the first one is to consider the size biased measure where the probability of a given environment is proportional to @xmath57 .
this gives us a nice characterization of the strong disorder regime @xmath58 : see lemma [ sizebi ] .
the second idea is to use a special construction of the size biased measure sometimes referred to as the _ spine construction_. it says that the environment under the size - biased law can be obtained by opening a self avoiding path of length @xmath1 chosen uniformly at random ( the spine ) , and opening every other edges in @xmath2 independently with probability @xmath10 : see lemma [ spine ] finally , and this is the most important step , we use this spine construction to show that additions of edges around the spines can generate a lot of open paths , showing that the size - biased measure and the original one are very different when @xmath10 is small ( proposition [ mainprop ] ) .
all these three ideas were used in an earlier work concerning oriented percolation @xcite , and the way to proceed for the two first step is exactly identical . however ,
here the final step is much more involved and requires some heavy - machinery for several reasons : the possibility of interaction between the walk and its past , the modification of the length when adding `` bridges on the walk '' etc ... at first sight , after reading our proof , one could think that the asymptotic lower - bound we find for @xmath42 is larger than @xmath23 for accidental reasons and that it might not be satisfied for some other type of high dimensional lattice .
this is however not the case : the fundamental reason for the difference between the two is that the contribution of `` square of open edges '' ( i.e. of regions where four edges forming a square are open ) is more important under the size - biased measure than under the original one . in the high dimensional limit , the contribution of these open square gives the leading asymptotic term for @xmath59 .
thus our result can be expected to be true for percolation models in the mean - field limit ( see for instance @xcite where the same result holds for directed percolation essentially for the same reasons ) .
we believe that @xmath7 in all dimensions but we currently have no argument to justify it for finite @xmath6 .
note finally that we did not try to explicit what @xmath6 large enough in theorem [ mainres ] means .
the reason for that is that quantitative estimates would be quite hard to derive and in any case , very suboptimal .
the size biazing technique consist in studying the process under a new measure , which gives a larger probability to environments @xmath60 for which the partition function @xmath0 is larger .
we introduce @xmath61 the renormalized partition function @xmath62 note that the inequality is strict if and only if @xmath61 decays exponentially fast .
we define @xmath63 the size biased - law , which is absolutely continuous with respect to @xmath16 and whose derivative is @xmath64 the exponential decay of @xmath65 under @xmath16 corresponds more or less to its exponential growth under @xmath63 .
this the content of the following lemma .
[ sizebi ] if there is a constant @xmath66 satisfying @xmath67{{\;\leqslant\;}}e^{-cn},\ ] ] then @xmath68 .
the result and proof are similar to proposition 4.2 in @xcite , and we include the proof here just for the sake of completeness . the inequality @xmath68 is equivalent to the almost sure exponential decay of @xmath61 .
assume that holds .
then from the definition of @xmath63 we have @xmath69 = { \widetilde}{{\ensuremath{\mathbb e } } } \left[(w_n)^{-1}{\mathbf{1}}_{w_n\in [ e^{-cn/2},e^{cn}]}\right ] \\ { { \;\leqslant\;}}e^{c/2n}{\widetilde}{{\ensuremath{\mathbb p } } } [ w_n{{\;\leqslant\;}}e^{cn } ] { { \;\leqslant\;}}e^{-cn/2}.\end{gathered}\ ] ] the markov inequality implies that @xmath70{{\;\leqslant\;}}e^{-cn}$ ] , and hence by the borel - cantelli lemma , we have @xmath71 the size biased measure can be obtained from the original one by opening all the edges along a randomly chosen self avoiding path .
similar constructions have been used for quite a while for the study of branching structure ( see @xcite ) or directed polymers @xcite . given a path @xmath72 and an environment
@xmath60 we define the environment @xmath73 with spine @xmath74 by @xmath75 then set @xmath76 [ spine ] the law of @xmath77 under @xmath63 , is the same as the law of @xmath78 under @xmath79 .
see e.g. @xcite lemma 1 . the most important step , and whose proof will be the focus of all the rest of the paper is to show that as soon as @xmath80 , with large probability , a lot of paths are open .
the idea is to use the spine as a backbone to construct a lot of open paths with large probability .
this result combined with the two lemmata 2.1 and 2.2 easily yields theorem [ mainres ] . [ mainprop ] given @xmath53 , when @xmath6 is large enough , there exists @xmath81 such that for all @xmath80 , for all @xmath1 large enough @xmath82<e^{-c({\varepsilon},d)n}.\ ] ] according to lemma [ spine ] and lemma [ sizebi ] , the theorem reduces to prove that there exists @xmath66 such that @xmath83<e^{-cn}.\ ] ] from the proposition [ mainprop ] , and the fact that ( cf .
the definition of the connective constant ) @xmath84 equation is satisfied for @xmath1 large enough if @xmath85 we use the following asymptotic expansion in @xmath6 of @xmath86 from @xcite @xmath87 to get that , for @xmath6 large enough , the inequality is satisfied if @xmath88
our strategy is to have a lower bound is to look only at paths that uses either edges of the spine or make very short excursions out of it ( we will call these excursions _ bridges _ ) .
we first have to restrict ourselves to a set of good spines @xmath74 and show that most spines are good .
then we show that when a spine is good , with large probability , we have a lot of open bridges , which allows to have a lot of open paths .
= 14 cm let us introduce some notation .
we set @xmath89 to be the @xmath90-th increment of @xmath74 , @xmath91 to be the set of indices when the spine makes a u - turn , @xmath92\ | \ s_{n}\sim s_{n-3}\},\ ] ] and @xmath93 the set of indices around which @xmath74 does not bend @xmath94\ | \ x_n= x_{n-1}\}.\ ] ] let @xmath95 to be the set of nearest neighbors of the origin .
we have to consider three types of bridges ( figure [ threebridges ] ) : * the edges @xmath96 and @xmath97 are @xmath60 open for some @xmath98 .
we call this an @xmath99-bridge over @xmath100 . *
the edge @xmath101 is open for some @xmath102 , we call this a @xmath103-bridge linking @xmath100 to @xmath104 . *
the edges @xmath105 , @xmath106 and @xmath107 are open for some @xmath108 , we call this a @xmath109-bridge over the edge @xmath110 ( the restriction of the direction @xmath111 is there because we do not want our bridge to use edges of the spine ) . to a bridge
we can associate a square of open edges in @xmath112 which is composed of @xmath44 , @xmath113 , or @xmath114 edges of the spine and the rest forms the bridge ( see figure [ threebridges ] ) .
we call the vertices of that are at the end of two edges of the bridge the free sites ( there is one free site for an @xmath99-bridge and two for a @xmath109-bridge and none for a @xmath103-bridge ) .
= 14 cm it can be remarked that if we have @xmath115 bridges of type @xmath99 , @xmath116 bridges of type @xmath103 and @xmath117 bridges of type @xmath109 such that the corresponding squares do not share edges nor free sites like on figure [ superpath ] where @xmath118 and @xmath119 ( see on figure [ overlapbri ] examples of configurations where bridges do overlap ) , then it is possible to construct @xmath120 distinct self - avoiding paths of length @xmath1 that are open for @xmath112 .
this is because if one selects a subset of the @xmath99-bridges and subsets of the @xmath103 and @xmath109-bridges of the same cardinality one can form an open path for @xmath112 that uses the edges of the selected bridges and the edges of the spine elsewhere else like on figure [ superpath ] ( we have to use the same number of @xmath103 and @xmath109 because these type of bridges modify the total length of the path and we want the length to be equal to @xmath1 ) . obviously different choices for the set of bridges give a different path .
= 14 cm thus our work will be focused on proving that with an overwhelmingly large probability we can construct a lot of bridges with squares that overlap only on spine sites .
the typical value of the number of bridges of type @xmath99 , @xmath103 and @xmath109 at the first order is not difficult to derive at the heuristic level if we accept that in high dimension the spine behaves like a simple random walk : given @xmath121 ( and in large dimension the proportion of indices not in @xmath93 tend to one ) , the probability of a bridge of type @xmath99 being open is @xmath122 so that the typical number of bridges of type @xmath99 is @xmath123n$ ] . for bridges
type @xmath103 we notice that the probability of making a @xmath124-turn for the self - avoiding walk should be roughly the same as for the simple random walk where it is asymptotically equivalent to @xmath125 when @xmath6 is large .
hence we typically have @xmath126n$ ] and each for each @xmath90 in @xmath91 , we have a probability @xmath51 to have an open bridge , giving a heuristic number of @xmath103-bridges of @xmath123n$ ] . for @xmath109
, we notice that for each edge , there are about @xmath127 directions in which a bridge of type @xmath66 can be open . considering that each bridge is composed of three edges and has a probability @xmath128 of being open , we get a similar asymptotic for the number of @xmath109-bridges .
hence from one should have typically @xmath129 however , to make this rigorous , we have to show that on the complement of a set of exponentially small probability , we can construct large sets of bridges that do not overlap with each other .
thus we would have that holds with large probability .
the remainder of the paper is devoted to that purpose .
we can separate the proof of proposition [ mainprop ] in two steps .
the first step is to reduce to a set of good spines for our construction , on which were there are lots of spaces to construct bridges that do not overlap .
we have to show that the set of good spine has a large probability .
the second step is to show that , given a good spine , the probability of being able to construct a lot of non - overlapping bridges is large .
introducing these two results requires some notation .
set @xmath130 resp .
@xmath131 to be the set of time for which the spine never comes back at a distance one ( resp .
two ) after more than one ( resp .
two)step , in both directions .
@xmath132 \ | \forall m{{\;\leqslant\;}}n , |m - n|>1 \rightarrow |s_m -s_n|>1\},\\ v^2_n&:=\{n\in[0,n ] \
| \forall m{{\;\leqslant\;}}n , |m - n|>2 \rightarrow |s_m -s_n|>2\},\}. \end{split}\ ] ] note that when @xmath1 is larger than @xmath133 we have @xmath134 .
given @xmath53 , we define @xmath135 the set of good spines @xmath136\right\}.\ ] ] then , we divide proposition [ mainprop ] into two results .
[ gudespine ] for fixed @xmath137 , for @xmath6 large enough , there exists @xmath138 such that for all @xmath1 large enough @xmath139 [ bridgeconstructor ] for fixed @xmath137 , for @xmath6 large enough , there exists @xmath138 such that for @xmath140 , for @xmath1 is large enough , and @xmath141 then @xmath142 { { \;\leqslant\;}}e^{-c({\varepsilon},d)n}.\ ] ] the remainder of the paper is divided in two section , each of which is devoted to the proof of one of the above propositions .
by monotonicity it is sufficient to prove the result for @xmath143 . hence in this whole section @xmath144 .
the strategy is exactly the one exposed above : we want to show that with a large probability there are a lot of bridges of each type and that moreover they are located at the right places .
we start with the bridges of type @xmath103 .
we define @xmath145 as a subset of the @xmath90 such that there is an open bridge of type @xmath103 linking @xmath100 and @xmath104 , more precisely where the second condition has to be checked only if @xmath147 ( otherwise @xmath148 is not an edge in @xmath2 ) .
this condition
@xmath148 is closed for @xmath60 " is present so that all the bridges in our set are well separated to avoid a situation such as @xmath149 on figure [ overlapbri ] where the two consecutive open bridges can not be used simultaneously .
we show that when the spine is good , @xmath150 can not be much smaller than @xmath151 or more precisely given the spine , @xmath150 depends only on the state of the edges @xmath154 . furthermore changing the state of one edge changes @xmath150 by at most one .
thus @xmath150 is a @xmath114-lipshitz function of @xmath155 equipped with the hamming distance .
thus we can apply mcdiarmid s inequality ( see @xcite or @xcite ) and get that for all positive @xmath21 the result then follows by choosing @xmath157-(1 - 2{\varepsilon})n/(2d)^2.\ ] ] we just have to check that @xmath21 is larger than some constant times @xmath1 .
we do so by noticing that as @xmath158 $ ] ( from the assumption @xmath141 ) , and that for @xmath102 , @xmath159 is either equal to @xmath160 or @xmath51 and hence @xmath161{{\;\geqslant\;}}|u_n|(2d)^{-1}(1-(2d)^{-1}){{\;\geqslant\;}}(1 - 3{\varepsilon}/2)n/(2d)^2,\ ] ] if @xmath6 is large enough .
now , we deal with bridges of type @xmath99 . for reasons exposed earlier
we do not want our bridges to overlap with briges of type @xmath103 and thus we have to choose their location carefully .
we define first a set @xmath162 where we can build the bridges .
we call the condition @xmath164\cap u_n=\emptyset$ ] is there to avoid interaction of bridges of type @xmath99 over @xmath90 , with @xmath165 with bridges of type @xmath103 ( like case @xmath166 in figure [ overlapbri ] ) . given the spine , we claim that the @xmath173 are iid bernoulli of parameter @xmath122 . the only thing to check it that there is no over overlap between the different bridges over @xmath90 , @xmath165 . for an edge to be used by two distinct bridges of type @xmath103 , it must be for two bridges over @xmath100 and @xmath174 . for an overlap to occur
we need that the free sites of the two bridges to be the same i.e. @xmath175 , or equivalently @xmath176 .
this implies @xmath177 , and thus @xmath178 and hence @xmath179 .
for this reason , bridges over @xmath100 , @xmath165 are edge disjoint .
the result then follows by choosing @xmath183-(1 - 3{\varepsilon})n/(2d)^2.\ ] ] it remains to check that this @xmath21 is proportional to @xmath1 by giving a lower bound on @xmath184 $ ] . by independence of the @xmath167 , given the spine , for @xmath185 , @xmath186 $ ] is either @xmath187 if @xmath188 or @xmath122 if not .
hence for @xmath141 @xmath190\cap(v^2_n+1)\setminus\left(t_n\cup \left(\bigcup_{i=0}^3 u_n - i\right)\right)\right|\\ { { \;\geqslant\;}}n-1-(n-|v^n_2|)-|t_n|-3|u_n|{{\;\geqslant\;}}n(1 - 3{\varepsilon}/2),\end{gathered}\ ] ] for large enough @xmath6 .
hence when @xmath6 large enough @xmath189{{\;\geqslant\;}}n(1 - 2{\varepsilon})(2d)^{-2},\ ] ] which is enough to conclude .
finally we treat the case of bridges of type @xmath109 .
we do not want these bridges to overlap with eachother nor with bridges of type @xmath103 and @xmath99 and thus we must be careful about their location .
define as before a set @xmath191 where we allow construction of bridges .
| \{n , n+1\}\subset v^2_n,\ [ n+1,n+3]\cap u_n=\emptyset \}.\ ] ] the condition @xmath193 is present because we do not want our bridges to overlap with some other distant bridge ( of type @xmath99 or @xmath109 see @xmath194 or @xmath195 of figure [ overlapbri ] ) , it also guarantees that the bridges over @xmath196 does not use edges or sites on the spine .
the condition @xmath197\cap u_n=\emptyset$ ] is there to avoid interaction with bridges of type @xmath99 .
however a bridge over the edge @xmath196 and another bridge over the edge @xmath198 may overlap .
we want to avoid this and set our definition of the set of good bridges accordingly .
for @xmath90 set @xmath199 the event that there exists an open bridge over @xmath196 ( recall that @xmath200 is the set of nearest neighbor in @xmath2 ) @xmath201 note that the possibilities for the direction @xmath111 have been reduced to avoid overlap with bridges of type @xmath99 .
define then the set of good bridges ( recall ) for @xmath210 , @xmath211 $ ] is not straight - forward to compute as it depends on the local configuration of the spine but one can easily get a bound on it .
notice that by definition @xmath167 and @xmath212 are independent of eachother and of @xmath213 .
for @xmath214 and @xmath215 to be one simultaneously , there are two possibilities , either @xmath216 edges forming two joint bridges are open : there are at most @xmath217 possibilities for this to occur , each one of probability @xmath218 ; or two disjoint bridges are open using a total of @xmath133 edges , there are at most @xmath219 option for choosing the bridges and the probability of both being open is @xmath220 and hence using union bound for all these events we have given @xmath74 , set @xmath224 to be the set of edges whose state might have an effect on @xmath204 , that is @xmath225 \ } \cup \ { ( s_n+e , s_{n+1}+e ) , n\in [ 0,n-1]\}.\ ] ] as changing the state of one edge can only affect the state of finitely many bridges , @xmath226 is a lipchitz function of @xmath227 ( @xmath60 restricted to @xmath224 ) with lipchitz constant @xmath228 .
hence by mcdiarmid s concentration inequality the definition of the injection is rather straight - forward : to each @xmath237 is associates the paths that uses the edge of the spines everywhere except where the bridges in @xmath238 , @xmath239 , @xmath240 are located ( recall figure [ superpath ] ) . because of the condition @xmath241 the length of the obtained path is indeed @xmath1 .
what there is to prove is that this construction is indeed possible , _
i.e. _ that it is possible to use simultaneously all theses bridges and that the obtained path is self - avoiding ( see figure [ overlapbri ] for possible complications ) . recall that when a bridge is open , it corresponds to a square of open edges for @xmath112 and that free sites are sites of those square that do not belong to any of the square s spine edges ( there is one free site for each bridge of type @xmath242 and two free site for each bridges of type @xmath243 ) . to prove that our injection is well defined , we have to show is that two of these square never have an edge in common and that the free sites of a given square do not belong to another square nor to the spine .
the reader can check that these assumptions are guaranteed by our definitions of @xmath205 , @xmath145 and @xmath204 .
the fact that free sites do not belong to the spine is guaranteed by the fact that for @xmath90 in @xmath205 , @xmath244 , and for @xmath245 , @xmath246 .
we leave to the reader to check that there is no edge overlap thanks to our assumptions as this would be rather tedious to develop this point here .
now we combine the inequality with the results of the three lemmata : when @xmath141 , with probability larger than @xmath247 , we have @xmath248 , and hence implies that for some appropriate constant @xmath66 the overall strategy to prove the proposition is to prove the exponential decay ( in @xmath1 ) of some bad events for walks that are markovian and then prove that these markovian walks are a good enough approximation of the self - avoiding walk ( which is not markovian ) .
let us introduce notation for the markovian approximation recall that @xmath250 to be the law of the self - avoiding walk of size @xmath1 .
set @xmath251 to be the law of the non - backtracking walk , that is the walk for which at each step @xmath252 is chosen uniformly at random among the neighbors of @xmath100 except that the choice @xmath253 is not allowed .
set @xmath254 the uniform law on path of length @xmath1 with no backtrack and no loops of length @xmath255 i.e. on @xmath256}\ | \ s_0=0,\ \forall n\in [ 0,n-1 ] , |s_{n}-s_{n+1}|=1 , \\
\forall n \in[0,n ] , s_n\notin \{s_{n-2},s_{n-4}\}\}.\end{gathered}\ ] ] and @xmath257 the markovian nearest - neighbor walk that at each step , jumps to a random neighbor chosen uniformly at random in @xmath258 under @xmath251 , @xmath259}$ ] is distributed uniformly at random among trajectories with no backtrack .
however , under @xmath257 , it is not true that @xmath259}$ ] is distributed uniformly on @xmath260 . on the contrary (
this is easy to check ) , each trajectory @xmath259}$ ] has a probability proportional to ( recall ) @xmath261 we will use @xmath262 and @xmath254 as approximations of @xmath250 .
the number of non - backtracking paths of length @xmath1 is equal to @xmath263 .
from @xcite , ( see also @xcite ) the number of paths of length @xmath1 with no four loops @xmath264 satisfies @xmath265 where asymptotic is taken for @xmath266 and @xmath267 is a constant .
furthermore , the asymptotic development of @xmath268 in @xmath6 is given by @xmath269 thus , considering the asymptotic development of the connective constant @xmath86 , one gets that there exists a constant @xmath270 such that for @xmath6 large enough and @xmath271 @xmath272 } \text { is self avoiding } \right)&=\left(1-\frac{1}{(2d)^2}+o(d^{-3})\right)^n,\\ \pi^2_n\left((s_n)_{n\in[0,n ] } \text { is self avoiding } \right)&=\left(1+o(d^{-3})\right)^n .
\end{split}\ ] ] we want to show first that under @xmath250 the number of u - turns up to step @xmath1 is roughly @xmath273 ( like for the simple random walk ) , and that @xmath274 is at most twice of the same order .
we also show that @xmath275 is small ( recall ) the case of @xmath93 is the simplest .
set @xmath281 .
then under @xmath262 the @xmath282 are iid bernoulli variables of parameter @xmath283 and @xmath284 . hence from cramer s theorem ( see e.g. @xcite exercise 2.2.23 ) one gets that for any @xmath285 and any @xmath53 @xmath286 where @xmath287 if @xmath276 is fixed , when @xmath6 is sufficiently large @xmath288 , so that for@xmath1 large enough @xmath289 hence the result follows ( with e.g. @xmath290 ) from @xmath291 } \text { is self - avoiding } \right)\\ { { \;\leqslant\;}}\frac{\pi^1\left(|t_n|{{\;\geqslant\;}}\delta n\right)}{\pi^1\left ( ( s_n)_{n\in [ 0,n ] } \text { is self - avoiding } \right ) } , \end{gathered}\ ] ] and . note that under @xmath262 , @xmath295 and @xmath296 are sums of iid bernoulli random variables of parameter @xmath297 ( this is because @xmath298 is independent of @xmath299 ) . thus using cramer s theorem one gets that for any @xmath300 and any @xmath53 small enough , for large @xmath1 @xmath301 where @xmath302 is defined by . for some fixed small @xmath276
, one can get for large @xmath6 that @xmath303 hence choosing @xmath137 small enough and using that @xmath304 we have for @xmath1 marge enough @xmath305 finally we can conclude ( with e.g. @xmath306 ) by using and @xmath307 } \text { is self - avoiding } \right)\\ { { \;\leqslant\;}}\frac{\pi^1\left(|u_n|/n<(1-\delta)/(2d)\right)}{\pi^1\left ( ( s_n)_{n\in [ 0,n ] } \text { is self - avoiding } \right)}. \end{gathered}\ ] ] the other bound for @xmath292 could be treated similarly but we will see that this is not needed . by invariance under time reversal of the self - avoiding walk , @xmath310 and @xmath311 have the same law under @xmath250 .
note that we also have @xmath312 so that it is sufficient to show that @xmath313 recall that hence to our purpose it is sufficient to have a good bound on the numerator .
let us define @xmath315 a sequence of stopping time by @xmath316 and @xmath317 they are the time at which @xmath74 comes to a neighborhood of its non - immediate past .
we want to show that under @xmath251 , @xmath319 is a stopping time so that by the strong markov property , a.s . on the event `` @xmath320}$ ]
is self avoiding '' , @xmath321}){{\;\leqslant\;}}\left(\frac{2(d-1)}{2d-1}\right).\ ] ] this is because at least one out of the @xmath322 available options for @xmath323 breaks self - avoidance . hence @xmath324 } \text { is self - avoiding } \
|\ ( s_n)_{n\in[0,\tau_{n-1}+1 ] } \
\text { is self - avoiding } ) \\ = \pi^1 ( s_{\tau_n}+1\ne s_n,\ \forall n{{\;\leqslant\;}}\tau_n\ | \ ( s_n)_{n\in[0,\tau_{n-1}+1 ] } \
\text { is self - avoiding } ) { { \;\leqslant\;}}\left(\frac{2(d-1)}{2d-1}\right).\end{gathered}\ ] ] iterating this inequality @xmath1 times gives . as a consequence of we have @xmath325 }
\text { is self - avoiding } \}\cap \ { |w_1^n|{{\;\leqslant\;}}{\varepsilon}n\}\right)\\ { { \;\leqslant\;}}\pi^1\left ( \{(s_n)_{n\in [ 0,\tau_{{\varepsilon}n-1}+1 ] } \text { is self - avoiding } \}\cap \ { \tau_{{\varepsilon}n-1}{{\;\leqslant\;}}n-1\}\right){{\;\leqslant\;}}\left(\frac{2(d-1)}{2d-1}\right)^{{\varepsilon}n-1}.\end{gathered}\ ] ] then , combining this result with and we obtain @xmath326 choosing @xmath327 for a fixed @xmath276 , we get that for @xmath6 large enough holds .
this gives also the bound for upper - deviation of @xmath292 as @xmath328 .
the only statement that we need to complete the proof of proposition [ gudespine ] is that with large probability @xmath329 is equivalent to one when the asymptotic in @xmath6 is concerned .
this is considerably more complicated than for @xmath330 because in this case @xmath262 is not a fine enough approximation of the measure to get a conclusion .
we first want to prove a result concerning the measure @xmath257 .
the reason why we do our proof for @xmath257 and not for @xmath335 is that having a markovian walks is a useful tool for the proof .
the plan is then to transform it into a result for @xmath254 , and finally to use to conclude . to this purpose
we define @xmath336 by @xmath337 , and approximately , @xmath339 is the sequence of time at which @xmath74 comes at distance two of its non - immediate past ( it is not exactly right because of the extra conditions @xmath340 and @xmath341 ) .
we want to prove that the idea is again to use the markov property similarly as what we did to prove . given a self - avoiding trajectory @xmath343}$ ]
, one can choose some @xmath344 such that @xmath345 ( there might be several choices but we fix the value of @xmath346 for what follows e.g. we take the largest possible @xmath346 ) .
there is at least one way to reach @xmath347 in two steps from @xmath348 .
we have to check that such a combination of two step is authorized for @xmath257 , and in fact because @xmath349 , it is sufficient to check the first step is not a backtrack and does not form a loop of length four . because @xmath350 , a step going to a neighbor of @xmath347 can not be a backtrack , and as @xmath351 it can not create a loop of length four in the first step anyway .
as there are at most @xmath322 possibilities for each step , we deduce that the possibility of reaching @xmath347 in two is at least @xmath352 .
averaging over all the possibilities for @xmath343}$ ] , we deduce that @xmath353 } \text { is self - avoiding } \ | \ ( s_n)_{n\in[0,\tau^2_{n-1}+2 ] } \text { is self avoiding } \right ) \\ { { \;\leqslant\;}}\pi^2\left ( ( s_n)_{n\in[0,\tau^2_n+2 ] } \text { is self - avoiding } \ | \
( s_n)_{n\in[0,\tau^2_{n } ] } \text { is self avoiding } \right){{\;\leqslant\;}}1-\frac{1}{(2d-1)^2}.\end{gathered}\ ] ] now , because of the conditions @xmath341 and @xmath340 in the sequences @xmath354 is not the generalized inverse of @xmath355 .
however combined with additional results can still be useful to get a bound on the tail distribution of @xmath356 . set @xmath357 we have @xmath358 this is because if @xmath359 , either it is equal to one of the @xmath360 or @xmath361 or it belongs to @xmath362 .
hence @xmath363 we want to get a bound on the right hand side for @xmath364 , and we know that the second term is small from
. we will use to bound the first term . notice that for @xmath1 large enough @xmath365 } \text { is self - avoiding } ) \\ { { \;\leqslant\;}}\pi^2((s_{n})_{n\in [ 0,\tau_{{\varepsilon}n-1}+2 ] } \text { is self - avoiding } ) \\ { { \;\leqslant\;}}\left(1-\frac{1}{(2d-1)^2}\right)^{{\varepsilon}n-1}{{\;\leqslant\;}}\exp(-{\varepsilon}n/(2d)^2).\end{gathered}\ ] ] @xmath367 } \text { is self - avoiding } ) ^2 \\ { { \;\leqslant\;}}\pi^2\left(\left(\frac{\partial \pi^2_n}{\partial \pi^2}\right)^2\right ) \pi^2 ( \tau^2_{{\varepsilon}n}{{\;\leqslant\;}}n\ , \ ( s_{n})_{n\in [ 0,n ] } \text { is self - avoiding } ) , \end{gathered}\ ] ] where @xmath368 denotes the expectation of @xmath369 under the probability measure @xmath370 .
we know that the second term of the above equation decays exponentially fast .
we focus on showing that the first term does not increase so fast as to counterbalance it . from we
have @xmath371 the second factor is not too hard to bound from below , using jensen s inequality and the expectation @xmath373 is not difficult to approximate : for @xmath374 $ ] we have @xmath375 , because there are at least @xmath376 choices for @xmath100 ( at most @xmath44 are forbidden ) and at most one for which @xmath377 is satisfied .
hence @xmath378 we use the decomposition and cauchy - schwarz to get @xmath381 then we remark that almost surely @xmath382 indeed there are @xmath383 possibilities for making a @xmath124-turn and at most one of them is made unavailable by the condition that @xmath255 loops are not allowed .
the probability of a fixed combination of two available steps is at least @xmath352 , and hence , ( recall @xmath384 ) @xmath385 by a trivial induction ( recall the definition ) @xmath386 the previous lemma together with equations and @xmath387 performing taylor expansions in @xmath51 , we see that @xmath388 and that @xmath389 and we can conclude that there exists @xmath270 ( which we might choose as large as we wish ) such that for all @xmath6 large enough , and all @xmath271 @xmath390 recall that from , fixing @xmath391 we have @xmath392 } \text { is self - avoiding } ) { { \;\leqslant\;}}\exp(-{\varepsilon}n/(2d)^2){{\;\leqslant\;}}\exp(-5c / d^3).\ ] ] this allows us to conclude from that f. comets , t. shiga , and n. yoshida , _ probabilistic analysis of directed polymers in a random environment : a review _ , advanced studies in pure mathematics * 39 * ( 2004 ) 115 - 142 . m. def _ mathematics _ , rawcus records , new york 1999 . a.b .
harris _ self - avoiding walks on random lattices _
, z. phys .
b cond . mat .
* 49 * ( 1983 ) 347 - 349 .
harris and y. meir _ self - avoiding walks on diluted networks _ ,
* 63 * ( 1989 ) 2819 - 2822 . c. mcdiarmid , _ on the method of bounded differences
_ , surveys in combinatorics * 141 * ( 1988 ) 148 - 188 .
g. slade , _ the self - avoiding walk : a brief survey _
, in surveys in stochastic processes , 181 - 199 , j. blath , p. imkeller , s. roelly eds .
( 2011 ) , european mathematical society , zrich . | let @xmath0 be the number of self - avoiding paths of length @xmath1 starting from the origin on the infinite cluster obtained after performing bernoulli percolation on @xmath2 with parameter @xmath3 .
the object of this paper is to study the connective constant of the dilute lattice @xmath4 , which is a non - random quantity .
we want to investigate if the inequality @xmath5^{1/n}$ ] obtained with the borel - cantelli lemma is strict or not .
in other words , we want to know the the quenched and annealed versions of the connective constant are the same . on a heuristic level
, this indicates whether or not localization of the trajectories occurs .
we prove that when @xmath6 is sufficiently large there exists @xmath7 such that the inequality is strict for @xmath8 .
+ 2000 _ mathematics subject classification : 82d60 , 60k37 , 82b44 . _
+ _ keywords : percolation , self - avoiding walk , random media , polymers , disorder relevance . _ |
a classical definition of young diagram@xcite is the following : the young diagram if size @xmath1 is a finite descending ideal in the lattice @xmath2 , i.e. a set of cells in positive quadrant , which contains a cell @xmath3 only if it contains as well all the sells under @xmath3 in terms of particular order . standard young table of size @xmath1 is a young diagram whose
cells are filled with numbers in the range from @xmath4 to @xmath1 , which increase in each row and each column . in other words ,
young table is a path in a lattice of young diagrams , which starts at the empty ( zero - sized ) diagram and ends in the given diagram .
young diagrams correspond to irreducible representations of @xmath0 ( see for example @xcite ) , and the number of young tables that fit into a given young diagram is equal to the dimension of the corresponding irreducible representation . for the sake of brevity
we call this number the dimension of the diagram @xmath5 and denote it as @xmath6 .
this paper is organized as follows : in section 2 we will present the results of numerical experiments on asymptotics of a typical ( by plancherel measure ) dimension of irreducible representation of the symmetric group .
this measure was introduced in @xcite , for further details see @xcite .
section 3 is devoted to the similar computations of a maximum dimension of irreducible representation of the symmetric group .
the results of section 2 should be considered as a support of a conjecture @xcite about the existence of a limit value of normalized dimension of a typical @xmath5 by plancherel measure .
this hypothetical limit value is denoted as @xmath7 and called in @xcite `` specific entropy of an irreducible representation '' . at the same time , basing on out computations in section 3 , nothing conclusive can be said about the asymptotic behavior of the maximum dimension .
that is , there is no claim that @xmath8 comes to plateau on approachable values of @xmath1 . in section 4 , we are concerned with another distribution on young diagrams : richardson distribution , which was studied in @xcite .
we give an experimental evidence of the proved theorem about a limit shape of typical young diagrams .
then we make the similar experiments in three dimensions ( this time the dimension has nothing to do with @xmath0 ) , and form a hypothesis on a 3-dimensional limit shape .
let @xmath9 be the collection of equivalence classes of complex irreducible representations of @xmath0 . for @xmath10 ,
we denote by @xmath11 the dimension of the representation @xmath12 , and by @xmath13 we denote the plancherel measure @xcite .
this is actually a probabilistic measure on @xmath9 , which we can derive from burnside s conjecture : @xmath14 the set @xmath9 and the dimension @xmath11 have the following interpretation @xcite : @xmath9 is the ensemble of all young diagrams of size @xmath1 , and @xmath11 is the number of standard young tables that fit into the diagram @xmath12 . in the present article , we make no difference between @xmath12 and the corresponding irreducible representation .
vershik and kerov @xcite stated the following conjecture : there exists a limit @xmath21 almost everywhere ( by plancherel measure ) on the set of infinite young tables .
infinite young table is an infinite sequence of nested young diagrams of sizes increasing one - by - one : @xmath22 .
the hypothetical limit value is denoted as @xmath7 and called `` specific entropy of an irreducible representation '' .
we are studying the behavior of coefficient @xmath17 w.r.t .
some experiments in this area were undertaken in @xcite , where the expectation and variance of @xmath17 are given for five values of @xmath1 , 1600 maximum ( for @xmath23 the sample was only 14 diagrams ) .
we have done the same computations on modern hardware , which is far more powerful than 25 years ago
. * note . *
after we did the described experiments , it was reported of a new paper by alexander bufetov , which gives a proof of existence of a limit @xmath24 of @xmath17 in @xmath25 by plancherel measure : @xmath26 the actual value of @xmath24 is not likely to be obtainable by methods used in bufetov s proof . in should be also noted that the conjecture @xcite about the existence of a limit if @xmath17 almost everywhere is still open , and if it happens that such a limit value @xmath7 exists , then clearly @xmath27 .
hook formula ( see for example @xcite ) allows to compute @xmath28 , avoiding the enumeration of all young tables that fit into @xmath5 .
@xmath29 where @xmath3 is a cell of young diagram , and @xmath30 is the length of the hook associated to that cell .
the hook of the cell @xmath3 consists of the cell itself and all cells that are in @xmath31-th row to the right from @xmath3 or in @xmath32-th column above @xmath3 ( fig .
[ hooks ] )
. the definition of random young diagram by richardson measure can be easily generalized to three - dimensional case .
there are no known results about the limit shape in this case , but the figure [ richardson - shape-3d ] , obtained by our computations , leads to assumption that the limit shape complies to similar equation @xmath33 .
this result shows that the limit shape of 3-dimensional young diagrams generated by richardson process are probably different from the limit shape for uniformly - distributed diagrams , which was studied in @xcite , and finally found in @xcite . | the article presents the results of experiments in computation of statistical values related to young diagrams , including the estimates on maximum and average ( by plancherel distribution ) dimension of irreducible representation of symmetric group @xmath0 .
the computed limit shapes of two - dimensional and three - dimensional diagrams distributed by richardson statistics are presented as well . |
glassy systems are characterized by very slow relaxations and non - equilibrium behavior . in general , there is however a high temperature ( or low density , pressure , etc ) equilibrium phase where all quantities relax to their equilibrium value in finite time . between the two is the putative glass transition , which has been identified as a true singularity in the free energy in spin glasses or in the dynamic free energy in mean field models of glass formers .
the main predictions of mean field models concerning the glass phase are the existence of an effective temperature @xmath2 at which the slowest degrees of freedom are `` thermalized '' and a specific form for the decay of the correlators at long times .
although effective temperatures have been reported in experiments and numerical simulations @xcite , the time evolution predicted by mean field models is in general not observed .
there is thus still a deep challenge in interpreting the measurements of slow relaxations in the glass phase as well as in describing the crossover between aging and equilibrium dynamics which take place when one is patient enough to wait for the system to reach equilibrium . among many interesting features of glassy materials , waiting time - temperature superposition ( wtts ) has been reported in various glass formers @xcite .
the aim of this letter is , in the context of a mean field model , to answer the following important questions .
are mean field predictions compatible with experimental observations ?
are slow dynamics in the glass phase intrinsically different from slow dynamics in the equilibrium phase ?
how does the crossover from aging to equilibrium occur at the critical point ? what is the origin of the observed wtts ?
the answer to these questions we will shed new light on the central role played by the relaxation of the energy . the letter is organized as follows : we first describe the model and the old solution of its dynamics , and then improve it with some new results .
next , we establish a correspondence between slow dynamics at equilibrium close to @xmath0 and aging dynamics below @xmath0 , which becomes an exact mapping at @xmath0 .
the consequences of this correspondence , wtts , interrupted aging and dynamical heterogeneity , are then discussed . some results presented here
follow from long and technical derivations , which will not be given here and will be detailed somewhere else @xcite .
we start discussing the dynamics of the spherical p - spin model , where exact statements can be made @xcite .
it can be seen as a ground for establishing rigorous results within the landscape approach following goldstein @xcite , and developed by many authors - see @xcite for instance .
the model consists of @xmath3 variables @xmath4 subject to the constraint @xmath5 , with hamiltonian @xmath6 , where the @xmath7 s are quenched gaussian random variables .
explicit equations can be written for the dynamics of correlators @xmath8 and response functions @xmath9 to an external field .
the solution of these equations , which can be found in @xcite , captures the features of glassy dynamics , where two well separated times scale emerge , corresponding to different kinds of degrees of freedom , some - local - responding as in equilibrium , and others - structural - being out of equilibrium . in the equilibrium time
sector , fluctuation dissipation theorem ( fdt ) holds . in the long time aging sector , fdt does not hold , but instead a modified form of it , where the bath temperature @xmath10 has to be replaced by an effective temperature @xmath11 which can be computed explicitly .
a fine analysis shows that slow degrees of freedom are indeed thermalized at @xmath2 rather than @xmath10 , coming from the contribution of the structural degrees of freedom to the entropy at some energy @xmath12 .
the threshold energy @xmath12 is the energy separating unstable and stable states , and corresponds to the asymptotic energy : at long times , a glassy system is close to marginal stability . in the equilibrium phase ,
the energy is @xmath13 is above @xmath14 , and at the approach of @xmath0 , the system gets closer to the threshold and slows down , almost trapped by marginally stable states .
similarly , in the glass phase , where the equilibrium energy is below the threshold , the system loses equilibrium when it gets close to @xmath15 and thus ages . in both cases ,
the closer the system is from the threshold level , the slower it is @xcite . before making this more precise , we briefly recall the known scalings for the correlator @xmath16 and the response function @xmath17 , using as usual @xmath18 , in the equilibrium and glass phases @xcite
. in the equilibrium phase , i.e. @xmath19 , there are two characteristic time scales , the @xmath20-decay around the plateau and the @xmath21-relaxation @xcite : @xmath22 and @xmath23 , @xmath24 , where @xmath25 and @xmath26 are well known mct exponents . there , the scalings for @xmath27 are , in order : @xmath28 and @xmath29 , where @xmath30 is the non - ergodicity parameter at @xmath0 . in the glass phase @xmath31 ,
two characteristic time sectors have also been identified @xcite .
the first time sector is stationary : when @xmath32 , there is time translation invariance ( tti ) : @xmath33 and fdt is still valid : @xmath34 .
the second - aging - time sector corresponds to @xmath35 @xcite .
there , fdt must be modified : @xmath36 and the scaling is @xmath37 .
an interesting but unpleasant feature is that the analysis of the aging equations at infinite @xmath38 is not sufficient to determine @xmath39 @xcite , which has remained unknown until recently . from this point of view , equilibrium and aging
dynamics seem to differ strongly .
we now improve an analysis made recently of the dynamical equations at large but finite @xmath38 @xcite and show that in fact both phases are very similar .
recently , it was shown that for @xmath31 , aging actually sets in around the plateau @xcite , a time sector which is usually just considered as the matching point of the tti and aging sectors .
it was shown that in this regime , the correlators have a scaling form @xmath40 , where the exponents are related through @xmath41 .
in addition , matching with the aging regime , the uncertainty on the function @xmath39 was considerably reduced to @xmath42 , where @xmath43 is a constant which can be safely absorbed into the relaxation time and set to @xmath44 , while the aging exponent @xmath45 verifies @xmath46 and now @xmath25 and @xmath26 are aging mct exponents . remark that @xmath45 is required for this solution to be consistent
. situations where this occurs have been reported in various experiments , e.g. @xcite .
values of @xmath47 exceeding @xmath44 ( super - aging , see e.g. @xcite ) are the signature of a different relaxation mechanism . after this analysis ,
the complete solution is still to be found , as the exponents are unknown although all determined by @xmath21 .
however , interestingly , in the same time the energy was shown to relax as @xmath48 , meaning that the whole slowness of the correlators is actually encoded in the energy relaxation towards the threshold level .
this is remarkable and will lead to the striking correspondence between equilibrium and aging dynamics , which we shall establish below .
last , but not least , the scaling of the response function is @xmath49 and the ratio @xmath50 interpolates smoothly between @xmath51 and @xmath52 , showing that all interesting features of the aging regime already occur in the plateau regime , which corresponds to the time scales where motion propagates from local to structural degrees of freedom .
we now analyze further and improve these results , in order to obtain a unified picture of equilibrium and glass phases .
first , we recall the simplest of the two equations verified by the scaling functions @xmath53 and @xmath54 around the plateau @xcite ( where @xmath55 is such that @xmath56 ) : @xmath57 when @xmath58 , @xmath59 , and eq .
( [ eq : g ] ) reduces to the equation for the scaling function @xmath60 of equilibrium dynamics @xcite .
second , writing @xmath61 , it is easy to verify that in order to have @xmath62 be of order @xmath44 , one needs that @xmath63 .
this has an important consequence : the scaling @xmath62 with a stretched exponential @xmath39 is in general very hard to observe except in the p - spin model @xcite , and in the large @xmath38 limit it differs only through tiny asymptotic corrections to the observed scaling @xmath64 .
this explains the many reported discrepancies between mean field predictions and measurements @xcite .
third , and more strikingly , the aging relaxation can also be related to the scaling function of the @xmath21-regime in the equilibrium phase , _ at all temperatures_. indeed , the scaling function @xmath65 was found in @xcite to verify @xmath66 we now show that the solution of this equation is in fact a familiar function of the equilibrium phase . writing @xmath67 , one gets , after a bit of algebra : @xmath68 with @xmath69 , which is exactly the equation verified by @xmath70 @xcite . using @xmath71 , and remarking that the small @xmath72 behavior must match the @xmath20-regime , we can determine unambigously @xmath73 , leading to : @xmath74 , where @xmath75 .
the interpretation of the preceding paragraph will be central result .
we introduce @xmath76 and obtain for @xmath31 : @xmath77 and @xmath78 . added to the above scalings
, this leads us to the conclusion that _ once temperature and @xmath38 dependences are expressed in terms of @xmath79 only , equilibrium and aging dynamics almost map onto each other . _ almost , because @xmath25 , @xmath26 and the scaling functions in the @xmath20-regime explicitly depend on @xmath10 below @xmath0 .
however , the mapping becomes exact at the transition , and is very accurate close to it ( see fig .
[ fig ] ) .
this is a rather unexpected and spectacular result , as it gives a route to studying aging dynamics from the knowledge of the deeply supercooled equilibrium phase - calculations in both phases are not intrinsically different .
in addition , it incorporates in a natural way the wtts reported in experiments .
an immediate consequence of that is interrupted aging .
indeed , the glass transition temperature @xmath80 is where a glass former seems to lose equilibrium , but where , with a bit of patience , it is still possible to have it equilibrate .
more precisely , there is a crossover waiting time @xmath81 such that for @xmath82 aging occurs , while for @xmath83 , the system has reached equilibrium and its dynamics is tti . by definition of @xmath80 , @xmath81 is large . a nave guess for the value of @xmath81 would be @xmath84 , the equilibrium relaxation time .
however , this is too nave , as this is the time for the system to have visited several equilibrium states , while @xmath81 is the time where aging stops and is in principe much smaller . in our model
, it is possible to compute @xmath81 using the previous analysis , from the following gedanken experiment : quench the system from @xmath85 down to @xmath86 , @xmath87 . during the first stage of the dynamics ,
the energy is close but not enough to its equilibrium value and thus relaxes as a power law : @xmath88 and the system ages .
this ends when the energy reaches its equilibrium value @xmath89 , which gives : @xmath90 .
this leads to the following scaling form for the @xmath21-relaxation time : @xmath91 , where @xmath92 at @xmath93 and @xmath94 .
remark that @xmath81 may also be expressed as @xmath95 , that is the exponent is the ratio of the ones of the relaxation time in both phases , which gives an easier way to measure it in the mct regime . using the previous results , it is possible to describe the general behavior of correlation functions generally used for describing dynamical heterogeneities ( dh ) .
we start with the now widely studied @xmath96 , where @xmath97 is a local correlator at position @xmath98 .
it is possible @xcite to extend the field - theoretic approach to @xmath99 used by biroli and bouchaud @xcite to the aging regime .
it results that @xmath100 is a sum of ladder diagrams built with the full correlator @xmath97 , which time dependence follows that of the previous paragraphs .
this has several immediate consequences .
first , all scaling results given in previous analysis of @xmath99 in the equilibrium phase can be directly applied to the glass phase , provided we replace @xmath1 by @xmath79 , providing mct predictions for dh in he glassy phase .
second , it was reported in @xcite that , once @xmath101 is plotted versus @xmath102 , all data - i.e. for all @xmath38 s - collapse onto a single master curve .
this could be obtained from general scaling considerations , but interestingly , it comes out naturally from the fact that all @xmath38-dependance of both quantities plotted is through @xmath79 , which gives the natural parameterization of the master curve .
in addition , close to @xmath0 , the full master curve should concide with the one of the equilibrium phase .
another quantity which has been shown to be interesting above @xmath0 is @xmath103 , which has been shown to have the same critical behavior as @xmath104 , and which is more easily accessed experimentally @xcite .
it is very simple to generalize this quantity to the glass phase , as @xmath105 is identical to @xmath106 , up to the non - singular multiplying factor @xmath107 .
it is thus natural to generalize @xmath108 to : @xmath109 in the equilibrium phase , @xmath110 reduces to @xmath111 , while in the glass phase , it becomes @xmath112 .
this expression makes @xmath113 as easy to measure in experiments as @xmath108 . in fig .
[ fig ] , @xmath113 at @xmath114 and @xmath108 at @xmath115 , with approximately the same value of @xmath1 are plotted versus @xmath72 . clearly , the peaks of the @xmath21-relaxation concide very well , while the @xmath20-regime slightly differ , having two different exponents , respectively @xmath116 and @xmath117 . when the same procedure is made at @xmath58 for @xmath113 ,
the curves fall on top of each other .
comparison of @xmath111 above @xmath0 ( straight line ) and @xmath118 well below @xmath0 ( dashed line ) , with the same value of @xmath1 .
the curves have been rescaled vertically , as the factor @xmath119 occurring in @xmath113 is not known . ]
in this letter , we have shown that , within a mean field model , dynamics in the ideal glass phase are essentially identical to equilibrium dynamics just above the glass transition , and that all important features in both phases may be absorbed in a single function @xmath79 , with important physical significance , being the distance in energy to threshold states .
this is a remarkable result , as it opens the door to interpreting the aging data obtained in the glass phase of molecular or collodal glass formers . in particular , it provides general scaling laws for multipoint correlation functions - relaxation , dynamical heterogeneities - as well as the natural parameterization to seek , @xmath79 , which may alternatively be an enthalpy difference .
it also naturally predicts wtts , which was reported in aging measurements @xcite . on the conceptual aspect ,
it is very unexpected , because it shows that aging dynamics is actually very similar from equilibrium dynamics in the deeply supercooled regime , which is rather far from common thinking about aging .
but this is not so surprising , and reflects the fact that both in liquid and glass phases , slow dynamics occurs because of the roughness of the energy landscape , which shape does not change qualitatively more than static quantities when crossing the glass transition . here , several strong predictions have been made , which should be tested in experiments and numerical simulations . doing so , one must keep in mind that in general , the exponents @xmath25 and @xmath26 both depend on temperature , and thus the predictions made here should in general be tested at close temperatures .
this would also be of valuable help for determining situations where the energy landscape approach gives a qualitatively correct picture . on this prospect , and
more speculatively , one may ask whether the effect of activation at low temperature may be also reabsorbed in the same way
. it would be also worth coming back to data from previous experiments , where disagreement with former mean field predictions where found , and reinterpret them using the predictions given in this letter .
this work is supported by the anr grant dynhet .
the author thanks g. biroli and j .- p .
bouchaud for stimulating discussions and careful reading of the manuscript , as well as j. kurchan .
this article was written during a stay at the columbia university , ny , and d. reichman is warmly thanked for his hospitality .
the numerical checks have been made using a code provided by k. miyazaki .
99 l. f. cugliandolo and j. kurchan , _ phys .
lett . _ * 71 * , 173 ( 1993 ) ; l. f. cugliandolo , in slow relaxations and nonequilibrium dynamics in condensed matter , les houches , j .- l .
barrat , m. feigelman , j. kurchan and j. dalibard editors , edp sciences : les ulis , springer : berlin ( 2002 ) .
p. wang , c. song and h. a. makse , _ nature physics _ * 2 * , 526 ( 2006 ) .
w. kob and j .- l .
barrat , _ phys .
lett . _ * 78 * , 4581 ( 1997 ) .
l. c. e. struick , _ physical aging in amorphous polymers and other materials _ , elsevier : houston ( 1978 ) .
p. a. oconnell and g. b. mc kenna , _ polym .
sci . _ * 37 * , 1485 ( 1997 ) .
p. lunkenheimer et al .
, _ journal of non - crystalline solids _ , * 352 * 4941 ( 2006 ) .
a. lefvre , _ unpublished_. m. j. goldstein , _ j. chem . phys . _ * 51 * , 3728 ( 1969 ) .
a. cavagna , i. giardinia and g. parisi , _ j. phys . a : math .
gen . _ * 34 * , 5317 ( 2001 ) .
s. sastry , _ nature _ * 409 * , 164 ( 2001 ) .
j. kurchan and l. laloux , _ j. phys . a : math .
gen . _ * 29 * , 1929 ( 1996 ) .
w. goetze in liquid , freezing and the glass transition , les houches , j. p. hansen , d. levesque , j. zinn - justin editors , north holland ( 1989 ) .
this was also noticed in @xcite , but not the exact connexion to equilibrium dynamics .
l. f. cugliandolo , j. kurchan and p. le doussal , _ phys .
* 76 * , 2390 ( 1996 ) .
this is actualy not correct when @xmath120 , as shown later in the text , but we recall it from ref .
a. andreanov and a. lefvre _ europhys .
lett . _ * 76 * , 919 ( 2006 ) .
b. kim and a. latz , _ europhys .
lett . _ * 53 * , 660 ( 2001 ) .
v. viasnoff , f. lequeux , and d. pine , _ rev .
inst _ * 73 * , 2336 ( 2002 ) ; el masri d. et al ,
_ j. stat .
_ , p07015 ( 2009 ) .
o. ronsin , c. caroli and t. baumberger , _ phys .
* 103 * , 138302 ( 2009 ) ; r. paul , g. schehr and h. rieger , _ phys .
* 75 * , 030104 ( 2007 ) .
g. biroli and j .- p .
bouchaud , _ europhys . lett._*67 * , 21 ( 2004 ) .
a. parsaeian and h. e. castillo , arxiv : 0811.3190 .
l. berthier et al . , _ science _ * 310 ( 5755 ) * , 1797 ( 2005 ) ; l. berthier et al . , _ j. chem .
_ , * 126 * , 184503 and 184504 ( 2007 ) . | we study and compare equilibrium and aging dynamics on both sides of the ideal glass transition temperature @xmath0 . in the context of a mean field model
, we observe that all dynamical behaviors are determined by the energy distance @xmath1 to threshold - i.e. marginally stable - states .
we furthermore show the striking result that after eliminating age and temperature at the benefit of @xmath1 , the scaling behaviors above and below @xmath0 are identical , reconciling _ en passant _ the mean field results with experimental observations . in the vicinity of the transition ,
we show that there is an exact mapping between equilibrium dynamics and aging dynamics .
this leads to very natural interpretations and quantitative predictions for several remarkable features of aging dynamics : waiting time - temperature superposition , interrupted aging , dynamical heterogeneity . |
the purpose of the e246 experiment is to measure the transverse component of the muon polarization in the decay @xmath0 ( @xmath1 ) , while we were also able to extract @xmath2 ( @xmath3 ) decay as a by - product .
the transverse muon polarization is a t - odd observable defined as @xmath4 where @xmath5 is used for @xmath1 and @xmath6 for @xmath3 , respectively . in the framework of the phenomenological consideration
, the transverse muon polarization can be related to the @xmath1 and @xmath3 form factors .
for @xmath1 the t - violating polarization is proportional to the imaginary part of the ratio of @xmath1 form factors : @xmath7 , where @xmath8 and @xmath9 , @xmath10 are defined through @xmath11 \cdot \left[\bar{u}_\mu\gamma_\lambda(1-\gamma_5)u_\nu\right]\ ] ] the standard model predicts a vanishing value of less than @xmath12 for @xmath13 in @xmath1 @xcite .
the calculations of @xmath13 due to the electromagnetic final state interactions result in a value of less than @xmath14 @xcite .
there are several non - standard models predicting sizeable value for @xmath13 : multi - higgs models , susy with squarks mixing , susy with r - parity violation , leptoquark models @xcite .
the values predicted in these models vary from @xmath15 to @xmath16 . in case of @xmath3
the transverse polarization is related to the decay form factors in a more complicated way : @xmath17\mbox{im}(\delta_p)$ ] where @xmath18 and @xmath19 $ ] are the functions of @xmath3 kinematic parameters ( shown in fig .
[ fig : sigma ] ) , $ ] ( left ) and @xmath20 $ ] ( right ) over the dalitz plot .
the standard @xmath3 kinematic variables @xmath21 and @xmath22.,title="fig:",width=264 ] $ ] ( left ) and @xmath20 $ ] ( right ) over the dalitz plot . the standard @xmath3 kinematic variables @xmath21 and @xmath22.,title="fig:",width=264 ] and @xmath23 are the contributions of non - standard interactions to the effective form factors @xcite .
although the standard model prediction for @xmath13 in @xmath3 is as small as for @xmath1 @xcite , the contribution of the final state interaction to this value is @xmath24 @xcite , i.e. considerably larger than in the case of @xmath25 decay .
the predictions for non - zero @xmath13 for @xmath3 come from the same non - standard models mentioned for @xmath1 and also from left - right symmetric models @xcite .
the expected values vary from @xmath26 to @xmath16 .
the noteworthy peculiarity of these predictions obtained in different models is the correlations between the expected values of @xmath13 for @xmath1 and @xmath3 @xcite : in the three higgs doublet model the @xmath13 expectations are related as @xmath27 ; in susy with squarks mixing the relation for @xmath13 induced by higgs exchange : @xmath28 , while for @xmath13 arising from w - boson exchange : @xmath29 , @xmath30 ; in susy with r - parity violation the relation is @xmath31 ; in left - right symmetric models we have : @xmath32 , @xmath30 .
the previous experimental results for @xmath1 came from the bnl experiment which used kaon decays in flight @xcite : @xmath33 , @xmath34 , as well as from e246 @xcite : the result obtained for the data collected during 1996 - 1997 period was @xmath35 , @xmath36 .
both results indicated no t - violation in @xmath1 . for @xmath3
there has been no @xmath13 measurement , so our result is the first one .
the e246 apparatus is shown in fig . [ fig : e246setup ] , and described in detail elsewhere @xcite . kaons with @xmath37 mev / c
are identified by a @xmath38erenkov counter , slowed in an al+beo degrader and then stopped in a target array of 256 scintillating fibers located at the center of a 12-sector superconducting toroidal spectrometer .
charged particles from kaon decays in the target were tracked by means of multi - wire proportional chambers at the entrance ( c2 ) and exit ( c3 and c4 ) of each magnet sector , along with the target and a scintillation ring hodoscope around the target .
the momentum resolution of @xmath39 mev / c at p=205 mev / c was obtained using mono - energetic products from the two - body decay @xmath40 ( @xmath41 ) . the energies and angles of the photons from @xmath42 decays
were measured by a csi(tl ) photon detector consisting of 768 modules .
the photon detector covers a solid angle of 3@xmath43 steradian , with openings for the beam entry and exit and 12 holes for charged particles to pass into the magnet gaps . to suppress accidental background from the beam , timing information from each crystal
was used . a good time resolution of 3.5 nsec ( rms ) at 100 mev was achieved .
energy resolution of @xmath44% at 200 mev , angular resolutions of @xmath45 and the invariant mass resolution of @xmath46 mev / c@xmath47 were obtained .
muons entering the polarimeter ( fig .
[ fig : e246setup]c ) were degraded by an al+cu block and stopped in a stack of pure al plates .
positrons from @xmath48 decays of stopped muons were detected by positron counters which were located azimuthally between the muon stoppers .
the trigger included the signals from @xmath38erenkov , target , tof and positron counters along with the requirement of at least one hit in csi(tl ) calorimeter . the t - violating asymmetry was extracted using a double ratio as : @xmath49 \label{atdr}\ ] ] here , @xmath50 and @xmath51 are the sums over all 12 sectors of counts of clockwise ( @xmath52 ) and counter - clockwise ( @xmath53 ) emitted positrons and @xmath54/@xmath55 denote events with the photon ( or @xmath42 ) going forward / backward with respect to beam direction .
the sign of @xmath56 for @xmath54 events is opposite to that of @xmath55 events that allows us to employ a double ratio method which reduces most systematic errors and enhances the effect . moreover , considerable reduction of systematic effects was achieved due to the azimuthal symmetry of the 12-sector detector .
the value of @xmath57 is related to @xmath58 by @xmath59 where @xmath60 is the analyzing power of the polarimeter , @xmath61 is an angular attenuation factor and @xmath62 is the overall fraction of backgrounds .
the extraction and analysis of @xmath1 and @xmath3 events comprised several procedures common for both decays .
the common stage included target analysis , charged particle tracking and tof analysis and the analysis of muon decay in polarimeter .
active target analysis included target energy deposition and target timing cuts to get rid of kaon decays in flight .
the momentum of charged particle reconstructed by four - point tracking procedure was used to suppress @xmath63 and @xmath41 decays by selecting events with @xmath64 mev / c .
the cut on @xmath65 for the charged particle track was used to suppress @xmath41 decays with @xmath66 decay in flight .
to separate muons and positrons ( thereby suppressing @xmath67 ) we used time - of - flight ( tof ) technique to calculate charged particle mass and then , on the scatter plot of the tof energy deposition versus tof - reconstructed mass square , we separated the muon cloud from the positron one .
finally , the common stage included the signal extraction from the positron time spectra in the polarimeter .
the second stage of event selection was specific for each decay mode .
for @xmath1 we selected one - photon events with @xmath68 mev and two - photon events which satisfy the constraints on the invariant mass of the two photons and on the missing mass ( reconstructed mass of missing neutrino ) : @xmath69 mev / c@xmath47 and @xmath70 mev@xmath47/c@xmath71 .
additionally , we used the cuts on the opening angles between two photons and between muon and @xmath42 : @xmath72 and @xmath73 to suppress kaon decays in flight and @xmath41 . the second stage for @xmath3 selected one - photon events with @xmath74 mev and
comprised three major cuts : a constraint on the neutrino missing mass @xmath75 mev@xmath47/c@xmath71 , a cut on the neutrino momentum @xmath76 mev / c and a cut on the opening angle between muon and photon @xmath77 .
these cuts suppressed the @xmath1 by a factor of 70 , while sustaining a @xmath3 loss by a factor of 2 .
[ fig : kmng ] selection .
the black dots with error bars show the experimental data .
mc simulation : dotted line @xmath78 , dashed line
@xmath25 , solid line
@xmath79.,title="fig:",width=264 ] selection .
the black dots with error bars show the experimental data .
mc simulation : dotted line
@xmath78 , dashed line
@xmath25 , solid line
@xmath79.,title="fig:",width=264 ] shows the spectra of @xmath80 and @xmath81 for both monte carlo ( mc ) ( @xmath1 and @xmath3 ) and experimental data .
the background fractions for extracted @xmath1 and @xmath3 events were estimated using experimental spectra along with the mc simulation data . for @xmath1
the major background contributions come from csi(tl ) accidental hits , @xmath67 and @xmath41 .
all these backgrounds do not induce spurious asymmetry and only dilute the sensitivity to @xmath13 .
the total background fraction for @xmath1 was estimated to be @xmath82% . in the case of @xmath3
the situation is much worse due to the predominant background of @xmath1 events with one photon escaping csi(tl ) detector .
such events almost completely mimic @xmath3 kinematics , so they can not be suppressed without a considerable loss of useful @xmath3 events .
the optimized @xmath3 cuts reduced the background fraction from @xmath1 to about 17% .
the background @xmath1 events might have non - zero @xmath13 thereby inducing spurious transverse asymmetry .
fortunately , we measure @xmath13 in @xmath1 with higher sensitivity in the same experiment and can reliably estimate this effect .
therefore , we can safely assume no spurious contribution to @xmath13 from @xmath1 background .
the second major source of background is accidental photons in the csi(tl ) detector .
it was suppressed to the level of @xmath83% by requiring a photon energy threshold of 50 mev and a coincidence between a signal from charged particle and a photon signal in the csi within a window @xmath84 ns .
other background modes are suppressed by the @xmath3-specific cuts to a negligible level .
the total background fraction for @xmath3 sample was estimated to be @xmath85% . the valuable part of the asymmetry analysis includes the extraction of the value of the normal asymmetry @xmath86 which is proportional to the t - even muon polarization , i.e. the in - plane component of the muon polarization normal to the muon momentum .
it can be measured by selecting events with @xmath42 ( or photon for @xmath3 ) moving into the left or right hemisphere with respect to the median plane of the given magnet sector .
the theoretical calculations indicate that the normal polarizations for @xmath1 @xcite and @xmath3 @xcite have opposite signs for the kinematic region where selected events are located , as shown in fig .
[ fig : pn ] .
( left ) and @xmath3 ( right ) .
the dots represent the @xmath3 experimental data.,title="fig:",width=264 ] ( left ) and @xmath3 ( right ) .
the dots represent the @xmath3 experimental data.,title="fig:",width=264 ] for one - photon @xmath1 events , we obtained @xmath87 while for @xmath3 we have @xmath88 .
this gives us a robust evidence of the sufficiently pure @xmath1 and @xmath3 selection .
the study of the dependence of the normal asymmetry on the energy of @xmath42 shows a sound agreement with the theoretical calculation . by comparing the values of the normal asymmetry and normal polarization obtained using monte carlo simulation using the relation @xmath89 , we are able to extract the value of the polarimeter analyzing power @xmath90 .
the main systematic uncertainties in @xmath13 come from magnetic field rotation , detector components misalignment , csi(tl ) accidental background , beam profile asymmetry and decay plane rotation .
the important point is that the sources of systematics are mostly the same for both measured decays , so we can estimate the systematic contributions using a large sample of @xmath1 data and then use the same error value for @xmath3 .
most of the systematic effects are canceled by the unique features of the set - up ( azimuthal symmetry , double ratio , etc ) .
we evaluated the contributions from all relevant sources using the mc simulation data and comparing them with the experimental distributions @xcite .
overall systematic error was estimated to be @xmath91 which is well below the level of the statistical error .
after the analysis of the entire data set collected in the period 1996 - 2000 we obtained the results for both @xmath1 and @xmath3 .
for @xmath1 we selected @xmath92 one- and two - photon forward / backward events and using the angular attenuation factor value extracted from mc simulation ( @xmath93 for @xmath94 events and @xmath95 for @xmath96 events ) the preliminary result @xmath97 is obtained . using the relation @xmath98 ( where @xmath99 is a kinematic factor , evaluated from mc simulation @xmath100 ) we get @xmath101 @xcite .
the dependences of the transverse asymmetry on the beam cycle , @xmath42 energy and magnet sector number indicate no systematic irregularity and thus confirm the robustness of our systematics study .
the results indicate no evidence of t - violation in @xmath1 and can be interpreted as limits on the measured quantities : @xmath102 at 90% c.l . and
@xmath103 at 90% c.l .
we have performed the first measurement of @xmath13 in the @xmath3 decay ( it is also the first @xmath3 measurement below @xmath41 peak ) .
the result obtained for 1996 - 1998 data was published in @xcite . here
we present the result of the analysis of the whole data set of @xmath104 forward / backward events .
we obtained the value @xmath105 with the evaluated attenuation factor of @xmath106 . similarly to the case of @xmath1 , we see no indication of t - violation in this decay and we can put the limit @xmath107 ( 90% c.l . )
we have performed the new measurement of t - violating muon polarization in two decays @xmath1 and @xmath3 for which several non - standard models predict non - zero @xmath13 values . for @xmath3 decay
our result is the first one . at the current level of the experimental sensitivity
, we see no evidence for t - violation and our results allowed us to impose constraints on several non - standard models : three higgs doublet model ( the most stringent experimental constraint ) , susy with squark mixing , susy with r - parity violation , leptoquark models , left - right symmetric models ( see , for example , ref .
much higher statistical sensitivity to @xmath13 of @xmath108 can be reached in a proposed experiment @xcite at the high intensity low energy separated kaon beam at j - parc @xcite .
in addition , the @xmath109 values in @xmath25 and @xmath78 can be measured with high accuracy in this experiment that provides a new sensitive method for determination of the kaon form factor values in these decays @xcite .
99 e. golowich and g. valencia , phys .
d * 40 * , 112 ( 1989 ) ; i. i. bigi and a. i. sanda , cambridge monogr . part .
cosmol .
* 9 * , 1 ( 2000 ) .
a. r. zhitnitsky , sov .
j. nucl .
* 31 * ( 1980 ) 529 [ yad .
* 31 * ( 1980 ) 1024 ] ; v. p.
efrosinin _ et al .
_ , phys .
b * 493 * , 293 ( 2000 ) [ arxiv : hep - ph/0008199 ] . c. q. geng and s. k. lee , phys .
d * 51 * , 99 ( 1995 ) ; g. h. wu and j. n. ng , phys . lett .
b * 392 * , 93 ( 1997 ) ; m. fabbrichesi and f. vissani , phys .
d * 55 * , 5334 ( 1997 ) ; g. belanger and c. q. geng , phys .
d * 44 * , 2789 ( 1991 ) .
m. kobayashi , t. t. lin and y. okada , prog .
phys . * 95 * , 361 ( 1996 ) [ arxiv : hep - ph/9507225 ] .
c. h. chen , c. q. geng and c. c. lih , phys .
d * 56 * , 6856 ( 1997 ) v. v. braguta _ et al .
_ , phys .
d * 66 * , 034012 ( 2002 ) [ arxiv : hep - ph/0205203 ] .
g. h. wu and j. n. ng , phys .
d * 55 * , 2806 ( 1997 ) [ arxiv : hep - ph/9610533 ] .
s. r. blatt _ et al .
_ , phys .
d * 27 * ( 1983 ) 1056 .
m. abe _ et al . _ [ kek - e246 collaboration ] , phys .
lett . * 83 * , 4253 ( 1999 ) ; m. abe _ et al . _ [ kek - e246 collaboration ] , nucl .
a * 663 * ( 2000 ) 919 .
m. abe _ et al . _
[ e246 kek ps collaboration ] , nucl .
instrum .
a * 506 * , 60 ( 2003 ) n. cabibbo and a. maksymowicz , phys .
* 9 * ( 1964 ) 352 ; errata ibid . , * 11 * ( 1964 ) 360 ; errata ibid.,*14 * ( 1965 ) 72 .
m. abe _ et al .
_ [ kek - e246 collaboration ] , nucl . phys . a * 721 * , 445 ( 2003 ) [ arxiv : hep - ex/0211049 ] . v. v. anisimovsky _ et al .
_ [ kek - ps e246 collaboration ] , phys .
b * 562 * , 166 ( 2003 ) .
f. l. bezrukov , d. s. gorbunov and y. g. kudenko , eur .
j. c * 30 * ( 2003 ) 487 [ arxiv : hep - ph/0304146 ] .
y. g. kudenko and a. n. khotyantsev , phys .
nucl . * 63 * , 820 ( 2000 ) [ yad .
* 63 * , 890 ( 2000 ) ] ; proceedings of the international workshop on jhf science , eds . j. chiba , m. furusaka , h. miyatake and s. sawada , v. i - iii , kek proceedings 98 - 5 , august 1998 .
f. l. bezrukov , d. s. gorbunov and y. g. kudenko , phys .
d * 67 * , 091503 ( 2003 ) [ arxiv : hep - ph/0302106 ] . | we report the results of the measurement of t - violating transverse muon polarization in the decays @xmath0 ( @xmath1 ) and @xmath2 ( @xmath3 ) performed in the experiment e246 at kek .
preliminary results obtained for the entire data set taken in the period 1996 - 2000 are consistent with no t - violation in both decays . |
ucu@xmath14 is a heavy - fermion antiferromagnet ( nel temperature @xmath15 k ) @xcite . in the alloy systems
ucu@xmath1pd@xmath2 and ucu@xmath1pt@xmath2 @xmath16 is suppressed to 0 in the neighborhood of @xmath6 , beyond which point non - fermi liquid ( nfl ) behavior is observed in both systems . for ucu@xmath1pt@xmath2 nfl signatures
are strong only for @xmath17 @xcite , whereas for ucu@xmath1pd@xmath2 nfl behavior is significant over a range of concentrations ( @xmath18 ) .
this raises the possibility that nfl behavior is strong only in ucu@xmath10pt because in this alloy series , unlike ucu@xmath1pd@xmath2 , the concentration @xmath17 is a quantum critical point @xcite . in these alloys
the transverse - field @xmath0sr ( tf-@xmath0sr ) relaxation rate ( i.e. , the width of the @xmath0sr line ) is due to an inhomogeneous spread @xmath19 of muon knight shifts @xmath20 @xcite .
this spread is in turn a measure of the spread @xmath21 in magnetic susceptibility @xmath22 from the basic relation @xmath23 .
the relaxation rate becomes very large at low temperatures in a number of nfl materials including ucu@xmath1pd@xmath2 and ucu@xmath10pt @xcite .
early nmr observations of this effect in ucu@xmath1pd@xmath2 @xcite were taken as evidence for the `` kondo disorder '' model @xcite for disorder - driven nfl behavior .
this motivated a similar tf-@xmath0sr study of the ucu@xmath1pt@xmath2 alloy system .
the @xmath0sr experiments reported here were carried out using the general purpose spectrometer at the paul scherrer institute , villigen , switzerland .
measured tf-@xmath0sr rates in ucu@xmath1pt@xmath2 , @xmath3 , 1.5 , and 2.5 , and ucu@xmath10pd are shown in fig .
1 . pt@xmath2 , @xmath3 , 1.5 , and 2.5 , and ucu@xmath10pd.,scaledwidth=45.0% ] the increase of exponential relaxation rate with decreasing temperature in ucu@xmath10pt is found to be about 40% less than that in ucu@xmath10pd after correction for the difference in measuring fields ( the relaxation rate , which is due to the inhomogeneous magnetization , is proportional to field ) .
the increase is weaker still for ucu@xmath1pt@xmath2 , @xmath7 and 2.5 , where the rates are practically independent of @xmath11 .
a heuristic model @xcite is useful in understanding this low - temperature increase .
assume a curie - weiss form @xmath24 where @xmath8 is a characteristic energy ( here @xmath25 ) that is inhomogeneously distributed .
then crudely ( to first order ) a spread @xmath12 in @xmath8 yields a spread @xmath26 in @xmath22 , where @xmath27 is the spatially - averaged susceptibility .
so we expect @xmath28 if disorder in the characteristic energy produces an inhomogeneous distribution of the susceptibility .
a plot of the fractional spread in muon knight shift @xmath29 vs. @xmath27 , with temperature an implicit parameter , is a useful presentation of these results . here
@xmath30 is an effective coupling constant that describes the interaction between @xmath31-ion spins and muons taking into account the spatial correlation of the disorder @xcite .
then @xmath29 is essentially the fractional spread @xmath32 in the inhomogeneous susceptibility . according to eq .
( [ eq : bernal ] ) @xmath29 should be roughly proportional to @xmath27 , with slope @xmath33 . the results for ucu@xmath1pt@xmath2 , @xmath3 , 1.5 , and 2.5 , and ucu@xmath10pd , shown in fig . 2 ,
exhibit this behavior ( the small feature for @xmath34 is mainly due to a low - temperature anomaly in @xmath22 , probably caused by paramagnetic impurities ) . vs. spatially - averaged ( bulk ) susceptibility @xmath27 , with temperature
an implicit parameter , in ucu@xmath1pt@xmath2 , @xmath3 , 1.5 , and 2.5 , and ucu@xmath10pd.,scaledwidth=45.0% ] here @xmath35 has been calculated assuming dipolar coupling , as is the case in ucu@xmath1pd@xmath2 to a good approximation @xcite .
experimentally @xmath29 is not quite proportional to @xmath22 , which is natural given the crude linear approximation of eq .
( [ eq : bernal ] ) . the departure from proportionality at small @xmath22 ( high temperatures )
may be due to thermally - activated muon diffusion , which motionally narrows the line and reduces @xmath19 , but this mechanism is activated and can not account for the gradual and roughly proportional increase of @xmath29 with increasing @xmath27 at lower temperatures .
this behavior , which as outlined above provides evidence for a broad distribution of characteristic energies , is observed for all samples .
we observe large values of @xmath29 ( @xmath362 ) at low temperatures in both ucu@xmath10pt and ucu@xmath10pd .
since @xmath37 , where @xmath38 is the spatially - averaged characteristic energy , we have @xmath39 from eqs .
( [ eq : curie])([eq : bernal ] ) .
thus we find a broad distribution of @xmath8 ( @xmath40 ) in the `` inhomogeneous characteristic energy '' scenario described above .
such a broad distribution is essential for nfl behavior in the disorder - driven kondo disorder @xcite and griffiths - phase @xcite models , since in these pictures nfl properties are due to uncompensated spins with @xmath41 @xcite . in ucu@xmath1pt@xmath2 , @xmath7 and 2.5 , the values of @xmath29 as @xmath42 ( i.e. , the data for highest susceptibilites in fig . 2 )
are smaller than for @xmath3 .
this result agrees with thermodynamic and transport measurements @xcite , which show a return to fermi - liquid behavior for these concentrations .
in contrast , in ucu@xmath1pd@xmath2 @xmath29 at low temperatures is similar for @xmath3 and 1.5 @xcite .
the slope @xmath33 is roughly independent of @xmath11 in ucu@xmath1pt@xmath2 , and a factor of two larger than the slope in ucu@xmath1pd@xmath2 .
why is @xmath33 larger in ucu@xmath1pt@xmath2 , @xmath7 and 2.5 , if nfl effects are smaller in these alloys ?
the zero - temperature susceptibility @xmath43 is a factor of two smaller in ucu@xmath10pt than in ucu@xmath10pd , so that the characteristic energy is a factor of two larger in ucu@xmath10pt .
the values of @xmath44 are comparable ( fig .
2 ) , so that we obtain a larger @xmath12 and a larger slope @xmath33 for ucu@xmath10pt than for ucu@xmath10pd . for @xmath45 and 2.5
@xmath43 is smaller still , so that even with the larger slope @xmath44 remains relatively small .
in ucu@xmath10pd we find @xmath46 k from the slope in fig .
this together with eq .
( [ eq : lowtemp ] ) gives @xmath47 k , in good agreement with the paramagnetic curie - weiss temperature = 133 k from @xmath48 @xcite . in ucu@xmath10pt a similar analysis yields @xmath49 k and @xmath50 k , in agreement with the paramagnetic curie - weiss temperature = 200 - 300 k from @xmath48 for @xmath512.5 .
the increase of @xmath38 with increasing @xmath11 is consistent with the observed decrease of the bulk susceptibility at low temperatures @xcite . in disorder - driven nfl theories
the important parameter is @xmath44 .
our results show that the increase of @xmath38 in ucu@xmath1pt@xmath2 , both in comparison with ucu@xmath1pd@xmath2 and as @xmath52 is increased , produces a decrease in @xmath44 , and hence of nfl behavior , for ucu@xmath1pt@xmath2 .
the data are consistent with ascribing the nfl behavior of both alloy systems to disorder effects , rather than to the proximity of a quantum critical point in one alloy system but not in the other .
we are grateful to a. amato , c. baines , and d. herlach for their help during the experiments .
one of us ( d.e.m . ) wishes to thank k. ishida , y. maeno , and the quantum materials group at kyoto university , and also k. nagamine and the msl group at kek , for their hospitality during visits when part of this work was carried out .
these experiments were performed at the swiss muon source , paul scherrer institute , villigen , switzerland .
the work was supported by us nsf grants dmr-0102293 ( riverside ) , dmr-0335173 ( san diego ) , and dmr-0203524 ( los angeles ) , by the u.s .
doe , contract fg02 - 04er46105 ( san diego ) , and by the netherlands nwo and fom ( leiden ) .
work at los alamos was carried out under the auspices of the us doe .
o. o. bernal , d. e. maclaughlin , a. amato , r. feyerherm , f. n. gygax , a. schenck , r. h. heffner , l. p. le , g. j. nieuwenhuys , b. andraka , h. v. lhneysen , o. stockert , h. r. ott , phys .
b 54 ( 1996 ) 130008 .
we have not attempted a detailed fit of either of these models to the data , in view of the difficulties both models have in explaining the low - frequency thermal fluctuation spectrum responsible for longitudinal - field muon spin relaxation @xcite . | transverse - field @xmath0sr shifts and relaxation rates have been measured in the non - fermi liquid ( nfl ) alloy system ucu@xmath1pt@xmath2 , @xmath3 , 1.5 , and 2.5 . at low temperatures the fractional spread in knight shifts
@xmath4 is @xmath5 for @xmath6 , but is only half this value for @xmath7 and 2.5 . in a disorder - driven scenario where the nfl behavior is due to a
broadly distributed ( kondo or griffiths - phase cluster ) characteristic energy @xmath8 , our results indicate that @xmath9 is similar for ucu@xmath1pd@xmath2 ( @xmath6 and 1.5 ) and ucu@xmath10pt , but is reduced for ucu@xmath1pt@xmath2 , @xmath7 and 2.5 .
this reduction is due to a marked increase of @xmath8 with increasing @xmath11 ; the spread @xmath12 is found to be roughly independent of @xmath11 .
our results correlate with the observed suppression of other nfl anomalies for @xmath13 in ucu@xmath1pt@xmath2 but not in ucu@xmath1pd@xmath2 , and are further evidence for the importance of disorder in the nfl behavior of both these alloy systems .
non - fermi liquid , @xmath0sr , susceptibility inhomogeneity , ucu@xmath1pt@xmath2 , ucu@xmath1pd@xmath2 |
the evolution of galaxies can not be understood without the simultaneous consideration of the surrounding gas .
galaxies drive powerful , galactic - scale outflows , and must also accrete gas from the intergalactic medium ( igm ) .
thus the process of galaxy evolution reflects the cycling of baryons in and out of galaxies .
the importance of galactic outflows is supported by a wide range of models and observations .
an incomplete list of phenomena that may be explained by galactic outflows includes the shape of the mass - metallicity relation and the metal enrichment of the igm ( e.g. @xcite ) , the quenching of star formation in massive galaxies ( e.g. @xcite ) , and the mismatch between the galaxy luminosity function and the mass function of dark matter halos ( e.g. @xcite ) .
powerful galactic outflows must also modulate the inflow of gas into galaxies , constraining the assembly of the baryonic component and regulating star formation ( e.g. @xcite ) . in the broadest terms ,
galactic outflows are powered by energy and momentum injected into the interstellar medium ( ism ) of galaxies by star formation and/or active galactic nuclei ( agn ) .
hot ( @xmath8 k ) , metal - enriched supernova ejecta may be expelled from galaxies , entraining significant amounts of the cool ( @xmath9 k ) ism in the process .
even before massive stars explode as supernovae , radiation pressure may accelerate the cool gas to large velocities @xcite .
the detailed physics of the outflows remains controversial , however , and observational constraints on the velocity of outflows as a function of radius , the mass outflow rate , the spatial extent of outflows , and the scaling of these quantities with galaxy mass and star formation rate are needed .
galactic winds are most commonly traced via the blueshifts of interstellar absorption lines from cool , outflowing gas , seen against the backlight of the stellar continuum . in the local universe
, outflows are found in starburst galaxies , from dwarf starbursts to ulirgs ( e.g. @xcite ) .
both observations and models suggest a critical star formation rate density in order to drive outflows @xcite , and outflow velocity is observed to increase with galaxy mass and star formation rate @xcite . at higher redshifts , @xmath10 , outflows are ubiquitous , with most galaxies observed showing absorption lines blueshifted by @xmath11 - 300 km s@xmath2 ( e.g. @xcite ) .
outflows in high redshift galaxies are also traced by @xmath12 , which is seen in redshifted emission as well as blueshifted absorption ; the emission is interpreted as resonant scattering of @xmath12 photons from receding gas on the far side of the galaxy . because the high redshift studies have been restricted to a smaller range in mass and star formation rate than the local samples ( which extend to low mass dwarfs ) , trends between outflow properties and galaxy properties have been more difficult to determine
; however , recent results at @xmath13 suggest similar scalings with galaxy mass and sfr @xcite .
a major difficulty in the study of outflows is the measurement of their spatial extent , since absorption line studies of galaxies offer no information on the location of the absorbing gas .
spatial information can be obtained by measuring absorption in the spectra of background objects ( qsos or galaxies ) with lines of sight passing near galaxies .
@xcite used composite spectra of pairs of foreground and background galaxies , stacked as a function of impact parameter , to show that significant metal line absorption is seen around galaxies to distances of @xmath14 kpc , while @xmath12 absorption extends to @xmath15 kpc .
searches for star - forming counterparts to absorption systems also suggest that galaxies drive outflows to large distances @xcite , while the observation that strong absorption systems are preferentially found along the minor axis of disk galaxies is further evidence that these systems are associated with outflows @xcite .
on the other hand , not all absorption systems have detectable counterparts @xcite and the absorption has also been interpreted as being due to infalling gas @xcite , a scenario potentially supported by the fact that an increase in absorption is also seen along the major axes of disk galaxies @xcite .
the extent of outflowing gas can also be measured if galactic winds can be observed in emission .
extended @xmath16 and x - ray emission is observed in local starbursts @xcite , but such emission becomes undetectable at higher redshifts . at @xmath17 , @xcite
detect emission at a radius of @xmath18 kpc from a starburst galaxy , while @xcite see diffuse @xmath12 halos extending to @xmath19 kpc around star - forming galaxies at @xmath43 , using stacked narrow - band images .
both the and the @xmath12 photons are interpreted as arising from resonant scattering in outflowing gas . in this study
we focus on and absorption and emission in galaxies at @xmath20 .
the @xmath5 2796 , 2803 doublet is a particularly useful diagnostic because it can be observed from the ground over a wide range in redshift , from @xmath21 to @xmath4 , thus in principle offering direct comparisons of outflow properties over @xmath22 gyr of cosmic time . in practice ,
samples across this redshift range have so far differed significantly in size , selection technique , and type of galaxy observed @xcite .
the study of and absorption and emission has also benefited from recent radiative transfer modeling of these transitions .
@xcite use a variety of different galactic outflow models to predict the strengths of these lines , emphasizing the importance of photon scattering and the effect of re - emitted photons on the line profiles .
photons absorbed in resonance lines such as are re - emitted near the systemic velocity , creating underlying emission which fills in the absorption and may affect inferences about the velocity and covering fraction of outflowing gas . because most of the resonance lines are coupled to fine structure transitions , as shown in the energy level diagram in figure [ fig : energylevels ] , the photons absorbed in the transitions may instead be re - emitted in the * fine structure lines .
this paper focuses on the effect of photon scattering on measurements of absorption line velocity centroids and equivalent widths , and on * and emission and absorption as tracers of outflowing gas . the paper is organized as follows .
we describe our sample and the acquisition and analysis of the data in section [ sec : obs ] , the determination of systemic redshifts in section [ sec : zsys ] , and the determination of stellar population properties in section [ sec : sedfitting ] .
results from individual spectra are reported in section [ sec : individual ] : in section [ sec : mg2uvcolor ] we discuss emission , and section [ sec : velw_indspec ] compares the velocities and equivalent widths of the and transitions . in section [ sec : composites ]
we construct composite spectra based on stellar population properties ( section [ sec : compspecs ] ) and on the presence or absence of emission .
we discuss the velocity centroids and equivalent widths of the transitions in sections [ sec : vel_compspec ] and [ sec : w_compspec ] and the maximum observed outflow velocities in section [ sec : maxvels ] .
the remainder of the paper , section [ sec : emission ] , is devoted to the examination of * and emission . in section
[ sec : summary ] we summarize our results .
we use a cosmology with @xmath23 , @xmath24 , and @xmath25 .
the galaxies observed in this study were drawn from the survey described by @xcite .
targets were photometrically pre - selected using the rest - frame uv bm " and bx " color critieria ; the bm criteria target galaxies in the redshift range @xmath26 , and the bx criteria select galaxies primarily at @xmath27 .
we observed galaxies in the q1623 ( @xmath28:25:45 , @xmath29:47:23 ) , q1700 ( @xmath30:01:01 , @xmath31:11:58 ) and q2343 ( @xmath32:46:05 , @xmath33:49:12 ) survey fields .
l l c c c c q1623 & 02 - 04 aug 2008 & 34,200 & 7500 & 1.0 & og550
+ q1700 & 22 - 24 aug 2009 & 27,000 & 7200 & 1.2 & gg495 + q2343 & 02 aug 2008 ; 21 - 24 aug 2009 & 39,600 & 7200 & 1.2 & og550 observations were conducted with the deep imaging multi - object spectrograph ( deimos ; @xcite ) on the keck ii telescope in august 2008 and august 2009 .
deimos is a medium resolution optical spectrometer with spectral coverage from @xmath7 4100 @xmath34 11,000 ; this range includes rest - frame near - uv features from @xmath35 to [ ] @xmath36 in galaxies with @xmath37 . with the use of a slitmask
, the 16.7 @xmath38 5.0 field of view allows the simultaneous observation of up to @xmath11 galaxies .
we observed one mask in each of the three fields , covering a total of 147 galaxies .
galaxies were selected for the masks based on the following priorities : 1 ) known redshift ( from the previous far - uv spectroscopy of @xcite ) in the range @xmath39 ; 2 ) the bm ( as opposed to bx ) color criteria , which target the redshift range for which features of interest fall within the spectral co verage ; and 3 ) brighter galaxies with @xmath40 . we used the 600 l / mm grating , which gives a dispersion of 0.65 / pixel .
exposure times ranged from 7.5 to 11 hrs , with average seeing fwhm @xmath41 .
details of the observations are given in table [ tab : obs ] .
most of the data reduction was done with the deep2 data processing pipeline @xcite , which produces one - dimensional air wavelength - calibrated galaxy , inverse variance and sky spectra .
we then removed the instrumental signature from the data by putting the galaxy and error spectra on a relative flux scale with the iraf software package , using standard stars observed in the same manner as the galaxies .
we also removed the strong telluric absorption bands due to molecular oxygen at @xmath42 and @xmath43 .
this correction is usually done by scaling the strength of the telluric absorption seen in a standard star spectrum to the absorption strength seen in an object s spectrum .
because of the long integration times used in this study , however , the galaxies were observed over a range of airmasses , making a single standard star observation taken at a single air mass a poor match to the telluric feature strength in the individual galaxy spectra .
a better measurement of the shape of the telluric features can be obtained from an average observed - frame galaxy spectrum for each field ( because the galaxies lie at different redshifts , combining in the observed frame averages over the galaxy properties but gives a good measurement of the sky features ) .
we used these average spectra to create telluric feature templates for each field .
each average galaxy spectrum was normalized and smoothed , and then portions of the spectrum which were outside of the telluric bands were forced to have a normalized flux value of one , leaving the telluric bands as the only deviations from continuum .
the resulting spectrum served as the template telluric spectrum for each field .
this template was then divided into each of the galaxy and error spectra .
finally , the individual spectra were converted to a vacuum wavelength scale and rebinned by a factor of two to increase s / n ;
the 600 l / mm grating is deimos lowest resolution mode and so results in oversampled data .
the final pixel scale is 1.25 per pixel in the observed frame , and the spectral resolution ( measured from the widths of the night sky lines at @xmath43 ) is @xmath44 ( fwhm ) or @xmath45 km s@xmath2 in the q1623 field , which was observed with 1 slits , and @xmath46 or @xmath47 km s@xmath2 in the q1700 and q2343 fields , which were observed with 1.2 slits .
we measured redshifts @xmath48 for 96 of the 147 galaxies observed , either from [ ] @xmath53727 , 3729 emission or and absorption as discussed in section [ sec : zsys ] below ( the remaining 51 galaxies were mostly among the fainter objects on the masks , and most had spectra too noisy to enable redshift determination ; 4 galaxies had @xmath49 ) . the analysis presented here is based on this sample of 96 galaxies . after measuring redshifts ,
the 96 spectra were normalized using the procedure outlined by @xcite .
normalization was done only in the rest - frame wavelength range 22003000 ; this range contains the and features of interest , and increasing sky noise at redder wavelengths makes normalization more difficult .
we shifted the spectra into the rest frame , identified regions of the spectrum which contained flux solely from the continuum , and fit a spline curve to both the mean wavelength and mean flux values within these defined continuum windows .
the same windows were used for each galaxy , with adjustment on an individual basis to avoid the gap between detectors or noise spikes . each galaxy spectrum
was then divided by the spline fit to its continuum .
the uncertainty in the continuum fits is @xmath50 for most galaxies .
these continuum normalized spectra were used for the remainder of the analysis .
we also make use of multi - wavelength imaging in order to model the spectral energy distributions of the galaxies .
the optical images , in the @xmath51 , @xmath52 and @xmath53 filters , are described by @xcite , the near - ir ( @xmath54 and @xmath55 ) images by @xcite , and the mid - ir _ spitzer _ irac data by @xcite and @xcite .
the systemic redshift of a galaxy ( i.e. , the redshift of its stars ) serves as the velocity zero - point for studying the kinematics of galactic and circumgalactic gas , and is used to align galaxies in the rest - frame for the construction of composite spectra , as described in section [ sec : composites ] . a reliable determination of the systemic redshift is therefore required for this study .
the preferred method of redshift determination , measurement of the wavelengths of strong nebular emission lines from regions , becomes increasingly difficult for galaxies with @xmath56 , as these lines redshift into the near - ir .
the centers of stellar photospheric absorption lines can also be used , but while these features are located at accessible wavelengths , they are broad and shallow and therefore hard to identify at the s / n of most high redshift galaxy spectra . if none of these features are available , the systemic redshift can be estimated from the wavelengths of the interstellar absorption lines .
in this study we use the first and third of the above methods . for 51 of the 96 galaxies in the sample , strong [ ] @xmath53727 , 3729 emission is included at the red end of the spectral region covered , allowing a precise determination of the systemic redshift from a double gaussian fit to the [ ] doublet , which is usually resolved .
these 51 galaxies have a mean redshift @xmath57 . for the remaining 45 galaxies in the sample ,
we turn to the interstellar absorption lines .
it is well - documented that these features are blueshifted with respect to the systemic redshift and that , while the degree of blueshifting varies for individual galaxies , the assumption of an average velocity offset between the systemic and interstellar absorption redshifts allows the construction of composite spectra which are effectively at zero velocity @xcite .
we follow the method of @xcite , who use a sample of star - forming galaxies at @xmath58 with both interstellar absorption lines and systemic redshifts from @xmath16 emission to find that , on average , the interstellar absorption lines are offset from the systemic redshift by @xmath59 km s@xmath2 .
we use the subsample of galaxies with measurements of both [ ] emission and @xmath602374 absorption to derive the average offset between nebular emission and interstellar absorption for our sample ( as discussed in sections [ sec : individual ] and [ sec : composites ] , this transition is expected to be least affected by the presence of emission , and therefore most representative of the kinematics of the outflow ) .
as described in more detail in section [ sec : individual ] , absorption line centroids are measured from a single gaussian fit using the idl routine _ mpfit.pro_ , which measures the center and equivalent width of each feature and their associated uncertainties . because [ ] and this transition are at opposite ends of the accessible wavelength range , there are 20 galaxies whose spectra include significant detections of both features ; the average uncertainty in the velocity measurements of this sample is 41 km s@xmath2 .
the relationship between the [ ] -defined systemic redshift and the redshift of the interstellar absorption is given by @xmath61 where @xmath62 km s@xmath2 is the average velocity offset ( we compute an unweighted mean to avoid the possibility of biasing the result toward brighter galaxies , which may also be more massive )
. the standard deviation is 87 km s@xmath2 , reflecting the significant scatter among individual measurements , while the error in the mean is 10 km s@xmath2 . for the higher s
/ n spectra with systemic redshifts from [ ] emission , we also measure velocity offsets through a joint fit to the @xmath602344 , 2374 and 2587 lines ( for more details , see @xcite ) .
the mean absorption line velocity measured in this way is @xmath63
km s@xmath2 with a standard deviation of 80 km s@xmath2 , in good agreement with the value of @xmath64 km s@xmath2 measured from @xmath602374 alone ; the joint and @xmath602374 velocities agree within 2@xmath6 in all cases , and within 1@xmath6 in half of the measured sample .
the magnitude of the velocity offset @xmath65 is a factor of @xmath66 smaller than the value of @xmath59 km s@xmath2 ( mean and standard deviation ) determined by @xcite .
we plot @xmath65 against redshift for the current sample and for the higher redshift sample of @xcite in figure [ fig : deltav_z ] ( see also figure 2 of @xcite ) , clearly showing the smaller offset from systemic velocity in the lower redshift sample .
the horizontal dotted line divides the @xcite sample into two redshift bins containing an equal number of galaxies , and the large light blue points show the mean velocity offset and redshift of each bin ; although the scatter is considerable , the average offset from systemic increases with increasing redshift , from @xmath67 km s@xmath2 at @xmath68 to @xmath69 km s@xmath2 at @xmath70 .
this increase is apparently driven by a decreasing fraction of objects with velocities near zero , as well as a higher fraction with @xmath71 km s@xmath2 .
our lower redshift sample follows the same trend of decreasing velocity centroid with decreasing redshift .
a full explanation of this trend requires more careful accounting of selection effects , however ; our galaxies are on average somewhat lower in mass than the @xcite sample , since the magnitude limit of the survey reaches intrinsically fainter objects at lower redshifts .
velocity offsets in the @xcite sample are measured from the far - uv interstellar absorption lines @xmath72 , @xmath73 and @xmath74 , whereas we use the near - uv lines .
the kinematics of these different lines have not been directly compared , although , and are expected to be similar as all of these transitions arise in cool outflowing gas , and @xcite find the kinematics of to be similar to those of the low ionization lines .
finally , a decrease in the blueshift of the velocity centroid does not necessarily correspond to a decrease in the outflow velocity , since the centroid may be influenced by the interstellar medium of the galaxy or by underlying emission near zero velocity .
for the 45 galaxies in the sample without [ ] emission , we use equation [ eq : zsys ] to determine the systemic redshift . where possible , we use @xmath602374 to define the absorption redshift , and when that feature is unavailable , we use @xmath602587 ( the choice of these two lines
is motivated by the fact that they are least likely to be affected by emission filling , which may shift the wavelength to the blue ) . in the 14 galaxies where these two features were not detected
, we use the far - uv absorption redshift from prior keck / lris spectroscopy @xcite .
the mean systemic redshift of the sample of 45 galaxies determined in this way is then @xmath75 .
given the possible trend of @xmath65 with redshift , there is some concern that the galaxies for which we apply equation [ eq : zsys ] lie at somewhat higher redshifts than the galaxies used to determine the equation ; however , the similar velocity offsets of our sample and the lower redshift half of the @xcite sample suggest that this is not a large effect . it is desirable to verify systemic redshifts obtained in this way via measurement of weak stellar or nebular features in composite spectra ; if systemic redshifts have been assigned correctly , these features should lie at zero velocity ( e.g. @xcite ) .
unfortunately our spectra do not contain features ideal for this purpose .
the strongest nebular emission line , ] @xmath76 , is a blend of several closely separated transitions whose ratios depend on the electron density , making it unsuitable for precise redshift determinations .
the composite spectrum of all 96 galaxies does contain a weak detection of the @xmath77 stellar photospheric line ( vacuum rest wavelength 2297.579 ) , used by @xcite to measure the systemic redshift of the lensed galaxy ms 1512-cb58 .
the low s / n of this line makes it of limited use , but its velocity in the composite spectrum is @xmath78 km s@xmath2 , consistent with zero .
we can also check the systemic redshifts by comparing the composite spectrum of the 51 galaxies with [ ] -based redshifts with that of the 45 galaxies with redshifts determined using equation [ eq : zsys ] .
for this test we compare the centroid and width of * @xmath79 , the strongest emission line appearing in both spectra ; we find that the line centroids in the two spectra agree to within 20 km s@xmath2 , while the line is @xmath80 ( 90 km s@xmath2 ) broader in the spectrum constructed using redshifts from equation [ eq : zsys ] .
such broadening is expected given the 87 km s@xmath2 dispersion in @xmath65 .
these tests validate our use of equation [ eq : zsys ] to determine the systemic redshifts of galaxies without [ ] emission .
for 71 of the 96 galaxies in the sample , we determine the stellar mass , age , extinction and star formation rate through modeling of the broadband spectral energy distribution ( sed ) .
we require photometry in the @xmath55 and/or _ spitzer _ irac 3.6 or 4.5 bands in order to perform the modeling , and most of the remaining 25 galaxies are not covered by this ir imaging ; we do not expect these 25 galaxies to be significantly different from the rest of the sample .
we model the sed as described by @xcite and @xcite ; briefly , we use the @xcite stellar population models , and assume constant star formation , the @xcite extinction law , and a @xcite initial mass function .
uncertainties in the fitted parameters are determined by monte carlo simulations which perturb the input photometry according to the photometric uncertainties and determine the best - fit sed for the perturbed colors ; typical fractional uncertainties are @xmath81 , 0.4 , 0.5 , and 0.2 in @xmath82 , age , sfr , and stellar mass , respectively .
systematic uncertainties are discussed further in section [ sec : compspecs ] .
we assume constant star formation histories because of the difficulty of obtaining robust constraints on the star formation history of galaxies from sed fitting alone ; fortunately , the stellar mass is relatively insensitive to the assumed star formation history @xcite .
the range of stellar population parameters for the 71 galaxies is shown in figure [ fig : sedhists ] .
we find a mean ( median ) stellar mass @xmath83 @xmath84(@xmath85 @xmath84 ) , mean ( median ) reddening @xmath86 ( 0.21 ) , mean ( median ) age 550 myr ( 360 myr ) , and mean ( median ) sfr 54 @xmath87 ( 22 @xmath87 ) . the degeneracies between age , extinction and star formation history involved in such modeling are well - known , as is the fact that the stellar mass is the most well - determined parameter ( e.g. @xcite ) .
we discuss the relationships between the modeled quantities further in section [ sec : composites ] , where we construct composite spectra based on stellar population properties .
representative examples of the individual spectra are shown in figures [ fig : ind_fespecs ] and [ fig : mg2profiles ] .
figure [ fig : ind_fespecs ] shows the two complexes of absorption and emission lines , at @xmath88 and @xmath89 ; the top two spectra are among the highest s / n in the sample , while the spectrum of q2343-bx410 in the bottom panel is more representative of the typical data quality .
we also show the @xmath52796 , 2803 profiles for these four galaxies and four additional objects in figure [ fig : mg2profiles ] . while the absorption lines show some variation in their strengths , we see much greater diversity in the profiles , which range from strong emission with very little absorption ( q2343-bx256 ) to pure absorption ( q2343-bx410 ) ; the pattern is suggestive of that seen in the far - uv spectra of high redshift galaxies , in which @xmath12 may appear in emission or absorption while the interstellar resonance lines appear in absorption .
we study the relationships between the strengths of these lines and the properties of the galaxies in this section and in the section following .
the features covered by the spectra are listed in table [ tab : lines ] .
many of these , particularly the weaker emission lines , are significantly detected only in the composite spectra discussed in the next section .
we measure the absorption and emission features by fitting a single gaussian profile to each line using the idl routine _ mpfit.pro_ ; given a spectrum and the error spectrum containing the pixel - to - pixel uncertainties , this routine measures the center and equivalent width of each line and their associated uncertainties .
although absorption lines associated with galactic outflows generally show asymmetric profiles
( e.g. @xcite ) , the s / n of the individual spectra in our sample is such that weak blue wings of the absorption lines are generally undetected and a gaussian provides a good fit .
we do not attempt to fit outflowing and zero - velocity components separately , as some similar studies have done ( e.g. @xcite ) .
l l l ] & 2326 & @xmath90 & ... & emission ; blend + [ ] & 2470.97 & @xmath91 & ... & emission ; blended with [ ] @xmath602471 + & 2471.09 & @xmath92 & ... & emission ; blended with [ ] @xmath602470 + & 2852.96 & @xmath93 & 1.83&absorption + & 2796.35 & @xmath94 & 0.6155 & absorption and emission + & 2803.53 & @xmath95 & 0.3058 & absorption and emission + feii & 2249.88 & ... & 0.00182&absorption + & 2260.78 & ... & 0.00244&absorption + & 2344.21 & @xmath96 & 0.114&absorption + & 2365.55 & @xmath97 & 0.0495 & emission + & 2374.46 & @xmath98 & 0.0313&absorption + & 2382.76 & @xmath99 & 0.320&absorption + & 2396.36 & @xmath100 & 0.279 & emission + & 2586.65 & @xmath101 & 0.0691&absorption + & 2600.17 & @xmath102 & 0.239&absorption + & 2612.11 & @xmath103 & 0.122 & emission + & 2626.45 & @xmath104 & 0.0455 & emission + & 2632.11 & @xmath105 & 0.087&emission we have seen from the spectra shown in figure [ fig : mg2profiles ] that the galaxies in our sample show a wide range in @xmath52796 , 2803 profiles , from strong emission to pure absorption . in order to separate the sample into galaxies with emission and galaxies without , we classify as emitters those galaxies with at least two adjacent pixels at least 1.5@xmath6 above the continuum level in either of the two transitions . with this classification , 33 of the 96 galaxies in the sample show emission , while 63 do not .
the classification clearly depends on the s / n of the individual spectra as well as the intrinsic strength of emission . in figure
[ fig : beta_mg2 ] we plot the rest - frame uv continuum slope @xmath106 ( @xmath107 ) against @xmath108 , the absolute magnitude at 1900 , for galaxies with and without emission , using larger symbols for galaxies with higher s / n in the region of the spectrum .
we measure @xmath106 from the @xmath109 color , which is uncorrelated with redshift for the galaxies in our sample . at the sample
s mean redshift of @xmath110 , the @xmath52 filter is centered at a rest wavelength of 1860 , and the @xmath53 filter lies at 2665 .
we calculate the absolute magnitude @xmath108 at @xmath111 , using the @xmath52 magnitude for @xmath112 , and composite magnitudes @xmath113 and @xmath114 for @xmath115 and @xmath116 respectively .
composite magnitudes are determined by averaging the flux in the two relevant filters .
it is clear that galaxies with significant emission tend to have bluer uv slopes .
quantitatively , 17 of the 33 galaxies with emission are found in the bluest third of the sample , with @xmath117 , while the remaining 16 are equally distributed between the two redder thirds .
a two - sample k - s test finds that the probability that the emitters and non - emitters are drawn from the same @xmath106 distribution is 0.025 .
this conclusion is somewhat strengthened when we consider only the 52 galaxies with @xmath118 in the region ; among this subsample , the 21 emitters have @xmath119 ( where 0.35 is the standard deviation ) , while the 31 non - emitters have @xmath120 .
histograms of these two subsamples are shown in the inset at the upper right of figure [ fig : beta_mg2 ] . repeating the k - s
test indicates that the probability that the two samples are drawn from the same distribution is 0.0097 .
in contrast to the @xmath106 distributions , the emitting and non - emitting samples are similarly distributed in uv luminosity .
emission was also identified in a sample of @xmath121 @xmath13 galaxies by @xcite , using criteria which select an excess of emission above the continuum level in a window between the two absorption lines .
the excess emission objects selected in this way constituted @xmath122% of the sample , and were attributed to narrow - line agn .
@xcite found that these excess emission objects had bluer @xmath123 colors and ( in contrast to our results ) were more luminous .
they also found that the excess emission was usually at the systemic redshift and not always accompanied by absorption .
our fraction of emitters is considerably higher than that of @xcite , although the different selection criteria make a direct comparison difficult .
we attribute the emission to resonant scattering in outflowing gas ; as shown in figure [ fig : mg2profiles ] , the emission is generally redshifted and accompanied by blueshifted absorption , the classic p cygni profile of an outflow .
the p cyngi profile is also seen in @xmath12 emission and absorption in galaxies at high redshift , and the relationship between uv color and @xmath12 emission is similar to the relationship between uv color and emission : galaxies with stronger @xmath12 emission are observed to have bluer uv slopes @xcite .
we discuss the relationship between emission and galaxy properties further in section [ sec : composites ] . as discussed in section [ sec : intro ] , most of the absorption lines are coupled to fine structure transitions , such that a photon absorbed at ( for example ) 2587 can be emitted in the * @xmath602612 or @xmath602632 emission lines .
the exception is @xmath124 ; the only allowed transition for a photon absorbed in this line is a return to the ground state .
thus , as discussed in detail by @xcite , one may expect significant rescattered emission to fill in " the absorption in this and other resonant transitions ; one may also expect the filling to be greater when the photon has no other means to return to a lower energy level , i.e. for transitions such as @xmath124 which are not coupled to fine structure emission lines .
evidence for this emission filling may be seen in the form of blueshifted velocity centroids , since the scattered emission is expected to occur near systemic velocity , or in a decrease in absorption equivalent width .
either of these effects may influence conclusions drawn regarding the kinematics , optical depth and covering fraction of outflowing gas .
the remainder of this section is dedicated to assessing the effect of this emission filling on measurements of velocity centroids and equivalent widths . in figure
[ fig : ind_velcomp ] we compare the velocities with respect to systemic for various transitions , as measured by the center of the gaussian fit .
we plot only galaxies with velocity uncertainties @xmath125 km s@xmath2 ( an absolute uncertainty is preferable to some threshold in @xmath126 in order to avoid biasing the sample away from galaxies with @xmath127 near zero ) .
because this measurement relies on a precise determination of the systemic redshift for each object , we include only galaxies for which we have determined @xmath128 from [ ] emission .
we plot the velocity centroids of @xmath124 and @xmath129 absorption , @xmath602796 absorption , and * @xmath130 emission , all against that of @xmath131 .
as discussed above , we may expect @xmath124 to show a larger blueshift than @xmath131 or @xmath129 , and this is indeed observed in the upper two panels of figure [ fig : ind_velcomp ] : on average , @xmath131 and @xmath129 show identical blueshifts of @xmath132 km s@xmath2 , while the average relative shifts of @xmath131 and @xmath124 are @xmath133 and @xmath134 km s@xmath2 respectively ( the two averages for @xmath131 are not identical because we include only galaxies with @xmath125 km s@xmath2 for both lines in each plot , and the two subsamples are slightly different ) .
we also compare the relative velocity shifts of @xmath131 and @xmath602796 , in the lower left panel of figure [ fig : ind_velcomp ] .
because both @xmath124 and @xmath602796 are resonance transitions coupled only to the ground state , we might expect emission filling to affect them similarly ; however , as also suggested by the diversity of profiles relative to in figures [ fig : ind_fespecs ] and [ fig : mg2profiles ] , the transition shows stronger evidence for the presence of emission .
the average blueshift of is @xmath135 km s@xmath2 , compared to the @xmath134 km s@xmath2 of @xmath124 . in the case of both lines
, the velocity comparison shows that only a fraction of the sample shows evidence of emission underlying the absorption ; about @xmath136 of the @xmath124 sample is significantly offset from the @xmath131 absorption , while the fraction of the sample showing an offset in @xmath602796 is higher at @xmath137 .
finally , we compare the velocity centroids of @xmath131 with those of the strongest fine structure emission line , * @xmath130 .
we discuss the origin of the fine structure emission lines in more detail in the following sections ; for now , we note that @xcite have proposed that * emission arises from photon scattering in the outflowing gas and should be observed at or near systemic velocity , assuming that the outflow is approximately isotropic and has little extinction . with a mean velocity centroid of @xmath138 km s@xmath2 ,
our observations are in general agreement with this scenario .
we turn next to a comparison of the equivalent widths of the absorption lines , measured by integrating the best - fit gaussian profiles . for an optically thin transition ,
the equivalent width @xmath139 of an absorption line depends on the covering fraction and column density @xmath140 of the ion and the oscillator strength of the transition @xmath141 such that @xmath142 .
when the transition becomes optically thick , the equivalent width depends instead on the velocity dispersion and covering fraction of the absorbing gas , and we then expect ( in the absence of emission filling ) different transitions of the same ion to have the same equivalent width .
it is therefore instructive to compare the equivalent widths of some of the observed absorption lines . in figure
[ fig : ind_ewcomp ] we compare the equivalent widths of @xmath602374 and @xmath602383 . with oscillator strengths of @xmath143 and @xmath144
respectively , these two lines should have very different equivalent widths in the optically thin case , with @xmath145 . as the absorbing gas becomes optically thick
, we expect @xmath146 .
an additional effect is the possible presence of emission filling : as discussed above , the two transitions differ in that @xmath131 is coupled to a fine structure transition , while @xmath124 is not , and we therefore expect that @xmath124 is more likely to be affected by the presence of emission underlying the absorption line
. such underlying emission would lead to a decrease in equivalent width , producing , in the optically thick case , otherwise unphysical line ratios of @xmath147 .
figure [ fig : ind_ewcomp ] shows that in most cases @xmath148 is consistent with unity or falls between the saturated and unsaturated limits . however , there are four objects ( q1623-bm95 , q1700-bm154 , q1700-bx1043 and q2343-bx364 , shown with large red circles in figure [ fig : ind_ewcomp ] ) for which the equivalent width ratio of the two lines can only be explained by a decreased @xmath124 equivalent width due to underlying emission . note that this represents only a lower limit on the number of objects affected by emission filling ; if the intrinsic line ratio falls between the unsaturated and saturated limits , as is the case for many of the galaxies in the sample , the equivalent width comparison can not distinguish between emission filling and an increase in optical depth .
there are also several additional objects that fall in the unshaded region with lower significance , so there may be additional galaxies in the sample with @xmath147 . as shown in figure [ fig : mg2profiles ]
, q1623-bm95 and q1700-bm154 also have strong emission ; q1700-bx1043 and q2343-bx364 ( not shown in figure [ fig : mg2profiles ] ) have lower s / n in the region , making the presence of emission more difficult to assess , but using the criterion described in section [ sec : mg2uvcolor ] bx1043 is also an emitter while bx364 is not .
we also note that the stellar masses of q1623-bm95 , q1700-bm154 , q1700-bx1043 and q2343-bx364 are @xmath149 @xmath84 , @xmath150 @xmath84 , @xmath151 @xmath84 and @xmath152 @xmath84 respectively , all lower than the average stellar mass of @xmath153 @xmath84 .
this small sample suggests that galaxies with strong emission are more likely to exhibit emission filling in the transitions as well , and that such emission filling may be more likely in low mass galaxies .
the fraction of galaxies significantly affected by this emission is relatively small , however , particularly for the lines .
in order to examine weak emission and absorption lines and study the variation of spectral features with galaxy properties more closely , we construct composite spectra based on the modeled stellar population properties and on the presence or absence of emission .
we also create a composite spectrum of all 96 galaxies in the sample .
the stacked spectra represent pixel - by - pixel averages of the normalized individual spectra , created with 3@xmath6 rejection .
the mean and standard deviation of the flux values at a given wavelength step were calculated , pixels with flux values @xmath154 were rejected , and the mean flux value of the remaining pixels was output as the final composite spectrum . because this was done for each pixel , the individual spectra contributing to the mean flux at a given pixel vary . the error spectrum for each composite
was computed by adding in quadrature the error spectra for the individual galaxies contributing to the mean flux at each wavelength step .
the final stacked spectra have 0.25 per pixel in the rest frame .
the composite spectrum of the full sample of 96 galaxies is shown in figure [ fig : all_fullspec ] , with the features listed in table [ tab : lines ] marked with vertical black lines .
in addition to the strong and absorption lines , we clearly detect five * emission lines and the broad blend of nebular ] emission at 2326 .
figure [ fig : all_fullspec ] also shows ( overplotted in orange ) the synthetic , normalized stellar spectrum of an integrated stellar population with constant star formation , smoothed to a resolution of @xmath155 km s@xmath2 ( fwhm ) to match the resolution of our observations .
the spectrum was produced by the stellar synthesis code starburst99 @xcite , and incorporates the library of theoretical uv spectra of massive stars described by @xcite .
the stellar spectrum assumes solar metallicity , an age of 100 myr ( the uv spectrum stabilizes after @xmath156 myr under conditions of constant star formation ) and a @xcite initial mass function .
this spectrum shows that the stellar contribution to the features of interest for the current study is negligible ; for constant star formation , the stellar continuum at near - uv wavelengths is dominated by stars too hot to have strong or absorption . as described in section [ sec : sedfitting ] , we model the broadband seds of 71 of the 96 galaxies in the sample to determine their stellar masses , ages , star formation rates and reddening .
we divide the 71 galaxies into three bins based on each of these four parameters and construct composite spectra of the galaxies in the highest and lowest bins , resulting in 8 composite spectra , each constructed from 24 individual spectra .
the mean stellar population properties of the galaxies in each composite are listed in table [ tab : sedcomps ] , along with the mean s / n per pixel of each composite spectrum .
l c c c c c c high mass & 24 & 14 & 24.3 & 0.20 & 28 & 1119 + low mass & 24 & 11 & 1.9 & 0.25 & 54 & 133 + high e(b - v ) & 24 & 12 & 12.0 & 0.31 & 118 & 363 + low e(b - v ) & 24 & 15 & 7.7 & 0.14 & 18 & 589 + high sfr & 24 & 15 & 9.5 & 0.28 & 128 & 205 + low sfr & 24 & 11 & 8.1 & 0.17 & 10 & 768 + high age & 24 & 11 & 21.7 & 0.18 & 16 & 1260 + low age & 24 & 12 & 2.7 & 0.28 & 118 & 55 there is , of course , considerable overlap between the composites because of covariance in the fitted parameters .
we quantify this overlap in table [ tab : overlap ] , in order to clarify the relationships between the stellar population properties considered .
this table shows the number of objects in common between any two composite spectra .
the largest overlap is between mass and age ; because the models assume constant star formation , the stellar mass is simply the product of the age and star formation rate .
as shown in the table , the mass and age composites have 18/24 objects in common in the massive , old sample and 16/24 in the low mass and young sample .
l c c c c c c c c low mass & ... & ... & 18 & 1 & 8 & 10 & 9 & 9 + high mass & ... & ... & 0 & 16 & 9 & 6 & 6 & 7 + low age & 18 & 0 & ... & ... & 5 & 15 & 3 & 15 + high age & 1 & 16 & ... & ... & 12 & 5 & 13 & 2 + low e(b - v ) & 8 & 9 & 5 & 12 & ... & ... & 13 & 3 + high e(b - v ) & 10 & 6 & 15 & 5 & ... & ... & 2 & 15 + low sfr & 9 & 6 & 3 & 13 & 13 & 2 & ... & ... + high sfr & 9 & 7 & 15 & 2 & 3 & 15 & ... & ... recent results indicate that the use of the @xcite extinction law for galaxies with young best - fit ages overestimates their star formation rates with respect to other indicators ; these galaxies may have less extinction than is suggested by their red uv slopes and be better modeled with a steeper , smc - like extinction law @xcite .
the result is that @xmath82 and the sfr may be overestimated for the youngest galaxies in the sample ( and because of the covariance between extinction and age in the modeling , the ages of these objects are also uncertain ) .
this systematic uncertainty primarily affects galaxies with best - fit ages ( calculated with models assuming the calzetti extinction law ) less than 50 myr , of which there are 16 in our sample of 71 objects with sed fits .
@xcite find that restricting the ages to be at least 50 myr ( approximately equal to the dynamical time ) and using the smc extinction law for such objects brings the sed - determined sfrs into agreement with sfrs determined from the combined unobscured uv + ir luminosities .
the effect of these changes can be significant ; the typical result is to decrease the best - fit @xmath82 by @xmath157 mag , and reduce the sfr by a factor of @xmath22 . because of these uncertainties , in the following discussions we give greatest weight to the composite spectra based on stellar mass .
we have also constructed composite spectra of the 33 galaxies with and 63 galaxies without emission , determined as described in section [ sec : mg2uvcolor ] .
table [ tab : sedcomps ] does not report mean stellar population properties for these composites , since not all of the galaxies have photometry sufficient for sed modeling ; however , the composites with and without emission have mean s / n per pixel of 16 .
the composite spectra based on the stellar population properties and on emission are shown in figure [ fig : allcomposites ] .
equivalent widths and velocity centroids are measured by direct integration of the flux above or below the continuum level , with @xmath158 uncertainties determined by monte carlo simulations in which we perturb each pixel of the original composite spectrum by an amount dependent on its uncertainty 5000 times to create 5000 artificial spectra .
the resulting velocity measurements are flux - weighted centroids , measured over a typical range of @xmath159 km s@xmath2 for the absorption lines and @xmath160 km s@xmath2 for the emission lines .
the uncertainties adopted for @xmath161 and velocity are the standard deviation of the 5000 measurements of these values made for each feature .
measurements of the equivalent widths and velocity centroids of the absorption features are given in table [ tab : abs ] , and of the emission features in table [ tab : em ] .
although we measured an average velocity shift of @xmath64 km s@xmath2 relative to systemic for the @xmath131 absorption line and then applied this shift to galaxies without [ ] detections in order to construct the composites , the velocity centroid of feii @xmath131 in the composite spectrum of all galaxies is @xmath162 km s@xmath2 .
this difference is due to the asymmetric absorption line profiles of the higher s / n composite , in which an extended blue tail of absorption shifts the centroids to larger velocities when they are measured with direct integration rather than gaussian fits which assume symmetry .
the peak optical depth of the line is consistent with the @xmath64 km s@xmath2 measurements from individual spectra .
the most striking feature of the composite spectra shown in figure [ fig : allcomposites ] is the difference in profile , particularly for the high and low mass composites .
the lowest mass galaxies have strong emission and weak absorption , while the most massive galaxies have stronger absorption and no apparent emission .
galaxies with young ages , low dust content , and ( marginally ) lower sfrs also show stronger emission , although the differences are not as striking as in the composites based on stellar mass .
similar trends were seen by @xcite , who saw increased emission in composite spectra of lower mass galaxies and galaxies with lower sfrs ( although @xcite removed objects with excess emission from their sample , additional emission became apparent when the symmetric , zero - velocity absorption component was removed from the profile ; we do not attempt such decomposition here ) .
the composite spectra also show more subtle differences in the strengths of the other absorption and emission features .
we discuss these differences below .
an inspection of the absorption line measurements in table [ tab : abs ] shows that the most consistent differences between the composite spectra arise when galaxies are divided based on stellar mass .
all of the lines are blueshifted to larger velocities in the high mass spectrum relative to the low mass , and most of the lines have higher equivalent widths in the high mass galaxies as well .
this larger blueshift suggests that more massive galaxies drive faster outflows , although we caution that the velocity centroid of the line is a poor proxy for outflow velocity because it may be influenced by absorption from the interstellar medium of the galaxy or by underlying emission .
we discuss the question of outflow velocity in more detail in section [ sec : maxvels ] . in figure
[ fig : compvels ] we compare the velocity centroids of several of the lines in the various composite spectra .
this figure should be compared with figure [ fig : ind_velcomp ] , in which we made the same comparisons for the individual spectra and found some evidence for the presence of underlying emission ( in @xmath136 of the sample ) in the form of blueshifted @xmath124 velocities relative to @xmath131 and @xmath129 ; recall that the latter two lines are coupled to fine structure transitions and therefore may be less affected by emission filling .
this signature blueshift of emission filling is less apparent in the composites .
we see generally good agreement between all three of the absorption lines , though the scatter is significant , particularly in the comparison of feii @xmath131 and feii @xmath129 .
we conclude that although underlying emission may affect the velocity centroids of some lines in some galaxies , the effect is not strong enough to be robustly detected in composite spectra .
this may be because the fraction of galaxies so affected is relatively small , or because the higher s / n of the composites in combination with the method of direct integration makes the measurements of the composites more sensitive to absorption at a wider range of velocities , thereby decreasing the effect of underlying emission near zero .
measurements of the individual spectra showed that the velocity centroid of @xmath163 is more affected by underlying emission than any of the lines , and this is also true of the composites .
in contrast to the lines studied , is blueshifted with respect to @xmath131 in all of the composite spectra .
this shift is probably due to underlying emission .
the largest velocity difference is observed in the composites of galaxies with and without emission : the average blueshift of @xmath163 in galaxies with no detectable emission is @xmath164 km s@xmath2 , while the average shift of those with emission is @xmath165 km s@xmath2 .
we observed above that low mass galaxies show stronger emission , and that the emission is somewhat stronger in galaxies with young ages , low dust content , and lower sfrs ( all shown in blue in figure [ fig : compvels ] ) as well .
figure [ fig : compvels ] suggests that this increased emission in low mass galaxies is apparent in the velocity of absorption as well , since the blue points tend to be more displaced than the red points from the line of equal velocities .
finally , we compare the velocity centroid of the * @xmath130 emission line with that of @xmath131 in the composites ( we note that the velocity of the * @xmath130 line is not affected by the nearby * @xmath166 line because the two lines are well - separated , as can be seen from figure [ fig : allcomposites ] ) . the individual spectra showed that the velocity of * @xmath130 scattered around zero , with a marginal tendency for greater blueshifts ; in the composities we see consistently blueshifted velocity centroids of about @xmath167 km s@xmath2 . as an additional check on this line , we measure the velocity offset of * @xmath130 in the composite spectrum of the 51 galaxies with [ ] redshifts and find it to be consistent with the other composites at @xmath168 km s@xmath2 .
the difference between the individual spectra and the composites may again be due to greater sensitivity to weak blueshifted emission in the composites ; see the discussion of the fine structure emission lines in section [ sec : emission ] below . in any case
, as with the individual spectra , * @xmath130 is on average observed at velocities closer to systemic than any of the absorption lines .
next we compare the equivalent widths of the @xmath131 and @xmath124 absorption lines in the composites , as we did for the individual spectra in section [ sec : velw_indspec ] and figure [ fig : ind_ewcomp ] . in the case of the individual spectra
, we saw that , with the exception of four low - mass galaxies , all of the objects in the sample have an equivalent width ratio consistent with the range expected based on the variation of optical depth , @xmath169 .
the ratio of equivalent width for all of the composites falls within this range as well , as shown in figure [ fig : comp_ew ] .
we do see systematic differences in the equivalent width ratio , however : low mass , young , and unreddened galaxies with low sfrs ( blue points ) have @xmath146 , while older , dustier , more massive galaxies with higher sfrs ( red points ) have @xmath170 .
these systematic differences may represent real differences in optical depth between the two samples , although it may perhaps be surprising that the lower mass and less reddened samples would show higher optical depth than the dustier and more massive galaxies .
the alternative explanation is that underlying emission is more likely to decrease the equivalent width of @xmath124 relative to @xmath131 in the low mass sample .
this second explanation is supported by the four low mass galaxies whose individual spectra require the presence of such underlying emission ( see figure [ fig : ind_ewcomp ] ) , and perhaps by the fact that emission is clearly stronger in low - mass galaxies .
however , such underlying emission must have relatively little effect on the velocity centroids of the lines , since systematic differences between the red and blue points are less apparent in the velocity comparison of feii @xmath131 and @xmath124 shown in figure [ fig : compvels ] than in the equivalent width comparison in figure [ fig : comp_ew ] .
we next examine the maximum blueshifted velocities of the absorption lines in composite spectra , a measurement we could not make for the individual objects due to limited s / n .
high resolution , high s / n spectra of lensed galaxies show that the outflow - related interstellar absorption lines extend to typical velocities of @xmath171 to @xmath172 km s@xmath2 @xcite , and similar maximum velocities are seen in composite spectra of galaxies at @xmath4 @xcite . @xcite
measured the velocity @xmath173 at which @xmath174 absorption reaches 90% of the continuum level in galaxies at @xmath175 , and found it to be correlated with both mass and star formation rate . although measurement of the maximum wind velocity is dependent on s / n , it is less likely to be affected by the interstellar medium or by underlying emission than the velocity centroid measurements discussed above .
we define the maximum velocity as the velocity corresponding to the wavelength at which the absorption profile first meets the normalized continuum ( i.e. , has a value of one ) on the blue side of the line .
because this measurement is highly dependent on s /
n , we use the 5000 monte carlo simulations described above , and take @xmath176 to be the average of the velocity at which the line meets the continuum in the 5000 realizations , with @xmath158 error given by the standard deviation .
we measure @xmath176 for the five strong absorption lines and for @xmath163 ( we do not measure @xmath177 because the blue wing is affected by the @xmath163 line ) .
the average value of @xmath176 for these six lines in the composite spectrum of all 96 galaxies is @xmath178 km s@xmath2 , in good agreement with other measurements of galaxies at @xmath4 .
as other authors have discussed ( e.g. @xcite ) , such a velocity is likely to be higher than the escape velocity of the galaxy , and it is therefore likely that at least some of the gas escapes the halo .
figure [ fig : termvels ] shows @xmath176 for each of the composite spectra .
we have averaged the measurements of @xmath176 for the six absorption lines , and plotted the result against the average value of the parameter on which the composite spectrum is based ( e.g. the average stellar mass of the spectrum of low mass galaxies ) .
the uncertainties in this measurement are large , but the maximum outflow velocity shows little dependence on dust content , age , sfr or the presence of emission .
we do , however , see a weak dependence on stellar mass ; this is the only parameter for which the maximum velocities of all six lines are higher in one composite relative to the other . in the high mass composite ,
we find @xmath179 km s@xmath2 , while in the low mass composite we measure @xmath180 km s@xmath2 , for a difference of @xmath181 km s@xmath2 .
thus , we find that the maximum outflow velocity in massive galaxies is @xmath182 times faster that that in low mass galaxies , with 1.8@xmath6 significance . for comparison ,
we find @xmath183 km s@xmath2 for the @xmath82 composites , @xmath184 km s@xmath2 for the sfr composites ,
@xmath185 km s@xmath2 for the age composites , and @xmath186 km s@xmath2 for the composites based on emission .
we also note that all of the composites have similar s / n of 1116 per pixel ( see table [ tab : sedcomps ] ) ; while the s / n of the high mass sample is somewhat higher than that of the low mass sample , this is also the case for several other pairs of composites , indicating that it is not likely to be the cause of the velocity difference observed between the high and low mass composites . using 1400 spectra of galaxies at @xmath13 from the deep2 survey
, @xcite found that @xmath187 and @xmath188 .
we clearly do not see the same dependence on star formation rate , as our measured values of @xmath176 are nearly identical in the high and low sfr composites . given the mean sfr values for the two composites ( 10 and 128 @xmath87 )
, we would expect the high sfr sample to have a velocity 2.1 times higher that that of the low sfr sample , a result inconsistent with the observed @xmath189 at the level of 3.7@xmath6 ( though given the systematic uncertainties on the sfr discussed above , it would be desirable to repeat this measurement with an independent estimate of the sfr ) .
similarly , in a sample of galaxies with and absorption at @xmath190 , @xcite find only weak ( @xmath191 ) evidence for a trend between outflow velocity and star formation rate .
some insight into the lack of correlation may be found by considering galaxies in the local universe .
using observations of absorption in a sample of nearby galaxies spanning a range of four decades in sfr , @xcite found a scaling of @xmath192 , while @xcite suggest that the trend of increasing velocity with sfr flattens for sfrs above @xmath193 @xmath87 . if this is the case , high redshift samples probing a limited range in sfr may be unlikely to detect a relationship with velocity .
we note , however , that the @xcite sample does not span a broad range in sfr ; their low sfr bin contains galaxies with @xmath194 @xmath87 , while their high sfr bins consists of galaxies with @xmath195 @xmath87 . turning to our observed mass dependence
, we find that it is entirely consistent with the results of @xcite , however ; for the mean stellar masses of the upper and lower mass bins , a dependence of @xmath196 would result in the massive galaxies having a maximum velocity 1.5 times larger than the low mass galaxies , exactly the difference between our two measured values .
given the uncertainties in our measurements of @xmath176 , however , a variety of mass dependencies would be consistent with our observations .
the remainder of this paper is devoted to the emission features in the composite spectra . as listed in table [ tab : lines ] and shown in figures [ fig : all_fullspec ] and [ fig : allcomposites ] , we detect emission from five * fine structure lines , and see significant emission in the @xmath5 2796 , 2803 doublet for about @xmath197 of the galaxies in the sample .
emission is more likely to appear in galaxies with blue uv slopes and in lower mass galaxies .
galaxies with emission also have stronger * emission ; this can be seen most clearly in the composite spectra of galaxies with and without emission shown in the bottom panels of figure [ fig : allcomposites ] and via the measured equivalent widths of the * lines in these spectra given in table [ tab : em ] . given the anti - correlation between emission and stellar mass , it is then unsurprising that low mass and low sfr galaxies also tend to have stronger * emission compared to their higher mass , higher sfr counterparts .
* emission is generally not seen in local star - forming and starburst galaxies @xcite .
these lines have also not been seen in samples of galaxies at redshifts similar to the current sample , due to spectral coverage in the case of @xcite and probably due to s / n limitations in the sample of @xcite .
the situation is similar to that of the * fine structure lines in the far - uv , which are seen in galaxies at @xmath44 @xcite but not in local starbursts @xcite .
@xcite detect both * and emission in a starburst galaxy at @xmath17 , and propose that both are generated by photon scattering in outflowing gas .
they also show that the emission is significantly spatially extended compared to the stellar continuum .
this model may explain the absence of fine structure emission in local galaxies , if the slits used to observe such objects cover only the central regions of the galaxies and capture little emission from an extended outflow @xcite .
@xcite use radiative transfer models to predict the strength of and absorption and emission for a variety of outflow models ; we compare our results to these models below . as a first test of the origin of the * emission , we compare the velocity structure of the strongest line , * @xmath130 , with that of the absorption line from which the emission is expected to arise , @xmath198 , using the composite spectrum of all 96 galaxies .
the comparison is shown in the left panel of figure [ fig : emabs_vel ] , with * @xmath130 emission in blue and @xmath198 absorption in red .
the absorption line shows the clear asymmetric profile of galactic outflows , with a blueshifted centroid and tail extending to large velocities .
* @xmath130 is a nearly perfect mirror of the absorption profile in emission , suggesting that the two lines do indeed arise from the same gas .
the * @xmath130 profile also indicates the presence of dust ; since the * lines are optically thin ( we see no absorption at their wavelengths ) , * emission should reach us from both sides of the outflow , rather than only the near side seen in absorption .
the fact that we see more emission from the blueshifted side than the redshifted side is probably due to extinction , as photons from the far side of the galaxy must pass through more dust in order to reach us @xcite . as a test of this scenario
, we also compare the * @xmath130 profiles in the high and low @xmath82 composite spectra , in the right panel of figure [ fig : emabs_vel ] .
if the asymmetric line profile is due to extinction , we would expect greater asymmetry in the composite spectrum of the dustier galaxies .
the lower s / n of the extinction - based composites makes a robust comparison difficult , but figure [ fig : emabs_vel ] does show that the red wing of the line extends to somewhat higher velocities in the low @xmath82 composite .
a comparison of the equivalent widths of the blue and red sides of the lines also offers support to the model in which the line asymmetry is due to extinction ; the red flux is @xmath199% of the blue flux in the high @xmath82 composite , and @xmath200% of the blue flux in the low @xmath82 composite . in this section
we examine the origins of * and emission in more detail , through photoionization modeling ( section [ sec : cloudy ] ) , a study of the spatial extent of the emission ( section [ sec : spatialextent ] ) , and a comparison of the relative strengths of the absorption lines and their associated * emission ( section [ sec : emabs ] ) . in this subsection
we use photoionization models to study the origins of * and emission . while previous work @xcite and the kinematics of the * @xmath130 line suggest that * fine structure emission arises in the outflow , other sources are also possible . in order to assess the possibility that the * emission originates in regions , we have constructed a suite of photoionization models which we use to examine the strength of * emission as a function of metallicity and ionization parameter .
the models were constructed with version 08.00 of cloudy , last described by @xcite .
we employ an input ionizing spectrum from starburst99 @xcite , assuming continuous star formation and a metallicity of @xmath201 , or @xmath202 .
we assume an electron density @xmath203 @xmath204 , consistent with the density inferred from the average value of the density - sensitive [ ] @xmath205/[]@xmath206 line ratio .
( we find @xmath207[]@xmath205/[]@xmath208 , implying densities between the low density limit and @xmath209 @xmath204 . )
we consider gas phase metallicities of 0.2 , 0.4 , 0.6 , 0.8 , and 1.0 @xmath210 .
we use cloudy s default region abundance set , and in addition consider models with the fe abundance reduced by factors of 0.8 , 0.6 , 0.4 , and 0.2 for each of the metallicities adopted .
each model is characterized by an ionization parameter @xmath211 , which is defined as the ratio of ionizing photon density to total hydrogen density , @xmath212 where @xmath213 is the rate of ionizing photons ( number s@xmath2 , given by the starburst99 model used to construct the input ionizing spectrum ) , @xmath214 is the distance from the ionizing source to the inner edge of the cloud , and @xmath215 is the hydrogen density .
the models assume a spherical region , but the geometry is effectively plane parallel if the thickness of the cloud is less than @xmath216 ; this is the case for all of our models except those with the highest values of the ionization parameter considered , @xmath217 .
the models use cloudy s default stopping criterion , in all cases stopping when the temperature falls below 4000 k. results of the photoionization modeling are shown in figure [ fig : cloudy ] .
we plot the ratio of each of the * lines to [ ] @xmath53626 , 3729 against the logarithm of the dimensionless ionization parameter @xmath211 , which represents the ratio of ionizing photons to particles .
( it is more standard to measure the strength of a line of interest against a recombination line such as @xmath218 , but [ ] is the only strong nebular emission line for which we have measurements . )
the [ ] / * ratio shows little metallicity dependence but decreases strongly with increasing ionization parameter , driven largely by the decrease in [ ] ( and corresponding increase in [ ] ) as ionization parameter increases .
grey horizontal lines show the measured ratios for individual spectra in our sample .
it is clear that , in order to match the observed ratios , a very high ionization parameter of @xmath219 is required .
we do not have ionization parameter measurements for the galaxies in our sample , but measurements of lensed galaxies at @xmath4 suggest that typical star - forming galaxies at these redshifts have @xmath220 @xcite , higher than values observed in most local galaxies but lower than needed to produced our observed [ ] / * ratios .
we also note two additional factors .
first , the measured line ratios are not corrected for dust extinction .
this is unlikely to be a large effect , given the relatively narrow wavelength separation , but the effect of dust would decrease the * lines relative to [ ] , making the observed [ ] / * ratios higher than the intrinsic ratios and further increasing the ionization parameter needed to match the data .
second , all of the models shown assume cloudy s default region o / fe ratio , which may not be appropriate given that an underabundance of the fe - peak elements relative to @xmath221-capture elements ( such as o ) has been observed in lensed galaxies at high redshifts @xcite .
we have tested the effect of a decreased abundance of fe relative to o in the cloudy models ; the effect is to weaken the * emission , increase the [ ] / * ratio , and thereby increase the ionization parameter needed to match the observations . in other words , a decreased abundance of fe relative to o shifts all of the colored points in figure [ fig : cloudy ] upwards . in summary
, it is unlikely that that the galaxies in our sample would have ionization parameters high enough to produce the required [ ] / * ratios , and it is therefore also unlikely that the observed * emission originates in the regions of the galaxies .
next we examine the question of .
in contrast to * ( and resonant ) emission , emission from is observed in local regions @xcite , so it is important to assess the likely sources of this emission in high redshift galaxies . in figure
[ fig : cloudy_mg2 ] we plot the predicted ratios of [ ] to @xmath174 and @xmath222 emission against the ionization parameter , color - coding by metallicity as in figure [ fig : cloudy ] .
the observed ratios are shown by the grey horizontal lines ; unlike the [ ] / * ratios , they are fully consistent with the expected ionization parameters of @xmath223 to @xmath224 .
the photoionization modeling therefore indicates that , in contrast to the * emission , much of the emission may arise in regions . because is a resonance line , however , photons produced in regions then scatter in outflowing gas , producing the observed blueshifted absorption and redshifted emission profiles .
we have suggested above ( section [ sec : individual ] ) that emission in star - forming galaxies at @xmath1902 is similar to @xmath12 emission at @xmath43 , in that it exhibits a complex profile of combined emission and absorption ; we have also seen that emission is more common in low mass and less dusty galaxies .
the photoionization modeling supports this connection , indicating that emission , like that of @xmath12 , arises from the resonant scattering of photons produced in regions .
thus the production of photons in regions is likely to account for the difference between the profiles of and that of lines not coupled to fine structure transitions such as @xmath225 . for example , the spectra of q1623-bm95 and q1700-bm154 , shown in figures [ fig : ind_fespecs ] and [ fig : mg2profiles ] , show weak to moderate @xmath225 absorption with no associated emission , while their profiles show strong redshifted emission accompanied by weak blueshifted absorption . under the assumption that emission is produced by the scattering of continuum photons produced in the outflow
, one would expect to see similar emission in the and @xmath225 lines ; instead we see much stronger emission in the transitions , accounted for by the production of photons in regions and their subsequent resonant scattering .
a comparison of the predicted strengths of @xmath174 and @xmath225 emission in the photoionization models supports this conclusion ; for ionization parameters @xmath223 to @xmath224 , emission from mgii @xmath174 is @xmath226 times stronger than @xmath225 emission . we therefore expect scattered emission from regions to be much more significant for than for @xmath225 .
more detailed photoionization modeling in the future may address the question of whether the [ ] and emission arise from the same parts of the regions ; according to @xcite , emission from regions arises primarily from recombinations in the outer layers .
an additional question which can not be addressed by our current models and data is that of additional sources of [ ] emission .
[ ] emission from diffused ionized gas is seen in local galaxies ( e.g. @xcite ) , and is likely present in higher redshift galaxies as well .
however , the cloudy models do not provide predictions for such emission or absolute [ ] luminosities with which to compare our measurements , and our observations do not have the spatial resolution necessary to address this question ( though see section [ sec : spatialextent ] below , where we show that the [ ] emission has the same spatial distribution as the stellar continuum , indicating that most of the emission does not arise in an extended outflow ) .
our observed [ ] / ratios are consistent with emission from both ions arising in regions , but other sources of [ ] emission can not be ruled out .
future , more detailed modeling and observations with higher sensitivity and spatial resolution will address this question . to summarize the results of this subsection
, the photoionization modeling indicates different mechanisms for the production of * and emission .
although both ultimately arise in the outflow , * emission arises from the re - emission of continuum photons absorbed in the resonance transitions , while the resonant scattering of photons originally produced in regions accounts for much of the emission .
if the * emission originates in a large - scale galactic outflow , it may have a more extended spatial distribution than the stellar continuum ; similarly , resonant scattering of photons in outflowing gas may produce spatially extended emission @xcite .
@xcite study * and emission in a galaxy at @xmath17 , finding that emission is broader than the stellar continuum and extends to distances of @xmath227 kpc .
likewise , @xcite use deep narrowband images to show that diffuse @xmath12 emission surrounds galaxies at @xmath4 to distances of @xmath19 kpc ; the @xmath12 transition is similar to in its radiative transfer properties . in order to assess the spatial extent of and * emission in our sample
we create two - dimensional composite spectra and compare the spatial extent of the line emission relative to the stellar continuum .
we also construct a 2d composite [ ] spectrum for comparison .
the 2d composite spectra are created by cutting out a 100 region around the line of interest in each individual spectrum .
these cutouts are also centered spatially ( but without spatial resampling ) on the object , using the object positions stored in the image header by the deep2 reduction pipeline .
the cutouts are then resampled to a common dispersion of 20 km s@xmath2 per pixel using linear interpolation , and the appropriate 2d wavelength solution is applied to each cutout . because the original two - dimensional spectra are rectified and because we consider only a short range in wavelength , the object traces are not curved and the wavelength does not depend on spatial location along the slit .
we then measure the centroid of the object position in each spectrum ( since the positions in the header are not exact ) , and use these measurements to construct spatial offsets used to combine the individual spectra .
further details of the combination depend on the line considered and are described below .
the composite includes only the 33 galaxies with emission , determined as described in section [ sec : mg2uvcolor ] .
because the peak velocity of emission varies significantly in individual spectra , we shift each 2d spectrum both spectrally and spatially , centering on the peak of the emission , and combine the individual spectra using a s / n - weighted average .
we measure the extent of * emission using the strongest of the * lines , * @xmath130 .
we combine the full sample ( 95 of the 96 galaxies in the sample have spectral coverage of the * @xmath130 line ) , aligning each spectrum spatially and again combining using a s / n - weighted average . for the [ ] composite , we combine the 51 galaxies with [ ] emission , again aligning each spectrum spatially and combining using a s / n - weighted average . in all three cases
we then average the two sides of the composites in order to increase s / n .
we determine uncertainties by creating 500 artificial 2d composite spectra using bootstrap resampling : we measure the average and standard deviation of the line and continuum profiles in the 500 artificial spectra , and take the @xmath158 uncertainties in the spatial profiles to be the standard deviation of their measurements in the sample of artificial spectra .
the two - dimensional composite spectra for the three lines are shown in figure [ fig:2dspecs ] .
spatial profiles of the line emission and stellar continuum are shown in figure [ fig : spatialprofiles ] .
the top two panels show emission : at left , we compare the emission profile ( red ) with that of the stellar continuum ( black ) , where the continuum has been scaled by a factor of 1.6 to have the same peak value as the emission .
the blue dotted line shows the point spread function , measured from the average spatial profiles of two stars appearing on the masks . in the upper right panel
we show the difference between the line and continuum emission .
the emission is slightly stronger at larger distances , reaching a peak excess of 40% greater than the scaled continuum at a radius of 0.8 arcsec . in the middle two panels we show the same comparison for the weaker * emission . in this case
we scale the stellar continuum by a factor of 1.15 to match the peak * flux , and find that the * emission is also somewhat more extended than the continuum , though with a smaller excess ; the difference between line and continuum peaks at @xmath228 arcsec , with a 10% excess in the line flux relative to the scaled continuum .
although this is a smaller excess than we see in the emission , we detect it with comparable significance because of the larger sample of galaxies used in the * composite . because the * emission lines are stronger in galaxies with emission ( see section [ sec : emission ] ) , a natural additional test would be to measure the excess * emission in galaxies with emission ; unfortunately
, however , the sample of emitters is too small to enable sufficient s / n to make this measurement .
the bottom two panels of figure [ fig : spatialprofiles ] show the comparison of the spatial profile of [ ] emission relative to the continuum .
[ ] emission is expected to arise primarily in regions , and because it is not a resonance line , its spatial distribution should trace that of the star - forming regions of the galaxy , with any differences with respect to the continuum emission arising from differing spatial distributions of the hot young stars producing the regions and the older stars responsible for the continuum ( or , at larger radii , the possible generation of [ ] emission in the outflow ) .
figure [ fig : spatialprofiles ] shows that the [ ] emission indeed follows the spatial distribution of the continuum closely .
there is no excess of line emission at @xmath229 arcsec , as is observed for and * emission ; instead the [ ] emission is slightly weaker than the continuum at @xmath230 arcsec , although the significance of this result is marginal .
we conclude from this test of the spatial distribution of [ ] emission that the observed excesses in and are not an artifact of the comparison of stronger line emission with weaker continuum . we have presented evidence that both and * ( but not [ ] ) emission are slightly more spatially extended than the underlying stellar continuum , with excess relative to the continuum detected at the @xmath1822@xmath6 level .
this supports the conclusion that these emission lines arise in outflowing gas , though higher s / n measurements are clearly needed .
the s / n of the stacked spectra we use to make this measurement is only just high enough to detect a difference in the spatial distribution of lines and continuum , but not high enough to measure line emission to significantly larger radii than the continuum .
detections of extended @xmath12 emission to distances of @xmath19 kpc @xcite suggest that such emission could be present at much larger distances , but the weakness of and * emission relative to @xmath12 mean that such detections at high redshift are likely out of the reach of current technology .
we conclude our discussion of the origins of * emission with a comparison of the relative strengths of the * emission lines and their associated absorption . as shown in figure [
fig : energylevels ] , each of the fine structure emission lines is coupled to an absorption line which provides the source of the emitted photons .
the generic expectation for the model in which * emission is produced by photons absorbed in transitions in the outflow is that * emission will increase as absorption increases .
@xcite provide extended discussion of this issue , and use radiative transfer models to predict the relative strengths of absorption and emission lines for a variety of outflow models . for a spherically symmetric outflow in the absence of dust , conservation of photons
requires that the total emission and absorption equivalent widths sum to zero ; thus , in the absence of dust or particular geometric effects , we expect the total emission and absorption strengths to be approximately the same
. note also that the absorbed photons may be re - emitted as emission at their original wavelength rather than through one of the fine structure transitions , producing a p cygni - like profile ; whether or not this is likely to happen depends on the probabilities of the transitions involved and on the details of the galaxy , including the spatial and velocity distribution of outflowing gas , the presence of an interstellar medium at zero velocity , and the dust content .
we show the expected relationship between absorption and emission in figure [ fig : pkr ] , which shows the predicted equivalent widths of the @xmath602600 and * @xmath602626 lines in all of the models considered by @xcite .
the points represent a wide variety of outflow scenarios rather than a range of galaxies within the same model ; thus they should be taken simply as a generic prediction of how increasing one line affects the other .
the left panel compares the strengths of @xmath602600 absorption and * @xmath602626 emission , and shows that , although there is considerable scatter , * @xmath602626 emission generally increases as @xmath602600 absorption increases .
the right panel shows that much of the scatter is due to the fraction of emission which escapes at 2600 rather than at 2626 ; except in cases with significant extinction or particular outflow geometry , the sum of the total emission and absorption is approximately zero .
the outliers ( points falling far from the solid line in the right panel of figure [ fig : pkr ] ) represent either non - spherically symmetric models or models with significant dust content .
these models show that either dust or geometry alone can cause large deviations from the simple conservation of photons . in figure
[ fig : emabs_ew ] we plot the equivalent width of each of the * emission lines against its associated absorption line , for each of the composite spectra ( note that because of the varying overlap of the samples used for the various composites , as discussed in section [ sec : compspecs ] , the points on these plots are not all independent ) .
we focus the following discussion primarily on the @xmath602600 absorption and * @xmath602626 emission lines , since * @xmath602626 is the strongest of the * lines , but ( with some exceptions noted below ) the paired lines follow similar trends . as already noted , * emission is stronger in less massive and less dusty galaxies and in galaxies with emission . it is also immediately apparent that we see a trend of _ decreasing _ * @xmath602626 emission with increasing @xmath602600 absorption , the opposite of what we expect based on the model described above .
we have not included @xmath602600 emission in this comparison , since it is not significantly ( @xmath231 ) detected in any of the composites ; however , examination of figure [ fig : allcomposites ] shows some evidence for redshifted p cygni @xmath602600 emission in the low mass and low @xmath82 composites .
these are the only spectra in which the emission is detected at @xmath232 significance ; we find @xmath233 in the low mass composite , and @xmath234 in the low @xmath82 composite .
these two composites already show relatively weak @xmath602600 absorption and strong * @xmath602626 emission , so including emission at 2600 simply increases the imbalance between total emission and absorption in these samples .
more generally , @xmath602600 emission decreases as the optical depth of the transition increases , and the @xcite models show that little to no emission is seen when an ism component is present in the galaxy .
our observations therefore suggest higher optical depth in higher mass and dustier galaxies , possibly due to a more significant interstellar medium .
@xcite also find that symmetric absorption centered at zero velocity is significantly stronger in more massive galaxies , and @xcite find more absorption at @xmath235 km s@xmath2 in galaxies with higher baryonic ( gas + stellar ) and dynamical mass at @xmath4 .
a more significant ism in more massive galaxies does not account for the observed decrease in * emission with increasing absorption , however , since the presence of an ism increases both the absorption and total emission , while shifting the emission from the resonance lines to the fine structure transitions .
the fact that the total absorption is stronger than the total emission in all spectra may be attributable to dust ; the @xcite models indicate that the effect of dust is to suppress the emission with relatively little effect on the absorption profile . as noted above , the * @xmath602626 emission profile in the composite spectrum of the full sample shows evidence for extinction in the lack of redshifted emission relative to blueshifted emission .
furthermore , extinction may at least in part explain the observed trend in @xmath236 vs. @xmath237 , if dust attenuates a larger fraction of the emission in massive galaxies .
this does not at first appear to be a satisfactory explanation , since the average value of @xmath82 is slightly higher in the low mass sample than in the high mass sample .
on the other hand , as discussed in section [ sec : compspecs ] , uncertainties in the extinction law for young galaxies suggest that @xmath82 for this sample may be overestimated , and independent measurements of extinction from _
24 imaging indicate that , on average , dust attenuation is larger in more massive galaxies @xcite . even if this is the case , however , the magnitude of the effect does not appear to be sufficient .
@xcite model the effects of dust as a function of the optical depth @xmath238 , finding that flux is reduced by a factor of @xmath239 ; in order to suppress the emission to the observed degree in the high mass and high dust samples , @xmath240 is required , while our values of @xmath82 ( which do not suffer from the systematic uncertainties affecting the lower mass sample ) indicate @xmath2412 for these samples .
@xcite also consider a dust+ism model .
as discussed above , the effect of an ism component is to increase both the absorption and total emission , while shifting the emission from the resonance lines to the fine structure transitions ; if the ism is also dusty , the emission is further reduced , particularly in the resonance lines .
however , the dust+ism model with @xmath242 still exhibits increased emission relative to the fiducial , dust - free model , indicating that the increase in both absorption and emission outweighs the effect of dust ; this is counter to our observations of weaker emission in more massive galaxies with more significant ism .
it may be that such a model with somewhat higher dust content would sufficiently suppress the emission without obscuring it entirely ; more detailed modeling is needed , and some degree of fine tuning of the dust content would seem to be required to match the observations . for the moment
we conclude that dust is an important factor but may not be the only explanation for the observed trend .
the remaining factor is the geometry of the absorbing and emitting regions .
anisotropic outflows viewed from certain directions may produce observed absorption to emission ratios that deviate from the expected conservation of photons ; in particular , a galaxy with a bi - conical outflow observed edge - on will show no absorption but strong emission .
thus observations of individual galaxies at a variety of inclinations could produce a trend similar to the observed @xmath236 vs. @xmath237 relationship .
this is unlikely to be the explanation for the current sample , however , since our data points are for composite spectra of galaxies selected without regard to inclination or morphology .
does not increase with apparent inclination ( d.r .
law , private communication ) , and in any case such a bias would result in stronger emission and weaker absorption for the dustier subsample , the opposite of what is observed . ]
the last issue we consider is the size of the * -emitting region relative to the area encompassed by the spectroscopic slit .
if fine structure emission occurs in outflowing gas at large distances from galaxies , the emitting region may be larger than the region subtended by the slit ; the broader the extent of the emission , the higher the fraction that may fall outside the slit . in this context ,
previous observations of low ionization absorption and fine structure emission in far - uv spectra at high redshift provide useful insights ; the resonant absorption lines are coupled to non - resonant * emission in the same fashion as the lines studied in this work .
@xcite detected * fine structure emission accompanying absorption in the composite spectrum of lyman break galaxies at @xmath243 , finding that the emission lines were too strong to be produced in the regions of the galaxies but much weaker than the absorption lines .
although they found no fully satisfactory explanation for the * lines ( outflow models were deemed inadequate due to the narrowness of the * lines relative to @xmath12 ) , they suggested that the weakness of the * lines could be due to the spectroscopic slit subtending only a small portion of the emitting region .
more recently , @xcite compare the composite spectra of lyman break galaxies at @xmath244 and @xmath243 , and find that galaxies at @xmath244 have stronger * emission but weaker absorption than galaxies at @xmath245 , the same pattern we observe in the lines in low and high mass ( and age , sfr , and dust ) galaxies at @xmath3 .
they interpret this pattern as evidence that photons are absorbed and re - emitted at larger distances at @xmath243 than at @xmath244 .
if slit losses are in part responsible for the weakness of the fine structure emission lines , significant absorption in the resonance lines must occur at large distances from the galaxies .
recent results indicate that this is likely to be the case . @xcite use stacked spectra of paired foreground and background galaxies to measure the strength of low ionization absorption as a function of impact parameter , finding @xmath246 at a mean distance of 31 kpc in both the @xmath247 and @xmath248 lines .
the strength of @xmath247 remains the same at 63 kpc , and significant absorption in @xmath12 is seen at @xmath249 kpc .
@xcite also see low surface brightness @xmath12 emission extending to @xmath19 kpc . in a study of absorption systems and [ ]
emission at @xmath250 , @xcite conclude that outflows are the primary source of absorption systems , and that outflowing gas can reach radii of @xmath156 kpc .
these results indicate that significant absorption and emission are likely to take place beyond the radius subtended by the slit ; the 1 and 1.2 slits used in our study subtend distances of 810 kpc for galaxies at @xmath251 .
the models of @xcite also assess the effect of slit loss .
the fraction of absorption and emission at large radius depends on the wind model considered ; for most models , they find that @xmath229 slits are sufficient to capture most of the emission in high redshift galaxies .
a notable exception is the radiation - pressure - driven wind model of @xcite , for which a slit subtending @xmath720 kpc is required to admit most of the scattered photons .
given the detection of significant resonant absorption at large radius in similar galaxies at @xmath4 , we suggest that , along with dust , increasing slit losses in massive galaxies may explain the observed trend of decreasing * @xmath602626 emission with increasing @xmath602600 absorption . in this scenario ,
the smaller ratio of emission to absorption strength observed in high mass ( age , dust , sfr ) galaxies is an indication that photons are absorbed and re - emitted at greater distances from the galaxy than in lower mass and less dusty objects .
this is consistent with the observation in section [ sec : maxvels ] that massive galaxies have higher wind velocities .
we also find support for this scenario in observations of a single low mass galaxy ; @xcite studied absorption and * emission in a compact , unreddened galaxy at @xmath252 , finding that the emission and absorption strengths were roughly equal . in this galaxy ,
very few of the * photons are lost to dust or fall outside the slit .
a prediction of this slit - loss model is that * @xmath602626 emission should be more extended in high mass galaxies than in low mass galaxies .
we test this by constructing two - dimensional composite spectra of the high and low mass samples as described in section [ sec : spatialextent ] , again determining uncertainties from the standard deviation of the spatial profiles of 500 artificial 2d composite spectra created using bootstrap resampling .
the 2d spectra are shown in figure [ fig : highlowmass_2d ] , and the spatial profiles are shown in figure [ fig : spatialprofiles_mass ] . in the top two panels of figure [ fig : spatialprofiles_mass ]
we show the * @xmath602626 spatial profiles at left and the difference between the profiles in the high and low mass samples at right .
although the uncertainties are significant , we find that at a radius of 1 the flux of the high mass sample is nearly 5 times that of the low mass sample ; the difference in flux at 1 is @xmath253 , for a significance of 1.9@xmath6 .
we also observe excess emission to @xmath182 , with lower significance .
this difference between the emission line profiles in the high and low mass galaxies can be further clarified by comparisons of the * line emission and the continuum for the high and low mass samples .
the relevant spatial profiles are shown in the lower two panels of figure [ fig : spatialprofiles_mass ] . a comparison of the line and continuum emission in the high mass sample shows that the emission line flux is @xmath254% higher at a radius of 0.81.1 arcsec ; this difference is not statistically significant , however , given the large uncertainties associated with the small high and low mass samples . a similar comparison of the line and continuum emission in the low mass sample shows that the two profiles are the same to within 5% .
finally , we compare the continuum profiles of the high and low mass samples .
massive galaxies are observed to be larger than low mass galaxies ( e.g. @xcite ) , and a comparison of the continuum profiles to the psf in the lower two panels of figure [ fig : spatialprofiles_mass ] suggests that this is also the case in our sample .
more quantitatively , at a radius of 0.81.1 arcsec the continuum emission from massive galaxies is @xmath255% higher than the continuum emission from low mass galaxies , but as with the line and continuum comparison above , this difference is not statistically significant .
thus the more extended line emission observed for high mass galaxies shown in the upper left panel of figure [ fig : spatialprofiles_mass ] appears to be due to a combination of extended line emission in massive galaxies and the larger sizes of massive galaxies ; neither of these effects are statistically significant alone , but in combination they produce the moderately more significant difference between the * profiles in high and low mass galaxies . in summary ,
the observation of more extended emission from the outflow in more massive galaxies supports the scenario in which slit losses are more significant for more massive galaxies .
a definitive test requires the detection of emission at much larger radii , but the weakness of the fine structure lines is likely to preclude this test for some time . a second important test of the model is a comparison of the strength of absorption as a function of impact parameter for both high and low mass galaxies , using pairs of foreground and background galaxies or galaxies with background quasars .
the scenario presented here predicts stronger low ionization metal line absorption at larger radii for more massive galaxies .
such results have already been found for galaxies associated with absorption systems at lower redshifts @xcite , and the coming years will see such studies extended to the redshifts probed by this sample and higher .
we have thus far focused discussion on the @xmath602600 absorption and * @xmath602626 emission lines , as * @xmath602626 is the strongest of the * lines . although we do not detect all of the * lines in all of the composites , the other lines in the uv1 multiplet show the same trend of stronger emission and weaker absorption in low mass galaxies ; we measure only upper limits on * @xmath602612 and * @xmath602632 in the high mass composites .
the uv2 multiplet presents a confusing picture , however . in this case
we observe weaker absorption and stronger emission in the high mass and high dust samples , although galaxies with emission also show weaker absorption and stronger emission .
the explanation for these mixed results is unclear . in the uv3 multiplet
we again observe weaker absorption in the low mass sample than in the high mass , but the * @xmath602365 line is too weak to lead to useful conclusions regarding the emission . as the uv2 observations partially conflict with our proposed model , further study of these lines is needed .
we have shown that the average velocity profile of * @xmath602626 emission , photoionization modeling , and the extent of the emission relative to the continuum support the scenario in which fine structure emission is generated by photon scattering in outflowing gas .
greater extinction and/or a greater spatial extent of outflowing gas in massive galaxies is required to explain the observed decrease in * emission with increasing absorption .
we have presented measurements of rest - frame near - uv and absorption and emission in a sample of 96 star - forming galaxies at @xmath256 .
the presence of large - scale galactic outflows is indicated by the blueshift of interstellar absorption lines relative to the systemic velocity .
our primary results are as follows . 1 .
for 51 of the 96 galaxies , we obtain systemic redshifts from the centroids of strong [ ] @xmath5 3726 , 3729 emission , which falls at the red end of the spectral range covered .
these galaxies have an average redshift @xmath257 .
interstellar absorption lines in these galaxies show an average offset from the systemic velocity of @xmath64 km s@xmath2 , less than the average offset of @xmath258 km s@xmath2 measured for galaxies at @xmath4 .
we use this offset of @xmath64 km s@xmath2 to estimate the systemic redshift of the remainder of the galaxies in the sample .
we also show that the average velocity offset decreases from @xmath259 km s@xmath2 at @xmath70 to @xmath260 km s@xmath2 at @xmath68 .
these higher redshift samples have a slightly higher average mass than our @xmath13 sample , and have velocity offsets measured from far - uv interstellar absorption lines rather than the near - uv lines used in this study .
more careful accounting of these selection effects , and of the effects of the interstellar medium on the velocity measurements , is therefore required to determine whether the decrease in velocity offset with decreasing redshift represents a real trend of decreasing outflow velocity .
2 . individual galaxies in the sample show more variety in the doublet than in the resonance lines .
is always seen in absorption , while ranges from strong emission to pure absorption .
emission is more common in galaxies with bluer uv slopes , echoing far - uv observations of galaxies at @xmath44 in which @xmath12 emission is stronger in bluer galaxies .
emission is also stronger in low mass galaxies , and fine structure * emission is stronger in galaxies with emission .
we compare the velocities and equivalent widths of the and transitions in individual spectra , in order to assess the effect of emission filling on measurements of outflow velocity .
underlying emission appears to shift the velocity centroid of the @xmath124 line to the blue in @xmath261 of the sample , while the @xmath163 line shows such a shift in @xmath137 of the sample .
studies of individual objects suggest that emission filling may be a stronger effect in low mass galaxies , but composite spectra show little difference in velocity centroids between the lines expected to be most and least affected by emission filling .
is much more clearly affected by underlying emission .
4 . the average maximum outflow velocity ( the maximum extent of the blue wing of the interstellar absorption lines in the composite spectrum of all of the galaxies in the sample ) is @xmath178 km s@xmath2 , in good agreement with other studies of star - forming galaxies at @xmath3 2 .
@xmath262 is 1.5 times larger in the high mass subset of galaxies than in the low mass subset , consistent with previous results indicating @xmath263 .
the velocity centroids @xmath264 are also blueshifted to higher velocities in the composite of high mass galaxies than in the low mass composite . unlike some previous studies
, we see no trend between @xmath176 and star formation rate .
5 . we present the following evidence in support of the model in which * and emission is generated by photon scattering in outflowing gas : * the velocity profile of * @xmath130 emission in the composite spectrum of the full sample of galaxies shows a blueshifted tail very similar to that of the @xmath265 absorption line , from which the emission is presumed to arise . *
photoionization modeling indicates that very high values of the ionization parameter , @xmath266 , are required in order for the * emission to originate in regions .
the ionization parameters of the galaxies in the sample are unlikely to be this high , implying that * emission is produced by the re - emission of continuum photons absorbed in the resonance transitions in outflowing gas . * in contrast to * emission , photoionization modeling shows that the strength of emission is consistent with its production in regions .
we conclude that photons from regions resonantly scatter in outflowing gas , producing the observed blueshifted absorption and redshifted emission profiles . *
we compare the spatial extent of and * emission relative to the continuum , and find that both emission lines are moderately more extended .
higher s / n measurements are needed to more robustly confirm this trend and measure the true extent of the emission . 6 . using composite spectra
, we observe a trend of _ decreasing _ * emission with increasing absorption .
this trend is the opposite of what one would naively expect , if the * photons are originally absorbed in the transitions .
we attribute this trend to a combination of increased extinction in more massive galaxies and more extended outflows in massive galaxies , leading to greater slit losses for * photons .
two - dimensional composite spectra of high and low mass galaxies indicate that emission from the outflow is stronger at a radius of @xmath193 kpc in high mass galaxies than in low mass galaxies .
our results suggest that and * emission provide a valuable but so far little - utilized tool for the study of galactic outflows .
in particular , as telescope capabilities develop , studies of the spatial extent of these lines may allow important measurements of the physical extent of outflowing gas .
these lines are accessible from the ground over a wide range in redshift , from @xmath21 to @xmath4 , and complement the far - uv lines studied in galaxies at @xmath267 and above . ,
s. m. , phillips , a. c. , kibrick , r. i. , alcott , b. , allen , s. l. , burrous , j. , cantrall , t. , clarke , d. , coil , a. l. , cowley , d. j. , davis , m. , deich , w. t. s. , dietsch , k. , gilmore , d. k. , harper , c. a. , hilyard , d. f. , lewis , j. p. , mcveigh , m. , newman , j. , osborne , j. , schiavon , r. , stover , r. j. , tucker , d. , wallace , v. , wei , m. , wirth , g. , & wright , c. a. 2003 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol .
4841 , society of photo - optical instrumentation engineers ( spie ) conference series , ed .
m. iye & a. f. m. moorwood , 16571669 , j. a. , cooper , m. c. , davis , m. , faber , s. m. , coil , a. l. , guhathakurta , p. , koo , d. c. , phillips , a. c. , conroy , c. , dutton , a. a. , finkbeiner , d. p. , gerke , b. f. , rosario , d. j. , weiner , b. j. , willmer , c. n. a. , yan , r. , harker , j. j. , kassin , s. a. , konidaris , n. p. , lai , k. , madgwick , d. s. , noeske , k. g. , wirth , g. d. , connolly , a. j. , kaiser , n. , kirby , e. n. , lemaux , b. c. , lin , l. , lotz , j. m. , luppino , g. a. , marinoni , c. , matthews , d. j. , metevier , a. , & schiavon , r. p. 2012 , arxiv:1203.3192 , b. j. , coil , a. l. , prochaska , j. x. , newman , j. a. , cooper , m. c. , bundy , k. , conselice , c. j. , dutton , a. a. , faber , s. m. , koo , d. c. , lotz , j. m. , rieke , g. h. , & rubin , k. h. r. 2009 , , 692 , 187 l c c c c c c c c c c c c @xmath140 & 96 & 24 & 24 & 24 & 24 & 24 & 24 & 24 & 24 & 63 & 33 + @xmath268 & @xmath269 & @xmath270 & @xmath271 & @xmath272 & @xmath273 & @xmath274 & @xmath275 & @xmath276 & @xmath277 & @xmath278 & @xmath279 + @xmath280 & @xmath281 & @xmath282 & @xmath283 & @xmath284 & @xmath285 & @xmath286 & @xmath287 & @xmath288 & @xmath289 & @xmath290 & @xmath291 + @xmath292 & @xmath293 & @xmath294 & @xmath295 & @xmath296 & @xmath297 & @xmath298 & @xmath299 & @xmath300 & @xmath301 & @xmath302 & @xmath303 + @xmath304 & @xmath305 & @xmath306 & @xmath307 & @xmath308 & @xmath309 & @xmath310 & @xmath311 & @xmath312 & @xmath313 & @xmath314 & @xmath315 + @xmath316 & @xmath317 & @xmath318 & @xmath294 & @xmath319 & @xmath320 & @xmath321 & @xmath322 & @xmath323 & @xmath324 & @xmath325 & @xmath326 + @xmath327 & @xmath328 & @xmath329 & @xmath330 & @xmath331 & @xmath332 & @xmath333 & @xmath334 & @xmath335 & @xmath336 & @xmath337 & @xmath338 + @xmath339 & @xmath340 & @xmath341 & @xmath342 & @xmath343 & @xmath344 & @xmath345 & @xmath346 & @xmath347 & @xmath348 & @xmath349 & @xmath350 + @xmath351 & @xmath352 & @xmath353 & @xmath354 & @xmath355 & @xmath356 & @xmath357 & @xmath358 & @xmath359 & @xmath360 & @xmath361 & @xmath362 + @xmath363 & @xmath364 & @xmath365 & @xmath366 & @xmath367 & @xmath368 & @xmath369 & @xmath370 & @xmath371 & @xmath370 & @xmath319 & @xmath372 + @xmath373 & @xmath374 & @xmath375 & @xmath376 & @xmath377 & @xmath378 & @xmath379 & @xmath380 & @xmath381 & @xmath382 & @xmath383 & @xmath384 + @xmath385 & @xmath386 & ... & @xmath387 & @xmath388 & ... & ... & ... & ... & ... & @xmath389 & ... + @xmath390 & @xmath391 & ... & ... & @xmath392 & ... & @xmath393 & ... & ... & ... & @xmath394 & ... + @xmath395 & @xmath396 & @xmath397 & @xmath398 & @xmath399 & @xmath400 & @xmath401 & @xmath402 & @xmath403 & @xmath404 & @xmath405 & @xmath406 + @xmath407 & @xmath408 & @xmath409 & @xmath410 & @xmath411 & @xmath412 & @xmath413 & @xmath414 & @xmath415 & @xmath416 & @xmath417 & @xmath418 + @xmath419 & @xmath420 & @xmath421 & @xmath422 & @xmath423 & @xmath424 & @xmath425 & @xmath426 & @xmath427 & @xmath428 & @xmath429 & @xmath430 + @xmath431 & @xmath432 & @xmath433 & @xmath408 & @xmath434 & @xmath435 & @xmath436 & @xmath437 & @xmath438 & @xmath439 & @xmath440 & @xmath441 + @xmath442 & @xmath443 & @xmath444 & @xmath445 & @xmath446 & @xmath447 & @xmath448 & @xmath449 & @xmath450 & @xmath451 & @xmath452 & @xmath453 + @xmath454 & @xmath455 & @xmath456 & @xmath457 & @xmath458 & @xmath459 & @xmath460 & @xmath461 & @xmath462 & @xmath463 & @xmath464 & @xmath465 + @xmath466 & @xmath467 & @xmath468 & @xmath469 & @xmath470 & @xmath471 & @xmath472 & @xmath473 & @xmath474 & @xmath475 & @xmath476 & @xmath477 + @xmath478 & @xmath479 & ... & @xmath480 & @xmath481 & ... & @xmath482 & @xmath483 & ... & @xmath484 & @xmath485 & @xmath486 l c c c c c c c c c c c c @xmath140 & 96 & 24 & 24 & 24 & 24 & 24 & 24 & 24 & 24 & 63 & 33 + @xmath487}\lambda2326}$ ] & @xmath488 & @xmath489 & @xmath490 & @xmath491 & @xmath492 & @xmath493 & @xmath494 & @xmath495 & @xmath496 & @xmath497 & @xmath498 + @xmath499 & @xmath500 & @xmath501 & @xmath502 & @xmath503 & @xmath504 & @xmath505 & @xmath506 & @xmath501 & @xmath507 & @xmath504 & @xmath508 + @xmath509 & @xmath510 & @xmath511 & @xmath512 & @xmath513 & @xmath514 & @xmath515 & @xmath516 & @xmath517 & @xmath518 & @xmath519 & @xmath520 + @xmath521 & @xmath522 & @xmath516 & @xmath502 & @xmath523 & @xmath524 & @xmath525 & @xmath526 & @xmath526 & @xmath527 & @xmath528 & @xmath529 + @xmath530 & @xmath531 & @xmath532 & @xmath533 & @xmath534 & @xmath535 & @xmath536 & @xmath537 & @xmath538 & @xmath539 & @xmath540 & @xmath541 + @xmath542 & @xmath543 & @xmath544 & @xmath545 & @xmath546 & @xmath547 & @xmath548 & @xmath549 & @xmath550 & @xmath551 & @xmath552 & @xmath553 + @xmath351 & @xmath554 & @xmath555 & @xmath556 & @xmath557 & @xmath558 & @xmath559 & @xmath560 & @xmath561 & @xmath562 & @xmath562 & @xmath563 + @xmath363 & @xmath564 & @xmath565 & @xmath556 & @xmath507 & @xmath566 & @xmath567 & @xmath568 & @xmath500 & @xmath568 & @xmath566 & @xmath569 + @xmath570}\lambda2326}$ ] & @xmath571 & @xmath572 & ... & @xmath573 & ... & @xmath574 & @xmath575 & @xmath576 & ... & @xmath577 & @xmath578 + @xmath579 & ... & @xmath580 & ... & @xmath581 & ... & @xmath582 & @xmath583 & @xmath584 & ... & ... & @xmath585 + @xmath586 & @xmath587 & @xmath588 & @xmath589 & @xmath590 & @xmath591 & @xmath592 & @xmath593 & @xmath594 & @xmath595 & @xmath596 & @xmath597 + @xmath598 & @xmath599 & @xmath600 & ... & @xmath601 & @xmath602 & @xmath603 & ... & ... & ... & @xmath604 & @xmath605 + @xmath606 & @xmath607 & @xmath608 & @xmath609 & @xmath610 & @xmath611 & @xmath612 & @xmath613 & @xmath614 & @xmath615 & @xmath616 & @xmath617 + @xmath618 & @xmath619 & @xmath620 & ... & @xmath621 & ... & @xmath622 & @xmath623 & @xmath624 & ... & ... & @xmath625 + @xmath454 & ... & @xmath626 & ... & @xmath627 & ... & ... & ... & @xmath628 & ... & ... & @xmath629 + @xmath466 & ... & ... & ... & ... & ... & @xmath630 & ... & ... & ... & ... & @xmath631 | we study large - scale outflows in a sample of 96 star - forming galaxies at @xmath0 , using near - uv spectroscopy of and absorption and emission .
the average blueshift of the interstellar absorption lines with respect to the systemic velocity is @xmath1 km s@xmath2 at @xmath3 , with standard deviation 87 km s@xmath2 ; this is a decrease of a factor of two from the average blueshift measured for far - uv interstellar absorption lines in similarly selected galaxies at @xmath4 .
the profiles of the @xmath52796 , 2803 lines show much more variety than the profiles , which are always seen in absorption ; ranges from strong emission to pure absorption , with emission more common in galaxies with blue uv slopes and at lower stellar masses .
outflow velocities , as traced by the centroids and maximum extent of the absorption lines , increase with increasing stellar mass with 23@xmath6 significance , in agreement with previous results .
we study fine structure emission from * , finding several lines of evidence in support of the model in which this emission is generated by the re - emission of continuum photons absorbed in the resonance transitions in outflowing gas .
in contrast , photoionization models indicate that emission arises from the resonant scattering of photons produced in regions , accounting for the differing profiles of the and lines .
a comparison of the strengths of the absorption and * emission lines indicates that massive galaxies have more extended outflows and/or greater extinction , while two - dimensional composite spectra indicate that emission from the outflow is stronger at a radius of @xmath710 kpc in high mass galaxies than in low mass galaxies . |
lll techniques & advantages & disadvantages + optical rotation in & no electric and magnetic fields & unavoidable systematic effects , + atomic vapor @xcite & are involved , no frequency & poor signal to noise ratio at zero + & measurements & crossing in the dispersion curve + stark interference & measurement procedure is & measured transitions are + in atomic vapor @xcite & relatively simple & doppler broadened + stark interference & doppler broadening is reduced , & limited by volume and time + in atomic beams & signal to noise ratio is larger & of interaction , coherence + @xcite & due to large no
. of atoms & time is short due to collision + @xmath0light - shift in single & absence of doppler broadening , & accurate determination of the + trapped and laser & tractable systematic , long & electric field of the light at the + cooled ion @xcite & coherence time , large signal to & position of the ion in the trap + & noise ratio & + @xmath0stark interference & large signal to noise ratio & less systematic from collision + with small number & & broadening + of atoms @xcite & & + weak interaction between atomic electron and the nucleus through the exchange of z@xmath1 boson leads to parity violation in atomic systems @xcite .
atomic parity violation ( apv ) has become a subject of keen interest as it has the potential to test the standard model ( sm ) of particle physics and to search for new physics beyond it @xcite .
several experiments have been performed over the last three decades on some heavy elements like cs @xcite , pb @xcite , tl @xcite , bi @xcite _ etc_. there are also some proposals with promising prospects on elements like yb @xcite , fr @xcite and atomic ions like ba@xmath2 @xcite and ra@xmath2 @xcite .
one of the most promising candidates is yb whose parity non - conserving ( pnc ) amplitude @xmath3 so far the largest .
this point has also been verified experimentally @xcite but the experimental precision needs to be improved in order to compete with the present bench mark value of cs pnc experiment @xcite .
the experiment on cs with an accuracy of @xmath4 , has successfully explained the sm of particle physics @xcite .
higher precision ( @xmath5 ) is required to search for new physics beyond sm @xcite .
the physical parameter that one seeks by combining these experiments and theory is the pnc transition amplitude @xmath3 . in table
[ tab1 ] presently available techniques have been mentioned along with their advantages and respective challenges .
a single trapped and laser cooled ion is free from unknown perturbations and it has long coherence time .
systematic uncertainties are easily tractable and therefore , the system is more favored for such experiment @xcite even though this has not yet been experimentally demonstrated .
single ion trapping and laser cooling are routinely done in radio frequency paul traps @xcite .
the possibility of apv experiment based on such a system was first put forward by fortson @xcite .
the overall idea has been reviewed here in brief focusing ra@xmath2 as a possible candidate . in fig .
[ fig1 ] the relevant energy levels of singly charged radium ( ra@xmath2 ) and barium ( ba@xmath2 ) have been shown .
after confining radium ion in an rf paul trap , it can be laser cooled by exciting the @xmath6 transition at @xmath7 nm .
a repumping laser at @xmath8 nm is necessary to bring the ion back to the cooling cycle from the metastable @xmath9 state .
atomic parity violation leads to mixing of different parity states with the ground @xmath10 state .
thus the ground state has a small contribution from @xmath11 state resulting in a non - zero probability of dipole transition between @xmath10 and @xmath9 states which is normally a forbidden electric dipole transition . and ( b ) ba@xmath2,scaledwidth=85.0% ]
a transitional dipole interacts with the electric field while a quadrupole interacts with the field gradient . in an experimental setup as shown in fig .
[ fig2 ] , it is possible to induce both a dipole transition ( due to apv ) as well as a quadrupole transition between @xmath10 and @xmath9 states . the interference term of these two leads to a measurable frequency change of the larmor frequency between the ground state zeeman sublevels in presence , as compared to , in absence of the laser fields .
one of the suitable laser field configurations that produce the needed apv frequency shift is @xmath12 & @xmath13 where @xmath14 and @xmath15 are the electric field amplitudes of the two lasers .
an ion placed at the antinode of @xmath16 field will suffer pnc induced electric dipole light - shift while the ion placed at the node of @xmath17 field , will show electric quadrupole light - shift .
the quadrupole light - shifts of the zeeman sublevels in the ground state due to the @xmath17 field are of the same magnitude and direction .
therefore , @xmath17 field will not lead to any change of the ground state larmor frequency defined by the energy difference between the zeeman sublevels of the ground state . on the contrary , the shifts due to @xmath16 field will increase the larmor frequency .
this change in larmor frequency is proportional to the magnitude of the @xmath16 field . in the experiment one measures the larmor frequency with and without these laser fields .
the difference of these two frequencies therefore , gives directly the apv light - shift @xmath18 which can be expressed as @xcite @xmath19 where @xmath20 and @xmath21 are the rabi frequencies for pnc and quadrupole induced transitions which are respectively proportional to the electric field amplitude and field gradient of the standing wave lasers , @xmath22 ; @xmath23 , @xmath24 are the zeeman sublevels of @xmath25 and @xmath26 states respectively .
the distinguished advantage of this technique is that measurement of @xmath18 is free from any fluctuation in the laser frequency and other sources of quadrupole shift ( @xmath27 ) .
the statistical uncertainty in the measurement of @xmath3 is given by @xmath28 where @xmath29 is an efficiency factor that depends on experimental conditions , @xmath30 and @xmath31 are the number of ions and coherence time respectively and @xmath32 is the time of observation . though @xmath33 in this experiment , longer coherence time improves the uncertainty in the measurement .
accurate determination of @xmath3 from measured @xmath18 depends on precise determination of the electric fields @xmath16 and @xmath17 at the position of the ion in the trap which is a challenge of this experiment and is a major source of systematic uncertainty . also placing
the ion at the antinode of @xmath16 field or node of @xmath17 field is a difficult task .
since pnc induced light - shift ( @xmath18 ) is very small ( @xmath34 cycles / s for @xmath35 v / m for ra@xmath2 ) , little fluctuations of the magnetic field resolving zeeman sublevels , will worsen the accuracy of the result . however , in the past decades several experimental techniques have been developed by which the above problems can be solved .
one can manipulate the ion position with respect to the standing wave .
recently , a technique has been reported @xcite by which the nodal point or the nodal line of the trap can be shifted upto few micrometers . thus the ion can be placed on the geometrical line of the standing wave .
systematic uncertainty originating from inaccurate positioning of the ion with respect to the electric field of the laser has been determined in the following .
since the ion is cooled to lamb - dicke regime , its motion is confined within its de broglie wavelength ( @xmath36 ) , typically @xmath37 nm for ra@xmath2 and ba@xmath2 .
the uncertainty in @xmath16 and @xmath17 fields for antinodal and nodal positions respectively are @xmath38 and @xmath39 these systematic uncertainties tabulated in table [ tab2 ] , have been estimated from the lamb - dicke parameter .
there are also some techniques for controlling magnetic field fluctuations .
recent improvement of high precision rf spectroscopic techniques @xcite opens up the possibility of success of the desired experiment in near future .
lll atomic properties & @xmath40ba@xmath2 & @xmath41ra@xmath2 + stability ( neutral specis ) & stable & @xmath42 years ( @xmath43 ) + @xmath3 in @xmath44 & @xmath45 @xcite & @xmath46 @xcite + pnc light - shift ( @xmath47 ) ( @xmath48 ) & @xmath49 & @xmath50 + coherence time @xmath51 ( s ) & @xmath52 & @xmath53 + @xmath0statistical uncertainty ( @xmath54 ) & @xmath5 & @xmath55 + @xmath56systematic uncertainty from @xmath16 field ( eq .
5 ) & @xmath57 & @xmath58 + @xmath56systematic uncertainty from @xmath17 field ( eq .
6 ) & @xmath59 & @xmath60 + quenching rate of @xmath61@xcite & 0.002 & 0.04 + quenching rate of @xmath62@xcite & 0.0033 & 3.36 + quenching rate of @xmath63@xcite & 0.0004 & 0.48 +
atomic parity violation effect scales little faster than @xmath64 @xcite for heavier element . in @xmath41ra@xmath2
it is @xmath65 times larger as compared to @xmath66ba@xmath2 @xcite and @xmath37 times larger than that of atomic cesium @xcite .
that is why at first sight @xmath41ra@xmath2 is seemed to be a promising candidate though there are other advantages over ba@xmath2 .
the most recent calculation shows that the pnc amplitude present in @xmath41ra@xmath2 is @xmath46 in the unit of @xmath44 @xcite , where @xmath67 is the weak charge . in table
[ tab2 ] the relevant atomic properties of @xmath40ba@xmath2 and @xmath41ra@xmath2 have been compared .
the lasers required for @xmath41ra@xmath2 are in visible and near infra - red region .
thus these lasers are available commercially as solid state diode lasers .
radium being a heavier element may be confined within smaller orbit than barium as the lamb - dicke parameter is inversely proportional to the square root of mass of the ion . in addition , the known relative systematic uncertainties for ra@xmath2 are three times smaller as compared to ba@xmath2 .
the element has a large number of isotopes with significant stability , thus opening up the possibility of the experiment to extract the effect of nuclear structure in apv @xcite .
pnc amplitude @xmath3 contains both nuclear spin dependent ( nsd ) and independent ( nsi ) parts @xcite . from nsd part of @xmath3 , nuclear anapole moment
can be measured @xcite .
however , in @xmath68 transition in ra@xmath2 or ba@xmath2 , the contribution of nsd part is smaller by few orders than nsi part and hence determination of anapole moment is difficult . to avoid nsi part , a similar experiment explained above can be performed using @xmath69 transition of nuclear spin non - zero ( @xmath70 ) isotopes of ra@xmath2 or ba@xmath2 .
pnc allowed @xmath69 transition in these isotopes contains only nsd part and may lead to a direct measurement of nuclear anapole moment .
@xmath71ra@xmath2 is more favored candidate for an apv experiment than @xmath66ba@xmath2 as pnc amplitude for @xmath72 transition is @xmath73 times larger in this isotope @xcite .
however , there are several disadvantages of choosing ra@xmath2 as a possible candidate . trapping and
cooling of ra@xmath2 has not been demonstrated so far .
the lack of spectroscopic data on ra@xmath2 is also a problem .
theoretical calculation needs to be more accurate ( below 1 % ) in order to compare with the experimentally obtained data .
the atomic structure of radium is not well studied .
coherence time is smaller for ra@xmath2 ( @xmath74 ) which will reduce the signal to noise ratio .
the systematic uncertainties originating from the determination of @xmath16 and @xmath17 at the ion position are too large for ra@xmath2 and demand some special experimental techniques to eliminate those .
the production of various isotopes of radium for the study of nuclear structure effects on apv demands well established facilities . at the kvi ,
groningen such a facility has been developed @xcite where some isotopes of radium may be produced with an aim for performing apv experiment based on single trapped and laser cooled ra@xmath2 .
atomic ra has been successfully trapped in a mot and laser cooled @xcite at argonne national laboratory in search of permanent electric dipole moment ( edm ) in atoms .
thus there is hope for details spectroscopic data on ra to be available shortly which will lead towards the implementation of apv experiment on ra@xmath2 .
with an aim to perform high precision rf spectroscopy in search of apv , work has been started by our group rcamos at iacs .
@xmath40ba@xmath2 has been chosen initially as barium is available commercially .
a linear paul trap has been designed .
the repumping laser at @xmath75 nm has been frequency stabilized using pound - drever - hall locking technique @xcite and frequency doubling of @xmath76 nm laser @xcite to produce cooling laser at @xmath77 nm is processing .
part of this work is being supported by dst - serc , india .
p. mandal and a. sen are thankful to csir , india for entertaining fellowship during the research work .
p. mandal acknowledges the organizing committee of laser 2009 workshop , poznan for sponsoring financial support to participate in the workshop .
spphys fortson , n. : phys .
lett . * 70 * , 2383 ( 1993 ) bouchait , m. a. , bouchait , c. : phys .
b * 48 * , 111 ( 1974 ) , rep . prog
. phys . * 60 * , 1351 ( 1997 ) ginges , j. s. m. , flambaum , v. v. : phys . rep . * 397 * , 63 ( 2004 ) , marciano , w. j. , j. l. rosner , j. l. : phys . rev
. lett . * 65 * , 2963 ( 1990 ) wood , c. s. , bennett , s. c. , cho , d. , masterson , b. p. , roberts , j. l. , tanner , c. e. , wieman , c. e. : science * 275 * , 1759 ( 1997 ) bennett , s. c. , wieman , c. e. : phys .
* 82 * , 2484 ( 1999 ) meekohof , d. m. , vetter , p. , majumdar , p. k. , lamoreaux , s. k. , fortson , e. n. : phys .
lett . * 71 * , 3442 ( 1993 ) vetter , p. a. , meekhof , d. m. , majumder , p. k. , lamoreaux , s. k. , fortson , e. n. : phys .
74 * , 2658 ( 1995 ) macpherson , m. j. d. , zetie , k. p. , warrington , r. b. , stacey , d. n. , hoare , j. p. : phys .
lett . * 67 * , 2784 ( 1991 ) demille , david : phys
lett . * 74 * , 4165 ( 1995 ) bouchait , m. a. : phys .
lett . * 100 * , 123003 ( 2008 ) koerber , t. w. , schacht , m. , nagourney , w. , fortson , e. n. : j. phys .
b : at . mol .
. phys . * 36 * , 637 ( 2003 ) tsigutkin , k. , dounas - frazer , d. , family , a. , stalnaker , j. e. , yashchuk , v. v. , budker , d. : phys .
lett . * 103 * , 071601 ( 2009 ) langacker , p. , luo , m. , mann , a. k. : rev .
* 64 * , 87 ( 1992 ) majumdar , p. k. , tsai , l. l. : phys .
a * 60 * , 267 ( 1999 ) leibfried , d. , blatt , r. , monroe , c. , wineland , d. : rev .
* 75 * , 281 ( 2003 ) herskind , p. f. , dantan , a. , albert , m. , marler , j. p. , drewsen , m. : j. phys .
b : at . mol .
phys . * 42 * , 154008 ( 2009 ) sherman , j. a. , andalkar , a. , nagourney , w. , fortson , e. n. : phys .
a * 78 * , 052514 ( 2008 ) sherman , j. a. , koerber , t. w. , markhotok , a. , nagourney , w. , fortson , e. n. : phys .
* 94 * , 243001 ( 2005 ) koerber , t. w. , schacht , m. h. , hendrickson , k. r. g. , nagourney , w. , fortson , e. n. : phys .
lett . * 88 * , 143002 ( 2002 ) sahoo , b. k. , chaudhuri , r. , das , b. p. , mukherjee , d. : phys .
rev . lett . * 96 * , 163003 ( 2006 ) dzuba , v. a. , flambaum , v. v. , ginges , j. s. m. : phys .
d * 66 * , 076013 ( 2002 ) wansbeek , l. w. , sahoo , b. k. , timmermans , r. g. e. , jungmann , k. , das , b. p. , mukherjee , d. : phys .
a * 78 * , 050501(r ) ( 2008 ) fortson , e. n. , pang , y. , wilets , l. : phys . rev . lett . *
65 * , 2857 ( 1990 ) geetha , k. p. , singh , a. d. , das , b. p. : phys .
a * 58 * , r16 ( 1998 ) koerber , t. w. , thesis , doctor of philosophy , university of washington ( 2003 ) shidling , p. d. , giri , g.s . ,
vanderhoek , d.j . ,
jungmann , k. , kruithof , w. , onderwater , c.j.g . , sohani , m. , versolato , o.o . ,
willmann , l. , wilschut , h.w .
instrum . meth .
a * 606 * , 305 ( 2009 ) guest , j. r. , scielzo , n. d. , ahmad , i. , bailey , k. , greene , j. p. , holt , r. j. , lu , z. t. , oconnor , t. p. , potterveld , d. h. : phys .
* 98 * , 093001 ( 2007 ) drever , r. w. p. , hall , j. l. , kowalski , f. v. , appl .
b * 31 * , 97 ( 1983 ) raab , c. , bolle , j. , oberst , h. , eschner , j. , schmidt - kaler , f. , blatt , r. : appl
b * 67 * , 683 ( 1998 ) | single trapped and laser cooled radium ion as a possible candidate for measuring the parity violation induced frequency shift has been discussed here .
even though the technique to be used is similar to that proposed by fortson @xcite , radium has its own advantages and disadvantages .
the most attractive part of radium ion as compared to that of barium ion is its mass which comes along with added complexity of instability as well as other issues which are discussed here .
example.eps gsave newpath 20 20 moveto 20 220 lineto 220 220 lineto 220 20 lineto closepath 2 setlinewidth gsave .4 setgray fill grestore stroke grestore |
after @xmath3and @xmath4 , the most abundant nucleus in a @xmath5 white dwarf ( wd ) interior is @xmath0 .
the reason for this starts with the physics of the cno cycle in the hydrogen burning phase of a star s life .
the slowest step in the cno cycle is the proton capture onto @xmath6and almost all of the cno catalyst nuclei end up as @xmath6at the completion of @xmath7burning .
the @xmath6@xmath8@xmath9@xmath10@xmath11@xmath12@xmath0reaction sequence during the subsequent helium burning then processes all the @xmath6into @xmath0 .
this results in a @xmath0mass fraction of @xmath13 for recently formed wds of @xmath5 ( see , for example , @xcite ) made from stars of initial mass @xmath14 .
the possibility that @xmath0could play a role in the energetics of a cooling wd was first noted by @xcite .
they discussed the possibility of gravitational energy release if @xmath0phase separates at core crystallization .
this was followed by more detailed studies @xcite which differed in their conclusions about the final state once crystallization was complete ; ranging from all the @xmath0 in the stellar center to most of it having an unchanged profile .
@xcite ( hereafter bh01 ) examined another aspect of @xmath0which had been previously overlooked ( except for a brief mention in bravo et al .
1992 ) , namely its ability to sink through the liquid wd interior .
as discussed in bh01 , there is an upward pointing electric field of magnitude @xmath15 in the degenerate wd interior , where @xmath1 is the local gravitational acceleration and @xmath16 is the proton mass . the net force on @xmath0is therefore @xmath17 , biasing the diffusion of @xmath0inward .
it is the excess neutrons of the @xmath0nucleus ( relative to the predominant @xmath18 nuclei , @xmath19 being the atomic mass number and @xmath20 the charge ) that make it special in this regard .
bh01 found that the gravitational energy released by sedimenting @xmath0is comparable to the thermal content of the wd at an age of a few gyrs .
this raised the question of whether @xmath0could settle fast enough to significantly affect the wd s cooling significantly .
the rate at which @xmath0sinks in the liquid c / o wd depends on the @xmath0diffusion coefficient , @xmath21 .
the conditions there are very non - ideal , as the average coulomb energy for the ions is comparable to the thermal energy , or @xmath22 , where @xmath23 .
as discussed by @xcite , there is substantial ( factors of many ) uncertainty in the diffusion coefficients in these liquid regimes , as the familiar notions of mean free path lose their meaning . in the absence of a definitive calculation of @xmath21 for this situation , bh01 proceeded by estimating @xmath21 for @xmath0 in a c / o plasma by the self - diffusion coefficient , @xmath24 , of the classical one - component plasma ( ocp ) . with this , bh01 then estimated the power released , @xmath25 , by @xmath0sinking for a fixed @xmath0profile .
this calculation suggested that @xmath0sedimentation might release sufficient energy to impact wd cooling .
we now take the next step in addressing this question by performing a self - consistent evolution of both the @xmath0density , @xmath26 , and the wd core temperature , @xmath27 , in wds composed of a single dominant ion species .
we find that @xmath0heating delays the time it takes a wd to cool to a given luminosity . _
the total increase in cooling age by the time the wd completely crystallizes ranges from 0.25 - 1.6 gyr , depending on the value of @xmath21 and the wd mass_. we also investigate the seismological impact of the @xmath0abundance profile at the time the wd crosses the zz ceti instability strip .
the gradient in @xmath0abundance produces a gradient in the electron mean molecular weight , @xmath28 , that provides an additional buoyancy and alters the brunt - visl frequency , @xmath29 .
_ this contribution alters the pulsation periods of high radial order g - modes by more than the measurement errors_. thus @xmath28 gradient contributions from sedimenting @xmath0cannot be ignored in precision wd pulsation work such as recent contraints on the interior abundance profiles @xcite or @xmath3@xmath12@xmath4reaction rate @xcite .
an unexpected result of our calculations is the interaction between the infalling @xmath0and the outward moving crystal / fluid boundary as the wd cools .
we assume that sedimentation halts in the crystalline interior , forcing @xmath0to accumulate at the crystal / fluid boundary and elevating the abundance there .
this abundance is then frozen as the crystal front moves outward .
the jump in @xmath0abundance at a given location depends on both the rate of @xmath0infall and the rate at which the crystal front moves outward .
this leads to an @xmath0abundance in the crystal regions of the star that _ increases _ outward .
whether or not this profile is subject to an elastic rayleigh - taylor instability remains an open question . in @xmath30
[ sec : methods ] , we cover the details of how we model the evolution of the @xmath0density , @xmath26 , and the wd core temperature , @xmath27 . we also discuss the current uncertainties in the ocp self - diffusion coefficient used to calculate our @xmath0flow rates .
the details of the numeric evolution of @xmath26 are in @xmath30[sec : diffusion ] . in [ sec : cool ] we detail the results of our self - consistent calculations .
we first discuss the evolution of @xmath26 , highlighting the influence of @xmath21 , wd mass , and crystallization on the adundance profiles .
we then turn to the thermal evolution of the wd and present our new cooling curves . with the @xmath0abundances in hand
, we estimate in [ sec : brunt ] the impact @xmath0can have on wd pulsations , focusing on high radial order g - modes .
we close in [ sec : conclusions ] with a summary of our results and a discussion of unanswered questions .
the appendix discusses the brunt - visl frequency in the deep interior .
the wd interior is a plasma of very degenerate electrons whose corresponding ions may be either in a liquid or crystalline state . for our calculation , we presume that the wd is composed a one component plasma ( ocp ) of ions whose @xmath19 is between 12 and 16 and whose @xmath31 .
we treat @xmath0as a trace species throughout the entire calculation , therefore neglecting its effect on the hydrostatic structure of the wd as the models evolve .
as can be seen from the initial composition profile of a realistic pre - wd model shown in figure [ fig : xinit ] , the @xmath0trace assumption is valid initially and we find that it remains valid at all times considered in our calculations .
the physical state of the ions is specified by the parameter , @xmath32 where @xmath33 g @xmath34 . for @xmath35 ,
the plasma is in a liquid state .
the plasma undergoes a phase transition from the liquid to crystalline states once @xmath36 @xcite .
the @xmath0concentration evolution is governed by mass continuity @xmath37 where @xmath26 and @xmath38 are the local @xmath0density and @xmath0flux respectively .
the flux has two pieces ; the diffusive flux caused by @xmath0density gradients and the drifting flux caused by the inward force , @xmath39 , on @xmath0 , @xmath40 where @xmath21 is the diffusion coefficient of @xmath0 in the background plasma and @xmath41 is the magnitude of the local @xmath0drift velocity .
we assume that @xmath0is no longer able to sink once the wd has crystallized and fix the @xmath0concentration of the crystal to be that of the fluid just prior to crystallization .
thus we do not consider concentration changes due to fractionation upon freezing ( see , for example , @xcite ) .
as mentioned in bh01 , the microphysical calculation of the diffusion coefficient of @xmath0through a plasma that is representative of wd interior conditions has not yet been performed .
a proper treatment involves the calculation of a trace species diffusing through a multi - component plasma in both the classical and quantum liquid regimes .
the latter will be important in the higher mass wds , where the core becomes a quantum liquid long before crystallization starts ( i.e. while the core is still liquid , @xmath42 ; @xcite ) . in light of these uncertainties , bh01 estimated @xmath21 in two ways . the first was using the stokes - einstein relation for a particle of radius @xmath43 ( taken to be the radius of the charge neutral sphere around @xmath0 ) undergoing brownian motion in a fluid of viscosity @xmath44 , which gives @xmath45 when the fluid is allowed to `` slip '' at the particle / fluid interface .
this estimate works well in liquids , where it is accurate at atomic dimensions @xcite .
the second was to use the self - diffusion coefficient @xmath46 calculated by @xcite for the ocp , where @xmath47 is the ion plasma frequency .
these two estimates differ by 20 - 40 % .
in addition , there is still uncertainty in the ocp self - diffusion coefficient in the regime of interest . to highlight these uncertainties in @xmath24
, we consider calculations of the ocp shear viscosity , @xmath44 , which are related to the ocp self - diffusion coefficient via the stokes - einstein relation .
a summary of shear viscosity calculations for the ocp are presented in figure [ fig : viscosity ] , where we show the reduced viscosity , @xmath48 as a function of @xmath49 .
the results represent a range in calculation techniques
monte carlo @xcite , kinetic theory @xcite , and molecular dynamics @xcite .
@xcite considered a glassy state above the normal crystal transition point , thus their data extends to @xmath50 .
error bars are displayed when there was sufficient information in the literature to determine them . between these results , there is disagreement on the 10 - 30% level .
the solid line displays the analytic result of the ocp ( no - screening ) limit of the yukawa system as calculated by @xcite .
the dotted line shows our fit to the data that reproduces the reduced viscosity implied from @xmath24 of equation ( [ eq : diff ] ) and the stokes - einstein relation in the regime @xmath51 while approximately fitting the data in the of tanaka & ichimaru for @xmath52 .
this fit is given by @xmath53 the first term in equation ( [ eq : etaglfit ] ) is the relation between @xmath54 and @xmath49 inferred from equation ( [ eq : diff ] ) .
we utilize this fit to explore the consequences of wd interiors not crystallizing but instead undergoing a transition to a glassy state . with all of these uncertainties in mind ,
we follow bh01 and take @xmath55 as our fiducial best guess for the @xmath0diffusion coefficient .
but , as we consider the actual value of @xmath21 to be uncertain even in the case of @xmath0diffusing in a single component plasma , we carry out calculations with @xmath55 , @xmath56 , and @xmath57 to determine to what extent a reasonable variation in the diffusion rate may alter the impact of @xmath0sedimentation . although the actual @xmath21 may be less than @xmath24 , we only consider @xmath58 here since for @xmath59 , the impact of @xmath0on wd evolution is irrelevant .
the variation of up to 10 times @xmath24 is probably large for the more typical low mass wds , but for the more massive wds , where the interior plasma can be a quantum liquid over a substantial portion of the wd before it crystallizes , the use of the classical ocp formula may be seriously in error .
once @xmath21 is specified , the downward drift velocity , @xmath41 , of the @xmath0ions under a macroscopic force @xmath60 is @xmath61 where the latter equality holds for @xmath62 .
this is the approximation we use to determine @xmath41 in our calculations . as @xmath63 over most of the stellar radius , the motion of @xmath0is dominated by falling at the local drift velocity set by equation ( [ eq : v ] ) ( bh01 ) .
equation ( [ eq : v ] ) also allows a rough estimate of the sedimentation timescale .
the time it takes a @xmath0ion to sink from the surface to the center of a constant density star when it falls at the rate given by equation ( [ eq : v ] ) is @xmath64 ( bh01 ) .
using the central densities as an estimate , a pure oxygen(carbon ) @xmath65 wd at @xmath66 has @xmath67 , so that complete @xmath0sedimentation is unlikely for the most common @xmath68 wds .
however , the strong density dependence means that this time decreases for massive wds .
for example , a pure oxygen @xmath69 wd at @xmath70 has @xmath71 .
one caveat of our specification of the relation between @xmath41 and @xmath21 should be mentioned .
our wd model is constructed using a fully degenerate eos everywhere ( with no inclusion of ion contributions in the structure calculation ) and we take initially the @xmath0mass fraction to be @xmath72 = 0.02 everywhere . this choice of model and @xmath0profile confines us to the wd interior where the electron fermi energy @xmath73 .
modeling the outer non - degenerate parts of the star requires a more accurate background model and electric field calculation .
since we are most interested here in the overall energetics , we can safely neglect the role of the thin outer layers .
our wd model consists of a plasma of a single predominant ion species plus a trace amount of @xmath0at a local number density of @xmath74 .
the energetics included are the heat provided by @xmath0sedimentation , latent heat released upon crystallization of the major species , and the wd s radiative losses .
since we are modeling the late ( @xmath75 myr ) evolution , we neglect the slight gravitational energy release as the wd contracts to its final radius .
also , other potential heat sources such as the potential h burning in the inner envelope and fractionation effects are not included for the sake of simplicity as our goal here is to demonstrate the potential impact of @xmath0sedimentation separated from other sources of uncertainty in wd cooling . the _
net _ power generated per @xmath0ion by the sedimentation process is @xmath76 , where @xmath77 is the bulk velocity of the @xmath0flow .
this quantity is different from the drift velocity since we must take into account both diffusive and drifting currents .
thus @xmath78 and the net gravitational energy release is given by @xmath79 where @xmath74 .
the latent heat released upon crystallization is @xmath80 per nuclei @xcite . for the energy lost to radiation
, we use the @xmath81 relations of @xcite . for the heat capacity of the liquid state ,
@xmath82 , we use equation ( 17 ) of @xcite which takes into account the ideal ion and ion - ion interactions .
we neglect the small electron contributions to the heat capacity .
quantum liquid effects are not taken into account . in the solid state
we utilize equation ( 5 ) of @xcite .
since the thermal conduction time across the wd core is much less than the evolution time , we treat all energy sources and sinks as global quantities .
the thermal evolution of the instantaneously isothermal wd at temperature @xmath83 is then @xmath84 where @xmath85 is the mass of the wd that has crystallized .
for a spherically symmetric flow , equation ( [ eq : cont ] ) reads @xmath86 we discretize equation ( [ eq : spcont ] ) according to the prescription @xmath87 @xcite . here
the superscripts refer to the time index and subscripts the spatial index and the factors of two come about from the averaging of @xmath88 between adjacent cells .
the first two terms give the diffusive fluxes through the top and bottom of the mass shell at radius @xmath89 .
the second two terms give the same for the drifting fluxes .
the boundary conditions are @xmath90 at @xmath91 , where @xmath92 is the wd radius .
we use @xmath26 instead of @xmath0concentration as our dependent variable since this simplifies the form of the continuity equation .
the rate at which @xmath0flows through the interior is completely determined by @xmath21 . in the liquid regions of the star , we set @xmath93 ( equation ( [ eq : diff ] ) ) .
the @xmath49 dependence of @xmath24 means that as the wd cools , @xmath21 , and hence @xmath41 , decrease and @xmath0flow slows .
we expect @xmath0diffusion to halt in regions of the wd that have crystallized .
there is thus an abrupt transition in the @xmath0flow rate at the crystal / liquid interface . combined with the fact that the neon concentration is fixed behind the crystal / liquid boundary
, the crystallizing of the wd leaves a significant imprint on the @xmath0concentration profile in the star , as we discuss in [ sec : cool : diffusion ] . in order to avoid numerical difficulties
, we handle the transition in the value of @xmath21 to zero in crystallizing regions by smoothing the transition with the formula @xmath94 here @xmath95 sets the factor by which @xmath21 is reduced from @xmath24 in the crystalline state and @xmath96 sets the width in @xmath49 over which the transition from liquid to crystal occur in the @xmath97 shell . by smoothing out the transition from liquid to crystal states
, we effectively resolve the propagation of the crystal / liquid boundary within a single shell of our grid in the sense that it now takes a finite time for the shell to go from fully liquid to fully crystalline .
this is important because the width of the transition region affects the @xmath0profile near the crystal / liquid boundary .
we choose @xmath98 so that the time it takes for each shell to transition from liquid to crystal ( the transition regime being defined by @xmath99 ) equals the amount of time it takes the crystal front to move across the shell . an easy way of approximating @xmath98 is to consider the @xmath27 at which the adjacent cells in our grid crystallize .
these two temperatures set the lower and upper bounds of the range in @xmath49 in the @xmath100 between which the cell undergoes crystallization . from this @xmath101
is determined by the simple relation @xmath102 where @xmath103 .
for our models , the right hand side of equation ( [ eq : dgamma ] ) is @xmath104 and @xmath105 .
we presume that the latent head is released instantaneously when @xmath106 in the shell . as mentioned prior
, we also utilize the diffusion coefficient implied by the stokes - einstein relation and equation ( [ eq : etaglfit ] ) to calculate the effects of @xmath0diffusion in wds if the interior plasma undergoes a glassy transition instead of crystallizing . in the glassy state
, we expect diffusion to continue .
thus @xmath0sedimentation will continue to generate energy long past the time when sedimentation would have stopped in a crystallized wd .
obviously , such a process has the potential to maintain a wd at a higher luminosity than it otherwise would have at very late times .
we explore this possibility in section [ sec : cool : cool ] . before including the thermal evolution of the wd
, we performed checks on the accuracy of our algorithm at constant @xmath27 in a @xmath107 mass wd .
the motion of @xmath0over most of the wd is dominated by falling at the local drift speed , @xmath41 , allowing an analytical calculation of the @xmath0concentration as a function of time by considering only the drift contributions to the flux .
we consider the @xmath0 in a thin shell ( at radius @xmath108 and thickness @xmath109 ) . from @xmath110
we calculate the position at later times of @xmath0ions starting at the inner and outer boundaries of this shell .
then , the new @xmath26 in the shell can be calculated from the old @xmath26 , @xmath108 , and @xmath109 and the new @xmath108 and @xmath109 .
the resulting @xmath72 at @xmath111 gyr is displayed in figure [ fig : xcomp1 ] along with the full numerical results at the same wd age .
the lower panel of the plot displays the relative difference between the two calculations , which show a good agreement in the inner regions of the wd .
we also compared the numeric and analytic results for the diffusive equilibrium of @xmath0 .
with @xmath21 and @xmath41 specified , the @xmath0profiles in diffusive equilibrium ( @xmath112 ) is given by @xmath113 which gives @xmath114\,,\ ] ] with @xmath115 normalized by total @xmath0mass .
a comparison between this expression and the equilibrium reached in the numeric integration shows very good agreement .
we utilize equation ( [ eq : numcont ] ) and the discretization of equation ( [ eq : tc ] ) @xmath116 to perform the mutual evolution of @xmath27 and @xmath26 numerically . here
@xmath117 , @xmath118 , @xmath119 , and @xmath120 are the luminosity , central temperature , total heat capacity of the wd , and latent heat released during the @xmath100 timestep . in what follows we consider a wd composed of a single ion species as the background .
figure [ fig : xinit ] is typical of the c / o ratio in wd interiors predicted by stellar evolution models .
we can approximate this composition profile by performing calculations taking the entire star to have a mean molecular weight @xmath121 throughout .
therefore we consider in greatest detail the results for models with this composition . in figure
[ fig : d1x22 ] we show the @xmath0profile evolution in the @xmath107 c / o wd when @xmath62 for 5 different epochs .
the initial @xmath0profile is a horizontal line at @xmath122 .
the major qualitative result of the @xmath0evolution is a significant depletion in the @xmath0concentration in the outer layers of the wd , where the rate of sinking is largest . elsewhere through the star , other than a gradual increase with time ,
the concentration is not much affected , at least as long as the wd remains in the liquid state .
this is to be expected since the sedimentation time for a low mass wd is much longer than the cooling time .
the sedimentation rate is affected by both @xmath21 and @xmath123 .
we show in figure [ fig : x221gyr ] a comparison of the @xmath0profiles in our 0.6 @xmath124 c / o wd at 1 gyr for various values of @xmath21 .
as expected , the larger values of @xmath21 lead to more rapid sedimentation and a faster depletion of @xmath0 in the outer layers of the wd and a stronger gradient in the @xmath0concentration as compared to @xmath62 .
since the sedimentation time decreases with density , @xmath0settles out at a faster rate in more massive wds .
figure [ fig : x22m1 ] shows the @xmath0concentrations at several ages for a 1.0 @xmath124 c / o wd .
two effects compared to the lower mass wd are apparent .
first , as expected , the rate of @xmath0sedimentation is greater leading to much faster depletion in the outer regions and accumulation in the central ones .
but , second , the wd begins crystallizing much sooner ( since the coulomb interactions are greater ) and the star finishes crystallizing at a much younger age .
this limits the extent to which @xmath0is able to settle inward before being frozen in place
. the depletion of @xmath0 in the outer layers of the wd may play a role in setting the inter - diffusion properties of @xmath3/@xmath4and @xmath125at the core - envelope boundary .
all three of these species have @xmath126 and the macroscopic forces acting them are approximately equal , at least in the degenerate regions of the wd . in these regions ,
significant mixing of @xmath125and the @xmath3/@xmath4 in the core should occur , at least on timescales much longer than typical diffusion times . however , with @xmath0present , the @xmath28 of the core is slightly larger than that of the envelope and @xmath125may be excluded from penetrating into @xmath0rich regions .
but , as @xmath0is evacuated from the outer layers of the core , the buoyant barrier created by @xmath28 gradients between the core and envelope is removed .
thus @xmath0sedimentation might set a natural limit on the depth to which @xmath125can inter - mix with the core plasma , but further work needs to be done to verify if this is truly the case .
the time , @xmath127 , it takes for the @xmath0to evacuate the outer layers of the wd can be approximated by integrating equation ( 4 ) of bh01 @xmath128 over the outer regions of the wd , where we can take @xmath1 constant and the pressure to be dominated by non - relativistic , degenerate electrons . if we make the further assumption that the layer is isothermal over the time it takes a @xmath0ion to fall between the limits of the integral , we find that @xmath129 where @xmath130 is the value of @xmath49 at the mass coordinate @xmath131 , @xmath132 is the mass outside of the depth of the ion , and @xmath123 is the wd mass . equation ( [ eq : deltm ] ) is equivalent to equation ( 6 ) from bh01 , but expressed in terms of mass coordinates .
of interest is the approximate time that it takes the @xmath0to be depleted from the outer @xmath133 of the wd that is from a region of roughly the same mass as the @xmath125layer on the wd .
taking a 50/50 mixture of @xmath3/@xmath4 in the outer core of the wd , equation ( [ eq : deltm ] ) gives @xmath134 , where @xmath135 at these times in the outer wd .
thus within several gyrs , any barrier to significant @xmath125penetration into the core should be removed .
crystallization leaves an imprint on the @xmath0profile .
the location of the outward moving crystal / fluid boundary can be seen in figure [ fig : d1x22 ] where the jump in @xmath0concentration in the 3 and 4 gyr profiles occurs .
this jump is a result of the halting of @xmath0sedimentation in crystallized regions of the star . at the crystal / fluid interface , @xmath0ions falling through the liquid portion of the star encounter the impenetrable crystal boundary and
as the star cools , the location of the crystal / fluid boundary moves outward , freezing the @xmath0concentration at the level to which it has accumulated .
the question then arises as to what determines the magnitude of the jump in @xmath0concentration at the crystal / fluid boundary , and therefore fixes the @xmath0profile in the crystal portions of the wd .
consider a thin region of thickness @xmath109 in front of the crystal / fluid boundary , located at a radius @xmath136 . at this location , the boundary position is moving outward at a rate @xmath137 , the @xmath0falls towards the interface at a velocity @xmath41 and has a density of @xmath26 . in the time
@xmath138 it takes the crystal boundary to cross the region @xmath109 , the total mass of @xmath0that flows into this region is @xmath139 the @xmath0density in the shell between @xmath140 and @xmath141 after freezing is then @xmath142 or defining @xmath143 @xmath144 since @xmath145 .
this analysis is valid as long as both @xmath26 and @xmath41 are constant over the time it takes to crystallize the shell .
both conditions are satisfied when @xmath146 , i.e. when @xmath147 .
equation ( [ eq : v ] ) gives @xmath41 as a function of @xmath148 , the velocity,@xmath149 , is calculated by considering the density , @xmath150 , at which @xmath151 for a given @xmath27 .
then @xmath149 is given by @xmath152 where @xmath153 is the inter - ion spacing defined in section [ intro ] .
this evaluates to @xmath154 where @xmath155 represents the net rate of energy loss in the wd due to all sources and sinks and @xmath156 is calculated from the wd s @xmath157 .
the ratio @xmath158 determines the @xmath0density jump across the liquid / crystal boundary .
how this ratio changes over the wd interior determines the gradient in the @xmath0concentration profile in the crystalline state .
a plot of @xmath158 for the models under consideration is shown in figure [ fig : crysstep ] . the increase in this ratio
outward is due to an increasing @xmath41 near the surface where @xmath149 decreases .
this leads to an inverted @xmath0concentration profile in the crystalline portion of the wd seen in figure [ fig : d1x22 ] .
this @xmath0abundance profile could well be destabilizing , as @xmath28 is increasing outwards .
however the linear stability is complicated by the elastic shear of the crystalline state , which might stabilize a profile that would otherwise be rayleigh - taylor unstable .
further calculations are needed to resolve this question .
one may also ask the question of how the increase in @xmath0at the crystal / fluid boundary affects the role that fractionation plays in wd cooling .
current thought is that under conditions typical of wds , a c / o/@xmath0plasma has liquid and solid phases of equal @xmath0concentration @xcite . when the wd starts to crystallize ,
the resulting phases are an o rich solid and c rich liquid .
the solid is the denser phase and an o rich solid core forms .
it is not until the temperature of the plasma is reduced to 1.12 times the freezing temperature of the pure carbon plasma that @xmath0fractionation begins to occur .
according to @xcite , by this time 70% ( by mass ) of the wd has already crystallized and the additional effects of ne distillation ( see @xcite ) on the cooling process is small .
if the results of @xcite ( i.e. his figure 1 ) are correct , the small increase in @xmath0concentration at the crystal / fluid boundary will not change this scenario . to do so
would require @xmath0concentrations of order @xmath159 , far above the additional increase in @xmath72 at the crystal / fluid boundary exhibited here .
thus , we do not expect @xmath0sedimentation to alter the picture developed to date of the role of @xmath0fractionation in wd cooling .
we have only considered wd models of a single ion species .
more realistic models with varying composition may have interesting effects on the @xmath0profile in the star .
for example , we expect that the viscosity of the interior plasma to increase with increasing @xmath4mass fraction which produces a corresponding change in @xmath41 . in regions where the composition is rapidly changing , for example near 0.5 @xmath124 in figure [ fig :
xinit ] , sharp spikes in the @xmath0composition may occur when the region crystallizes .
the rapid variation in composition may effect the @xmath0flow rates sufficiently in the liquid state to cause local increases in @xmath0concentrations at this stage also .
another way in which the @xmath0profile could be altered is through fractionation of the c / o/@xmath0plasma when crystallization occurs , an effect that we have ignored in this paper .
a full calculation including @xmath0sedimentation and full consideration of the 3 component plasma phase transition is needed to answer the question of how much this may change the final @xmath0profile in the wd .
the effects of @xmath0sedimentation on wd cooling are summarized in figures [ fig : tclgcrys ] and [ fig : ltcrys ] .
figure [ fig : tclgcrys ] shows the instantaneous power generated by @xmath0sedimentation , @xmath25 , in comparison with the luminosity of the wd ( dash - dot line ) as a function of core temperature . shown
are the @xmath25 produced for pure @xmath3and pure @xmath4(solid and dashed lines respectively ) stars as well as a comparison between calculations performed with @xmath62 and @xmath160 ( lower and upper sets of curves ) .
the results for a c / o wd are between the @xmath3and @xmath4wd curves .
the effect of crystallization in halting the @xmath0sedimentation is seen in the sharp change in slope for the @xmath25 curve near @xmath161 k. since @xmath162 , it is seen that this wd always cools .
thus , in the case of low mass wds , the possibility for @xmath0sedimentation to provide a wd s luminosity seems unlikely , at least if wds crystallize as they cool .
in addition , the results for a c / o wd that has a phase change to a glassy state at high @xmath49 values is shown for both masses by the dash - dot line .
as mentioned before , @xmath21 in this case was calculated from equation ( [ eq : etaglfit ] ) and the stokes - einstein relation . that sedimentation continues to generate energy at late time is quite obvious , and in the case of a @xmath124 wd
can dramatically affect the wd s late evolution .
concentrating on the 0.6 @xmath124 , @xmath62 curves for a moment , @xmath25 is substantially less than @xmath163 over most of the evolution of the wd and the effects of @xmath0sedimentation on wd cooling take some time to become apparent .
this is seen in figure [ fig : ltcrys ] where a comparison between the cooling with and without @xmath0settling is shown . not until after about 4 gyr is there a substantial difference in the cooling age of this model .
the fact that the low mass wds crystallize slowly combined with slow @xmath0sedimentation cause the delays from @xmath0sedimentation to increase until the wd is completely crystalline ( which occurs at about @xmath164 in this model ) .
the total delay in cooling that occurs for this model is about 0.25 gyr by 10 gyr .
we also indicate the approximate location of the zz ceti instability strip in this figure by the horizontal dashed line .
the @xmath0profile at the time at which the wd crosses this location is needed to determine the potential effects that @xmath0sedimentation may have on the seismological properties of the star ( see [ sec : brunt ] ) . as found by bh01 ,
the competition between the heating due to @xmath0and the energy output of the wd is much closer in the more massive wds , as exemplified by the 1.0 @xmath124 model in the lower panel of figure [ fig : tclgcrys ] .
however , in all cases we examined ( including calculations with @xmath165 in the 1.0 @xmath124 model ) , the power generated by @xmath0sedimentation never exceeded the luminosity of the wd .
thus it appears improbable in all but maybe the most massive wds that @xmath0sedimentation would be able to power the wd .
however , the early competition between the @xmath0heating and radiative losses does show up in the cooling curves as will be seen in a moment . the overall impact of @xmath0on the cooling of the more massive wds due to this rapid sedimentation is mitigated by the equally rapid crystallization in these stars , as exhibited by the earlier turnover in the @xmath25 curves in these models .
we can also see a qualitative differences in the effect of @xmath0sedimentation on differing mass models from this plot . in the lower mass wds ,
@xmath0sedimentation is a slower , enduring process which takes substantial time to compete with cooling at a significant level .
but , @xmath25 in this case stays competitive with @xmath163 for a longer period of time . for the higher mass stars , sedimentation is fast and @xmath25 is large initially but drops off quickly . in the 1.0 @xmath124 model ,
the @xmath165 exhibits the effects of rapid @xmath0depletion in the outer layers of the wd .
the extremely rapid sedimentation in this model leads to a large delay in the cooling time early on as compared to the other models which include sedimentation effects .
but , as the @xmath0is depleted from the outer layers in this model , energy generation due to sedimentation decreases compared to the models with smaller diffusion coefficients and the relative delays between them decrease at late times .
the effects of these differences are seen in the cooling curves shown in figure [ fig : ltcrys ] . for clarity
, data is only shown for wds with @xmath121 ; results for pure @xmath3and @xmath4wds bracket these results .
both plots show the effects of varying @xmath21 over the range of @xmath166 along with a model that neglects @xmath25 ( the bottom curve in each ) and a model that undergoes a transition to a glassy state instead of crystallizing ( dotted curve ) .
as could be expected from figure [ fig : tclgcrys ] , the delays introduced in the @xmath107 model only grow significant at late times , whereas those of the @xmath167 model are apparent early .
these delays continue until the wds have almost fully crystallized , which occurs by the time @xmath168 for the @xmath169 models .
a summary of the increase in cooling ages , @xmath127 , for these models is in figure [ fig : delays ] .
each panel shows the difference between the age of our @xmath121 models with @xmath170 ( solid lines , bottom to top , respectively ) and one without @xmath0diffusion included .
in the 1.0 @xmath124 model , the early termination of increases in @xmath127 in the @xmath171 and @xmath165 models as compared to the @xmath62 model is the results of rapid depletion of @xmath0 in the outer layers of the wd .
so , there is little @xmath0left to sediment out in the remaining liquid regions of the wd before the wd completely crystallizes .
the dash - dot line in figure [ fig : delays ] shows the difference in age between the reference model and a wd that undergoes a glassy transition . for a period
while the reference model is undergoing crystallization , the glassy wd model is actually _ younger _ than the reference model .
the explanation for this is that the glassy wd does not liberate latent heat , which is a much more significant quantity of energy while the reference star is crystallizing .
once the crystallization process is finished , the continued release of energy from @xmath0sedimentation in the glassy wd produces the age increases seen ( at least in combination with the rapid cooling that the crystallized wd undergoes once it enters the debye cooling regime ) . in the glassy wd models , the increase in wd age for a given luminosity continue to increase indefinitely and would lead to an abundance of faint wds greater than would otherwise be expected . from equation ( [ eq : lg ] ) ,
the power generated by @xmath0sedimentation is proportional to @xmath0number density .
the delays in cooling time produced by @xmath0sedimentation will thus vary with the metallicity of the wd progenitor .
a particular example where this may be observationally relevant is the open cluster ngc 6791 .
ngc 6791 is interesting both because is it old ( @xmath172 gyr ) and metal _ rich _ , @xmath173=0.4 $ ] @xcite .
figure [ fig:06ltczcomp ] shows a comparison in the @xmath25 generated in the 0.6 @xmath124 @xmath121 wd models whose progenitor s metallicity were solar ( as in the prior calculations ) or @xmath174 , as is appropriate for wds within ngc6791 . shown in figure [ fig:06ltzcomp ]
is the comparison of cooling curves between these wds .
as is readily seen , wds whose progenitors were metal rich will be more affected by the effects of @xmath0sedimentation .
this will be especially true if @xmath21 is significantly larger than our fiducial best guess of @xmath24 .
can this seen ?
most likely , yes , if a massive ( @xmath175 ) is found , as these should be equal in age to the cluster since their massive progenitors had such short lives .
figure [ fig : ngc6791cmd ] shows the location of a @xmath176 wd in the cluster cmd for different values of @xmath21 but all at a fixed age of 8 gyr .
the deceleration of the aging process is evident , as the @xmath62 model is @xmath177 mags brighter than the model without diffusion included . for comparison the effects on the more typical lower mass wds
are also shown .
the fact that @xmath0sedimentation s effects on the wd cooling sequence is metallicity dependent also provides the most straightforward means of separating it from other uncertainties in wd cooling . to our knowledge
, no other uncertainty has as significant dependence on the metal content of the wd as @xmath0sedimentation .
the only other effect that might is the fractionation of @xmath0during crystallization .
but , according to @xcite , the extent to which this effect matters is much smaller than that due to c / o fractionation and in any case occurs only after the larger portion of the wd has crystallized .
separating the delays in wd cooling due to @xmath0from other contributions is difficult in the absence of observational comparisons between metal rich and metal poor wds . in the low mass wds , at the point where @xmath0 s effects become apparent ,
the star has already begun crystallizing and the uncertain energy releases from fractionation can occur . in the high mass stars , especially for large @xmath0diffusion rates , there is a window between 1 - 4 gyrs in which @xmath0sedimentation can have an extremely significant impact on the cooling age of the wd . to be fair , here again ,
crystallization has already started in these stars , but the potential magnitude of @xmath0contributions may allow separation of @xmath0and fractionation effects . although not yet feasible , potentially the clearest test of @xmath0sedimentation s impact on wd cooling in particular , and wd cooling theory in general , would be the construction of empirical cooling curves for wds of specified masses .
as the overall morphology of the cooling curves can depend on which effects are included and their magnitudes , this may provide at least a qualitative means of determining empirically which phenomenon are operative in wds .
such a comparison would require a large sample of confirmed wds homogeneous in mass and metallicity and whose cooling ages are determined independently of their luminosity .
typical @xmath0abundance profiles at the time our wd models reach the zz ceti instability strip are shown in figure [ fig : neprofiles ] .
the abundance gradient in @xmath0creates a gradient in the mean molecular weight per electron , @xmath28 , that provides an extra restoring force for the g - modes .
we now show that this modifies the pulsation periods at levels comparable to the uncertainties of the measurements . to see this ,
we start with the brunt - visl frequency @xmath178 \;,\ ] ] @xcite where @xmath179 is one of the adiabatic indices . for an isothermal background of degenerate , relativistic electrons , and an ideal ion gas ( the coulomb corrections to @xmath180 are small ; @xcite ) and a @xmath28 gradient that depends only on the gradients of the trace @xmath0 , we find ( see appendix ) @xmath181 where @xmath182 and @xmath183 is the ion mean molecular weight .
the first piece is the thermal contribution to @xmath180 ( previously known to be small in the deep interior ; @xcite ) while the second is the @xmath0contribution .
figure [ fig:10brunt ] shows @xmath180 for a 1 @xmath184 wd with @xmath185 for @xmath62 and @xmath186 at the point its @xmath187 k ( the ages of the two models are 1.2 and 1.5 gyr respectively ) . for this plot , we improved upon equation ( [ eq : n2rel ] ) by explicitly calculating @xmath188 using the fully degenerate equation of state instead of using either limiting approximations ( see appendix ) .
the high pressure limit is set by the location of the crystal / fluid boundary , which effectively acts as a hard - sphere boundary @xcite .
the solid horizontal lines show selected mode periods of the only solar mass zz ceti known , bpm 37093 @xcite .
it is clear that these observed modes probe the region of the wd interior in which @xmath0gradients make their largest contribution to @xmath180 .
the contribution from the @xmath0abundance profile affects @xmath29 over the interior and thus the periods of the longer period modes that penetrate deeply .
we estimate the extent to which @xmath0sedimentation affects the mode frequencies by performing the integral @xmath189 through the liquid wd interior ( the envelope s contribution to the integral is @xmath190 ) . using only the thermal contribution to @xmath180 gives @xmath191 through the wd core .
the same core integral evaluated taking into account the @xmath0gradient gives @xmath192 for @xmath62 and @xmath193 for @xmath186 , or a total change in the interior integral of 8 - 30% .
adding the envelope contribution tells us that @xmath0sedimentation can affect pulsation periods by greater than 1% , or over 100 times larger than the measurement error .
the effects of @xmath0sedimentation in a @xmath194 wd are about 0.6% , or 4 seconds for 700 second modes , at the levels where work is already in progress ( e.g. @xcite ) . the question of whether or not the seismological effects of @xmath0gradients can be separated from those due to the uncertainties in the c / o profile and envelope layer thicknesses will have to wait until detailed models are made that include the @xmath0contributions .
it seems probable that once this is done , it will be seen that distinguishing between theoretical mode spectra calculated with or without @xmath0gradient contributions is possible based on the fact that the @xmath0gradients introduce a new spike in the brunt - visl frequency in the wd interior . whether or not pulsation studies will be able to confirm the presence of such gradients in real wds remains to be answered .
with two extra neutrons ( over and above the @xmath195 ratio of the background ions ) , the diffusion of @xmath0is biased inward in the liquid interior of wds .
the impact of this sedimentation on wd cooling was first estimated by bh01 .
we have extended their work by calculating the mutual evolution of the @xmath0density , @xmath26 , and core temperature , @xmath27 , for wd models composed of a single background ion species .
the heating produced by @xmath0increases the cooling age of our models on order of 0.2 - 1.5 gyr ( see figure [ fig : delays ] ) depending on the value of @xmath21 .
we have also performed initial estimates of the effects that @xmath0abundance gradient can have on the pulsation modes of wds .
although more precise work needs to be done , these initial calculations indicate that the corrections to mode periods due to @xmath28 gradients impact the results at the current level of observational precision .
the uncertainty in the actual value of the diffusion coefficient produces a large uncertainty in the possible impact @xmath0may have on wd evolution .
the significant impact that @xmath0sedimentation may have on ages of recently formed wds provides clear astrophysical motivation for authoritative calculations for the @xmath21 of @xmath0through a multicomponent plasma .
hopefully further theoretical work in this direction will eliminate this source of uncertainty .
observations can constrain the diffusion rate of @xmath0 in wd interiors .
for example , say the age of a wd is known because of its membership in a cluster .
its luminosity will then depend on the rate at which @xmath0sinks and on the overall amount of @xmath0present in the wd ( see equation ( [ eq : lg ] ) ) .
also the amount by which the brightness of the wd will be affected is very dependent on the wds mass , as can be seen by the differences between the 0.6 and 1.0 @xmath124 cooling curves in figure [ fig : ltcrys ] .
having an independent measure of these four quantities ( @xmath123 , @xmath163 , wd age , and @xmath72 ) can allow meaningful tests of our theory .
the main uncertainty in wd cooling , namely fractionation effects , do not come into play until a large portion of the wd has crystallized and do not seem to be a function of @xmath0concentration ( at least for @xmath196 ; see @xcite ) .
increases in wd luminosity over that expected from the canonical cooling theory for times prior to crystallization or which show a strong metallicity dependence are most readily interpreted as the effects of @xmath0sedimentation .
the question is whether or not @xmath0can affect the wd luminosity to a measurable degree in the regimes where fractionation is unimportant .
one place we might look are massive young wds . for ages between 14 gyrs ,
the @xmath163 of such objects is strongly dependent on @xmath21 .
but , on the other hand , these wds begin to crystallize in this time frame . for @xmath197 , c / o fractionation effects may dominate over @xmath0sedimentation , at least if @xmath198 ( @xcite and see also figure [ fig : delays ] ) . for larger values of @xmath21
though , these wds will be maintained at a high enough temperature that fractionation becomes significantly less important at these times .
the other place we might look for the effects of @xmath0are in wds born from metal rich progenitors . in this case , the effects on @xmath0sedimentation even in the lower mass wds are more readily apparent and may be observable . the data required for either of the above two observational programs can be obtained for wds that are members of open clusters . in this case
, the age of the wd can be inferred from the difference between the cluster s turn - off age and the main sequence lifetime of it progenitor , if the wd mass is known .
the cluster s metallicity constrains the @xmath0content of the wds . to date , though , the actual quantity of data for confirmed wd cluster members is rather scarce .
a compilation of such objects was performed by von hippel in 1998 @xcite . in this sample ( which is still seemingly complete )
, there is only one wd ( g152 in ngc 2682 ) with a mass or @xmath1 determination in a cluster older than 1 gyr @xcite .
the other 27 objects all reside in clusters younger than this . of these , there are about 5 - 6 wds with masses greater than 0.9 @xmath124 and several ( most notably the two wd members of ngc 2168 ) whose mass determination are highly suspect @xcite .
obviously , it would be ideal to increase the number of known wds in clusters with a wider range of ages than we have currently .
such a program should be a priority if for no other reason than the need for a direct observational test of wd cooling theory in general . in the past seven years or so ,
the number of wd candidates in open clusters have grown substantially due to the series of deep ( limiting magnitudes of 24 - 26 in the @xmath199 band ) photometric studies of open clusters that several groups have undertaken @xcite . the goal of these studies is to provide a determination of cluster ages using the age of the oldest wd candidates found in each cluster .
identification of an object as a wd is based on the object s location in the cluster s color - magnitude diagram and on it having a spatial morphology that is stellar in nature ( as opposed to one that is galactic
contamination of the wd sequence with faint blue galaxies in these studies is one of the difficulties encountered)@xcite . to date
, no follow up spectroscopy to confirm these objects aras wds and to make mass determinations if they are has been done , although the cfht group has future plans to do so @xcite .
overall the number of new wd candidates added by these studies is around 300 objects .
the clusters studied so far range in age from 0.5 - 7 gyrs .
it is worth noting that only one of the 19 clusters in the cfht survey has an age greater than 1 gyr @xcite .
follow up spectroscopy on clusters outside of the cfht sample is thus highly desirable . as mentioned earlier , the cluster ngc 6791 is an extremely interesting target due to its high metal content . to our knowledge
, there have been no wd candidates identified in this cluster and discovery of a wd sequence in ngc 6791 could provide a ready test of our theory since the effects of @xmath0on wd cooling will be greatly amplified there .
we would like to thank peter heflich for providing the pre - wd evolutionary models from which we based our discussion of wd composition , leandro althaus for providing the @xmath200 relations used in our calculations , and james liebert for a critical reading of our manuscript .
this work was supported by the nsf under grants phy99 - 07949 , ast01 - 96422 , and ast02 - 05956 .
l. b. is a cottrell scholar of the research corporation .
the brunt - visl frequency is given by the expression @xmath201\ ] ] which can be expressed in the form @xmath202 + g \left(\frac{\chi_t}{\chi_{\rho } } { \ensuremath{\frac{d \,\ln t}{dz } } } + \frac{\chi_{\mu_e}}{\chi_{\rho } } { \ensuremath{\frac{d \,\ln \mu_e}{dz } } } \right)\ ] ] where @xmath203 and @xmath28 is the electron mean molecular weight .
as @xmath204 can be written @xmath205 and putting @xmath206 along with using hydrostatic balance , @xmath207 , equation ( [ eq : bv2 ] ) can be expressed in the form @xmath208\;,\ ] ] showing the explicit relation between our expression and the more common form of equation ( 13 ) from @xcite . in extremely degenerate matter , @xmath209 and if not careful the important term is missed .
we thus rewrite the expression for the adiabatic exponent in terms of @xmath188 plus a small correction term @xmath210 in degenerate matter,@xmath215 is between 4/3 and 5/3 depending on the relativity parameter .
finally @xmath216 where the last step is valid only if @xmath0is a trace . with this expression
we have @xmath217 so that , for relativistic matter , @xmath218 99 althaus , l. & benvenuto , o. , 1989 mnras 296 , 206 andreuzzi , g. , richer , h. b. , limongi , m. , & bolte , m. , 2002 , a&a ( in press ) bildsten , l. & cutler , c. , 1995 , apj , 449 , 800 bildsten , l. & hall , d. , 2001 , apj , 549 , l219 bradley , p. a. 2001 , apj , 552 , 326 brassard , p. , fontaine , g. , wesemael , f. , kawaler , s.d . , &
tassoul , m. , 1991 , apj , 367 , 601 bravo , e. , isern , j. , canal , r. , & labay , j. , 1992 , a&a , 257 , 534 chaboyer , b. , green , e. , & liebert , j. , 1999 , aj , 117 , 1360 chabrier , g. , 1993 apj , 414 , 695 chabrier , g. , ashcroft , n. , & dewitt , h. , 1992 , nature , 360 , 49 dewitt , h. & slattery , w. , 1999 , plasma phys 39 , 97 donk , z. , & nyri , b. , 2000 , physics of plasmas , 7 , 45 farouki , r. t. & hamaguchi , s. 1993 , phys .
e 47 , 4330 hansen , j .-
& mcdonald , i. r. , 1986 , _ theory of simple liquds _ , london , academic press hansen , j .-
p . , mcdonald , i. r. , & pollack , e. l. 1975 , phys .
a , 11 , 1025 isern , j. , mochkovitch , r. , garcia - berro , e. , & hernanz , m. , 1991 , a&a , 241 , l29 kanaan , a. , kepler , s. o. , giovannini , o. , winget , d.e . ,
montgomery , m. , & nitta , a. , 1998 , balt .
7 , 183 kalirai et al . , 2001a , aj 122 , 257 2001b , aj 122 , 266 2001c , aj 122 , 3239 koester , d. & reimers , d. , 1988 , a&a , 202 , 77 landsman , w. , et al .
, 1998 , aj , 116 , 789 metcalfe , t. s. , winget , d. e. & charbonneau , p. 2001 , apj , 557 , 1021 mochkovitch , r. , 1983 , a&a , 233 , 456 montgomery , m. h. , klumpe , e. w. , winget , d. e. , & wood , m. a. , 1999 , apj , 525 , 482 montgomerey , m. h. & winget , d. e. , 1999 , apj , 526 , 976 murillo , m. s. , 2000 , phys . rev .
e , 62 , 4115 nitta , a. , kanaan , a. , kepler , s. o. , koester , d. , montgomery , m. h. , & winget , d. e. , 2000 , balt .
9 , 97 ogata , s. , iyetomi , h. , ichimaru , s. , & van horn , h. , 1993 phys .
e , 48 , 1344 paquette , c. , pelletier , c. , fontaine , g. , & michaud , g. , 1986 , apj , 61 , 177 potekhin , a. & chabrier , g. , 2000 , phys rev e , 62 , 8554 press , w. h. , teukolsky , s. a. , vetterling , w. t. , & flannery , b. p. , 1992
, _ numerical recipies in c , the art of scientific computing _ , cambridge university press richer , h. b. , fahlman , g. g. , rosvick , j. , & ibara , r. , 1998 , apj , 116 , l91 salaris , m. , garca - berro , e. , hernanz , m. , isern , j. , & saumon , d. , 2000 , apj 544 , 1036 segretain , l. , chabrier , g. , hernanz , m. , garca - berro , e. , isern , j. , & mochkovitch , r. , 1994 , apj , 434 , 641 segretain , l. , 1996 , a&a , 310 , 485 tanaka , s. & ichimaru , s. , 1987 phys rev a 35 , 4743 .
umeda , h. , nomoto , k. , yamaoka , h. , & wanajo , s. , 1999 , apj , 513 , 861 unno , w. , yoji , o. , ando , h. , saio , h. , & shibahashi , h. , 1989 , _ nonradial oscillations of stars , 2nd ed .
of tokyo press viellefosse , p. , & hansen , j.p . , 1975 phys rev a 12 , 1106 von hippel , t. , gilmore , g. , & jones , d.h.p .
, 1995 , mnras , 273 , l39 von hippel , t. , 1998 , aj , 115,1536 von hippel , t. & gilmore , g. , 2000 , aj , 120,1384 von hippel , t. & sarajedini , a. , 1998 , aj , 116,1789 wallenborn , j. & baus , m. , 1978 , phys . rev .
a , 18 , 1737 xu , z .w .
& van horn , h. m. , 1992 apj , 387 , 662 | we assess the impact of the trace element @xmath0on the cooling and seismology of a liquid c / o white dwarf ( wd ) . due to this elements neutron excess
, it sinks towards the interior as the liquid wd cools .
the subsequent gravitational energy released slows the cooling of the wd by 0.251.6 gyrs by the time it has completely crystallized , depending on the wd mass and the adopted sedimentation rate .
the effects will make massive wds or those in metal rich clusters ( such as ngc 6791 ) appear younger than their true age .
our diffusion calculations show that the @xmath0mass fraction in the crystallized core actually increases outwards .
the stability of this configuration has not yet been determined . in the liquid state ,
the settled @xmath0enhances the internal buoyancy of the interior and changes the periods of the high radial order @xmath1-modes by @xmath2 1% . though a small adjustment , this level of change far exceeds the accuracy of the period measurements . a full assessment and comparison of mode frequencies for
specific wds should help constrain the still uncertain @xmath0diffusion coefficient for the liquid interior . |
electromagnetic propagation through metamaterials has stimulated , in the last decade , an intense research activity with two main purposes : the identification of artificial structures exhibiting anomalous values of the permittivity @xmath2 and the permeability @xmath3 @xcite and the investigation of the electromagnetic phenomenology resulting from the unusual electromagnetic properties .
leading examples of the novel class of phenomena supported by metamaterials are superlensing @xcite , optical cloaking @xcite and photonic circuits @xcite .
recently , a good deal of attention has been devoted to the investigation of metamaterials characterized by a very small dielectric permittivity ( @xmath2-near - zero materials ) since they host an electromagnetic regime where the magnetic field displays static features @xcite and the tunneling of electromagnetic energy through sub - wavelength channels has been proposed @xcite and experimentally observed @xcite .
metamaterials exhibiting remarkable nonlinear properties have been investigated as well @xcite and various soliton manifestations @xcite have been considered .
the investigation of metamaterial nonlinear properties is particularly important in that it can lead to overcoming one of the fundamental limit of nonlinear optics , the fact that most of the optical materials have a relatively weak nonlinear response .
the main idea is that the local electromagnetic fields of the inclusions in the metamaterial can be much larger than the average value of the field thus producing an enhancement of the nonlinear response @xcite .
the problem of achieving a substantial enhancement of the nonlinear response has also been considered within the more general subject of composite structures homogenization @xcite , and the strategy is always that of conceiving a microscopic inhomogeneous structure concentrating the field within the nonlinear constituents .
the full exploitation of the nonlinear response is possible only if the nonlinear polarization is not a small perturbation to the linear part of the electric displacement field and generally this is achieved through nonlinearity enhancement or by means of resonant processes or photorefractive processes where the large nonlinearities come at the cost of a large time response . however , as shown in this paper , the interplay between the linear part of the electric displacement field and the nonlinear polarization can be made efficient even by following the opposite route , i.e. by reducing the linear polarization .
we therefore devise a nonlinear medium with a very small linear dielectric constant since it is a natural setting for the observation of the electromagnetic regime where the nonlinear response does not play the role of a mere perturbation .
in this paper we consider a periodic layered composite whose slabs are filled either with linear media with arbitrary permittivity and permeability or by standard isotropic focusing or defocusing kerr media . exploiting a suitable extension of the well known technique generally used for describing the homogenization of linear layered composites ,
we show that the homogenized medium is characterized by effective constitutive relations formally coinciding with those of a standard kerr medium .
we note that the parameters characterizing such an effective response can be simply tailored through a suitable choice of the composite underlying constituents and that , due the large freedom in choosing both constituent media and their volume filling fraction , the design of linear and nonlinear properties can be independently performed .
therefore our composite medium allows a full and efficient engineering of the kerr nonlinear response .
more interestingly , we prove that the effective response parameters span very wide ranges encompassing values not available in standard media .
the standard isotropic kerr response ( in the the frequency domain ) generally depends on two parameters usually denoted with @xmath4 and @xmath5 @xcite in term of which we define @xmath6 and @xmath7 and the available values of @xmath8 belong to the range @xmath9 ( depending on the actual mechanism supporting the kerr response ) . in this paper
we prove that the parameter @xmath8 appearing in the effective kerr nonlinear response can , in principle , assume any value ( encompassing negative and very large values ) and we discuss the impact of the exotic values of @xmath8 on the electromagnetic phenomenology . confining our attention to transverse magnetic ( tm ) fields ,
we show that it is possible to design a nonlinear kerr medium whose linear dielectric permittivity is much smaller than one and therefore able to host the electromagnetic regime where nonlinearity can not be regarded as a perturbation ( extreme nonlinear regime ) . as a first general electromagnetic effect characteristic of the extreme nonlinear regime , we discuss the transverse power flow reversing , i.e. the fact that , for an electromagnetic beam , the poynting vector on the beam propagation axis can be antiparallel to poynting vector on the beam lateral sides @xcite . within the extreme nonlinear regime , the full exploitation of the nonlinear response can be achieved and , in order to discuss the consequent novel electromagnetic phenomenology , we consider a class of nonlinear guided waves admitting full analytical treatment . a number of the obtained nonlinear guided waves exhibit , as expected , transverse power flow reversing .
we prove that , in the situation where @xmath10 , @xmath11 and @xmath12 ( or in the equivalent case @xmath13 , @xmath14 and @xmath15 ) these waves are allowed to propagate through the medium if @xmath16 , in striking contrast to the fact that , in a linear medium with the same permittivity and permeability , any wave would be evanescent .
this is possible since nonlinearity overcomes the linear contribution to the electric displacement field producing an effective nonlinear dielectric permittivity able to support propagating waves .
in addition we identify a mechanism which , combining the extreme nonlinear regime with the properties of the effective medium response with @xmath17 , produces an hyper - focusing of the considered nonlinear guided waves . in the situation @xmath13 , @xmath11 and @xmath15 ( or in the equivalent case @xmath10 , @xmath14 and @xmath12 ) the nonlinear guided waves exist in the wider range @xmath18 and @xmath19 thus allowing to probe the kerr nonlinear response for large values of @xmath20 .
since the phase difference between the components of the considered waves is @xmath21 , we predict that , for @xmath22 , an extreme compensation occurs ( within the effective nonlinear polarization ) and supports a field whose transverse @xmath23 and longitudinal @xmath24 components are such that @xmath25 , i.e they share the same profile .
the paper is organized as follows . in sec.ii
we investigate the homogenization of a one - dimensional periodic layered medium comprising both linear ( metamaterials ) and nonlinear ( kerr media ) slabs and we derive the constitutive relations characterizing the effective medium . in sec.iii
we focus on the transverse magnetic field configuration and we discuss the effective nonlinear response engineering and the allowed extreme nonlinear regime , together with the transverse power flow reversing effect . in sec.iv we investigate a class of nonlinear guided waves belonging to the extreme nonlinear regime and we discuss a number of their peculiar properties .
the existence of the considered nonlinear guided waves is investigated in appendix a.
consider a monochromatic electromagnetic field ( whose time dependence is @xmath26 ) propagating through a metamaterial layered medium consisting of periodically repeating , along the @xmath27axis , @xmath28 layers of different media of thicknesses @xmath29 ( @xmath30 ) , so that the structure spatial period is @xmath31 ( see fig.1 where the case @xmath32 is depicted ) . each one of these @xmath28 media can be either a linear metameterial with arbitrary dielectric and magnetic properties or a cubic standard nonlinear medium so that , the electromagnetic response of the @xmath33th medium is modelled by the general constitutive relations @xmath34 , \nonumber \\ { \bf b}_j & = & \mu_0 \mu_j { \bf h}_j,\end{aligned}\ ] ] where @xmath35 and @xmath36 are the vacuum permittivity and permeability constants , whereas @xmath37 , @xmath38 , @xmath39 , @xmath40 are the complex amplitude of the local electromagnetic field vectors , @xmath41 and @xmath42 are the relative permittivity and permeability of the @xmath33th layer , @xmath43 and @xmath44 are the standard parameters characterizing the isotropic cubic nonlinear response of the @xmath33th layer @xcite . evidently , @xmath45 if the @xmath33th layer is filled by a linear medium and @xmath46 for a nonlinear dielectric .
the linear dielectric constants @xmath41 are here regarded as arbitrary complex numbers since the layers can be filled with lossy and active media . if the field vacuum wavelength is much greater than the spatial period ( @xmath47 ) the considered periodically nonlinear stratified medium can be homogenized , i.e. its electromagnetic response can be shown to coincide with that of a suitable homogeneous medium . in order to obtain the overall effective response , note that any component of the local electromagnetic field can be assumed , within each layer , independent on @xmath48 since the layers are extremely small ( @xmath49 ) .
although this is a very reasonable physical assumption , it can be rigorously proven exploiting the well known powerful two - scale expansion method @xcite .
each of the physical observable ( macroscopic ) electromagnetic field vectors , say @xmath50 ( @xmath51 ) , is obtained by averaging ( along the @xmath27 axis ) the layer local fields over the period @xmath52 , so that @xmath53 where @xmath54 is the volume filling fraction of the @xmath33th medium ( @xmath55 ) and the averaging has been performed by exploiting the uniformity along @xmath48 of the local fields @xmath56 . within each unit cell
( of thickness @xmath52 ) , at each plane interface between the @xmath33th and @xmath57th layer , the local fields have to satisfy the electromagnetic boundary conditions ( continuity of the tangential component of electric and magnetic fields and continuity of the normal components of the displacement and magnetic induction fields ) so that the local fields are joined by the relations @xmath58 where @xmath59 . averaging the local fields
@xmath60 , @xmath61 and @xmath62 and exploiting the second of eqs.([micr ] ) together with magnetic boundary conditions of eqs.([boundary ] ) , it is straightforward to prove that the macroscopic fields @xmath63 and @xmath64 are related by @xmath65 where the relative magnetic permeability @xmath66 is the diagonal tensor @xmath67 $ ] , which is a very well - established result concerning the homogenization of layered media @xcite . as far as the dielectric response
is concerned , exploiting the first of eqs.([micr ] ) together with the fact that @xmath68 , @xmath69 and @xmath70 ( obtained by combining the electric boundary conditions of eqs.([boundary ] ) and eq.([average ] ) ) we obtain @xmath71 \nonumber \\ & + & \epsilon_0 \left [ \langle \chi_j |e_{jy}|^2 \rangle { \bf e}_\perp + \langle \chi_j \gamma_j e_{jy}^2 \rangle { \bf e}_\perp^ * \right ] , \nonumber \\ d_y & = & \epsilon_0 \epsilon_j e_{jy } \nonumber \\ & + & \epsilon_0 \chi_j \left [ ( { \bf e}_\perp \cdot { \bf e}_\perp^ * ) e_{jy } + \gamma_j ( { \bf e}_\perp \cdot { \bf e}_\perp ) e_{jy}^ * \right ] \nonumber \\ & + & \epsilon_0 \chi_j \left [ ( 1+\gamma_j ) |e_{jy}|^2 e_{jy } \right]\end{aligned}\ ] ] where @xmath72 and @xmath73 are the transverse parts of the macroscopic displacement and electric fields , respectively and @xmath74 is the @xmath27component of the macroscopic displacement field . note that in the second of eqs.([dd ] ) no averaging has been performed . as a consequence of the layers nonlinear dielectric behavior , the first of eqs.([dd ] ) contains terms where the squares of @xmath75 is suitably averaged . in order to derive the effective medium response
, the second of eqs.([dd ] ) has to be solved to express @xmath75 ( @xmath30 ) as a function of the macroscopic fields @xmath76 and @xmath74 .
this can be pertubatively done by noting that , since we are considering the standard kerr effect for each nonlinear layer @xcite , the terms containing @xmath43 are much smaller than the term @xmath77 so that , up to the first order in the field cubic terms , from the second of eqs.([dd ] ) we obtain @xmath78 \nonumber \\ & - & \chi_j \frac{1+\gamma_j}{\epsilon_0 ^ 2 \epsilon_j^4 } |d_y|^2 d_y.\end{aligned}\ ] ] averaging eq.([ee ] ) we obtain a relation joining the macroscopic field @xmath79 and the fields @xmath80 and @xmath74 so that this relation , if its nonlinear contributions are much smaller than the leading linear term , can be inverted to perturbatively yield @xmath74 as a function of @xmath81 and @xmath79 . therefore ,
up to the first order in the field cubic terms , we obtain @xmath82 e_y \nonumber \\ & + & \epsilon_0 \left [ \frac{\langle \chi_j \gamma_j \epsilon_j^{-2 } \rangle } { \langle \epsilon_j^{-1 } \rangle^{2 } } { \bf e}_\perp \cdot { \bf e}_\perp + \frac{\langle \chi_j ( 1+\gamma_j ) \epsilon_j^{-4 } \rangle}{2 \langle \epsilon_j^{-1 } \rangle^{4 } } e_y^2 \right ] e_y^*. \nonumber \\\end{aligned}\ ] ] substituting the fields @xmath75 of eqs.([ee ] ) into the first of eqs.([dd ] ) and using eq.([dy ] ) ( neglecting everywhere the terms containing powers of @xmath43 higher than one ) we obtain @xmath83 { \bf e}_\perp \nonumber \\ & + & \epsilon_0 \left [ \langle \chi_j \gamma_j \rangle { \bf e}_\perp \cdot { \bf e}_\perp + \frac{\langle \chi_j \gamma_j \epsilon_j^{-2 } \rangle}{\langle \epsilon_j^{-1 } \rangle^{2 } } e_y^2 \right ] { \bf e}_\perp^*.\end{aligned}\ ] ] equations ( [ dy ] ) and ( [ dp ] ) solely contain the macroscopic fields so that they are the constitutive dielectric relations characterizing the effective medium obtained by the homogenization of the considered 1d layered medium .
the obtained dielectric response shows that the effective medium behaves like an anisotropic kerr medium whose nonlinear properties can be tailored by suitably choosing the underlying layered composite .
note that , in the specific case of linear layers ( i.e. @xmath84 for all @xmath85 ) , eqs . ( [ dy ] ) and ( [ dp ] ) yield @xmath86 where the relative dielectric permittivity @xmath87 is the diagonal tensor @xmath88 $ ] , reproducing a well known result concerning homogenization of linear layered media @xcite .
electromagnetic propagation through the homogenized medium of sec.ii is described by the macroscopic maxwell equations @xmath89 and the constitutive relations of eqs.([bbb ] ) , ( [ dy ] ) and ( [ dp ] ) .
hereafter , we focus on transverse magnetic ( tm ) electromagnetic fields ( see fig.1 ) of the form @xmath90 for which the constitutive relations become @xmath91 , \nonumber \\ { \bf b } & = & \mu_0 \mu { \bf h},\end{aligned}\ ] ] where @xmath92^{-1 } , \nonumber \\
\chi = & \displaystyle \sum_{j=1}^n f_j \chi_j , \quad \quad \gamma & = \frac{1}{\chi } \sum_{j=1}^n f_j \chi_j \gamma_j.\end{aligned}\ ] ] therefore the effective dielectric response experienced by a tm field coincides with the standard isotropic kerr response . in analogy to what happens in linear layered media
we note , from eqs.([parameters ] ) , that the permittivity @xmath2 and the permeability @xmath3 are the weighted and harmonic weighted means , respectively , so that the former is bounded by the minimum and maximum of its microscopic values whereas the latter is unbounded since some @xmath42 can be negative .
it is worth stressing that the averaging of the microscopic linear parameters additionally allows an efficient loss management since , combining lossy and gain media , one can design an effective medium whose effective parameters @xmath2 and @xmath3 have negligible imaginary parts @xcite . on the other hand , @xmath93 is the weighted mean of the constituents nonlinear kerr coefficients and this is in agreement with the results of ref.@xcite .
more interestingly , @xmath8 is generally not the weighted mean of its microscopic values since some @xmath43 can be negative ( corresponding to defocusing nonlinear layers ) , so that @xmath8 can assume any positive and negative value .
this result is particularly interesting since , for standard cubic isotropic materials , there is , in general , a small number of available @xmath8 ( depending on the physical mechanism supporting the nonlinear response @xcite ) whereas , for composite materials , it has been shown in ref.@xcite that @xmath8 can span the whole range @xmath9 .
even though eqs.([tmconst ] ) formally coincides with the standard isotropic kerr response , the considered layered material can support a nonlinear electromagnetic phenomenology very different from that observed in a standard kerr medium , as discussed in the following three subsections .
consider a layered medium for which @xmath2 is such that @xmath94 , @xmath95 and the effective nonlinear coefficient @xmath93 is of the same order of magnitude of underlying coefficients @xmath43 ( i.e. @xmath96 .
an effective dielectric permittivity with a very small real part and a negligible imaginary part can be achieved by combining positive and negative standard dielectric layers ( @xmath97 ) together with gain media ( generally unavoidable since absorption due to negative dielectrics can not be in principle neglected @xcite ) . for the sake of simplicity let us consider the case @xmath98 .
if the electromagnetic field propagating through the medium is such that @xmath99 so that , in the first of eqs.([micr ] ) , the nonlinear part is much smaller than the linear contribution @xmath100 .
therefore , the nonlinear layers are in the presence of a field whose intensity is sufficiently small for their response to be purely cubic and , as a consequence , the microscopic responses of eqs.([micr ] ) hold and the macroscopic constitutive relation in the first of eqs.([tmconst ] ) , holds as well . on the other hand , combining eq.([extr ] ) and the first of eq.([tmconst ] ) we conclude that , as opposed to what happens in standard nonlinear materials , the considered layered medium is able to support the extreme nonlinear regime where the electromagnetic field is such that the linear contribution @xmath101 and the nonlinear term @xmath102 in the overall dielectric response have the same order of magnitude .
it is evident that the same argument , with some slight changes , allows to prove that the extreme nonlinear regime is observable for any value of @xmath8 .
even though the advantages of such an extreme nonlinear regime are self - evident , it is worth to compare it to the standard paraxial nonlinear optical situation where the overall refractive index is @xmath103 where @xmath104 is the background linear refractive index and @xmath105 is the kerr nonlinear term ( where @xmath106 is the optical intensity ) . in this situation
the field satisfies the equation @xmath107 ( @xmath108 ) and , within the paraxial regime , the field is of the form @xmath109 . at this point
one chooses @xmath110 since , substituting the paraxial field into eq.([helmo ] ) the term @xmath111 ( responsible for the fast spatial variation of the field ) is removed and the paraxial equation @xmath112 is readily obtained by neglecting the term containing @xmath113 .
therefore , in standard paraxial nonlinear optics , @xmath114 can play an important role in eq.([helmo ] ) since the large linear contribution proportional to @xmath115 is suppressed by the presence of the carrier plane wave .
in other words , the nonlinear kerr contribution @xmath116 is not requested to compete against the whole linear part @xmath117 of the dielectric response . as opposed , in the case of the extreme nonlinear regime discussed in this section , this competition can happen and therefore the nonlinearity is not confined to solely drive the slowly varying amplitude @xmath118 .
the extreme nonlinear regime can support propagation of beams characterized by exotic properties such as the reversing of the electromagnetic power flow along the beam transverse profile @xcite . in order to discuss this phenomenon , consider a tm field describing a beam mainly propagating along the @xmath119-axis , i.e. a field of the form @xmath120 with the requirement @xmath121 ( @xmath122 being a wave - vector depending on the actual electromagnetic configuration ) . in this case
maxwell equations of eqs.([maxwell ] ) yield @xmath123 where it has been assumed that @xmath124 and @xmath125 . exploiting eqs.([hy ] ) , the @xmath119-component of the time averaged poynting vector @xmath126 can be expressed through the equivalent relations @xmath127.\end{aligned}\ ] ] if , for example , @xmath128 , @xmath11 , @xmath15 and @xmath129 , for an electromagnetic beam whose peak electric field strength is greater than @xmath130 ( extreme nonlinear regime ) , it is evident from the first of eqs.([tmconst ] ) that @xmath131 and @xmath23 are antiparallel around the propagation axis ( i.e. where @xmath132 ) and parallel elsewhere so that , from the first of eqs.([sz ] ) , @xmath133 is negative near the beam axis and positive elsewhere .
in other words the beam is characterized by a power flow whose direction reverses its sign along the transverse profile and this is due to sign flipping of @xmath131 along the wave transverse profile while @xmath23 does not change its sign .
note that the sign flipping of @xmath131 corresponds to a sign flipping of @xmath134 ( see the first of eqs.([hy ] ) ) so that , considering the second of eqs.([hy ] ) , this can happen without sign flipping of @xmath23 only if @xmath135 is not negligible with respect @xmath136 .
therefore power flow reversing can take place only if the field has a transverse size comparable with @xmath137 .
it is worth stressing that the discussed power flow reversing is very different from the effect that , in left handed metamaterials , the poynting vector is antiparallel to the carrier wave vector which is a consequence of the fact that , in such media , @xmath138 and @xmath14 ( with @xmath139 ) . on the other hand , in our case ,
@xmath140 and the sign of the power flow is not uniform being controlled by the field intensity through the nonlinearity . in order to discuss the impact of the wide ranges of the parameters of eqs.([parameters ] ) , made possible by linear and nonlinear design , on electromagnetic phenomenology , we consider the situation where the effective dielectric permittivity @xmath2 and magnetic permeability @xmath3 are real , so that , after substituting the expression for the tm field of eqs.([tmfield ] ) into eqs.([maxwell ] ) , eliminating the magnetic field and using eqs.([tmconst ] ) we get @xmath141 , \nonumber \\ & & \frac{\partial ^2 u_x}{\partial \xi \partial \zeta } - \frac{\partial ^2 u_z}{\partial \xi^2 } \nonumber \\ & & = \sigma_\epsilon \sigma_\mu u_z + \sigma_\mu \left [ ( 1+\gamma ) |u_z|^2 u_z + \left ( u_z u_x^ * + \gamma u_z^ * u_x \right ) u_x \right ] , \nonumber \\\end{aligned}\ ] ] where dimensionless variables and fields have been introduced according to @xmath142 and @xmath143 whereas @xmath144 and @xmath145 are the signs of the products @xmath146 and @xmath147 , respectively . as opposed to the linear regime where electric field behavior solely depends on the sign of @xmath148 @xcite , from eqs.([system ] ) we note that , in the present approach , the nonlinear dynamics separately depends on the signs of @xmath146 and @xmath147 , i.e. the presence of the nonlinearity breaks the symmetry between the roles played by the signs of @xmath2 and @xmath3 @xcite . from eqs.([dimension ] ) we note that @xmath2 , @xmath3 and @xmath93 scale the actual physical size and amplitude of the field so that , since the effective parameters can in principle be independently chosen ( see eqs.([parameters ] ) ) , for each solution of eqs.([system ] ) , a suitable layered medium can be designed in such a way that the actual electromagnetic field has prescribed geometrical size and intensity ( see eqs.([dimension ] ) ) . as an example , in the case of beam propagation , such an electromagnetic scaling freedom can allow to observe nonparaxial feature of a beam whose transverse width is much greater than the vacuum wavelength ( if @xmath149 ) or , on the contrary , to observe the standard paraxial phenomenology for beams whose transverse width is much smaller than the vacuum wavelength ( if @xmath150 ) and , remarkably , this can be done by avoiding any unfeasible requirement on the intensity .
the parameter @xmath8 plays a role fundamentally different since it can not be generally removed from eqs.([system ] ) with a field transformation . setting @xmath151 and @xmath152 ( where @xmath153 , @xmath154 , @xmath155 and @xmath156 are real ) , the nonlinear terms of eqs.([system ] ) , namely @xmath157 and @xmath158 , can be written as @xmath159 a_x e^{i \phi_x } , \nonumber \\ n_z & = & \left [ \left(1+\gamma e^{2i(\phi_x-\phi_z)}\right ) a_x^2 + ( 1+\gamma ) a_z^2 \right ] a_z e^{i \phi_z } , \nonumber \\\end{aligned}\ ] ] from which we note that @xmath8 is responsible for a nonlinear coupling between the two field components which is highly sensitive to the phase difference @xmath160 .
the case @xmath161 is particularly interesting since in eqs.([nonli ] ) the terms proportional to @xmath162 vanish so that each field component is nonlinearly driven only by the other component and such coupling is fully sensitive the phase difference . also intriguing is the regime @xmath163 since , two major opposite situations exist .
if the phases of @xmath164 and @xmath165 are equal , from eqs.([nonli ] ) it is evident that the overall nonlinear polarization is proportional to @xmath162 so that , if @xmath166 , the effective nonlinear response can be enhanced .
if , on the other hand , the phase difference between @xmath164 and @xmath165 is @xmath21 , the overall nonlinear response of eqs.([nonli ] ) solely contains the terms @xmath167 and @xmath168 , so that , if @xmath163 ( for realistic bounded fields ) the compensation @xmath169 has to occur in order to prevent the divergence of the term proportional to @xmath8 in eqs.([system ] ) .
as explained in the above section , the nonlinear response of the proposed kerr metamaterial is easy to manage and , since the accesible ranges of its parameters are very wide , one can devise situations where the medium supports a genuinely novel nonlinear electrodynamical phenomenology . in order to probe the novel regime , we consider here a class of fields which are sufficiently simple to admit full analitical treatment and , at the same time , rigged with enough structure to show many of the novel effects .
more specifically we consider nonlinear guided waves propagating along the @xmath170axis of the form @xmath171 where @xmath172 is a real constant and the amplitudes @xmath173 and @xmath174 are real .
substituting the field of eqs.([solit ] ) into eqs.([system ] ) we obtain @xmath175 u_x , \nonumber \\ & & \beta \frac{d u_x}{d \xi } - \frac{d^2 u_z}{d \xi^2 } = \sigma_\mu \left [ \sigma_\epsilon + \psi_z \right ] u_z\end{aligned}\ ] ] where @xmath176 using eqs.([ga1 ] ) , it is worth noting that , from the first of eqs.([tmconst ] ) , the vector @xmath177 can be expressed as @xmath178 from which we note that @xmath179 act , for the fields of eqs.([solit ] ) , as effective nonlinear dielectric permittivity .
the quantities of eqs.([ga1 ] ) play a fundamental role in our discussion since it is evident that , if the conditions @xmath180 and @xmath181 do not hold along the profile of a nonlinear guided wave , the linear and nonlinear contribution in eq.([dx ] ) are comparable , i.e. the considered nonlinear guided wave belongs to the extreme nonlinear regime we have discussed in sec.[enr ] .
we consider solutions of eqs.([syste ] ) with definite parity where @xmath173 and @xmath174 are spatially even ( @xmath182 ) and odd ( @xmath183 ) , respectively and , as a consequence , we adopt the boundary conditions @xmath184 it is worth stressing that , due to the feasible possibility of arbitrary choosing the effective linear and nonlinear parameters characterizing the effective nonlinear medium , we consider here solutions of eqs.([syste ] ) and ( [ boundary1 ] ) for @xmath185 , @xmath186 and , remarkably , for any real @xmath8 . in appendix
a we show that the system of eqs.([syste ] ) is integrable and we derive the existence conditions characterizing the nonlinear guided waves satisfying eqs.([boundary1 ] ) , i.e , for each possible combinations of @xmath187 and @xmath188 , we derive a @xmath8 dependent range of @xmath189 ( i.e. @xmath190 ) corresponding to nonlinear waves existence .
.guided waves existence ranges of @xmath189 depending on @xmath8 , @xmath187 and @xmath188 . [ cols="^,^,^",options="header " , ] the resulting phenomenology is reported in table i. for each obtained nonlinear guided wave the propagation constant @xmath172 and the asymptotical value @xmath191 are given by ( see appendix a ) @xmath192 , \nonumber \\ u_{z \infty}^2 & = & \frac{-\sigma_\epsilon - ( 1-\gamma ) u_{x \infty}^2}{1+\gamma}\end{aligned}\ ] ] where the second expression for @xmath172 is obtained from the first by exploiting the first of eqs.([ga1 ] ) .
it is worth noting that , for each @xmath193 , these values of @xmath172 and @xmath191 are obtained by requiring that the nonlinear guided wave is asymptotically spatially uniform ( i.e. by annulling the derivatives in eqs.([syste ] ) for @xmath194 ) and , exploiting eqs.([eeff ] ) , this implies that @xmath195 from a physical point of view , the second of eqs.([efflim ] ) states that the considered nonlinear waves can propagate through the medium only if , asymptotically , the nonlinear term exactly balances the linear one to yield an overall vanishing dielectric constant , i.e. @xmath196 this proves that the nonlinear guided waves we are considering always belong to the extreme nonlinear regime , i.e. they can not be observed in standard kerr media .
it is worth stressing that the sub - family of nonlinear guided waves with @xmath197 do not require such an asymptotical compensation mechanisms since , for @xmath194 , the second of eqs.([syste ] ) vanishes together with @xmath191 . to sum up
, an asymptotically spatially uniform wave can exist only if @xmath198 ( since @xmath199 ) and this can be achieved by requiring either @xmath197 or @xmath200 , the second situation being investigated in the present paper .
since the waves we are investigating always belong to the extreme nonlinear regime , power flow reversing discussed in sec.[pfr ] is expected . for the field of eqs.([solit ] ) , the time averaged poynting vector @xmath201 $ ] is given by @xmath202 \hat{\bf e}_z , \nonumber \\
{ \bf s } & = & s_0 \frac{\textrm{sign}(\chi)}{\beta } \left(\sigma_\epsilon + \psi_x \right ) u_x^2 \hat{\bf e}_z\end{aligned}\ ] ] where @xmath203 and the second expression is obtained from the first one by exploiting the first of eqs.([syste ] ) to eliminate the derivative of @xmath174 and the the first of eqs.([ga1 ] ) .
note that eqs.([ssz1 ] ) coincide with eqs.([sz ] ) ( with the identification @xmath204 ) , the difference lying in the fact that , for the nonlinear guided waves we are considering , eqs.([ssz1 ] ) are exact due to the waves propagation invariance .
the situation @xmath205 and @xmath206 is very intriguing since @xmath207 and therefore , in the absence of nonlinearity , no propagation can occur since the medium can support only evanescent waves . on the contrary
, we have shown that , in this situation ( see first column of table i ) , the nonlinear kerr metamaterial can support propagating nonlinear guided waves ( i.e. with real propagation constant @xmath172 ) . in order to grasp this fact ,
we note , from the first of eqs.([beta ] ) , for @xmath205 and @xmath208 , we obtain @xmath209 from which it is evident that , in the absence of the nonlinearity , @xmath210 , i.e. only evanescent waves exist .
however , combining the first of eqs.([ga1 ] ) and the second of eqs.([beta ] ) , we obtain @xmath211 from which it is evident that , for @xmath212 and @xmath213 ( see the first column of table i ) , @xmath214 so that @xmath215 and the nonlinear wave can propagate through the medium . from a physical point of view
, the same result can be understood by regarding the nonlinear guided wave as a background infinite plane wave with a distortion around @xmath216 .
the background nonlinear plane wave can propagate through the medium since its amplitude is such that @xmath217 so that the effective nonlinear dielectric permittivity in the first of eqs.([eeff ] ) is such that @xmath218 > 0 $ ] . in other words , in the extreme nonlinear regime ,
the nonlinear polarizability overcomes the linear dielectric contribution in such way that the sign of the effective overall dielectric permittivity is opposed to that of the linear dielectric permittivity and the waves are consistently not evanescent .
( subplot ( a ) ) and @xmath219 ( subplot ( b ) ) at different values of @xmath193 for @xmath220 , @xmath221 , @xmath222.,scaledwidth=45.0% ] and @xmath174 of various nonlinear guided waves with @xmath223 and @xmath224 and with @xmath8 spanning the range @xmath225 .
all the waves are characterized by the same asymptotical value @xmath226 . for each @xmath8 two waves
exist , the first being reported in panel ( a ) and ( b ) and the second in panel ( c ) and ( d),scaledwidth=45.0% ] component @xmath227 evaluated for the nonlinear guided waves of fig.(3 ) . here
@xmath228 , @xmath229 and @xmath230 .
panel ( a ) contains @xmath227 evaluated for the waves reported in panel ( a ) and ( b ) of fig.(3 ) .
panel ( b ) contains @xmath227 evaluated for the waves reported in panel ( c ) and ( d ) of fig.(3 ) . in the inset of panel ( a ) , the width @xmath231 ( root - mean - square deviation ) of the various @xmath227 reported in panel ( a ) is plotted as a function of @xmath8.,scaledwidth=45.0% ] in fig.(2 ) we plot various profiles of @xmath232 and @xmath219 for @xmath233 and for @xmath234 . in fig.(3 ) we plot the profiles of @xmath232 and @xmath219 for different nonlinear guided waves , in the range @xmath235 , each characterized by the same asymptotical value @xmath236 .
as explained in appendix a , more than one nonlinear guided wave ( for a given @xmath8 ) can be generally found for each @xmath193 and , in the situation of fig.(3 ) , there are specifically two waves , the first being reported in panel ( a ) and ( b ) and the second in panel ( c ) and ( d ) .
the most striking feature emerging from fig.(3 ) is that , for a given @xmath193 the closer @xmath8 to @xmath237 , the sharper the profile of @xmath173 , i.e. @xmath8 produces an hyper - focusing effect for this waves when it approaches the value @xmath161 .
as expected ( see the discussion of sec.[design ] ) the situation @xmath161 is peculiar displaying a phenomenology absent for other values of @xmath8 . from a physical point of view the hyper - focusing effect results from the combination of two different mechanisms .
for the first one , we note that , since @xmath200 ( see the second of eqs.([efflim ] ) ) , if @xmath8 is very close to @xmath237 , from the second of eqs.([eeff ] ) and the second of eqs.([ga1 ] ) , we conclude that @xmath191 is much greater than one ( in agreement with the second of eqs.([beta ] ) evaluated for @xmath17 ) .
therefore , @xmath238 is much greater than one everywhere apart a small region around @xmath239 where @xmath174 is very small ( since @xmath240 ) and @xmath241 is very large .
the second mechanism supporting hyper - focusing is based on the fact that , for @xmath242 the field @xmath243component ( @xmath173 ) is nonlinearly driven solely by @xmath174 ( see the discussion of sec.[design ] ) .
therefore , in the region where @xmath174 is very small and @xmath241 is very large , from the first of eqs.([syste ] ) , it is evident that @xmath232 displays a very pronounced peak , i.e. @xmath173 is tightly squeezed by @xmath174 for @xmath17 . in fig.(4 )
we report the @xmath119-component of the normalized poynting vector @xmath227 ( see eqs.([ssz1 ] ) ) evaluated for the waves reported in fig.(3 ) .
note that the above discussed hyper - focusing effectively corresponds to a tight energy localization around @xmath239 for @xmath8 close to @xmath237 .
this effect is particularly evident from the inset of panel ( a ) of fig.(4 ) where we plot the peak width @xmath231 of @xmath227 as a function of @xmath8 .
the power flows reported in panel ( b ) of fig.(4 ) clearly displays the transverse power flow reversing discussed in sec.[pfr ] since there is a region around @xmath244 where @xmath245 whereas @xmath246 elsewhere . note that the power flows reported in panel ( a ) of fig.(4 ) does not exhibit transverse power flow reversing and this can be easily understood considering the structures of the two waves reported in fig.(3 ) .
the @xmath170component of the first wave is such that @xmath247 ( see panel ( b ) of fig.(3 ) ) so that , from the first of eq.([ssz1 ] ) , the two bell shaped contributions @xmath248 and @xmath249 are both positive yielding the @xmath250 . on the other hand ,
the second wave is such that @xmath251 ( see panel ( d ) of fig.(3 ) ) and therefore , in the first of eq.([ssz1 ] ) , the two contributions have different signs and @xmath227 can flip its sign along the transverse profile . for @xmath205 and @xmath252 ,
waves are allowed to propagate in the linear regime and the nonlinear guided waves can propagate if @xmath253 ( see the first of eqs.([beta ] ) ) , i.e. the background nonlinear plane wave can not produce a nonlinear dielectric response overcoming the linear part .
( a ) , @xmath174 ( b ) and @xmath254 ( c ) of various nonlinear guided waves with @xmath223 and @xmath255 and @xmath256 .
all the waves are characterized by the same asymptotical value @xmath226.,scaledwidth=45.0% ] ( a ) , @xmath174 ( b ) and @xmath254 ( c ) of various nonlinear guided waves with @xmath223 and @xmath255 and @xmath257 .
all the waves are characterized by the same asymptotical value @xmath226.,scaledwidth=45.0% ] as opposed to the case discussed in sec.[ngw1 ] , in the present situation there are two distinct families of nonlinear guided waves corresponding to the ranges @xmath256 and @xmath257 ( see the second column of table i ) . in fig.5
we report different nonlinear guided waves profiles together with their power flows for different values of @xmath258 and for @xmath236 .
from panel ( a ) of fig.5 we note that , in this regime , the profile of @xmath173 has a peak and a hole ( around @xmath239 ) for small and large values of @xmath8 , respectively .
this behavior is easily understood since for @xmath259 we have @xmath260 ( from the first of eqs.([beta ] ) ) and @xmath261 ( from the first of eqs.([syste ] ) evaluated at @xmath239 and for @xmath205 ) whereas , for @xmath262 we have @xmath263 and @xmath264 ( from the limit @xmath265 of the first of eqs.([syste ] ) for @xmath205 ) . as a consequence
there must be a value of @xmath8 for which @xmath173 is uniform . requiring that @xmath266 , eqs.([syste ] ) can be casted in the form @xmath267 so that the solution of the second of eqs.([syste1 ] ) fulfilling the boundary conditions of eqs.([boundary1 ] ) is @xmath268\ ] ] or , in other words , the longitudinal component @xmath174 is an exact electromagnetic dark soliton .
however , the obtained @xmath174 has to satisfy the full maxwell system so that , substituting the field in eq.([dark ] ) into the first of eqs.([syste1 ] ) , it is straightforward to prove that this is possible only if @xmath269.\ ] ] for @xmath236 we obtain @xmath270 which is the value of @xmath8 at which @xmath173 is uniform in panel ( a ) of fig.5 . note that the situation where @xmath173 is uniform and @xmath174 is a dark soliton is possible only in the extreme nonlinear regime where the nonlinearity @xmath271 can compensate the linear part in the second of eqs.([syste ] ) . from panel ( c ) of fig.5 it is evident that every considered waves exhibit the transverse power flow reversing discussed in sec.[pfr ] .
in fig.6 we report different nonlinear guided waves profiles together with their power flows for different values of @xmath272 and for @xmath273 .
let us consider , in this case , the behavior of the nonlinear guided waves for large values of @xmath20 . from panel ( a ) and ( b ) of fig.6
it is evident that a kind of saturation occurs , i.e. @xmath173 and @xmath174 approaches their asymptotic profiles for @xmath274 .
this is consistent with the fact that , taking the limit @xmath275 of eqs.([beta ] ) we obtain @xmath276 i.e. @xmath277 and @xmath278 asymptotically approaches two finite asymptotic values and it is relevant that @xmath279 . in order to obtain the asymptotic profiles of @xmath173 and @xmath174 , we note that , taking the limit @xmath275 of eqs.([syste ] ) and consistently assuming that the profiles remains everywere finite , consistency requires that @xmath280 so that , asymptotically , @xmath281 for @xmath282 and @xmath283 for @xmath284 ( since @xmath173 and @xmath174 are spatially even and odd , respectively ) . exploiting this property , eqs.([syste ] ) yields , for @xmath275 , the equation and @xmath174 ( solid lines ) of nonlinear guided waves for @xmath285 and @xmath286 . the asymptotical ( @xmath275 )
profiles of eqs.([asympt ] ) are also reported ( dotted line).,scaledwidth=45.0% ] @xmath287 so that , the asymptotic nonlinear guided wave profiles are @xmath288 note that the asymptotic profile of @xmath173 is singular at @xmath239 ( i.e. it continuous but not differentiable ) as a consequence of the unrealistic assumption @xmath275 .
even though the profiles of eqs.([asympt ] ) are not physical , they show that both @xmath248 and @xmath289 asymptotically approaches the square of an electromagnetic dark solitons and they provide a very good description of the above mentioned saturation for @xmath290 .
the profiles of @xmath173 and @xmath174 for @xmath285 and @xmath286 together with the asymptotical profiles of eqs.([asympt ] ) are plotted in fig.(7 ) from which it is evident that , for @xmath291 , the asymptotical profiles of eqs.([asympt ] ) accurately describe the actual field profiles ( apart a small region around @xmath239 for @xmath173 ) . we conclude that , in the extreme nonlinear regime , the compensation between @xmath292 and @xmath271 and the linear terms , if @xmath163 , forces @xmath248 and @xmath289 almost to coincide .
in conclusion , we have investigated the effective response of a nonlinear kerr metamaterial obtained by homogenizing a one dimensional layered periodic structure .
the effective response formally coincides with that of a standard nonlinear kerr medium with the important difference that its parameters ( both linear and nonlinear ) can be independently designed and that they can assume even values not achievable in standard material . as a consequence we can choose the linear dielectric permittivity to be much smaller than one thus allowing the observation of the regime where the nonlinear polarization can not be regarded as a small perturbation ( extreme nonlinear regime ) . as a leading general phenomenon characterizing the extreme nonlinear regime
we have discussed the transverse power flow reversing effect , i.e. the fact that the power flow can change its sign along the transverse beam profile . combining the extreme nonlinear regime and
the fact the effective kerr response can be tailored in an unconventional way , we have discussed a number of novel phenomena exploiting a class of fields ( nonlinear guided waves ) admitting full analytical description .
examples of such novel effects are the fact that the nonlinear waves can propagate even if the medium linear properties ( dielectric permittivity and magnetic permeability ) would forbid propagation , hyper - focusing induced by the phase difference between the field components ( in the case @xmath293 ) and extreme compensation between the field components if they are @xmath21 out of phase in the limiting situation @xmath294 .
in order to derive the existence conditions of nonlinear waves satisfying eqs.([syste ] ) and eqs.([boundary1 ] ) , it is convenient to cast eqs.([syste ] ) into the standard form of a first order system of differential equations .
differentiating the first of eqs.([syste ] ) and substituting the obtained expression of @xmath295 into the second of eqs.([syste ] ) , we obtain , after some algebra @xmath296 u_x , \nonumber \\ \beta \frac{d u_x}{d \xi } & = & \frac { \beta^2 \sigma_\epsilon + ( 1-\gamma)\left\ { ( 2\sigma_\epsilon \sigma_\mu - \beta ^2 ) + 2 \sigma_\mu \left [ ( 1+\gamma ) u_x^2 + ( 1-\gamma ) u_z^2 \right ] \right\ } u_x^2 + \beta^2 ( 1+\gamma ) u_z^2}{\sigma_\epsilon + \left [ 3(1+\gamma ) u_x^2 + ( 1-\gamma ) u_z^2 \right ] } u_z\end{aligned}\ ] ] which is a system of ordinary differential equations equivalent to maxwell equations provided the relation @xmath297 \neq 0\ ] ] holds along the whole profile @xmath232 , @xmath219 @xcite .
the system of eqs.([syst ] ) can be fully analytically investigated since it admits the first integral @xmath298 \nonumber \\ & & -\frac{1}{2 } \sigma_\mu ( 1+\gamma ) ( u_x^4+u_z^4)- \sigma_\mu ( 1-\gamma ) u_x^2 u_z^2 \nonumber \\ & & -\frac{1}{\beta^2 } \left\ { ( \beta^2 - \sigma_\epsilon \sigma_\mu ) - \sigma_\mu \left[(1+\gamma)u_x^2+(1-\gamma)u_z^2 \right ] \right\}^2 u_x^2 \nonumber \\\end{aligned}\ ] ] or , in other words , the relation @xmath299 holds for any solution @xmath300 of eqs.([syst ] ) .
considering the boundary conditions of eqs.([boundary1 ] ) , since @xmath232 and @xmath219 have to asymptotically approach two constant values , their first and second derivatives vanish for @xmath301 so that we require the right hand sides of eqs.([syst ] ) to vanish at @xmath302 and @xmath303 . as a consequence
we obtain @xmath304 which are the relations expressing the propagation constant @xmath172 and the asymptotical amplitude @xmath191 as functions of the asymptotical amplitude @xmath193 .
in addition , exploiting the fact that the @xmath305 is constant along the wave profile , the relation @xmath306 in which @xmath172 and @xmath307 have been eliminated using eqs.([bet ] ) , is a cubic equation for @xmath308 which can be solved to yield the peak amplitude @xmath309 as a function of the asymptotical amplitude @xmath193 . for each @xmath193 ,
the existence of the corresponding guided waves has to be assured by a number of requirements . in first place
@xmath172 and @xmath191 have to be real so that , from eqs.([bet ] ) , we obtain @xmath310 which are necessary inequalities for guided waves existence .
analogously , it is necessary that @xmath309 , obtained by eq.([cubic ] ) , is real . on the other hand ,
the above boundary conditions for @xmath301 imply that , after substituting the expression of @xmath277 of the first of eqs.([bet ] ) into eq.([f ] ) , the function @xmath311 has a stationary point at @xmath312 .
therefore , since the curve @xmath313 of the plane @xmath314 has to reach the point @xmath312 , it has to be required that @xmath311 has a saddle point at @xmath312 and this leads to the necessary inequality @xmath315 \nonumber \\
\times \left[2 ( 1-\gamma ) u_{x \infty}^4 + \sigma_\epsilon ( 3-\gamma ) u_{x \infty}^2 + 1 \right ] > 0.\end{aligned}\ ] ] we conclude that the existence of the nonlinear guided waves we are considering is assured by the condition of eq.([excond ] ) , together with the reality of @xmath309 ( obtained by solving eq.([cubic ] ) ) and the inequalities of eqs.([ineq1 ] ) and ( [ ineq2 ] ) .
the fulfilment of all these requirements leads to the existence conditions reported in table i. aa s. ramakrishna and t. m. grzegorczyk , _ physics and applications of negative refractive index materials _
( crc press taylor & francis group , new york , 2009 ) . w. cai and v. shalaev , _ optical metamaterials : fundamentals and applications _
( springer , dordrecht , 2010 ) j. b. pendry , phys .
* 85 * , 3966 ( 2000 ) .
n. fang , h. lee , c. sun and x. zhang , science * 308 * , 534 ( 2005 )
. j. b. pendry , d. schurig and d. r. smith , science * 312 * , 1780 ( 2006 ) .
d. schurig , j. j. mock , b. j. justice , s. a. cummer , j. b. pendry , a. f. starr , d. r. smith , science * 314 * , 977 ( 2006 ) .
n. engheta , science * 317 * , 1698 ( 2007 ) .
a. alu , m. g. silveirinha , a. salandrino and n. engheta , phys .
b * 75 * , 155410 ( 2007 ) .
m. silveirinha and n. engheta , phys .
* 97 * , 157403 ( 2006 ) .
r. liu , q. cheng , t. hand , j. j. mock , t. j. cui , s. a. cummer and d. r. smith , phys . rev . lett . * 100 * , 023903 ( 2008 ) b. edwards , a. alu , m. e. young , m. silveirinha and n. engheta , phys . rev .
* 100 * , 033903 ( 2008 ) .
a. a. zharov , i. v. shadrivov and y. s. kivshar , phys .
91 * , 037401 ( 2003 ) .
i. v. shadrivov and y. s. kivshar , j. opt . a : pure appl
* 7 * , 68 ( 2005 ) .
n. lazarides and g. p. tsironis , phys .
e * 71 * , 036614 ( 2005 ) .
n. a. zharova , i. v. shadrivov , a. a. zharov and y. s. kivshar , opt .
express * 13 * , 1291 ( 2005 ) .
r. s. hegde and h.t g. winful , opt . lett . * 30 * , 1852 ( 2005 ) .
j. b. pendry , a. j. holden , d.j .
robbins and w.j .
stewart , ieee transactions on microwave theory and techniques * 47 * , 2075 ( 1999 ) .
i. v. shadrivov , s. k. morrison and y. s. kivshar , opt .
express * 14 * , 9344 ( 2006 ) .
s. obrien , d. mcpeake , s. a. ramakrishna , and j. b. pendry , phys .
b * 69 * , 241101(r ) ( 2004 ) .
d. stroud and van e. wood , j. opt .
b * 6 * , 778 ( 1989 ) .
a. e. neeves and m. h. birnboim , j. opt .
b * 6 * , 787 ( 1989 )
. j. e. sipe and r. w. boyd , phys .
a * 46 * , 1614 ( 1992 ) .
g. l. fischer , r. w. boyd , r. j. gehr , s. a. jenekhe , j. a. osaheni , j. e. sipe and l. a. weller - brophy , phys .
* 74 * , 1871 ( 1995 ) .
j.p . huang and k.w .
yu , phys .
rep . * 431 * , 87 ( 2006 ) .
a. ciattoni , c. rizza and e. palange , `` transverse power flow reversing of confined waves in extreme nonlinear metamaterials '' , submitted for publication on prl r. w. boyd , _ nonlinear optics _ ( academic press , new york , 1994 ) .
g. allaire , siam j. math .
23 * , 1482 ( 1992 ) .
d. felbacq , g. bouchitt , b. guizal , a. moreau , j. nanophoton .
* 2 * , 023501 ( 2008 ) .
a. i. cabuz , d. felbacq , and d. cassagne , phys .
lett . * 98 * , 037403 ( 2007 ) .
p. a. belov and y. hao , phys .
b * 73 * , 113110 ( 2006 ) . s. a. ramakrishna and j. b. pendry , phys .
b * 67 * , 201101(r ) ( 2003 ) .
w. boyd and j. e. sipe , j. opt .
b * 11 * , 297 ( 1994 ) .
m. scalora , m. s. syrchin , n. akozbek , e. y. poliakov , g. daguanno , n. mattiucci , m. j. bloemer and a. m. zheltikov , phys .
lett . * * 9**5 , 013902 ( 2005 ) . y. hu , s. wen , h. zhuo , k. you and d. fan , opt .
express * 16 * , 4774 ( 2008 ) .
a. ciattoni , b. crosignani , p. di porto and a. yariv , j. opt .
b * 22 * , 1384 ( 2005 ) . | we consider a sub - wavelength periodic layered medium whose slabs are filled by arbitrary linear metamaterials and standard nonlinear kerr media and we show that the homogenized medium behaves as a kerr medium whose parameters can assume values not available in standard materials .
exploiting such a parameter availability , we focus on the situation where the linear relative dielectric permittivity is very small thus allowing the observation of the extreme nonlinear regime where the nonlinear polarization is comparable with or even greater than the linear part of the overall dielectric response .
the behavior of the electromagnetic field in the extreme nonlinear regime is very peculiar and characterized by novel features as , for example , the transverse power flow reversing . in order to probe the novel regime
, we consider a class of fields ( transverse magnetic nonlinear guided waves ) admitting full analytical description and we show that these waves are allowed to propagate even in media with @xmath0 and @xmath1 since the nonlinear polarization produces a positive overall effective permittivity . the considered nonlinear waves exhibit , in addition to the mentioned features , a number of interesting properties like hyper - focusing induced by the phase difference between the field components . |
in free space , a @xmath8 particle decays dominantly associated with a pion in the final state as @xmath8 @xmath9 . in the case of @xmath8 bound in a nucleus , a @xmath8 hypernucleus , @xmath8 is not only possible to decay as in the free space ( mesonic weak decay ) , but is also able to stimulate a nucleon in the nucleus as @xmath8``@xmath10''@xmath11 ( non - mesonic weak decay , nmwd ) .
the nmwd gives a unique opportunity to study the weak interaction between baryons , because this strangeness non - conserving process is purely attributed to the weak interaction . in the nmwd of a @xmath8 hypernucleus
, there are two decay channels , @xmath12 @xmath13 ( @xmath6 ) and @xmath14 @xmath15 ( @xmath4 ) .
the ratio of those decay widths , @xmath4/@xmath6 , is an important observable used to study the isospin structure of the nmwd mechanism . for the past 40 years
, there has been a longstanding puzzle that the experimental @xmath4/@xmath6 ratio disagrees with that of theoretical calculations based on the most natural and simplest model , the one - pion exchange model ( ope ) . in this model ,
the @xmath16 reaction is expressed as a pion absorption process after the @xmath17 decay inside the nucleus .
since the ope process is tensor - dominant and the tenser transition of the initial @xmath18 pair in the @xmath19-state requires the final @xmath20 pair to have isospin zero , the @xmath21 ratio in ope process becomes close to 0 .
however , previous experimental results have indicated a large @xmath21 ratio ( @xmath221 ) @xcite .
this large discrepancy between the ope - model predictions and the experimental results has stimulated many theoretical studies : the heavy meson exchange model @xcite , the direct quark model @xcite and the two - nucleon ( 2@xmath10 ) induced model ( @xmath23 ) @xcite .
after k. sasaki @xmath24 pointed out an error about the sign of the kaon exchange amplitudes in 2000 @xcite , those theoretical values of the @xmath21 ratio have increased to the level of [email protected] @xcite . on the other hand
, the experimental data still have large errors ( @xmath21 = 0.93 @xmath25 0.55 for @xmath0he @xcite ) , and it is hard to draw a definite conclusion on the @xmath21 ratio . so far , several nucleon energy spectra from hypernuclear decay have been reported .
most of the experiments measured only the proton energy spectra , because of the difficulty in detecting neutrons .
the @xmath21 ratio has been estimated through a comparison of the measured proton spectrum with that of the intra - nuclear cascade calculation by changing the @xmath21 ratio @xcite . in this method , there are large uncertainties due to the indirect evaluation of neutrons .
recently , the neutron energy spectra from @xmath1c and @xmath26y were measured with high statistics @xcite . for @xmath1c ,
the @xmath21 ratio was extracted from the neuron - to - proton yield ratio , @xmath27 , estimated by the neutron energy spectrum combined with a proton energy spectrum measured in another experiment @xcite .
the energy loss of protons in the target , however , was not able to be corrected for .
since there were ambiguities coming from different detection energy thresholds in the two measurements , the value might be affected by a large systematic error .
we therefore simultaneously measured the neutron and proton spectra from the decay of @xmath8 hypernuclei with a proton energy loss correction , and with much higher statistics over those of previous experiments .
when we compare the `` measured '' @xmath21 ratio with that obtained in theoretical calculations , the most serious technical problem was a treatment of the re - scattering effect in the residual nucleus , the so - called final state interaction ( fsi ) . the nucleon energy spectrum emitted from the two - body nmwd process , @xmath16 , must originally have a broad peak at half of the q - value ( @xmath22 76 mev ) .
the fsi effect distorts the nucleon spectra enhancing the low energy region , in which the strength of the effect depends on the mass number of the hypernuclei . in order to minimize the fsi effect , we selected a light @xmath19-shell hypernucleus , @xmath0he . in the @xmath19-shell hypernucleus
, initial relative @xmath18 states must be @xmath28 states , whereas in a @xmath29-shell hypernucleus it is possible to be @xmath30 states . according to ref .
@xcite , the transitions coming from the @xmath30 states were predicted to be small , though we need experimental confirmation for that . in order to investigate the @xmath29-wave effect
, we also performed the same experiment for a typical light @xmath29-shell hypernucleus , @xmath1c . in this paper , we show those energy spectra for both hypernuclei and present the neuron - to - proton yield ratios , @xmath27 , with a high energy threshold ( 60 mev ) in order to get rid of the contribution from the fsi effect
. on the other hand , the nucleon yields in the low energy region should be sensitive to the strength of the fsi effect .
in the proton measurement there is a cutoff energy , typically at 30 @xmath22 40 mev , due to the energy loss inside the target and the detector , whereas in the neutron measurement we have experienced sensitivity in the whole energy region . in this paper
, we also show the mass number dependence of the `` neutron '' energy spectra for a = 5 and 12 , with that for a = 89 @xcite , which provides valuable information to investigate the fsi effect .
the possible existence of a multi - nucleon induced process in which the q - value can be distributed over three or more nucleons in the final state has been discussed theoretically ( such as 2@xmath10-induced process ) , though there has been no experimental evidence so far ; the nucleons emitted from this process should also distribute in the low energy region .
thus , the whole shape of the neutron energy spectrum also brings useful knowledge on the multi - nucleon induced nmwd mechanism .
the present experiments ( kek - ps e462/e508 ) were carried out at the 12-gev proton synchrotron ( ps ) in the high energy accelerator research organization ( kek ) .
hypernuclei , @xmath0he and @xmath1c , were produced via the ( @xmath2,@xmath3 ) reaction at 1.05 gev/@xmath31 on @xmath32li and @xmath33c targets , respectively . since the ground state ( g.s . ) of @xmath34li is above the threshold of @xmath0he @xmath35 , the g.s .
promptly decays into @xmath0he emitting a low - energy proton .
the @xmath36li ( @xmath2,@xmath3 ) @xmath34li reaction was therefore employed to produce @xmath0he .
the hypernuclear mass spectra were calculated by reconstructing the momenta of incoming @xmath2 and outgoing @xmath3 using a beam - line spectrometer composed of the qqdqq system and the superconducting kaon spectrometer ( sks ) @xcite , respectively .
particles emitted from the decays of @xmath8 hypernuclei were detected by the decay - particle detection system , as shown in fig .
[ fig : e462setup ] .
it was composed of plastic scintillation counters ( t1 , t2 , t3 and t4 ) and a multi - wire drift chamber ( pdc ) .
the neutral particles , neutrons and gammas , were identified by means of the time - of - flight ( tof ) technique between t1 and t4 ; in addition , t3 was used as a veto counter in order to reject charged particles .
the charged particles , protons and charged pions , were identified by utilizing @xmath37 on t2 , total energy deposit on sequentially fired counters ( t2 , t3 and t4 ) , and the tof between t2 and t3 .
their trajectories were measured by pdc .
figure [ fig : ex ] shows the inclusive excitation - energy spectra of @xmath34li and @xmath1c .
peak is clearly seen in our excitation spectra for both hypernuclei .
we selected events from the hypernuclear production by gating the g.s .
peak , as shown in the figure .
the yields of @xmath0he and @xmath1c are , respectively , about 4.6 @xmath38 10@xmath39 and 6.2 @xmath38 10@xmath39 , which were one order - of - magnitude higher than those of previous experiments .
there is a constant background below the g.s .
peak ; the background for @xmath0he and @xmath1c within the g.s .
gate were estimated to be as small as 8% and 3% , respectively .
we estimated the contribution from this constant background in the nucleon spectra and subtracted it from the g.s .
gated nucleon spectra .
captypefigure in the neutral decay particle analysis , the tof between the reaction vertex and the t4 counter , @xmath40 , is expressed as @xmath41 , where @xmath42 and @xmath43 are the timings of t4 and t1 ( time zero ) , respectively ; @xmath44 means the flight time of the beam pion ; @xmath45 corresponds to the time delay due to the hypernuclear lifetime .
since @xmath45 can not be determined for each event individually , we replaced @xmath45 with the mean lifetime of the hypernuclei to avoid a systematic energy shift in the neutron energy spectra . here , we used the lifetimes of @xmath46 ps for @xmath0he and @xmath47 ps for @xmath1c , improved through the lifetime analysis of the present experiment @xcite .
the inverse of the neutron velocity , 1/@xmath48 , was calculated as 1/@xmath48 = @xmath40 / @xmath49 , where @xmath50 is the flight length and @xmath31 is the light velocity .
the top of fig .
[ fig : pid ] shows the measured 1/@xmath48 spectra with a light - output threshold of 2 mevee ( mev electron equivalent ) for @xmath1c .
the energy scale is displayed on the upper horizontal axis .
the gate region of the neutron was defined as 1.97 @xmath51 9.73 , which corresponds to 5 @xmath52 150 mev of neutron energy .
the spectrum shows good @xmath53/neutron separation .
the extremely low background level of the @xmath54 0 region in the spectrum indicates that the accidental background is negligibly small .
the neutron energy resolutions for @xmath0he and @xmath1c were estimated from the width of the @xmath53-ray peak .
they were , respectively , about 10 mev and 7 mev in full width at half maximum ( fwhm ) at @xmath5575 mev . the bottom of fig .
[ fig : pid ] shows the particle identification ( pid ) of the charged particles emitted from the decay of @xmath0he .
this plot was made by the pid function derived from the @xmath37 , the total energy and the particle velocity .
the charged particles ( charged pions , protons and deuterons ) were separated well .
it is noteworthy that deuterons were separated from the protons for the first time in such counter experiments for the nmwd .
the proton peak was fitted by a gaussian function , and the proton gate was set at the @xmath25 2 @xmath56 region of the gaussian , as shown in the figure .
the loss of protons ( 4.6% ) by this tight cut was taken into account in the analysis efficiency .
the proton kinetic energy was determined by the tof between t2 and t3 , which must be affected by the energy losses inside the target and in the t2 counter .
those energy losses were estimated event - by - event from the path length inside the target and the t2 counter , which was calculated from the reaction vertex and the trajectory .
captypefigure the numbers of the neutron and proton per nmwd at energy @xmath57 , @xmath58 and @xmath59 , are expressed as @xmath60 where @xmath61 is the number of g.s .
formations in the inclusive spectrum ; @xmath62 and @xmath63 are the numbers of detected neutrons and protons from the decay . the acceptances for the neutron and proton detection system , @xmath64 and @xmath65 ,
were estimated respectively to be 26.6 @xmath25 0.8% and 10.0 @xmath25 0.3% by a monte - carlo simulation . only for the proton analysis , we selected the central parts of the segmented t2 and t3 counters of the top and bottom coincidence arms in order to suppress ambiguity of the acceptance estimation due to escape from the side edge of the t4 counter .
the difference between @xmath64 and @xmath65 is attributed to this selection .
the non - mesonic decay branching ratios of @xmath0he and @xmath1c , @xmath66 , were evaluated as 0.429 @xmath25 0.012(stat ) @xmath25 0.005(sys ) and 0.768 @xmath25 0.005(stat ) @xmath25 0.012(sys ) , respectively , by @xmath67 and @xmath68 branching - ratio analyses of the present experiment @xcite .
the energy - dependent detection efficiency for the neutron , @xmath69 , was estimated by the monte - carlo simulation code demons @xcite , which is based on cecil @xcite and is applicable to multi - element neutron detectors .
we compared the calculated efficiencies with those of various existing experimental data @xcite , and found that the systematic error does not exceed 6% for integrated yields above 10 , 20 , 30 and 40 mev . according to a monte - carlo simulation for the proton
, there is no experimental sensitivity to a proton below about 30 mev , due to the energy loss , and the detection efficiency was gradually reduced from 50 mev to 30 mev .
the energy - dependent detection efficiency for the proton , @xmath70 , was calculated using the simulation .
there is a non - negligible nucleon background due to the pion absorption process in which @xmath67 s from the mesonic decay of @xmath8 hypernucleus are absorbed by the materials around the target .
the background was estimated by assuming that the shape of the nucleon spectra from this @xmath67 absorption process is the same as that from the @xmath67 decay of @xmath8 ( @xmath8 @xmath71 ) formed via the quasi - free formation process . here
, we regarded the @xmath67 branching ratio of the quasi - free @xmath8 decay as that of @xmath8 in the free space , and estimated the contribution using the @xmath67 branching ratios , 0.359 @xmath25 0.009 for @xmath0he @xcite and 0.099 @xmath25 0.011(stat ) @xmath25 0.004(sys ) for @xmath1c @xcite .
the top and middle of fig .
[ fig : npene ] show the neutron / proton energy spectra from the decay of @xmath0he and @xmath1c , respectively .
they were normalized per nmwd .
it seems that the neutron spectra for both hypernuclei have a similar shape to those of protons above the proton energy threshold of 30 mev , and the neutron yields were about twice higher than those of protons .
thanks to the event - by - event energy - loss correction for proton , we can directly compare the neutron energy spectra with those of the protons at the same energy threshold .
captypefigure it should be noted that the partial decay rate of the @xmath0he @xmath72 process , which emits a monochromatic high - energy neutron ( @xmath22 135 mev ) , is not negligible .
the decay rate was reported as 0.049 @xmath25 0.01 @xmath73 @xcite , where @xmath73 is the @xmath67 decay width of @xmath0he .
though it is hard to observe the corresponding peak due to the limited resolution , @xmath74 mev ( fwhm ) at such a high - energy region , one can see an appreciable neutron yield at the high - energy region ( 120 @xmath22 150 mev ) in the neutron spectra of @xmath0he , which can be considered to be the events from this process . the fsi process and the possible multi - nucleon induced process ( such as the 2@xmath10-induced process ) tend to enhance the low - energy region in the nucleon energy spectra . in order to reduce the effects from those processes
, we set the energy threshold to be as high as 60 mev , which is near to half of the q - value ( @xmath22 76 mev ) of the @xmath16 decay process .
the ratios of the yields between neutrons and protons , @xmath27 , for @xmath0he and @xmath1c above 60 mev , were obtained respectively as follows : @xmath75 where we set the upper energy limit of 110 mev only for @xmath0he in order to avoid the contamination from the @xmath0he @xmath72 process .
systematic errors are predominantly coming from an ambiguity of the absolute neutron detection efficiency ( @xmath226% ) . due to the similar spectrum shape for neutrons and protons , the @xmath27 ratios of both hypernuclei are quite stable for a change of the energy thresholds . recently ,
an independent experimental result of the @xmath27 on @xmath1c was reported as @xmath76 @xcite , which agrees with our result within the errors .
if the fsi effect and the contribution of the 2@xmath10-induced process can be neglected , the relation between the @xmath21 and the @xmath77 for @xmath16 decay process is approximately expressed as @xmath78 when we applied the above relation to the obtained @xmath77 , the @xmath21 ratios for both hypernuclei were obtained as about [email protected] .
the result excludes earlier experimental results that the @xmath21 is close to unity @xcite .
it also rules out a theoretical calculation based on the ope model , which predicts that the value should be as small as 0.1 .
our result supports recent calculations based on short - range interactions , such as the heavy - meson exchange and the direct quark exchange .
moreover , we could not observe significant difference of the @xmath21 ratio between the @xmath19-shell and @xmath29-shell hypernuclei .
it is suggested small contribution from the @xmath29-wave in this process . the bottom of fig .
[ fig : npene ] shows the neutron spectrum per nmwd of @xmath26y obtained in a previous experiment @xcite . with this spectrum ,
the mass - number dependence of the neutron energy spectra for a = 5 , 12 and 89 is shown .
the integrated numbers of neutrons per nmwd of @xmath0he , @xmath1c and @xmath26y are listed in table [ table : numneunmwd ] for energy thresholds of 10 , 20 , 30 and 40 mev .
the high - energy region is suppressed with the increase of mass number while enhancing the low - energy component .
this tendency can be naturally interpreted as being the fsi effect .
.total number of neutrons per nmwd of @xmath79he , @xmath1c and @xmath26y @xcite with energy thresholds of 10 , 20 , 30 and 40 mev .
[ cols="^,^,^,^ " , ] if the 1@xmath10-induced process ( @xmath80 ) dominates the nmwd and the fsi effect is negligible , the neutron energy spectrum should have a broadened peak , due to the fermi motion , at about one half of the q - value of the process ( @xmath2275 mev ) .
we simply expected that this would be the case for @xmath0he due to its small mass number .
garbarino @xmath24 reported a calculated nucleon energy spectrum for @xmath0he with the fsi effect and assuming that @xmath81 and @xmath82 , where @xmath83 and @xmath84 denote the decay widths of the 1@xmath10- and 2@xmath10-induced processes @xcite .
it is overlaid on the top of fig .
[ fig : npene ] ( dashed histogram ) on the same scale .
the observed spectrum was significantly different from the calculated energy spectrum .
the calculated spectrum still has its maximum at about q/2 , whereas the experimental spectrum levels off in the region from 20 to 80 mev .
it should be noticed that there are so far two types of the 2@xmath10-induced nmwd models .
one is a three - body reaction in which the q - value can be distributed over three nucleons in the final state .
the other is based on the correlated two - nucleon absorption of a virtual pion emitted from the weak vertex of @xmath85 , in which the energy of the emitted nucleon is small , since the virtual pion is close to on - shell @xcite . since most of the q - value in this 2@xmath10-induced process
is distributed to two nucleons , the shape of the energy spectra should be close to that in the 1@xmath10-induced process ( @xmath80 ) .
thus , the latter model adopted in the calculated spectra has a much smaller effect to enhance the low - energy region than the former model .
this suggests the importance of the former 2@xmath10-induced model .
another speculation could be that the fsi effect is much stronger than that in the calculation by garbarino @xmath24
we measured the energy spectra of neutrons and protons emitted from the non - mesonic decays of @xmath0he and @xmath1c with high statistics .
both the neutron and proton spectra show a similar shape .
the obtained neutron - to - proton yield ratios , @xmath86 , for both hypernuclei are approximately equal to two with high energy thresholds , which are consistent with a simple estimation assuming @xmath87 0.5 .
this result rules out a theoretical calculation based on the ope model , and supports recent calculations based on short - range interactions . in the present experiment ,
the high statistics of the observed nucleons has enabled us to carry out a coincidence analysis of the two nucleons , @xmath88 and @xmath89 , which is now in progress .
the analysis will provide a definitive @xmath21 ratio . in the obtained neutron spectrum for @xmath0he ,
the spectral shape was not consistent with the simple expectation that there is a broad peak at one half of the q - value .
this indicates the importance of the multi - nucleon induced process in the nmwd or / and a large fsi effect , even for @xmath0he .
this result calls for more detailed studies concerning the 2@xmath10-induced nmwd process and the fsi effect .
we are grateful to prof . k. nakamura and the kek - ps staff for support of the present experiment .
we also acknowledge the staff members of the cryogenic group at kek . this work was supported in part under the korea - japan collaborative research program of kosef(r01 - 2000 - 000 - 00019 - 0 ) and krf(2003 - 070-c00015 ) . | we have simultaneously measured the energy spectra of neutrons and protons emitted in the non - mesonic weak decays of @xmath0he and @xmath1c hypernuclei produced via the ( @xmath2,@xmath3 ) reaction with much higher statistics over those of previous experiments .
the neutron - to - proton yield ratios for both hypernuclei at a high energy threshold ( 60 mev ) were approximately equal to two , which suggests that the ratio of the neutron- and proton - induced decay channels , @xmath4(@xmath5)/@xmath6(@xmath7 ) , is about 0.5 . in the neutron energy spectra , we found that the yield of the low - energy component is unexpectedly large , even for @xmath0he . , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , @xmath8 hypernuclei , non - mesonic weak decay , nucleon energy spectrum 21.80.+a , 13.30.eg , 13.75.ev |
the recent announcement of bicep2 observations @xcite of the tensor - to - scalar ratio @xmath3 claim the b - mode polarisation signatures to be produced completely due to the primordial gravity waves arising from the early inflation .
however , we think that this belief is misplaced and the observed value of @xmath4 must include the contributions not only from the early inflation but also the contributions from the present acceleration of the universe .
in fact , this acceleration being of the recent origin and of a much longer duration ( by many orders ) must contribute significantly , and so the later contributions on the b - mode polarisation of the cosmic microwave background radiation ( cmbr ) must be more marked than due to the early inflation , whose sole contribution was reported by planck @xcite .
we recall that the cosmological and astrophysical observations support the fact our universe is in accelerated expansion phase @xcite , albeit the exact form of scale factor has not yet been fixed by any observations . heading from this motivation
previously we adopted the quasi - exponential expansion that can also produce the significant tensor fluctuations of spacetime @xcite . in @xcite the cosmological constant @xmath5 with the energy density of the self - interaction of scalar bosons bound in a condensate is identified . in this approach , @xmath5 decay provides with a dynamical realization of spontaneous symmetry breaking .
the detailed kinematics of @xmath5 decay and the back reaction of the decay products on the @xmath5 dynamics are given in @xcite .
the slow evaporation regime is found here for a wide range of possible parameters of particle interactions .
the present paper is organised into five sections . in section
@xmath6 we study the mutual interaction among the higgs - like tachyon field components wherein the gravity waves unleashed by the earlier inflation may be further boosted by the present acceleration of the universe , and may appear beyond the planck upper bounds @xcite .
section @xmath7 is laid for @xmath1-diagnostic and section @xmath8 is devoted towards @xmath9-diagnostic and statefinder parameters for interacting components of tachyonic field .
the lagrangian for the tachyonic scalar field appears from string theory @xcite and is given from the action @xmath10 as @xmath11 whereas the energy - momentum tensor @xmath12 leads to pressure and energy density of the tachyonic scalar field as @xmath13 and @xmath14 for spatially homogeneous tachyonic scalar field we have @xmath15 @xmath16 here , we assume that radiation with equation of state @xmath17 also exists as one inherent component of same tachyonic scalar field . due to some physical mechanism
not known in detail at present to us , but that may be like a higgs mechanism , we can split the expressions ( [ n6 ] ) and ( [ n7 ] ) as @xmath18 @xmath19 from ( [ n8 ] ) and ( [ n9 ] ) it is seen that when we include radiation in tachyonic scalar field then one new exotic component also appears ( say , exotic matter since its energy density is negative ) with zero pressure .
thus , the tachyonic scalar field resolves into three components say @xmath20 , @xmath21 and @xmath22 .
the pressure and energy density of @xmath20 is given as @xmath23 this is nothing but the ` true ' cosmological constant because of its equation of state being @xmath24 .
the second component with @xmath25 can be identified as radiation with @xmath26 the last component is characterised by @xmath27 this component mimics dust matter but has negative energy density
. the exotic matter may include the dirac fermions as well as the majorana fermions whence the negative energy states turn into the positive energy states@xcite . in our earlier work
@xcite we allowed a small time dependent perturbation in the equation of state(eos ) of the cosmological constant with @xmath28 thus , with the perturbed eos , the true cosmological constant becomes a shifted cosmological parameter .
this has a bearing upon the eos of radiation and exotic matter , both .
therefore , these two entities turn into shifted radiation and shifted exotic matter respectively . with fixed energy density of field components ,
the expressions for the energy density and pressure of each component are given as below .
for the shifted cosmological constant one has @xmath29 @xmath30 and @xmath28 for shifted radiation , we have @xmath31 @xmath32 with @xmath33 in presence of perturbation the zero pressure of exotic matter turns into negative non - zero pressure due to shifted exotic matter which would also accelerate the universe like dark energy .
thus , the energy density and pressure for shifted exotic matter are now , respectively , given as @xmath34 @xmath35 with @xmath36
the interacting dark energy models have been recently proposed by several authors @xcite .
why must the components of tachyonic scalar field interact ?
this is one of the interesting questions about interaction . since all components are in thermodynamic non - equilibrium , therefore , to achieve an equilibrium state they must fall into mutual interaction .
we propose that the currently ongoing acceleration must also boost the tensor - scalar fluctuations caused by the early inflation , and the interaction among the components may be responsible for this boost beyond the planck bounds @xcite . with this motivation we study the interaction of these components assuming that even though the total energy of the perturbed field ( spatially homogeneous ) is kept conserved , yet
during interaction it can get reasonably violated for individual components .
the equations for conservation of energy with interaction are given as @xmath37 @xmath38 @xmath39 where @xmath40 and @xmath41 are the interaction strengths . in the above expressions ( [ n16]),([n17 ] ) and ( [ n18 ] ) the following broad conditions must govern the dynamics .
* condition(i ) * @xmath42 .
this corresponds to the following cases , ( i ) : : if @xmath43 , @xmath44 then the right hand side of ( [ n16 ] ) is negative while ( [ n17 ] ) and ( [ n18 ] ) are positive , respectively .
this means that there is energy transfer from shifted cosmological parameter to shifted radiation and shifted exotic matter , respectively .
thermodynamics allows this kind of transfer of energy .
( ii ) : : @xmath45 , @xmath46 implies that there is an energy transfer to shifted cosmological parameter from shifted radiation and shifted exotic matter .
* condition(ii ) * @xmath47 .
( i ) : : @xmath44 , @xmath43 would make the right hand side of ( [ n17 ] ) as positive and ( [ n16 ] ) and ( [ n18 ] ) as negative .
this shows that there is an energy transfer to shifted radiation from shifted cosmological parameter and shifted exotic matter .
thermodynamics again does not allow this kind interaction .
( ii ) : : @xmath45 , @xmath46 makes way for the energy transfer from shifted radiation to shifted exotic matter and shifted cosmological parameter .
* condition(iii ) * if @xmath48 then we have the following possibility ( i ) : : @xmath49 leads to an energy transfer to shifted radiation from shifted cosmological parameter , while the shifted exotic matter remains free from interaction with its energy density held conserved .
this type of interaction holds compatibility with the laws of thermodynamics .
( ii ) : : if @xmath50 , energy would flow from shifted radiation to shifted cosmological parameter , whereas the shifted exotic matter does not get involved in interaction mechanism .
thus , the conservation of energy for shifted exotic matter holds good .
( iii ) : : as an alternative , @xmath51 would pull the components of tachyonic scalar field out of mutual interaction like the standard @xmath5cdm model .
the second case of condition ( i ) and condition(ii ) violate the laws of thermodynamics , therefore , we are not interested in these types of interactions . the interaction of type condition ( iii ) has been previously discussed for two components of tachyonic scalar field in our earlier work @xcite .
the positivity of the quantity @xmath52 implies that @xmath40 should be large and positive . for
if @xmath40 had been large and negative then the second law of thermodynamics would have been violated and the cosmological constant ( as the dark energy candidate ) would have dominated much earlier withholding the structure formation against the present observations . also , @xmath53 should be positive and small since if it is negative and large then conservation of energy of tachyonic field is violated .
the interaction strengths @xmath40 and @xmath53 should depend on temperature also , but due to mathematical simplification following the occam s razor , we consider the interaction strength as independent of temperature . on the left hand side of energy conservation equation
are hubble parameter ( reciprocal of dimension of time ) and energy density thus it is natural choice that the interaction strength should be function of hubble parameter and energy density .
several authors have proposed different forms of @xmath54 @xcite .
owing to the lack of information regarding the exact nature of dark matter and dark energy ( as the cosmological constant or else ) we can not yet fix the exact form of interaction strength .
thus , with this motivation we present the form of interaction term heuristically as function of time rate of change in energy densities as 1 .
@xmath55 2 . @xmath56 where @xmath57 are proportionality constant .
the total conservation of energy of field is ensured by @xmath58 with @xmath59 and hubble parameter @xmath60 given from @xmath61\label{n21}.\end{aligned}\ ] ] from ( [ n16 ] ) , ( [ n17 ] ) and ( [ n18 ] ) , the functional form of energy density with redshift @xmath62 is given as @xmath63 where @xmath64 @xmath65 and @xmath66\nonumber\\-\left(\frac{\beta\bar{\rho}^{0}_{r}(4 + 3\varepsilon)}{4 + 3\varepsilon-\eta+\eta\beta}\right)[x^{4 + 3\varepsilon/1-\beta}-x^{\eta}]\label{n24}\end{aligned}\ ] ] where @xmath67 ( constant ) is defined as @xmath68 it is clearly seen that in the absence of perturbation , @xmath69 does not play any role in the evolution of the matter component and so delinks it from the coextensive evolution of the shifted cosmological parameter . in that case
, we have @xmath70 then , ( [ n24 ] ) would turn into @xmath71\ ] ] which , in the absence of interaction(@xmath72 ) , further yields @xmath73 as expected in @xmath5cdm model .
plot for variation of shifted exotic matter energy density @xmath74 with redshift ( @xmath75 for range @xmath76 to @xmath77 ) .
we assume @xmath78 , @xmath79 , @xmath80 and @xmath81 for @xmath82 and @xmath83 .
the red , green , blue and black curves correspond to @xmath84 and @xmath85 , respectively.,scaledwidth=50.0% ] variation of shifted radiation energy density @xmath86 with redshift ( @xmath75 for range @xmath76 to @xmath77 ) .
we assume @xmath78 , @xmath79 , @xmath80 and @xmath87 and @xmath88 for @xmath89 .
the red and green curves stand for @xmath87 and @xmath88 respectively .
, scaledwidth=50.0% ] variation of shifted radiation energy density @xmath86 with redshift ( @xmath75 for range @xmath76 to @xmath77 ) .
we assume @xmath78 , @xmath79 , @xmath80 and @xmath90 and @xmath85 for @xmath89 .
the blue and black curves refer to @xmath90 and @xmath85 , respectively .
, scaledwidth=50.0% ] variation of energy density for shifted cosmological parameter @xmath91 with redshift ( @xmath75 for range @xmath76 to @xmath77 ) .
we assume @xmath78 , @xmath79 , @xmath80 for @xmath83 .
the plot is independent of the chosen values of @xmath92 , scaledwidth=50.0% ]
recently , sahni _ et al _ @xcite introduced the redshift dependent function @xmath93 where @xmath94 is the normalized hubble function @xmath95 where the constants @xmath96 and @xmath97 are respectively given as @xmath98 @xmath99 and @xmath100 is the present density parameter in the spatially flat @xmath101 universe .
normalized hubble parameter @xmath102 with redshift ( @xmath75 for range @xmath76 to @xmath77 ) .
we assume @xmath78 , @xmath79 , @xmath80 and @xmath103 and @xmath88 for @xmath82 and @xmath83 .
blue and green curves correspond to @xmath87 and @xmath104 , respectively.,scaledwidth=50.0% ] normalized hubble parameter @xmath102 with redshift ( @xmath75 for range @xmath76 to @xmath77 ) .
we assume @xmath78 , @xmath79 , @xmath80 and @xmath105 and @xmath85 for @xmath82 and @xmath83 .
green and red curves refer to @xmath90 and @xmath106 , respectively.,scaledwidth=50.0% ] in our scenario ( perturbation + interaction ) the @xmath1-diagnostics is given as @xmath107 where @xmath108 @xmath109 and @xmath110 plot of @xmath1-diagnostics with redshift ( @xmath75 for range @xmath76 to @xmath77 ) .
we assume @xmath78 , @xmath79 , @xmath80 and @xmath87 and @xmath88 for @xmath83 , @xmath82 and @xmath89 .
the plot is independent of the chosen values of @xmath92.,scaledwidth=50.0% ] plot of @xmath1-diagnostics with redshift ( @xmath75 for range @xmath76 to @xmath77 ) .
we assume @xmath78 , @xmath79 , @xmath80 and @xmath111 and @xmath85 for @xmath83 , @xmath82 and @xmath89 .
green and black curves are plotted for @xmath90 and @xmath85 respectively .
, scaledwidth=50.0% ] the difference of the squares of the normalized hubble parameter @xmath102 at two different redshifts @xmath112 and @xmath113 is given as @xmath114 where @xmath115 measuring the values of @xmath116 from observations we can estimate the proportionality constants .
lll @xmath2 & @xmath117 & @xmath112 + @xmath118 & 0.1 & 1.1 + @xmath119 & 0.4 & 1.4 + @xmath120 & 1.3 & 2.3 + @xmath121 & 1.75 . &
2.75 + we further take a set of values @xmath2 at different redshifts given in table 1 @xcite . from these values , we obtain six non - linear equations .
thus , from ( [ m1 ] ) with @xmath122 @xcite we obtain six values of @xmath116 mentioned in table 2 .
lll @xmath116 & numeric value & @xmath123 + @xmath124 & @xmath125 & @xmath126 + @xmath127 & @xmath128 & @xmath129 + @xmath130 & @xmath131 & @xmath132 + @xmath133 & @xmath134 & @xmath135 + @xmath136 & @xmath137 & @xmath138 + @xmath139 & @xmath140 & @xmath141 + with the help of these two tables we have following six non - linear equations for @xmath116 @xmath142 + \nonumber \\
a_{2}[(1.1)^{4 + 3\varepsilon /1-\beta}-(1.4)^{4 + 3\varepsilon /1-\beta } ] \nonumber
\\ + c[(1.1)^{\eta}-(1.4)^{\eta}]=-0.783\label{m2}\end{aligned}\ ] ] @xmath143 + \nonumber \\
a_{2}[(1.1)^{4 + 3\varepsilon /1-\beta}-(2.3)^{4 + 3\varepsilon
/1-\beta } ] \nonumber \\ + c[(1.1)^{\eta}-(2.3)^{\eta}]=-4.308\label{m3}\end{aligned}\ ] ] @xmath144 + \nonumber \\ a_{2}[(1.1)^{4 + 3\varepsilon
/1-\beta}-(2.75)^{4 + 3\varepsilon /1-\beta } ] \nonumber \\ + c[(1.1)^{\eta}-(2.75)^{\eta}]=-6.618\label{m4}\end{aligned}\ ] ] @xmath145 + \nonumber \\ a_{2}[(1.4)^{4 + 3\varepsilon /1-\beta}-(2.3)^{4 + 3\varepsilon /1-\beta } ] \nonumber
\\ + c[(1.4)^{\eta}-(2.3)^{\eta}]=-3.525\label{m5}\end{aligned}\ ] ] @xmath146 + \nonumber
\\ a_{2}[(1.4)^{4 + 3\varepsilon /1-\beta}-(2.75)^{4 + 3\varepsilon /1-\beta } ] \nonumber
\\ + c[(1.4)^{\eta}-(2.75)^{\eta}]=-5.835\label{m6}\end{aligned}\ ] ] @xmath147 + \nonumber
\\ a_{2}[(2.3)^{4 + 3\varepsilon /1-\beta}-(2.75)^{4 + 3\varepsilon /1-\beta } ] \nonumber
\\ + c[(2.3)^{\eta}-(2.75)^{\eta}]=-2.310\label{m7}\end{aligned}\ ] ] where @xmath148 @xmath149
the two point @xmath1-diagnostics may be defined as@xcite @xmath150 and @xmath151-point @xmath1-diagnostics as @xmath152 using ( [ n32 ] ) and ( [ n33 ] ) we have @xmath9-diagnostic @xmath153 @xmath154 , @xmath155 and @xmath156 can be calculated from ( [ n28 ] ) at three different redshifts .
we would take up this study in future . here , we use the statefinders for constraining the interaction .
in addition to the cosmological constant several other candidates for dark energy ( quintom , quintessence , brane - world , modified gravity etc . , e.g. @xcite ) have been proposed . to differentiate different models of dark energy sahni _ et.al _
@xcite proposed statefinder diagnostics based on dimensionless parameters @xmath157 which are the functions of scale factor and its time derivative .
these parameters are define as @xmath158 and @xmath159 with deceleration parameter @xmath160 thus , re - writng the statefinders as , @xmath161 @xmath162 \label{n36}.\end{aligned}\ ] ] considering for interaction + perturbation scenario the statefinder parameters may be calculated as from ( [ n28 ] ) .
let @xmath163 where @xmath164 @xmath165 where @xmath166 @xmath167 from ( [ n39]),([n40 ] ) and ( [ n41 ] ) we can find the other statefinder parameter @xmath168 defined by ( [ n36 ] ) as given below @xmath169 \label{n42 } \end{aligned}\ ] ] where @xmath170 given as @xmath171 for non - interacting model i.e. , @xmath172 and @xmath72 and in the absence of perturbation these turn out to be the same as in @xmath5cdm model @xmath173 for @xmath174 and @xmath175 with negligible contribution of radiation .
we argue that the contributions from the recent acceleration of the universe must generate the tensor fluctuations in a way similar to early inflation .
it would amplify the tensor - to - scalar ratio .
in fact , being robust and of recent and much longer duration this may explain the high value of @xmath4 recently observed by bicep2 , since it must represent a the sum of inflationary as well as well as the post - inflationary current acceleration .
we have used the single tachyonic scalar field which splits due to some unknown ( apparently higgs - like ) mechanism into three components ( cosmological constant , radiation and dust matter ) which interact to achieve a thermodynamical equilibrium .
the interaction among the components at present may boost the tensor - to - scalar amplitude ratio caused by the early inflation and must appear as the enhanced signature beyond the bounds set by the planck observations .
the entire evolution of the universe arises from this process of interaction . due to consideration of radiation in this field
the dust matter appears with negative energy . a small perturbation allowed in eos of cosmological constant changes its status from a true cosmological constant to a shifted cosmological parameter .
similarly , status of radiation and dust matter changes to shifted radiation and shifted exotic matter . with
the perturbation shifted exotic matter gains non - zero negative pressure which also helps ( along with dark energy ) in the accelerated expansion of the universe .
total energy of field stays conserved but the field components mutually interact with interaction strength parameter @xmath54 resulting in local violation of energy conservation .
the interaction reflects in diagnostics with our choice of interaction strengths which are constrained by the astrophysical data of hubble parameter at different redshifts chosen ( @xmath176 and @xmath177 ) ( albeit a narrower range of redshifts would be more useful in determining dynamics at an epoch ) .
the @xmath178 diagnostics and statefinder parameters may be further explored for interaction @xmath179 perturbation model to check the important constraints .
p. a. r. ade _
et al_. , b. collaboration ( 2014 ) , arxiv:1403.3985 .
p. a. r ade _ et al_. , planck collaboration ( 2013 ) , arxiv:1303.5082 .
a. g. riess _
et al_. , astron.j . * 116 * , 1009 ( 1998 ) .
s. perlmutter _
el al_. , astrophys.j .
* 517 * , 565(1999 ) .
b. p. schmidt _
et al_. , astrophys.j . * 507 * , 46(1998 ) .
l. amendola , phys .
d * 62 * , 043511(2000 ) .
w. zimdahl and d. pavon , phys .
b * 521 * , 133(2011 ) .
m. m. verma , astrophys .
space sci . * 330 * , 101(2010 ) .
m. m. verma , texas symposium on relativistic astrophysics
texas 2010 , december 06 - 10 , 2010 , heidelberg , germany , ( _ proceedings of science _ , * 218 * , ( 2010 ) .
sissa , italy ) .
b. wang , y. g. gong and e. abdalla , phys .
b * 624 * , 141(2005 ) .
s. d. campo , r. herrera and d. pavon , ijmp d vol.20 , * 4 * 561(2011 ) .
h. wei and r. g. cai , phys .
d * 71 * , 043504(2005 ) .
h. wei and s. n. zhang , phys .
b * 644 * , 7(2007 ) . | it is proposed that the recently announced bicep2 value of tensor - to scalar ratio @xmath0 can be explained as containing an extra contribution from the recent acceleration of the universe . in fact
this contribution , being robust , recent and of much longer duration ( by a large order of magnitude ) may dominate the contribution from the inflationary origin . in a possible scenario , matter ( dark or baryonic ) and radiation etc .
can emerge from a single higgs - like tachyonic scalar field in the universe through a physical mechanism not yet fully known to us .
the components interact among themselves to achieve the thermodynamical equilibrium in the evolution of the universe .
the field potential for the present acceleration of the universe would give a boost to the amplitude of the tensor fluctuations of gravity waves generated by the early inflation and the net effects may be higher than the earlier planck bounds . in the process , the dark energy , as a cosmological constant decays into creation of dark matter .
the diagnostics for the three - component , spatially homogeneous tachyonic scalar field are discussed in detail .
the components of the field with perturbed equation of state are taken to interact mutually and the conservation of energy for individual components gets violated .
we study mainly the @xmath1 diagnostics with the observed set of @xmath2 values at various redshifts , and the dimensionless state - finders for these interacting components .
this analysis provides a strong case for the interacting dark energy in our model . |
the kondo - lattice model ( klm ) is one of the most analyzed models in solid state theory due to its large variety of applications .
quite generally , the klm describes the interaction between localized spins and a band of conduction electrons .
one particular class of systems , where such a situation can be realized , are the transition metal oxides , which have received considerable interest due to their rich phase diagram comprising different types of magnetic and orbital order and even supercondunctivity.@xcite this complexity stems from the interplay between the formation of narrow @xmath1-bands leading to a delocalization of these states on the one hand and the local part of the coulomb interaction between the @xmath1-electrons tending to localize them.@xcite a particularly interesting example is la@xmath2ca@xmath3mno@xmath4@xcite . in this cubic perovskite the five - fold degenerate @xmath1 level is split by crystal field into three - fold degenerate @xmath5 , which have the lower energy , and two - fold degenerate @xmath6 states .
these states have to be filled with @xmath7 electrons , nominally yielding a metal even for @xmath8 .
however , taking into account the local coulomb interaction , three of these electrons will occupy the @xmath5-states forming an @xmath9 high - spin state due to hund s coupling , which interacts ferromagnetically with the electron occupying the @xmath6 states . besides its complicated phase diagram with a large variety of paramagnetic and magnetically ordered metallic and insulating phases one finds a colossal magneto - resistance ( cmr).@xcite the physics just described
can be covered in great parts by the klm , in this context usually called double exchange model ( de).@xcite another class of materials which can be addressed by the klm with ferromagnetic exchange interaction between conduction states and the localized spin degrees of freedom are magnetic semiconductors or semi - metals in the series of the rare earth monopnictides and monochalcogenides.@xcite here the focus lies on the magnetic respectively magneto - optic properties of for example eus , euo , gdn or thin films of such systems.@xcite last but not least the ferromagnetic klm can be applied to investigate the magnetic properties of diluted magnetic semiconductors such as ga@xmath2mn@xmath3as . in these materials the iii - v semiconductor gaas becomes ferromagnetic by introducing magnetic ions,@xcite such as mn .
the local spin of the mn - ions couples ferromagnetically to the states of the semiconductor , a setup which can be modeled by the klm .
note , however , that in these materials disorder effects are expected to play a vital role.@xcite a completely different realization of the klm starts from the periodic anderson model ( pam),@xcite a model for heavy fermion physics.@xcite heavy fermion physics manifests itself in a number of lanthanide and actanide compounds , which have a very large effective mass in common .
the low - temperature physics of these compounds is determined by a partly filled f shell and hybridization induced spin flip scattering between the f and conduction electrons . by a schrieffer - wolff - transformation@xcite
the periodic anderson model can be mapped onto the klm with antiferromagnetic exchange interaction .
note that in these materials one typically expects a competition between the heavy fermion physics , driven by the kondo effect due to the antiferromagnetic coupling , and the formation of antiferromagentism.@xcite in most of the approaches one does not include coulomb interaction among the conduction band electrons . especially for manganites
this is quite likely an insufficient approximation , because all the physics takes place in a correlated d - band , as explained above .
there is no reason to ignore the local coulomb correlations in the itinerant subshells , in particular because estimates of its magnitude typically lead to values of the order or even larger than the bandwidth of the d states at the fermi level.@xcite therefore , we want to address the influence of these local coulomb correlations among the conduction band on the magnetic properties of the klm .
generally , the 5-fold or 14-fold degenerate d- respectively f - shells split in a crystalline environment into more or less well separated subshells . here , we assume that the different crystal field levels are well separated on the scale of relevant low - energy structures this is typically true in transition metal compounds , but less obvious in rare earth systems and focus on a situation , where we have one subshell well localized , hosting a spin @xmath10 , while the remaining levels are split sufficiently to leave one relevant , spin - degenerate state close to the fermi energy .
our model thus consists of a conventional one - band hubbard - model,@xcite where the itinerant states are in addition coupled to local spins @xmath10 , i.e. @xmath11 @xmath12 describes the ordinary hubbard model , @xmath13 where @xmath14 creates ( annihilates ) an electron at lattice site @xmath15 with spin @xmath16 .
while the inclusion of an arbitrary hopping @xmath17 presents no principle problem , we focus on the simplest case of next - neighbor - hopping parametrized by an amplitude @xmath18 .
finally , @xmath19 denotes the density operator for a @xmath20electron at site @xmath15 and @xmath21 parametrizes the local coulomb interaction .
the band electrons are in addition coupled to a local spin @xmath22 at each lattice site by an exchange interaction @xmath23 where @xmath24 is the spin operator for the band states at site @xmath15 .
as discussed before , such a term can arise through hund coupling , in which case @xmath0 is ferromagnetic , or through a hybridization , leading to an antiferromagnetic @xmath0.@xcite note that both effects can appear simultaneously , thus partially compensating each other.@xcite even without coulomb interaction @xmath21 this model is not exactly solvable in general , thus approximations have to be made .
for @xmath25 , a first approach can consist of a perturbative treatment of the exchange interaction @xmath0 , leading to the well - known effective rudermann - kittel - kasuya - yoshida ( rkky ) interaction@xcite with a characteristic , dimension - dependent dependence on distance .
although generally accepted as an at least proper ansatz for a qualitative discussion of magnetic properties of models like the klm , it has not yet been studied in detail how well this approximation works for increasing @xmath0 .
furthermore , for the antiferromagnetic kondo lattice model , where heavy fermion physics can play an essential role , or in the presence of additional correlations in the band states , the validity of the use of the rkky arguments is far from clear .
thus one aspect of the present paper is to investigate to what extent the rkky exchange indeed leads to a reasonable description of the low - temperature properties of the klm . as approximation to study the klm while leaving as much of the local correlations induced by both the coulomb interaction @xmath21 and the exchange @xmath0 intact as possible we use the dynamical mean field theory ( dmft),@xcite mapping the lattice model onto an effective impurity problem . in former treatments of the model ( [ model ] ) , especially within dmft,@xcite classical spins
were assumed to avoid the sign problem of the quantum monte - carlo treatment of the effective impurity problem employed there .
the importance of a fully quantum mechanical treatment of the local spins even in impurity problems has been addressed by @xcite and the effects of quantum spins in the klm by @xcite . to achieve such a fully quantum mechanical treatment we use wilsons s numerical renormalization group ( nrg)@xcite as impurity solver .
the nrg can handle the whole interaction and temperature regime and is also able to calculate green s function even in ordered phases .
we use a recent improved technique for calculating green s functions@xcite within nrg .
therefore we are able to treat a large bandwidth of values of both on - site interactions @xmath0 and @xmath21 .
the paper is organized as follows . in the next section
we discuss the form of the perturbative rkky exchange in the limit spatial dimension @xmath26 appropriate for the dmft . in section
iii we present our results for the magnetic phase diagrams of the model ( [ model ] ) at half - filling and finite doping and different values of @xmath21 and discuss their dependence on the sign and magnitude of @xmath0 .
a summary will conclude the paper .
conventionally the interaction between localized spins in metals is analyzed in terms of the rkky interaction.@xcite this effective exchange interaction shows a characteristic dependence on distance @xmath27 and local exchange coupling @xmath0 of the form @xmath28 where @xmath29 is the fermi momentum of the host and @xmath30 some positive , dimension - dependent number .
this distance - dependence is due to the sharp fermi surface in metals and , strictly speaking , valid only in the limit @xmath31 .
note that , because @xmath32 , the sign of @xmath0 does not matter .
furthermore , it is a priori not evident , how the distance - dependence looks like in the limit spatial dimension @xmath26 , which sets the framework for the construction of the dmft . in particular , in this limit there is no proper definition of @xmath29 , which controls the dependence of @xmath33 on the occupancy of the band .
thus , in order to be able to study to what extent the rkky interaction controls the magnetic properties of the klm in dmft , we will derive this interaction and especially its dependency on the filling for the limit @xmath26 in the following . to this end
we consider the extended hubbard model ( [ model ] ) on a hypercubic lattice . in the paramagnetic phase with @xmath34 ,
an effective hamiltonian for the local spins can be calculated perturbatively in @xmath0 by formally tracing out the band states .
the resulting expression in lowest order in @xmath0 reads @xmath35 s_j^\alpha[0]\rangle_{j=0 } \nonumber\\ & = & -j^2\sum\limits_{ij}\sum\limits_{\alpha=1}^3s_i^\alpha s_j^\alpha\chi_{ij}^{\alpha\alpha}\nonumber\;\;,\end{aligned}\ ] ] where @xmath36 denotes the static susceptibility of the bare hubbard model . in the paramagnetic case , @xmath36 does not depend on @xmath30 , and we are free to evaluate it for @xmath37 , for example .
this leads to an effective spin - spin interaction @xmath38 let us evaluate the latter sum for a model with nearest - neighbor hopping in the limit @xmath26 .
as has been shown by mller - hartmann,@xcite the @xmath39-dependence then enters only via @xmath40 where @xmath41 is the lattice parameter . due to inversion symmetry
, we have @xmath42 with @xmath43\;\;.\ ] ] for a nontrivial result in the limit @xmath26 we need @xmath44 to be finite .
we thus will evaluate @xmath45 directly .
following again the arguments by mller - hartmann , we first rewrite @xmath46 and obtain @xmath47 expanding the exponential in terms of @xmath48 and observing that @xmath49 and so on for terms involving @xmath50 for @xmath51 , we obtain for @xmath52 @xmath53\nonumber\\ & & \cdot \cos\left(q\cdot(r_i - r_j)\right)\nonumber\\ & = & -i\delta_{|r_i - r_j|,a } \int\limits_{-\infty}^\infty\frac{ds}{2\pi}e^{ixs } s+o\left(\frac{1}{d}\right)\nonumber\\ & = & -\delta_{|r_i - r_j|,a}\,\frac{d}{dx}\delta(x)\,+o\left(\frac{1}{d}\right)\nonumber \;\;.\end{aligned}\ ] ] with this result we find in the limit @xmath26 for the rkky exchange @xmath54 note that the right hand side of ( [ eq : rkky_final ] ) is already correctly scaled to obtain nontrivial results in the limit @xmath26 .
furthermore , the rkky exchange acts only on nearest neighbors , the oscillatory structures arising from fermi surface singularities in finite dimensions are absent in @xmath26 .
nevertheless , the dominant nearest - neighbor exchange constant which through its sign determins the type of order can be obtained . for @xmath25
, the susceptibility is given by the simple bubble , which can be evaluated exactly albeit only numerically , leading to the behavior for @xmath33 depicted in fig .
[ rkkystrength ] .
note that the sign change from @xmath55 ( afm ) to @xmath56 ( fm ) appears for @xmath57 .
we will discuss this figure in connection with numerical results in the next sections .
for @xmath25 as function of the occupancy @xmath58,scaledwidth=45.0% ] for finite @xmath59 , a similar evaluation is not possible , because the susceptibility entering now is the full lattice susceptibility . while @xmath60 is the local susceptibility , which can be obtained from the effective impurity problem directly , its derivative determining @xmath33 involves neighboring @xmath61 values , which can not be calculated within nrg .
we content ourselves here by noting that finite coulomb correlations typically lead to a more asymmetric distribution of spectral weight , which is known to tend to enhance ferromagnetic correlations .
we thus expect that the root of @xmath33 will shift to larger values of the occupancy @xmath58 for finite @xmath21 .
in this section we present our results for the magnetic phases of the extended klm ( [ model ] ) . up to now the local spin @xmath10 was completely arbitrary and can in principle take any value .
although it is surely interesting to study the effect of increasing spin quantum number on the results,@xcite we restrict the discussion to the case @xmath62 here .
the calculations were done for a bethe lattice employing a bipartite subdivision to accommodate antiferromagnetic order.@xcite we used a bethe lattice instead of a hypercubic one for computational reasons .
however , we did not find any significant differences between the two lattices in test calculations .
as discretization parameter for the nrg we used @xmath63 and typically kept @xmath64 states in each nrg step . as our unit of energy
we choose the bandwidth @xmath65 .
let us begin by examining the magnetic order at half - filling , @xmath66 . clearly fig .
[ rkkystrength ] states that the interaction between the localized spins is negative which is supposed to result in antiferromagnetic order . as function of the coupling @xmath0 between spins and band electrons .
circles were calculated for @xmath67 , corresponding approximately to @xmath25 , squares for @xmath68.,scaledwidth=45.0% ] in fig .
[ antihalf ] are collected results for the polarization @xmath69 of the band electrons and @xmath70 of the local spin .
one can see in fig .
[ antihalf ] two curves corresponding to @xmath71 ( circles ) and @xmath68 ( squares ) . for large negative @xmath0
there is no magnetic order .
the system forms a kondo - insulator , locally quenching all moments .
within dmft we find for the critical value of the coupling at @xmath25 @xmath72 which is consistent with long - known results.@xcite with increasing @xmath21 , this value is shifted to somewhat smaller absolute values @xmath73 .
for @xmath74 antiferromagnetic order can emerge before all moments have been quenched locally . quite generally , we observe that the polarization @xmath75 of the band states and @xmath70 of the local spins are opposite in sign for @xmath55 . around the critical value @xmath76 the resulting total polarization
@xmath77 thus is very small , although both contributions can already have rather sizable values .
this result is quite interesting , in particular in view of the so - called small - moment antiferromagnetism observed in several rare - earth compounds .
obviously , for @xmath25 there is no antiferromagnetic order at @xmath78 ; however , even for very small @xmath79 the local spins are almost fully polarized , @xmath80 . on the other hand ,
the polarization of the conduction electrons @xmath75 goes smoothly through zero .
we can thus identify this range of @xmath0 values as corresponding to the rkky regime , where the local spins are fully polarized and the band electrons show a polarization proportional to the `` effective field '' @xmath81 provided by the local spins . in this region
we also observe that the magnetic properties of the system are roughly independent of the sign of @xmath0 , as predicted by rkky . on the other hand ,
the behavior in the vicinity of @xmath76 can not be understood in terms of rkky any more , although @xmath82 is still significantly smaller than @xmath83 and one might expect the perturbational arguments to be still valid .
however , the physical properties are radically different for @xmath55 and @xmath56 , as no critical point exists for @xmath56 and the model always is in the fully polarized state . for @xmath59 there is antiferromagnetic order even at @xmath78 , which represents the pure hubbard model.@xcite as mentioned before , for @xmath55 spin and electron polarizations are antiparallel while for @xmath56 both orientations are parallel . due to to numerical errors we were not able to determine the behavior as @xmath84 .
even with as many as @xmath85 states kept in each nrg step dmft+nrg calculations did not converge but showed strong fluctuation for @xmath86 .
the results obtained nevertheless suggest a jump at @xmath78 . again , as @xmath87 , the net polarization @xmath77 is strongly reduced from the almost full values for each individual part . and temperature @xmath88.,scaledwidth=45.0% ]
[ spekhalf ] shows the majority spectral functions for @xmath67 and different @xmath0 .
all calculations show a gap at @xmath89 , characteristic for an insulating behavior .
the solid line represents the results for a large antiferromagnetic coupling , i.e. in the kondo insulating state .
the spectral function agrees with the typical result for paramagnetic phase of a kondo - insulator .
the remaining lines correspond to @xmath74 .
the dotted and dashed lines correspond to the same value @xmath90 , however with opposite signs .
the general shape is that of a weak - coupling antiferromagnet@xcite with the characteristic square - root van - hove singularities at the gap edges .
note that for @xmath55 we find a shallow shoulder for @xmath91 which quite likely is the remnant of a quasiparticle peak due to the kondo effect expected in this regime.@xcite this is again a sign , that already for moderate values of @xmath0 the physical properties are significantly different for @xmath55 and @xmath56 , i.e. can not be understood by pure rkky physics .
the dotted - dashed line in fig . [ spekhalf ] corresponds to a large ferromagnetic coupling . here
, life - time effects due to a large self - energy already substantially dominate the spectrum . and @xmath92 as function of filling and temperature .
the upper ( lower ) panel shows the results for ferromagnetic ( antiferromagnetic ) coupling .
lines are meant as guide to the eyes . for ferromagnetic coupling a magnetic phase exists between @xmath93 ,
whose nature could not be determined ( see text ) .
the number of states kept during nrg calculation was increased at phase boundaries.,scaledwidth=45.0% ] figure [ phase001 ] shows the magnetic phases of the kondo - lattice model at @xmath25 as function of filling @xmath58 and temperature calculated within dmft for a kondo coupling @xmath94 .
note that this value is already larger than the critical value @xmath76 at half filling , thus we expect that for @xmath55 the physics is governed by local quenching of the moments close to half filling . the upper panel in fig .
[ phase001 ] shows the model with ferromagnetic coupling , while the lower one displays the results for antiferromagnetic coupling . in the ferromagnetic kondo - lattice model we find an antiferromagnetic phase around half - filling @xmath95 , and a ferromagnetic phase for @xmath96 .
the location of phase boundary of the ferromagnet at @xmath97 appears to agree very well with rkky prediction from fig.[rkkystrength ] , where the coupling between the localized spins within dmft becomes ferromagnetic for @xmath96 .
we have focused in our work on homogeneous ferromagnetic- , antiferromagnetic nel- and paramagnetic states
. between @xmath93 calculations showed also significant magnetic response .
but neither a ferromagnetic nor an antiferromagnetic nelstate could be stabilized .
the real nature of this phase could not yet be clarified in our calculations , but one might argue that one should expect some type of incommensurate order .
in fact , the magnetic phase diagram of the kondo - lattice model has been analyzed by a number of other authors.@xcite in addition to the conventional homogeneous ferromagnet and nel antiferromagnet other magnetic phases like incommensurate , chiral and short range ordered phases were analyzed for @xmath98 in these studies and found to be stable in the regime where our calculations fail to converge .
for different temperatures . other parameters as in fig .
[ phase001].,scaledwidth=45.0% ] for the antiferromagnet close to @xmath66 , several authors@xcite also reported phase separation between antiferromagnetism at half - filling and phases away from half - filling .
figure [ antiferro001 ] shows our results for the occupancy and polarization in the antiferromagnetic phase as function of the chemical potential @xmath99 for different temperatures .
we see that with decreasing temperature the slope of the curves for @xmath100 get larger and larger . however , even for the lowest temperature we could stabilize every occupation number @xmath101 , even if there is a large slope for @xmath100 . increasing the numerical accuracy by for example keeping
more states per nrg step also tend to increase the slope further .
however , we did never observe a clear jump of the occupation at a critical value of the chemical potential .
note , however , that we can not rule out phase separation at @xmath97 from our numerical results .
the lower panel in figure [ phase001 ] shows the phase diagram of the antiferromagnetic kondo - lattice model . here
our results yield only evidence for a ferromagnetic phase .
in contrast to the ferromagnetic kondo - lattice model , this phase reaches to significantly higher occupations and temperatures ; the antiferromagnetic coupling seems to stabilize ferromagnetic order .
this behavior as well as the rather large critical value for @xmath58 at @xmath97 can not be explained within the rkky picture .
furthermore , the quite likely incommensurate phase neighboring the ferromagnet for @xmath56 is absent here .
for @xmath102 we could not stablilize a magnetic phase any more due to numerical problems connected with the small filling .
.,scaledwidth=45.0% ] the previous observations are further substantiated by the results in fig .
[ ferro001 ] , where we show the ferromagnetic polarizations for ferromagnetic ( upper panel ) and antiferromagnetic ( lower panel ) coupling parameter .
the circles correspond to a very low temperature , squares to a temperature near the transition into the paramagnetic state .
full lines represent @xmath75 , dashed lines @xmath70 .
for @xmath56 and temperatures above the critical temperature for the `` incommensurate '' phase , both @xmath75 and @xmath70 show a similar behavior as function of doping .
in particular , they vanish smoothly as @xmath103 , indicating a second order phase transition .
for very low @xmath104 , on the other hand , @xmath70 remains almost constant at the fully polarized value until we reach the phase boundary to the `` incommensurate '' phase at @xmath105 . at the same time
, @xmath75 monotonically increases with filling up to @xmath106 .
although we can not determine the true structure of the phase for @xmath107 , its existence seems to be connected to the peculiar behavior of @xmath75 rather than with @xmath70 , i.e. is primarily driven by the itinerant electrons .
we are currently further investigating this phase transition by including magnetic structures other than the nel state in our calculations . for antiferromagnetic coupling ( lower panel of fig .
[ ferro001 ] ) we find a critical @xmath106 increasing with temperature , where the ferromagnetic metal becomes a kondo insulator .
note that for @xmath108 we are effectively in the ground state .
further lowering the temperature does not change the phase diagram any more . both @xmath75 and @xmath70
appear to vanish smoothly , too , thus again indicating a second order phase transition .
choosing @xmath109 does not modify the general structure of the phase diagram for ferromagnetic coupling .
the results for antiferromagnetic coupling , of course , do change , in particular close to half filling , where now an antiferromagnetic phase appears , too . while the overall structure of the phase diagram is now similar to the case @xmath56 in fig .
[ phase001 ] upper panel , the values for @xmath110 are still enhanced with respect to the rkky prediction . and @xmath92 as function of filling and temperature .
the upper ( lower ) panel shows results for ferromagnetic ( antiferromagnetic ) coupling .
lines are meant as guides to the eye .
for ferromagnetic coupling a magnetic phase exists between @xmath112 , whose nature could not be determined ( see text ) .
the number of states kept during nrg calculation was increased at phase boundaries.,scaledwidth=45.0% ] let us now turn to the model with finite on - site coulomb interaction .
we present results for fixed @xmath113 , changing these parameters modifies the observations quantitatively but not qualitatively .
the phase diagrams for ferromagnetic ( upper panel ) and antiferomagnetic ( lower panel ) coupling are shown in fig .
[ phase0101 ] . as for @xmath25 , we find both an antiferromagnetic phase around half filling and a ferromagnetic phase for larger doping in the ferromagnetic kondo - lattice model .
the critical values have changed to @xmath114 and @xmath115 for @xmath97 , indicating that local correlations additionally stabilize ferromagnetic order but destabilize antiferromagnetic for @xmath56 , in accordance with the anticipated variation of @xmath33 with @xmath21 . for the ferromagnetic kondo lattice model we again find between the ferromagnetic phase for @xmath116 and the antiferromagnetic for @xmath117 a magnetic phase whose nature could not be determined for the same reasons as for @xmath25 .
guided by our observations for @xmath25 and a comparison to other studies by other groups , one can again assume the appearance of an incommensurate magnetic phase like a spin wave or a chiral phase .
from quantum monte - carlo studies @xcite for the conventional hubbard model an incommensurate phase can indeed be anticipated in this parameter regime . for different temperatures .
chemical potential @xmath99 .
other parameters as in fig .
[ phase0101].,scaledwidth=45.0% ] figure [ antisepa ] shows occupation and polarization as function of the chemical potential @xmath99 for different temperatures . for high - temperatures , squares and triangles , there is a smooth crossover from half - filling to lower occupancies , as for @xmath25 ( cf .
[ antiferro001 ] ) . however , in contrast to the indecisive results at @xmath25 , we observe clear evidence for phase separation at @xmath118 , circles , this time .
there exists a critical value @xmath119 , where both occupation and polarization jump from their values at half - filling to a smaller occupancy and polarization .
we therefore have phase separation between the antiferromagnetic insulator at half - filling and an antiferromagnetic metal with a lower occupancy .
note that one also finds phase separation in the conventional hubbard model , which here however occurs between an antiferromagnetic insulator with @xmath66 and a paramagnet with @xmath120.@xcite .,scaledwidth=45.0% ] the dependence of the polarization on @xmath58 and @xmath104 in the ferromagnetic phase for the present values of @xmath21 and @xmath73 is shown in fig .
[ ferro01 ] for @xmath56 ( upper panel ) and @xmath55 ( lower panel ) . as in fig .
[ ferro001 ] , full lines represent @xmath75 and dashed lines @xmath70 . for high temperatures ( squares )
we again find second order phase transitions , indicated by the smooth simultaneous vanishing of both @xmath75 and @xmath70 , while for low @xmath104 ( circles ) and @xmath56 the peculiar behavior at the phase boundary to the `` incommensurate '' phase already seen in fig .
[ ferro001 ] is observed .
a markable difference to the behavior for @xmath25 collected in fig . [ ferro001 ] can be observed for antiferromagnetic coupling @xmath55 .
here , the polarization now vanishes discontinously at @xmath110 for @xmath121 , indicating a first order phase transition in contrast to the second order type observed for @xmath25 .
thus , while the results at @xmath25 would predict a quantum critical point at @xmath110 , such a scenario is obvioulsy destroyed by coulomb correlations in the conduction band , at least within dmft .
note that for the bare hubbard model no ferromagnetic phase exists in this parameter regime for a bipartite lattice .
it is thus clearly the spin - electron interaction which leads to ferromagnetism .
our results are consistent with earlier studies by other groups , who however treated the electron - electron interaction in an effective medium approach,@xcite while we are able to treat the on - site correlations exactly within dmft and nrg .
especially our curie temperatures are located in the same range as in these earlier studies .
a possible reason for the tendency of both finite @xmath21 and negative @xmath0 to stabilize ferromagnetism is the strong asymmetry induced by @xmath21 and @xmath0@xcite . as can be seen from fig .
[ dos_n06 ] , this effect is most pronounced , @xmath122 , @xmath25 , @xmath123 and @xmath124.,scaledwidth=45.0% ] for @xmath55 .
for @xmath56 , a weaker , but nonetheless significant redistribution of spectral weight is observed , which can explain the increase of @xmath110 here , too .
in this paper we discussed the magnetic properties of the extended kondo lattice model , where in addition to an exchange coupling between itinerant states and localized spins the band electrons are subject to a local coulomb interaction .
such a model is expected to describe the properties of e.g. transition metal compounds , where part of the @xmath125-states are localized in the crystalline environment due coulomb correlations and form a local spin which couples to the remaining , usually itinerant @xmath125-states either due to hund s exchange or hybridization .
these materials typically have a rich magnetic phase diagram , and it is an interesting question , how these two interactions cooperate or compete in driving different magnetic phases .
a particularly interesting question in this respect is , to what extent , qualitatively and quantitatively , the concept of the rkky approximation of an effective interaction between the local spins mediated by the itinerant conduction states holds .
we treat the model within dmft , using nrg as solver for the effective impurity problem .
the calculations were done for a bipartite lattice , allowing for homogeneous ferromagnetism and nel antiferromagnetism . at half filling , we find an antiferromagnetic phase , as expected , for all ferromagnetic kondo couplings @xmath56 . for antiferromagnetic coupling @xmath55 , there exists a critical @xmath76 , where the system becomes a paramagnetic kondo insulator.@xcite for @xmath74 , this antiferromagnetic phase prevails for finite doping up to a critical filling @xmath126 depending on @xmath0 and the local coulomb interaction @xmath21 . for finite @xmath21 and @xmath56 ,
we in addition find phase separation between the antiferromagnetic insulator with @xmath66 and an antiferromagnetic metal with @xmath127 . for @xmath25
we find no convincing evidence that such a phase separation exists , too , but the numerical results are not sufficiently clear to rule it out either . for @xmath128
no antiferromagnetic phase exists at all . below a second critical filling @xmath110 a homogeneous ferromagnet
quite interestingly , the extent and stability of this phase is increased by both a finite @xmath21 and an antiferromagnetic @xmath0 .
the reason for this stabilization can quite likely be traced back to the introduction of a strong asymmetric redistribution of spectral weight in the electronic spectral function .
an interesting aspect in connection with quantum critical points in e.g. ferromagnetic heavy fermion compounds is that for antiferromagnetic kondo exchange a finite coulomb correlation among the conduction states has the tendency to change the order of the phase transition at @xmath110 from second order without correlation to first order with correlations .
we find that for ferromagnetic kondo coupling @xmath0 the predictions by rkky can be used to at least qualitatively account for the different phases . for moderate and large @xmath55
, however , rkky can not even qualitatively predict the phases correctly , underestimating the ferromagnetic phase grossly .
moreover , the variations of the polarizations of the local spin and bandelectrons seem to follow rkky for very small @xmath73 , but already deviate substantially for moderate values .
we want to thank helpful discussions with andreas honecker , akihisa koga and dieter vollhardt .
this work was supported by the dfg through pr298/10 .
computer support was provided by the gesellschaft fr wissenschaftliche datenverarbeitung in gttingen and the norddeutsche verbund fr hoch- und hchstleistungsrechnen . | we study magnetic ordering of an extended kondo - lattice model including an additional on - site coulomb interaction between the itinerant states .
the model is solved in the dynamical mean - field theory using wilson s numerical renormalization group approach as impurity solver . for a bipartite lattice we find at half filling the expected antiferromagnetic phase . upon doping this phase is gradually suppressed and hints towards phase separation are observed . for large doping the model exhibits ferromagnetism , the appearance of which can at first sight be explained by rudermann - kittel - kasuya - yosida interaction .
however , for large values of the kondo coupling @xmath0 significant differences to a simple rudermann - kittel - kasuya - yosida picture can be found .
we furthermore observe signs of quantum critical points for antiferromagnetic kondo coupling between the local spins and band states . |
observations are accumulating about the early universe and the interstellar medium that challenge the standard models of cosmology , galaxy formation , star formation , planet formation and comet formation .
for many years no data were available to contradict the jeans 1902 gravitational structure formation criterion that in a gas of uniform density @xmath8 , gravitational condensation can not occur on scales smaller than @xmath9 where @xmath10 @xmath11 is newton s gravitational constant , @xmath12 is the gas constant , @xmath13 is boltzmann s constant , @xmath14 is the gas particle mass , and @xmath15 is the temperature . because @xmath16 is @xmath17 for an ideal gas , jeans 1902 and
others have assumed that the jeans acoustic scale @xmath9 defined on the left and the jeans hydrostatic scale @xmath18 defined on the right of ( [ eq1 ] ) are equivalent .
they are not .
it has been shown that the jeans criterion is unreliable @xcite : non - acoustic density fluctuations are absolutely unstable to structure formation .
hydrostatic equilibrium is achieved only at small scales @xmath18 determined by the largest viscous , turbulent , or diffusive schwarz scale ( table 1 ) of hgt and various planetary and star formation processes that are induced at positive non - acoustic nuclei , with @xmath19 . using viscous forces , turbulent forces and diffusion it was shown from hgt - cosmology that the baryonic dark matter is dominated by small frozen planets ( @xmath20 primordial fog particles , or pfps ) that fragmented by self gravity at the plasma to gas transition 300,000 years after the big bang , @xmath21 .
this theoretical prediction was immediately and independently supported by observations of twinkling frequencies of a lensed quasar , suggesting the lensing galaxy mass is dominated by rogue planets `` likely to be the missing mass '' @xcite . the gibson @xmath22 schild 1996 prediction and observational demonstration that the baryonic dark matter is primordial micro - brown - dwarfs has since been confirmed by independent observations in the same and other lensed quasars , and by other observational evidence , @xmath23 .
hgt describes the formation of structure by self - gravitational condensations and rarefactions for cosmology , astrophysics and astronomy .
the first gravitational structures to form were proto - supercluster - voids in the plasma epoch by fragmentation at 30,000 years after the big bang when viscous and weak turbulent forces first matched gravity forces at the horizon scale @xmath24 , followed by proto - cluster - voids and proto - galaxy - voids filled with the weakly collisional non - baryonic - dark - matter ( nbdm ) by diffusion @xcite .
cold dark matter condensations ( cdm - halos ) are excluded by hgt for scales smaller than @xmath25 in the plasma epoch , and are not necessary to make structure in the gas epoch ( after 300,000 years ) by hierarchical - clustering - cosmology ( hcc ) , which is also excluded by hgt .
gravitational clustering of collisionless cdm - halos would produce steep core cusps in galaxies that are not observed @xcite .
the most distant galaxies observed are clustered @xcite and the closest clusters observed are fragmenting @xcite , contrary to hcc .
the early clusters predicted by hgt serve as gravitational lenses and produce arcs and multiple arcs of background galaxies at high rates of incidence that contradict all the flat - cdm cosmologies by orders of magnitude @xcite .
it is necessary to invent special `` super - lensing '' clusters to explain the multiple arcs observed in the recent red - sequence cluster survey and other studies , which would otherwise exclude the presently popular flat-@xmath26cdm cosmology by orders of magnitude @xcite .
drag forces from the expansion of the universe rapidly separated proto - galaxies and helped fragment proto - galaxy - clusters at the plasma - gas transition @xcite , making galaxy mergers quite rare rather than mandatory as in hcc , and explaining alignments of bright central supercluster galaxies with background qsos by perspective that otherwise would confute cdmhcc , the big bang , and standard physics @xcite without hgt @xcite .
two of these rare galaxy mergers were imaged by the new hst / acs wide angle camera in the tadpole complex vv29abcdef @xcite and in mice , showing star trails of young - globular - clusters ( ygcs ) produced as star formation is triggered from pgc - pfps by the mergers of galaxies and their baryonic - dark - matter ( bdm ) halos .
high resolution wide angle images by hst / acs in the helix planetary nebula and other nearby pne in the galaxy support the hgt prediction that galaxy disks are formed by accreted pgcs , and that the ism has the high primordial pgc - pfp density fossilized at the time of first structure and expected in star forming regions @xcite . from hgt , density perturbations @xmath27 on scales
@xmath28 in a uniform density gas are absolutely unstable to the formation of gravitational structure , and grow or decrease exponentially with time depending on the sign of the perturbation ( see reyden 2003 @xmath29 but ignore @xmath30 ) .
let @xmath31 inside a sphere of radius @xmath32 and constant @xmath8 outside , where @xmath33 and @xmath34 with @xmath35 the particle collision length .
as correctly shown in @xmath29 , self gravitation causes the density perturbation to either increase in magnitude and shrink in size ( gravitational condensation , @xmath36 ) $ ] , with @xmath37 ) , or increase in magnitude and grow in size ( gravitational void formation , @xmath38 ) $ ] , with @xmath39 ) .
non - acoustic density fluctuations are absolutely unstable to self - gravitational formation of structure , just as shear layers are absolutely unstable to the formation of turbulence due to inertial - vortex forces @xmath40 , where @xmath41 is velocity and @xmath42 is vorticity @xcite .
both processes are highly non - linear so neither can be reliably described by linear perturbation stability analysis , and both couple very large scales to very small scales so they are difficult to simulate numerically .
it is stated without proof in @xmath30 @xcite that pressure will prevent the exponential growth of density perturbations on length scales @xmath43 .
this is not true .
the idea that `` pressure support '' or `` thermal support '' prevents structure formation on scales smaller than @xmath9 is an unfortunate legacy of the erroneous jeans 1902 criterion for structure formation .
the time @xmath44 required for pressure waves ( sound ) to move a distance @xmath45 is @xmath46 , where @xmath17 is the speed of sound . for @xmath47 , @xmath48 , so @xmath49 for @xmath50 and @xmath51 .
density perturbations @xmath52 of size @xmath53 grow to large values for @xmath54 as we have seen .
pressure forces have the wrong sign to resist gravitational condensation except near the gravitational stagnation point because they arise from bernoulli s equation , where @xmath55 .
pressure forces overcome condensational self gravitational forces @xmath56 on scales @xmath57 only at late stages in a collapse ( @xmath58 ) due to local irreversible processes ( eg : turbulence stresses , viscous heating , condensation of the gas to liquid form , freezing of the liquid , star formation ) , and have no effect on small scale self gravitational void formation forces ( @xmath59 ) until the void size reaches @xmath9 .
the maximum speed of void growth is the maximum speed of a rarefaction wave , which is the speed of sound .
self - gravity has the effect of a negative density diffusivity @xmath60 in the vicinity @xmath61 ( see table 1 and @xmath62 ) of a non - acoustic density maximum or density minimum , with the resulting absolute instability .
high resolution numerical simulations of stagnant gas show fragmentation and condensation at scales smaller than the jeans scale , but in calculations to date these have been systematically filtered out as numerical artifacts @xcite .
the authors assume that pressure support or thermal support prevent sub - jeans - scale self - gravitational condensations and fragmentations in astrophysical fluids and adjust their numerical simulations to conform to these assumptions .
the large gassy jovian planets beyond mars in the solar system are a mystery from the jeans criterion because they violate the jeans limit @xmath63 @xcite . if we compute the jeans mass @xmath64 and solve for the jeans temperature @xmath65 we find @xmath66 where @xmath67 m@xmath6 s@xmath68 k@xmath2 is the gas constant and @xmath69 is the gas molecular weight . substituting a primordial h -
he gas constant @xmath70 m@xmath71 s@xmath68k@xmath2 and the primordial ( first structure ) gas density @xmath72 kg m@xmath4 for jupiter mass @xmath73 kg with @xmath74 m@xmath75 kg@xmath2 s@xmath68 gives a jeans temperature @xmath76 k that is impossibly cold .
why are jupiter , saturn , uranus and neptune all contrary to the jeans limit ? numerous extra - solar planets with jupiter mass
have now been detected that must also be composed of h - he gas .
how are these possible ?
how are rocky planets possible ?
how are rocks possible ?
all are impossible by the jeans criterion , and all are easily explained by hgt , @xmath62 . according to hgt , all large planets are formed by a hierarchical clustering of pfps to form jovian - pfp - planets ( jpps ) and all stars are formed by further accretion of pfp - jpps and from their fragments and gas formed by the close proximity to the proto - star .
dust from supernovas is collected by numerous pfp - jpps in the ism and turned into rocks near their cores and the cores of the numerous large planets and brown dwarfs ( jpps ) formed as the pfps accrete .
multiple hydrogen freeze - thaw cycles of wandering pfp - jpps produces many crushing central pressure episodes and layerings .
rocky planets like earth are thus rocky cores of jupiters evaporated as the star grows by jpp accretion .
long period comets with random oort cloud orbits @xmath77 m matching a typical pne radius suggest that the trillions of oort comets are from a few hundred pfps constantly drizzling in from the sun s accretion hole in the ism ( its oort cloud inner radius according to hgt ) that were not captured by the sun and its jovians in their first accretional attempt and have not yet been completely fragmented by tidal forces and evaporated by radiation @xcite .
orbit perturbations of 82 new ( first - entry ) class i ( accurate ) oort comets imply the existence of a sirtf - infrared and vla - radio detectable @xmath78 kg jpp at @xmath79 m @xcite . from hgt
the inner radius of the oort cloud is @xmath80 m , corresponding to a solar mass accreted from a pgc , and the outer radius is @xmath81 m , corresponding to the average extent of a pgc . without hgt
the origin of the huge numbers of fragile , frozen , oort comets with their eccentric orbits , specific distance scale , and random directions is mysterious .
what about the inter - stellar - medium ( ism )
? if there are really 30,000,000 pfps per star in the average galaxy , should nt some of these be revealed by their evaporated gas near very hot objects ; for example , near carbon white dwarfs formed when ordinary stars have burned most of their h - he gas to form planetary nebulae , or in the ism after novas and supernovas ?
evidence that mysterious `` free - floating planets pepper our galaxy '' ( june 2003 astronomy cover page ) is rapidly accumulating @xcite .
the number of rogue jupiters greatly exceeds predictions of `` failed brown dwarf '' and star expulsion models .
why do mysterious knots , comets and herbig - haro objects appear from the dark ism whenever young stars form plasma jets ? where does the huge amount of gas in the halos of planetary nebulae come from ?
are ordinary stars really so inefficient that most of their mass is wasted forming pne halos ?
see @xmath23 for a discussion of hst images of the helix planetary nebula @xcite , which is the closest pne to earth , and its thousands of cometary globule pfp and jpp candidates .
hgt contradicts cold - dark - matter hierarchical - clustering - cosmologies ( cdmhcc ) .
cdm `` halos '' can not remain gravitationally bound because the weakly collisional cdm particles are intrinsically much too diffusive , with @xmath82 during the plasma epoch .
hence , small primordial cdm `` halos '' ( the key misconception ) can not cluster to provide massive gravitational potential wells to collect the baryonic matter and hierarchically form galaxies , clusters and superclusters with ever increasing mass as proposed by cdmhcc .
instead , hgt forms protosuperclusters first in the hot plasma epoch and these gently fragment to proto - clusters and then proto - galaxies as the plasma cools , starting at 30,000 years and finishing at the plasma to gas transition at 300,000 years .
the gentle fragmentation from viscous - gravitational beginnings in a uniformly expanding universe explains the remarkable angular and geometric correlations observed between galaxies and quasi - stellar - objects ( qsos ) with highly discordant red shifts that are used to contradict the big bang hypothesis @xcite .
compact galaxy clusters like stephan s quintet ( sq ) with highly discordant redshifts support the hypothesis that redshifts may be intrinsic @xcite and that all the sq galaxies are at the same distance and were ejected by galaxy ngc 7331 located at the same redshift @xmath83 as the nearest member ngc 7320 of the sq cluster .
no physical explanation has been put forth for intrinsic redshifts and qso ejections . the observations supporting these hypotheses can be easily explained by hgt without inventing new physics by recognizing that the galaxies and qsos are aligned by perspective , and stay aligned along narrow lines of sight as the universe expands because their sticky beginnings allow only small transverse velocities .
cdmhccs are contradicted by the accumulation of evidence showing strong angular coincidences of qsos and bright galaxies @xcite .
stephan s quintet is presented as an example in @xmath23 .
standard cdm cosmologies are flawed by their adoption of simplistic fluid mechanical equations that fail to incorporate significant forces , and by their inappropriate use of collisionless fluid mechanics .
the ill - posed jeans 1902 theory ( neglecting non - acoustic density perturbations ) is far too limited a basis for discussion of structure formation because it neglects viscous forces , turbulence forces , non - acoustic density fluctuations , particle collisions , and the effects of diffusion on gravitational structure formation .
jeans did linear perturbation stability analysis ( neglecting turbulence ) of euler s equations ( neglecting viscous forces ) to reduce the problem of gravitational instability of a nearly uniform ideal gas with density @xmath8 only a function of pressure ( the barotropic assumption ) to one of gravitational acoustics .
furthermore , to reconcile his equations with the linearized collisionless boltzmann s equations and the resulting poisson s equation for the gravitational potential @xcite , jeans assumed the density was zero in a maneuver appropriately known as the `` jeans swindle '' .
the critical wave length for stability with all these questionable assumptions is the jeans length scale @xmath9 given in eq . 1 ( see table 1 ) .
density fluctuations in natural fluids are generally not barotropic as assumed by jeans 1902 except in small regions and short times near powerful sound sources , but are dominated by non - acoustic ( non - barotropic ) density variations from turbulent mixing of temperature or chemical species concentrations @xcite . even in the context of jeans theory ( without the jeans swindle ) , any gravitational condensation on an acoustical density maximum rapidly converts to a non - acoustical density maximum because the accreted mass retains the momentum of the presumed motionless ambient gas . at this point one
faces the more realistic problem of gravitational structure formation in a moving , viscous , diffusive , possibly turbulent gas that was addressed by gibson 1996 .
turbulence or viscous forces will dominate gravitational forces at small distances from a non - acoustic point of maximum or minimum density , but gravitational forces will dominate turbulent or viscous forces at larger distances if the gas does not diffuse away faster than it can condense or rarify due to gravity .
the jeans 1902 analysis fails because the problem was ill posed .
self - gravitational structure formation in nature occurs on non - acoustic density maxima and minima , not sound wave crests , but jeans started by assuming a uniform density , and was forced to set the density to zero in the jeans swindle to avoid the nonlinear problems of non - acoustic density maxima forming at acoustic wavecrests .
the geometry of evolving non - acoustic density fields at scales @xmath84 is shown in figure 1 , with ( top ) and without ( bottom ) self - gravitational instability . without self - gravity a quasi - equilibrium at the batchelor length scale ( table 1 ) exists , where the competition between diffusive smoothing of density gradients and strain rate steepening produces a universal turbulent mixing scale @xmath85 at points of maximum and minimum density @xcite that begin to decay by diffusion as soon as turbulence scrambles a uniform density gradient to form such extrema with zero density gradients , as shown by experiments and numerical simulations @xcite . without gravity , points of maximum density always decrease by diffusion and points of minimum density increase , always moving and diffusing back toward the original configuration with uniform gradient and no density extrema .
evidence of primordial turbulence in cmb anisotropies @xcite suggests non - acoustic density maxima necessary to seed primordial gravitational structures were available in the plasma epoch . assuming the same initial density conditions as fig .
1 ( top ) , the evolution of density with self gravity ( bottom ) is very different .
the reason is that gravity has the effect of causing a negative diffusivity @xmath86 for distances @xmath87 close to the points of zero density gradient and maximum or minimum density value .
negative density diffusivity causes absolute gravitational instability at non - acoustic density extrema . with self gravity ( bottom )
density maxima increase their density values toward infinity and density minima decrease their density values toward zero . both experience exponential growth to form condensates and voids for all length scales @xmath88 unless limited by turbulent , viscous or diffusive effects at the appropriate schwarz length scales .
consider a well posed problem in self - gravitation .
a mass perturbation @xmath89 with scale @xmath90 suddenly placed in a motionless fluid of constant density @xmath8 at time @xmath91 will increase as @xmath92 $ ] unless viscous , turbulent , other forces or diffusion arise to prevent the exponential growth @xcite .
everything slowly begins to move toward or away from the mass perturbation for @xmath93 and everything happens at once at @xmath94 .
for @xmath95 the radial velocity @xmath96 from newton s law of gravity , so the mass flow @xmath97 and density @xmath8 are independent of radius .
thus , @xmath98 , so @xmath99 . since everything happens within @xmath45 of the perturbation at @xmath100 , any `` pressure support '' mechanism is confined to this small space - time region and has no affect on the weakly accelerated fluid that slowly gains momentum at scales @xmath101 and @xmath102 for times @xmath103 .
since it takes a time @xmath100 for a sound wave to reach a distance @xmath104 in time @xmath105 , any hypothetical acoustic `` pressure support '' mechanism of the jeans criterion would be too late to prevent the exponential density singularity at @xmath100 .
note that the time required for self - gravitational structure formation depends almost entirely on the mean density @xmath8 and only very slightly on the size of the density perturbation @xmath106 .
the diffusion velocity is @xmath107 for diffusivity @xmath108 at distance @xmath45 and the gravitational velocity is @xmath109 .
the two velocities are equal at the diffusive schwarz length scale @xmath110^{1/4}.\ ] ] weakly collisional particles such as the hypothetical cold - dark - matter ( cdm ) material can not possibly form potential wells for baryonic matter collection because such particles have large diffusivity and will disperse , consistent with observations @xcite .
diffusivity @xmath111 , where @xmath112 is the particle speed and @xmath113 is the collision distance for particles with number density @xmath114 , @xmath115 is the collision cross section , and @xmath14 is the particle mass .
cdm particles have @xmath116 m@xmath71 or less compared to baryonic @xmath117 m@xmath71 values .
weakly collisional particles thus have large collision distances @xmath35 and therefore large diffusivities @xmath108 and large diffusive schwarz lengths @xmath118 .
therefore , the non - baryonic dark matter ( possibly neutrinos ) must be the last material to fragment by self gravity and not the first as assumed by cdm cosmologies . in hgt galaxy formation @xcite , the nbdm diffuses to form either an outer halo of the galaxy
if the galaxy is isolated , or diffuses completely away from the baryonic galaxy to become part of a galaxy cluster halo or galaxy supercluster halo at mpc scales @xmath119 m or more .
the baryonic matter is subject to large viscous forces , especially in the hot primordial plasma from photon viscosity and gas states from large @xmath112 values existing when most gravitational structures first formed , where @xmath120 if @xmath121 .
the viscous forces per unit volume @xmath122 dominate gravitational forces @xmath123 at small scales , where @xmath124 is the kinematic viscosity and @xmath125 is the rate of strain of the fluid .
the forces match at the viscous schwarz length @xmath126 which is the smallest size for self gravitational condensation or void formation in such a flow .
turbulent forces may require even larger scales of gravitational structures .
turbulent forces @xmath127 match gravitational forces at the turbulent schwarz scale @xmath128 where @xmath129 is the viscous dissipation rate of the turbulence .
thus in strongly turbulent regions with large @xmath129 values one expects large stars to form because the turbulent schwarz scales are larger than any other hydro - gravitational scale . in the cold , turbulent , dense molecular clouds of the milky way galaxy disk where the jeans mass is near @xmath130 or less , few stars form
compared to the rate expected by the jeans criterion because @xmath131 .
the jeans criterion is nearly irrelevant to gravitational structure formation triggered by non - acoustic density maxima and minima where structures can form at scales that are either larger or smaller than @xmath9 . by jeans criterion no structures can form in the primordial plasma because @xmath9 during this hot , dense
, epoch is always larger than the scale of causal connection @xmath132 , where @xmath133 is the speed of light and @xmath134 is the time since the big bang .
however , the viscous and turbulent schwarz scales @xmath135 became smaller than @xmath9 during this period , and first matched the more rapidly increasing @xmath24 at time @xmath136 seconds @xcite , well before @xmath137 seconds which is the time of plasma to gas transition ( 300,000 years ) . because the expansion of the universe inhibited condensation but enhanced void formation in the weakly turbulent plasma , the first structures were proto - supercluster - voids . at @xmath138 s the reynolds number was near critical so that @xmath139 where @xmath118 applies to the non - baryonic component and @xmath140 , @xmath141 , and @xmath142 apply to the baryonic component .
as proto - supercluster fragments formed the voids between filled with non - baryonic matter by diffusion , inhibiting further structure formation by decreasing the gravitational driving force . the baryonic mass density @xmath143 kg/@xmath144 and rate of strain @xmath145 @xmath146 were preserved as hydrodynamic fossils within the proto - supercluster fragments and within proto - cluster and proto - galaxy objects resulting from subsequent fragmentation as the photon viscosity and @xmath140 decreased prior to the plasma - gas transition and photon decoupling @xcite .
as shown in eq .
5 , the kolmogorov scale @xmath147^{1/4}$ ] at the time of first structure nearly matched the horizon scale @xmath132 and the viscous and turbulent schwarz scales , freezing in the density , strain - rate , and spin magnitudes and directions of the proto - cluster and proto - galaxy fragments of proto - superclusters .
remnants of the strain - rate and spin magnitudes and directions of the weak turbulence at the time of first structure formation are forms of fossil vorticity turbulence @xcite .
thus hgt naturally explains observed bright galaxy spin alignments and close angular associations with each other and quasars without assuming intrinsic red shifts and mutual ejections .
the quiet condition of the primordial gas is revealed by measurements of temperature fluctuations of the cosmic microwave background radiation , which show an average @xmath148 , much too small for any turbulence to have existed at that time of plasma - gas transition .
this is to be expected because any turbulent plasma motions would have been strongly damped by buoyancy forces after the first gravitational fragmentation time @xmath138 s. viscous forces in the plasma are inadequate to explain the lack of primordial turbulence ( kinematic viscosity @xmath124
@xmath149 m@xmath6 s@xmath2 is required versus only @xmath150 m@xmath6 s@xmath2 from gibson 2000 ) .
the gas temperature , density , viscosity , and rate of strain are all precisely known at transition , so the gas viscous schwarz mass @xmath151 is @xmath152 kg , the mass of a small planet , or @xmath153 . from hgt , soon after the cooling primordial plasma turned to gas at @xmath137 s ( 300,000 yr ) , the entire baryonic universe condensed to a fog of planetary - mass primordial - fog - particles ( pfps ) .
these gas - cloud objects gradually cooled , formed h - he rain , and eventually froze solid to become the baryonic dark matter and the basic material of construction for stars and everything else , @xmath154 rogue planets per star .
the pfp mass @xmath155 is above the evaporation stability limit @xmath156 @xcite .
the jeans mass @xmath157 of the primordial gas at transition was @xmath158 , the mass of a globular star cluster .
proto - galaxies fragmented at the pfp scale but also at this proto - globular - star - cluster pgc scale @xmath9 , although not for the reason given by the jeans 1902 theory . density fluctuations in the gaseous proto - galaxies were absolutely unstable to void formation at all scales larger than the viscous schwarz scale @xmath140 .
pressure can only remain in equilibrium with density without temperature changes in a gravitationally expanding void on scales smaller than the jeans scale . from the second law of thermodynamics ,
rarefaction wave speeds that develop as density minima expand due to gravity to form voids are limited to speeds less than the sonic velocity .
cooling would therefore occur and be compensated by radiation in the otherwise isothermal primordial gas when the expanding voids approached the jeans scale .
gravitational fragmentation of proto - galaxies will then be accelerated by radiative heat transfer to these cooler regions , resulting in fragmentation at the jeans scale and isolation of proto - globular - star - clusters ( pgcs ) with the primordial - gas - jeans - mass .
these @xmath159 kg pgc objects were not able to collapse from their own self gravity because of their internal fragmentation at the viscous schwarz scale to form @xmath160 kg pfps .
the fact that globular star clusters have precisely the same density and primordial - gas - jeans - mass from galaxy to galaxy proves they were all formed simultaneously soon after the time of the plasma to gas transition @xmath137 s. the gas has never been so uniform since , and no mechanism exists to recover such a high density , let alone such a high uniform density , as the fossil turbulent density value @xmath161 kg/@xmath144 .
young globular cluster formation in bdm halos in the tadpole , mice , and antennae galaxy mergers show that dark pgc clusters of pfps are remarkably stable structures , persisting without disruption or star formation for more than ten billion years .
so the proto - galaxies ( pgs ) before stars were rapidly fragmented at both the jeans and viscous schwarz scales to form pgcs , a million per galaxy , with embedded pfps , a trillion per pgc and @xmath162 per pg .
viscous forces inhibited the formation of larger planetoids by accretion of the pfps except at the cores of the pgcs and the cores of the pgs .
it was a very quiet and gentle time .
the gentleness of this time of first star formation is reflected in the small , uniformly distributed , long - lived , population ii stars of the ancient globular - star - clusters observed in the milky way galaxy .
the time to form the first of these stars was probably one million years ( the primordial density free fall time ) , not the 275 million `` dark age '' years required by the standard cdm model while the cdm halos formed by frictionless clustering of cdm seeds .
open star clusters formed from captured and disrupted pgcs in the galaxy disk are exposed to higher levels of radiation and tidal friction than pgcs in the bdm halo , and form large irregularly distributed stars .
the large sizes of the disk stars and their slow rate of formation reflect the high turbulence levels of disk star forming regions ( from supernovas ) compared to the low turbulence levels existing where stars are formed in isolated pgcs . from hgt , the thin disks of spiral galaxies are accretion disks forming within large , massive , spherical bdm halos surrounding the original smaller denser @xmath163 m galaxy cores from which the pgcs diffused , rather than collapsed gaseous pancakes in cdm halos as envisaged in cdm standard models .
relatively low gas levels are observed in spiral galaxy cores according to hgt , where excess gas is frozen out on the ambient pfps .
with the expansion of the universe and further cooling , the pgcs and pfps cooled as well , reaching the 14 k freezing point of hydrogen at about a redshift of 30 at a time about 80 my . because most of the gas is isolated in individual pfps separated from their neighbors by distances of about @xmath164 m corresponding to the primordial density , and because most pfps within a pgc are initially surrounded homogeneously by identical objects with the same spacing , the tendency without external forcing would be for them to slow down due to friction from the inter - pfp gas and remain isolated indefinitely .
after freezing , the pfp collision cross section will decrease but the inter - pfp gas density and friction will also decrease , permitting speedup of the pfps and a new meta - stable state of equilibrium .
the gunn - peterson trough phenomenon shows intergalactic gas effectively vanishes for red shifts smaller than 6 ( @xmath165 700 my ) in quasar spectra with their lyman-@xmath166 forests of intervening galaxies .
this suggests that most baryonic - dark - matter galaxy halo gas has frozen out on pfps embedded in the bdm halos at times less than a billion years after the big bang .
as the inter - pgc gas density decreases with time since pfp creation , these pgc clumps of condensing baryonic dark matter become increasingly collisionless and diffusional .
those formed near the core of the proto - galaxy have a higher likelihood of mutual capture , interaction , disruption , and conversion to stars , but those formed on the periphery of the proto - galaxy will tend to either capture each other to form proto - dwarf - galaxies or diffuse to larger and larger orbits . the inner pgcs should be the most agitated by tidal forces and therefore the most likely to form stars and eventually the full population of @xmath167 stars observed in a globular star cluster .
those further out might form only a few thousand stars and drift to the outskirts of the luminous portion of the galaxy , sometimes being captured or disrupted by the disk , but the largest number diffusing farther out to form the baryonic - dark - matter ( bdm ) halo of the galaxy , invisible unless disturbed by intruders as in the case of the tadpole merging galaxy system ( vv29=arp 188=ugc 10214 ) .
figure 3 shows the hst / acs image with its combination of wide field and high resolution , which clearly reveals the existence of star wakes , a dark dwarf galaxy and young - globular - clusters triggered into existence from the dark baryonic - dark - matter halo by the merging cluster of galaxy components vv29cdef as they spiral in to the central galaxy vv29a and merge @xcite .
sparse globular clusters such as palomar 5 with luminous mass @xmath168 and palomar 13 with luminous mass @xmath169 observed in the milky way are likely dim rather than disrupted pgcs with 98% to 99.9% of their mass intact as dark - matter pfps .
nearly half of the stars of pal 5 are observed as long tidal tails along its orbit about the galaxy center @xcite , suggesting that the same tidal forces that produced the tails may also be producing the stars from a large supply of remaining pfps .
the amount of dark matter in dim globular clusters like pal 5 and pal 13 in the milky way , and in galactic dwarf spherical galaxies ( dsphs ) , is likely underestimated in virial estimates @xmath170 of the dark mass @xmath171 because friction from pfp gas is neglected that reduces the velocity variance @xmath172 of the stars within radius @xmath173 . a mass - to - light ratio @xmath174 for palomar 13
was recently reported @xcite , which is the first estimate from @xmath172 measurements in dim globular star clusters and is much higher than @xmath175 expected . in another estimate ( cote et al .
14b ) @xmath176 fitting to the luminosity relation for dsphs which is closer to the value @xmath177 expected from hgt .
large @xmath178 values measured in dsph galaxies have been assumed to reflect non - baryonic cdm halos @xcite , but from tadpole we see pgc and dsph dark matter is mostly baryonic because it forms stars when agitated .
the hydro - gravitation - theory ( hgt ) scenario presented here is significantly different from standard cdmhc cosmology based on jeans 1902 precepts .
gravitational structures in the baryonic matter fragmented during the plasma epoch with supercluster to galaxy masses .
the protosuperclusters have the observed uniform supercluster mass of about @xmath179 kg ( including their non - baryonic halos ) , with a uniformity not expected for hierarchical clustering of cdm halos assumed in the standard cdmhc model .
weak primordial turbulence is damped by buoyancy forces of the first structures to form fossil vorticity turbulence @xcite .
the density and rate - of - strain values at the time of first structure ( @xmath138 s ) are preserved ( fossilized ) , and can still be detected in the density of globular star clusters and the masses of pfps .
the non - baryonic matter diffused to fill the protosuperclustervoids , limiting the amplitude of the density contrast formed to the small values observed in the cmb .
superclusters , clusters and galaxies never contracted or collapsed but continue to fragment and gently expand at slower rates than the rest of the universe .
evidence supporting the hydro - gravitational theory of structure formation is gradually accumulating .
high resolution telescope images in star forming regions and supernova remnants are full of poorly explained small objects , knots and `` ejecta '' that are likely to be evaporating pfps .
cometary globules and the huge masses of gas observed by the hst near hot dying stars in planetary nebulae are much more easily explained as the result of evaporating ambient pfps in the interstellar medium than they are as ejecta .
plasma beams from young stars produce herbig - haro objects with cometary tails , and the plasma beams themselves pulsate as though the accreted material forming the star contained planetary mass chunks .
numerous brown dwarfs and jupiters have been detected whose origin is mysterious without hgt .
the total mass of extra - solar planets discovered increases with decreasing planet - mass , even though low mass extra - solar planets are more difficult to detect , as expected if extra - solar planets are formed by accretion of ambient pfps before their capture by stars .
the closest planetary nebula is helix , figure 2 , where some 6500 cometary globules have been identified , each with mass about @xmath180 kg and size @xmath137 m , and separated by about @xmath164 m @xcite .
radial tails point away from the hot , central , white dwarf star at @xmath181 m. the mass @xmath182 kg and separation distances @xmath183 m of the helix cometary globules give a halo density @xmath184 kg m@xmath4 , suggesting that pfps originally in a sphere with radius @xmath181 m have accreted to form the central star out of an ism with mass dominated by pfps , with primordial density @xmath184 kg m@xmath4 , leaving pfps at the hole boundary ( oort cloud ) evaporated and illuminated by the hot dying star .
measured evaporation rates @xcite are too high for the globules shown in figure 2 to be pfps , so they must be jpps with at least jupiter mass to have survived this long @xcite .
if the ism surrounding the helix represents a dark proto - globular - star - cluster ( pgc ) , the pn represents the luminous boundary of a hole formed by accreted pfps to form the illuminating central star , with mass @xmath185 of the surrounding ism pgc baryonic dark matter cloud of pfps .
the gas in the nebula is of order the mass of the star , but the total mass of the atmosphere of the dying star before it is ejected should be less than about @xmath186 .
the mass density for the helix globules @xmath187 kg/@xmath144 is three orders of magnitude larger than the density of any atmosphere ejected by the white dwarf star .
shock waves can increase densities by no more than factors @xmath188 , not the factors @xmath189 required to explain the objects of figure 2 .
these cometary globules are usually explained as due to rayleigh - taylor instability , where a dense gas shell ejected first is accelerated by a lighter shell ejected more rapidly later , but rt instability can not increase the density of the dense shell by many orders of magnitude or decrease the density in the rt fingers to form wakes as observed .
rt fingers only form in a narrow range of reynolds numbers near critical , but this is not the case for the helix objects . if the cometary globule separations in helix reflects that of the ambient pfp population , the mass density of pfps is near the value @xmath3 kg/@xmath144 expected for a pgc .
the pgc may have supplied the pfps from which the helix star was formed , consistent with the size of the @xmath190 m region surrounding the white dwarf that is observed to be empty of cometary globules @xcite . however , the most conclusive and convincing evidence for the pgc - pfp hgt scenario is still the technically challenging one of quasar microlensing . the schild 1996 interpretation that the qso 0957 + 561a , b quasar lens galaxy must be dominated by planetary mass objects
is based on the observed high twinkling frequency of the two quasar images a , b .
the initially controversial schild time delay has now been confirmed by several independent observers , and is in the process of further refinement by an international team of observers using telescopes at sufficient points on the earth to provide continuous coverage and independent confirmation of the twinkling qso 0957 + 561a , b light curves ( colley et al . 2002 , 2003 ) . at this time
four other lensed quasars have been observed with sufficiently precise time delays of the mirage images to extract the planetary mass twinkling frequencies , and these have been observed and reported ( burud et al .
2000 , 2002 ; hjorthe et al .
2002 , schechter et al .
the specific microlensing profile discovered in q0957 by colley and schild ( 2003 ) with high statistical significance shows that the lens galaxy g1 halo population has a significant optical depth of planetary mass objects , and the double - ring halo model of schild and vakulik ( 2003 ) shows how such rapid microlensing plausibly follows from simulations .
evidence of substructure in several lensing galaxies is reported by dalal and kochanek 2002 from the difference in brightness of the background source mirages , with substructure masses in the range @xmath191 .
this evidence is presented to suggest that the missing cdm halos of the local group of galaxies is not a crisis for the hierarchical cdm model of galaxy formation that overestimates the numbers of dwarf galaxies by orders of magnitude compared to observations because the actual galaxy substructures may be dark .
however , the substructure evidence @xcite of dark non - baryonic cdm halos may also be interpreted as evidence that significant fractions of the lens galaxies central masses are in dark baryonic clumps .
dark pgc clumps of pfps dominating central galaxy masses is predicted by hgt .
clumps of dark pgcs triggered to form clumps of ygcs by the star formation wake of vv29c are indicated in the vv29b filament of tadpole by the hst / acs images @xcite , figure 3 .
merging galaxy systems such as the tadpole , mice , and antennae are characterized by the appearance out of the dark of young - globular - clusters ( ygcs ) and large quantities of gas .
using the hst and a wide variety of other telescopes to cover a broad range of frequencies @xcite , nearly a thousand ygcs have been identified and studied in the antennae system ( ngc 4038/9 ) , which is the closest merging galaxy system at only @xmath192 m. at this distance the clusters are well resolved .
the mass function inferred from the luminosity for the young ( 5 myr ) globular clusters in antennae @xcite is very different than that for old ( @xmath193 gyr ) globular clusters ( ogcs ) , falling with power law slope -2 from a maximum near @xmath194 rather than showing the narrow lognormal peak near @xmath195 widely observed for ogcs .
the brightest ygcs have the same mass as the ogcs .
we suggest that the low masses of the most numerous ygcs is simply because most of the mass of the pgcs from which these ygcs are forming is still tied up in dark matter pfps that are rapidly accreting to form stars under the influence of the tidal merger .
figure 4 shows an hst image of stephan s quintet ( sq , hgc 92 , arp 319 , vv 288 ) which is number 92 in the hickson catalog of compact clusters .
like over 40@xmath196 of the hgcs , the sq galaxies @xcite have highly discordant redshifts .
these are easily explained by hgt , where the sq and other hgc galaxies with discordant redshifts represent galaxy clusters that have gently separated only recently due to the expansion of the universe @xcite .
the star wakes in figure 4 show luminous evidence of the baryonic dark matter cluster halo and its boundary .
no evidence of a galaxy merger is shown , in contrast to the tadpole merger in figure 3 where the merged galaxy is revealed .
numerous young globular clusters ( ygcs ) have been identified in sq @xcite , confirming the hgt prediction that the baryonic dark matter consists of pgc clumps of pfps that can be triggered to form stars by radiation and tidal forces .
figure 5 shows the tomographic mass distribution of the rich galaxy cluster abell 1689 determined from the gravitational distortion of 6000 background galaxy images by 4000 foreground galaxies in the cluster @xcite .
a mass to light ratio of about 400 suggests the cluster halo is non - baryonic dark matter with an extent determined by the nbdm particle diffusivity @xmath197 and the large cluster density .
the schwarz diffusive scale @xmath198 from the mass profile is @xmath199 m , giving @xmath200 m@xmath6 s@xmath2 from the cluster density @xmath201 kg m@xmath4 .
such a large diffusivity excludes any possibility that the outer dark matter halo is baryonic .
it fixes the value of the @xmath202 ratio to a value @xmath203 kg m@xmath68 . because the maximum collision cross section @xmath115 for non - baryonic matter is about @xmath204 m@xmath6 ,
the maximum particle mass for the nbdm is @xmath205 kg , excluding wimps with @xmath206 kg as nbdm candidates by large factors .
for example , the neutralino mass is about @xmath207 kg , giving a @xmath115 value of @xmath208 m@xmath6 , which is so large that neutralino - nbdm would be easily detected .
the nbdm becomes unstable to gravitational fragmentation when @xmath118 becomes less than the horizon scale @xmath24 .
this occurs at @xmath209 , soon after the plasma - gas transition at 300,000 years , for the indicated diffusivity .
outer halos scale with @xmath210 so the galaxy nbdm outer - halo - radius is @xmath211 m , or about four times the observed bdm inner - halo - radius for the tadpole galaxy @xcite .
since baryons can be no more than @xmath212 of the mass of a flat universe from nucleosynthesis , non - baryons can be no more than @xmath213 times the mass of baryons . thus ( neglecting any @xmath26 contribution ) galaxy outer halo densities are dominated by non - baryons and galaxy inner halo densities are dominated by baryons .
high resolution images of the helix nebula , the tadpole merger and the stephan quintet ( sq ) compact cluster support the hydro - gravitational theory ( hgt ) prediction that primordial - fog - particles ( pfps ) in proto - globular - star - clusters ( pgcs ) dominate baryonic dark matter ( bdm ) inner - galaxy - halos and the interstellar medium ( ism ) .
young globular clusters and star wake alignments in tadpole and sq are explained by hgt and are not explained by previous models .
evidence accumulates showing high frequency twinkling of quasar images lensed by foreground galaxies , confirming that the galaxy missing baryonic mass is dominated by pfp mass `` rogue planets '' @xcite .
large numbers of extra - solar planets have been discovered with jupiter mass which must be made of h - he gas , but which are shown to be impossible by the jeans 1902 gravitational structure formation criterion in @xmath214 .
such planets can not be formed by condensation on rocky cores because the formation of rocks is also impossible by the jeans criterion .
jupiters and rocks form naturally using hydro - gravitational theory @xcite by accretion of pfps to form larger planets and stars , and by gravitational accumulation of the dust of supernovas by pfps and larger planets to form rocky cores .
the helix nebula and other planetary nebula show evidence that the ism mass is dominated by pfps as predicted by hgt .
these accrete to form jpps ; jupiters , brown dwarfs and stars .
evidence that the total mass of small extra - solar jpps exceeds the total mass of larger extra - solar jpps , as expected from hgt , suggests that the total mass of brown dwarfs in a galaxy exceeds the mass of stars .
this prediction of hgt can be tested by the space infrared telescope facility ( sirtf ) when it is launched , possibly this year .
stephan s quintet is an example of many compact clusters of galaxies with discordant red shifts that refute cdmhcc and support hgt .
hst images confirm the pgc - pfp baryonic dark matter halo model indicated by the tadpole merger , but show star wakes of cluster galaxies separating through the bdm cluster halo rather than merging as shown by tadpole and as predicted by hgt .
angular clustering of qsos and bright galaxies @xcite also rule out cdmhcc and support either hgt without new physics or intrinsic red shifts and qso ejection by agn galaxies @xcite and a possible failure of the big bang hypothesis @xcite .
the forty year authoritative accumulation of inexplicable angular clustering of galaxies , clusters and qsos with anomalous red shifts are readily explained by hgt as the result of perspective and gentle straining along lines of sight by the expansion of the universe , with small transverse velocities reflecting the viscous origins and frictional effects of fragmenting clusters in gassy , baryonic , galaxy and cluster halos . from hgt and the mass distribution
@xcite computed for a rich galaxy cluster we find the nbdm particle mass is too small to be a cdm candidate , with fragmentation scales for non - baryonic - dominated outer - galaxy - halos larger than radius values observed for baryon - dominated inner - galaxy - halos .
cusps at galaxy cores expected from small , clustered , collisionless , cdm - halos are not observed @xcite .
hierarchical clustering cosmologies are contradicted by the incidence of strong - lensing clusters @xcite .
the most distant observable galaxies are already clustered @xcite and the closest clusters @xcite are fragmenting .
hgt and the preponderance of accumulated evidence suggest that cold - dark - matter , hierarchical - clustering - cosmologies ( cdmhccs ) , and the jeans 1902 gravitational structure formation criterion are physically incorrect and contrary to observation , and should be abandoned .
gibson , c. h. and schild , r. e. 2003a , astro - ph/0304107 gibson , c. h. and schild , r. e. 2003b , astro - ph/0210583 gibson , c. h. and schild , r. e. 2003c , astro - ph/0306467 gladders , m. d. , hoekstra , h. , yee , h. k. c. , hall , p. b. , and barrientos , l. f. 2003 , astro - ph/0303341v2 . + schwarz diffusive & @xmath118&@xmath217^{1/4}$ ] & @xmath218 balances @xmath219 + schwarz viscous & @xmath140&@xmath220^{1/2}$ ] & viscous force balances gravitational force + schwarz turbulent & @xmath141&@xmath221^{3/4}$ ] & turbulence force balances gravitational force + kolmogorov viscous & @xmath142&@xmath222^{1/4}$ ] & turbulence force balances viscous force + batchelor diffusive & @xmath223&@xmath224^{1/2}$ ] & diffusion balances strain rate + collision & @xmath225&@xmath226 & distance between particle collisions + horizon , hubble & @xmath227&@xmath228 & maximum scale of causal connection + bdm & baryonic dark matter&pgc clumps of pfps from hgt + cdm & cold dark matter & questioned concept + cdmhcc & cdm hcc & questioned concepts + hcc & hierarchical clustering cosmology & questioned concept + hgc & hickson compact galaxy cluster & stephan s quintet ( sq = hgc 92 ) + hgt & hydro - gravitational theory & modifies jeans 1902 + ism & inter - stellar medium & mostly pfps and gas from pfps + jpp&jovian pfp planet&planet formed by pfp accretion + nbdm & non - baryonic dark matter&possibly neutrinos + ogc & old globular star cluster & pgc forms stars at
@xmath229 yr + pfp&primordial fog particle&planet - mass protogalaxy fragment + pgc & proto - globular star cluster & jeans - mass protogalaxy fragment + ygc & young globular star cluster & pgc forms stars at @xmath230 now + | observations are compared to conflicting predictions about self - gravitational structure formation by the hydro - gravitational theory ( hgt ) of gibson 1996 - 2003 versus cold - dark - matter hierarchical - clustering - cosmology ( cdmhcc ) and the jeans 1902 criterion . according to hgt , gravitational structures form immediately after mass - energy equality by plasma fragmentation at 30,000 years when viscous and weak turbulence forces first balance gravitational forces within the horizon @xmath0^{1/2}$ ] , contrary to the jeans 1902 criterion .
buoyancy forces fossilize the @xmath1 s@xmath2 rate - of - strain and the @xmath3 kg m@xmath4 baryonic density .
the non - baryonic dark matter ( nbdm ) diffuses into the voids rather than forming cold - dark - matter ( cdm ) halos required by cdmhcc . from hgt ,
supercluster - mass to galaxy - mass fragments exist at the plasma to gas transition , and these fragment further to form proto - globular - star clusters ( pgcs ) and planetary - mass primordial - fog - particles ( pfps ) : the baryonic dark matter of the interstellar - medium and inner - galaxy - dark - matter - halos , from which all planets and stars are formed by accretion ( gibson 1996 , schild 1996 ) . from hgt and a rich cluster mass profile @xcite , @xmath5 m@xmath6 s@xmath2 , @xmath7 kg , and the nbdm forms outer - galaxy halos after 300,000 years . |
the models with multispin interactions have attracted much attentions , such as the eight - vertex model @xcite which involves both two- and four - spin interactions , the ising model with four - spin interactions @xcite and the baxter - wu model @xcite which is a non - trivial ising model with three - spin interactions on a triangular lattice .
the hamiltonian of the baxter - wu model is@xmath0where @xmath1 denotes a triplet of nearest neighbor sites of a triangular lattice , @xmath2 is the coupling constant , and @xmath3 _ _ _ _ is the ising spin located at site @xmath4 . besides the two - dimensional ising model @xcite , the baxter - wu model is another exactly solvable one which exhibits spontaneous symmetry breaking .
the system has been studied by series - expansion analysis griffiths73,watts74,adler , monte carlo @xcite , monte carlo renormalization group @xcite and conformal invariance approaches @xcite , etc .
it has the same critical temperature as the ising model with three - spin interactions in one direction on a square lattice @xcite , and belongs to the same universality class as two - dimensional 4-state potts system with the equal critical exponents @xcite , namely , their specific heat critical exponent and the correlation length exponent are equal to 2/3 @xmath5 , and correlation function exponent is @xmath6 .
in addition , costa _
et al_. @xcite have studied the spin-1 baxter - wu model with a crystal field and found that the phase diagram presents a tricritical point . in 1990s ,
due to describing incommensurate phases of quantum one - dimensional magnetics , tsvelik constructed a modified xxz model @xcite with hamiltonian@xmath7where @xmath8 are pauli matrices , @xmath9 is the total number of spins , @xmath10 is the three - spin interactions strength . with competing of the two and three - spin interactions , the model exhibits several quantum phase frahm1992,lou2004 . in this paper , imitating eq .
( [ mxxz ] ) we define a quantum ising model with three - spin interactions on a triangular lattice .
it is well - known that the blume - capel ( bc ) model @xcite and the blume - emery - griffiths ( beg ) model @xcite present much interesting critical phenomenon : tricritical point , multicritical phase diagrams @xcite , the reentrant phenomenon @xcite , metastable and unstable states lapinskas , ekiz .
similar to the bc model and beg model , we study the spin-1 quantum ising systems with a crystal field and biquadratic interactions by mean - field method .
our emphases are the thermodynamic properties and phase diagrams of the systems .
we expect our system can exhibit that interesting proprieties .
the outline of this paper is as follows . in section [ model ] , using the mean - field method , we study the spin-1 quantum ising system with a crystal field . in sec .
[ beg ] , the spin-1 quantum ising system with biquadratic interactions is studied in detail , and sec .
[ summary ] is a brief summary .
the spin-1 quantum ising model with three - spin interactions in presence of a crystal field on a triangular lattice is defined by the hamiltonian @xmath11 where @xmath12 , @xmath13 @xmath14 are components of the spin-1 operator at site @xmath4 , and @xmath15 plays the role of a crystal field .
@xmath16 are @xmath17 in the following discussion , the coupling constant @xmath18 , that is , a ferromagnetic case is considered .
based on the mean - field approximation fittipaldi , sun , the three - spin cluster hamiltonian @xmath19 can be written as@xmath20 where @xmath21 is the average magnetization of the cluster , @xmath22 . in the representation of the direct product of @xmath23 , @xmath24 and @xmath25 ,
@xmath19 can be written as a form of @xmath26 matrix .
thus the partition function @xmath27 tr@xmath28 $ ] ( @xmath29 , where @xmath30 is the boltzmann constant , @xmath31 denotes the temperature of the system ) can be obtained as@xmath32 in which@xmath33@xmath34 \text{,}\ ] ] where @xmath35 is the reciprocal of the reduced temperature , and @xmath36 is the reduced crystal - field parameter .
the reduced free energy @xmath37 of the three - spin cluster in the system as a function of @xmath38 and @xmath21 is further given by@xmath39 \text{.}\ ] ] according to the equilibrium condition of the system @xmath40 , we can obtain the self - consistent equation of the magnetization@xmath41 where@xmath42 by solving the self - consistent eq .
( [ m1 ] ) , numerically , we obtain the thermal variations of the magnetization for several values of @xmath43 ( fig . 1 ) .
we obtain the stable , metastable and unstable branches , which are represented by the solid , dot and dash - dotted lines , respectively .
it can be seen that for @xmath44 , @xmath45 and @xmath46 the stable magnetization are continuous , indicating the second - order phase transitions . for @xmath47 and @xmath46
, we also find unstable solutions . on the other hand , the stable magnetization drops discontinuously from a finite value @xmath48 to zero at a temperature @xmath49 , which characterizes a first - order phase transition .
the first - order phase transition temperature @xmath50 can be obtained by the free energy , when its local minimum at @xmath51 is equal to the local minimum at @xmath52 @xcite . when the reduced crystal - field parameter is slightly larger than the tricritical point @xmath53
, the system can undergo the first - order phase transition , such as , for @xmath54 the first - order phase transition temperature is @xmath55 .
figure 2 shows the phase diagram of the spin-1 quantum ising system in the presence of a crystal field .
the critical line separates the ferromagnetic phase from the paramagnetic phase .
we can see that the second - order phase transition ( solid line ) occurs for finite values of the crystal field .
the second - order phase transition temperature falls smoothly with the reduced crystal - field parameter @xmath43 increasing and reaches the tricritical point .
the value of the tricritical point is _ _
_ _ @xmath56 _ _ _ _ and _ _ _ _ @xmath57 , _ _
_ _ which is different from that in ref [ 13 ] _
_ _ _ @xmath58 _ _ and _ _ _ _ @xmath59__. _ _ then , the first - order phase transition line falls smoothly from the tricritical point to zero .
this phenomenon is similar to the spin-1 baxter - wu model @xcite .
it need to clear that the last point of the curve in fig . 2 is _ _
_ _ @xmath60 _ _ _ _ and _ _ _ _ @xmath61__. _ _
the hamiltonian of the spin-1 quantum ising system with biquadratic interactions on a triangular lattice is defined as @xmath62 where @xmath63 indicates all nearest - neighbor pairs of sites of a triangular lattice and @xmath64 denotes the biquadratic interactions parameter .
different from that in sec .
[ model ] , for solving this model , we need two order parameters : the magnetization @xmath21 and quadrupolar moment @xmath65.@xmath66using the similar method as in sec . [ model ] , the three - spin cluster hamiltonian @xmath67has the following expression:@xmath68 where @xmath21 has been defined in sec . [ model ]
, @xmath65 is the average squared magnetization of the three - spin cluster , @xmath69 . according to the definition of _ _ _ _ the partition function , it is obtained as @xmath70 in which@xmath71@xmath72@xmath73@xmath74 where @xmath75 is the reduced biquadratic parameter .
the reduced free energy of the three - spin cluster can be written as@xmath76 with the equilibrium conditions ( @xmath77 and @xmath78 ) of the system , the self - consistent equations of the two order parameters are@xmath79 and@xmath80 in which@xmath81@xmath82 where@xmath83\text{.}\ ] ] by numerically calculating eqs .
( [ self - consistent - m ] ) and ( self - consistent - q ) , the behavior of the magnetization @xmath21 and quadrupolar moment @xmath65 as functions of temperature are obtained ( see figs . 3 and 4 ) .
we obtain the stable ( m1 and q1 ) , unstable ( m2 and q2 ) and metastable ( m3 and q3 ) branches . as in sec .
[ model ] , the stable branches of the order parameters undergo the first - order or second - order phase transition . what is interesting is that in figs .
3(d ) and 3(e ) , the stable branches of the order parameters undergo one first - order phase transition and one second - order phase - transition successively .
4(b ) shows that the stable branches of the order parameters undergo two second - order phase transitions and one first - order phase transition .
otherwise , what should be mentioned is that the stable branch of the quadrupole order parameter does not undergo any phase transition when the stable branch of the magnetization is equal to zero @xcite ( see fig .
seen from fig . 3 , at @xmath84 , the system can be in the ground state with @xmath85 , @xmath86 _ _ _ _ or @xmath87 , which is similar to the ground state of the @xmath88 ising model on triangular lattice @xcite .
the actual ground state depends on the competition between the reduced biquadratic interaction and crystal field .
figure 5 shows three phase diagrams for different biquadratic parameter @xmath89 .
it can be seen that the first - order phase transition can occur either in the range of @xmath90 ( fig .
5(a ) ) or @xmath91 ( fig .
at fixed reduced biquadratic and crystal - field parameter in fig .
5(a ) , when the temperature decrease the transition from paramagnetic phase ( p ) to ferromagnetic phase ( f ) is encountered , and if the temperature is lowered further , the reentrant phase transition from ferromagnetic phase to paramagnetic phase can take place .
5(c ) shows a double reentrant phenomena , that is , at fixed reduced biquadratic and crystal field parameter , as the temperature decrease , the p@xmath92f@xmath93p@xmath94f sequence of phase is encountered .
it may be determined by the competition among quantum effect , crystal - field and the biquadratic interaction .
in this paper , we have investigated the spin-1 quantum ising systems with three - spin interactions on two - dimensional triangular lattices by mean - field approximation .
the thermal variations of the magnetization @xmath21 and quadrupolar moment @xmath65 are obtained . by comparing the free energy of these solutions ,
we obtain the stable , unstable and metastable branches of the order parameters .
in addition , we find that the quantum ising systems present tricritical points for finite values of the crystal field , and the system with biquadratic interactions exhibits single reentrant and double reentrant phase transitions . this work was supported by the national natural science foundation of china under grant no .
10775088 , the shandong natural science foundation ( grant no .
y2006a05 ) , and the science foundation of qufu normal university .
1the magnetization @xmath21 versus temperature @xmath95 for the spin-1 quantum ising system with a crystal field on a triangular lattice .
curves a , b , c , d represent the stable branches for @xmath44 , @xmath45 , @xmath46 , @xmath96 , respectively . curves b@xmath97 , c@xmath97 , d@xmath98 represent the unstable branches and curve d@xmath99 is the metastable branch .
2the phase diagram of spin-1 quantum ising system with a crystal field on a triangular lattice .
the solid line represents the second - order phase transition and dot line represents the first - order phase transition . the big real point ( solid dot )
is the tricritical point .
p and f represent the paramagnetic and ferromagnetic phases , respectively .
3the magnetization @xmath21 and quadrupolar moment @xmath65 as functions of temperature @xmath100and crystal - field parameter @xmath43 with @xmath101 for the spin-1 quantum ising model with biquadratic interactions on a triangular lattice .
m1 , q1 ( solid line ) are the stable branches of the magnetization @xmath21 and quadrupolar moment @xmath65 , respectively ; m2 , q2 ( dot line ) represent the unstable branches ; m3 , q3 ( dash - dot line ) represent the metastable branches . @xmath102 and
@xmath103 indicate the first - order and second - order phase transition temperature , respectively .
4the nonzero order parameters @xmath21 and @xmath65 as functions of temperature @xmath100and crystal - field parameter @xmath43 with @xmath104 for the spin-1 quantum ising model with biquadratic interactions on a triangular lattice .
m1 , q1 ( solid line ) represent the stable branches ; m2 , q2 ( dot line ) represent the unstable branches . @xmath102 and
@xmath105 indicate the first - order and second - order phase transition temperatures , respectively .
@xmath106 is the unstable critical temperature .
5phase diagrams of the spin-1 quantum ising model with biquadratic interactions on a triangular lattice for different @xmath107 .
the solid and dot lines represent the second - order and first - order phase transition lines and the solid dot represents the tricritical point .
p and f denote the paramagnetic and ferromagnetic phase , respectively .
( a ) shows a p@xmath108f@xmath94p reentrant phase transitions .
( c ) shows a p@xmath108f@xmath94p@xmath94f double reentrant phase transitions . | the spin-1 quantum ising systems with three - spin interactions on two - dimensional triangular lattices are studied by mean - field method .
the thermal variations of order parameters and phase diagrams are investigated in detail .
the stable , metastable and unstable branches of the order parameters are obtained . according to the stable conditions at critical point , we find that the systems exhibit tricritical points . with crystal field and biquadratic interactions ,
the system has rich phase diagrams with single reentrant or double reentrant phase transitions for appropriate ranges of the both parameters . |
sterile neutrinos @xmath3 can be produced in the early universe or in supernova ( sn ) cores if they mix with one of the active flavors @xcite .
even if the mixing angle @xmath7 is very small , repeated collisions of the active component allow for an efficient @xmath3 production .
the sn 1987a neutrino signal duration implies @xmath8 to avoid excessive energy losses @xcite . in detail , this limit depends on whether the dominant mixing is with @xmath9 or one of the other active flavors .
this result assumes @xmath10 kev for matter effects on neutrino propagation to be negligible compared with the vacuum mass . for smaller masses
, the matter effect typically suppresses the effective mixing angle and thus diminishes the limit on @xmath11 . for @xmath12
kev even maximal mixing is allowed .
however , no detailed treatment of the sn bound exists for @xmath13 , where matter and resonance effects are important @xcite . on the other hand , this is precisely the mass range where sterile neutrinos could play an interesting warm dark matter role in cosmology @xcite . while sterile neutrinos can be produced in the early universe by different mechanisms @xcite ,
the production by oscillations and collisions once more depends on the active - sterile mixing angle , so one naturally wonders about the sn bound on @xmath7 .
in particular , we study the role of feedback of sterile neutrino emission on the emission rate itself .
when matter effects are important , the mixing angle is resonantly enhanced in some range of neutrino energies @xcite . at first the emission of @xmath5
is more efficient than @xmath3 because in the neutrino sector the mixing angle is suppressed for all energies . as a consequence , the active flavor , being trapped in the sn core , builds up a @xmath14 excess .
the opposite could happen for larger mixing angles , when most antineutrinos @xmath5 are trapped rather than freely escape and the emission of @xmath3 is more efficient than that of @xmath5 . to be specific we use @xmath2 as the active flavor because there is no initial @xmath2-@xmath4 asymmetry in a sn core and @xmath15 is so large that charged @xmath16 leptons never play any role .
our main point is that the depletion of @xmath2 or @xmath4 relative to the other always goes in the direction of quenching the initial emission rate , implying that in the 1100 kev - mass range the @xmath7 bounds are indeed suppressed .
our conclusion differs somewhat from previous discussions where it was suggested that the buildup of a @xmath2-@xmath4 asymmetry goes in the direction of reducing the matter effect and leads to restrictive sn limits on @xmath7 @xcite .
while we agree that the matter effect can be modified in this direction , the positive @xmath2-@xmath4 asymmetry also implies a depletion of source @xmath4 to be converted to @xmath5 relative to source @xmath2 to be converted to @xmath3 .
it is also possible in some region of the parameter space that the matter effects are enlarged due to a negative @xmath2-@xmath4 asymmetry . in both cases ,
the compound effect is a reduction , not an enhancement , of the energy loss . in sec .
[ sec : matter ] , we briefly review the matter effects on active - sterile neutrino mixing in the sn core .
the development of a @xmath2-@xmath4 asymmetry is discussed in sec .
[ sec : state ] , where we identify a stationary state with equal neutrino and antineutrino emission rates and estimate the time scale to reach it .
[ sec : snbounds ] is devoted to the calculation of the energy - loss rate caused by sterile neutrinos and the sn bound on sterile neutrino masses and mixing angles . finally , we summarize our conclusions in sec . [
sec : conclusions ] .
the dispersion relation of neutrinos will be modified in matter due to the coherent forward scattering of neutrinos off background particles @xcite .
this matter effect can be described by an effective potential @xmath17 for each kind of active neutrino @xmath18 , @xmath19 and @xmath2 .
the effective potentials for antineutrinos have the opposite signs , i.e. , @xmath20 . in the case of @xmath2-@xmath3 oscillation in matter , the effective hamiltonian is @xmath21 where @xmath22 , @xmath23 with @xmath7 being the vacuum mixing angle and @xmath24 the oscillation frequency in vacuum . as far as the kev - mass sterile neutrinos
are concerned , we have @xmath25 with @xmath1 being the sterile neutrino mass .
in contrast with the flavor conversions of ordinary neutrinos in the sn , there are no collective effects in active - sterile neutrino oscillations . in sn cores ,
the main ingredients of matter are protons @xmath26 , neutrons @xmath27 , electrons @xmath28 , some muons @xmath29 as well as active neutrinos @xmath30 and antineutrinos @xmath31 .
we ignore the possibility of a meson condensate or hyperons .
in such a medium , the effective potential for tau neutrinos is @xmath32 \ ; , \ ] ] where @xmath33 is the fermi constant , @xmath34 the baryon number density , and @xmath35 with @xmath36 and @xmath37 being the number densities of particle @xmath38 and its antiparticle @xmath39 .
because of charge neutrality @xmath40 , we have @xmath41 .
while @xmath30 and @xmath31 of all flavors can be produced in pairs , electron neutrinos @xmath9 can also be generated in beta processes such as @xmath42 , and similar for muon neutrinos .
the muon mass @xmath43 is comparable to the average thermal energy @xmath44 with @xmath45 , thus a small population of muons is unavoidable .
beta equilibrium @xmath46 implies the relation among chemical potentials @xmath47 , where @xmath48 typically lies in the range 50100 mev , depending sensitively on the equation of state . noting that the initial @xmath29 lepton number is vanishing and taking @xmath49
, one finds @xmath50 and @xmath51 @xcite . on the other hand
, the @xmath9 chemical potential will be much larger due to the electron lepton number trapped during infall . for simplicity
, we consider a sn core just after the bounce and assume the temperature @xmath45 and matter density @xmath52 to be constant .
we furthermore take a typical value of electron lepton number fraction @xmath53 , which leads to @xmath54 and @xmath55 due to beta equilibrium with @xmath56 .
it is straightforward to verify that @xmath57 and @xmath58 , so henceforth we simply set @xmath59 .
tau neutrinos initially follow the fermi - dirac distribution without chemical potential , but later an asymmetry develops due to @xmath3 emission . with these simplifications , the effective potential in eq .
( [ eq : potential ] ) is @xmath60 it is now evident that @xmath61 is negative for @xmath62 , implying that the mikheyev - smirnov - wolfenstein resonance occurs in the antineutrino sector @xcite . given the effective hamiltonian in eq .
( [ eq : hamiltonian ] ) , one immediately obtains the effective mixing angle @xmath63 where the upper sign refers to @xmath14 and the lower to @xmath64 .
the resonant energy @xmath65 is @xmath66 where @xmath67 is the matter density @xmath68 in units of @xmath69 and @xmath70 . as indicated by eq .
( [ eq : angle ] ) , the mixing angle @xmath71 for the whole energy range is always suppressed by matter effects , while @xmath72 can be resonantly enhanced for @xmath73 . note that the vacuum limit " with @xmath74 is reached for large sterile neutrino masses @xmath75 , while the `` medium limit '' with reduced mixing angle @xmath76 is obtained for small masses @xmath77 . for intermediate masses
, we have a resonance in the antineutrino sector .
sterile neutrinos are produced in the sn core via oscillations and collisions of tau neutrinos .
if the effective mixing angles @xmath71 and @xmath72 are small enough , @xmath3 and @xmath5 can escape from the core immediately after production .
since the mixing angle of antineutrinos is always larger than that of neutrinos , the emission rate of antineutrinos exceeds that of neutrinos .
consequently , a @xmath2-@xmath4 asymmetry arises from these different emission rates . as we can observe from eq .
( [ eq : resonant - energy ] ) , it might turn out that a relatively large @xmath2-@xmath4 asymmetry is achieved , i.e. , @xmath78 , so as to drive the resonant energy @xmath79 to infinity , leading to the `` vacuum limit '' even for small masses . put another way , the @xmath2-@xmath4 asymmetry seems to develop in the direction of reducing the matter effects , resulting in restrictive sn limits on the vacuum mixing angle @xmath7 for both large and small sterile neutrino masses .
it has been argued @xcite that there exists a stationary state with @xmath80 , which can be achieved rapidly and thus validates the vacuum limit " and restrictive bounds on @xmath7 . however , the state with @xmath80 can not be stationary , because the population of @xmath2 at this moment is larger than that of @xmath4 , implying that more neutrinos than antineutrinos are ready to be emitted and thus the condition @xmath80 breaks down .
furthermore , it is even possible that the emission rate of @xmath3 exceeds that of @xmath5 , since the effective mixing angle @xmath72 can be so large that most of @xmath5 are trapped in the core . therefore one may obtain a negative asymmetry @xmath81 , driving the system towards the `` medium limit . '' in the following sections , we shall examine how the @xmath2-@xmath4 asymmetry actually develops , and explore its implications on the anomalous energy - loss rate of the sn core and thus the sn bounds on sterile neutrinos .
the matter density of sn cores is so high that both neutrino oscillations and frequent collisions with background particles are important .
an elegant method to treat neutrino flavor conversions in this case is to implement the matrix of occupation numbers @xmath82 , where @xmath83 denotes the annihilation operator for a neutrino of flavor @xmath84 and momentum @xmath85 , and to derive the non - abelian boltzmann equations of @xmath86 @xcite .
the diagonal elements @xmath87 are the usual occupation numbers @xmath88 , while the off - diagonal ones encode the phase information .
the analogous definition for antineutrinos is @xmath89 with @xmath90 being the annihilation operator for an antineutrino of flavor @xmath84 and momentum @xmath85 . in general , the equations of motion for @xmath86 and @xmath91 are complicated by the nonlinear nature of the collision integrals . for kev - mass sterile neutrinos , the problem can be much simplified by taking the weak - damping limit , which is usually valid in sn cores . to be more explicit
, we estimate the neutrino oscillation length in matter @xmath92 and the mean free path of tau neutrinos @xmath93 where @xmath94 is the cross section of neutrino - nucleon scattering via the neutral - current interaction .
the weak - damping limit is @xmath95 , meaning that the active - sterile neutrino oscillations take place many times before a subsequent collision of active neutrinos with the nucleons .
it deserves mention that the weak - damping limit is violated for smaller masses @xmath96 , however , the effective neutrino mixing angle in this case is highly suppressed by matter effects and thus the energy - loss rate is negligibly small .
we shall always assume the weak - damping limit in the mass range of our interest . in the weak - damping limit , @xmath86 is averaged over many cycles of oscillations and then can be parametrized by the neutrino occupation numbers @xmath97 and @xmath98 with the neutrino energy @xmath99 and similar for antineutrinos @xcite .
note that we consider a homogenous and isotropic ensemble of active neutrinos and antineutrinos , which are trapped in the sn core and stay in thermal equilibrium with ambient matter , so the momentum direction is irrelevant . assuming that sterile neutrinos freely escape
, we obtain the evolution equation of the @xmath2 density @xmath100 where @xmath101 denotes the target particle , and @xmath102 the transition probability for @xmath103 via the neutral - current interaction . in a similar way
, we can derive the evolution equation of the @xmath4 number density , involving the mixing angle @xmath72 , the occupation number @xmath104 and the transition probability @xmath105 .
however , only the neutrino energies ensuring @xmath106 are taken under the integration , where the critical mixing angle @xmath107 can be estimated by requiring the mean free path of sterile neutrinos to equal the core radius @xmath108 .
on the other hand , the transition probabilities @xmath102 and @xmath109 of neutrino - nucleon scattering dominate over those of neutrino - electron scattering if the sterile neutrino mass is lying in the range of @xmath13 .
hence we consider only the neutral - current @xmath14-@xmath110 scattering in the leading - order approximation . in order to describe the @xmath2-@xmath4 asymmetry
, we assume the occupation number @xmath111^{-1}$ ] for @xmath2 and @xmath112^{-1}$ ] for @xmath4 , where @xmath113 is the temperature of the sn core and @xmath114 is the degeneracy parameter . in the absence of active - sterile neutrino mixing ,
the spectra of @xmath2 and @xmath4 just follow the fermi - dirac distribution with a vanishing chemical potential , namely @xmath115 at @xmath116 . substituting the occupation numbers into eq .
( [ eq : number - rate ] ) , taking account of the neutral - current @xmath2-@xmath110 scattering and subtracting the corresponding equation for antineutrinos , one arrives at @xmath117 \mathscr{g}^{-1}(\eta)\ ; , \ ] ] where the relevant functions are defined as follows @xmath118 and @xmath119/{\rm d}\eta$ ] with @xmath120 being the fermi - dirac integral of order two . in addition
, we have introduced @xmath121 , @xmath122 , and the box function @xmath123 , which equals one for @xmath124 $ ] and vanishes otherwise .
here @xmath125 denotes the unit step function , i.e. , @xmath126 for @xmath127 and @xmath128 for @xmath129 .
note that eq .
( [ eq : function ] ) has been cast into a compact form so as to include both @xmath130 and @xmath131 cases .
taking typical values of the matter density @xmath132 and the core temperature @xmath45 , we can rewrite eq .
( [ eq : evolution ] ) as @xmath133 with @xmath134 and @xmath135 \mathscr{g}^{-1}(\eta)$ ] . the time evolution of the degeneracy parameter @xmath114 depends crucially on the initial difference between neutrino and antineutrino emission rates , i.e. , @xmath136 , as well as the evolution of @xmath137 with respect to @xmath138 . in fig . 1
, we show the initial rate @xmath139 in the @xmath140 plane , where the blank region with
@xmath141 " denotes a strong excess of @xmath5 over @xmath3 emission while that with
@xmath142 " represents the opposite case .
the cyan regions on the left - hand and right - hand side indicate @xmath143 and @xmath144 , respectively , but the magnitude of @xmath145 is extremely small for both cases .
the reason for the former case is just that the effective mixing angle is too small , and for the latter case is that most of @xmath3 and @xmath5 are trapped . , where the blank region with @xmath141 " stands for a strong excess of @xmath146 over @xmath3 emission while that with
@xmath142 " is the other way round .
the narrow strip indicates a rapid and continuous transition between these two regions . ] if the initial emission rate of @xmath5 is larger than that of @xmath3 , i.e. , @xmath147 , the degeneracy parameter increases from zero to a positive stable value @xmath148 , at which the emission rates of @xmath3 and @xmath5 are equal , namely @xmath149 or equivalently @xmath150 . in this case , we are finally left with a positive @xmath2-@xmath4 asymmetry when such a stationary state is reached .
if the initial emission rate of @xmath3 exceeds that of @xmath5 , i.e. , @xmath151 , the degeneracy parameter decreases from zero to a negative stable value @xmath148 .
at this moment , we have equal neutrino and antineutrino emission rates as well , but a negative @xmath2-@xmath4 asymmetry .
the important point here is feedback of the established @xmath2-@xmath4 asymmetry or a finite degeneracy parameter . in the case of @xmath147 ,
a positive @xmath138 suppresses the @xmath4 population and shifts the resonant energy to a larger value .
the combined result is just to reduce the @xmath5 emission rate .
meanwhile , the population of @xmath2 is accordingly increased and the mixing angle @xmath71 becomes less suppressed , enhancing the @xmath3 emission rate .
similar arguments apply to the case of @xmath151 .
it is also possible that @xmath152 , which lies in the narrow strip in fig . 1
, the system remains in its initial state with a vanishing @xmath2-@xmath4 asymmetry .
hence we have no feedback effect in this special case . , which is determined by @xmath149 , implying equal neutrino and antineutrino emission rates .
] $ ] to reach the stationary state , where the contours are labeled by the logarithmic values @xmath153 . ] in fig . 2 ,
the degeneracy parameter @xmath148 of the stationary state has been solved from @xmath154 , no matter whether such a stationary state can be reached .
the large values of @xmath148 appear in the region where the initial emission rate of @xmath3 is significantly different from that of @xmath5 .
this condition can be satisfied for ( 1 ) small mixing angles and intermediate masses , when the mixing angle @xmath72 is resonantly enhanced so that @xmath155 ; ( 2 ) large mixing angles and large masses , when more @xmath5 than @xmath3 are trapped in the core such that @xmath156 .
however , it is obvious that the @xmath3 and @xmath5 emission rates in both cases are extremely small .
the timescale for the system to achieve the stationary state can be determined by numerically solving eq .
( [ eq : evolution ] ) . for a rough estimate
, we take @xmath157 and then obtain the timescale @xmath158 $ ] . in fig . 3 , we show the estimated timescale @xmath16 in the @xmath140 plane . in the mass range @xmath159 , where sterile neutrinos can be warm dark matter , the timescale is larger than the neutrino diffusion time @xmath160 for small mixing angles @xmath161 . for larger
mixing angles , the stationary state can be achieved within a fraction of a second , but @xmath148 in this case is quite small as shown in fig . 2 .
hence we expect that the feedback effects are negligible . in this connection ,
the most interesting parameter space should be @xmath162 and @xmath163 , where both sizable @xmath148 and @xmath164 are expected .
since @xmath165 decreases from the initial value @xmath166 to zero , as the absolute value of the degeneracy parameter @xmath167 increases from zero to @xmath168 , the relaxation time may be underestimated .
however , this rough estimate has already shown the main features of the relaxation timescale .
and mixing angles @xmath7 , where the purple region is excluded by the energy - loss argument while the green one by the energy - transfer argument .
the excluded region will be extended to the dashed ( red ) line if the build - up of degeneracy parameter is ignored , i.e. , @xmath169 .
the dot - dashed ( green ) line represents the sterile neutrinos as dark matter with the correct relic abundance @xmath170 . ]
the emission of sterile neutrinos @xmath3 and @xmath5 may cause rapid energy losses from the sn cores , which can significantly shorten the duration of neutrino signals @xcite . in order to avoid conflict with the observation of sn 1987a neutrinos
, we require the energy - loss rate per unit mass to be smaller than @xmath171 , which can be translated into the volume energy - loss rate @xmath172 for @xmath132 .
given the emission rate of neutrinos in eq .
( [ eq : number - rate ] ) and the counterpart for antineutrinos , the energy - loss rate is @xmath173 \ ; , \ ] ] with @xmath174 where the definitions of relevant parameters are given below eq .
( [ eq : function ] ) .
the energy - loss rate @xmath175 depends on time through the degeneracy parameter @xmath114 , for which the time evolution has been discussed in last section . to constrain the sterile neutrino mass and mixing angle
, we evaluate the emission rate @xmath176 averaged over the neutrino diffusion timescale @xmath177 . beyond the diffusion timescale
, one may expect that all active neutrinos have already diffused out of the core and thus the emission of sterile neutrinos is physically meaningless .
our strategy is to follow the time evolution of @xmath114 for each point in the @xmath140 parameter space , and then calculate the averaged energy - loss rate in eq .
( [ eq : average ] ) .
we show in fig . 4 the contour plot of the averaged energy - loss rate @xmath178 in the @xmath140 plane , where the purple region corresponds to @xmath179 and is thus excluded .
some comments are in order .
first , the initial energy - loss rate with @xmath115 is shown in fig . 4 for comparison . except for @xmath180 and small mixing angles
, the averaged energy - loss rate @xmath181 can not be distinguished from the initial one @xmath182 .
the reason is that either the stationary state has not been reached within @xmath183 , or the asymptotic value @xmath148 for the stationary state is quite small . in the mass range @xmath184 ,
the buildup of a @xmath2-@xmath4 asymmetry is efficient and strongly reduces the energy - loss rate .
second , kev - mass sterile neutrinos can be produced in the early universe and contribute as dark matter to the total energy density . in a nonresonant production scheme without large primordial lepton asymmetries ,
the relic sterile neutrino abundance can be estimated as @xcite @xmath185 the correct dark matter abundance @xmath170 is shown in fig . 4 , where one can see that the masses around @xmath186 have been excluded by the sn bound if the degeneracy parameter is assumed to be vanishing . in a realistic situation , this region
is retrieved because the energy - loss rate is reduced as the @xmath2-@xmath4 asymmetry builds up .
however , the warm dark matter range @xmath159 is essentially unconstrained . for sterile neutrinos of masses below @xmath187 ,
even maximal mixing is allowed , because the total energy - loss rate is highly suppressed by matter effects . finally , our discussion was based on the energy - loss argument , ignoring the sterile neutrinos trapped in the sn core .
for this reason , the top - right green region in fig . 4 , where both the sterile neutrino mass and vacuum mixing angle are large , was not excluded . however , the mean free path of these trapped sterile neutrinos is much larger than that of @xmath2 , so the energy transfer in the sn core will be more efficient .
the energy in a sn core is carried by those trapped particles with the largest mean free path , a role played here by the sterile states .
once more the neutrino burst duration is shortened too much if the sterile mfp is larger than a few times the one for @xmath2 @xcite .
therefore , the large mixing angle region is actually excluded in the spirit of the energy - transfer argument .
since kev - mass sterile neutrinos are a promising candidate for warm dark matter , we have revisited the supernova bound on the sterile neutrino masses and mixing angles by studying the case of @xmath2-@xmath3 mixing in the sn core and requiring no excessive energy losses induced by sterile neutrinos .
it turns out that the warm dark matter range is essentially unconstrained , while sterile neutrinos of masses around @xmath188 receive the most stringent constraint , i.e. , @xmath189 . for even larger masses
@xmath190 , the sn limit on the mixing angles is @xmath191 , which is about one order of magnitude weaker than that for @xmath192 .
it is the matter effects that render these constraints quite different .
we have identified a mass range @xmath193 where a sizable @xmath2-@xmath4 asymmetry can be established due to the strong excess of @xmath5 over @xmath3 emission or vice versa .
the build - up of this asymmetry feeds back on the emission rates , leading to a stationary state where the neutrino and antineutrino emissions become equal . for proper mixing angles ,
such a stationary state can be achieved within the neutrino diffusion timescale @xmath160 . as a consequence , the energy - loss rate will be significantly reduced , and thus the bounds are relaxed . as for the @xmath19-@xmath3-mixing case ,
our discussions about the feedback effects are essentially applicable .
however , the charged - current interactions of @xmath19 and @xmath194 should be taken into account , and the change of @xmath19-@xmath194 asymmetry will be redistributed between muon neutrinos and charged muons .
the @xmath9-@xmath3 mixing in sn cores is more involved because of the large trapped electron number and high @xmath9 degeneracy . besides energy loss , deleptonization by sterile neutrino emission is an effect to be taken into account
this case requires a dedicated investigation .
we have performed a `` single zone '' analysis by assuming a homogenous and isotropic sn core with constant matter density and temperature .
this treatment should capture the dominant feedback effect .
however , the local variation of these quantities may modify the final results , for example smearing out the resonance in the energy - loss rate .
such a refinement is also left for future works .
this work was partly supported by the deutsche forschungsgemeinschaft under grants no .
tr-27 and no .
exc-153 and by the alexander von humboldt foundation .
a. kusenko , phys .
rept . * 481 * , 1 ( 2009 ) [ arxiv:0906.2968 ] .
a. boyarsky , o. ruchayskiy and m. shaposhnikov , ann .
nucl . part .
sci . * 59 * , 191 ( 2009 ) [ arxiv:0901.0011 ] .
k. kainulainen , j. maalampi and j. t. peltoniemi , nucl .
b * 358 * , 435 ( 1991 ) .
g. raffelt and g. sigl , astropart .
* 1 * , 165 ( 1993 ) [ astro - ph/9209005 ] .
j. t. peltoniemi , astron .
astrophys . * 254 * , 121 ( 1992 ) .
x. shi and g. sigl , phys .
b * 323 * , 360 ( 1994 ) , ( e ) _
ibid_. b * 324 * , 516 ( 1994 ) [ hep - ph/9312247 ] .
a. d. dolgov , s. h. hansen , g. raffelt and d. v. semikoz , nucl .
b * 590 * , 562 ( 2000 ) [ hep - ph/0008138 ] .
a. d. dolgov , s. h. hansen , g. raffelt and d. v. semikoz , nucl .
b * 580 * , 331 ( 2000 ) [ arxiv : hep - ph/0002223 ] .
a. d. dolgov and s. h. hansen , astropart .
* 16 * , 339 ( 2002 ) [ hep - ph/0009083 ]
. k. abazajian , g. m. fuller and m. patel , phys .
d * 64 * , 023501 ( 2001 ) [ astro - ph/0101524 ] .
s. dodelson and l. m. widrow , phys .
lett . * 72 * , 17 ( 1994 ) [ hep - ph/9303287 ] .
x. d. shi and g. m. fuller , phys .
* 82 * , 2832 ( 1999 ) [ astro - ph/9810076 ] .
t. asaka , m. laine and m. shaposhnikov , jhep * 0701 * , 091 ( 2007 ) [ arxiv : hep - ph/0612182 ] .
l. wolfenstein , phys .
d * 17 * , 2369 ( 1978 ) .
s. p. mikheyev and a. yu . smirnov , sov .
j. nucl .
phys . *
42 * , 913 ( 1985 ) [ yad .
fiz . * 42 * , 1441 ( 1985 ) ] .
t. k. kuo and j. t. pantaleone , rev .
* 61 * , 937 ( 1989 ) .
s. hannestad , h. t. janka , g. g. raffelt and g. sigl , phys .
d * 62 * , 093021 ( 2000 ) [ astro - ph/9912242 ]
. l. stodolsky , phys .
d * 36 * , 2273 ( 1987 ) .
g. raffelt , g. sigl and l. stodolsky , phys .
lett . * 70 * , 2363 ( 1993 ) , ( e ) _ ibid . _ * 98 * , 069902 ( 2007 ) ] [ hep - ph/9209276 ] . g. sigl and g. raffelt , nucl
. phys .
b * 406 * , 423 ( 1993 ) .
g. g. raffelt , phys .
rept . * 198 * , 1 ( 1990 ) . | active - sterile neutrino mixing is strongly constrained for @xmath0 kev to avoid excessive energy losses from supernova cores . for smaller
@xmath1 , matter effects suppress the effective mixing angle except for a resonant range of energies where it is enhanced .
we study the case of @xmath2-@xmath3 mixing where a @xmath2-@xmath4 asymmetry builds up due to the strong excess of @xmath3 over @xmath5 emission or vice versa , reducing the overall emission rate . in the warm dark matter range
@xmath6 kev the mixing angle is essentially unconstrained . |
we consider the vorticity formulation of @xmath0 incompressible euler equation on the torus @xmath4 : @xmath5 where the velocity @xmath6 is determined from @xmath7 by the biot - savart law @xmath8 with the convention @xmath9 .
it is well known that is well - posed in @xmath10 for @xmath11 and in @xmath12 when the initial data is smooth .
if @xmath13 for @xmath14 then @xmath15 is lipschitz in @xmath16 and the following ordinary differential equation @xmath17 defines the area - preserving flow maps @xmath18 along which the vorticity is transported ; @xmath19 in this note , we will provide a simple proof of the _ strong illposedness _ of the euler equation in the critical sobolev space @xmath20 , which was obtained recently by bourgain and li @xcite .
the following result shows that so - called _ norm inflation _ occurs in @xmath2 for smooth initial data .
[ thm : growth ] for any @xmath21 , there exists an initial data @xmath22 and a time moment @xmath23 such that @xmath24 where @xmath25 is the ball of radius @xmath26 around the origin .
next , we show that a localized solution which is initially small in @xmath1 can immediately escape @xmath2 for @xmath27 , which have also appeared in @xcite : [ thm : pat ] for any @xmath28 such that @xmath29 and @xmath30 , there is @xmath31 which is @xmath32 away from the origin that for any @xmath33 , @xmath34 therefore , for @xmath30 and @xmath29 , there are initial data in @xmath35 which escapes not only @xmath35 but also @xmath2 for @xmath27
. actually our proof of theorem [ thm : pat ] can be modified a little bit to prove ill - posedness statements in @xmath35 for all values of @xmath36 : there exists @xmath31 whose @xmath35-norm becomes instantaneously infinite for @xmath27 . in the work of bourgain - li @xcite ,
the existence of localized initial data which escapes @xmath2 was obtained by carefully `` patching '' together an infinite sequence @xmath37 of @xmath32 data whose support becomes smaller but grows in @xmath2 with a larger rate in a shorter period of time as @xmath38 .
it is possible that the @xmath32 solutions that we construct in theorem [ thm : growth ] can be patched together to obtain the desired statement as well .
however , this seems to require a rather involved analysis , and we have chosen to establish theorem [ thm : pat ] via exhibiting a simple explicit initial data in @xmath1 , see .
the problem of well - posedness will not be an issue in the above statements as there is a unique , global - in - time solution of the extended system in @xmath39 ( so - called yudovich solutions ) for @xmath40 just in @xmath41 , even though in this case , @xmath15 is only log - lipschitz in general . a simple proof of this fact may be found in @xcite , for instance .
the space @xmath2 is called _ critical _ since we barely can not close the standard energy estimate @xmath42 as @xmath43 or even @xmath44 does not guarantee that @xmath6 is lipschitz .
this failure of lipschitz regularity is at the heart of the possibility of rapid growth of vorticity gradient .
it is explicit in the bahouri - chemin example @xcite : take @xmath45 on @xmath46 ^ 2 $ ] and extend it to @xmath47 ^ 2 $ ] as an odd function in both variables .
this defines a stationary solution of @xmath0 euler in the sense of equations , and the flow near the origin is `` hyperbolic '' in the following specific sense :
for @xmath48 small , it can be computed that for some absolute constant @xmath49 ( see denissov @xcite ) @xmath50 with some smooth functions @xmath51 .
certain perturbations of this stationary solution were utilized in the works of denissov @xcite , kiselev - verk @xcite , and zlato @xcite ( in chronological order ) to obtain growth of vorticity gradient in the maximum norm @xmath52 .
the growth rates of @xmath53 obtained in @xcite and @xcite were double exponential for arbitary long but finite time and exponential for all time , respectively . the groundbreaking work @xcite settled the possibility of double exponential growth of @xmath54 for all time , when the domain is a disc .
the `` key lemma '' of kiselev and verk ( which was also utilized in @xcite ) is an essential tool in our arguments as well ( see below lemma [ lem : zlatos ] ) . while our basic strategy to obtain growth of @xmath7 in @xmath20 is similar to that of the aforementioned works ,
there are a number of notable differences in our setting .
first , while the idea of `` linearizing '' around the bahouri - chemin stationary solution makes sense when considering only bounded vorticities , this solution does not belong to @xmath20 .
hence , we needed to consider a different type of `` background '' vorticity , and our choice was to take a suitably localized version of the following function : @xmath55 with @xmath56 .
the advantage of this initial vorticity is that it belongs to @xmath57 and the corresponding velocity @xmath58 satisfies @xmath59 .
second , since we want localized solutions , it is not clear if the specific hyperbolic picture of the type near the origin will be sustained , even for a very short periodic of time . in view of this ,
our strategy is to take an initial vorticity which extends over two different length scales @xmath60 and @xmath61 , and to show that vorticity outside the @xmath62-region , in the special time scale of @xmath63 , is sufficient to generate a hyperbolic flow which stretches the vorticity gradient on the @xmath62-region . here
, a caveat is that we could not exclude the possibility of our initial vorticity chunk getting `` squeezed '' in the angular direction even earlier than the scale @xmath63 , in which case we do not have a good lower bound on @xmath64 .
hence our actual proof is based on a contradiction argument .
this difficulty vanishes when the domain has a boundary : see remark [ rem : disc ]
. closing the introduction , let us mention that similar ill - posedness statements were recently established for the integer based @xmath65 spaces with @xmath66 of the velocity field @xmath6 , independently in the works of elgindi - masmoudi @xcite and bourgain - li @xcite .
we refer the interested readers to the works @xcite , @xcite , and @xcite for an extensive list of references on the problem of well - posedness of the euler equations .
let us use the notation @xmath67 for @xmath68 $ ] for simplicity .
we use letters @xmath69 to denote various absolute positive constants , and their values may change from line to line .
when a constant depends on some parameters , we explicitly indicate dependence as subscripts .
we use superscripts to refer to components of a vector : for example , @xmath70 and @xmath71 .
our initial vorticity @xmath40 will be odd both in the variables @xmath72 and @xmath73 .
since this symmetry persists for all time , we may view @xmath74 as defined just on @xmath46 ^ 2 $ ] . pick a large integer @xmath75 , and let us define the initial vorticity on @xmath46 ^ 2 $ ] as follows : @xmath76 where @xmath77 and @xmath78 are smooth bump functions . more specifically
, they satisfy @xmath79 \\ 0 & \mbox{for } r \notin \left[n^{-1}/2 , 2n^{-1/2}\right ] , \end{cases } \qquad \mbox{and } \qquad \psi(\theta ) : = \begin{cases } 1 & \mbox{for } \theta \in \left[\pi/4 , \pi/3\right ] \\ 0 & \mbox{for } \theta \notin \left[\pi/6 , 5\pi/12\right].\end{cases}\ ] ] since @xmath80 the main contribution of @xmath81 comes from the angular variation : @xmath82 as @xmath83 .
as mentioned in the introduction , we need to work with a special time scale . given @xmath84 and @xmath75 , we set @xmath85 and we shall track the evolution of initial data on the time interval @xmath86 $ ] . to get an idea of how this scale appears , recall that the main idea is to stretch vorticity in the @xmath62 region using the chunk of vorticity `` behind '' . since initially @xmath87 while @xmath82
, we need to stretch the @xmath2-norm in the local region by a factor of @xmath88 to obtain norm inflation . in view of @xmath89
, we achieve this goal once we sustain this lower bound on the velocity gradient during an interval of time @xmath86 $ ] .
it is important that in this time scale , fluid particles can move only up to a factor of @xmath90 , see and below .
our main technical tool is the following expression for the velocity due to kiselev and verk @xcite ; we use a version by zlato ( * ? ? ?
* lemma 2.1 ) which works in the case of the torus @xmath91 .
[ lem : zlatos ] let @xmath74 be odd in @xmath72 and @xmath73 . then for @xmath92
, we have @xmath93 with @xmath94 for @xmath95 .
there are several striking features of this lemma , which we would like to emphasize .
first , the expression essentially gives a _ pointwise _ control over the velocity gradient , just under the assumption that @xmath96 .
it is surprising that such a control is available , especially because the formula is applicable even in situations where @xmath97 is unbounded .
next , the integral in is _ monotone _ in @xmath74 , so that for the purpose of obtaining a lower bound on the velocity gradient , it suffices to find a region in space where vorticity is uniformly bounded from below . on the other hand
, one should note that lemma [ lem : zlatos ] is applicable only when the integral term in dominates the remainder term @xmath98 .
the following estimates are standard ( cf .
@xcite ) and will play a complementary rle of the previous lemma .
[ lem : standard ] let @xmath99 to be the solution triple for the @xmath0 euler equations in @xmath100 with initial data @xmath40 .
the velocity is log - lipschitz @xmath101 and the flow maps @xmath18 for @xmath102 satisfy quasi - lipschitz estimates of the form @xmath103 note that the argument of the logarithm in is always greater than 1 since @xmath104 in our torus @xmath105 .
although these estimates are well - known , we provide a proof of ( assuming the bound in ) , as it appears throughout the arguments given below . for simplicity , we set @xmath106 , and from the definition of flow we have @xmath107 and applying the estimate gives a bound @xmath108 which implies @xmath109 denoting @xmath110 and @xmath111 as the unique solution of the respective ode system @xmath112 on the time interval @xmath113 $ ] with initial data @xmath114 , we obtain the desired estimates as @xmath115 . , where the two rectangles represent the region where @xmath116 ( inner ) and @xmath117 ( outer ) .
the right figure shows possible evolution of the set @xmath117 under the flow .
the shaded region represents @xmath118.,title="fig:",height=226 ] , where the two rectangles represent the region where @xmath116 ( inner ) and @xmath117 ( outer ) .
the right figure shows possible evolution of the set @xmath117 under the flow .
the shaded region represents @xmath118.,title="fig:",height=226 ] given lemmas above , we present the proof of theorem [ thm : growth ] .
we will instead show the following statement : * claim .
* for any @xmath119 , there exists some @xmath120 depending only on @xmath121 such that for all @xmath122 , the solution associated with the initial data as in satisfies , with an absolute constant @xmath123 , @xmath124 once it is established , we simply use the scaling symmetry of the euler equation : given a solution @xmath125 and @xmath126 , @xmath127 is another solution with initial data @xmath128 , and we can pick @xmath129 to achieve the statements of the theorem . given @xmath119 , we fix @xmath130 , where @xmath131 is a constant depending only on @xmath121 to be defined below . in several places of the following argument , it is implicitly assumed that @xmath75 is sufficiently large with respect to @xmath121 and some absolute constants appearing in the proof .
consider the annulus @xmath132 . during the time interval
@xmath86 $ ] , particles starting from the arc @xmath133 remain in the region @xmath134 under the flow @xmath135 .
similarly , particles from @xmath136 can not escape @xmath137 .
both statements follow from applied with @xmath138 and @xmath139 ( where @xmath140 ) : we have @xmath141 and since @xmath142 for some @xmath143 , we obtain @xmath144 with constants @xmath145 uniform over @xmath146 . in particular
, it implies that any line segment @xmath147 should evolve in a way that it intersects each circle @xmath148 for @xmath149 .
take the domain @xmath150 and consider the region @xmath151 which is a curvilinear rectangle whose two opposite edges are bounded by @xmath152 ( see figure [ fig : rectangles ] ) .
note that @xmath153 on @xmath118 . for each @xmath154 $ ] ,
consider the closed set @xmath155 and let us denote its lebesgue measure by @xmath156 . to show that the @xmath2-norm grows , we are led to consider two different scenarios .
* case i*. assume that there exists a time moment @xmath157 such that for more than half ( with respect to the haar measure @xmath158 ) of @xmath154 $ ] , we have @xmath159 . if @xmath154 $ ] is such that @xmath159 , then we can definitely pick some @xmath160 such that the points @xmath161 and @xmath162 with @xmath163 satisfy @xmath164 and @xmath165 .
this implies a lower bound @xmath166 which in turn gives that @xmath167 with some absolute constant @xmath168 . we have established the * claim * in this case , recalling that @xmath169 .
* case ii*. for _ all _ @xmath170 $ ] , for at least half ( again with respect to the measure @xmath158 ) of @xmath154 $ ] , we have @xmath171 . in this scenario , we will track the evolution of the following segment @xmath172 for the time interval @xmath86 $ ] , where @xmath173 is an absolute constant to be determined below . since @xmath174 for all @xmath175 , to show growth of the @xmath2-norm of @xmath7 , it is enough to demonstrate that @xmath176 is close enough to the vertical segment ( where @xmath7 vanishes ) . in the remaining part of the proof
, we will always assume that @xmath177 and @xmath178 $ ] .
as a first step , we collect simple bounds on the trajectory of @xmath179 , which will in particular guarantee the applicability of lemma [ lem : zlatos ] . to begin with
, applying with @xmath180 gives @xmath181 ( recall that @xmath182 and @xmath183 are related by @xmath184 ) .
next , to obtain a lower bound on @xmath185 , we use the log - lipschitz estimate : @xmath186 and since @xmath187 proceeding exactly as in the proof of the estimate of lemma [ lem : standard ] gives that @xmath188 hence , for @xmath177 , we have @xmath189 for @xmath190 . on the other hand , with our assumption on @xmath191 , we estimate the integral appearing in lemma [ lem : zlatos ] at the point @xmath192 : @xmath193 and upon setting @xmath194 } } \int_i \sin\theta\cos\theta d\theta,\ ] ] we obtain @xmath195 for some @xmath196 , whenever @xmath197 .
therefore , we conclude that the @xmath198-term can be neglected in lemma [ lem : zlatos ] ( by possibly adjusting the value of @xmath199 in ) as long as we apply it to the trajectory of @xmath200
. that is , @xmath201 for @xmath177 and @xmath202 for some constant @xmath203 depending only on @xmath121 .
similarly , we deduce that @xmath204 on the same time interval for @xmath177 . from these bounds
, it follows that the curve @xmath205 is contained in the region @xmath206 the flow estimate further gives that @xmath207 intersects the circles @xmath208 and @xmath209 .
we could have taken @xmath210 so that @xmath211 ( where @xmath212 are constants from the estimate ) . then , for each @xmath213 , we may find a point ( in polar coordinates ) of the form @xmath214 on @xmath207 such that @xmath215 .
therefore , we deduce that @xmath216 and integrating over @xmath217 against @xmath158 with the choice @xmath218 gives @xmath219 which gives the desired lower bound in . [
rem : disc ] this construction carries over to the setting of the whole domain and a bounded open set , with minor modifications .
in the case when the fluid domain is a disc ( or more generally , a bounded open set with an axis of symmetry ) , we can utilize the boundary to achieve theorem [ thm : growth ] without relying on a contradiction argument . to be more specific ,
assume for simplicity that our domain is the upper half - plane @xmath220 .
take @xmath40 which is odd in @xmath72 and equals a smoothed out version of the indicator function on the polar rectangle @xmath221 \times [ 0,\pi/4]$ ] in the positive quadrant .
then it can be shown that for the time interval that we consider , we do not run out of angles ; i.e. * case i * does not happen .
the same can be said for the proof of theorem [ thm : pat ] , and actually one can even show continuous - in - time loss of regularity of the solution .
we expand on this point in our forthcoming work @xcite . inspecting the proof
, one can check that @xmath222 works , and so that we may choose @xmath223 as @xmath224 .
in other words , the initial data in grows at least by a multiple of @xmath225 ( in both scenarios ) .
again , when we have a boundary available , it is not necessary to introduce @xmath121 and we obtain growth by a factor of @xmath226 for any @xmath173 as long as @xmath75 is sufficiently large .
this time , we consider an odd initial vorticity defined on @xmath227 by @xmath228 where @xmath229 is the same angular bump function as in and @xmath230 is a smooth bump function which identically equals 1 for @xmath231 and vanishes for @xmath232 . clearly , @xmath40 is a bounded continuous function and by choosing @xmath21 small enough , we may assume that @xmath233 .
given @xmath234 and @xmath235 satisfying @xmath29 and @xmath30 , we can find a value of @xmath236 so that @xmath237 : note that @xmath238 so that given @xmath29 , @xmath239 it can be shown that the solution associated with the initial data remains @xmath32-smooth away from the origin for all time ( see proposition [ prop : smooth ] below ) . hence , if we denote the solution by @xmath74 , its @xmath2-norm can be unambiguously defined by @xmath240 which can take the value @xmath241 .
we will show that there exists a sequence of positive time moments @xmath242 and a sequence of radii @xmath243 , such that @xmath244 , @xmath245 , and for a fixed absolute constant @xmath246 , @xmath247 for each fixed @xmath248 , the function @xmath249 is continuous in time and provides a lower bound for @xmath250 .
therefore , the existence of sequences satisfying above gives the statement in theorem [ thm : pat ] .
the proof we present is strictly analogous to that of theorem [ thm : growth ] , as @xmath40 in can be viewed as a `` continuum '' version of data from our previous proof . to be more specific ,
pick some large number @xmath75 and radially truncate the function at length scales @xmath60 and @xmath61 .
then this is essentially a scalar multiple of the smooth initial data @xmath251 from the previous section , and recalling the scaling symmetry of the euler equation , it follows that this truncated initial data grows in @xmath2 by a factor which diverges with @xmath75 at some time moment @xmath252 which converges to 0 as @xmath253 .
therefore it is intuitively clear that the data would escape @xmath2 immediately .
given @xmath119 , we consider the time moment @xmath254 where @xmath255 is to be determined later .
it will be implicitly assumed that @xmath75 is sufficiently large with respect to @xmath121 and a few absolute constants . in particular , as @xmath256 , it guarantees that @xmath257 . throughout the proof
, it will be always assumed that the variable @xmath258 take values in the interval @xmath86 $ ] . the outline of the argument is as follows : we identify a `` bulk '' region which initially extends over length scales @xmath61 and @xmath259 , and a `` local '' region near @xmath60 . in the special time interval that we consider ,
if there is too much angular squeezing of the bulk , then we are done . otherwise , the bulk region has enough mass which stretches vorticity in the local region .
we note in advance that , compared to the situation of theorem [ thm : growth ] , we have less precise information on the local particle trajectories , so we should apply lemma [ lem : zlatos ] in a very careful manner . to begin with , using the basic estimate ( recall that @xmath260 ) , we take @xmath261 such that the fluid particles starting from @xmath262 at @xmath263 can not cross the circle @xmath264 within @xmath86 $ ] , as @xmath182 can be taken to be much smaller than a few absolute constants . similarly using the same estimate
, we can ensure that the particles starting on the circle @xmath265 can not escape the annulus @xmath266 in the same time interval .
indeed , taking the logarithms of ( assuming @xmath267 small enough ) , @xmath268 so that in the time interval that we consider , @xmath269 is equivalent to @xmath270 up to absolute constants which can be assumed arbitrarily close to 1 , uniformly in @xmath258 and @xmath271 . given these bounds , take the polar rectangle @xmath272 and consider intersections of the form @xmath273 note that on the `` angular '' sides of @xmath117 , @xmath40 takes the values @xmath274 and @xmath275 respectively , for some @xmath276 . for each @xmath178 $ ] and @xmath277 $ ] , we consider the ( non - empty ) set of angles @xmath278 we again consider two cases ; introducing the set of radii with `` enough '' angles @xmath279 : |i(t , r)| \ge m^{-1 } \left ( \ln \frac{1}{r } \right)^{-\alpha/3 } \ } \end{split}\ ] ] ( note the power @xmath280 ) and first , assume that there exists some @xmath281 such that @xmath282 } \left ( \ln \frac{1}{r } \right)^{-5\alpha/3 } \frac{dr}{r } \le c \left ( \ln n \right)^{1 - 5\alpha/3}. \end{split}\ ] ] in this case , we argue exactly as * case i * of the previous proof : whenever @xmath283 , we integrate over angle to get @xmath284 where we have used that when @xmath285 , @xmath286 for @xmath287 having the form @xmath288 in polar coordinates and similarly @xmath289 when @xmath290 , and that @xmath291 and @xmath292 are equivalent up to some absolute constants arbitrarily close to 1 ( relative to the difference between @xmath293 and @xmath294 ) .
integrating the above lower bound over @xmath283 gives the desired estimate @xmath295\backslash a(t_{cr } ) } \left ( \ln \frac{1}{r } \right)^{-5\alpha/3 } \frac{dr}{r } \ge cm ( \ln n ) ^{1 - 5\alpha/3 } .
\end{split}\ ] ] therefore , we may assume that for all @xmath170 $ ] , we have a lower bound @xmath296 under this hypothesis , we shall track the evolution of the diagonal segment in the `` local '' region : @xmath297 we may assume that the trajectories of the two endpoints of @xmath200 are trapped in the annuli @xmath298 and @xmath299 , respectively . assume for a moment that we have @xmath300 see figure [ fig : rectangles2 ] .
on the set @xmath176 , @xmath301 , and on the vertical line , @xmath302 .
therefore , for each @xmath303 , we have a lower bound @xmath304 where @xmath214 is a point on @xmath176 .
this gives @xmath305 and integrating over @xmath303 against @xmath158 , @xmath306 since @xmath307 , we obtain the desired lower bound . hence our goal now consists of establishing the containment in .
let us begin by obtaining a lower bound on the integral term appearing in lemma [ lem : zlatos ] .
we fix a reference point @xmath308 .
then , @xmath309 where we have again used the observation that @xmath310 and @xmath311 are comparable . now recalling that @xmath312 for @xmath313 we have an estimate @xmath314 which gives ( under our hypothesis on the size of @xmath315 ) @xmath316 we may now apply lemma [ lem : zlatos ] .
from now on , we reserve the letter @xmath16 for points in the diagonal segment @xmath200 . to begin with , on the diagonal , the error terms @xmath317 in lemma [ lem : zlatos ]
are bounded by an absolute constant , and since @xmath318 , we can ensure that the trajectory of @xmath16 stays in the region @xmath319 .
then in turn , this information guarantees that @xmath320 is positive , as @xmath321 is bounded by an absolute constant whenever @xmath319 , which gives in particular @xmath322 .
fix some @xmath177 , and assume for the sake of contradiction that @xmath323 is not contained in the region @xmath324 , for _ all _ @xmath197 .
it guarantees that @xmath325 and in particular the @xmath326 term is dominated by @xmath327 .
hence , @xmath328 and therefore integrating in time over @xmath86 $ ] with the choice @xmath329 ( for some absolute constant @xmath246 ) , we deduce @xmath330 .
combined with @xmath331 , @xmath332 this is a contradiction , so there must exist @xmath333 for which @xmath334 however , observe that for any point on the line @xmath335 , we have @xmath336 , so that the trajectory of @xmath16 for @xmath337 can not escape the region @xmath338 unless @xmath339 becomes larger than @xmath340 , which is impossible during the time interval @xmath86 $ ] .
this finishes the proof .
one does not face the restriction @xmath341 in the presence of a boundary .
for the convenience of the reader , we give a proof that @xmath74 in the case of theorem [ thm : pat ] actually stays @xmath32 away from the origin .
[ prop : smooth ] consider @xmath342 which is @xmath32 away from a closed set @xmath343 .
then , the unique solution @xmath344 of the 2d euler equation stays @xmath32 away from @xmath345 for all @xmath27 .
we may assume that @xmath346 .
once we show that @xmath74 is smooth away from @xmath345 for @xmath347 $ ] with some absolute constant @xmath348 then we may iterate the argument to extend the statement to any finite time moment .
take an open set @xmath349 which is separated from @xmath152 .
it suffices to show that for @xmath347 $ ] , there exists some @xmath350 that @xmath74 is uniformly @xmath351 in @xmath352 , for any integer @xmath353 and some open set @xmath354 .
we deduce this by inducting on @xmath355 . for the base case of @xmath356 , take some open set @xmath357 which is still separated away from @xmath152
. then we simply write @xmath358 which is bounded by an absolute constant via the hlder estimate once we choose @xmath359 where @xmath360 is the constant from .
now we assume that @xmath74 is @xmath351 in @xmath352 with some @xmath353 , where @xmath361 and @xmath362 .
we first pick some open set @xmath363 which satisfies @xmath364 in particular , @xmath363 is separated away from @xmath152 . for each @xmath365 , take a smooth cutoff function @xmath366 which equals 1 on @xmath367 and vanishes outside of @xmath352 .
then , for @xmath368 , with the biot - savart kernel @xmath210 , we write @xmath369 regarding the first term , a classical singular integral estimate gives @xmath370 the second term is indeed @xmath32 in @xmath367 simply because @xmath371 is @xmath32 away from the origin .
hence we deduce that @xmath15 is uniformly @xmath372 in @xmath367 . at this point
we may extend @xmath15 to be @xmath372 on the entire domain @xmath100 to obtain @xmath373 . then solving @xmath374
gives that @xmath375 is a @xmath372 flow , which coincides with @xmath323 whenever @xmath368 and @xmath376 .
this can be done by obtaining an a priori estimate for @xmath377 and then argue along a ( smooth ) sequence of approximate solutions .
note that @xmath378 so @xmath379 is invertible and the inverse function theorem gives that @xmath380 is also a @xmath372 diffeomorphism of the domain . from @xmath381
differentiating both sides @xmath382 times , on the right hand side we obtain terms which contains up to the @xmath382th derivatives of @xmath40 ( composed with @xmath383 ) multiplied with some factors of @xmath383 also up to the @xmath382th derivatives . since each such factor is @xmath384 ,
we conclude that @xmath74 is @xmath372 .
this finishes the proof .
in this section we discuss a few interesting open problems .
we have shown that there are yudovich solutions of the @xmath0 euler equations on @xmath100 which are initially in the class @xmath2 but which do not belong to the class @xmath385 for any @xmath386 .
one could ask whether even worse behavior is possible .
in fact , the existing estimates do not rule out the existence of yudovich solutions with @xmath2 data which leave @xmath387 for every @xmath388 in finite time .
[ losingestimate ] let @xmath389 be mean - zero .
then , the unique yudovich solution satisfies the following estimate : @xmath390 for every @xmath391 with @xmath392 solving the ode : @xmath393 for some large universal constant @xmath123 . by the john - nirenberg lemma , @xmath394 for some small universal constant @xmath360 . by a generalized young s inequality , @xmath395
now , by passing to the lagrangian formulation ( and suppressing the composition with the flow maps ) we see : @xmath396 hence , @xmath397 now upon integrating and using our inequality for @xmath398 , we have : @xmath399 finally , choosing @xmath400 we get @xmath401 the bound in lemma [ losingestimate ] is sharp , in the sense that there exist yudovich solutions which continuously lose regularity .
it seems that proving the conjecture is true ( for a short time ) is much more difficult on @xmath100 than on a domain with a boundary .
though there have recently been numerous results on ill - posedness for the euler equations in critical spaces , it seems as though the case of @xmath403 vorticity ( or @xmath404 velocity ) is still open .
consider the @xmath0 euler equations with partial viscosity on @xmath405 $ ] : @xmath406 notice that we have put viscosity only in the horizontal variable and , hence , we only need the no - slip boundary condition : @xmath407 on @xmath408 and @xmath409 .
when @xmath410 , we see that @xmath2 data can leave @xmath2 initially using a modification of the proof of theorem [ thm : pat ] .
in fact , it can be shown that @xmath411 becomes infinite . when @xmath412 this is no longer possible due to the energy equality : @xmath413 there a few regimes where one could study the behavior of the solutions of the partially viscous problem as @xmath414 .
the first regime is when @xmath415 . depending upon the relative sizes of @xmath416 and @xmath258 ,
different behaviours can be observed . in particular
, one would expect that if @xmath417 that we could see @xmath2 growth immediately . on the other hand , if @xmath418 we should nt see any growth .
determining the exact dynamics in this regime seems interesting . by the same token
, one could consider the inhomogenous problem and study the limit @xmath419 and @xmath414 .
we thank the anonymous referee for numerous suggestions and comments , which have significantly improved the quality of this article . | for the @xmath0 euler equation in vorticity formulation , we construct localized smooth solutions whose critical sobolev norms become large in a short period of time , and solutions which initially belong to @xmath1 but escapes @xmath2 immediately for @xmath3 .
our main observation is that a localized chunk of vorticity bounded in @xmath1 with odd - odd symmetry is able to generate a hyperbolic flow with large velocity gradient at least for a short period of time , which stretches the vorticity gradient . |
the lynds 1630 ( l1630 ) dark cloud is part of the orion b molecular cloud . at a distance of approximately 450 pc ,
it is one of the nearest giant molecular clouds .
its mass is estimated to be more than @xmath7 m@xmath8 ( maddalena et al . 1986 ) .
l1630 shows many signs of on - going star formation : ( _ a _ ) dense cores of molecular gas and high velocity outflows ( snell & edwards 1982 ; gibb & heaton 1993 ; gibb et al .
1995 ) ; ( _ b _ ) bright reflection nebulae ( e.g. , ngc 2068 and ngc 2071 ) ; ( _ c _ ) t tauri stars ( herbig & kuhi 1963 ; wiramihardja et al .
1989 ) ; ( _ d _ ) herbig - haro ( hh ) objects , including the morphologically complex hh2426 ( herbig 1974 ; solf 1987 ; mundt , ray , & raga 1991 ; eislffel & mundt 1997 ) ; ( _ e _ ) luminous far - infrared and submillimeter sources ( cohen et al . 1984 ; cohen & schwartz 1987 ; gibb & heaton 1993 ; schwartz , burton , & herrmann 1997 ; mitchell et al .
2001 ) ; ( _ f _ ) radio continuum sources ( bontemps , andr , & ward - thompson 1995 ; verdes - montenegro & ho 1996 ; reipurth et al .
2002 ) ; and ( _ g _ ) dense clusters of optically obscured near - infrared sources ( strom et al .
1975 ; strom , strom , & vrba 1976 ) .
the number density of young stars in l1630 , 450 pc@xmath9 , is comparable with that of the @xmath10 ophiuchus star - forming region ( lada et al .
1991 ) .
starting with the work of the stroms , l1630 has been studied in ever increasing detail at infrared wavelengths .
particular attention has been focused recently on its many outflows and hh features .
evidence for emission in the 2@xmath6 near - infrared lines of h@xmath11 from parsec - scale hh flows and jets has been presented by zealey et al .
( 1992 ) , and more extensively by davis et al .
( 1997 ) and chrysostomou et al .
images of hh24 in narrow emission lines at optical and infrared wavelengths trace out at least three intersecting , collimated , bipolar flows ( e.g. , mundt et al .
1991 : fig . 1719 ; davis et al . 1997 : fig .
the flows emanate from the vicinity of hh24b , at the location of the bright near - infrared sources ssv63e and ssv63w .
both objects are presumed from their steep spectral energy distributions to be class i protostars ( cohen et al .
1984 ) , that is , they are in a very early stage of evolution preceding the t tauri star ( tts ) or so - called class ii phase , and in the process of actively accreting material from their protostellar disks . the dynamical age of hh24 is of the order of @xmath12 yr ( eislffel & mundt 1997 ; snell & edwards 1982 ) , which implies that the class i objects powering those flows must be very young , and also in a very short - lived phase of evolution . in the past , the bright hh24a knot was interpreted as the `` working surface '' or bow shock for one of those outflows ( mundt et al . 1991 ; solf 1987 ) , but it is now recognized ( e.g. , eislffel & mundt 1997 ) to be the terminus of an independent outflow , which originates in a region immediately to the southwest , at the location of a deeply embedded class 0 protostar , hh24-mms . class 0 objects like hh24-mms represent the main accretion phase for protostars and are extremely rare because their lifetimes are so short .
hh24-mms is observed directly at near - infrared wavelengths only in shocked h@xmath11 ( bontemps , ward - thompson , & andr 1996 ; davis et al .
it was discovered as a bright source at submillimeter and radio wavelengths ( chini et al .
1993 ; ward - thompson et al . 1995 ; bontemps et al . 1995 ) .
greybody fits suggest a color temperature of 20 k and a total luminosity of @xmath13520 l@xmath8 , of which @xmath14 l@xmath8 originates from an unresolved and possibly disk - like central source ( ward - thompson et al .
the compact source is surrounded by a much more extended envelope , from which it is most likely accreting material at a very high rate ( chandler et al .
1995b ) . the detection of an h@xmath11 jet , blue - shifted co emission ( bontemps et al .
1996 ) , and indications of disk structure suggest that hh24-mms , even at its remarkably young age , has already begun to drive a powerful collimated outflow into the surrounding molecular cloud . various theories and mechanisms have been advanced in the past to explain such highly focused ( and frequently bipolar ) jets . in the class of magnetic models , for example
, it has been proposed that the outflow is confined and driven by interactions between the magnetosphere of the protostar and the primordial field that threads its accretion disk ( e.g. , shu et al .
1994 ) . south of hh24 , high resolution radio , millimeter , and
submillimeter maps have revealed a number of strong molecular peaks and bipolar outflows in the proximity of hh25 and hh26 ( gibb & heaton 1993 ; gibb et al .
1995 ; gibb & davis 1998 ) .
the surrounding areas have been extensively mapped in the @xmath15(1 ) line of h@xmath11 at 2@xmath6 ( davis et al .
1997 ; chrysostomou et al .
2000 ; davis et al .
a class 0 object , hh25-mms , was discovered @xmath16 south of the optically - visible knot hh25a as both a 1.3-mm and a 3.6-cm source ( bontemps et al .
strong 450 and 850@xmath6 continuum emission has been observed from the same location ( mitchell et al .
2001 ; phillips , gibb , & little 2000 ) .
the brightest near - infrared objects in the vicinity of hh2526 are ssv59 , ssv60 , and hh26-ir ( strom et al .
1976 ; davis et al .
none of the three has been shown conclusively to be the origin of a large - scale flow , although the weight of the evidence for hh26 ( in particular , the discovery of a compact h@xmath11 jet : davis et al .
2002 ) now points most convincingly to hh26-ir .
the latter , as well as ssv59 , is generally regarded as a class i protostar ( e.g. , bontemps et al . 1995 , davis et al . 2002 ) .
the status of ssv60 is unknown ; unlike the other two sources , it is associated with neither a submillimeter peak ( lis , menten , & zylka 1999 ; mitchell et al .
2001 ; phillips et al . 2001 ) nor a molecular core ( e.g. , torrelles et al . 1989 ; gibb & heaton 1993 ) . the same process that drives the outflows and jets of young stellar objects ( ysos ) is expected to result in the repeated shearing and reconnection of the magnetic field lines that attach such a star to its circumstellar disk , and thereby to generate a substantial amount of x - ray emission ( hayashi , shibata , & matsumoto 1996 ; shu et al .
1997 ; goodson , winglee , & bhm 1997 ) .
the numerical simulations of hayashi et al . and goodson et al .
in particular demonstrated that as a consequence of reconnection events some of the magnetically entrapped gas that is close to an accreting yso ( not necessarily its outflowing jet ) will reach temperatures as high as 100 mk and will be flung outwards , away from the star .
such high temperatures should give rise to hard x - ray spectra .
x - ray emission at equally high temperatures can also be produced in an accretion shock at the footpoints of the magnetic field , where infalling material from the disk streams down onto the surface of the forming star ( uchida & shibata 1984 ) .
the hh2426 region was observed in x - rays with asca for a net exposure time of 30 kiloseconds ( ks ) by ozawa et al .
the strongest source in the soft x - ray band ( 0.72.0 kev ) in that observation was ssv61 , a t tauri star ( tts ) which is embedded in an unusual patch of nebulosity situated mid - way on the sky between hh24 and hh2526 ( e.g. , scarott , gledhill , & warren - smith 1987 ) .
strom et al . (
1975 ) classified the star as an m2m4 giant from image - tube spectra .
a ccd spectrum of ssv61 , kindly made available to us by g. herbig , exhibits h@xmath0 emission and @xmath17 absorption equivalent widths that are typical of the class of weak - lined t tauri stars ( wtts ) , ew(h@xmath0 ) = @xmath18 and ew ( ) = 480 m ( cf .
walter 1986 ; martn et al .
the brightest emission in the hard asca band ( 2.08.0 kev ) is located north of ssv61 , in the vicinity of hh24 .
the high - energy peak lies nearest ssv63 , but can not be identified unambiguously with either of its infrared components due to the low spatial resolution of asca .
ozawa et al . noted the great difficulty of separating the x - ray flux of ssv63 from that of ssv61 , given their offset of just @xmath19 , at roughly the half - power radius of asca .
the quiescent luminosity @xmath20 ergs s@xmath3 they derive for ssv63 places it above the top end of the range thus far observed for all but a few class i objects , by factors of 50100 ( carkner , kozak , & feigelson 1998 ; preibisch , neuhuser , & stanke 1998 ) . whether ssv63 is truly one of the most x - ray luminous ysos in the sky , however , has awaited confirmation from high resolution imaging that resolves not only ssv61 from ssv63 , but also the individual infrared components of ssv63 itself .
that task , including the requisite high precision astrometry , has now been accomplished by the _ chandra _ observations reported here .
as for the class 0 sources in l1630 , ozawa et al . found no sign of x - ray emission at the position of hh24-mms , located in the saddle between the soft and hard x - ray peaks in their image , nor any emission at the position of hh25-mms , which was also in the asca field of view .
their @xmath21 upper limit of @xmath22 ergs s@xmath3 is not a stringent limit on the emission , however , since ysos elsewhere in orion and in other star forming regions have been detected at much fainter levels , @xmath23 ergs s@xmath3 ( e.g. , feigelson et al .
2002 ; imanishi , koyama , & tsuboi 2001 ) . using the _ chandra observatory _
, we have undertaken a far more sensitive search for the onset of x - ray emission in the class 0 objects within the l1630 cloud .
our objective was to establish an evolutionary progression in x - ray properties within a single cloud , and to examine the possible relationship between protostellar x rays , accretion disks , magnetically confined jets , and large - scale molecular outflows .
taking advantage of the lower background , greater sensitivity , and unsurpassed spatial resolution of _ chandra _ , we are now able to set a 10 times smaller bound on possible x - ray emission from the three established class 0 protostars in the l1630 cloud , hh24-mms , hh25-mms , and lbs17-h .
l1630 was observed with the advanced ccd imaging spectrometer array ( acis : garmire et al .
2002 ) on the _ chandra x - ray observatory _ for a net exposure time ( livetime ) of 62.8 ks .
the date of observation was 1415 november 2002 .
all six detector chips of the acis - s camera , s0s5 , were activated during the observation .
the aim point of the observation was chosen to place ssv61 close to the center of the back - illuminated s3 chip and to put hh24 and hh2526 on the same device within a few arcminutes of the optical axis of the telescope , thereby ensuring the best optical performance for our highest priority targets within the acis field of view .
the roll angle of the observation , @xmath24 , was selected to position the ngc 2068 reflection nebula , located to the northeast of hh2426 , on the back - illuminated s1 chip of the detector array and the class 0 object , ngc 2068/lbs17-h ( gibb & little 2000 ) , on the adjacent , front - illuminated s2 chip . the majority of the field of view is shown in fig .
1 , superposed on the red image of the digitized sky survey of the space telescope science institute ( dss ) .
the approximate locations of hh objects in the neighboring areas are marked , as are the positions of the two dozen x - ray sources we have detected in our _ chandra _ observation .
to analyze the _
image , we initially filtered the level-2 x - ray event file provided by the pipeline data processing system , retaining only the photon events that were assigned asca grades 0 , 2 , 3 , 4 , and 6 and that were within the broad ( 0.510 ) kev energy band in which the acis detector has its highest response and the background radiation is at its lowest levels .
subsequent steps in the analysis , including photometry and astrometry , were carried out following various data analysis threads within ciao , the interactive _ chandra _ data analysis package .
for source detection , we applied the celldetect sliding - cell detection algorithm .
twenty - five sources , all of them point - like , were found on the six chips . to extract x - ray counts for each source
, we used the dmextract tool .
we employed a circular extraction region with a radius of 10 pixels ( @xmath25 ) to measure the brightness of sources located within 7 of the optical axis of the telescope , a radius twice as large for sources more than 7 off - axis , and an annulus at least 10 times as large to evaluate the background emission .
the results of the photometry are given in table 1 .
listed there are the right ascension and declination ( j2000 ) of each source , its off - axis angle , the net source counts extracted within the 0.510 kev band , and the statistical errors in the source counts . in the final column
we tabulate the count rate for the effective exposure time of each source , which takes account of the variations in the sensitivity of the camera across the field of view , as indicated by our exposure map calculations .
the adjustments are negligible for the nearly on - axis sources on the s3 chip that are the main interest of the present study , but may be as great as 50% for the few objects that are located either near the edge of a detector chip or else far away from the optical axis of the telescope
. we also list possible optical and infrared counterparts , which were found by cross - matching the x - ray coordinates with entries in the simbad database and by visual comparison with red images from the dss as well as infrared images from the 2mass sky survey .
the identifications include several h@xmath0 emitters ( herbig & kuhi 1963 ) , named infrared sources ( strom et al . 1976 ) , anonymous near - infrared objects from the 2mass survey , and entries in the _ hubble space telescope _
guide star catalog - ii ( identified here with the prefix of s ) .
the correspondence of each x - ray and optical / infrared position is within 1 (= 2 acis pixels ) or less .
the position of x - ray source no .
4 coincides with the location of a previously known non - thermal , extragalactic radio source ( bontemps et al .
1995 ; anglada et al .
a search of the nasa / ipac extragalactic database ( ned ) turned up no other background objects in the acis field of view .
ssv63 is resolved into three separate x - ray components , which correspond to the vla 3.6-cm radio positions of ssv63e , 63w , and 63ne ( reipurth et al .
an overlay of the radio source positions on the 28 kev hard - band x - ray image from _ chandra _ is shown in fig .
2 . within the uncertainties in the absolute coordinates of the radio and infrared sources ( see table 1 of bontemps et al .
1995 ) , and the astrometric error in _
observations ( nominally 1 pixel , or @xmath26 , for positions close to the optical axis of the telescope ) , the x - ray centroids coincide with the infrared counterpart of ssv63e and the brighter , southern , infrared component of ssv63w .
the fainter infrared component of ssv63w , located 2 farther north ( see fig .
2 of davis et al .
2002 ) , appears dark in x rays .
ssv63ne is a source of both radio emission and x - ray emission , but has no near - infrared counterpart ( as we will show below in 2.2 ) .
if not a distant background object , it may qualify on that basis as an extreme class i or possible class 0 yso .
all three components of ssv63 have hard x - ray spectral indices : 96% of the x - ray photons detected from ssv63w have pulse - height energies greater than 2 kev , while all of those detected for ssv63e and ne have energies of 2 kev or more .
we performed individual @xmath27 timing analyses for the three x - ray components , using the xronos software package from the high energy astrophysics science archive research center ( heasarc ) of the nasa / goddard space flight center .
we found no indication of variability on timescales of 6 ks or longer , certainly nothing comparable in amplitude or duration to the order - of - magnitude outburst in x - ray flux recorded by asca ( ozawa et al .
the x - ray count rates listed in table 1 therefore likely correspond to the quiescent activity levels of those sources .
changes on timescales shorter than 6 ks were not investigated owing to the faintness of the sources .
the strongest x - ray source in the _ chandra _ field of view was the wtts , ssv61 , for which we accumulated very nearly 13,000 counts in the 0.510 kev bandpass .
the x - ray light curve of ssv61 was stable throughout the entire observation ; various @xmath27 tests performed with xronos found no evidence of variability on timescales greater than 300 s. to analyze the x - ray spectrum of ssv61 , we used the heasarc xspec spectral analysis package in order to fit simple multi - temperature thermal plasma models to the pulse height distribution of the extracted photons .
the quality of the fit was determined by @xmath27 minimization . in the calculations
, we removed a small ( @xmath28 ) contribution of emission from the background and also updated the instrumental response function to compensate for the on - orbit decline in the low - energy sensitivity of the acis detectors . in terms of the reduced @xmath27 value of the fit
, no satisfactory model was found .
typical of the results we obtained are the following parameters for a two - temperature , reduced - metal mekal fit ( mewe , kaastra , & liedahl 1995 ) , which we matched to the observed counts in the 0.57.0 kev pulse - height energy channels : reduced @xmath29 ; @xmath30 kev , @xmath31 kev ; one - third solar elemental abundance , z = @xmath32 ; hydrogen column density , n@xmath33 @xmath34 ; volume emission measure , @xmath35 , three - quarters of which are due to the high - temperature component ; broadband flux at earth , @xmath36(0.310 kev ) = @xmath37 ergs @xmath34 s@xmath3 ; and intrinsic ( absorption - corrected ) x - ray luminosity for an assumed distance of 450 pc , @xmath1(0.310 kev ) = @xmath38 ergs s@xmath3 . here , @xmath39(=254 ) is the number of degrees of freedom in the model .
closely similar fits were obtained with both the xspec pileup model ( see davis 2001 ) , which we used to account for the @xmath40% pileup fraction in our spectrum of ssv61 , and also the apec plasma emission code described by smith et al .
an example of an apec model fit is shown in fig .
3 . except for a few possible camera artifacts ,
the largest residuals in all the model fits are attributable to excess emission in the observed spectrum near photon energies of 1 kev .
the apparent excess may be due to lines of ne ix and x or to inner shell transitions of high ionization stages of fe .
a similar excess has also been observed in the acis spectra of pre - main sequence stars in other star - forming regions , e.g. , orion and @xmath10 ophiuchus ( feigelson et al .
2002 ; imanishi , tsujimoto , & koyama 2002 ) .
an evaluation of the intrinsic 0.310 kev brightness of ssv61 for a variety of xspec models produced very nearly the same results for the x - ray flux and for the luminosity .
the various models also yielded closely similar amounts for the hydrogen column density . from the models and the empirical relation of bohlin , savage , & drake ( 1978 ) , we infer a visual extinction of @xmath41 mag .
our estimate , drawn from the _ chandra _ spectrum of ssv61 , is a factor of four less than the extinction ozawa et al .
( 1999 ) derived from their asca observation , but is consistent with the conclusion reached by strom et al .
( 1975 ) on the basis of infrared colors that the foreground reddening to this star is relatively small .
infrared spectroscopy ( discussed below ) further confirms the low extinction for this star .
the three x - ray sources identified with ssv63 were too faint for detailed spectral model fitting , and so we limited ourselves to a few simple models of the strongest component , ssv63w .
for that source we chose an isothermal , one temperature ( 1-t ) mekal model , constrained it to a sub - solar abundance of z = 0.5 , and allowed xspec to search for values of n@xmath42 and @xmath43 that minimized the @xmath27 of the fit .
the best - fit values derived by xspec were n@xmath44 @xmath34 ( equivalent to 48 mag of visible extinction ) and @xmath45 kev .
the resulting hydrogen column density is at the low end of the range of values deduced by ozawa et al .
( 1999 ) , while our temperature is a factor of 3 lower than their best - fit value for a quiescent interlude in the asca observation . a reddened bremsstrahlung model , which we chose as an alternative fit , produced a similar @xmath27 , temperature , and n@xmath42 as the mekal model . both models , the mekal and the bremsstrahlung , yielded a comparable 0.310 kev flux and absorption - corrected luminosity for ssv63w : @xmath46 ergs @xmath34 s@xmath3 and @xmath47 ergs s@xmath3 . raising the temperature to a fixed value of @xmath48 kev for consistency with the asca result reduced the column density to @xmath49 @xmath34 , corresponding to an extinction of @xmath50 mag , but yielded a much poorer fit to the _ chandra _ pulse - height spectrum in terms of its @xmath27 value .
no model analysis was possible for the two remaining components of ssv63 , given the small numbers of counts recorded for them during our _ chandra _ observation .
to estimate x - ray luminosities for those sources , we scaled from ssv63w in proportion to the background - adjusted number of counts detected for each source . in doing so , we implicitly assumed that the optimal xspec models of ssv63e and ssv63ne would yield the same parameters as the best - fitting model for ssv63w .
our normalization procedure is equivalent to using a counts - to - energy conversion factor ( ecf ) of @xmath51 ergs @xmath34 s@xmath3 per count s@xmath3 for the x - ray flux observed at earth , and @xmath52 ergs s@xmath3 per count s@xmath3 for the extinction - corrected broadband luminosity .
anticipating the results of 3.2 , we note that sub - millimeter observations of l1630 corroborate the high extinction value provided by our x - ray analysis of ssv63w . on the other hand , spectroscopy of the infrared ice band in the direction of ssv63e and w ( discussed in 2.3 ) suggests that the foreground extinction towards both sources may be only half as large , @xmath53 mags .
the apparent discrepancy raises the possibility that the x - ray sources may be more deeply obscured than are the near - infrared sources , reflecting , perhaps , an anomalous gas - to - dust ratio along the line of sight .
the x - ray luminosities for ssv61 and ssv63 are summarized in table 2 .
our @xmath1 value for ssv61 is 15% higher than the asca result of ozawa et al .
( 1999 ) but lies substantially above the high end of the range found in an extensive survey of wtts in taurus by rosat ( neuhuser 1997 ) and is well above the range of x - ray luminosities measured by _
chandra _ for low - mass stars in both ophiuchus ( imanishi et al .
2001 ) and the orion nebula cluster ( feigelson et al . 2003 ) .
our @xmath1 value for ssv63w falls near the top end of those three luminosity distributions , while the values for the two other detected class i ysos , ssv63e and 63ne , are in the top quartile of each range .
also listed in the table are upper limits on the x - ray emission of the near - infrared objects ssv60 and 59 , the former lying between hh25c and 25d along the curving ribbon of emission knots that define hh25 ( being closer to d ) , the latter lying just north of hh26a .
the upper limits on the count rates given in the table are three times the poisson uncertainties in the total counts ( uncorrected for background emission ) measured through a 10 diameter circular detection cell , which we centered on the respective 2mass positions listed in the second and third columns of the table .
based on eqn . 10 in gehrels ( 1986 )
, the significance level of these upper limits corresponds to @xmath4 , i.e. , to a confidence level of 0.95 .
the upper limits obtained for a smaller 5 diameter detection cell , which still encircles more than 95% of the energy in the _ chandra _ point spread function , are @xmath54 dex lower at a confidence level of 0.95 , in accordance with either gehrels ( 1986 ) or the bayesian approach of kraft , burrows , & nousek ( 1991 ) , but may be subject to a somewhat greater uncertainty due to the imprecision of the infrared coordinates of both objects . for lack of a more compelling choice , the conversion from count rate to luminosity in table 2 is based on the ecf derived from our xspec modelling of ssv63w .
the upper limits cited on the luminosity assume both objects lie at a distance of 450 pc within the l1630 cloud , which may not be true for ssv60 , as we discuss further in 2.3 . in similar fashion
, we derived @xmath4 upper limits on the possible x - ray emission from the three class 0 objects in the l1630 cloud , hh24-mms , hh25-mms , and lbs 17-h .
those results are also given in table 2 .
in each case , the detection cell was centered at the known long - wavelength or radio position of the object .
the j2000 coordinates of hh24-mms and hh25-mms were precessed from the b1950 vla positions of bontemps et al .
( 1995 ) and gibb ( 1999 ) ; those of lbs 17-h were precessed from the b1950 location of the peak in the 450@xmath6 dust continuum map of phillips et al .
( 2001 ) . finally , we list an upper limit for our x - ray non - detection of hh26-ir , the possible driving source for the hh26 outflow ( schwartz et al .
1997 , wherein the object is known as irs 4 ) .
the b1950 location for that source was taken from the discovery paper of davis et al .
( 1997 ) and precessed to an equinox of j2000 .
two sets of _ jhk _ near - infrared images were obtained of the hh2426 region on 18 november 2002 , using the university of hawaii quirc hgcdte camera array ( hodapp et al .
the instrument was mounted at the f/10 cassegrain focus of the 2.2-m telescope on mauna kea .
the image scale was 019 pixel@xmath3 , the field of view was 32 .
one set of images was centered on ssv63 and hh24 , the other mid - way between ssv59 and 60 in the vicinity of hh2526 . at both locations ,
a sequence of @xmath55 s exposures was obtained at each of 20 separate positions , offset by 10 in right ascension and declination , to produce a heavily sampled dither - pattern .
the same cycle was repeated in each of the three filter bands .
we alternated the observations of both hh fields with similar integrations on a pair of flanking fields 150 e and w , from which the bright stars were subsequently removed in order to create suitable sky frames .
flat - field frames were created by median - averaging observations of the 2.2-m dome under illumination by incandescent lamps . in the reductions ,
the dithered hh frames were flat - fielded and sky - subtracted , then aligned by cross - matching the positions of bright stars to form an enlarged median - averaged mosaic .
the cumulative exposure time in the center of the final co - added image in each filter is @xmath56 s. taking into account the intermittent , non - photometric observing conditions , we estimate brightness limits of 20.9 mags , 20.1 mags , and 18.5 mags in @xmath57 , @xmath58 , and @xmath59 , respectively .
the typical seeing in the course of the observations was 05 in @xmath59 ( fwhm ) .
however , the psf of stellar images in the co - added data proved to be somewhat broader than this , and also slightly distorted , as a result of a misalignment of the primary mirror of the 2.2-m telescope .
our images of hh24 in the three colors are shown in fig .
the positions of the individual components of ssv63 , as listed in table 2 , are marked on the @xmath57-band image .
we have also identified the expected locations of the two most prominent hh emission knots , 24a and 24b .
the position from table 2 of the class 0 object , hh24-mms , is also noted . that source is faintly visible in the @xmath59 image ,
the bandpass of which includes a number of h@xmath11 lines , and is also very weakly present in the @xmath58 image , whose bandpass includes [ ] @xmath60 , a low - excitation emission line that is observed in the jets of some hh objects ( e.g. , reipurth et al .
as noted earlier , the elongated jet associated with hh24-mms is very prominent in the continuum - subtracted , narrow - band h@xmath11 images of bontemps et al .
( 1996 , their fig .
1a ) and davis et al .
( 1997 , their fig .
2 ) , and also in the fabry - perot image of davis et al .
( 2002 , their fig .
the n s components of ssv63w are cleanly separated in our @xmath57-band image , but are heavily exposed in the longer - wavelength images , as is true of ssv63e .
an extended halo around the latter component , pointing westward , becomes increasingly evident with increasing wavelength .
contrary to previous reports ( moneti & reipurth 1995 ; terquem et al .
1999 ) , we see no evidence to suggest this star is double or that the emission peak in @xmath58 lies 1 west of the peak in @xmath59 . in our images , there is no shift in the position of the peak with color to a limit of 1 pixel or 02 .
the coordinates of the radio and x - ray source ssv63ne place it adjacent to two very red patches of nebulosity known from earlier work ( e.g. , zealey et al .
1992 ) . a false - color image that combines the three separate panels of fig .
4 exhibits a noticeable gradient in color from w to e , possibly originating at ssv63e .
the variation in color suggests that both patches of infrared nebulosity may be illuminated by an embedded source to the west , either the class i source , ssv63e , which scarott et al .
( 1987 ) proposed as the illuminating source for the diffuse optical nebulosity of hh24 , or else a much more deeply embedded infrared object at the location of 63ne .
the radio and x - ray coordinates of ssv63ne coincide to within 33 with the position for the predicted but unobserved `` jet g star '' of jones et al .
( 1987 ) , which those authors proposed to be the driving source of the hh24 g outflow ( see also mundt et al .
1991 , in particular the highly detailed [ ] image in their fig .
the absence in fig .
4 of any obvious point source at the same location to a limit of @xmath61 mag implies that this star , presuming it does exist , lies exceedingly deep within the molecular cloud .
given our detection of x - ray emission at the radio position , there can be no doubt that ssv63ne , if it is a protostar of class 0 , has already evolved past the stage of isothermal collapse .
submillimeter maps of the region obtained with the scuba bolometer array on the jcmt ( phillips et al .
2001 ; mitchell et al .
2001 ) show extended emission at the position of ssv63 but are unable to resolve 63ne from the two other radio components , 63e and 63w .
consequently , although it seems quite likely that ssv63ne is a bona fide yso , its exact status is not yet entirely settled .
our infrared images of the southern region containing hh25 and 26 are presented in fig .
5 . again , for orientation , the positions of the notable objects in the field of view are marked .
the class 0 protostar , hh25-mms , is invisible in both the @xmath57 and @xmath58 images , although a hint of emission ( presumably from ro - vibrational lines of h@xmath11 ) can be seen at the radio position of this object in the @xmath59-band image .
hh25a and ssv59 are both immersed in nebulosity .
nebulae of similar appearance have been seen in high resolution images of other embedded protostars , e.g. , hl tau ( close et al . 1997 , especially their figs . 2 and 4 ) and interpreted as the outline of bipolar cavities .
ssv59 has been classified as a class i object .
no star has been observed before at the location of hh25a at optical wavelengths . however
, this hh knot is uncommonly bright in the @xmath57 and @xmath58 frames in fig . 5 to be merely the `` working surface '' of a high - velocity outflow from hh25-mms , located immediately to the south .
our underexposed 1.92.5@xmath6 infrared spectrum of hh25a displays an array of very strong emission lines of h@xmath11 , which likely arise from shock - heated gas , but no spectral signatures of an underlying star .
a very red , compact emission knot is observed at the location of hh25c , in approximate alignment with hh25a , ssv60 , and hh25d . as mentioned earlier , we believe ssv60 is a heavily reddened background giant , not a member of l1630 .
no other plausible candidate for the exciting source of hh25 is revealed by our images .
infrared spectra of ssv59 , 60 , 61 , 63e , 63w ( both the north and south components combined ) , and hh26-ir were acquired with the spex cross - dispersed , medium - resolution spectrograph ( rayner et al . 2003 ) at the 3.0-m nasa irtf telescope on mauna kea on 3 january and 13 february 2003 . spex covers the full wavelength range from 0.8 to 5.5@xmath6 in two grating settings .
we observed each source in the 1.94.2@xmath6 wavelength region , using the lxd setting ; the two brightest objects , ssv60 and 61 , were also observed in the 0.82.4@xmath6 region with the sxd setting .
observing conditions were photometric and dry throughout , with t ypical seeing values near 05 fwhm ( measured in @xmath59-band ) .
slit widths of 03 and 05 at the short- and long - wavelength settings provided a nominal spectral resolving power , @xmath62 , of 2000 and 1500 , respectively .
our on - source integration times ranged from 300 to 1200s and resulted in signal - to - noise ratios of 2050 in both the @xmath59 and @xmath63 bands .
each observing sequence began with a nearby ( less than 0.05 difference in airmass ) a0v standard star , moved to the target object , and finished with a set of internal calibration spectra . to cancel the sky background , we observed in the conventional way by nodding along the 15-long slit .
the data were reduced with the spextool software package ( see cushing et al .
2003 for details ) .
spectra were extracted using a 12 aperture along the slit , corrected for telluric features , and calibrated in flux by means of the standard star spectra according to the techniques that have been described by vacca , cushing , & rayner ( 2003 ) .
the processed spectra are shown in fig .
6 , where they are plotted on a logarithmic flux scale .
the infrared continua of ssv60 and 61 come to a peak blueward of the @xmath59-band , near 1.9 and 1.5@xmath6 , respectively .
absorption lines from a variety of neutral and ionized metals , including , , , , , and , can be seen in the spectra of both stars .
no emission lines are observed in the spectrum of ssv60
. the br@xmath0 line of at 4.05@xmath6 appears in emission in the spectrum of ssv61 ; its equivalent width is entirely consistent with the observed strength of h@xmath0 , case b recombination , and very little foreground reddening .
both spectra exhibit deep absorption bands from the first overtone ro - vibrational transitions of co near 2.32.5@xmath6 . assuming the observed atomic and molecular features are photospheric in origin , we have used their relative strengths to estimate the spectral type and luminosity class of the underlying stars in the manner described by kleinmann & hall ( 1986 ) , hodapp & deane ( 1993 ) , and greene & meyer ( 1995 ) .
here we have adopted aspin s ( 2003 ) formulation , in which an atomic index formed by the sum of the equivalent widths of the doublet at 2.21@xmath6 and the triplet at 2.26@xmath6 is compared to a molecular index represented by the equivalent width of the @xmath64=20 co bandhead at 2.29@xmath6 ( see fig .
according to their atomic and molecular index values , ssv60 ( 4.2 , 10.0 ) is definitely a giant , whereas ssv61 ( 5.9 , 9.4 ) is intermediate to a dwarf and giant ( a not uncommon finding for a tts ) .
we have combined the spectral indices , the strengths of the various photospheric absorption lines of the metals , the deep co bandheads , and the general continuum shapes to estimate spectral classifications of k5m0iii for ssv60 and m0m2iii / v for ssv61 .
those spectral types , resting solely on the infrared spectra , are consistent with the locations of both stars in a ( @xmath65 , @xmath66 ) color - color plot based on their 2mass photometry , and also with the effective temperatures we have derived following the procedure of greene & meyer ( 1995 ) , which makes use not only of the absorption line strengths but the broadband colors as well .
the latter technique provides , among other things , an estimate of the fraction of excess light from veiling continuum emission , which we find to be constrained to low values for both stars ( @xmath67 , @xmath68 at @xmath58 and @xmath59 ) .
the association of ssv61 with the l1630 molecular cloud is hardly in question , but the same can not be said for ssv60 . with its bright , attendant nebula , ssv61 would be a rare star indeed to be in the far foreground of this region , while the absence of any ice absorption at 3.08@xmath6 in its infrared spectrum ( fig .
6 ) argues against the likelihood it lies far behind the molecular cloud at a very considerable distance from earth . for an early m giant ,
ssv60 is clearly much more heavily absorbed than ssv61 , as is apparent from its spectral energy distribution shortward of 2@xmath6 and its strong ice absorption band .
the broad ice feature is common to the spectra of other sources which are either background to or deeply embedded within molecular clouds , where the ice has condensed onto dust grains but not undergone significant processing at high temperature ( smith , sellgren , & tokunaga 1989 ) . for ssv60 , the optical depth at band center , @xmath69 = 0.96 , leads ( according to the results of whittet et al .
1988 ) to an optical extinction of roughly @xmath70 mag .
such heavy extinction is not to be found in the foreground to l1630 . on the other hand
, ssv60 shows none of the signposts of youth or activity that are expected for an embedded class i protostar no x - ray emission , no high velocity outflow or jet in co or h@xmath11 , no far infrared or sub - mm emission .
we conclude , therefore , that ssv60 is a background giant star that is unconnected either with l1630 or with hh25 , on which it is apparently coincidentally projected .
the strong red continua and deep ice bands of hh26-ir , ssv59 , 63e , and 63w , on the other hand , suggest that each of those objects is deeply embedded within the l1630 cloud .
their observed ice - band optical depths range from @xmath71 ( for hh26-ir ) to @xmath72 ( for ssv63e and w ) , corresponding to approximate @xmath73 values of 10 to 20 mags . at close inspection ,
the ice band spectra of ssv59 and 63e also show a broad , shallow depression near 3.45@xmath6 .
that feature , according to smith et al .
( 1989 ) , can be identified with the c
h stretching modes of hydrocarbons ( in the ch@xmath11 and ch@xmath74 subgroups ) .
emission lines of are visible in the spectra of all four objects , including br@xmath0 4.05@xmath6 , br@xmath75 2.17@xmath6 , and ( in the case of ssv63e ) @xmath76=85
pf@xmath75 3.74@xmath6 .
the emission is a clear indication of ionized gas in the close vicinity of each star , but whether the warm gas is inflowing or outflowing is a matter of continuing debate ( e.g. , najita , carr , & tokunaga 1996 ; folha & emerson 2001 ) . there are no signs in our spectra of any photospheric absorption lines , which we suspect may be hidden beneath a large continuum veiling flux .
emission lines from ro - vibrational transitions of h@xmath11 are present throughout the @xmath59-band spectra of ssv63w and hh26-ir , the 2.12@xmath6 @xmath77(1 ) line being the strongest ( fig .
the same lines are often observed in the spectra of class i objects ( greene & lada 1996 ) .
they are also commonly observed in the spectra of high velocity outflows ( e.g. , gredel 1994 , 1996 ) and have been attributed to excited h@xmath11 molecules that form behind shocks which arise as jet material collides with the ambient gas in a cloud ( garden , russell , & burton 1990 ) .
the h@xmath11 emission in our spectral images does not appear to be spatially extended in a n s direction along the slit .
however , it has already been shown by davis et al .
( 2002 ) that the h@xmath11 emission of hh26-ir has a modest extension along its jet in the ne
sw direction .
the @xmath59-band spectra of ssv59 , 63e , and hh26-ir ( fig .
8) display the first overtone ro - vibrational co bands in emission .
various explanations have been offered for the co lines : they may be formed in the innermost portion of a keplerian disk or in the neutral component of a stellar wind or outflow ( carr 1989 ; chandler , carlstrom , & scoville 1995a ; najita et al .
alternatively , the emission may be produced by accretion of disk material onto the central star via a magnetic `` funnel flow '' ( martin 1997 ; see , however , najita et al .
2003 ) . in any case , biscaya et al .
( 1997 ) have demonstrated that in some cases the co bands can be highly variable , perhaps even exhibiting periodic behavior .
carr ( 1989 ) , in fact , observed the co bands of ssv59 in absorption , whereas they are definitely in emission in our spex spectra , as can be readily seen from fig .
we detected x rays from none of the three class 0 ysos within the _ chandra _ field of view .
two of those embedded objects , hh24-mms and hh25-mms , are continuum radio sources and therefore must have a component of ionized gas ; the third object , lbs17-h , has not been detected at radio wavelengths and hence may be in an earlier stage of evolution than the others , or else its radio emission is much more optically thick and therefore below current detection levels ( gibb 1999 ) .
the radio spectral indices of 24-mms and 25-mms are slightly positive ( anglada et al .
the indices are compatible with thermal free - free radiation ( reynolds 1986 ) , but do not definitively rule out a possible non - thermal gyrosynchrotron origin .
we note that the spectral index for gyrosynchrotron emission is most often strongly negative but it can also be weakly positive , as for a thermal source ( e.g. , gdel 2002 ) . in the case of hh24-mms , the radio source is spatially resolved ( bontemps et al .
1996 ; reipurth et al .
therefore , it is almost certainly free - free emission from a warm ionized jet , although some portion of it may be thermal emission from dust ( ward - thompson et al .
1995 ) .
we can suggest several plausible explanations to account for the dearth of x rays from the class 0 objects in l1630 : ( a ) assuming the three ysos are in the main phase of protostellar accretion , they may not yet have developed accretion shocks that are strong enough to heat infalling gas to the million degree temperatures that are needed to generate x rays , or else their magnetic fields may not be strong enough or sufficiently well organized on large scales to produce intense reconnection heating of the infalling gas ; ( b ) alternatively , if hot gas is now present around the star , the temperature of the gas may be sufficiently low ( below 12 mk ) that it emits only a very soft spectrum of x - ray photons , which is unable to penetrate the dense envelope that surrounds the growing protostar ; or ( c ) if there is x - ray emitting gas now present at temperatures that are high enough to form hard x rays with energies of 210 kev , the extinction in the surrounding envelope and cloud may be sufficiently large to absorb most ( or all ) of the x - ray photons and to depress the emergent radiation below levels that _ chandra _ can detect . here
we address the last possibility .
since , by definition , class 0 protostars are especially faint at near - infrared wavelengths , we do not have @xmath78 colors or irtf spectra for the class 0 ysos in l1630 from which we might estimate the foreground extinction to those objects . instead
, we have estimated the line - of - sight extinction to each source by making use of the relationship between visible extinction and 850@xmath6 brightness given by mitchell et al .
( 2001 ) , @xmath79 .
applying that expression in turn to the submillimeter data for hh24-mms , hh25-mms , and lbs17-h , we find @xmath73 estimates of 200 , 100 , and 86 mags , respectively . in each case , the derived extinction is larger than the value of 48 mags we previously adopted from our xspec model of the _ chandra _ spectrum of ssv63w . given the foregoing extinction estimates , we have used the heasarc webpimms tool to predict a _ chandra _ detection limit for each class 0 yso , computing for each one the intrinsic x - ray luminosity that yields the observed @xmath4 count rate limit listed in table 2 . in our simulations , we have adopted a single - temperature thermal model for the x - ray emission and assumed a normal value of the gas - to - dust ratio in order to convert the submillimeter - based @xmath73 into a column density of n@xmath42 , following bohlin et al .
guided by our xspec calculations , we have considered two different source temperatures which should bracket an adequate range of spectral hardness , @xmath48 kev and @xmath80 kev . in the high - temperature case , the tabulated count rate limits for hh24-mms , hh25-mms , and lbs17-h correspond to @xmath4 detection limits in @xmath81 of 30.30 , 30.00 , and 29.75 , respectively .
the limits in @xmath81 for the low - temperature models are 0.3 dex higher . for both temperatures ,
the predicted luminosity limits are far less than the intrinsic @xmath1 we derived from xspec models of the class i object ssv63w , which we detected at @xmath82 .
to have escaped detection at that @xmath1 , i.e. , to have resulted in fewer than @xmath83 net x - ray counts in our exposure time of 63 ks , the class 0 objects would each have to suffer more than 400 mags of extinction ( or else be intrinsically very much softer x - ray sources than we have assumed ) .
therefore , to a very high significance level , it appears all three class 0 objects are intrinsically much fainter in x rays than 63w , even when the latter source is quiescent rather than flaring .
the x - ray luminosity of the eastern component of ssv63 , the class i yso 63e , which we also detected with _
, is comparable in strength to the luminosity upper limits for the class 0 objects .
thus , we can further assert that the class 0 sources are fainter than 63e at the @xmath4 confidence level .
on the other hand , because the two other established class i objects in the same observation , ssv59 and hh26-ir , were not detected by _
chandra _ , the x - ray brightness of individual class i sources must have a wide range of values , even within the same star forming region . _ chandra _ observations of class i ysos elsewhere in orion ( tsujimoto et al .
2002 ) and in ophiuchus ( imanishi et al .
2001 ) bear that point out .
thus , although the class 0 objects in l1630 are clearly fainter than _ some _ of the class i objects in the same cloud , they may not be fainter than _ all _ of them . in terms of an evolutionary trend , until more is known about the lower range of @xmath1 values for class i objects , and the extent to which such objects are variable , it is premature to draw any firm conclusions about whether the class 0 sources as a group are systematically less luminous in x rays than are the more advanced class i sources which are their descendants . the submillimeter map of mitchell et al .
( 2001 ) leads to an @xmath73 estimate ( 47 mags ) for the combined ( spatially unresolved ) ssv63ew that is entirely in agreement with the extinction we have derived from our xspec modelling of the _ chandra _ spectrum of 63w .
both the e and w infrared components of ssv63 were detected in x rays ; they are also both thermal radio emitters ( anglada et al .
hh 26-ir was not detected in x rays , but may be a weak radio source according to gibb ( 1999 ) .
ssv59 is neither an x - ray source nor a radio source .
however , the near - infrared spectrum of ssv59 , as well as that of hh26-ir ( fig .
6 ) , exhibits emission in the brackett lines of hydrogen , implying that both objects are associated with ionized gas .
the presence of a resolved h@xmath11 jet appended to hh 26-ir ( davis et al .
2002 ) suggests that its radio emission may originate from the jet and be thermal in nature ; the radio spectrum of ssv59 may be optically thick and thus undetectable .
it is perhaps significant that the hh filaments and knots identified with the northern class i ysos ( the various alphabetical components of hh24 ) are much more highly developed than the ones in the south ( hh2525 ) .
this is perhaps an indication that the southern region is relatively younger and has not yet reached the age where its embedded ysos have formed strong radio sources or possess the million degree plasma required to produce x rays .
the weak - line t tauri star ssv61 is by far the brightest source in our _ chandra _ image .
it is also a highly variable radio source , which suggests a non - thermal origin for its radio flux ( anglada et al .
1998 ; gibb 1999 ) . as we noted in 2.3
, there is very little chance of our having greatly overestimated the distance and consequently the @xmath1 of this star . for an assumed distance of @xmath84 pc ,
the radio and x - ray luminosities of ssv61 closely fit the @xmath85 relation for normal late - type giants , lying among the tidally locked binaries at the top end of the activity range ( drake , simon , & linsky 1989 ; dempsey et al .
this may be an indication that ssv61 is itself a close binary . in any event , it is safe to say that the emissions of this star at radio as well as x - ray wavelengths are coronal in origin .
the near - infrared spectrum and 2mass colors of ssv61 in a ( @xmath65 , @xmath66 ) two - color plot are consistent with those of a lightly reddened and lightly veiled early m giant star , for which @xmath86 mags .
however , the observed infrared magnitudes , combined with the bolometric correction for an early m giant ( @xmath87 , b.c.@xmath88 = 2.60 mags ) , imply a very high luminosity , @xmath89 , which places the star well above the stellar birthline at a logarithmic effective temperature of @xmath90 = 3.53.6 ( stahler 1983 ; mercer - smith , cameron , & epstein 1984 ) . at that @xmath91 ,
the ratio of the x - ray to the bolometric luminosity of ssv61 is at the saturation limit for the coronal x - ray emission of late - type stars , @xmath92 . that maximum x - ray brightness limit is well established from observations of many different kinds of late - type stars , among them wtts like ssv61 , and so any substantial reduction in our @xmath91 estimate for
ssv61 would make it all the more difficult to understand the origin of its intense x - ray brightness . on the other hand , as one might surmise from the optical polarimetry of scarrott et al .
( 1987 ) , the 2mass photometry of ssv61 may be heavily contaminated by scattered light from the nebulosity that is present on all sides of the star . a large proportion of the near - infrared light of the more deeply embedded class i objects
is thought to be contributed by scattering in the protostellar envelope ( kenyon et al .
in that instance the scattering adds no continuum veiling to the spectrum of the central star itself ( calvet et al .
the same scattering mechanism , applied to ssv61 , would therefore be consistent with the low veiling ratio we have determined from our spex observation of ssv61 . in the submillimeter map of mitchell et al .
( 2001 ) , there is an 850@xmath6 peak that coincides in position with ssv61 .
the value of @xmath73 implied by the submillimeter flux is 40 mags or more , which is inconsistent with the near absence of reddening established by the infrared colors and ice - band spectrum of ssv61 .
this suggests that the cold dust producing the 850@xmath6 peak is most likely located behind ssv61 and that the star sits in front of the cloud , possibly illuminating its front face . in the infrared color - color plot , the 2mass colors of ssv60 place this star on the reddening line for late - type giants , consistent with the spectral classification we obtained from our irtf spectrum .
however , the bright near - infrared magnitudes of this star ( @xmath93 ) , coupled with the large extinction implied by the deep 3@xmath6 ice band we have observed ( @xmath70 ) , can not be reconciled with membership in l1630 .
the best match we can find for the spectral type , the ( @xmath94 , @xmath95 ) colors , the large amount of reddening , and the near - infrared magnitudes are for an early m background giant , which is located beyond the l1630 cloud at a distance of @xmath96 kpc and with a luminosity of @xmath97 .
we have acquired a deep x - ray image of the hh2426 region in the l1630 cloud with the acis - s camera on _ chandra _ , which has resulted in the detection of a number of anonymous 2mass infrared sources and optically visible h@xmath0 emission - line stars , including the x - ray luminous coronal source , the wtts ssv61 .
also detected at what appear to be their quiescent levels of activity are two deeply embedded class i protostars ( ssv63e and w ) , which have been identified in earlier work as the source of energetic molecular and ionized flows in the region .
the two were blended together in an earlier detection by asca , but are well separated at the much higher spatial resolution of _ chandra_. two other hh - associated class i objects ( ssv59 and hh26-ir ) as well as three class 0 ysos in the same cloud ( hh24-mms , hh25-mms , and lbs 17-h ) were not detected .
although x - ray observations of several different star forming regions thus far reveal a striking lack of class 0 detections , no firm conclusions can be drawn from our small sample of ysos regarding a possible evolutionary trend in the intrinsic x - ray luminosities of the class 0 and class i protostars .
x - ray emission was detected from a deeply embedded continuum radio source ( hh24ne ) , whose location within hh24 suggests that it may be the highly obscured driving source for one of the long hh filaments in this area .
the object has no visible or near - infrared counterpart , and therefore is potentially a class 0 or extreme class i object . +
this research has made use of the simbad database , operated at _ cds _ ,
strasbourg , france , and is based in part on observations made with the _ chandra x - ray observatory_. support for this work was provided by the national aeronautics and space administration through chandra award number go2 - 3007x issued by the _ chandra x - ray observatory _ center , which is operated by the smithsonian astrophysical observatory for and on behalf of the national aeronautics and space administration under contract nas8 - 39073 .
the digitized sky surveys ( dss ) were produced at the space telescope science institute under nasa grant nag w-2166 and are based on photographic data obtained using the oschin schmidt telescope on palomar mountain and the uk schmidt telescope with the permission of those institutions .
the guide star catalogue - ii is a joint project of the space telescope science institute and the osservatorio astronomico di torino .
space telescope science institute is operated by the association of universities for research in astronomy , for the national aeronautics and space administration under contract nas5 - 26555 .
the participation of the osservatorio astronomico di torino is supported by the italian council for research in astronomy .
additional support is provided by european southern observatory , space telescope european coordinating facility , the international gemini project and the european space agency astrophysics division .
the two micron all sky survey ( 2mass ) is a joint project of the university of massachusetts and the infrared processing and analysis center ( ipac)/california institute of technology , funded by the national aeronautics and space administration and the national science foundation .
lccrlrr 1 ...... & 05 45 46.3 & @xmath98 15 15 & 8.5 & & @xmath99 & @xmath100 + 2 ...... & 05 45 51.7 & @xmath98 16 33 & 8.5 & & @xmath101 & @xmath102 + 3 ...... & 05 45 53.1 & @xmath98 13 26 & 6.1 & gsc2 s02000123390 ( = kh@xmath0174 ) & @xmath103 & @xmath104 + 4 ...... & 05 45 57.8 & @xmath98 09 29 & 3.8 & & @xmath105 & @xmath106 + 5 ...... & 05 46 00.3 & @xmath98 08 26 & 3.5 & ssv65 & @xmath107 & @xmath108 + 6 ...... & 05 46 04.2 & @xmath98 05 59 & 4.5 & 2mass j05460427 - 0005591 & @xmath109 & @xmath110 + 7 ...... & 05 46 07.8 & @xmath98 10 01 & 1.3 & ssv63w & @xmath111 & @xmath112 + 8 ...... & 05 46 07.8 & @xmath98 11 57 & 2.4 & ssv61 & @xmath113 & @xmath114 + 9 ...... & 05 46 08.1 & @xmath98 05 32 & 4.6 & & @xmath115 & @xmath116 + 10 .... & 05 46 08.5 & @xmath98 10 03 & 1.2 & ssv63e & @xmath117 & @xmath118 + 11 .... & 05 46 08.9 & @xmath98 09 56 & 1.0 & ssv63ne & @xmath119 & @xmath120 + 12 .... & 05 46 09.5 & @xmath98 03 31 & 6.5 & 2mass j05460960 - 0003312 & @xmath121 & @xmath122 + 13 .... & 05 46 09.8 & @xmath98 05 59 & 4.0 & 2mass j05460984 - 0005591 & @xmath123 & @xmath124 + 14 .... & 05 46 11.6 & @xmath98 02 20 & 7.6 & gsc2 s02000125079 & @xmath125 & @xmath126 + 15 .... & 05 46 11.6 & @xmath98 06 28 & 3.5 & 2mass j05461162 - 0006279 & @xmath127 & @xmath128 + 16 .... & 05 46 12.3 & @xmath98 08 08 & 1.8 & 2mass j05461226 - 0008078 & @xmath129 & @xmath130 + 17 .... & 05 46 13.0 & @xmath98 18 27 & 8.5 & & @xmath131 & @xmath132 + 18 .... & 05 46 16.3 & @xmath98 01 19 & 8.7 & & @xmath133 & @xmath134 + 19 .... & 05 46 17.7 & @xmath98 16 12 & 6.4 & & @xmath135 & @xmath136 + 20 .... & 05 46 18.9 & @xmath98 05 38 & 4.6 & gsc2 s020001258 & @xmath137 & @xmath138 + 21 .... & 05 46 19.5 & @xmath98 05 20 & 4.9 & ssv64 ( = lkh@xmath0301 ) & @xmath139 & @xmath140 + 22 .... & 05 46 20.2 & @xmath98 10 20 & 1.8 & 2mass j05462018 - 0010197 & @xmath141 & @xmath142 + 23 .... & 05 46 22.4 & @xmath98 08 53 & 2.6 & ssv68 ( = lkh@xmath0302 ) & @xmath143 & @xmath144 + 24 .... & 05 46 25.9 & @xmath98 10 15 & 3.2 & & @xmath145 & @xmath146 + 25 .... & 05 46 33.3 & @xmath147 02 54 & 13.8 & ssv13 & @xmath148 & @xmath149 lcccccc ssv59 ...........
& i & 05 46 04.8 & @xmath98 14 17 & @xmath150 & @xmath151 & @xmath152 + ssv60 ........... & gk / m & 05 46 08.4 & @xmath98 14 25 & @xmath153 & @xmath154 & @xmath155 + ssv61 ........... & wtts & 05 46 07.8 & @xmath98 11 57 & @xmath156 & @xmath157 & @xmath158 + ssv63w ....... & i & 05 46 07.8 & @xmath98 10 01 & @xmath112 & @xmath159 & @xmath160 + ssv63e ......... & i & 05 46 08.5 & @xmath98 10 03 & @xmath118 & @xmath161 & @xmath162 + ssv63ne ...... & i/0 & 05 46 08.9 & @xmath98 09 56 & @xmath120 & @xmath163 & @xmath164 + hh24-mms ... & 0 & 05 46 08.3 & @xmath98 10 43 & @xmath153 & @xmath154 & @xmath155 + hh25-mms ... & 0 & 05 46 07.3 & @xmath98 13 30 & @xmath165 & @xmath166 & @xmath167 + hh26-ir ........ & i & 05 46 03.9 & @xmath98 14 53 & @xmath165 & @xmath166 & @xmath167 + lbs17-h ........ & 0 & 05 46 30.8 & @xmath98 02 40 & @xmath168 & @xmath169 & @xmath170 | the hh24 - 26 star forming region within the lynds 1630 ( l1630 ) dark cloud in orion contains a remarkable collection of rare class 0 and class i protostars , collimated molecular and ionized jets , and a luminous but spatially unresolved asca x - ray source . to study the x - ray properties of the embedded protostar population of that region
, we have obtained a deep x - ray image with the acis - s camera on board the _ chandra x - ray observatory_. a number of h@xmath0 emission - line objects were detected in the areas surrounding hh2426 , of which the weak - line t tauri star ssv61 was the brightest source , at a steady luminosity of @xmath1(0.310 kev)@xmath2 ergs s@xmath3 .
two class i protostars aligned with optical jets in hh24 , ssv63e and ssv63w , were also detected , as was the continuum radio source ssv63ne , which is very likely an extreme class i or class 0 object .
we observed no x rays from the class 0 protostars hh24-mms and hh25-mms , nor any from regions of the cloud bounded by hh25 and hh26 , at a @xmath4 upper limit of @xmath5 ergs s@xmath3 .
hh26-ir , the class i object thought to be the origin of the hh26 flow , was not detected .
near - infrared spectroscopy obtained at the nasa irtf reveals 3@xmath6 ice bands in the spectra of ssv59 , 63e , 63w , and hh26-ir , and 2.3@xmath6 co overtone absorption bands for ssv61 .
ssv60 , which lies astride one end of the great arc of nebulosity forming hh25 , exhibits a deep infrared ice band and co absorption , but is not an x - ray source , and is most likely a distant background giant of late spectral type . |
interstellar grains affect starlight which passes through them by absorbing and scattering photons .
these two physical processes produce the interstellar extinction which depends on the properties of dust grains , e.g. , size distribution and composition .
the description of the average extinction curve of our galaxy as a function of the wavelength from the infrared to ultraviolet can be found in savage @xcite .
extinction curve shows some evident features : it rises in the infrared , it shows a slight knee in the optical , it is characterized by a bump at 2175@xmath1 , and it rises in the far - ultraviolet .
these features are common between different environments .
the properties of interstellar grains are different in diffuse and dense interstellar medium and thus also the extinction changes .
ccm @xcite found an average extinction law valid over the wavelength range @xmath2 , which is applicable to both diffuse and dense regions of the interstellar medium .
this extinction law depends on only one parameter @xmath3 .
the @xmath0 parameter ranges from about 2.0 to about 5.5 ( with a typical value of 3.1 ) when one goes from diffuse to dense interstellar medium and thus @xmath0 characterizes the region that produces the extinction .
if one knows the value of @xmath0 along a particular line of sight , one can obtain the extinction curve from the infrared to ultraviolet using the ccm law ( see figure [ fig1 ] ) : @xmath4 where @xmath5 , and @xmath6 and @xmath7 are the wavelength - dependent coefficients
. there are different ways to obtain @xmath0 using the nir or uv extinction data .
wegner @xcite computed @xmath0 values for the sample of 597 ob stars with known near - infrared magnitudes .
he assumed that , in the infrared spectral region , the normalized extinction curve is proportional to @xmath8 or @xmath9 . extrapolating the ir interstellar extinction curve to @xmath10 he derived @xmath0 as : @xmath11 _ { \lambda \longrightarrow \infty } \label{rvwegner}\ ] ] gnaciski & sikorski @xcite applied the @xmath12 minimization method to compute the @xmath0 values for a sample of ultraviolet ( uv ) extinction data using the linear relation ( [ ccmlaw ] ) .
geminale & popowski @xcite extended the analysis from @xcite by using non - equal weights derived from observational errors to determine @xmath13 and @xmath0 values toward a sample of stars with known ultraviolet color excesses .
in this paper we use both optical / infrared ( @xmath14 ) and uv extinction data to obtain @xmath0 values for a sample of 436 lines of sight .
we arrive at two main conclusions : ( i ) there are lines of sight with extinction in the optical / ir or / and uv which generically do nt follow the ccm law , and ( ii ) there are lines of sight which show an extinction curve that can not be reproduced with a single @xmath0 value in the whole wavelength range .
we normalize the extinction in a standard way : @xmath15 the absolute extinction may be deduced from the relative extinction by using a total - to - selective extinction ratio @xmath0 : @xmath16 then : @xmath17 for each individual band , equation ( [ ccmlaw ] ) and ( [ eps ] ) can be combined to derive an @xmath0 value .
more generally , the @xmath12 minimization can be used to obtain the @xmath0 value that provides the best ccm fit to all observed extinction data .
gnaciski & sikorski @xcite suggested the following @xmath12 , which can be minimized to derive @xmath0 : @xmath18 \right \ } ^2,\ ] ] where @xmath19 and @xmath20 are the coefficients of the ccm law .
following our previous work @xcite we use an improved weighted formula to find @xmath0 .
we minimize the following @xmath12 : @xmath21 \ } ^2 ~e^2(b - v ) \label{chiweighted}\ ] ] where @xmath22 are the weights associated with each band . minimizing equation ( [ chiweighted ] ) with respect to @xmath0 ,
we find : @xmath23 where here @xmath24 values are taken from wegner @xcite and they were computed according to : @xmath25 = \left [ \frac{1}{e(b - v ) } \right]^2 \ , \left \{\sigma^2[e(b - v ) ] + [ e(\lambda_i - v)]^2 \ , \sigma^2[e(\lambda_i - v ) ] \right \ } \label{sigma}\ ] ] we estimate the error in @xmath0 from : @xmath26
we use a sample of 436 lines of sight with optical / ir and uv extinction data reported by wegner @xcite . in wegner s
computation the ultraviolet photometry is taken from wesselius et al .
@xcite and based on _ astronomical netherlands satellite _ ( ans ) ; infrared magnitudes in @xmath27 passbands are originate mostly from the catalog of gezari , schmitz and mead @xcite and gezari et al . @xcite .
the @xmath28 and @xmath29 magnitudes with accuracy of @xmath30 mag are taken from johnson @xcite and fernie @xcite . the spectral classification and @xmath31 data
which have the accuracy of @xmath30 mag come from the simbad database .
the effective wavelengths of the optical / ir bands are : @xmath32 , @xmath33 , @xmath34 , @xmath35 , @xmath36 , @xmath37 , @xmath38 , @xmath39 ; whereas the effective wavelengths of the uv bands are : 0.1549 , 0.1799 , 0.2200 , 0.2493 , and 0.3294 @xmath40 . from the original sample ,
we exclude 20 lines of sight because of their negative value of @xmath0 .
negative @xmath0 values are typically the result of noisy data for the lines of sight with small @xmath41 .
a useful method to test if all extinction data points are well fitted by the ccm law is to compute @xmath12 based on equation ( [ chiweighted ] ) normalized to the number of degrees of freedom , which is equal to the number of observed points minus the number of fitted parameters ( in our case the only parameter is @xmath0 ) .
we note that our average @xmath42 is not equal to the expected value of 1 ; since the errors are taken from wegner ( 2002 ) and we do not want to modify them , we do not renormalize our errors requesting @xmath43 . therefore it is safer to treat our @xmath42 as a measure of a relative rather than absolute quality of the ccm fit in the optical / ir and uv wavelength ranges .
we select outliers considering the tail of our @xmath42 distribution .
we assume that the values @xmath44 are indicative of the lines of sight with extinction curves not well fitted by the ccm law .
we find that 25% of the lines of sight of our sample shows this disagreement with the ccm law .
figure [ fig2 ] displays two extinction curves representative of this group . .
the long dashed line represents the ccm curve which is the best fit to the ir data and the point - short dashed line represents the ccm curve which best follows the uv data.,title="fig:",width=302,height=359 ] .
the long dashed line represents the ccm curve which is the best fit to the ir data and the point - short dashed line represents the ccm curve which best follows the uv data.,title="fig:",width=302,height=359 ] we further analyze the lines of sight with @xmath45 for which we expect the ccm law to fit well all observed extinction data from optical / ir to uv .
the usual assumption is that the knowledge of the @xmath0 value obtained from the ir part of the extinction curve may be used to obtain the entire extinction curve by using the ccm law .
we critically test this assumption using two sets of @xmath0 values for each line of sight : the infrared @xmath0 ( @xmath46 ) and ultraviolet @xmath0 ( @xmath47 ) .
we compute the following statistic : @xmath48 + \sigma^2[^{\rm{ir}}\!r_{v}]}}. \label{deviation}\ ] ] specifically , when @xmath49 we classify the line of sight as anomalous in the sense that the ccm law is not able to reproduce the whole extinction curve with a single value of @xmath0 .
values obtained from optical / ir and uv extinction data . the points with error bars are those for which @xmath49.,width=321 ] values are necessary to obtain a good fit to the entire extinction curve.,title="fig:",width=302,height=377 ] values are necessary to obtain a good fit to the entire extinction curve.,title="fig:",width=302,height=377 ] figure [ fig3 ] shows the comparison between the @xmath0 values obtained from optical / ir and uv extinction data . there is a large scatter around the 1-to-1 relationship
; however , only the points with the error bars shown are those which deviate from this relation significantly ( @xmath49 ) . these 25 lines of sight need two different values of @xmath0 to match the optical / ir and uv part of the extinction curve properly .
therefore , in these cases we can not universally use the ccm law to represent the entire extinction curve .
six cases out of 25 have the @xmath50 higher than the @xmath51 one , and the other 19 have the @xmath50 lower than the @xmath51 one .
figure [ fig4 ] shows two examples of such anomalous extinction curves .
we use a @xmath12 minimization method to compute the @xmath0 values for a sample of 436 lines of sight .
we exclude 20 cases because of their negative @xmath0 and analyze the final sample of 416 lines of sight .
we derive our @xmath0 values assuming ccm law .
this law aims at reproducing the entire extinction curve by using only one parameter : @xmath0 .
we analyze the goodness of the ccm fit for all lines of sight using the @xmath42 statistic and test the universality of the ccm law by computing @xmath0 values separately for the optical / ir and uv part of the extinction curve .
we find that for 69% of our original sample , the ccm law is able to fit well both the optical / ir and uv data .
we divide the remaining 31% of cases into two groups , according to two main peculiarities : a ) the optical / ir or / and uv extinction data points can not be fitted by the ccm law ( 25% of the entire sample ) ; b ) @xmath0 values are significantly different for the two spectral regions : optical / ir and uv ( 6% of the entire sample ) .
unless caused by faulty data , peculiar extinction curves result from unusual properties of dust grains .
therefore , theoretical modeling of these extinction curves ( e.g. , mishchenko @xcite ; saija et al .
@xcite ; weingartner & draine @xcite ) may help us to understand the processes which modify the properties of interstellar grains .
50 savage b d and mathis j s 1979 _ ara&a _ * 17 * 73 cardelli j a , clayton g c and mathis j s 1989 _ apj _ * 345 * 245 wegner w 2003 _ astronomische nachrichten _ * 324 * 219 gnaciski p and sikorski j 1999 _ acta astron . _ * 49 * 577 geminale a and popowski p 2004 _ acta astron . _
* 54 * 375 wegner w 2002 _ baltic astronomy _ * 11 * 1 wesselius p r van duinen r j de jonge
a r w aalders j w g luinge w and wildeman k j 1982 _ a&as _ * 49 * 427 gezari d y schmitz m and mead j m 1984 _ nasa ref .
1118 gezari d y schmitz m pitts p s and mead j m 1993 _ nasa ref .
1294 johnson h l 1966 _ ara&a _ * 4 * 193 fernie j d 1983 _ a&as _ * 52 * 75 mishchenko m. i. 1989 _ soviet astr .
* 15 * 299 saija r iat m a borghese f denti p aiello s and cecchi pestellini c 2001 _ apj _ * 559 * 993 weingartner j c and draine b t 2001 _ apj _ * 548 * 296 | we present the analysis of a large sample of lines of sight with extinction curves covering wavelength range from near - infrared ( nir ) to ultraviolet ( uv ) .
we derive total to selective extinction ratios based on the cardelli , clayton & mathis ( 1989 , ccm ) law , which is typically used to fit the extinction data both for diffuse and dense interstellar medium .
we conclude that the ccm law is able to fit most of the extinction curves in our sample .
we divide the remaining lines of sight with peculiar extinction into two groups according to two main behaviors : a ) the optical / ir or / and uv wavelength region can not be reproduced by the ccm formula ; b ) the optical / nir and uv extinction data are best fit by the ccm law with different values of @xmath0 .
we present examples of such curves .
the study of both types of peculiar cases can help us to learn about the physical processes that affect dust in the interstellar medium , e.g. , formation of mantles on the surface of grains , evaporation , growing or shattering . |
all graphs in this paper are finite and simple .
a graph @xmath2 _ contains _ a graph @xmath3 if an induced subgraph of @xmath2 is isomorphic to @xmath3 .
a graph @xmath2 is _ @xmath3-free _ if @xmath2 does not contain @xmath3 . for a set of graphs @xmath4
, @xmath2 is _ @xmath5-free _ if it is @xmath3-free for every @xmath6 .
an _ element _ of a graph is a vertex or an edge .
when @xmath7 is a set of elements of @xmath2 , we denote by @xmath8 the graph obtained from @xmath2 by deleting all edges of @xmath7 and all vertices of @xmath7 .
a _ wheel _ is a graph formed by a chordless cycle @xmath0 and a vertex @xmath1 not in @xmath0 that has at least three neighbors in @xmath0 .
such a wheel is denoted by @xmath9 ; @xmath1 is the _ center _ of the wheel and @xmath0 the _
rim_. observe that @xmath10 is a wheel ( in some papers on the same subject , @xmath10 is not considered as a wheel ) .
wheels play an important role in the proof of several decomposition theorems .
little is known about wheel - free graphs .
the only positive result is due to chudnovsky ( see @xcite for a proof ) .
it states that every non - null wheel - free graph contains a vertex whose neighborhood is made of disjoint cliques with no edges between them .
no bound is known on the chromatic number of wheel - free graphs .
no decomposition theorem is known for wheel - free graphs .
however , several classes of wheel - free graphs were shown to have a structural description . *
say that a graph is _ unichord - free _ if it does not contain a cycle with a unique chord as an induced subgraph .
the class of \{@xmath10 , unichord}-free graphs is a subclass of wheel - free graphs ( because every wheel contains a @xmath10 or a cycle with a unique chord as an induced subgraph ) , and unichord - free graphs have a complete structural description , see @xcite . *
it is easy to see that the class of graphs that do not contain a subdivision of a wheel as an induced subgraph is the class of graphs that do not contain a wheel or a subdivision of @xmath10 as induced subgraphs . here again , this subclass of wheel - free graphs has a complete structural description , see @xcite .
* the class of graphs that do not contain a wheel as a subgraph does not have a complete structural description so far .
however , in @xcite ( see also @xcite ) , several structural properties for this class are given .
* a _ propeller _ is a graph formed by a chordless cycle @xmath0 and a vertex @xmath1 not in @xmath0 that has at least _ two _ neighbors in @xmath0 .
so , wheels are just special propellers , and the class of propeller - free graphs is a subclass of wheel - free graphs . in @xcite , a structural description of propeller - free graphs is given .
interestingly , every graph that belongs to one of the four classes described above is 3-colorable ( this is shown in cited papers ) .
one might conjecture that every wheel - free graph is 3-colorable , but this is false as shown by the graph represented on figure [ fig : r ] ( it is wheel - free and has chromatic number 4 ) . also , the four classes have polynomial time recognition algorithms , so one could conjecture that so does the class of wheel - free graphs .
but it is proved in @xcite that it is np - hard to recognize them .
all this suggest that possibly , no structural description of wheel - free graphs exists .
, the unique graph @xmath2 satisfying @xmath11 , @xmath12 and @xmath13.[fig : r ] ] in this paper , we study planar wheel - free graphs . a _ clique cutset _ of a graph @xmath2 is a clique @xmath14 such that @xmath15 is disconnected .
when the clique has size three , it is referred to as a _
@xmath16-cutset_. when @xmath17 is a graph , the line graph of @xmath17 is the graph denoted by @xmath18 defined as follows : the vertex - set of @xmath18 is @xmath19 , and two vertices @xmath20 and @xmath21 of @xmath18 are adjacent if they are adjacent edges of @xmath17 .
we prove the following theorems .
[ th : struct ] if @xmath2 is a 3-connected wheel - free planar graph , then either @xmath2 is a line graph or @xmath2 has a clique cutset .
we now give a complete description of 3-connected wheel - free planar graphs , but we first need some terminology .
a graph is _ basic _ if it is the line graph of a graph @xmath22 such that either @xmath22 is @xmath23 , or @xmath22 can be obtained from a 3-connected cubic planar graph by subdividing every edge exactly once .
we need to name four special graphs : the claw , the diamond , the butterfly and the paw , that are represented in figure [ f : cdbp ] .
basic graphs have a simple characterization given below .
, height=37 ] ' '' '' , height=37 ] ' '' '' , height=37 ] ' '' '' , height=37 ] [ th : equiv ] let @xmath2 be a graph .
the following statements are equivalent . 1 .
@xmath2 is basic .
2 . @xmath2 is a 3-connected wheel - free planar line graph .
@xmath2 is 3-connected , planar and \{@xmath10 , claw , diamond , butterfly}-free . with theorems [ th : struct ] and
[ th : equiv ] , we may easily prove the complete description of 3-connected wheel - free planar graphs . by the jordan curve theorem ,
a simple closed curve @xmath0 in the plane partitions its complement into a bounded open set and an unbounded open set .
they are respectively the _ interior _ and the _ exterior _ of @xmath0 .
[ complete ] the class @xmath24 of 3-connected wheel - free planar graphs is the class of graphs that can be constructed as follows : start with basic graphs and repeatedly glue previously constructed graphs along cliques of size three that are also face boundaries . by theorem
[ th : equiv ] , a basic graph is in @xmath24 . also gluing along cliques of size three that are also face boundaries preserves being in @xmath24 ( in particular
, it does not create wheels , because wheels have no clique cutset ) .
it follows that the construction only constructs graphs in @xmath24 .
conversely , let @xmath2 be a graph in @xmath24 .
we prove by induction on @xmath25 that @xmath2 can be constructed as we claim .
if @xmath2 is a line graph , then theorem [ th : equiv ] implies @xmath2 is basic .
so by theorem [ th : struct ] we may assume that g has a clique cutset .
since @xmath2 is 3-connected , this clique must be a triangle @xmath14 whose edges form a closed curve in the plane since @xmath2 is planar .
so , @xmath2 is obtained by gluing along @xmath14 the two induced subgraphs of @xmath2 that are drawn respectively on the closure of the interior and on the closure of the exterior of @xmath14 .
these two graphs are easily checked to be 3-connected because @xmath2 is 3-connected .
it follows by induction that @xmath2 can be constructed from previously constructed graphs by gluing along a triangle that is also a face boundary .
a consequence of our description is the following .
[ th : col ] every wheel - free planar graph is 3-colorable .
we have no conjecture ( and no theorem ) about the structure of wheel - free planar graphs in general ( possibly not 3-connected ) . in figure [ f:2c ]
three wheel - free planar graphs of connectivity 2 are represented .
it can be checked that they belong to none of the four classes described above ( each of them contains a cycle with a unique chord , an induced subdivision of @xmath10 , a wheel as a subgraph and a propeller ) .
so we do not understand them .
we leave the description of the most general wheel - free planar graph as an open question . , title="fig:"],title="fig:"],title="fig : " ] section [ sec : struct ] gives the proof of theorem [ th : struct ] , and in fact of a slight generalization that we need in section [ coloration ] . in section [ sec : equiv ] , we prove theorem [ th : equiv ] . theorem [ th : col ] is proved in section [ coloration ] .
we use notation and classical results from @xcite .
let @xmath2 be a graph , @xmath26 and @xmath27 .
we denote by @xmath28 $ ] the subgraph of @xmath2 induced on @xmath29 , by @xmath30 the set of neighbors of @xmath1 , and by @xmath31 the set of vertices of @xmath32 adjacent to at least one vertex of @xmath29 ; and we define @xmath33 .
we sometimes write @xmath34 instead of @xmath35 .
when @xmath36 is an edge of @xmath2 , we denote by @xmath37 the graph obtained from @xmath2 by contracting @xmath36 . a _ path _
@xmath38 is a graph with @xmath39 vertices that can be numbered @xmath40 , and with @xmath41 edges @xmath42 for @xmath43 .
the vertices @xmath44 and @xmath45 are the _ end - vertices _ of @xmath38 , and @xmath46 is the _ interior _ of @xmath38 .
we also say that @xmath38 is a _
@xmath47-path_. if @xmath48 are paths , disjoint except that they have one end - vertex @xmath49 in common , then their union is a path and we often denote it by @xmath50 . if @xmath51 are vertices of a path @xmath38 , we denote the subpath of @xmath38 with end - vertices @xmath51 by @xmath52 .
a _ cycle _ @xmath0 is a graph with @xmath53 vertices that can be numbered @xmath54 , and with @xmath55 edges @xmath42 for @xmath56 ( where @xmath57 ) .
let @xmath58 be a path or a cycle in a graph @xmath2 .
the _ length _ of @xmath58 is the number of its edges .
an edge @xmath59 of @xmath2 is a _ chord _ of @xmath58 if @xmath60 , but @xmath61 is not an edge of @xmath58 .
a chord is _ short _ if its ends are joined by a two - edge path in @xmath58 .
we need the following . [ harary ] a graph is the line graph of a triangle - free graph if and only if it is \{diamond , claw}-free .
a graph @xmath2 is _ almost 3-connected _ if either it is 3-connected or it can be obtained from a 3-connected graph by subdividing one edge exactly once . for a 2-connected graph drawn in the plane ,
the boundary of every face is a cycle .
we need the following consequence .
[ planar ] let @xmath2 be an almost 3-connected graph drawn in the plane , and let @xmath20 be a vertex of @xmath2 such that all its neighbors have degree at least three .
let @xmath17 be the face of @xmath62 in which @xmath20 is drawn .
then the boundary of @xmath17 is a cycle @xmath0 , and @xmath0 goes through every vertex of @xmath63 . in this section ,
we prove the theorem below , which clearly implies theorem [ th : struct ] .
we prove the stronger statement below because we need it in the proof of theorem [ th : col ] .
[ claw ] if @xmath2 is an almost 3-connected wheel - free planar graph with no clique cutset , then @xmath2 is a line graph .
the proof is by contradiction , so suppose that @xmath2 is an almost 3-connected wheel - free planar graph that has no clique cutset and that is not a line graph .
[ trianglefact ] let @xmath64 be a clique of size three in @xmath2 , and let @xmath38 be a chordless path of @xmath65 with one end @xmath66
. then at least one of @xmath67 has no neighbor in @xmath68 .
suppose @xmath67 both have neighbors in @xmath68 , and @xmath69 be the minimal subpath of @xmath38 such that @xmath70 , and both @xmath71 and @xmath72 have neighbors in @xmath73 .
we may assume that @xmath69 is from @xmath66 to @xmath20 , @xmath20 is adjacent to @xmath71 , and @xmath71 has no neighbor in @xmath74 .
then @xmath75 is an induced cycle , say @xmath0 .
now since @xmath72 is adjacent to @xmath66 and @xmath71 , and has a neighbor in @xmath73 , it follows that @xmath76 is a wheel , a contradiction .
[ diamond ] @xmath2 is diamond - free .
suppose that @xmath77 induces a diamond of @xmath2 , and @xmath78 . since @xmath79 is not a cutset of @xmath2 , there exists a chordless @xmath61-path @xmath38 in @xmath80 , contrary to ( [ trianglefact ] ) .
a vertex @xmath36 of @xmath2 is a _ corner _ if @xmath36 has degree two , and there exist four vertices @xmath81 such that @xmath82)=\{ab , ac , bc , cd , de , eb\}$ ] . [ corner ] no vertex of @xmath2 is a corner .
suppose that @xmath83 is a corner and let @xmath81 be four vertices as in the definition . since @xmath84 is not a cutset of @xmath2 , there exists a chordless @xmath85-path @xmath38 in @xmath86 .
but now the path @xmath87 contradicts ( [ trianglefact ] ) .
[ c : claw ] @xmath2 contains a claw .
otherwise , by ( [ diamond ] ) and theorem [ harary ] , @xmath2 is a line graph , a contradiction .
the rest of the proof is in two steps .
we first prove the existence of a special cutset , called an `` i - cutset '' ( defined below ) .
then we use the i - cutset to obtain a contradiction .
let @xmath88 be a cutset of size three of @xmath2 .
a component of @xmath89 is said to be _ degenerate _ if it has only one vertex , or it has exactly two vertices @xmath51 and @xmath90 $ ] has the following edge - set : @xmath91 , and _ nondegenerate _ otherwise .
a cutset @xmath88 of size three of @xmath2 is an _ @xmath92-cutset _
if @xmath93 $ ] has at least one edge and @xmath89 has at least two connected components that are non - degenerate .
[ icutset ] @xmath2 admits an @xmath92-cutset .
fix a drawing of @xmath2 in the plane . by ( [ c : claw ] )
, @xmath2 contains a claw .
let @xmath1 be the center of a claw .
let @xmath94 ( @xmath95 ) be the neighbors of @xmath1 , in cyclic order around @xmath1 , where @xmath96 have degree at least three .
if @xmath97 has degree two , let @xmath98 be its neighbor different from @xmath1 , and otherwise let @xmath99 .
deleting @xmath1 , and also deleting @xmath97 if @xmath97 has degree two , yields a 2-connected graph , drawn in the plane , and therefore , the face @xmath17 of this drawing in which @xmath1 is drawn is bounded by a cycle @xmath0 . by theorem [ planar
] @xmath100 all belong to @xmath0 , and are in order in @xmath0 . for @xmath101 , let @xmath102 be the unique @xmath103-path included in @xmath0 that contains none of @xmath104 except @xmath105 and @xmath106 ( subscripts are taken modulo @xmath55 ) .
assume that @xmath61 is a chord of @xmath0 .
vertices @xmath20 and @xmath21 edge - wise partition @xmath0 into two @xmath61-paths , say @xmath69 and @xmath107 .
since @xmath17 is a face of @xmath108 or of @xmath109 , it follows that @xmath88 is a cutset of @xmath2 that separates the interior of @xmath110 from the interior of @xmath111 .
if @xmath61 is not a short chord , then both these interiors contain at least two vertices and therefore @xmath88 is an @xmath92-cutset of @xmath2 .
so we may assume that @xmath61 is short .
if @xmath112 both belong to @xmath102 for some @xmath113 , then @xmath114 is a clique - cutset of @xmath2 , a contradiction .
thus we may assume that for every chord @xmath61 of @xmath0 , there exists @xmath115 such that @xmath116 , @xmath117 and both @xmath118 and @xmath119 are edges .
* claim*. @xmath120 and @xmath121 has degree two . to prove the claim , assume by way of contradiction that @xmath1 has at least three neighbors of degree at least @xmath122 .
since @xmath2 is wheel - free , @xmath0 must have chords .
let @xmath61 be a chord , and choose @xmath123 such that @xmath118 and @xmath119 are edges of @xmath0 .
suppose first that we can not choose @xmath61 and @xmath113 such that @xmath105 is adjacent to @xmath1 .
consequently @xmath124 , and @xmath97 has degree two ; moreover , the cycle obtained from @xmath0 by replacing the edges @xmath125 and @xmath126 by @xmath61 is induced . since in this case
@xmath127 , it follows that @xmath1 has at least three neighbors in this cycle and so @xmath2 contains a wheel , a contradiction .
we can therefore choose @xmath61 and @xmath113 such that @xmath105 is adjacent to @xmath1 .
it follows that @xmath128 are not consecutive in @xmath0 , since @xmath1 is the center of a claw .
we claim that there are no edges between @xmath129 and @xmath130 , except @xmath61 . for suppose
such an edge exists , say @xmath131 . since @xmath128 are not consecutive in @xmath0 , it follows that @xmath131 is a chord of @xmath0 .
since every chord of @xmath0 is short , it follows that @xmath132 and that @xmath133 and @xmath134 are both edges of @xmath0 .
but now , @xmath135 $ ] is a wheel or , in case one of @xmath136 , @xmath137 or @xmath138 equals @xmath139 and @xmath121 has degree @xmath140 , @xmath141 $ ] is a wheel .
hence there are no edges between @xmath129 and @xmath130 except @xmath61 and thus , @xmath142 or , in the case where @xmath143 and @xmath121 exists , @xmath144 is an induced cycle containing three neighbors of @xmath105 , a contradiction .
this proves the claim .
observe that the claim implies that every center of a claw in @xmath2 has degree three and is adjacent to @xmath121 since @xmath2 has at most one vertex of degree two .
let @xmath112 be the neighbors of @xmath145 in @xmath146 , @xmath147 respectively .
note that possibly @xmath148 .
observe that , since @xmath1 is the center of a claw , @xmath149 .
since every center of a claw is adjacent to @xmath97 , it follows that @xmath145 is not the center of a claw and thus @xmath61 is an edge .
now , @xmath150 must admit a chord , for otherwise @xmath145 is the center of a wheel of @xmath2 .
hence @xmath151 is an edge .
let @xmath152 be the neighbor of @xmath153 in @xmath147 . since @xmath153 is not the center of claw
, @xmath154 is an edge and thus @xmath97 is a corner , a contradiction to ( [ corner ] ) .
[ path ] let @xmath88 be an @xmath92-cutset of @xmath2 where @xmath61 is an edge and let @xmath0 be a nondegenerate connected component of @xmath155 . then there exist @xmath156 and a path @xmath38 of @xmath157 $ ] from @xmath1 to @xmath49 , such that the vertex of @xmath158 has no neighbor in @xmath159 .
since @xmath2 does not admit a clique cutset , it follows that @xmath1 is non - adjacent to at least one of @xmath112 . if @xmath1 is adjacent to exactly one vertex among @xmath20 and @xmath21 , then the claim holds .
so we may assume that @xmath1 is adjacent to neither @xmath20 nor @xmath21 . since @xmath2 is \{diamond , @xmath10}-free , at most one vertex of @xmath2 is adjacent to both @xmath20 and @xmath21 .
let @xmath66 be such a vertex , if it exists .
let @xmath160 if @xmath66 exists , and let @xmath161 otherwise .
since @xmath14 is not a clique cutset in @xmath2 , we deduce that @xmath1 has a neighbor in every component of @xmath162 .
suppose first that there is a component @xmath163 of @xmath162 containing a neighbor of one of @xmath112 .
let @xmath38 be a path with interior in @xmath163 , one of whose ends is @xmath1 , and the other one is in @xmath114 , and subject to that as short as possible . then
only one of @xmath112 has a neighbor in @xmath164 , and ( [ path ] ) holds .
so we may assume that no such component @xmath163 exists , and thus neither of @xmath112 has neighbors in @xmath165 .
let @xmath166 if @xmath66 exists , and otherwise let @xmath167 .
then @xmath168 is a cutset in @xmath2 separating @xmath169 from @xmath112 .
since @xmath2 is almost 3-connected , it follows that @xmath166 , and @xmath169 consists of a unique vertex of degree two , so @xmath0 is degenerate , a contradiction . for every @xmath92-cutset @xmath88 , some nondegenerate component @xmath170 of @xmath155
has no vertex with degree two in @xmath2 ; choose an @xmath92-cutset @xmath88 and @xmath170 such that @xmath171 is minimum .
we refer to this property as the _ minimality of @xmath170_. put @xmath172 $ ] , and @xmath173 .
assume without loss of generality that @xmath61 is an edge , and let @xmath174 be another nondegenerate component . from ( [ path ] ) and the symmetry between @xmath112
, we may assume without loss of generality that there is a chordless path @xmath58 of @xmath175 from @xmath1 to @xmath20 such that @xmath21 has no neighbor in @xmath176 , and in particular @xmath177 are non - adjacent .
also , since @xmath177 both have neighbors in @xmath178 , there is a chordless path @xmath17 of @xmath175 between @xmath177 not containing @xmath20 .
since @xmath177 both have neighbors in @xmath170 , there is a chordless path @xmath38 of @xmath179 between @xmath177 not containing @xmath20 .
consequently the union of @xmath48 and the edge @xmath61 is a cycle @xmath7 .
let @xmath180 be the disc bounded by @xmath7 .
suppose that some edge of @xmath179 incident with @xmath20 is in the interior of @xmath180 , and some other such edge is in the exterior of @xmath180 . by adding these two edges to an appropriate path within @xmath181 $ ]
, we obtain a cycle @xmath182 drawn in the plane , such that the path formed by the union of @xmath61 and @xmath38 crosses it exactly once ; and so one of @xmath183 is in the interior of the disc bounded by @xmath182 , and the other in the exterior .
but this is impossible , because @xmath183 are also joined by the path @xmath17 , which is included in @xmath175 and thus is disjoint from @xmath184 .
we deduce that we may arrange the drawing so that every edge of @xmath179 incident with @xmath20 belongs to the interior of @xmath180 .
in addition we may arrange that the edge @xmath61 is incident with the infinite face .
subject to this condition ( and from now on with the drawing fixed ) , let us choose @xmath38 so that @xmath180 is minimum . since @xmath185 , every component of @xmath186
has vertex set either a subset of @xmath170 or disjoint from @xmath170 .
suppose that some vertex @xmath72 of @xmath170 is drawn in the interior of @xmath180 , and let @xmath14 be the component of @xmath186 containing it . from the choice of @xmath38
, it follows that there do not exist two non - consecutive vertices of @xmath38 both with neighbors in @xmath14 and , since @xmath187 ( because @xmath2 is almost 3-connected and all vertices in @xmath170 are of degree at least three ) , and @xmath188 , we deduce that @xmath189 , and @xmath190 say , where @xmath51 are consecutive vertices of @xmath38 . from the minimality of @xmath170 ,
@xmath191 is not an i - cutset , and thus @xmath14 is degenerate .
hence , since @xmath14 has no vertex of degree @xmath140 , @xmath192 , i.e. @xmath193 .
therefore @xmath72 has degree three , with neighbors @xmath194 .
but then @xmath72 has three neighbors in @xmath7 , and so @xmath2 contains a wheel , a contradiction .
thus no vertex in @xmath170 is drawn in the interior of @xmath180 .
so , since all edges of @xmath179 incident with @xmath20 belong to the interior of @xmath180 , every neighbor of @xmath20 in @xmath170 belongs to @xmath38 . since @xmath2 is almost 3-connected and all vertices of @xmath170 are of degree at least 3 , @xmath20 has at least one neighbor in @xmath170 .
since @xmath195 is a chordless cycle , it follows that @xmath20 has at most two neighbors in @xmath38 ( counting @xmath21 ) , and so only one neighbor in @xmath170 .
let @xmath196 be the unique neighbor of @xmath20 in @xmath170 . since @xmath197
, there is a vertex @xmath198 different from @xmath196 in @xmath170 , and since @xmath2 is almost 3-connected , there are two paths of @xmath2 , from @xmath198 to @xmath177 respectively , vertex - disjoint except for @xmath198 , and not containing @xmath196 .
consequently both these paths are paths of @xmath179 , and so there is a path of @xmath179 between @xmath177 , containing neither of @xmath199 .
we may therefore choose a chordless path @xmath69 of @xmath179 between @xmath177 , containing neither of @xmath199 .
it follows that the union of @xmath200 and the edge @xmath61 is a chordless cycle @xmath201 say , bounding a disc @xmath202 say ; choose @xmath69 such that @xmath202 is minimal .
since @xmath196 is in @xmath38 and @xmath61 is incident with the infinite face , it follows that @xmath196 is in the interior of @xmath202 .
let @xmath203 be the set of vertices in @xmath204 that are drawn in the interior of @xmath202 .
we claim that every vertex in @xmath203 has degree three , and is adjacent to @xmath196 and to two consecutive vertices of @xmath69 . for let @xmath205 , and
let @xmath14 be the component of @xmath206 that contains @xmath72 . from the choice of @xmath69 , no two non - consecutive vertices of @xmath69 have neighbors in @xmath14 , and so as before , @xmath207 , where @xmath51 are consecutive vertices of @xmath69 , and @xmath192 .
it follows that every vertex in @xmath203 has degree three and is adjacent to @xmath196 and to two consecutive vertices of @xmath69 .
let @xmath196 have @xmath208 neighbors in @xmath69 .
thus @xmath196 has at least @xmath209 neighbors in the chordless cycle @xmath201 , and consequently @xmath210 since @xmath2 does not contain a wheel .
the degree of @xmath196 equals @xmath211 , and since @xmath196 has degree at least three and @xmath210 , we deduce that @xmath212 , and either @xmath213 , or @xmath214 and @xmath215 .
choose @xmath216 , and let @xmath152 be adjacent to @xmath217 , where @xmath218 are in order in @xmath69 .
we claim that @xmath196 is adjacent to neither @xmath1 nor @xmath21 . for suppose
@xmath196 is adjacent to @xmath1 or @xmath21 .
since @xmath196 is the unique neighbor of @xmath20 in @xmath170 , @xmath219 is a cutset of @xmath2 separating @xmath220 from the rest of the graph .
so , since it is not an @xmath92-cutset and since all vertices in @xmath170 have degree at least @xmath122 , @xmath221 and thus @xmath222 . since @xmath152 has degree at least @xmath122 , @xmath152 is adjacent to @xmath1 , @xmath21 and @xmath196 and , since @xmath223 , @xmath224 .
hence @xmath225 $ ] is a diamond , a contradiction to ( [ diamond ] ) or else @xmath196 is adjacent to both @xmath1 and @xmath21 and @xmath225 $ ] is a wheel , a contradiction .
so @xmath196 is adjacent to neither @xmath1 nor @xmath21 . if @xmath196 has a neighbor in @xmath226 ( a unique neighbor because @xmath227 ) between @xmath1 and @xmath66 , say @xmath49 , then @xmath152 has three neighbors in the chordless cycle formed by the union of @xmath228 , the subpath of @xmath69 between @xmath49 and @xmath21 , and the edges @xmath229 and @xmath230 .
on the other hand , if @xmath196 has a neighbor in @xmath69 between @xmath71 and @xmath21 , say @xmath49 , then @xmath196 has three neighbors in the chordless cycle formed by the union of @xmath228 , the subpath of @xmath69 between @xmath49 and @xmath1 , the path @xmath58 and the edge @xmath230 .
thus @xmath196 has no neighbor in @xmath69 , and so @xmath231 and @xmath232 .
let @xmath233 , adjacent to @xmath234 say , where @xmath235 are consecutive vertices of @xmath69 , and @xmath236 are in order on @xmath69 . from planarity , @xmath237 , and
so we may assume that @xmath238 are in order on @xmath69 .
but then @xmath239 has three neighbors in the chordless cycle formed by the path @xmath240 and the subpath of @xmath69 between @xmath71 and @xmath21 , a contradiction .
we need the following . [
th : sed ] if @xmath22 is a graph of maximum degree at most three , then @xmath241 is planar if and only if @xmath22 is planar .
we now prove the following implications between the three statements of theorem [ th : equiv ] . *
( @xmath242)*. suppose that @xmath2 is a basic graph .
from the definition , @xmath2 is a line graph of a planar graph @xmath17 of maximum degree at most 3 .
moreover , it is easy to see that @xmath2 is 3-connected . by theorem
[ th : sed ] , @xmath2 is planar .
it remains to check that is wheel - free .
if @xmath243 , then @xmath2 is obviously wheel - free . otherwise , @xmath17 is obtained from a 3-connected cubic planar graph by subdividing every edge exactly once .
suppose for a contradiction that @xmath244 is a wheel in @xmath2 .
since @xmath245 , @xmath1 is an edge of @xmath17 , and we set @xmath246 where @xmath20 has degree @xmath122 and @xmath21 has degree @xmath140 .
let @xmath247 be the other neighbor of @xmath21 ( so , @xmath247 has degree 3 in @xmath17 ) . in @xmath17 , there are two edges @xmath36 and @xmath248 different from @xmath61 and incident to @xmath20 . and there are two edges @xmath249 and @xmath250 different from @xmath251 and incident to @xmath247 .
since @xmath1 ( seen as a vertex of @xmath2 ) has degree 3 , the cycle @xmath0 of @xmath2 must go through @xmath36 , @xmath248 and @xmath252 ( also seen as vertices of @xmath2 ) .
but to go in and out from the vertex @xmath252 of @xmath2 , the only way is through @xmath249 and @xmath250 that are adjacent .
it follows that @xmath0 has a chord , a contradiction . *
( @xmath253)*. suppose that @xmath2 is a 3-connected wheel - free planar line graph , say @xmath245 .
since @xmath2 is a line graph , it is claw - free .
since @xmath2 is wheel - free , it is @xmath10-free .
suppose for a contradiction that @xmath2 contains a diamond .
it follows that @xmath17 contains a paw ( see figure [ f : cdbp ] ) , say a triangle @xmath254 and vertex @xmath208 adjacent to @xmath20 and to none of @xmath21 or @xmath152 .
since @xmath2 is 3-connected , the removal of the edge @xmath255 in @xmath17 keeps @xmath17 connected .
it follows that in @xmath17 , there is a path @xmath38 from @xmath208 to @xmath21 or @xmath152 , that does not use the edge @xmath256 .
the edges of @xmath38 , together with the edges @xmath256 , @xmath61 , @xmath257 and @xmath258 form a wheel in @xmath2 , a contradiction .
suppose finally that @xmath2 contains a butterfly .
the vertex of degree 4 in the butterfly is an edge @xmath61 in @xmath17 , and both @xmath20 and @xmath21 have degree at least 3 ( because of the butterfly ) , and in fact exactly 3 ( because @xmath2 contains no @xmath10 ) .
since @xmath2 is 3-connected , the removal of the vertex @xmath61 of @xmath2 makes a 2-connected graph .
it follows that in @xmath17 , there exists a cycle through @xmath20 and @xmath21 that does not go through the edge @xmath61 .
hence , the edges of this cycle form the rim of a wheel in @xmath2 ( the center is the vertex @xmath61 of @xmath2 ) .
this is a contradiction . *
( @xmath259)*. by theorem [ harary ] , @xmath2 is the line graph of a triangle - free graph @xmath17 . since @xmath2 is @xmath10-free , every vertex of @xmath17 has degree at most @xmath122 . in particular , since @xmath2 is planar , by theorem [ th : sed ]
, @xmath17 must be planar . also , if @xmath17 has a cutvertex @xmath20 , at least one pair of edges incident to @xmath20 form a cutset ( of vertices ) in @xmath2 , because @xmath2 has at least four vertices since it is 3-connected .
this is a contradiction to the 3-connectivity of @xmath2 .
it follows that @xmath17 is 2-connected .
if two adjacent vertices of @xmath17 have degree 3 , then @xmath2 contains a diamond or a butterfly , a contradiction .
hence , @xmath17 is edge - wise partitioned into its branches , where a _ branch _ in a graph is a path of length at least 2 , whose ends have degree at least 3 and whose internal vertices have degree 2 .
in fact , every branch of @xmath17 has length exactly 2 because a branch of length at least 3 would yield a vertex of degree 2 in @xmath2 , a contradiction to its 3-connectivity .
suppose that there is a pair of vertices @xmath260 of degree 3 in @xmath17 such that at least two distinct branches @xmath261 have ends @xmath20 and @xmath21 .
we denote by @xmath36 ( resp .
@xmath248 ) the edge incident to @xmath20 ( resp .
@xmath21 ) that does not belong to @xmath38 or @xmath58 .
now , @xmath262 is disconnected ( contradicting @xmath2 being 3-connected ) , unless @xmath36 and @xmath248 are the only edges of @xmath17 that do not belong to @xmath38 and @xmath58 .
but in this case , @xmath263 .
so , from here on , we may assume that for all pairs of vertices @xmath260 from @xmath17 , there is at most one branch of @xmath17 with ends @xmath20 and @xmath21 .
it follows that by suppressing all vertices of degree 2 of @xmath17 , a cubic graph @xmath264 is obtained ( _ suppressing _ a vertex of degree 2 means contracting one the edges incident to it ) .
suppose that @xmath264 is not 3-connected .
this means that @xmath265 is disconnected where @xmath20 and @xmath21 are vertices of @xmath264 .
since @xmath264 is cubic , for at least one component @xmath266 of @xmath265 , @xmath20 has a unique neighbor @xmath247 in @xmath266 .
also , @xmath21 has a unique neighbor @xmath267 in some component @xmath268 .
now , @xmath269 and @xmath270 are two edges of @xmath264 whose removal disconnects @xmath264 .
these two edges are subdivided in @xmath17 , but they still yield two edges whose removal disconnects @xmath17 .
this yields two vertices in @xmath2 whose removal disconnects @xmath2 , a contradiction to @xmath2 being 3-connected .
we proved that @xmath17 is obtained from a 3-connected cubic graph ( namely @xmath264 ) by subdividing once every edge .
a _ coloring _ of @xmath2 is a function @xmath271 such that no two adjacent vertices receive the same color @xmath272 . if @xmath273 , we say that @xmath274 is a _ k - coloring _ of @xmath2 .
an _ edge - coloring _ of @xmath2 is a function @xmath275 such that no two adjacent edges receive the same color @xmath272 .
if @xmath273 , we say that @xmath274 is a _ k - edge - coloring _ of @xmath2 .
observe that an edge - coloring of a graph @xmath22 is also a coloring of @xmath241 .
a graph @xmath17 is _ chordless _ if every cycle in @xmath17 is chordless .
a way to obtain a chordless graph is to take any graph and to subdivide all edges .
it follows that basic graphs are in fact line graphs of chordless graphs .
this is the property of basic graphs that we rely on in this section .
it is proved in @xcite that for all @xmath276 and all chordless graphs @xmath2 of maximum degree @xmath277 , @xmath2 is @xmath277-edge - colorable ( for @xmath278 , a simpler proof is given in @xcite ) .
unfortunately , this result is not enough for our purpose and we reprove it for @xmath278 in a slightly more general form .
a graph is _ almost chordless _ if at most one of its edges is the chord of a cycle . for suppose that there exists an edge @xmath282 such that @xmath1 and @xmath49 belong to @xmath281 . from the minimality of @xmath2
there exists a 3-edge - coloring of @xmath283 . since @xmath284 , it is easy to extend the 3-edge - coloring of @xmath285 to a 3-edge - coloring of @xmath2 , a contradiction .
otherwise @xmath2 has a cut - vertex @xmath49 , so @xmath286 partitions into two nonempty sets of vertices @xmath170 and @xmath178 with no edges between them .
a 3-edge - coloring of @xmath2 can be obtained easily from 3-edge - colorings of @xmath287 $ ] and @xmath288 $ ] , a contradiction .
suppose there exists two disjoint edges @xmath291 and @xmath292 such that @xmath293 is not connected ; then @xmath294 partitions into two nonempty sets of vertices @xmath170 and @xmath178 with no edges between them . by ( [ c:2con ] )
we may assume that @xmath295 and @xmath296 .
for @xmath297 , let @xmath298 be the graph obtained from @xmath299 $ ] by adding a vertex @xmath300 adjacent to both @xmath105 and @xmath301 .
if @xmath179 contains a cycle @xmath0 with a chord @xmath131 , then @xmath131 is a chord of a cycle of @xmath2 ( this is clear when @xmath0 does not contain @xmath302 , and when @xmath0 contains @xmath302 , the cycle is obtained by replacing @xmath302 by a @xmath303-path included in @xmath178 that exists by ( [ c:2con ] ) ) .
it follows that @xmath179 and symmetrically @xmath175 are almost chordless .
moreover they both clearly have maximum degree at most three and , by ( [ y ] ) , both @xmath170 and @xmath178 contain vertices of degree three , so @xmath179 and @xmath175 have fewer edges than @xmath2 . therefore @xmath179 and @xmath175 admit 3-edge - colorings .
let @xmath304 and @xmath305 be 3-edge - colorings of respectively @xmath179 and @xmath175 .
we may assume without loss of generality that @xmath306 and @xmath307 .
now , the following coloring @xmath274 is a 3-edge - coloring of @xmath2 : @xmath308 , @xmath309 , @xmath310 if @xmath311 and @xmath312 if @xmath313 , a contradiction .
suppose that @xmath61 is an edge of @xmath28 $ ] such that @xmath314 is not 2-connected .
then , there exists a vertex @xmath315 such that @xmath316 is disconnected .
let @xmath266 and @xmath268 be the two components of @xmath316 , where @xmath317 and @xmath318 . since @xmath315 is of degree at most three , @xmath315 has a unique neighbor @xmath319 in one of @xmath320 , say in @xmath266 .
if @xmath321 , then @xmath20 is a cut - vertex of @xmath2 ( because @xmath322 since @xmath20 has degree three ) , a contradiction to ( [ c:2con ] ) .
so @xmath323 and hence @xmath324 are disjoint , a contradiction to ( [ edgecutset ] ) .
therefore , for every edge @xmath61 of @xmath28 $ ] , @xmath314 is 2-connected .
so , if such an edge exists , by menger s theorem there exists a cycle @xmath0 going through both @xmath20 and @xmath21 in @xmath314 , and thus @xmath61 is a chord of @xmath0 .
since @xmath2 is almost chordless , there is at most one such edge . if @xmath2 is chordless , then by ( [ y ] ) and ( [ x ] ) , @xmath325 forms a bipartition of @xmath2 , so by a classical theorem of knig , @xmath2 is 3-edge - colorable , a contradiction .
so let @xmath61 be a chord of a cycle of @xmath2 .
let @xmath247 and @xmath326 be the two neighbors of @xmath20 distinct from @xmath21 and let @xmath267 and @xmath327 be the two neighbors of @xmath21 distinct from @xmath20 . by ( [ x ] ) , @xmath247 , @xmath326 , @xmath267 and @xmath327 are all of degree 2 and by ( [ y ] ) , they induce a stable set . if @xmath328 , then @xmath2 is the diamond and thus is 3-edge - colorable .
if @xmath329 , say @xmath330 and @xmath331 , then @xmath332 are disjoint and their deletion disconnects @xmath2 , a contradiction to ( [ edgecutset ] ) .
hence @xmath247 , @xmath326 , @xmath267 and @xmath327 are pairwise distinct .
let @xmath333 ( resp .
@xmath334 , @xmath335 , @xmath336 ) be the unique neighbor of @xmath247 ( resp .
@xmath326 , @xmath337 , @xmath338 ) distinct from @xmath20 ( resp .
@xmath21 ) .
let @xmath339 be the graph obtained from @xmath2 by deleting the edge @xmath61 and contracting edges @xmath269 , @xmath340 , @xmath270 and @xmath341 .
we note @xmath20 the vertex resulting from the contraction of @xmath269 and @xmath340 , and @xmath21 the vertex resulting from the contraction of @xmath270 and @xmath341 .
since @xmath339 has maximum degree at most three and is bipartite by ( [ y ] ) and ( [ x ] ) , it follows that @xmath339 has a 3-edge - coloring @xmath342 by knig s theorem .
assume without loss of generality that @xmath343 , @xmath344 , @xmath345 and @xmath346 where @xmath347 .
since @xmath348 , we may assume without loss of generality that @xmath349 , so @xmath350 .
let us now extend this coloring to a 3-edge - coloring @xmath274 of @xmath2 . for any edge @xmath36 of @xmath2 such that its extremities are not both in @xmath351 , set @xmath352 .
set @xmath353 , @xmath354 , @xmath355 and @xmath356 .
now we can set @xmath357 , @xmath358 , @xmath359 , @xmath360 and @xmath361 .
so @xmath274 is a 3-edge - coloring of @xmath2 .
we argue by induction on @xmath25 .
suppose first that @xmath2 admits a clique cutset @xmath14 .
let @xmath170 be the vertex set of a component of @xmath362 and @xmath363 . by induction @xmath364 $ ] and @xmath365 $ ] are both 3-colorable and thus @xmath2 is 3-colorable .
so we may assume that @xmath2 has no clique cutset .
if @xmath2 has a vertex @xmath1 of degree two , then we can 3-color @xmath108 by induction and extend the coloring to a 3-coloring of @xmath2 .
so we may assume that every vertex of @xmath2 has degree at least three .
assume now that @xmath2 is 3-connected .
by theorem [ claw ] , there exists a chordless graph @xmath22 of maximum degree three such that @xmath366 .
hence , by theorem [ almostchordless ] , @xmath22 is 3-edge - colorable and thus @xmath2 is 3-colorable .
so we may assume that the connectivity of @xmath2 is two .
let @xmath367 be such that @xmath368 is disconnected .
we choose @xmath369 to minimize the smallest order of a component of @xmath370 , and let @xmath0 be the vertex set of this component . if @xmath371 , then the vertex in @xmath0 is of degree two in @xmath2 , a contradiction .
so @xmath372 .
let @xmath373 be the graph obtained from @xmath374 $ ] by adding the edge @xmath131 ( that did not exist since @xmath2 has no clique cutset ) .
let us prove that @xmath373 is 3-connected . since @xmath372 and @xmath373 therefore has at least four vertices , we may assume by contradiction that @xmath373 admits a 2-cutset @xmath114 .
let @xmath375 ( @xmath376 ) be the vertex sets of the components of @xmath377 . since @xmath131 is an edge of @xmath373 , @xmath66 and @xmath71
are included in @xmath378 $ ] for some @xmath379 , say @xmath380 . hence @xmath114 is a cutset of @xmath2 and @xmath170 is a component of @xmath381 that is a proper subset of @xmath0 , a contradiction to the minimality of @xmath0 .
so @xmath373 is 3-connected .
( but it might not be wheel - free . )
let @xmath382 be the graph obtained from @xmath373 by subdividing @xmath131 once , and let @xmath383 be the vertex of degree two of @xmath382 .
since @xmath373 is 3-connected , @xmath382 is almost 3-connected .
suppose that @xmath382 admits a wheel @xmath384 .
since @xmath2 is wheel - free , @xmath383 must be a vertex of @xmath385 .
since @xmath383 is of degree two , @xmath383 is in @xmath17 , and so @xmath386 is a subpath of @xmath17 .
since @xmath2 is 2-connected , there exists a chordless @xmath131-path @xmath38 in @xmath387 .
hence by replacing @xmath386 by @xmath38 , we obtain a wheel in @xmath2 , a contradiction .
therefore @xmath382 is an almost 3-connected wheel - free planar graph . by theorem [ claw
] , there exists a chordless graph @xmath22 of maximum degree three such that @xmath388 .
we are now going to prove there exist two ways to 3-edge - color @xmath22 , one giving the same color to @xmath66 and @xmath71 ( that are edges of @xmath22 ) , and the other giving distinct colors to @xmath66 and @xmath71 .
this implies that there exist two ways to 3-color @xmath374 $ ] , one giving the same color to @xmath66 and @xmath71 and the other giving distinct colors to @xmath66 and @xmath71 . since by the inductive hypothesis there exists a 3-coloring of @xmath389 , it follows that this 3-coloring can be extended to a 3-coloring of @xmath2 .
we first prove that there exists a 3-edge - coloring @xmath274 of @xmath22 such that @xmath390 .
observe that both ends of @xmath383 are of degree two in @xmath22 .
hence , @xmath391 is also a chordless graph with maximum degree at most three .
therefore there exists a 3-edge - coloring @xmath274 of @xmath392 and clearly @xmath274 statisfies @xmath393 .
it is easy to extend @xmath274 to a 3-edge - coloring of @xmath22 by giving a color distinct from @xmath394 and @xmath395 to @xmath383 .
let us now prove that there is a 3-edge - coloring of @xmath22 such that @xmath396 .
let @xmath397 , @xmath398 and @xmath399 .
we claim that @xmath400 is not an edge of @xmath22 . for
if @xmath400 is an edge of @xmath22 , then there exists a vertex @xmath20 in @xmath382 adjacent to both @xmath66 and @xmath71 . since @xmath382 is almost 3-connected and @xmath383 is the only vertex of degree @xmath140 in @xmath382 , @xmath401 is connected , and thus there exists a path @xmath38 between @xmath66 and @xmath71 avoiding @xmath20 and @xmath383 . since @xmath66 and @xmath71 are not adjacent , @xmath38 is of length at least @xmath140 .
naming @xmath1 the vertex of @xmath38 adjacent to @xmath66 , @xmath1 is adjacent to @xmath20 , otherwise @xmath402 $ ] is a claw of @xmath382 , contradicting the fact that @xmath382 is a line graph .
hence @xmath20 has at least three neighbors in the chordless cycle formed by the path @xmath38 and the edges @xmath403 and @xmath404 , a contradiction to the fact that @xmath382 is wheel - free .
so @xmath400 is not an edge of @xmath22 .
let @xmath405 be the graph obtained from @xmath22 by deleting the vertices @xmath406 and @xmath407 and adding the edge @xmath400 .
if an edge @xmath61 distinct from @xmath400 is the chord of a cycle @xmath58 , then since it is not a chord in @xmath22 , @xmath58 must contain @xmath400 . then by replacing @xmath400 by @xmath408
, we deduce that @xmath61 is also the chord of a cycle in @xmath22 , a contradiction .
hence @xmath405 is almost chordless and thus , by theorem [ almostchordless ] , @xmath405 admits a 3-edge - coloring @xmath342 .
assume that @xmath409 .
then setting @xmath410 and @xmath411 , we obtain a 3-edge - coloring of @xmath22 satisfying @xmath412 .
this completes the proof of theorem [ th : col ] . | a _ wheel _ is a graph formed by a chordless cycle @xmath0 and a vertex @xmath1 not in @xmath0 that has at least three neighbors in @xmath0 .
we prove that every 3-connected planar graph that does not contain a wheel as an induced subgraph is either a line graph or has a clique cutset .
we prove that every planar graph that does not contain a wheel as an induced subgraph is 3-colorable .
ams classification : 05c75 |
it is evident to all workers in these fields that the relationship between operator theory and the theory of analytic functions is the source of many deep results . in recent work @xcite of the author
a transfer function , which is in fact a multi - analytic operator , has been introduced in the context of noncommutative markov chains .
these can be thought of as toy models for interaction processes in quantum physics .
the theory of multi - analytic operators , pioneered by popescu @xcite and others in the late 1980 s , has developed into a very successful generalization of the relationship mentioned above .
hence it is natural to expect that noncommutative markov chains and their transfer functions open up a possibility to apply these tools in the study of models in quantum physics .
this paper is the result of an effort to discover the common geometric underpinning which ties together these at first sight rather different settings .
it is found in the tree - like structure of wandering subspaces of row isometries , more precisely : the transfer function describes the relative position of two such trees .
this is worked out in section 1 below .
one of the main results in section 1 is a geometric characterization of pairs of subspaces with a multi - analytic operator as their transfer function .
with this work done we are in a position to discuss the existing applications in a new light which highlights common features . in section 2
we give , from this point of view , an expository treatment of characteristic functions , both the well - known characteristic function of a contraction in the sense of sz.nagy and foias @xcite and the less well - known characteristic function of a lifting introduced by dey and gohm @xcite . in section 3
we explain in the same short but expository style the transfer function of a noncommutative markov chain from @xcite and sketch a generalization which is natural in the setting of this paper .
we hope and expect that this presentation is helpful for operator theorists to find their way into an area of potentially interesting applications .
let @xmath0be a hilbert space and @xmath1 a row isometry on @xmath2 . recall that this means that the @xmath3 are isometries with orthogonal ranges . here
@xmath4 and additionally we also include the possibility of a sequence @xmath5 of such isometries , writing symbolically @xmath6 in this case .
let @xmath7 be the free semigroup with @xmath8 generators ( which we denote @xmath9 ) .
its elements are ( finite ) words in the generators , including the empty word ( which we denote by @xmath10 ) .
the binary operation is concatenation of words .
let @xmath11 , with the @xmath12 , be such a word .
we denote by @xmath13 the length of the word @xmath14 .
further we define @xmath15 ( @xmath16 is the identity operator ) .
by @xmath17 we mean @xmath18 .
we refer to @xcite for further background about this type of multi - variable operator theory .
we want to establish an efficient description of the relative position of pairs of subspaces and their translates under a row isometry @xmath1 on @xmath2 .
suppose @xmath19 and @xmath20 are hilbert spaces and @xmath21 and @xmath22 are isometric embeddings into @xmath2 .
further we write @xmath23 and @xmath24 , similarly @xmath25 and @xmath26 , where @xmath27 . to describe the relative position of @xmath28 and @xmath29 we consider the contraction @xmath30 note that @xmath31 is nothing but the orthogonal projection onto @xmath29 restricted to @xmath28 .
the embeddings introduced above allow us to represent these contractions for varying @xmath32 and @xmath33 on common hilbert spaces @xmath19 and @xmath20 .
[ lem : toeplitz ] @xmath34 for varying @xmath32 and @xmath33 is a _ multi - toeplitz kernel _
, i.e. , @xmath35 such that @xmath36 if @xmath37 then @xmath38 similarly if @xmath39 then @xmath40 otherwise the orthogonality of the ranges of the @xmath41 forces @xmath34 to be @xmath10 .
multi - toeplitz kernels , in the positive definite case , have been investigated by popescu , cf .
@xcite . for more recent developments see also @xcite
. our focus will be on the analytic case , see theorem [ thm : transfer ] below .
let us introduce further notation and terminology .
we define @xmath42 @xmath43 @xmath44 is the smallest closed subspace invariant for all @xmath41 containing @xmath45 , and @xmath46 is invariant for all @xmath47 . a subspace @xmath48 is called _ wandering _ if @xmath49 for @xmath50 ( @xmath51 ) .
we suppose from now on that @xmath45 is wandering . then @xmath52 ( orthogonal direct sum ) , @xmath53 for all @xmath54 and @xmath55 for all @xmath56 .
we can identify the space @xmath44 with @xmath57 , the @xmath19-valued square - summable functions on @xmath7 , in the natural way .
if @xmath58 is also wandering then we can associate a multi - toeplitz operator @xmath59 with a matrix given by the multi - toeplitz kernel @xmath60 from lemma [ lem : toeplitz ] .
in fact , using the identifications of @xmath52 with @xmath57 and of @xmath61 with @xmath62 we see that @xmath63 is nothing but the orthogonal projection onto @xmath64 restricted to @xmath44 .
hence @xmath63 is a contraction which describes the relative position of @xmath44 and @xmath64 .
we are interested in the case where the multi - toeplitz kernel @xmath60 ( resp .
the multi - toeplitz operator @xmath63 ) is _ multi - analytic _
, i.e. , @xmath65 for all @xmath56 .
we note that the notion of multi - analytic operators has been studied in great detail by popescu , cf .
for example @xcite .
the following theorem gives several characterizations of multi - analyticity in our setting .
the notation @xmath66 for the orthogonal projection onto a subspace @xmath67 is used without further comments .
[ thm : transfer ] suppose that @xmath45 is wandering for the row isometry @xmath68 on @xmath2 and let @xmath58 be any subspace of @xmath2 .
then the following assertions are equivalent : * @xmath60 is multi - analytic .
* @xmath69 for all @xmath70 * @xmath71 * @xmath72 for all @xmath73 * @xmath74 for all @xmath70 assertions ( 1)-(5 ) imply the following assertion : * @xmath75 for all @xmath76 and @xmath77 if in addition @xmath58 is also wandering for @xmath68 then ( 6 ) is equivalent to ( 1)-(5 ) and can be rewritten as * @xmath78 for all @xmath76 , + where @xmath79 and @xmath80 are the row shifts obtained by restricting @xmath68 to @xmath44 and @xmath64 and @xmath81 is the multi - toeplitz operator introduced above .
let us describe the relative position of the embedded subspaces @xmath19 and @xmath20 characterized in theorem [ thm : transfer ] by saying that there is an _
orthogonal @xmath20-past_. this terminology is suggested by ( 5 ) and some additional motivation for it is given at the end of this section .
in fact , @xmath83 means exactly that @xmath84 is orthogonal to @xmath58 or , equivalently , that @xmath45 is orthogonal to @xmath85 . @xmath86
. if @xmath87 is not satisfied then there exists @xmath88 and @xmath70 such that @xmath89 .
but then @xmath90 contradicting @xmath91 . @xmath92 . because for @xmath54 @xmath93 we conclude from @xmath71 that @xmath94 , hence @xmath72 .
@xmath95 follows from the fact that @xmath46 is invariant for the @xmath47 and @xmath96 is obvious .
it is elementary that @xmath98 for all @xmath76 . to get ( 6 ) , that is @xmath75 for all @xmath76 and @xmath77
, it is therefore enough to consider all vectors of the form @xmath99 where @xmath88 and @xmath100 is a word which does not begin with @xmath14 and to show that such vectors are always orthogonal to @xmath101 where @xmath77 . by ( 3 )
we have @xmath71 which implies , because @xmath53 for all @xmath54 , that @xmath99 is contained in the span of @xmath46 and of all @xmath102 where the word @xmath103 does not begin with @xmath14 .
this is indeed orthogonal to @xmath101 because @xmath45 is wandering .
conversely we prove , under the additional assumption that @xmath58 is wandering , the implication @xmath104 . if @xmath91 is not satisfied then there exists @xmath105 and @xmath70 such that @xmath106 is not orthogonal to @xmath58 .
hence @xmath107 on the other hand , from @xmath58 wandering , we get @xmath108 and hence @xmath109 .
note that if @xmath58 is not wandering then in general ( 6 ) does not imply ( 1)-(5 ) , in other words ( 6 ) may be true without @xmath60 being multi - analytic .
choose @xmath110 for example .
though in this paper we are mainly interested in pairs of wandering subspaces it is very useful to observe that all the other implications in theorem [ thm : transfer ] hold more general
. for example it can be convenient in applications to start with a bigger subspace @xmath58 and to restrict only later to a suitable wandering subspace . + now consider the following operators : @xmath111 note that the assumption that @xmath45 is wandering is needed to show that the @xmath112 map @xmath19 into @xmath46 . if @xmath60 is multi - analytic then it is determined by these operators .
in fact , it is elementary to check that @xmath113 these formulas suggest an interpretation from the point of view of linear system theory .
( 15,3.5 ) ( 0.0,1)(0,1)2 ( 0.0,1)(1,0)2.9 ( 2.9,1)(0,1)2 ( 0.0,3)(1,0)2.9 ( 4.4,1)(0,1)2 ( 4.4,1)(1,0)3 ( 7.4,1)(0,1)2 ( 4.4,3)(1,0)3 ( 8.9,1)(0,1)2 ( 8.9,1)(1,0)2.8 ( 11.7,1)(0,1)2 ( 8.9,3)(1,0)2.8 ( 4.4,1.7)(-1,0)1.5 ( 6.2,1.7)(-1,0)0.6 ( 8.9,1.7)(-1,0)1.5 ( 0.3,2.0)output space @xmath20 ( 9.2,2.0)input space @xmath19 ( 4.7,2.0)internal space @xmath46 ( 3.5,1.2)@xmath114 ( 5.7,1.2)@xmath115 ( 8.0,1.2)@xmath112 ( 2.7,0.5)(1,0)6.6 ( 2.7,0.5)(0,1)0.5 ( 9.3,0.5)(0,1)0.5 ( 5.8,0.2)@xmath116 in fact , if we interpret @xmath117 as an input then we can think of @xmath118 as a family of outputs originating from it , stored in suitable copies of @xmath20 . motivated by these observations we say , in the case of an orthogonal @xmath20-past , that the associated multi - analytic kernel @xmath60 ( or the multi - analytic operator @xmath63 if available ) is a _ transfer function _ ( for the embedded spaces @xmath19 and @xmath20 ) .
we remark that the scheme is close to the formalism of ball - vinnikov in @xcite , compare for example formula ( 3.3.2 ) therein .
essentially the same construction , but in a commutative - variable setting , appears in @xcite . in the later section on markov chains in this paper
we describe another reappearance of this structure which has been observed in @xcite . for the moment , to make our terminology even more plausible , let us consider the simplest case where @xmath45 and @xmath58 are both wandering and @xmath119 ( i.e. , @xmath68 is a single isometry ) .
let @xmath120 resp .
@xmath121 denote the @xmath19-valued resp .
@xmath20-valued hardy space on the complex unit disc @xmath122 .
for example a function in @xmath120 has the form @xmath123 there is a natural unitary from @xmath124 onto @xmath120 , taking the summands as coefficients ( similar for @xmath20 ) .
it can be used to move operators from one hilbert space to the other . for more details
see for example @xcite , chapter ix .
this allows us to summarize the previous discussions in this special case as follows .
[ cor : transfer ] if @xmath45 and @xmath58 are a pair of wandering subspaces ( for an isometry @xmath68 ) with orthogonal @xmath20-past then @xmath125 moved to the hardy spaces becomes a contractive multiplication operator @xmath126 with @xmath127 here @xmath128 and @xmath129 , the unit ball of the algebra of bounded analytic functions on @xmath122 with values in @xmath130 , the bounded operators from @xmath19 to @xmath20 .
this means that in this case @xmath63 is an analytic operator in the sense of @xcite ( except for the insignificant fact that it operates between different hilbert spaces ) . in the general noncommutative case
we can similarly encode all the entries @xmath131 ( as described above ) into a formal power series which fully describes a multi - analytic operator @xmath63 .
[ cor : transfer2 ] if @xmath45 is a wandering subspace for a row isometry @xmath1 and @xmath58 is another subspace then , with indeterminates @xmath132 which are freely noncommuting among each other but commuting with the operators , @xmath133 where @xmath134 , the transpose indicating that @xmath135 and @xmath136 should be interpreted as ( operator - valued ) column vectors .
further @xmath137 if @xmath138 is a word of length @xmath139 . such a formalism is explained in more detail and used systematically in @xcite .
using the language of system theory we have all the relevant information in the socalled _ system matrix _
@xmath140 let us put these results into the context of other work already done in operator theory and focus on the case @xmath119 again .
we could have extended the isometry @xmath68 to a unitary @xmath141 on a larger hilbert space .
if we now define @xmath142 also for @xmath143 then it is natural to call @xmath144 the @xmath20-past . in this extended
setting the fact that we have orthogonal @xmath20-past ensures that @xmath141 is a coupling in the sense of @xcite , chapter vii.7 , between the right shifts on the orthogonal spaces @xmath145 and @xmath144 .
further our operator @xmath63 can now be interpreted as the contractive intertwining lifting of the zero intertwiner between the two shifts which is canonically associated to the coupling @xmath141 .
see @xcite , chapter vii.8 , for this construction .
we do nt go into this here , the book @xcite contains detailed discussions how analytic functions arise in the classification of such structures .
we remark that in the case @xmath146 it is more complicated to develop the analogue of such a ` two - sided ' setting but this has been worked out in @xcite within a theory of haplitz kernels and cuntz weights .
for the purposes of this paper it turns out that the simpler ` one - sided ' setting , as presented in this section and in particular in theorem [ thm : transfer ] , is sufficient .
the examples in this section are well known and the main emphasis is therefore to show that they fit naturally into the scheme developed in the previous section and that thinking about them in this way simplifies the constructions . for further simplification
we only work through the details of the case @xmath119 , i.e. , a single isometry @xmath147 .
suppose that @xmath45 and @xmath58 are a pair of wandering subspaces with orthogonal @xmath20-past and with system matrix @xmath148 for the adjoint @xmath149 we obtain from the definition of @xmath150 : @xmath151 \oplus i^*_0 p_{{\ensuremath{{\mathcal u}}\xspace}_0 } \big [ vh + j_0(y ) \big]\ ] ] let us consider a special case of the previous setting where @xmath152 .
then @xmath149 is isometric , i.e. , @xmath153 is a coisometry .
conversely , for any hilbert spaces @xmath154 let @xmath155 be any coisometry .
now define the hilbert space @xmath156 with the @xmath157 copies of @xmath19 , the embeddings @xmath158 and an isometry @xmath68 by @xmath159 and acting as a right shift on @xmath160 .
then @xmath45 and @xmath58 are a pair of wandering subspaces with orthogonal @xmath20-past and with system matrix @xmath153 .
in fact , orthogonal @xmath20-past is clear from @xmath71 and theorem [ thm : transfer ] and then @xmath58 wandering is an immediate consequence of @xmath152 and the specific form of @xmath68 .
this situation occurs in the sz.nagy-foias theory of characteristic functions for contractions .
let us sketch briefly how this fits in .
let @xmath161 be a contraction .
then we have defect operators @xmath162 and @xmath163 with defect spaces @xmath164 and @xmath165 defined as the closure of their ranges .
the reader can easily check that the construction above applies with @xmath166 and with the unitary rotation matrix @xmath167 then @xmath68 is the minimal isometric dilation of @xmath168 and the transfer function for the pair @xmath45 and @xmath58 given by @xmath169 is nothing but the well - known sz.nagy-foias _ characteristic function _ of @xmath168 .
in fact it is characteristic in the sense that it characterizes @xmath168 up to unitary equivalence only if @xmath168 is completely non - unitary ( cf .
@xcite or @xcite ) .
so in the general case it may be better to refer to @xmath170 as the transfer function associated to @xmath168 .
it is possible to handle the multi - variable case ( @xmath171 ) , first studied by popescu in @xcite , in a very similar way and the result , if expressed in the notation explained for corollary [ cor : transfer2 ] , is very similar : the transfer function associated to a row contraction @xmath172 is @xmath173 where @xmath174 . it is shown in ( @xcite , 5.4 ) that @xmath170 is characteristic in the sense of being a complete unitary invariant if @xmath168 is completely non - coisometric .
it is further shown in ( @xcite , 5.3.3 ) that to get a complete unitary invariant in the class of completely non - unitary row contractions one can consider a characteristic pair @xmath175 where @xmath176 is a cuntz weight .
but there are other possibilities to obtain a pair of wandering subspaces @xmath45 and @xmath58 with orthogonal @xmath20-past than the scheme explained in the previous example .
we go back to the case @xmath119 and assume again that @xmath156 and that an isometry @xmath68 is given on @xmath2 which acts as a right shift on @xmath124 .
now suppose further that @xmath177 is a subspace of @xmath46 which is @xmath178-invariant .
then for any subspace @xmath58 satisfying @xmath179 it follows that @xmath45 and @xmath58 are a pair of wandering subspaces with orthogonal @xmath20-past .
in fact , because @xmath180 we have for @xmath181 that @xmath182 , but also @xmath183 so that @xmath184 .
hence @xmath185 for all @xmath181 , i.e. , @xmath58 is a wandering subspace .
together with @xmath186 and theorem [ thm : transfer ] this establishes the claim .
this situation occurs in the theory of characteristic functions for liftings ( cf .
@xcite ) . as this is less well known than the sz.nagy-foias theory in the previous example and the presentation in @xcite gives the general case @xmath187 using a different approach and a different notation we think it is instructive to work out explicitly some details of this transfer function in the case @xmath119 with the methods of this paper . as in the previous subsection let @xmath161 be a contraction , @xmath188 , @xmath189 the minimal isometric dilation and we still have @xmath190 and @xmath191 . but now suppose that @xmath192 such that @xmath177 is invariant for @xmath193 , in other words @xmath168 is a block matrix @xmath194 with respect to @xmath192 .
we also say that @xmath161 is a _ lifting _ of @xmath195 .
then @xmath68 is also an isometric dilation of @xmath196 , i.e. , @xmath197 for all @xmath198 , and it restricts to the minimal isometric dilation @xmath199 of @xmath196 on a reducing subspace .
the subspace @xmath177 is invariant for @xmath178 and we obtain a situation as described in the beginning of this subsection by putting @xmath200 and for @xmath201 @xmath202 hence we have orthogonal @xmath20-past and @xmath45 and @xmath58 are both wandering .
it is well known about contractive liftings such as @xmath168 that we always have @xmath203 with a contraction @xmath204 ( cf .
@xcite , chapter iv , lemma 2.1 ) .
we conclude that @xmath205 to compute @xmath206 more explicitly note that for @xmath207 @xmath208 [ the latter because @xmath209 ) ] .
with @xmath210 we can compute @xmath116 as follows : @xmath211 ( using @xmath212 in the last line ) .
hence we get a transfer function @xmath213 the domain of @xmath214 ( for each @xmath215 ) is @xmath216 .
we gain additional insights by evaluating @xmath214 on @xmath217 with @xmath218 and on @xmath219 with @xmath201 .
@xmath220 ( d_r h_r)\end{aligned}\ ] ] which shows that @xmath214 restricted to @xmath221 is nothing but @xmath222 times the transfer function associated to @xmath223 in the sense of sz.nagy and foias , as discussed in the previous subsection .
its presence can be explained by the fact that @xmath68 restricted to @xmath224 also provides an isometric dilation of @xmath223 . for the other restriction @xmath225
we find , using @xmath226 , @xmath227 ( h_s ) \\
& = & \big [ d_{s } - \gamma d_{r^*}\ , \sum_{n\ge 0 } ( z r^*)^n q \big ] ( h_s ) \\ & = & \big [ i - \gamma d_{r^ * } ( i - z r^*)^{-1 } d_{r^ * } \gamma^ * \big ] d_{s } h_s \\\end{aligned}\ ] ] again the multi - variable case ( @xmath228 ) can be handled similarly and yields similar results .
here we investigate a row contraction @xmath229 which is a lifting in the sense that @xmath230 for all @xmath54 .
all the formulas for transfer functions derived above have been written in a form which makes sense and which is still valid for the multi - variable case if we simply replace the variable @xmath215 by @xmath231 ( as in corollary [ cor : transfer2 ] ) and the vectors @xmath232 by @xmath8-tuples of vectors .
these transfer functions have been introduced in @xcite ; see section 4 therein , in particular formulas ( 4.6 ) and ( 4.5 ) , for an alternative approach to the facts sketched above . among other things
it is further investigated in @xcite in which cases such transfer functions are characteristic for the lifting , i.e. , characterize the lifting @xmath168 given @xmath196 up to unitary equivalence .
there is another way how transfer functions as described in section 1 appear in applications , namely in the theory of noncommutative markov chains .
this has been observed in @xcite and to work out a common framework in order to facilitate the discussion of similarities has been a major motivation for this paper . we quickly review the setting of @xcite as far as it is needed to make our point , referring to that paper for more details .
an _ interaction _ is defined as a unitary @xmath233 where @xmath234 are hilbert spaces . in quantum physics it is common to describe the aggregation of different parts by a tensor product of hilbert spaces and in this case we may think of @xmath235 as one step of a discretized interacting dynamics .
( for such an interpretation we may take @xmath236 and think of @xmath237 and @xmath238 as describing the same part before and after the interaction .
but mathematically it is more transparent to treat them as two distinguishable spaces . )
if @xmath46 represents a fixed quantum system , say an atom , and interactions take place with a wave passing by , say a light beam , then it is natural , at least as a toy model , to represent repeated interactions ( @xmath139 steps ) by @xmath239 where the @xmath240 ( resp .
@xmath241 ) are copies of @xmath237 ( resp .
@xmath238 ) , and @xmath242 acts as @xmath235 from @xmath243 to @xmath244 , identical at the other parts .
( 15,3.5 ) ( 0.5,2 ) ( 2.0,2)(-1,0)1 ( 2.0,1)(0,1)2 ( 3.0,1.5)(0,1)1.5 ( 4.0,1.5)(0,1)1.5 ( 5.0,1)(0,1)2 ( 5.1,1 )
( 2.4,1.9)@xmath245 ( 3.4,1.9)@xmath246 ( 4.4,1.9)@xmath247 ( 0,1)atom ( 3.0,1)beam ( 6.8,1)@xmath248 ( 7,1.5)(7.9,2)(8.5,1.5 ) ( 7,1.5)(8.8,3)(10,1.5 ) ( 7,1.5)(9.7,4)(11.5,1.5 ) ( 8.4,1.7)@xmath249 ( 9.6,2)@xmath250 ( 10.8,2.3)@xmath251 choosing unit vectors @xmath252 and @xmath253 we can also form infinite tensor products along these unit vectors and obtain @xmath254 for every @xmath198 on a common hilbert space . such a toy model of quantum repeated interactions can mathematically be thought of as a noncommutative markov chain .
we refer to @xcite for some motivation for this terminology by analogies with classical markov chains .
it is proved in @xcite ( in a slightly different language ) that if we have another unit vector @xmath255 such that @xmath256 ( we call these unit vectors _ vacuum vectors _ in this case ) then we obtain a pair of wandering subspaces @xmath45 and @xmath58 with orthogonal @xmath20-past for a row isometry @xmath68 , notation consistent with section 1 , as follows : @xmath257 i.e. , the latter subspace of @xmath2 is identified with @xmath46 .
the row isometry @xmath68 on @xmath2 is of the form @xmath258 where @xmath259 is the number of elements in an orthonormal basis of @xmath238 .
let @xmath260 be such an orthonormal basis of @xmath261 , fixed from now on .
then for @xmath262 and @xmath263 @xmath264 note that @xmath265 is shifted to the right in the tensor product and appears as @xmath266 on the right hand side .
it is immediate that @xmath68 is a row isometry .
to get used to this definition the reader is invited to verify the formula @xmath267 where @xmath268 with @xmath269 and on the right hand side @xmath265 now appears as @xmath270 .
it becomes clear that the properties of the repeated interaction are encoded into properties of the row isometry @xmath68 .
+ finally we define the pair of embedded subspaces : @xmath271 @xmath272 from the specific form of the isometries @xmath41 it is easy to check that @xmath45 is wandering and that @xmath273 .
the proof that @xmath58 is wandering can be found in @xcite or deduced from proposition [ prop : wandering ] below ( which covers a more general situation ) . from theorem [ thm : transfer ]
we have an associated transfer function which can be made explicit as a multi - analytic kernel @xmath60 or as a ( contractive ) multi - analytic operator @xmath63 .
it may be called a transfer function of the noncommutative markov process . with @xmath274 ( here
we identify @xmath46 with @xmath275 ) we find the operators @xmath276 appearing in the system matrix @xmath153 to be related to the interaction @xmath235 by @xmath277 ( where we have to identify @xmath278 and @xmath20 for the last equation ) .
it is further discussed in @xcite how for models in quantum physics these operators and the coefficients of the transfer function built from them can be interpreted , and it is shown that the transfer function can be used to study questions about observability and about scattering theory ( outgoing cuntz scattering systems @xcite and scattering theory for markov chains @xcite ) .
let us finally indicate that the additional ideas introduced in this paper provide a flexible setting for generalizations .
let us consider the situation above but without assuming the existence of vacuum vectors .
with @xmath252 being an arbitrary unit vector we can easily check that @xmath45 as defined above is still a wandering subspace for @xmath68 .
hence for an arbitrary subspace @xmath58 of @xmath279 we conclude , by theorem [ thm : transfer ] , that we have orthogonal @xmath20-past and there exists a corresponding transfer function corresponding to a multi - analytic kernel @xmath60 .
when is @xmath58 wandering ?
a sufficient criterion generalizing the situation with vacuum vectors is provided by the following [ prop : wandering ] let @xmath177 be a subspace of @xmath46 such that @xmath280 .
then any subspace @xmath281 is wandering .
+ ( here we adapt the convention to omit a tensoring with @xmath282 in the notation . )
let @xmath283 be any vector orthogonal to @xmath284 .
our first observation is that for all @xmath54 the vectors @xmath285 are orthogonal to @xmath284 too . in fact , we can write @xmath286 where @xmath287 with @xmath288 and with @xmath263 orthogonal to @xmath289 and @xmath290 . using the specific form of @xmath41 it follows immediately that @xmath291 is orthogonal to @xmath284 and the same is also true for @xmath292 taking into account the assumption @xmath280 , in the form : @xmath293 is orthogonal to @xmath284 .
the second observation is that for all @xmath54 the vectors @xmath285 are orthogonal to @xmath294 .
as @xmath295 can be approximated by a finite sum @xmath296 with @xmath297 and @xmath298 we may assume for simplicity that @xmath295 is of this form .
but then @xmath299 is orthogonal to @xmath300 and now an application of @xmath301 gives us the result .
1 j. ball , c. sadosky , v. vinnikov , _ scattering systems with several evolutions and multidimensional input / state / output systems .
_ integral equations and operator theory * 52 * ( 2005 ) , 323 - 393 j. ball , v. vinnikov , _ functional models for representations of the cuntz algebra .
_ in : operator theory , system theory and scattering theory : multidimensional generalizations .
operator theory , advances and applications , vol .
* 157 * , birkhuser ( 2005 ) , 1 - 60 j. ball , v. vinnikov , _ lax - phillips scattering and conservative linear systems : a cuntz - algebra multidimensional setting .
_ memoirs of the ams , vol .
* 837 * ( 2005 ) s. dey , r. gohm , _ characteristic functions of liftings . _
journal of operator theory * 65 * ( 2011 ) , 17 - 45 c. foias , a.e .
frazho , _ the commutant lifting approach to interpolation problems . _ operator theory , advances and applications , vol . *
44 * , birkhuser , 1990 .
r. gohm , _ noncommutative markov chains and multi - analytic operators .
_ j. math .
, vol . * 364*(1 ) ( 2009 ) , 275 - 288 b. kmmerer , h. maassen , _ a scattering theory for markov chains . _ inf .
analysis , quantum prob . and related topics , vol.*3 * ( 2000 ) , 161 - 176 g. popescu , _ isometric dilations for infinite sequences of noncommuting operators . _ trans .
amer . math .
316 * ( 1989 ) , 523 - 536 .
g. popescu , _ characteristic functions for infinite sequences of noncommuting operators .
_ j. operator theory * 22 * ( 1989 ) , 51 - 71 .
g. popescu , _ multi - analytic operators on fock spaces .
* 303 * ( 1995 ) , no . 1 , 31 - 46 .
g. popescu , _ structure and entropy for toeplitz kernels .
paris ser .
* 329 * ( 1999 ) , 129 - 134 .
m. rosenblum , j. rovnyak , _ hardy classes and operator theory .
_ oxford university press , 1995 b. sz .- nagy , c. foias , _ harmonic analysis of operators . _ north holland , 1970 during my visit to mumbai in @xmath303 results closely related to proposition [ prop : wandering ] were communicated to me by santanu dey .
these were in the back of my mind when i worked out a version fitting into the setting of this paper .
further i want to thank the referee for constructive comments leading to a better presentation and for pointing out additional links to the existing literature . | to a pair of subspaces wandering with respect to a row isometry we associate a transfer function which in general is multi - toeplitz and in interesting special cases is multi - analytic .
then we describe in an expository way how characteristic functions from operator theory as well as transfer functions from noncommutative markov chains fit into this scheme . |
with the discovery of heavy quark symmetry ( for a review see ref .
@xcite and references therein ) , it has become clear that the study of exclusive semileptonic @xmath1 decays close to zero recoil allows for a reliable determination of the cabibbo
maskawa matrix element @xmath7 , which is free , to a large extent , of hadronic uncertainties @xcite@xcite .
model dependence enters this analysis only at the level of power corrections , which are suppressed by a factor of at least @xmath8 .
these corrections can be investigated in a systematic way using the heavy quark effective theory @xcite .
they are found to be small , of the order of a few per cent . until recently ,
this method to determine @xmath0 was limited by large experimental uncertainties of about 1520% , which were much larger than the theoretical uncertainties in the analysis of symmetry - breaking corrections . however , three collaborations have now presented results of higher precision @xcite@xcite .
it is thus important to reconsider the status of the theoretical analysis , even more so since the original calculation of power corrections in ref .
@xcite has become the subject of controversy @xcite .
the differential decay rate for the process @xmath9 is given by @xcite @xmath10\,|\,v_{cb}|^2\,\eta_a^2\,\widehat\xi^{\,2}(w ) \,,\end{aligned}\ ] ] where @xmath11 denotes the product of the meson velocities .
i have factorized the hadronic form factor for this decay into a short - distance coefficient @xmath12 and a function @xmath3 , which contains the long - distance hadronic dynamics . apart from corrections of order @xmath13
, this function coincides with the isgur wise form factor @xcite .
luke s theorem determines the normalization of @xmath3 at zero recoil ( @xmath14 ) up to second - order power corrections @xcite : @xmath15 the strategy is to obtain the product @xmath16 from a measurement of the differential decay rate , and to extrapolate it to @xmath14 to extract @xmath17 the task of theorists is to provide a reliable calculation of @xmath12 and @xmath18 in order to turn this measurement into a precise determination of @xmath0 .
the short - distance coefficient @xmath12 takes into account a finite renormalization of the axial vector current in the region @xmath19 .
its calculation is a straightforward application of qcd perturbation theory . at the one - loop order
, one finds @xcite @xmath20 the scale of the running coupling constant is not determined at this order .
choosing @xmath21 between @xmath22 and @xmath23 , and using @xmath24 , one obtains values in the range @xmath25 .
the scale ambiguity leads to an uncertainty of order @xmath26 ^ 2\sim 2\%$ ] .
the calculation can be improved by using the renormalization group to resum the leading and next - to - leading logarithms of the type @xmath27^n$ ] , @xmath28^n$ ] , and @xmath29^n$ ] to all orders in perturbation theory @xcite@xcite .
a consistent scheme for a next - to - leading - order calculation of @xmath30 has been developed in ref .
the result is @xmath31 where @xmath32 , and @xmath33 .
the numerical result is very stable under changes of the input parameters . for @xmath34 gev ( for 4 flavours ) and @xmath24 , one obtains @xmath35 .
the uncertainty arising from next - to - next - to - leading corrections is of order @xmath36 . taking this into account ,
i think it is conservative to increase the error by a factor 2.5 and quote @xmath37
hadronic uncertainties enter the determination of @xmath0 at the level of second - order power corrections , which are expected to be of order @xmath38 . for a precision measurement , it is important to understand the structure of these corrections in detail .
falk and myself have derived the exact expression @xcite @xmath39 which depends upon five hadronic parameters : @xmath40 and @xmath41 parametrize the deficit in the `` wave - function overlap '' between @xmath42- and @xmath43-flavoured pseudoscalar ( _ p _ ) and vector ( _ v _ ) mesons , for instance @xmath44 the parameter @xmath45 is proportional to the kinetic energy of the heavy quark inside a heavy meson , @xmath46 gev@xmath47 is determined by the vector
pseudoscalar mass splitting , and @xmath48 parametrizes certain matrix elements containing double insertions of the chromo - magnetic operator . with the exception of @xmath49 , estimates of these parameters
are model - dependent . in ref .
@xcite , we made the simplifying assumptions that @xmath50 , and that the corrections represented by @xmath48 are negligible .
the latter one is based on the observation that these corrections involve a double insertion of an operator that breaks the heavy quark spin symmetry .
using then reasonable values such as @xmath51 gev@xmath47 and @xmath52 gev@xmath47 , one obtains @xmath53 . here and in the following ,
i take @xmath54 gev and @xmath55 gev for the heavy quark masses . in ref .
@xcite , the error in the estimate of @xmath18 has been increased to @xmath56 in order to account for the model dependence and higher - order corrections .
a very similar result , @xmath57 , has been obtained by mannel @xcite .
recently , shifman _
et al_. have suggested an alternative approach to obtain an estimate of @xmath18 , which is based on bounds derived using sum rules and the operator product expansion @xcite .
these bounds imply the inequalities @xmath58 the upper bound for @xmath18 implies that @xmath59 . in ref .
@xcite , this number is quoted as 0.94 . in the same reference , the authors give an `` educated guess '' @xmath60 corresponding to @xmath61 .
however , the arguments presented to support this guess are not very rigorous .
it is possible to combine the above approaches to reduce the theoretical uncertainty in the estimate of @xmath18 @xcite .
the idea is to use the sum rules to constrain the hadronic parameters in ( [ delm2 ] ) in a threefold way : ( i ) the first relation in ( [ sumruls ] ) implies that @xmath62 excluding some of the values for the parameter @xmath63 used in previous analyses .
( ii ) comparing the third relation in ( [ sumruls ] ) with ( [ delm2 ] ) in the limit @xmath64 , one finds that @xmath65 ( iii ) finally , @xmath40 and @xmath41 are correlated in such a way that @xmath66 if @xmath48 is not too large . to illustrate this last point ,
let me define new parameters @xmath67 in terms of these , @xmath68 using the inequalities ( [ sumruls ] ) , one can show that @xmath69 and @xmath70 , with @xcite @xmath71 where @xmath72 .
there are thus three effects , which decrease @xmath18 with respect to the estimate given in ref .
@xcite : a large value of @xmath73 , a positive value of @xmath48 , and the fact that for small @xmath48 the difference @xmath74 is centred around @xmath75 ( i.e. @xmath76 is centred around 0 ) .
= 9.5 cm in evaluating ( [ final ] ) , i will take @xmath77 gev@xmath47 , which is consistent with the bound in ( [ lam1bou ] ) and with the value @xmath78 gev@xmath47 obtained from qcd sum rules @xcite .
varying @xmath79 in the range between 0.36 and 0.5 gev@xmath47 does not alter the results very much .
the main uncertainty comes from the unknown values of the parameters @xmath80 and @xmath48 . as a guideline
, one may employ the constituent quark model of isgur _
et al_. @xcite , in which one uses non - relativistic harmonic oscillator wave functions for the ground - state heavy mesons , for instance @xmath81 , where @xmath82 is the reduced mass .
one then obtains @xmath83 gev@xmath47 , where i use @xmath84 gev for the light constituent quark mass .
however , this estimate of @xmath80 is probably somewhat too low .
lattice studies of heavy - light wave functions suggest an exponential behaviour of the form @xmath85 @xcite , which leads to @xmath86 gev@xmath47 .
values much larger than this are unlikely , since i use a rather large constituent quark mass @xmath87 .
in fact , adopting the point of view that the sum rules for @xmath40 and @xmath41 are saturated to approximately 50% by the ground - state contribution @xcite , one would expect @xmath88 gev@xmath47 , which seems a very reasonable value to me . in fig .
[ fig1 ] , i show the allowed regions for @xmath18 as a function of @xmath48 for two values of @xmath80 .
i think it is reasonable to assume that @xmath48 is of a magnitude similar to @xmath49 or smaller .
thus , i conclude that for all reasonable choices of parameters the results are in the range @xmath89 which is consistent with the previous estimates in refs .
@xcite at the @xmath90 level
. a more precise determination of the parameter @xmath80 would help to reduce the uncertainty in this number .
in the extrapolation of the differential decay rate ( [ bdrate ] ) to zero recoil , the slope of the function @xmath3 close to @xmath14 plays an important role . one defines a parameter @xmath91 by @xmath92 it is important to distinguish @xmath91 from the slope parameter @xmath93 of the isgur wise function .
they differ by corrections that break the heavy quark symmetry . whereas the slope of the isgur wise function is a universal , mass - independent parameter , the slope of the physical form factor depends on logarithms and inverse powers of the heavy quark masses . on the other hand
, @xmath91 is an observable quantity , while the value of @xmath93 depends on the renormalization scheme . to illustrate this last point ,
let me neglect for the moment @xmath13 corrections and work in the leading logarithmic approximation .
then the relation between the physical slope parameter @xmath91 and the slope parameter @xmath94 of the regularized isgur
wise function is @xcite @xmath95 where @xmath96 is the renormalization scale , and @xmath97 is an undetermined ( at this order ) scale between @xmath98 and @xmath99 . the last equation can be used to define the @xmath96-independent slope of the renormalized isgur wise function ( see ref .
@xcite for the generalization of this definition to next - to - leading order ) . if next - to - leading logarithmic corrections are taken into account , the scale ambiguity related to the choice of @xmath97 is resolved , and one obtains @xmath100 the @xmath13 corrections to this relation have been investigated and are found to be negative . however , any such estimate is model - dependent and thus has a large theoretical uncertainty .
the result is @xmath101 @xcite .
predictions for the renormalized slope parameter @xmath93 are available from qcd sum rules including a next - to - leading - order renormalization - group improvement .
one obtains @xmath102 @xcite .
i thus expect @xmath103
using the updated values @xmath104 and @xmath105 , i obtain for the normalization of the hadronic form factor at zero recoil : @xmath106 three experiments have recently presented new measurements of the product @xmath107 . when rescaled using the new lifetime values @xmath108
ps and @xmath109 ps @xcite , the results obtained from a linear fit to the data are @xmath110 where the first error is statistical and the second systematic .
i will follow the suggestion of ref .
@xcite and add @xmath111 to these values to account for the curvature of the function @xmath3 . taking the weighted average of the experimental results and using the theoretical prediction ( [ etaxi ] ) ,
i then obtain @xmath112 which corresponds to a model - independent measurement of @xmath0 with 7% accuracy .
this is by far the most accurate determination to date . neglecting @xmath13 corrections ,
i have related the physical slope parameter @xmath91 to the slope of the isgur wise function and obtain the prediction @xmath4 .
it compares well with the average value observed by experiments , which is @xmath113 @xcite@xcite .
t. browder ( cleo collaboration ) , talk presented at the 27th international conference on high energy physics ( ichep94 ) , glasgow , scotland , july 1994 ; b. barish _
et al_. ( cleo collaboration ) , cornell preprint clns 94/1285 , submitted to ichep94 . | i discuss the theoretical uncertainties in the extraction of @xmath0 from a measurement of the @xmath1 decay rate close to zero recoil .
in particular , i combine previous estimates of the @xmath2 corrections to the normalization of the hadronic form factor at zero recoil with sum rules derived by shifman _
et al_. to obtain a new prediction with less uncertainty .
i also give a prediction for the slope of the form factor @xmath3 at zero recoil : @xmath4 . using the most recent experimental results ,
i obtain the model - independent value @xmath5 .
epsf cern - th.7396/94 + hep - ph/9409306 * theoretical uncertainties in the + extraction of @xmath6 + from @xmath1 decays near zero recoil * matthias neubert + _ theory division , cern , ch-1211 geneva 23 , switzerland + ( e - mail : [email protected])_ _ invited talk presented at the 27th international conference on high energy physics + glasgow , scotland , 2027 july 1994 _
cern - th.7396/94 + september 1994 |
let us discuss the case of superposition of @xmath2 number of pure states of a qudit . considering a @xmath0-dimensional referential state @xmath111 ,
whose overlap ( magnitude ) with each of the constituent state is known .
therefore , assuming @xmath112 , where @xmath155 .
let @xmath104 be the desired weights for creating superposition of @xmath0-dimensional states @xmath105 , @xmath106 , @xmath107 @xmath108 respectively .
we begin with the initial state , @xmath156 where @xmath120 is the normalization factor , which is equal to @xmath157 .
this state belongs to a @xmath158 dimensional hilbert space , where the primed coefficients are , @xmath159 this initial state is then made to undergo a series of controlled - swap operations , @xmath160 where state of first spin acts as control . in order to describe the action of this operation , let us reconsider the set of bases vectors of the control spin , ( @xmath161 , @xmath162 ) in @xmath2-dimensional hilbert space , whenever first qu@xmath2it is in state @xmath161 , states of second and the @xmath163 qudits get swapped .
the resulting state is of the form , @xmath164 this is then acted upon by a set of projection operators constructed using the referential state @xmath111 .
operator performing @xmath165 number of projections on qudits numbered 3 to @xmath2 is given as , @xmath166 , where @xmath167 represents the qudit number .
this helps to remove the global phases that may be occurring with the constituent states ( @xmath168 s ) .
the resulting state is given as , @xmath169 tracing out states of qudits numbered 3 to @xmath170 , we are left with a @xmath114-dimensional state . also , shedding the global phases , the state in eq .
( [ eq15 ] ) is written in a simple manner , @xmath171 where @xmath172 . in case of superposition of two qubits with weights @xmath127 and @xmath128 , above equation
is reduced to , @xmath173 which is the expected final state after superposing two single - qubits , the additional factor @xmath174 .
thus we reduce the existing three - qubit based protocol described in @xcite to the present two - qubit based protocol described in the maintext .
it is to be noted that the state in eq .
( [ eq12 ] ) has already taken care of the global phases of states ( @xmath29 and @xmath30 ) .
[ [ b - comparison - between - general - two - qubit - and - three - qubit - based - protocols - appb ] ] ( b ) comparison between general two - qubit and three - qubit based protocols [ appb ] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ in this section , we discuss the comparison between the present scheme and the previously proposed scheme in @xcite to carry out the superposition of two single - qubit states .
present scheme ( described in the maintext ) for superposition can be obtained from the three - qubit based scheme as described in the encoding scheme ( section ( a ) ) , giving rise to the superposed state , @xmath175 the success probability in this case is given as @xmath176 . here
@xmath49 is the normalization factor for state @xmath177 ( where @xmath178 ) .
recalling the treatment in a three - qubit based protocol @xcite , the resultant state in that case is given as , @xmath179 the success probability in this case , @xmath180 . comparing the success probabilities resulting from these two protocols , we have , @xmath181 where @xmath131 , @xmath132 , and @xmath133 .
same value of success probabilities ( @xmath182 and @xmath183 ) result , in case the overlaps , @xmath184 or the superposition is obtained with equal weights , i.e. @xmath185 .
figure [ plot ] shows the variation @xmath186 vs @xmath187 at different values of @xmath140 .
it is interesting to note that our two - qubit based protocol outperforms the three - qubit based protocol ( in terms of success probabilities ) in the range @xmath135 ( when @xmath136 ) and in the range @xmath137 ( when @xmath138 ) .
some of the examples of pair of states following these criteria for weighted superposition are discussed in the maintext .
is shown with the ratio of overlaps , @xmath188 corresponding to different values of @xmath140 .
different curves correspond to different values of @xmath140 , that are specified on the right side of the plot .
[ plot ] ] we re - consider the problem of superposition of two arbitrary single qubit states with known non - zero overlaps , @xmath143 and @xmath144 with the referential single - qubit state @xmath34 . thus one can obtain the overlaps of the constituent states with @xmath100 ( single - qubit state orthogonal to @xmath34 ) .
we have , @xmath145 and @xmath146 .
let us begin with a three - qubit initial state , similar to the one given in eq .
( [ eq21 ] ) , @xmath189 this state is then acted upon by the same three - qubit controlled - swap operation as described in section ( a ) , such that the resulting state is , @xmath190 consider the action of the identity operator @xmath191 on the three - qubit state given in eq .
( [ eq31 ] ) .
the resultant state is given as , @xmath192 \otimes \vert \chi \rangle \nonumber \\ & & + \left [ a \langle \chi^{\bot } \vert \psi_2 \rangle \vert 0 \rangle \vert \psi_1 \rangle + b \langle \chi^{\bot } \vert \psi_1 \rangle \vert 1 \rangle \vert \psi_2 \rangle \right ] \otimes \vert \chi^{\bot } \rangle .
\nonumber \\
\label{eq33 } \end{aligned}\ ] ] another controlled unitary operation is implemented on the first qubit , where state of third qubit acts as control .
subject to the state of the third qubit ( @xmath34 or @xmath100 ) , the action of this controlled operation is described ( on the first qubit ) as , @xmath193 eq . ( [ eq33 ] ) thus leads to , @xmath194 application of the projection operator , @xmath195 then leads to , @xmath196 thus we obtain the weighted superpositions of single - qubit states @xmath29 and @xmath30 .
if state of second qubit here is @xmath34 , the superposed state , @xmath197 is obtained with a success probability , @xmath198 . while corresponding to second - qubit state @xmath100 , the superposed state , @xmath199 is resulted with a success probability , @xmath200 .
@xmath201 and @xmath202 are the normalization factors of the first qubit state when states of the second qubit are @xmath34 and @xmath100 respectively in eq .
( [ eq36 ] ) .
states given in eqs .
( [ eq37 ] ) and ( [ eq38 ] ) are weighted superpositions of the same constituent states @xmath29 and @xmath30 .
but they may be different because of their possibly different relative phases .
the situation of our interest arises when @xmath203 varies from @xmath204 only upto a global case .
following are few special cases discussing such scenarios .
[ [ both - the - states - belong - to - same - longitudinal - plane - on - the - bloch - sphere ] ] both the states belong to same longitudinal plane on the bloch sphere ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ assume now that both @xmath29 and @xmath30 lie in the same longitudinal plane on the bloch sphere as shown in fig . [ sphere ] .
more explicitly , for @xmath205 , eq . ( [ eq36 ] ) takes the form , @xmath206 tracing out the second qubit , we obtain , @xmath207 which is the desired superposed state .
this superposed state is obtained with success probability , @xmath208 , where @xmath209 and @xmath210 .
putting another constraint , @xmath148 , we obtain @xmath211 which gives rise to the desired superposed state with a success probability , @xmath212 [ [ both - the - states - belong - to - same - transverse - plane - on - the - bloch - sphere ] ] both the states belong to same transverse plane on the bloch sphere ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ in this case , we have @xmath150 , which implies @xmath151 .
( [ eq35 ] ) thus leads to , @xmath213 further , assuming both the states occupying diametrically opposite locations on the bloch sphere , the azimuthal angles of the two states may be considered as @xmath152 and @xmath153 . under the action of projection operator , @xmath214
( [ eq41 ] ) gives rise to the desired superposed state , @xmath215 with a success probability , @xmath216 .
note that with the projection operator , @xmath217 eq .
( [ eq41 ] ) gives rise to the desired superposed state , @xmath218 with a success probability , @xmath219 .
the total success probability obtained in above two instances , @xmath220 thus for arbitrary orthogonal input states , @xmath221 . | the principle of superposition is an intriguing feature of quantum mechanics , which is regularly exploited at various instances .
a recent work [ prl * 116 * , 110403 ( 2016 ) ] shows that the fundamentals of quantum mechanics restrict the superposition of two arbitrary pure states of a quantum system , even though it is possible to superpose two quantum states with partial prior knowledge .
the prior knowledge imposes geometrical constraints on the choice of input pure states .
we discuss an experimentally feasible protocol to superpose multiple pure states of a @xmath0 dimensional quantum system and carry out an explicit experimental realization to superpose two single - qubit pure states on a two - qubit nmr quantum information processor .
_ introduction : _ postulates of quantum theory seems to have allowed superposition of an arbitrary pair of pure states of a quantum system unless there is any super - selection rule @xcite .
but recent study shows that there exists no protocol allowed by quantum mechanics to create superposition of a completely unknown pair of pure quantum states with specified coefficients @xcite .
quantum states which are equivalent to each other upto a global phase , represent the same physical state ( @xmath1 ) .
therefore the addition of unknown quantum states with varying global phases , may develop a relative phase between these states , thus changing the resultant superposed state .
however some partial prior knowledge about the states can be used to achieve the restricted type of superposition as suggested in a recent work @xcite . as per this protocol , two unknown quantum states are superposed , each of which has a fixed non - zero overlap with a known referential pure state .
experimental implementation of such protocol is interesting , not only as a quantum computational task , but also as a fundamental principle .
there has always been a growing interest for more feasible , and robust experimental quantum computation models @xcite . in this regard , attempts have been made using different experimental systems such as photons @xcite , nuclear spins @xcite , and super - conducting qubits @xcite .
implementation using photonic qubits @xcite , is based on a system of two photons , wherein controlled - swap implementation is considered to be a challenge .
therefore an effective controlled - swap operation is implemented which includes post - selection , and is a non - unitary operation .
another attempt has been made using three nuclear spins , wherein controlled - swap gate is implemented via numerically optimized pulses while the projection operator implementations are avoided by theoretically performing the partial trace operations @xcite .
an approximate quantum adder is realized using superconducting qubits on an ibm quantum computer @xcite .
this scheme implements an optimal quantum circuit obtained using genetic algorithm techniques , but its operation is limited to a specific region of the hilbert space of two - qubits .
present work proposes a fully experimentally implementable protocol to perform the desired superpositions of pure states of a quantum system .
moreover , this is a two - qubit based scheme to superpose two single - qubit states , in contrary to the existing three - qubit based implementation @xcite .
the extension of this scheme to superpose @xmath2 number of higher - dimensional quantum states is also discussed .
we also analyse the enhancement in success probabilities in this scheme for different prior informations . _
theoretical scheme : _ beginning with an explicit analysis for the superposition of two single - qubit pure states , we consider a system of two coupled spin @xmath3 particles ( denoted here as a and x ) under the action of a hamiltonian @xmath4 where @xmath5 , @xmath6 correspond to resonance frequencies of the first and second spins respectively , and @xmath7 represents the scalar coupling constant .
@xmath8 and @xmath9 are the @xmath10 components of angular momenta for spins @xmath11 and @xmath12 respectively .
@xmath13 ( @xmath14 ) are the eigen vectors of @xmath8 ( @xmath9 ) corresponding to eigenvalues @xmath15 respectively .
the single qubit pure states of our system are encoded in the eigenbasis @xmath16 of the hamiltonian @xmath17 .
we use the eigenstates @xmath18 , @xmath19 of @xmath17 to store the single - qubit input state of the system , @xmath20 , where @xmath21 , while remaining two levels are used to store the input state of the system , @xmath22 , where @xmath23 .
the state of the two - qubit system ( @xmath24 ) is represented here as ( assuming @xmath25 , where @xmath26 and @xmath27 are the weights for superposition ) , @xmath28 thus , we use a system of two - qubits to encode the single qubit states @xmath29 and @xmath30 .
in eq .
( [ coded ] ) , the first qubit is considered as ancilla , and the second qubit is considered as system - qubit . assuming the global phases @xmath31 and @xmath32 with states @xmath29 and @xmath30 respectively , a more general state of this two - qubit system may be taken as , @xmath33 given any fixed state @xmath34 of the system qubit , prior knowledge of the non - zero overlaps @xmath35 and @xmath36
is exploited to find the phases : @xmath37 and @xmath38 . using this information ,
we construct a phase gate ( @xmath39 ) , that implements a @xmath40rotation on the first qubit by angle @xmath41 , leading to the state , @xmath42 thus the global phases of the individual single - qubit states are modified , and appears as a global phase of the two - qubit state . _ supplementary material _
@xcite contains an alternative protocol to encode the states @xmath29 , @xmath30 and to get rid of the global phases @xmath31 , @xmath32 respectively .
further , a hadamard gate implementation on the first - qubit in eq .
( [ eq3 ] ) leads to the state ( ignoring the global phase @xmath43 ) , @xmath44 depending upon the state of the first qubit , one can choose between the sum or difference of the single - qubit states @xmath29 and @xmath30 : a measurement on the first qubit in the basis @xmath45 gives rise to the state , @xmath46 of the second qubit ( in case of outcome @xmath47 ) which is the desired superposed state , @xmath48 ( @xmath49 being the normalization constant ) , obtained with a success probability @xmath50 .
thus , with the help of only one ancillary qubit , we are able to superpose two single - qubit states .
+ _ experimental implementation : _ the nmr pulse sequence to carry out weighted superposition of two single - qubit states is shown in fig
. [ pp ] , where the first channel corresponds to the ancillary - qubit @xmath11 and the second channel corresponds to the system qubit @xmath12 ( here labeled as @xmath51 and @xmath52 respectively ) , initialized in state @xmath53 .
a single qubit rotation by angle @xmath54 about @xmath55axis is applied on the ancillary qubit , which gives rise to the state @xmath56 ( with @xmath57 and @xmath58 ) .
this is followed by two two - qubit controlled operations , that encode second qubit with state @xmath29 , when first qubit is in state @xmath47 and with state @xmath30 when first qubit is in state @xmath59 .
each controlled - operation is achieved by a controlled - rotation of second - qubit by angle @xmath60 where state of the first qubit , @xmath61 ( @xmath62 ) is the control . the axis of rotation , @xmath63 .
possible global phases of states @xmath29 and @xmath30 are taken care of by applying a @xmath40pulse of angle @xmath64 on the first qubit .
this is followed by pseudo - hadamard gate implementation on the first qubit .
a partial read out of the system qubit leads to the expected superposed state .
in all the experiments , the referential state ( @xmath34 ) is chosen as @xmath47 . .
two channels show the operations on ancilla ( @xmath51 ) and system qubits ( @xmath52 ) respectively .
pulse sequence is divided into various steps , which are numbered as @xmath65 , @xmath66 , @xmath67 and @xmath68 in the end .
radio - frequency pulses are shown as rectangles , with respective angles of rotations mentioned at the top and the axes of rotations specified at the bottom .
the arbitrary axes of rotations are defined as : @xmath69 , @xmath70 , @xmath71 , and @xmath72 . at the end of the sequence
, a single - qubit measurement is performed on the system qubit .
[ pp ] ] as discussed in the theoretical scheme , measurement is composed of a projective measurement on the first qubit ( @xmath73 ) , followed by a partial - trace operation that retains the state of the second qubit . thus measurement here in fact selects the subspace spanned by eigenvectors @xmath74 , @xmath75 of @xmath17 .
experimentally , this is realized by partial quantum state tomography of the system qubit ( @xmath52 ) .
here we do this by observing nmr transition between levels @xmath53 and @xmath76 ( of @xmath52 ) .
the desired single - qubit density operator is obtained by a set of two operations : ( i ) direct readout , and ( ii ) @xmath77 ( gradient operation followed by a @xmath78 pulse about @xmath55axis on the second qubit ) .
the resultant single - qubit density operator is un - normalized , which is expected as per the protocol .
normalization constant for the desired part of the density operator is obtained experimentally by measuring the relative populations .
this is achieved by applying a gradient to dephase the coherences , followed by a spin - selective @xmath78 pulse on the first qubit ( @xmath79 ) . a readout of the resultant nmr spectrum of the first - qubit provides the normalization constant for the desired subspace .
this normalization factor is then used to completely characterize the final state density operator of the superposed state .
pulse sequence shown in fig .
[ pp ] is implemented experimentally on a sample consisting of @xmath52 labeled chloroform in deutrated acetone .
the experiments are performed on 500 mhz bruker avance ii nmr spectrometer mounted with qxi probehead .
all the pulses used are high power , short duration rf pulses on @xmath51 and @xmath52 channels respectively .
both @xmath51 and @xmath52 spins in this doubly rotating frame are chosen to be on - resonance .
spin - spin relaxation times ( @xmath80 ) of @xmath51 and @xmath52 spins are estimated to be 540ms and 170ms respectively .
nuclear spin - systems at thermal equilibrium constitute a mixed state .
the system is thus initialized in a pseudo - pure state , @xmath53 , which is obtained by spatial averaging technique @xcite with a fidelity of @xmath81 . starting from this pseudo - pure state , various pairs of single - qubit states ( @xmath29 and @xmath30 )
are encoded on a two - qubit system ( as described earlier ) . in order to ensure the accuracy of this experimental implementation ,
two - qubit density operators are tomographed at the end of step @xmath66 and @xmath68 of the pulse sequence ( fig .
[ pp ] ) , thus obtaining the state after encoding ( @xmath82 ) and the state before the measurement ( @xmath83 ) respectively .
a two - qubit state is completely constructed with a set of four operations : @xmath84 , where @xmath85 refers to spin - selective @xmath78 pulse along @xmath86-axis .
the fidelity between the theoretically expected ( @xmath87 ) and the experimentally obtained ( @xmath88 ) states are measured using the following expression , @xmath89 the experimentally superposed state is obtained by tomography of the subspace spanned by eigenvectors @xmath90 .
the results of various experiments are given in table [ table1 ] , where columns 2 and 3 show the single qubit pure states to be superposed , and column 5 contains the fidelity ( @xmath91 ) with which experimentally superposed states match with the theoretically expected ones . in datasets numbered 1 - 4 of table [ table1 ] , we have , @xmath92 , and @xmath93 , with @xmath94 .
each of these pairs corresponds to the same two conic - sections as per their bloch sphere representations .
similarly , datasets numbered 5 and 6 of the table [ table1 ] show the superposition between two pairs of states from same respective conic - sections . a detailed tomographic analysis corresponding to dataset numbered 3 ( table [ table1 ] ) is shown in fig .
[ tomo ] .
we also perform the superposition of the same constituent states with different weights , as given in datasets numbered 5 , 7 and 8 of table [ table1 ] . for completeness , the experiments are performed with assumed global phases of the input states
. global phase of a quantum state is not physically realizable . in the set of experiments described in datasets 9 and 10 of table [ table1 ]
, global phases are intentionally introduced while encoding the states @xmath29 and @xmath30 . in fig .
[ pp ] , the pulse of angle @xmath54 is applied about an axis @xmath95 that is aligned with @xmath55axis by an angle @xmath96 .
the encoded state is thus of the form , @xmath97 .
experiments are performed for two pairs of states shown in datasets numbered 9 and 10 in table [ table1 ] . in both of these cases , @xmath98 and rest of the parameters
are same as that of set of states numbered 5 and 6 in table [ table1 ] .
now compare the datasets 5 with 9 and 6 with 10 ( table [ table1 ] ) .
as expected , presence or absence of the global phase does not affect the final superposed state .
efficacy of this experimental scheme does not directly depend upon the values of the overlaps ( @xmath99 ) .
this is signified by the dataset 11 of table [ table1 ] , where @xmath30 is very close to @xmath100 ( orthogonal to @xmath34 ) .
table [ table1 ] shows that even if we choose the pair of input states ( @xmath101 ) outside the set @xmath102 , still the aforesaid method of superposition does give rise to the expected superposed state @xmath103 with quite high accuracy .
.various experimental outcomes .
[ table1 ] [ cols="<,<,<,<,<",options="header " , ] _ superposition of multiple qudits : _ the aforesaid procedure is extended to superpose arbitrary pure states of @xmath2 number of qudits ( @xmath0-dimensional quantum system ) .
let @xmath104 be the desired coefficients for creating superposition of @xmath2 number of states @xmath105 , @xmath106 , @xmath107 @xmath108 of the @xmath0-dimensional system .
this requires a hybrid @xmath109dimensional qu@xmath2it - qudit system , where the qu@xmath2it ( @xmath110dimensional quantum system ) acts as an ancilla ( as before ) and the qudit acts as the system .
consider now a @xmath0-dimensional referential state @xmath111 , whose non - zero overlaps , @xmath112 , ( @xmath113 ) are known . following the same protocol as before ,
states of each of @xmath2 individual qudits are encoded on the @xmath114 number of basis vectors of the hybrid qu@xmath2it - qudit system : @xmath115 where @xmath116 .
the global phase of the constituent states are taken care of by using the information of overlaps of respective constituent states with the referential state @xcite .
this is then followed by fourier transformation of the qu@xmath2it , which is in fact the generalization of the hadamard operation in higher - dimensions .
the resultant state , which is a generalization of two - qubit state in eq .
( [ eq6 ] ) , is given as , @xmath117 where @xmath118 , is the @xmath119 root of unity and @xmath120 is the normalization constant .
an arbitrary superposition of @xmath2 pure states of a qudit is then obtained by the projective measurement @xmath121 subsequently tracing out the state of first party ( qu@xmath2it ) .
the final state is a superposition of @xmath2 number of @xmath0-dimensional states , which alongwith the information of global phases of the constituent ( @xmath2-qudits ) states is given as @xcite , @xmath122 where @xmath123 is a constant that normalizes the un - normalized state obtained after the projective measurement .
the superposed state @xmath124 ( eq . ( [ eq16 ] ) )
is obtained with a success probability , @xmath125 + _ discussion : _ the superposed state of two qubits obtained here with a success probability , @xmath126 $ ] ( from eq .
( [ prob ] ) , taking @xmath127 and @xmath128 ) is compared with the success probability of the existing superposition protocol : @xmath129 @xcite for the case of superposition of two single - qubits .
for comparison of these two protocols , we take the ratio of their respective success probabilities @xcite , @xmath130 where @xmath131 , @xmath132 , and so @xmath133 .
it is interesting to note that @xmath134 in the range @xmath135 ( when @xmath136 ) and in the range @xmath137 ( when @xmath138 ) .
two such pairs of states satisfying these criteria are shown in datasets 7 and 8 of table [ table1 ] . in both of these datasets ,
@xmath139 , while the values of @xmath140 are chosen as @xmath141 and @xmath142 respectively . in general
, there is an interplay between the success probability with which the desired superposed state is obtained and the amount of prior information of the constituent states .
we impose certain constraints on the constituent states and observe its impact on the success probabilities . re - considering the problem of superposition of two single qubit states having fixed non - zero overlaps , @xmath143 and @xmath144 with the referential single - qubit state @xmath34 .
thus one can obtain the overlaps of the constituent states with @xmath100 ( single - qubit state orthogonal to @xmath34 ) .
we have , @xmath145 and @xmath146 . therefore one can use the overlap of the input states with @xmath34 and @xmath100 to increase the success probability of superposition for certain class of input states @xcite .
we mainly study two types of constraints ; @xmath65 both @xmath29 and @xmath30 lie in the same longitudinal plane of the bloch sphere .
the desired superposed state is obtained with success probability , @xmath147 for @xmath148 , the success probability , @xmath149 , becomes double to that of the ordinary case .
@xmath66 both the states belong to same transverse plane on the bloch sphere . in this case , we have @xmath150 , which implies @xmath151 .
further , assuming both the states occupying diametrically opposite locations on the bloch sphere , the azimuthal angles of the two states may be considered as @xmath152 and @xmath153 . the total success probability obtained under these conditions is given by : @xmath154 in case both the states lie in the equatorial plane , i.e. arbitrary pair of orthogonal input states , the success probability is @xmath3 .
+ _ conclusions : _ we experimentally create superposition of single - qubit states in the defined framework , covering all possible aspects , i.e. ( i ) creation of various single - qubit states and obtaining their superposition , ( ii ) superposition with arbitrary weights , and ( iii ) superposition of single - qubit states in the presence of assumed global phases . all the experimental results are obtained with good fidelities .
this protocol is also extended for the superposition of multiple states of a qudit .
moreover special cases are discussed where one can obtain the desired superposed state with an enhanced value of the success probability . + _ acknowledgements : _ sd acknowledges the financial support by the institute of mathematical sciences chennai india , technische universitt dortmund germany , and support by the international collaborative research centre trr 160 coherent manipulation of interacting spin excitations in tailored semiconductors , " funded by the deutsche forschungsgemeinschaft .
all the experiments were performed in nmr lab , technische universitt dortmund germany .
sg , gt , and sd would like to thank somshubhro bandyopadhyay , manik banik , prathik cherian j. , guruprasad kar , samir kunkri , and ramij rahaman for useful discussions .
14ifxundefined [ 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 @noop _
_
( , , ) @noop ( ) , link:\doibase 10.1103/physrevlett.116.110403 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop _ _ , quantum information , decoherence , quantum computation , cryptography ( , , ) @noop _ _ ( , ) @noop _ _ ( , ) @noop * * , ( ) link:\doibase 10.1103/physreva.94.033844 [ * * , ( ) ] @noop ( ) , @noop ( ) , @noop \doibase http://dx.doi.org/10.1016/s0167-2789(98)00046-3 [ * * , ( ) ] |
following @xcite we make the following definitions : given a hyperbolic @xmath1-orbifold @xmath2 with cusps @xmath3 we say that cusps @xmath4 are _ geometrically isolated _ from cusps @xmath5 if any deformation of the hyperbolic structure on @xmath2 induced by dehn filling cusps @xmath6 while keeping cusps @xmath7 complete does not change the euclidean structure at cusps @xmath4 .
we say that cusps @xmath8 are _ first order isolated _ from cusps @xmath9 if the map from the space of deformations induced by dehn filling cusps @xmath6 while keeping cusps @xmath7 complete to the space of euclidean structures at cusps @xmath10 has zero derivative at the point corresponding to the complete structure .
we say that cusps @xmath8 are _ strongly geometrically isolated _ from cusps @xmath9 if for any fixed dehn filling on cusps @xmath8 , the geometry of the ( possibly filled ) cusps @xmath8 is unchanged by any dehn filling on cusps @xmath9 .
it is immediate from these definitions that strong isolation implies geometric isolation , which in turn implies first order isolation .
in @xcite it is shown that first order isolation and strong isolation are symmetric conditions in the sets @xmath7 and @xmath9 , and that they can be given an analytic definition in terms of the @xmath11 function defined in @xcite .
neumann and reid in @xcite construct a @xmath0-cusped hyperbolic orbifold @xmath12 with geometrically isolated cusps .
based on numerical evidence , they conjectured that the cusps are strongly isolated .
the orbifold @xmath12 is particularly interesting for two reasons : firstly it is arithmetic , and secondly , all other known examples of manifolds with strongly isolated cusps contain a rigid totally geodesic separating surface which forces " the isolation . there is evidence that this example does not contain such a surface , based on the fact that its volume is so small . in this paper
we find explicit formulae describing how the tetrahedra making up an ideal triangulation of @xmath12 are distorted as the hyperbolic structure on @xmath12 is deformed which suggests that the existence of such a surface is unlikely .
@xmath13 is a @xmath14cusp link complement in @xmath15 .
@xmath12 is obtained from @xmath13 by @xmath16 dehn surgery on two of the links .
it is a curious fact , which we will exploit in our calculations , that @xmath13 is a double cover of @xmath12 , obtained by a twofold branching of @xmath15 over the two filled cusps .
we can draw @xmath13 as a @xmath17 complement .
the two removed cusps are linked as in figure 2 .
the two cusps in the figure we label as @xmath18 and @xmath19 .
@xmath20 is the cusp @xmath21 , and @xmath22 the cusp @xmath23 . as in @xcite
we let @xmath24 and @xmath25 denote the logarithm of the holonomy of @xmath26 and @xmath27 , the meridian and longitude respectively of cusp @xmath28 .
rotating through an angle @xmath29 gives the same link complement after isotopy .
in fact , we can see that @xmath13 retains its @xmath14fold symmetry even after cusps @xmath18 and @xmath19 are @xmath30filled .
hence the cusp shapes at @xmath21 and @xmath31 are constant . by taking the quotient
we see that in @xmath12 the two cusps are geometrically isolated .
note that this technique can be generalized .
let @xmath2 be any @xmath14fold symmetric @xmath17 2link complement for which there is an orientation preserving isometry permuting each pair of cusps .
quotienting by this isometry we obtain an orbifold in which one cusp is geometrically isolated from the other .
alternatively , let @xmath2 be any @xmath14fold symmetric @xmath17 knot complement .
then the cusp corresponding to this knot is geometrically isolated from the other two cusps .
it is not difficult to construct infinitely many examples this way .
@xmath13 has a hyperbolic structure with fundamental domain given by two regular ideal hyperbolic octahedra identified along their boundaries . to see this , note that @xmath13 consists of an alternating link complement in @xmath17 , and therefore by general properties of alternating links , a fundamental domain for @xmath13
can be found by two coloring the surface on which the link projection lies ( in this case a torus ) , and gluing the regions above and below this surface with a twist of @xmath29 or @xmath32 according to the coloring ( see @xcite ) .
the link in @xmath13 separates the dividing torus into two squares .
the suspension on each of these squares for which the suspension points are cusps @xmath20 and @xmath22 is an octahedron , from which the vertices are removed , since they lie on the cusps .
gluing the top four faces of each octahedron in the obvious manner , the bottom four faces of each are identified with a twist of @xmath33 , relative to the top identification .
the pattern of identifications is given in figure 3 . to prove that @xmath12 is strongly geometrically isolated , it suffices to show that @xmath34 and @xmath35 are functions solely of @xmath36 and of @xmath37 respectively ; i.e. that we can define @xmath38 and @xmath39 .
we subdivide each octahedron into @xmath14 ideal tetrahedra by cutting along the two faces @xmath40 and @xmath41 , so that each tetrahedron has vertices @xmath42 and @xmath20 .
we denote the simplex parameters of the four tetrahedra from the first octahedron as @xmath43 relative to the central edge , and for the tetrahedra from the second octahedron as @xmath44 .
the triangulated cusp shapes for the complete structure are given in figure 4 .
in the notation of @xcite we have the consistency relations @xmath45 @xmath46 the holonomies of @xmath47 and @xmath48 for the four cusps are @xmath49 @xmath50 @xmath51 @xmath52 since we are lifting dehn surgeries from @xmath12 , we require that @xmath53 from which , and from the dehn surgery equations it will follow that @xmath54 as in @xcite we denote @xmath55 and @xmath56 .
reparameterizing , we find that we can write everything in terms of two complex variables , @xmath57 and @xmath58
. we can define the simplex parameters in terms of these two variables as follows : @xmath59 @xmath60 @xmath61 @xmath62 with these definitions , it can easily be verified that @xmath63 and @xmath64 are functions only of @xmath58 , and that @xmath65 and @xmath66 are functions only of @xmath57 ; for example , @xmath67 , which is a function only of @xmath58
. explicitly we can calculate @xmath68 @xmath69 we therefore have the following : the orbifold @xmath12 has strongly isolated cusps .
there is a nice geometric definition of the parameterization in terms of the two complex numbers @xmath57 and @xmath58 , which we now describe . given a square @xmath70 in the complex plane of side length @xmath71 , with vertices at @xmath72
, we choose a point @xmath73 at the complex number @xmath57 .
we construct the points @xmath74 , @xmath75 , @xmath76 and @xmath77 so that the triangles @xmath78 , @xmath79 , @xmath80 and @xmath81 are similar with angles @xmath82 and @xmath83 , as illustrated in figure 5 .
then it is an easy calculation to see that as complex numbers , @xmath84 and @xmath85 .
therefore the octagon @xmath86 tiles the complex plane .
the four triangles @xmath87 , @xmath88 , @xmath89 and @xmath90 are precisely ( up to similarity ) the horoball sections of the simplices @xmath91 , @xmath92 , @xmath93 and @xmath94 respectively after dehn filling cusps @xmath22 and @xmath20 .
there is a similar , though `` mirror reversed '' picture describing how @xmath95 , @xmath96 , @xmath97 and @xmath98 are determined as a function of @xmath57 .
from the fact that this octagon tiles the plane , the consistency relations @xmath99 and @xmath100 are geometrically obvious . from the mirror - reversed picture , the consistency relations @xmath101 and @xmath102 can similarly be deduced . a relation of the form @xmath103 is equivalent to the geometric fact that the orientation preserving similarity taking @xmath104 to @xmath105 takes @xmath74 to the midpoint of @xmath106 .
from simple relations such as these , the independence of @xmath107 and @xmath108 from @xmath57 and of @xmath109 and @xmath110 from @xmath58 can be seen .
there is an interesting relationship between the parameterization of dehn surgery parameter space by real pairs @xmath111 , and by @xmath112 . in particular , we have the following [ circle_par ] the circle @xmath113 corresponds to the values of @xmath114 lying on the square with vertices at @xmath115 .
the circle @xmath116 passes through the square in the complex plane with vertices at @xmath117 .
therefore , by the description in section 4 , it corresponds precisely to the values of @xmath58 for which four of the simplices are degenerate . that is to say , the simplex parameters @xmath118 are real . by symmetry
, we need only consider the case that @xmath58 lies on a fixed `` quadrant '' of the circle , for the sake of argument , the arc between @xmath119 and @xmath120 in figure 5 .
however we know that @xmath121 @xmath122 and moreover , that @xmath123 but then , by inspecting figure 5 it can be seen that for @xmath58 taking values on this arc , @xmath124 , @xmath125 and @xmath126 , so that @xmath127 must take the values of @xmath128 respectively .
hence , @xmath129 .
it is a simple matter to check that @xmath130 , and therefore that @xmath131 . by a continuity argument
, it can be seen that vertices @xmath120 and @xmath119 correspond to the points @xmath132 and @xmath133 in @xmath134 space , and the proof follows .
notice of course that a corresponding fact holds for @xmath135 and @xmath57 .
@xmath136 for @xmath137 taking values on the square with vertices as in theorem [ circle_par ] . by a similar argument
we can show the `` circle '' @xmath138 corresponds to the values of @xmath137 lying on the square with vertices @xmath139 for the sake of argument , the infinite circle can be understood as a limit of larger and larger circles , in the hausdorff topology on the riemann sphere @xmath140 .
it is easy to see that as @xmath141 , the simplex parameters @xmath142 ( that is to say , their orientation and volume is negative ) .
this implies that each of @xmath143 approaches the value of @xmath144 or @xmath145 according to which branch of the logarithm is taken .
this depends on which homotopy path @xmath58 takes from @xmath146 to @xmath147 in the @xmath14-punctured sphere @xmath148 .
we restrict the values of @xmath58 for the moment to the simply connected domain consisting of @xmath149 minus the four infinite rays emanating from the four corners of the square .
the appropriate branch of the logarithm can then be determined by observing how the simplex parameters deform as @xmath141 and then requiring that @xmath150 be continuous .
it remains to check that @xmath151 in the limit . without loss of generality
, we choose @xmath58 with @xmath152 .
then @xmath153 , by the argument above .
it suffices to check that @xmath154 as @xmath141 .
but by the parameterization in terms of @xmath58 , it is immediate that @xmath155 .
the case for other large absolute values of @xmath58 follows by symmetry , and a continuity argument for the cases that @xmath156 .
the proof follows .
one must be wary about assuming that geometric structures corresponding to the case that @xmath137 lie in the interior of the square with vertices as in theorem [ circle_par ] exist . for the case that @xmath137
lie on this square , four of the tetrahedra are flat and disjoint except at the vertices .
the other four tetrahedra can be glued to the flat four , to produce an incomplete hyperbolic manifold , by the consistency equations .
the fact that each flat tetrahedra has a ( geometric ) bicollared neighborhood in the manifold so obtained not intersecting any of the other four tetrahedra implies that one can `` pump some air '' into each of the four flat tetrahedra without changing the homeomorphism type of the manifold .
that is to say , the incomplete manifold so obtained is homeomorphic to @xmath13 , and the complete manifold obtained by performing geometric dehn surgery is homeomorphic to a manifold obtained by topological dehn surgery on @xmath13 .
note by a theorem proved in @xcite one can find geometric structures corresponding to @xmath137 in an open neighborhood of @xmath157 that is to say , including some cases in which four of the tetrahedra are negatively oriented .
notice that as @xmath58 approaches one of @xmath117 that one of the @xmath158 triangles in a horoball section gets arbitrarily small with respect to the other three .
this is to say , the length of the filled geodesic in a complete structure ( if one were to exist ) goes to infinity .
this suggests that the orbifolds @xmath159 contain incompressible tori or annuli , a fact which can be verified geometrically .
the author wishes to express his gratitude to walter neumann and iain aitchison with whom he had many very fruitful conversations regarding the material in this paper .
he particularly wants to thank walter neumann for his generosity in taking the time to make numerous useful remarks on the presentation and details of this material .
99 i. aitchison and j. rubinstein , _ canonical surgery on alternating link diagrams _ , in knots 90 , proceedings of the international conference on knot theory and related topics , held in osaka , august 1519 , 1990 . ed .
akio kawauchi ; pp .
543558 c. hodgson , _ degeneration and regeneration of geometric structure on three manifolds _ , phd thesis , princeton university ( 1986 ) w. neumann and a. reid , _ rigidity of cusps in deformations of hyperbolic @xmath1orbifolds _ , math .
ann * 295 * ( 1993 ) pp .
223237 w. neumann and d. zagier , _ volumes of hyperbolic three manifolds _ , topology * 24 * no . 3 ( 1985 ) pp . | neumann and reid in @xcite describe a @xmath0cusped hyperbolic @xmath1orbifold in which the cusps are geometrically isolated .
based on numerical evidence provided by jeff weeks snappea program , they conjecture that the cusps are strongly geometrically isolated , a fact which we establish here .
we also give a parameterization of the dehn surgery space of this orbifold which has amusing properties .
[ thm]corollary |
nowadays , it is widely regarded that the cosmological constant @xmath0 or more generally dark energy is the most responsible candidate which explains accelerating expansion of universe .
nonetheless the details of cosmological constant @xmath0 or dark energy are still far from clear , then it is preferable and worthy to clarify the validity of this hypothesis by means of not only cosmological observations but also another astronomical / astrophysical ones . among such attempts , it would be the most natural idea to investigate the role of cosmological constant @xmath0 in the classical tests of general relativity , e.g. the perihelion advance of planetary orbit and the bending of light path .
so far , it was shown that the cosmological constant @xmath0 causes the perihelion shift of planets at least in principle , even though its contribution is too small to detect in the current measurement technique ( see @xcite and the references therein ) . while it has been believed for a long time that @xmath0 does not contribute to the light deflection because there is no @xmath0 in the second - order ordinary differential equation ( ode ) of photon .
however recently , rindler and ishak @xcite pointed out that @xmath0 does affect the bending angle by using the schwarzschild - de sitter or kottler metric and the invariant formula of cosine .
subsequently many authors argued its appearance in diverse ways and generality assisted the fact that there appears @xmath0 in the deflection angle @xmath1 , see @xcite for review and the references therein and also @xcite .
however , it seems that the conclusion has not converged yet ; for instance whether the leading order effect of @xmath0 is coupled with the mass of central body @xmath5 or not and so on . in order to clear up the confusion
, we will revisit the effect of the cosmological constant on the light deflection and its role in the cosmological lens equation .
before discussing the influence of @xmath0 on bending angle @xmath1 , we shall begin with re - considering the solution of photon trajectory in the schwarzschild spacetime . from the schwarzschild metric in the schwarzschild
coordinates , @xmath6 and the condition for null geodesic @xmath7 , we have the geodesic equation for equatorial plane ( @xmath8 ) , @xmath9 in which the two constants of motion , @xmath10 and @xmath11 are total energy and angular momentum , respectively .
alternatively , eq . ( [ eq : geodeq1 ] ) can be expressed in the form of second - order ode as , @xmath12 nevertheless , hereinafter we use eq .
( [ eq : geodeq1 ] ) instead of eq .
( [ eq : geodeq2 ] ) . in order to obtain the photon trajectory up to the second - order in @xmath2 ,
let us put the solution of eq .
( [ eq : geodeq1 ] ) @xmath13 as , @xmath14 where @xmath15 and @xmath16 are , respectively , first @xmath17 and second @xmath18 order correction to zeroth - order solution @xmath19 ( straight line ) .
hence @xmath15 and @xmath16 satisfy the following differential equations , @xmath20 .
\nonumber\end{aligned}\ ] ] noting that the integration constants of eqs .
( [ eq:1st - eq ] ) and ( [ eq:2nd - eq ] ) are chosen such that maximizing @xmath13 ( or minimizing @xmath21 ) for @xmath22 , then we obtain the trajectory of photon up to the second - order in @xmath2 as , @xmath23 where @xmath24 is the impact parameter which represents the minimum value of @xmath21-coordinate for the undeflected light ray , i.e. , @xmath25 . the bending angle @xmath1 is shortly derived from eq .
( [ eq : trajectory1 ] ) and it coincides with the famous formula by @xcite , @xmath26 a simple derivation of @xmath1 is given in appendix [ b ] .
it should be mentioned about the validity of the solution for light trajectory .
the appropriateness of our solution , eq .
( [ eq : trajectory1 ] ) can be verified readily by the direct substitution into eq .
( [ eq : geodeq1 ] ) and it is found that the residual terms are order @xmath27 , then it is perfectly valid up to the order of @xmath28 .
however , the photon trajectory given in previous works such as eq . ( 18 ) of @xcite and eq .
( 16 ) of @xcite are incorrect ; in fact , there appears @xmath18 order residual term in the solution of photon trajectory in @xcite .
now , let us investigate the contribution of @xmath0 on light ray . for this purpose
, we adopt the schwarzschild - de sitter or kottler metric @xcite , @xmath29 in the same way as the schwarzschild case , the differential equation of light is given by @xmath30 it should be emphasized here that the geodesic equation of light eq .
( [ eq : geodeq3 ] ) does obviously include @xmath0 .
therefore , previous arguments , such as `` @xmath0 does not appear in the geodesic equation of light '' , would be overstated . actually , the second - order ode derived from eq .
( [ eq : geodeq3 ] ) reduces to eq .
( [ eq : geodeq2 ] ) , nevertheless its solution of light trajectory should be obtained in such a way that the integration constants satisfy eq .
( [ eq : geodeq3 ] ) .
furthermore , the impact parameter is the minimum value of the coordinate @xmath21 if the light ray were undeflected , i.e. , @xmath25 .
it is obvious from eq .
( [ eq : geodeq3 ] ) that the impact parameter @xmath31 in this case is defined by @xmath32 then , the form of eq .
( [ eq : geodeq3 ] ) completely coincides with eq .
( [ eq : geodeq1 ] ) , except that the impact parameter @xmath24 is replaced by @xmath31 . therefore , the solution of eq .
( [ eq : geodeq3 ] ) becomes , @xmath33 and deflection angle is , @xmath34 it is worthy to note that the contribution of @xmath0 is incorporated in eqs .
( [ eq : trajectory2 ] ) and ( [ eq : def_angle2 ] ) through eq .
( [ eq : impact ] ) . as a consequence
, it is found that the cosmological constant @xmath0 does appears in both the geodesic equation and its solution , that is the trajectory of photon .
however , the effect of @xmath0 is completely absorbed into the definition of the the impact parameter ( see eq .
( [ eq : impact ] ) ) .
hence it is difficult to distinguish the influence of @xmath0 from the observed deflection angle . when we expand eq .
( [ eq : def_angle2 ] ) by using @xmath35 and remain @xmath36 terms , it follows that @xmath37 in which second term coincides with the previous results , e.g. eq . ( 5 ) and below in @xcite and the third term of eq . ( 15 ) in @xcite .
hence it is found that these results are included in eq .
( [ eq : def_angle2 ] ) as a limiting case .
it is clear that the trajectory of photon eq .
( [ eq : trajectory2 ] ) strictly satisfies eq .
( [ eq : geodeq3 ] ) up to the second - order in @xmath2 based on the result in previous section .
finally , we consider the contribution of @xmath0 in the cosmological lens equation . under the assumption that the thin lens approximation is valid , the lens equation relates the observed image position angle @xmath38 to the unlensed position angle @xmath39 of the source as @xmath40 where @xmath41 denotes the angular diameter distance from the redshift @xmath42 to @xmath43 , and the arguments @xmath44 and @xmath45 are the redshift of the lens and the source , respectively . for the distance formula @xmath4 in the unperturbed friedmann - lematre - robertson - walker ( flrw ) universe with @xmath0 ,
see , e.g. , @xcite . up to the first order in @xmath2 , the bending angle @xmath1 in the present case is @xmath46 the impact parameter @xmath31 is related to the image position angle @xmath38 by @xmath47 then , from eqs .
( [ eq : lenseq1 ] ) , ( [ eq : lensangle ] ) , and ( [ eq : cosmolens ] ) , the lens equation is finally @xmath48 therefore , the contribution of @xmath0 is completely involved in the form of the angular diameter distance @xmath4 .
no modifications due to @xmath0 appear even if the term of @xmath18 in @xmath1 is included . it should also be noted that eq .
( [ eq : lenseq2 ] ) is exactly the same form appeared in @xcite , where the authors have shown that the gravitational lensing effects are strongly dependent on the value of the cosmological constant and hence they provide us with useful means to test the cosmological constant . here
, we mention that in paper @xcite , sereno introduced the relation , @xmath49 ( see eq .
( 8) below of @xcite ) where @xmath50 is the impact factor of `` schwarzschild lens '' ( see eq .
( 8) of @xcite ) , @xmath51 is the angular - diameter distance , and
@xmath52 is the angular separation . since in the case of cosmological lens , the cosmological distance , such as the angular diameter distance ,
is defined with @xmath0 , then @xmath50 should be replaced by @xmath31 , instead .
we briefly summarize our conclusions .
contrary to the previous arguments , ( a ) the cosmological constant @xmath0 does appear in the orbital equation of light , ( b ) nevertheless the bending angle of light @xmath1 does not change its form even if @xmath3 since the contribution of @xmath0 is thoroughly absorbed into the definition of the impact parameter , and ( c ) the effect of @xmath0 is completely involved in the angular diameter distance @xmath4 .
then , no modifications due to @xmath0 appear even if the second - order terms in @xmath2 are included .
it should be mentioned that the similar conclusions are reached by other authors @xcite .
we hope that this paper provides a simpler and clearer explanation of the role of @xmath0 in the bending of light and the cosmological lens equation than the above - mentioned studies .
rindler and ishak @xcite have derived the angle between the radial direction and the light trajectory from the invariant cosine formula . just for a reference , here we give a more intuitive derivation . consider a spherically symmetric space .
the metric on the equatorial plane @xmath53 is given by @xmath54 in this space , the infinitesimal proper lengths along the radial direction and that perpendicular to the radial direction are given by @xmath55 and @xmath56 , respectively .
then the tangent of the angle @xmath57 between the radial direction and the light trajectory is @xmath58 which is the same form as eq .
( 16 ) in @xcite .
once we obtain the solution @xmath59 for the light trajectory , eq .
( [ eqa2 ] ) gives the angle at any point @xmath60 on the trajectory .
is @xmath61 .
[ fig1],scaledwidth=42.0% ]
using eqs . ( [ eq:1 ] ) , ( [ eq : trajectory1 ] ) and ( [ eqa2 ] ) , the angle @xmath57 at the point @xmath62 is @xmath63 consider a flat space which is tangent to the asymptotically flat region of the schwarzschild geometry ( see fig .
[ fig1 ] ) .
it is apparent from the property of an isosceles triangle in the euclidean geometry that the angle @xmath57 at the point @xmath64 is a half of the bending angle @xmath1 ( see fig .
[ fig2 ] ) .
then , @xmath65 .
the impact parameter @xmath24 and the bending angle @xmath1 are defined on this space.[fig2 ] , scaledwidth=42.0% ] between the radial direction and the light trajectory at @xmath66 on schwarzschild - de sitter geometry.[fig3],scaledwidth=42.0% ]
using eqs . ( [ eq:9 ] ) , ( [ eq : trajectory2 ] ) , and ( [ eqa2 ] ) , the angle @xmath57 between the radial direction and the light trajectory at the point @xmath67 on the schwarzschild - de sitter geometry is ( see fig . [ fig3 ] ) @xmath68 when linearized with respect to @xmath0 , @xmath69 which is essentially a half of eq .
( 18 ) in @xcite , although they drop the last term in eq .
( [ eqc2 ] ) .
it should be emphasized since the euclidean property does not hold on the schwarzschild - de sitter geometry , the angle @xmath57 in eq .
( [ eqc2 ] ) is _ not _ a half of the bending angle @xmath1 . 99 a. w. kerr , j. c. hauck , b. mashhoon , class .
grav . , * 20 * , 2727 ( 2003 ) .
w. rindler , relativity : special , general , and cosmological ( 2nd ed . ) , oxford univ . press , new york ( 2006 ) .
l. iorio , adv .
268647 ( 2008 ) .
w. rindler , m. ishak , phys .
d , * 76 * , i d .
043006 ( 2007 ) .
m. ishak , w. rindler , gen .
grav . , * 42 * , 2247 ( 2010 ) .
k. lake , phys .
d. * 65 * , i d .
087301 ( 2002 ) . m. park , phys .
d , * 78 * , i d .
023014 ( 2008 ) . i. b. khriplovich , a. a. pomeransky , int . j. mod .
d , * 17 * , 2255 ( 2008 ) .
f. simpson , j. a. peacock , a. f. heavens , mnras , * 402 * , 2009 ( 2010 ) .
a. bhadra , s. biswas , k. sarkar , phys .
d , * 82 * , i d . 063003
h. miraghaei , m. nouri - zonoz , gen .
42 * , 2947 ( 2010 ) t. biressa , j. a. de freitas pacheco , gen
grav . , * 43 * , 2649 ( 2011 ) .
r. epstein , i. i. shapiro , phys .
d , * 22 * , 2947 ( 1980 ) .
j. bodenner , c. m. will , am .
j. phys . ,
* 71 * , 770 ( 2003 ) .
f. kottler , annalen .
, * 361 * , 401 ( 1918 ) .
m. sereno , phys .
lett . , * 102 * , i d . 021301 ( 2009 ) .
m. fukugita , t. futamase , m. kasai , e. l. turner , astrophys . j. , * 393 * , 3 ( 1992 ) .
m. fukugita , t. futamase , m. kasai , mon . not .
soc . , * 246 * , 24 ( 1990 ) . | we revisit the effect of cosmological constant @xmath0 on the light deflection and its role in the cosmological lens equation .
first , we re - examine the motion of photon in the schwarzschild spacetime , and explicitly describe the trajectory of photon and deflection angle @xmath1 up to the second - order in @xmath2 .
then the discussion is extended to the contribution of the cosmological constant @xmath0 in the schwarzschild - de sitter or kottler spacetime .
contrary to the previous arguments , we emphasize the following points : ( a ) the cosmological constant @xmath0 does appear in the orbital equation of light , ( b ) nevertheless the bending angle of light @xmath1 does not change its form even if @xmath3 since the contribution of @xmath0 is thoroughly absorbed into the definition of the impact parameter , and ( c ) the effect of @xmath0 is completely involved in the angular diameter distance @xmath4 . |
the ginzburg - landau parameter of a superconductor , @xmath0 , is generally defined as the ratio of its magnetic penetration depth ( @xmath1 ) to its coherence length ( @xmath2 ) at the thermodynamical critical temperature ( @xmath3 ) .
the parameter relates the two fundamental length scales of the material s superconducting phase , distinguishes whether the phase transition is first- or second - order , and characterizes the material s response to applied magnetic fields in the superconducting state .
theoretically , for clean materials , it is given by the bcs @xmath0 = 0.96 @xmath4(0)/@xmath5 where @xmath4(0 ) is the london penetration depth at zero temperature and @xmath5 is the pippard coherence length @xcite .
the parameter , although often referred to , is infrequently measured and little tabulated owing to its dependence on sample purity and structure . where measured ,
@xmath0 has been determined @xcite from independent measurements of @xmath1 and the thermodynamic critical field @xmath6 , as well as from magnetization measurements on thin films and foils of various materials @xcite . for type - i materials
( @xmath7 1/@xmath8 ) , another technique has been from measurements of the supercritical fields of microspheres .
this technique has been used in determining the @xmath0 of aluminum @xcite , cadmium @xcite , gallium @xcite , mercury @xcite , indium @xcite , lead @xcite , tin @xcite , thallium @xcite and zinc @xcite , as well as a series of alloys @xcite .
inexplicably , the results are generally @xmath9 factor 2 smaller than the thin film / foil determinations as well as the bcs - defined @xmath0 , as shown in table i. more recent measurements , using a fast pulse induction technique with thin foils , provide results in better agreement with those from the microspheres @xcite .
.comparison of ginzburg - landau parameter determinations ; `` - '' denotes no experimental measurement . [ cols="^,^,^,^,^,^,^",options="header " , ]
the supercritical fields determinations of all reports are obtained from hysteresis curves of the first order tin and indium transitions .
different reports have however assumed different definitions of the critical fields within the hysteresis curves , as shown schematically in fig .
5 . as evident
, these definitions alone can lead to an appreciable variation in the reported @xmath10 . in general , the @xmath11 , @xmath12 parameters of fm should be consistently higher than sbc , which is not observed in the case of fm @xmath11 .
in the case of single grain measurements , variations in the supercritical fields arise mainly from the grain metallurgy .
fm performed careful investigations of defects as part of their study .
the hysteresis curves were measured as a function of the orientation of the spheres in the applied field , and the @xmath11 , @xmath12 taken from the direction yielding the lowest values ( i.e. highest @xmath13 , lowest @xmath14 ) .
the results are strongly dependent on @xmath3 . in principle , for @xmath15 , the coherence length is sufficiently large that these effects are negligible ; at lower temperatures , the coherence length is smaller than the defect and the effects are more noticeable . for this reason ,
fm also employed @xmath16 ( @xmath14/@xmath13)@xmath17 as a comparison for @xmath18 , which has the advantage of being insensitive to @xmath3 since it is independent of @xmath6 . in the case of the suspension measurements ,
the deformation of the hysteresis from the rectangular single sphere cycle is due largely to the interplay between defects and diamagnetic interactions between the spheres @xcite . at lower temperatures , defects in the sphere metallurgy
may significantly reduce the superheating capacity of the material , resulting in a lowered transition field such as observed in the indium data of sbc . however , the defects should also serve as nucleation centers for `` early '' normal ( n ) @xmath19 superconducting ( s ) transitions at lower temperatures , which is not observed . on the other hand
, the advantage of the multisphere measurement lies in the averaging over the defect contribution - the last sphere to transition to the normal state is the most defect - free sphere of the suspension . the recent work of pearanda et .
@xcite has demonstrated significant diamagnetic interactions even for filling factors as small as 5% .
these may vary from one suspension to another of the same material as a result of inhomogeneous distributions .
although the last sphere to transition in the s @xmath19 n case is effectively free of diamagnetic interactions , this is not true for the n @xmath19 s where diamagnetic interactions are a maximum .
as the field decrease continues , the new n @xmath19 s transitions enhance the local diamagnetic contribution to the local fields of those spheres still normal .
this would suggest that the @xmath18 determinations with suspensions are lower than actual by some amount proportional to the diamagnetic interactions , since the local field is higher . as seen in fig .
1(a ) , the sbc h@xmath20 for tin is essentially the same as those of the fm single sphere
. for indium however , the sbc @xmath21 are systematically lower than those of fm by @xmath9 10% for @xmath22 0.8 .
hysteresis curve schematic , indicating the various field definitions for @xmath13 , @xmath14 employed by sbc and fm.,width=8 ] all results were moreover obtained by variation of the magnetic field , rather than temperature , which as shown by chaddah and roy @xcite results in an enhanced @xmath21 , apart from any diamagnetic interaction , as a result of the thermal fluctuations induced by the field changes themselves .
the variation of @xmath0 with temperature suggests a variation of the transition order with temperature , which has possibly important ramifications since @xmath0 is then less a fundamental property of the superconductor than a simple ratio between the two characteristic lengths in the description , both of which vary with temperature and yield results consistent with the observed @xmath0 determinations .
variation of the order of the transition with the temperature is predicted in recent renormalization - based reformulations of basic superconductive theory @xcite , which include fluctuations in the involved gauge and scalar fields , and result in a dividing line between type - i and -ii behavior at @xmath23 with a magnetic response which can be varied between type - i and type - ii simply by temperature change .
this variation has been seen in magnetization measurement of nitrogen - doped ta ( @xmath0 = 0.665 ) @xcite , but is otherwise unconfirmed .
reexamination of previous ginzburg - landau parameter determinations from superheating field measurements of microspheres of tin and indium shows the long - standing discrepancy in reported @xmath24 of tin to derive from the different analyses methods employed , rather than whether single- or multi - grain determinations , or grain metallurgy .
although the tin results of sbc can be made to agree with those of fm ( and larrea et .
al . ) , the customary analyses via eq . ( 2 ) is in disagreement with that performed numerically .
this is not the case of indium , where the discrepancy reflects severe differences in the supercritical field measurements themselves .
the similar reexamination of the supercooling field measurements uncovers no divergences capable of explaining the discrepancies with the measurements of ref .
@xcite . to the contrary ,
the variation of @xmath14 with @xmath25 appears to confirm the latter , raising severe questions as to the true significance of @xmath0 as a descriptor of superconductors .
the resolution of these is however beyond the scope of this work .
the available @xmath26 1 data is of insufficient precision to permit more than suggestions as to @xmath0 .
analyses of all superheating results via the pad approximation yields a convergence of @xmath24 , @xmath18 to @xmath0 , with values for each material only slightly different from those previously reported on the basis of @xmath21 alone .
these previous results however are likely larger than actual as a result of diamagnetic interactions between the superconducting sphere population . apart from nonlocal effects ,
the large @xmath11 for @xmath26 0.8 is likely due to defect presence , which causes nucleation of the normal state to occur at fields below the maximal @xmath27 .
these become unimportant as @xmath28 1 , increasing @xmath13 and generating a rapid drop in @xmath11 . whether the proper analyses is by eq .
( 2a ) , numerical or pad remains in question . nevertheless , all of the materials analyzed in ref .
@xcite ( tin , indium , thallium , lead and mercury ) follow from the sbc fig . 1 ; in contrast , the analyses of the remaining refs . has proceeded via eq .
( 2 ) , which is seen to underestimate the respective @xmath10 . in view of the significance of the ginzburg - landau parameter ,
the uncertainties inherent to this technique and the variation in experimentally - obtained results , and technological advances over the last thirty years , careful remeasurements of the supercritical fields over the full temperature range , and their analysis within a definitive description of the @xmath29 plane , would seem to be indicated . the discrepancy in @xmath0 between spheres and thin film / foil determinations has been known for some decades , but to the best of our knowledge remains unexplained .
the thin film / foil results are in fact a factor @xmath9 1.7 larger than those from the microspheres ; they are also in better agreement with @xmath0 derived from @xmath30 via @xmath31 , and also agree in general with the lower temperature results of the microspheres .
we thank j. seco for preliminary investigations of the discrepancy which stimulated the re - analyses herein , a. baptista for assistance in last phases of the analysis , and g. waysand for critical comments .
the work was supported in part by grants praxis / fis/10033/1998 and pocti / fnu/39067/2001 of the foundation for science and technology of portugal , co - financed by feder . | long - standing discrepancies within determinations of the ginzburg - landau parameter @xmath0 from supercritical field measurements on superconducting microspheres are reexamined . the discrepancy in tin
is shown to result from differing methods of analyses , whereas the discrepancy in indium is a consequence of significantly differing experimental results .
the reanalyses however confirms the lower @xmath0 determinations to within experimental uncertainties . |
optical lattices are known to be an effective tool for manipulation of matter waves .
the induced spatial periodicity allows , in particular , for the existence of various localized structures , not available in homogeneous systems .
such excitations , which are called localized modes ( also gap solitons depending on the limit they are considered in ) have been intensively studied during the last few years ( see e.g. recent reviews @xcite and references therein ) and observed experimentally in a bose - einstein condensate ( bec ) of @xmath0rb atoms @xcite . while main attention was devoted to single - component becs , existence of localized modes was found to be a generic property of multicomponent systems , as well .
more specifically , emergence of coupled bright solitons , from the modulational instability of binary mixtures of becs in optical lattices ( ols ) , was observed numerically in @xcite . in @xcite , it was shown that ols can be used to modify the interactions both within and between bec components , leading to the creation of two - component gap solitons .
existence and stability of 2d and 1d coupled gap solitons in a binary bec with repulsive interactions was analyzed by means of the variational approximation and numerically in @xcite .
it was also found that localized modes can exist in atomic - molecular condensates @xcite , in bose - fermi mixtures in ol @xcite and in three - component spinor condensates @xcite .
all the results on binary mixtures of becs reported so far , concern excitations of the two components characterized by similar scales , i.e. by spatial localization regions , and numbers of particles of the different species having the same order . also earlier studies were restricted to modes having density distributions in both components of the same symmetry , whenever the components belong to the same gap ( gap solitons belonging to different gaps were addressed in @xcite ) .
in the present paper we report a diversity of novel states observable in two - component becs embedded in ols with different symmetry properties of the density distributions with respect to the periodic potential and therefore termed below as _ mixed symmetry _ modes ( or also gap solitons of mixed symmetry ) .
the interaction between the two components of the condensate is crucial for the existence of these modes since they are formed from single component modes which have different stability properties ( one stable and the other unstable ) in absence of interaction .
the interspecies interaction allows for stabilization of the unstable mode in the pair and subsequently for formation of a stable bound state .
mixed symmetry modes exist both for attractive and repulsive inter and intra species interactions and represent very stable new excitations of two component becs in ols .
stabilization phenomena induced by the interspecies interaction are also possible for equal symmetry states ( both symmetric or antisymmetric with respect to a minimum or a maximum of the potential ) . in this case
we find that modes which are both unstable in absence of interaction can form stable equal symmetry bound states in presence of interspecies interaction .
besides stationary states of mixed symmetry we show that non stationary states of mixed symmetry which give rise to breather - like oscillations are also possible . when the inter and the intra
species interactions are all equal , the two - component breather modes can be constructed from a linear superposition of two mixed symmetry modes with an unbalanced number of atoms in the two components . on the other hand , we show that breather modes with a balanced numbers of atoms and with a very regular dynamics which persist on a long time scale without any apparent emission of matter , are also possible .
mixed symmetry modes , both of stationary type and of breather type , considerably enlarge the number of stable localized excitations which one can find in bec mixtures , this showing the richness of these systems with respect to single component becs .
the paper is organized as follows . in sec .
[ sec_model ] we recall the system of coupled gross - pitaevskii ( gp ) equations describing the mixture , specify the parameters to be used for particular numerical simulations , and give a simple physical picture suggesting existence of asymmetric modes . in sec .
[ sec_numerics ] we discuss on particular examples the localized modes of mixed symmetry in the limit ( @xmath1 ) and study stability of the modes by direct numerical simulations . in sec .
[ largeampl ] we consider large amplitude localized modes both of equal symmetry and of mixed symmetry type .
[ breathers ] is devoted to breathing modes , both with balanced and unbalanced number of atoms , which are obtained by combining the developed approach with the symmetry of the coupled gp equations . in the conclusion
the main outcomes are summarized .
at the zero temperature and a large number of atoms , when quantum fluctuations can be neglected , a diluted binary mixtures of becs is well described by the coupled gp equations ( @xmath2 ) @xcite @xmath3 here @xmath4 is the atomic mass of the @xmath5 component , @xmath6 is the reduced mass , and @xmath7 are the s - wave scattering lengths of the binary interactions .
the macroscopic wave functions , @xmath8 , are normalized to numbers of atoms @xmath9 : @xmath10 , which do not depend on time .
the trap potential for the component @xmath11 , @xmath12 , consists of a superposition of a magnetic trap and an ol along the @xmath13-direction : @xmath14 here @xmath15 is the transverse linear oscillator frequency of the component @xmath11 , @xmath16 is the aspect ratio ( it is assumed to be equal for both components ) , @xmath17 is the lattice constant , @xmath18 , @xmath19 is the amplitude of ol measured in terms of the respective recoil energy @xmath20 , and @xmath21 is a @xmath22-periodic dimensionless function ( the lattice profiles are considered to be the same for both components ) .
we consider cigar - shaped condensates , i.e. the limit @xmath23 , at low densities .
to specify the scaling we introduce a small parameter of the problem @xmath24 , where @xmath25 is an average transverse oscillator length defined as @xmath26 , where @xmath27 .
then the applicability of the theory is determined as smallness of a number of particles on the scale of the scattering length , i.e. as @xmath28 , where @xmath29 is the linear atomic density and @xmath30 is the total number of atoms .
this is the case when the multiple - scale analysis can be applied to reduce ( [ nls_gp ] ) to a system of coupled 1d gp equations ( see e.g. @xcite ) . by introducing slow dimensionless independent variables @xmath31 and @xmath32 and dimensionless order parameter @xmath33 , through the expansion ( @xmath2 ) @xmath34 with the normalized transverse eigenfunctions @xmath35
we obtain a system of coupled 1d gp equations @xmath36 here @xmath37 and the nonlinear coefficients have the following form @xmath38 looking for stationary solutions @xmath39 , we obtain the system @xmath40 with @xmath41 and @xmath42 being the chemical potentials of the first and second components , respectively . the goal of this paper is to present some families of localized modes and breathers of eqs .
( [ eigen ] ) which are typical for multicomponent becs and which posses interesting symmetry properties .
these solutions have been found numerically by means of different methods .
the typical difficulty which arise in this problem is to get _ a priori _ idea on possible shapes of localized modes which can be used for numerical iterative algorithm . to overcome this difficulty we have used two alternative approaches .
the first involves a search of bifurcations for specific solutions of ( [ eigen ] ) using shooting method ( see @xcite ) as the main tool .
the second is the self - consistent approach which has been used before for applications to single - component becs ( see @xcite ) .
the self consistent method , indeed , is very convenient for exploring the band structure in presence of nonlinearity and to address to specific modes for which good initial guesses are available .
basically , the shooting method allows one to make a systematic exploration of the whole families of solutions in terms of a single ( for single component becs ) parameter ( shooting parameter ) .
similar exploration of the case of coupled gp equations is quite complicated , since in this case one has to scan a semi - infinite domain of two shooting parameters . in the present paper
we use the shooting method to investigate localized modes of the mixed symmetry type in the limit @xmath1 ( see sec . [ sec_numerics ] ) and resort to self consistent approach to investigate bec mixtures with balanced number of atoms ( see sec .
[ largeampl ] ) .
when one of the components is dominant , i.e. it has much larger number of atoms than the second component , the last one can be considered as bifurcating from the linear spectrum .
such modes will be referred to as _ unbalanced _ or strongly asymmetric in number of atoms . in this way after constructing the localized solution for the first component in the absence of the second one , the mode can be constructed by means of some iterative procedure . combining this approach with the symmetry of the system we will also be able to construct ( numerically ) exact breathing localized modes .
another regime we will deal with corresponds to the situation when the both components are large amplitude solutions having numbers of particles of the same order , @xmath43 . in this case
we will show existence of not only simplest fundamental modes with equal symmetry but also modes and breathers with mixed symmetry .
we remark that mixed symmetry modes were also found in atomic - molecular condensates @xcite . in this case , however , the mean - field equations had an intrinsic asymmetry due to the different properties of atoms and molecules .
first of all it is desirable to establish a convenient terminology for the identification of modes . to this end
we observe that multiplying ( [ eigen ] ) by @xmath44 and integrating with respect to @xmath45 one arrives at the _ necessary condition _ for existence of a stationary localized mode ( @xmath46 ) : @xmath47 for the physical interpretation of this condition , we notice that a solution of ( [ eigen ] ) minimizes the hamiltonian @xmath48 where @xmath49 @xmath50 , and we have defined @xmath51 considering @xmath52 with @xmath53 , which for a stable solution must lead to increase of the energy , we immediately obtain @xmath54 the first of the obtained constrains is exactly eq .
( [ necessary ] ) .
let us now require the potential to have a definite symmetry , i.e. to be either even or odd , and concentrate on the solutions possessing a specific symmetry , i.e. @xmath55 if the potential is odd , @xmath56 , such solutions strongly localized about @xmath57 can not satisfy ( [ necessary ] ) .
this allows us to conjecture nonexistence of stable solitons localized about the point @xmath57 for odd potentials .
in particular , in the case of cos - like potential @xmath58 explored below in the present paper : localized modes with a defined symmetry can be excited only in the vicinity of one of the points @xmath59 where @xmath60 is an integer .
considering solutions localized about @xmath57 for an even potential @xmath61 one can identify four types of the modes .
they can be classified in terms of the symmetry properties acquired with respect to the periodic potential , considered earlier for a atomic - molecular bec @xcite and for a single - component bec @xcite .
the diversity of modes makes it is desirable to establish a convenient terminology for their identification . to this end
we classify states according to their symmetry properties with respect to the minima of the periodic potential around which are localized .
a classification of this type was introduced in refs .
@xcite , for atomic - molecular becs and single - component becs , respectively , in terms of four single component states .
these states are the os _ on - site symmetric _ and oa _ on - site antisymmetric _ modes , which are respectively symmetric and antisymmetric with respect to a minimum of the ol , hereafter referred as _ lattice site _ , and the is _ inter - site symmetric _ and ia _ inter - site anti - symmetric _ modes which are respectively symmetric and antisymmetric with respect to a maximum of the ol ( the center of the symmetry is the middle point between two lattice sites ) .
notice that these symmetry properties are the same as for intrinsic localized modes of nonlinear lattices , as one could expect from the tight - binding approximation based on the properties of wannier functions @xcite and reducing the periodic gp equations to discrete lattices @xcite .
except for the os solution , which is always stable , the other types of gap solitons are usually unstable in the one component case @xcite .
however , from ( [ h - deriv ] ) one can expect that gap solitons which have the same symmetry properties in both components should exist and may be stable .
the respective modes will be characterized by a double indexing , like for example os - os , os - oa , etc .
, the first and second pairs of letters referring to the first and second components , respectively .
the modes os - os , oa - oa , is - is , and ia - ia will be called the _
`` equal - symmetry '' two - component modes _ , while modes os - oa , is - ia , etc . will be called mixed symmetry modes , with an evident meaning in the notation .
although our results have general character , and can be applied to any binary mixture of becs , either spinor or of different species , to be specific we focus on spinor becs .
such condensates are available experimentally . as example we mention mixtures of the hyperfine states @xmath62 and @xmath63 as well as of the states @xmath64 and @xmath65 of @xmath0rb atoms produced in @xcite and @xcite , respectively . for a spinor mixture @xmath66 , @xmath67 , and the potential ( [ cos ] ) the equations ( [ eigen ] )
are simplified [ eigen_spinor ] @xmath68 where nonlinear coefficients are given by @xmath69 , @xmath70 , and @xmath71 $ ] with @xmath72 .
in order to fix parameters we recall that the thermodynamical stability of a homogeneous binary mixture is determined by the condition @xcite @xmath73 , where @xmath74 using this fact as the reference point we will consider the following three situations @xmath73 , @xmath75 and @xmath76 ( we however emphasize that , generally speaking , these condition should not be considered as the stability / instability conditions for localized modes ) . before
going into details of numerical simulations , we observe that interchange of the regimes with different @xmath77 can be experimentally achieved either by means of feshbach resonance affecting scattering lengths @xmath78 or by means of using transverse magnetic trap , i.e. geometry of the trap , resulting in different @xmath79 for different components , say , in @xmath80 for the mixtures of hyperfine states @xmath64 and @xmath65 of @xmath0rb atoms in a magnetic trap @xcite .
let us consider the limit @xmath82 .
for the first step , neglecting contribution of the second component , @xmath83 , one can solve equation ( [ eigen1_spinor ] ) in the form @xmath84 to find localized modes for a given value of the chemical potential @xmath85 . in what follows we focus only on symmetric modes belonging to the lowest branch of the solutions , and therefore referred to as the _ fundamental mode_. as the second step , in the same approximation , i.e. neglecting @xmath86 , one can consider ( [ eigen2_spinor ] ) with @xmath87 found from ( [ mode1 ] ) : @xmath88 where the effective potential is defined as : @xmath89 after finding discrete eigenvalues @xmath90 for a given @xmath85 , one can scan the gap for the first component varying @xmath85 and obtaining all eigenvalues @xmath91 . in fig .
[ fig_w1w2 ] we present the results of this procedure for the cases @xmath73 and @xmath75 .
the physical meaning of the above procedure consists in the approximation of a localized modes @xmath92 , @xmath93 in the limit @xmath82 . in this limit
, the wavefunction of the second component , @xmath93 , appears as a _ defect mode _ and , strictly speaking , does not describe the atomic distributions at a finite @xmath94 .
since however , the solution at @xmath95 is known , departing from it and using an iterative procedure one can construct the solutions for @xmath96 .
this approach is based on the following important statement ( whose arguments are presented in appendix [ app:1 ] ) : _ if @xmath85 and @xmath90 as well as @xmath97 and @xmath98 are fixed , then the localized modes of ( [ eigen ] ) are isolated_. when @xmath99 varies one can speak about _ families _ of the nonlinear localized modes . before performing the search of localized modes ,
let us discuss in more details , the results of the above analysis of defect modes . in fig .
[ fig_w1w2 ] three branches of the solutions are shown for different signs of @xmath77 . the branches a correspond to the case where the chemical potentials are approximately equal . in order to understand their occurrence
let us consider a solution of eqs .
( [ eigen_spinor ] ) in the form @xmath100 where @xmath101 is a constant , which without loss of generality can be considered real .
such a solution exists for @xmath102 and @xmath103 and below will be referred to as a _
trivial solution_. in order to obtain the linear limit ( [ mode1 ] ) , ( [ mode2_eigen ] ) one has to consider @xmath104 , what is possible only for @xmath105 . in this last case
the branches a would coincide with the diagonals of the boxes in fig .
[ fig_w1w2 ] . in the cases at hand @xmath106 , what explains deviations of the lines a from the diagonals .
the most interesting results , however , are shown in the insets of fig .
[ fig_w1w2 ] , where one can observe the two other branches of the solution of the eigenvalue problem ( [ mode2_eigen ] ) branches b and c. these modes appear in the vicinity of the opposite gap edges for the first and the second components , as this is illustrated in fig .
[ fig_veff](a ) for the case @xmath73 : the eigenvalues are located near the top of the gap for the first component and in the vicinity of the bottom of the gap for the second component . for @xmath96 ( but small ) these eigenvalues give rise to os and oa modes which are characterized by a small difference in energies .
their shapes are shown in fig [ fig_prof ] . as we have already mentioned these solutions
appear as defect modes , what is illustrated in fig .
[ fig_veff ] where we show the fundamental localized mode of the first component near the upper edge of the gap ( panel ( b ) ) , corresponding potential @xmath107 and the respective energy levels , which can be interpreted as defect modes ( panel ( c ) ) .
the two eigenvalues b and c can be viewed as a splitting of discrete defect level ( similar to the level splitting in a double - well potential ) .
a number of the defect modes depends on the amplitude of the potential @xmath19 . in all our cases the energy differences between the modes were very small ,
what makes it convenient to speak about appearance of the pairs of the modes .
then , one can determine @xmath108 , such that for @xmath109 there exist @xmath60 pairs of the defect modes ( simultaneously with the fundamental mode ) .
in particular , for the cos - like potential and @xmath73 ( @xmath75 ) we have found numerically that @xmath110 , @xmath111 ( @xmath112 , @xmath113 ) .
we have introduced eqs .
( [ mode1 ] ) , ( [ mode2_eigen ] ) for the sake of approximation of localized modes in the limit @xmath114 . in order to construct the solutions
emerging from the branches b and c we use an iterative procedure starting with the solutions obtained for @xmath115 .
[ fig_prof ] shows examples of the obtained results . the mode for the second component bifurcates from the zero - solution .
its amplitude grows with the detuning of the chemical potential @xmath90 towards the center of the gap [ see fig .
[ fig_prof](c)-(f ) and fig .
[ fig_veff ] ( a ) ] , what is accompanied by the increase of the number of particles @xmath94 . in fig .
[ fig_n_mu ] we present the dependence of the number of particles in the second component , @xmath94 on @xmath90 deviating from the corresponding values b and c ( see fig . [ fig_w1w2 ] ) .
we notice that in our procedure of search of the modes the profile of the fundamental component also changes with @xmath90 .
the change , however , is very small , it corresponds to the change of the number of particles in the first component of order of @xmath116 , and therefore is practically invisible on the scale of fig .
[ fig_prof](a ) , ( b ) . to study dynamical stability of the obtained modes we performed direct numerical integration of equations ( [ gp_1d_1 ] ) with @xmath117 and @xmath118 .
as initial conditions we used perturbed stationary solutions obtained using the iterative procedure , the amplitude of perturbation being of the order of 5@xmath119 . first , in fig .
[ fig_dynamics_mag_par ] we show the case @xmath73 .
one can see that the time evolution of the two - component soliton is stable . by switching off the coupling parameter
, i.e. setting @xmath120 , one obtains the system of two uncoupled gp equations .
the dominant fundamental component is close , in some sense , to an os mode of the single gp equation with the periodic potential and that is why it is stable .
the second component , representing two in - phase humps and originally coupled by the effective potential due to the first component , becomes unstable at @xmath120 .
the two humps in the absence of first component , which the played role of the effective barrier between them , start to move towards each other and overlap .
however due to the large velocity after collision two wave packets start to move outwards the center .
the described behavior is observed at the initial stages of the evolution shown in fig .
[ fig_dynamics_mag_chi0 ] ( a ) , ( b ) . at latter times , reaching the turning point they change the direction of motion towards the center .
after several periods of damped oscillations the one - hump os mode is created .
the out - of - phase distribution of the second component , shown in fig .
[ fig_dynamics_mag_chi0 ] ( d ) is stable representing a single - component stable oa mode .
next we investigated the case @xmath75 .
typical results are shown in fig .
[ fig_dynamics_opt_par ] , where one observes instability of the both os - os ( panels ( a ) and ( b ) ) and os - oa ( panels ( c ) and ( d ) ) modes .
it has been checked that the time at which the instability is developed increases when @xmath90 approaches the lowest edge of the gap , tending to infinity as @xmath90 approaches @xmath121 and respectively as @xmath82 .
let us now turn to the study of two - component localized modes with _ a balanced _
i.e. with the same or with similar number of particles in each component .
these modes can be of equal symmetry or of mixed symmetry type .
some of the equal symmetry modes were also found in ref .
balanced mixed symmetry modes have not been discussed before .
in contrast to the equal symmetry solutions , which for many aspects are similar to standard modes of the single component gpe , the mixed state symmetries are more specific for multicomponent bec systems in ol displaying interesting stability properties .
( left panel ) and all repulsive @xmath123 ( right panel ) interactions . for both cases
the amplitude of the ol is @xmath124 and the number of atoms is @xmath125 .
open circles refer to the first component while filled squares refer to the second component .
notice that all equal symmetry modes are degenerated .
also notice the presence of a bound state near the edge of the upper band ( left panel ) and a bound state n the second gap ( right panel ) .
[ fig1ms],title="fig:",width=160,height=160 ] ( left panel ) and all repulsive @xmath123 ( right panel ) interactions .
for both cases the amplitude of the ol is @xmath124 and the number of atoms is @xmath125 .
open circles refer to the first component while filled squares refer to the second component .
notice that all equal symmetry modes are degenerated .
also notice the presence of a bound state near the edge of the upper band ( left panel ) and a bound state n the second gap ( right panel ) . [ fig1ms],title="fig:",width=160,height=160 ] ( all attractive case ) .
thick and thin lines refer to first and second component , respectively ( open circles and filled squares in fig .
[ fig1ms ] ) . to identify the symmetry of the solutions , lattice potentials ( scaled by a factor 3 )
have been shown by thin dotted line.,title="fig:",width=100,height=100 ] ( all attractive case ) .
thick and thin lines refer to first and second component , respectively ( open circles and filled squares in fig .
[ fig1ms ] ) . to identify the symmetry of the solutions , lattice potentials ( scaled by a factor 3 )
have been shown by thin dotted line.,title="fig:",width=100,height=100 ] ( all attractive case ) .
thick and thin lines refer to first and second component , respectively ( open circles and filled squares in fig . [ fig1ms ] ) . to identify the symmetry of the solutions , lattice potentials ( scaled by a factor 3 )
have been shown by thin dotted line.,title="fig:",width=100,height=100 ] ( all attractive case ) .
thick and thin lines refer to first and second component , respectively ( open circles and filled squares in fig .
[ fig1ms ] ) . to identify the symmetry of the solutions , lattice potentials ( scaled by a factor 3 )
have been shown by thin dotted line.,title="fig:",width=100,height=100 ] ( all attractive case ) .
thick and thin lines refer to first and second component , respectively ( open circles and filled squares in fig .
[ fig1ms ] ) . to identify the symmetry of the solutions , lattice potentials ( scaled by a factor 3 )
have been shown by thin dotted line.,title="fig:",width=100,height=100 ] ( all attractive case ) .
thick and thin lines refer to first and second component , respectively ( open circles and filled squares in fig .
[ fig1ms ] ) . to identify the symmetry of the solutions , lattice potentials ( scaled by a factor 3 )
have been shown by thin dotted line.,title="fig:",width=100,height=100 ] ( all repulsive case ) .
thick and thin lines refer to first and second component , respectively ( open circles and filled squares in fig .
[ fig1ms ] ) . to identify the symmetry of the solutions , lattice potentials ( scaled by a factor 3 )
have been shown by thin dotted lines.,title="fig:",width=100,height=100 ] ( all repulsive case ) .
thick and thin lines refer to first and second component , respectively ( open circles and filled squares in fig .
[ fig1ms ] ) . to identify the symmetry of the solutions , lattice potentials ( scaled by a factor 3 )
have been shown by thin dotted lines.,title="fig:",width=100,height=100 ] ( all repulsive case ) .
thick and thin lines refer to first and second component , respectively ( open circles and filled squares in fig .
[ fig1ms ] ) . to identify the symmetry of the solutions , lattice potentials ( scaled by a factor 3 )
have been shown by thin dotted lines.,title="fig:",width=100,height=100 ] ( all repulsive case ) .
thick and thin lines refer to first and second component , respectively ( open circles and filled squares in fig .
[ fig1ms ] ) . to identify the symmetry of the solutions ,
lattice potentials ( scaled by a factor 3 ) have been shown by thin dotted lines.,title="fig:",width=100,height=100 ] ( all repulsive case ) .
thick and thin lines refer to first and second component , respectively ( open circles and filled squares in fig .
[ fig1ms ] ) . to identify the symmetry of the solutions , lattice potentials ( scaled by a factor 3 )
have been shown by thin dotted lines.,title="fig:",width=100,height=100 ] ( all repulsive case ) .
thick and thin lines refer to first and second component , respectively ( open circles and filled squares in fig . [ fig1ms ] ) . to identify the symmetry of the solutions , lattice potentials ( scaled by a factor 3 )
have been shown by thin dotted lines.,title="fig:",width=100,height=100 ] thus , while the oa and the oa - oa modes are unstable for the one- and two - component system , respectively , the coupling of the oa mode with an os mode gives rise to a stable os - oa mixed state of two - component bec in ol .
the same situation occurs with the is and ia modes . in the following
we show that these mixed symmetry solutions exist for both attractive and repulsive inter and intra specie interactions and together with the on site os - os solution , they represent stable localized excitation of a two - component bec in ol . , in absence ( @xmath120 , left panels ) and in presence ( @xmath126 , right panels ) of interaction between the two components.,title="fig:",width=158,height=207 ] , in absence ( @xmath120 , left panels ) and in presence ( @xmath126 , right panels ) of interaction between the two components.,title="fig:",width=158,height=207 ] but for the is - ia mode of the all repulsive case in fig .
[ fig3ms].,title="fig:",width=160,height=207 ] but for the is - ia mode of the all repulsive case in fig .
[ fig3ms].,title="fig:",width=160,height=207 ] in fig .
[ fig1ms ] we depict the lowest part of the band structure of eqs .
( [ eigen_spinor ] ) obtained for the cases of all attractive @xmath127 ( left panel ) and all repulsive @xmath128 ( right panel ) interactions .
we see that the nonlinearity induces in both cases a number of bound states in the band gap structure , located in the forbidden zone below the first band for the attractive case or in the gap between the first two bands for the repulsive case .
similar bound states can also form inside the upper band gaps ( see right panel of fig .
[ fig1ms ] ) . in this figure
bound state levels are labeled by symmetry type symbols close to the pair of the corresponding levels .
thus the os - os symbol denotes a bound state consisting of an os state in both components with the corresponding chemical potentials labeled by the open circle and the full square close to the symbol .
we find that the two component bound states of same symmetry type are degenerate in energy ( chemical potential ) both for all attractive and all repulsive interactions .
in contrast to the states os - os and oa - oa which are found to be stable also in absence of the interspecies interaction ( @xmath120 ) , the states ia - ia is stable only in presence of interaction between the two species and the state is - is is unstable also in presence of the interspecies interaction ( in this case the state decays a bit faster without interaction ) . in figs .
[ fig2ms ] and [ fig3ms ] we depict the eigenstates corresponding to the bound states shown in fig .
[ fig1ms ] .
in addition to the equal symmetry two - component bound states there are also states with mixed symmetry such as the oa - os and the is - ia bound states .
these states are very stable and represent new excitations of the two - component system since they can exist only in presence of the interspecie interaction . in particular we find that the mixed symmetry states oa - os and ia - is exist and are stable under time evolution only due to the interspecie interaction .
this is shown in figs .
[ fig4ms ] and [ fig5ms ] for the is - ia mode of the attractive and repulsive cases in fig .
[ fig2ms ] and fig .
[ fig3ms ] , respectively .
we see that in both cases modes are stable in both components only in presence of interaction and the switching off of the interaction makes the second component to decay into an os mode with emission of matter .
( left panel ) and as @xmath129 ( right panel ) .
other parameters are @xmath124 , @xmath130 .
open circles refer to the first component , filled squares to the second one .
notice that in this case the equal symmetry modes are non degenerated .
also notice the presence of localized modes near the edge of the upper bands.,title="fig:",width=160,height=160 ] ( left panel ) and as @xmath129 ( right panel ) .
other parameters are @xmath124 , @xmath130 .
open circles refer to the first component , filled squares to the second one .
notice that in this case the equal symmetry modes are non degenerated .
also notice the presence of localized modes near the edge of the upper bands.,title="fig:",width=160,height=160 ] .
thick and thin lines refer to first and second component , respectively ( open circles and filled squares in fig .
[ fig7ms ] ) . to identify the symmetry of the solutions , lattice potentials ( scaled by a factor 3 )
have been shown by thin dotted lines.,title="fig:",width=100,height=100 ] .
thick and thin lines refer to first and second component , respectively ( open circles and filled squares in fig .
[ fig7ms ] ) . to identify the symmetry of the solutions ,
lattice potentials ( scaled by a factor 3 ) have been shown by thin dotted lines.,title="fig:",width=100,height=100 ] .
thick and thin lines refer to first and second component , respectively ( open circles and filled squares in fig .
[ fig7ms ] ) . to identify the symmetry of the solutions , lattice potentials ( scaled by a factor 3 )
have been shown by thin dotted lines.,title="fig:",width=100,height=100 ] .
thick and thin lines refer to first and second component , respectively ( open circles and filled squares in fig .
[ fig7ms ] ) . to identify the symmetry of the solutions , lattice potentials ( scaled by a factor 3 )
have been shown by thin dotted lines.,title="fig:",width=100,height=100 ] .
thick and thin lines refer to first and second component , respectively ( open circles and filled squares in fig . [ fig7ms ] ) . to identify the symmetry of the solutions ,
lattice potentials ( scaled by a factor 3 ) have been shown by thin dotted lines.,title="fig:",width=100,height=100 ] .
thick and thin lines refer to first and second component , respectively ( open circles and filled squares in fig .
[ fig7ms ] ) . to identify the symmetry of the solutions , lattice potentials ( scaled by a factor 3 )
have been shown by thin dotted lines.,title="fig:",width=100,height=100 ] we have also investigated the case of mixed interactions @xmath131 ( see figs .
[ fig7ms],[fig8ms ] ) .
in contrast with previous cases where bound state levels were located in the semi - infinite gap or in the first band gap , in the case of mixed interactions the bound states are all below the first band , in spite of the fact that the second component has a repulsive character . in absence of the interaction , however , the bound states of the second component are indeed located in the first band gap as expected ( see left panel of fig .
[ fig7ms ] ) . by increasing the strength of the interspecies interaction these eigenvalues
get shifted below the first band , this being a consequence of the attraction between the two components ( for repulsive interspecies interactions the levels would be repelled into the gap making the formation of bound states more difficult ) .
also notice that in contrast with the all attractive and all repulsive cases , the equal symmetry localized modes ( as well as the mixed symmetry ones ) are non degenerate ( the components of the bound state have different chemical potentials ) , this being due to the fact that the unperturbed levels belong to different band gaps . by changing the values of @xmath132 ( by means of feshbach resonances ) one can indeed change the position of the bound state levels with respect to the band structure , this affecting the properties ( in particular the stability ) of the corresponding localized state .
similarly properties are found also for other parameter choices and for two - component modes states of higher band gaps .
the method of construction of localized modes based on continuation from the limit @xmath95 , can be used for obtaining rather different solutions of the coupled gp equations with a periodic potential , when all the nonlinearity coefficients are equal @xmath133 .
indeed , in that case the system ( [ eigen ] ) takes the form @xmath134 and is invariant under transformation @xmath135 let us now assume that one of the component , @xmath136 , is much larger than the other one , say @xmath1 . by using the method described in the previous section we can find stationary solutions of the coupled gp equations in the form @xmath137 ( notice that the chemical potential of the large - amplitude component is located near the top of the gap while that of the small - amplitude component is near the bottom of the gap , such that @xmath138 is of the order of the energy gap .
thus , substituting the stationary solutions @xmath139 into ( [ invar ] ) we readily obtain new solutions @xmath140 characterized by the amplitudes of the same order and by _ two _ chemical potentials , i.e. representing breathers ( in the sense of the periodically oscillating density of the condensate ) . in fig .
[ breath ] we show the dynamics of such breathing solution which appears to be dynamically stable .
besides two - components breather modes with unbalanced number of atoms , we also find breather modes with mixed symmetry which are very stable even for relatively large values of the number of atoms in both components .
the existence of these modes is expected both from previous considerations and from the existence and stability of the mixed symmetry modes with balanced number of atoms discussed in the previous section . and
for number of atoms @xmath125 .
the top panels show ten snapshots of the density profiles of the first ( first panel ) and second ( second panel ) component of the gap - breather taken at regular intervals of time during one period @xmath141 of oscillation .
the bottom panel show the amplitude of the left peaks of both components as a function of time .
the arrows indicate that the oscillation of the two components occurs in opposite phase .
, width=311,height=311 ] in fig .
[ fig6ms ] a gap - breather mode of the oa - os type is depicted for the case of all repulsive interactions .
we see that the oscillations of the two components are out of phase and are very regular , persisting for very long time without any emission of matter , this indicating that the mode is indeed very close to an exact breather solution of the two - component gpe system .
in the present paper we have presented a variety of localized modes , having stationary localized density distribution , and breathers , characterized by periodically varying density , in binary mixtures of bose - einstein condensates .
in particular , we have considered modes having either close or very different numbers of atoms ( we termed them balanced and unbalanced modes ) and characterized by the same or different symmetries of the atomic distributions in each components .
we have suggested algorithms of systematic finding families of the modes .
we exploited the fact that unbalanced modes bifurcate from the linear defect modes , the defect being understood as a localized excitation of one of the components in the absence of another component .
this allowed us construct exact breather modes in a systematic way .
we have introduced the classification of the modes on the basis of four possible distributions of each of the components , what in the general case results in the modes of 8 different types , which can be designated as bc@xmath142-bc@xmath143 , where b stands either for `` o '' in the case of the distribution centered in a local minimum of the periodic potential or for `` i '' in the case of the mode centered in the maximum of the potential , and c@xmath144 stands for s or a depending whether the wavefunction of the the @xmath11-th component is even ( s ) or odd ( a ) .
although our analysis was concentrated on spinor condensates , it allows for straightforward generalization for arbitrary two - component system described by coupled nonlinear schrdinger equations with periodic coefficients .
hac acknowledges support of the fct thorough the grant sfrh / bd/23283/2005 .
vab was supported by the fct grant sfrh / bpd/5632/2001 .
vvk acknowledges support from ministerio de educacin y ciencia ( mec , spain ) under the grant sab2005 - 0195 .
m. s. acknowledges partial financial support from the miur , through the inter - university project prin-2005 , and from the ( cnism ) consorzio nazionale interuniversitario per le scienze fisiche della materia .
the work of hac , vab , vvk , and gla was supported by the fct and feder under the grant poci / fis/56237/2004 .
in this appendix we briefly outline the implementation of the shooting method for finding localized solutions of ( [ eigen ] ) .
let @xmath99 and @xmath97 , @xmath98 be fixed and @xmath145 belong to gaps of the potentials @xmath146 . in the limit @xmath147 the equations in ( [ eigen ] ) separate . both of them become linear hill equations and the asymptotics of @xmath148 as @xmath147
are given by the formula ( @xmath2 ) @xmath149 where @xmath150-periodic functions @xmath151 and values @xmath152 are uniquely defined by the parameters of equations .
the parameters @xmath153 specify the solutions of ( [ eigen ] ) which tend to zero at @xmath147 . if @xmath153 are fixed the solution @xmath154 , subject to the conditions ( [ ascond ] ) at @xmath147 , can be found by shooting backward starting from large enough value of @xmath45 where this asymptotics is valid .
let us consider , for the sake of definiteness , _ even _ localized solutions of ( [ eigen ] ) .
they have to satisfy the conditions @xmath155 this is the system of two equations for two variables @xmath156 and @xmath157 .
generically , the solutions of this system are _ isolated _ [ in the sense that if @xmath158 and @xmath159 is a solution there is some neighborhood of @xmath160 which does not contain other solutions of ( [ eigen ] ) ] .
similar arguments hold for odd and also for non - symmetric solutions of ( [ eigen ] ) . in this way
one arrives at the conclusion formulated in the text .
myatt , _ et al .
rev . lett . * 78 * , 586 ( 1997 ) ; d.s .
hall , _ et al .
lett . * 81 * , 1539 ( 1998 ) ; d.s .
hall , _ et al .
_ , _ ibid_. * 81 * , 1543 ( 1998 ) ; m.r .
et al . _ , phys .
lett . * 81 * , 243 ( 1998 ) ; j. stenger , _ et al .
_ , nature ( london ) * 396 * 345 ( 1998 ) ; h .- j .
miesner , _ et al .
lett . * 82 * , 2228 ( 1999 ) ; g. modugno , _ et al .
_ , science * 294 * , 1320 ( 2001 ) ; m.r .
matthews , _ et al .
lett . * 83 * , 3358 ( 1999 ) ; m.r .
matthews , _ et al .
lett . * 83 * , 2498 ( 1999 ) ; j. williams , _ et al .
a. * 59 * , r31 ( 1999 ) . | we study localized modes in binary mixtures of bose - einstein condensates embedded in one - dimensional optical lattices .
we report a diversity of asymmetric modes and investigate their dynamics .
we concentrate on the cases where one of the components is dominant , i.e. has much larger number of atoms than the other one , and where both components have the numbers of atoms of the same order but different symmetries . in the first case
we propose a method of systematic obtaining the modes , considering the `` small '' component as bifurcating from the continuum spectrum .
a generalization of this approach combined with the use of the symmetry of the coupled gross - pitaevskii equations allows obtaining breather modes , which are also presented . |
: : h : hermitian hamiltonian : : h : non hermitian hamiltonian : : @xmath0 : metric for inner product of the state space of h : : @xmath1 : square root of @xmath0 : : @xmath2 and @xmath3 and @xmath4 are state spaces of h and h and @xmath5 respectively : : @xmath6 and @xmath7 are observables for h and h respectively .
: : @xmath8 and @xmath9 are density matrices for h an h respectively . : : @xmath10 and @xmath11 are projection operators for h and h respectively . : : v ( ) is von - neumann entropy and e ( ) is entanglement entropy : : @xmath12 and @xmath13 are eigenvalues of @xmath14 and @xmath5 : : @xmath15 and @xmath16 are states for hermitian h , @xmath17 and @xmath18 for pseudo - hermitian h and @xmath19 and @xmath20 for pseudo - hermitian @xmath5
quantum physicists has always known that a non - dissipating physical system can always be given by a hermitian hamiltonian which has all the ingredients of reality of eigenvalues ( as they represent energy ) and norm preserving time evolutions ( as no particle or energy exchange is happening ) . in 1998
bender et .
al , @xcite studied eigenvalues of non - hermitian hamiltonian , numerically .
they found to their surprise that these hamiltonians had a spectrum of real eigenvalues in a certain parameter range .
moreover , these hamiltonians were pt ( parity , time reversal)-symmetric .
however , the state space of such a system was not complete orthogonal ( hence , it could nt form a hilbert space ) , as it is needed to represent a non dissipative quantum physical system ( which was until now always represented by hermitian hamiltonians ) .
several attempts were made by people to solve this difficulty,@xcite .
it was however shown by , mostafazadeh,@xcite that rather than focussing on pt - symmetry one should focus on the non - hermitian property of these hamiltonians , and he provided a general framework which showed that all non - hermitian hamiltonians in the real eigenvalue regime will show an exact behavior like systems represented by hermitian hamiltonians , by changing the metric of state space of these hamiltonians . however , in the complex eigenvalue regime it can be seen that the system does not have norm preserving time evolutions even after changing metric and has complex eigenvalues , which is interpreted as onset of dissipation in this paper and , @xcite and many others . now , the use of non - hermitian mathematical framework has been considered by , @xcite where they have used it in classical optics by taking two waveguides , one of it loosing and other gaining , but overall the system is non - dissipating . they have specifically taken up the idea of symmetry breaking of non - hermitian systems , and used it in optical waveguide theory .
we in this paper follow that the formalism of pseudo - hermiticity can be used to describe non - dissipating regime of a system whose dissipating nature depends on certain parameters , @xcite and it has been verified in the experiment that by going above certain threshold given by this theory ( which , @xcite called optical gain / loss coefficient ) , a non - hermitian system starts dissipating .
the outline of the paper is as follows , section 1,2,3,4,5 gives necessary and sufficient proofs for establishing the iso - spectrality between a pseudo - hermitian system and a hermitian hamiltonian system and addresses important subtleties creating confusion regarding this theory .
section 6 establishes the theory of measurement in these systems and section 7,8 describes non - composite pseduo - hermitian systems in density operator language .
section 8,9,10,11 describe composite systems involving pseudo - hermitian systems and establish important quantum measures and clarifications to .@xcite we will first of all clarify certain ideas related to this field and direct to appendix for particular examples of the framework .
there are certain ideas which need to be clarified using geometry of projection of state space of a 2@xmath212 non - hermitian quantum system with real eigenvalues , which we have provided in appendix .
also , hence forward we will designate non - dissipating systems in real eigenvalue regime as pseudo - hermitian .
here we give minimal amount of proofs needed to establish this theory consistently .
we start by asking what is the relationship between eigenvalues of non - hermitian matrix and its hermitian conjugate .
consider an operator @xmath14 and @xmath5 s.t ( such that ) @xmath22 and they obey an eigenvalue problem , then the eigenvalues of @xmath14 and @xmath5 , denoted by @xmath12 and @xmath13 respectively are related as @xmath23 ( where * denotes complex conjugation ) . we know from given information that , @xmath24 and @xmath25 where @xmath18 and @xmath20 are eigenvectors of h and @xmath5 respectively i.e @xmath26 @xmath27 and @xmath28 @xmath29 where @xmath27 and @xmath29are vector spaces with inner product @xmath30.now , taking conjugate transpose of ( 1)and then multiplying ( 1 ) by @xmath31 from right we have @xmath32 using ( 2 ) we see that @xmath33 hence we can see that canceling @xmath34 that @xmath35 hence we can see that when eigenvalues are real they are equal @xmath36 + we know that matrices with same eigenvalues are similar @xcite , i.e. they are related by similarity transformation
. we now check what will be the similar transformation between these non - hermitian matrices in real eigenvalue regime . in the real eigenvalue regime of h ,
h and @xmath5 are related by @xmath0h@xmath37=@xmath5 ; s.t .
@xmath38 where n is the dimension of h , @xmath39 and @xmath40 .
note : @xmath27 and @xmath29 are just state space of @xmath14 and @xmath5 and they are not yet rendered into hilbert space .
we know @xmath41=@xmath42 now conjugate transposing we get @xmath43 as proved earlier in real eigenvalue regime , @xmath44 we have @xmath45 multiplying @xmath46 from left we have @xmath47 also multiplying @xmath41=@xmath42 by @xmath48 from right we have equating with ( 7 ) that @xmath49 summing over i=1 to n gives @xmath50 where @xmath38 let us check the properties of the matrix @xmath0 . note : we have not yet proved that @xmath3 i.e state space of h forms a hilbert space ( complete orthogonal vector space ) .
we have not yet proved @xmath0 is invertible , such that we can always use the relation @xmath51=@xmath5 for finite dimensions . if we have @xmath52 and eigenvalues of @xmath14 are necessarily real , then ( @xmath18,@xmath20 ) will be orthogonal to each other under @xmath53 as inner product and where , @xmath39 and @xmath40 we know that @xmath0h@xmath37=@xmath5 in real eigenvalue regime @xmath54 @xmath55 previous theorem gives @xmath56 hence , by ( 10 ) we see that @xmath57 therefore , @xmath58 has to be @xmath59 so that , @xmath60 we can see that if we use the operator @xmath0 which we previously saw was the operator relating h and h@xmath61 by similarity transform , that @xmath62 @xmath63 hence @xmath0 renders the state space of h , @xmath27 to be now a complete orthonormal vector space , i.e a hilbert space with sesquilinear form @xmath64 as the inner product .
let us now on denote it by @xmath3 .
it can also be verified by example that gram - schmidt orthogonalisation will not work for non - hermitian systems if it uses identity as metric .
we can easily see that @xmath0=@xmath65 : @xmath66 as ( ab)@xmath61=b@xmath61a@xmath61 , where a and b are matrices .
hence , we have @xmath67 @xmath0 is always invertible see , @xcite . as @xmath0 has a full rank because , @xmath68 and all @xmath69 are linearly independent of each other as they are eigenvectors of h@xmath61 .
hence , the relation @xmath70 will always exist for finite dimensions .
some emphasis has to be given to square roots of matrices , because they will be useful to elaborate isospectrality .
a matrix has square root iff it is diagonalizable , and a hermitian matrix is a always diagonalizable , hence square root of a hermitian matrix always exists s.t .
a@xmath61a = b , where a is square root of b. a matrix has many roots .
a point to be noted is not all square roots of a hermitian matrix be hermitian ( not properly emphasized even in , @xcite ) , example : @xmath71 we can see that @xmath72 a@xmath73a@xmath61 but b = b@xmath61 where a , b@xmath74 .
also , it can be easily checked that as @xmath0 is always invertible it s root is also always invertible .
this property has implications as we will see in section 5 .
we already saw that @xmath0h@xmath37=@xmath5 ; we define this property as pseudo - hermiticity , for hermitian h we see that @xmath0=@xmath75 , identity matrix . analogous to unitary matrix u@xmath61=u@xmath76 similarly a matrix p we define pseudo - unitarity as @xmath77 s.t . @xmath78 note : unitary transformations preserve norm for hermitian systems , then pseudo - unitary transformations preserve norm for pseudo - hermitian systems .
we saw that @xmath0 is hermitian and hence it s root always exists .
@xmath54 we can always write @xmath79 suppose , we define an operator h by a similarity transform on pseudo - hermitian operator h then , @xmath1h@xmath80=h prove that h is hermitian .
we know @xmath1h@xmath80=h , then , h@xmath81=(@xmath80)@xmath61h@xmath61@xmath82 , but h@xmath61=@xmath0h@xmath83 so @xmath84 @xmath85 as , @xmath1 always exists and hence h always will exist such that h is hermitian , h and h are isospectral partners because similarity transformation always preserves the eigenvalue of the operator .
moreover , we will see that the observables in pseudo - hermitian theory are different in form than that of hermitian theory , but their values always exactly match that of isospectral hermitian partner .
hence , there will be no way to distinguish such systems in real eigenvalue regime .
+ moreover , we saw that for finite dimensions @xmath1 will always exist without exception as @xmath0 will always exist , and thus there will always exist a @xmath86 for corresponding @xmath14 . in conventional theory an observable is defined as an operator which provides real expectation values , moreover adjointness of the operator is also important , usually in textbooks , hermiticity and self - adjointness are synonymous with each other but it is not true .
an , operator @xmath87 is called adjoint of @xmath6 if , @xmath88 moreover , @xmath6 is called self - adjoint if @xmath89 .
so , as we see in hermitian theory , h = h@xmath90 and hence a is also self - adjoint in hermitian theory . in pseudo - hermitian theory as @xmath0h@xmath83=h@xmath90 , h is called pseudo - hermitian , it is also called as pseudo - self - adjoint @xmath91 . in a similar fashion , all other observables in pseudo - hermitian theory are pseudo - self - adjoint i.e. @xmath92 this , can be seen in the example of 2@xmath212 matrix given in appendix , that pauli matrices do not remain usual pauli matrices in the pseudo - hermitian theory , otherwise we would get complex expectation values of spin .
now , we will see that the states of pseudo - hermitian hamiltonian have same expectation values of observables as given by states of it s hermitian iso - spectral partners .
similar to h it can be seen that all observables are related to their hermitian iso - spectral observables by transformation , @xmath93 where @xmath7 is an observable for a pseudo - hermitian hamiltonian system while @xmath6 is an observable for a hermitian hamiltonian system , and @xmath94 . as , we saw earlier , @xmath1h@xmath80=h , where h = h@xmath90 and @xmath95=@xmath0 , @xmath0 being the metric of hilbert space of h. so , @xmath96 , where @xmath97 and @xmath98 .
@xmath99 hence , we can see that their expectation values match and there is no way to distinguish a hermitian system from a system with non - hermitian hamiltonian in the real eigenvalue regime .
however , as stated in appendix the structure of the state space might seem different from that of it s isospectral hermitian partner , but it is not ( see appendix and section 8 of , @xcite ) .
@xcite mentions that composite systems might provide a way to distinguish these systems , in this paper we show that even in these kind of systems we can find an isospectral hermitian hamiltonian , also we provide an information theoretic framework and show that von - neumann entropy of h and it s partner h provide same value provided the form of von - neumann entropy in pseudo - hermitian theory is changed in form just like the form of observables are changed .
we also , show that entanglement entropy of composite systems of hermitian@xmath100pseudo - hermitian and pseudo - hermitian@xmath100pseudo - hermitian will not change if they are transformed to their isospectral partners under similarity transform .
for this in later sections we will have to properly develop pseudo - hermiticity in density operator language and try to leave no room for error .
the method given in , @xcite to effectively describe a dissipative system involves , complex hamiltonian @xmath101 where , @xmath102 denotes the hamiltonian for the system and @xmath103 is hermitian part ( hence always has real eigenvalues ) and @xmath104 is the part through which we can control dissipation and is dependent on parameter @xmath105 .
now , for the bath the hamiltonian is given as , @xmath106 s.t .
@xmath107 where h is hermitian and has real eigenvalues ( hence overall it is not dissipating ) .
it can be seen that if only the system hamiltonian is considered , then it is given as ( 18 ) , and in the real eigenvalue regime as stated in previous two sections , the system acts like a hermitian system .
note that it is not equivalent to making @xmath108 but even if @xmath109 the system acts as if it is not dissipating .
it can be seen using the example @xmath110 matrix given in appendix .
also , it has been experimentally verified in .@xcite
in hermitian theory we have two major types of measurements i.e. pvm and povms.@xcite pvms : these are measurements where the observation does not destroy the observed state , repeated measurements provide same results .
these are idealised measurements and are extremely rare to create .
mathematically , in hermitian theory , these measurements are given by projection operators which have following two important properties , @xmath10 .
@xmath111 and @xmath112also @xmath113 now , @xmath114 where , @xmath115 are basis of h ; s.t .
h = h@xmath61 .
if we define @xmath116 then @xmath117 all the above properties can be verified for @xmath118 .
any state @xmath119 where @xmath120 are the probability amplitudes for eigenvectors @xmath115 of observable @xmath6 and they have complex values .
the probability of obtaining particular ith state is .
@xmath121 after . the measurement @xmath15 changes to @xmath122 , if a measurement in such a type of scheme is performed again we get @xmath123 and it is same as @xmath122 @xcite .
these types of measurements are called pvm or projection operator valued measurements. @xmath124 as the inner products are changed the outer products also need to be changed for consistency . hence , the projection operator for pseudo - hermitian hamiltonian is given as @xmath125 s.t .
@xmath126 @xmath127 are probability amplitudes , and the probability of obtaining ith state , upon measurement , when a system is in state @xmath128 is @xmath129 usually in hermitian theory we write @xmath130 but this statement is more subtle than this .
the point is that for complex matrices we reserve the definition @xmath131 for @xmath132 and in hermitian theory we have @xmath133 hence @xmath134 as a fact of idempotency .
however if we follow the same definition of @xmath135 in pseudo - hermitian theory it creates ambiguities .
but it can be seen that in pseudo - hermitian theory even if @xmath136 @xmath137 and hence it can be seen that operating projection operator @xmath11 twice does not affect the previous measurement result .
note that @xmath11 is a pseudo - hermitian operator .
we will see what happens when repeated measurements affect the state after measurement .
povm ( positive operator valued measure ) are more general type of measurements than pvms .
unlike , pvm repeated measurements do not provide same measurement values , ex : a photon falling on a photographic plate will have been altered drastically to be measured again .
mathematically , these are represented by more general positive semidefinite operators , which have following properties , @xmath138 always , m is just the index for denoting mth state .
post - measurement state is given by @xmath139 upto normalisation .
but repeated measurement will not give same value .
for , pseudo - hermitian theory , @xmath140 are selected such that they do not from @xmath141 .
first , we develop a density operator formalism for pseudo - hermitian theory .
this is done because this formalism provides more flexibility to talk about mixed states .
later we consider , composite systems and quantify certain properties of these quantum systems , like entanglement . under this formalism
we see that , there is no way that new phenomenon can emerge from composite systems of pseudo - hermitian@xmath100hermitian systems(denoted h@xmath100h ) and pseudo - hermitian@xmath100pseudo - hermitian ( denoted h@xmath142 @xmath100h@xmath143 ) systems .
also , it is shown that the form of entanglement entropy(e ( ) ) as well as of von neumann entropy ( v ( ) ) has to be changed and under this new form it can be shown ( for both cases stated above ) that the values e ( ) of an entangled state in h and of an isospectral entangled state in h will be same , hence the entanglement entropy does not reduce as stated in , @xcite due to isospectrality . also , it is easy to prove that rate of entanglement of both of these types of systems will be same .
let us usually denote hermitian systems by h and it s eigenvectors as @xmath144 which form an orthogonal basis for @xmath2 hilbert space for h. @xmath145 .
@xmath146 is the density operator for pure states of h. see , @xcite . where , @xmath147 , where z denotes state of the system and second index denotes the basis vector of the system being considered , also @xmath148 , are probability amplitudes . following
relations always hold for such systems , @xmath149 and tr@xmath150 , and tr@xmath151 , because @xmath152 .
also , expectation values of any observable @xmath6 is given by @xmath153=tr@xmath154 all these properties can be checked taking a state as @xmath155 . for pseudo - hermitian system h , with eigenvectors @xmath156 and having a hilbert space @xmath3 with metric @xmath0 , s.t .
@xmath68 where @xmath157 , n being dimension of h. analogously for pure states of pseudo - hermitian systems , we define @xmath158 sticking to this definition , we can see that @xmath159 , where m denotes the state of system .
note , @xmath160 .
but it can be checked that @xmath161 i.e. it is pseudo - hermitian . tr@xmath162=@xmath163=1 this happens because @xmath164=1 . also ,
@xmath165,but @xmath166 the expectation value of an observable @xmath7 is given as @xmath167 now , using the iso - spectrality property we can see that , @xmath168 where @xmath1 is the similarity transformation operator between pseudo - hermitian hamiltonian h to it s iso - spectral hermitian hamiltonian h. @xmath169 and it can also be checked that , using cyclic property of trace , @xmath170 @xmath171 h and h are indistinguishable .
there are two cases which arise for statistical mixtures , ( pseudo - hermitian and pseudo - hermitian ) and ( pseudo - hermitian and hermitian ) mixtures .
firstly , revising hermitian systems theory , state of mixtures can not be written as @xmath172 .
they are formed due to incomplete information .
example : a screen which is impinged upon by 70% of vertically polarised light from one source and 30% from other being horizontally polarised .
they can only be written in density operator form .
@xmath173 , where p@xmath174 denote probability of finding a certain state forming the mixture and @xmath175 , l is the number of systems forming the mixture.(ex:@xmath176 ) .
75% are impinged upon from system 1 as @xmath177 and other 30% from system 2 as @xmath178 .
tr@xmath179 and tr@xmath180 .
density operator in such conditions is given as @xmath181 where , @xmath182 are the metric for corresponding systems of which mixture is made of , @xmath175 where p denote the percentage of the individual systems which make the mixture , l is the number of individual systems making up the mixture and @xmath183 be individual density operators .
analogous to hermitian case , @xmath184 . due to the iso - spectrality
it can be seen that individual systems @xmath185and we can write , @xmath186 and we get the expectation values , by substituting previous results @xmath187 out of total l mixtures consider a mixture of e pseudo - hermitian systems and l - e hermitian systems .
then , the density operator is defined as , @xmath188 where @xmath189 and @xmath190 and @xmath182 are metric for corresponding hilbert spaces of pseudo - hermitian systems . here , as in previous case we define @xmath183 and @xmath191 , also as usual notation h correspond to non - hermitian and h correspond to hermitian , also @xmath192 it can also be easily checked that , tr@xmath193=1 .
also as we saw in previous section , @xmath194 and and the relation @xmath195 , where these new @xmath196 belong to hermitian isospectral system of h@xmath174s .
now , we get using all these results that , @xmath197 @xmath198 @xmath199 it can be seen in all the scenarios of a non - composite system , the open system in real eigenvalue regime acts like a hermitian system(non - dissipating system ) .
as we changed the form of observables in a hermitian setting from @xmath6 to @xmath7 , by relation @xmath200 .
an example is properly elaborated in the appendix that the representation of pauli matrices need to be changed to properly get real expectation values of spin .
it is a simple fact that due to isospectrality the values of entropy of h and h will be same because the von - neumann entropy is defined as @xmath201 , where @xmath202 are the eigenvalues of @xmath9 and @xmath8 both .
this happens due to similarity equivalence between them , and similarity transformations preserve eigenvalues .
this is contrast to ideas taken up by , @xcite where they have ignored that the two equivalent formulae for von - neumann entropy do not follow in the same way as in pseudo - hermitian theory .
this can be proved in density matrix form using the iso - spectrality condition . in hermitian
setting , another way of quantifying von - neumann entropy is given as @xmath203 where logarithm is to the base 2 .
for , pseudo - hermitian hamiltonians however the proper form of von - neumann entropy should be given as below , so that there is consistency in definition of von - neumann entropy @xmath204 then it can be easily seen that due to the transformation .
@xmath205 , that @xmath206
let us first revise composite systems in hermitian theory .
usually if two systems are to be represented simultaneously , then these systems are to be represented using tensor product @xmath100 of the matrices representing them .
let us consider for simplicity of case of bipartite ( i.e 2 systems ) composite system , which can be generalized to many more systems ( multipartite ) by generalization .
as we are for revision considering bipartite hermitian composite system , we denote the hamiltonians of these systems as , @xmath207 and @xmath208 , which form a composite system @xmath86 denoted by , @xmath209 .
we will denote the dimensions of these systems by e and f. any state residing in the hilbert space of such a system is given by , @xmath210 where @xmath211 and @xmath212 and @xmath213 and @xmath214 are probability amplitudes as in @xmath215 .
we will particularly focus on what happens to inner product metric for such systems , it is not always explicitly emphasized that metric of inner product of a bipartite systems is @xmath216 .
these metrics are such that @xmath217 and @xmath218 . @xmath219 is called seperable if it can be decomposed as , @xmath220 .
if it can not be decomposed as earlier then it is called entangled. a few important results while using tensor products are as such , @xmath221 a , b , c , d can be any column , row , square matrices .
therefore , @xmath222 @xmath217and @xmath218 and using the properties stated above . for simplicity , we will consider henceforward , that we are considering systems in pure states .
density matrix of a composite system is given as,@xmath223 where , @xmath224 . and
@xmath225 von neumann entropy is given explicitly as @xmath226 where @xmath227 and we are considering that @xmath8 is density matrix for a pure state , it can easily be generalised for mixed states and @xmath228 are eigenvalues of the density operator @xmath8 .
s is the dimension of the square density matrix where e@xmath21f = s , where e is dimension of @xmath207 and f is dimension of @xmath208 .
also , it is known that this quantity is invariant under unitary similarity transformation .
partial trace is a mathematical tool of describing individual properties of a composite system .
partial trace of a density matrix , @xmath8 is denoted by @xmath229 and @xmath230 , where @xmath227 .
the prescription to find it is given as @xmath231 where @xmath232 example : @xmath233 here @xmath234 @xmath235 and @xmath236 and @xmath237 , e = f=2 .
entanglement entropy of composite systems is defined as , @xmath239 or another definition @xmath240 where @xmath219 is an entangled state and @xmath241 is the density operator for that state and , @xmath242 and @xmath243 are found by the prescription given above . in explicit way
they can be written as , @xmath244 this can be written due to the simple fact that @xmath245 is always unitary and entropy is invariant under unitary transformations .
it can also be equivalently given as , @xmath246 .
writing entanglement entropy this way provides clarification later .
consider , case of a bipartite composite system of two pseudo - hermitian hamiltonians given as @xmath247 and having eigenvectors @xmath248 and @xmath249 s.t .
@xmath250 form a biorthogonal basis under metric @xmath251 ( analogous to @xmath252 ) . @xmath253 and @xmath254 where @xmath255 and @xmath256 .
any state of hilbert space of such a system is given as , @xmath257 where @xmath258 and @xmath259 . and @xmath260 .
@xmath261 and @xmath262 .
now , analogous to hermitian case @xmath128 is called seperable if , @xmath263 .
otherwise it is called entangled .
@xmath264 as we know , @xmath261 and @xmath262 . we have , @xmath265 which applies to both entangled and separable states .
density matrix of such a composite system is given as @xmath266 s.t .
@xmath267 another important result is that as @xmath268 and as we already saw , @xmath269 and @xmath270 then we have already seen that for a particular non composite system hamiltonian , @xmath271 and @xmath272 , where after transformation @xmath273 we have @xmath274 .
now we check for composite system , that @xmath9 is equal to @xmath275 as we know , @xmath276 . then @xmath9 becomes @xmath277 last two brackets become @xmath278 defining @xmath279 and @xmath280 we have @xmath281 .
hence , there exists a similarity transformation@xmath282 s.t .
now , von neumann entropy as defined earlier gives @xmath283 substituting @xmath271 it gives @xmath284 hence , we can see that given a composite system on pseudo - hermitian systems their entropies are same as that of composite system made of isospectral partners of individual pseudo - hermitian systems .
this can also be seen for entanglement entropy . for
that we first need to formalise the act of partial tracing for such a theory .
the density matrix of a system making a composite system is found for a pseudo - hermitian system by partial tracing as given below @xmath285 as seen earlier for hermitian case it can be reduced to , @xmath286 and @xmath287 using definition of von neumann entropy for single pseudo - hermitian systems as formalized earlier and defining entanglement entropy for case 1 of composite systems containing these systems we see that , entanglement entropy needs to be defined as @xmath288 as we have already seen that @xmath289 and @xmath290 we have @xmath291 where @xmath292 and therefore , entanglement entropy @xmath293 becomes @xmath294 hence we can see that the value of entanglement entropy of an entangled state from composite system of 2 pseudo - hermitian systems will be same as entangled entropy of an entangled state from a hermitian isospectral system .
all of these properties can be checked using a bipartite composite system @xmath295 consider , the bipartite composite system of a pseudo - hermitian ( @xmath296 ) and a hermitian ( @xmath208 ) hamiltonian systems .
the eigenvectors of @xmath297 will be @xmath298 s.t .
the , metric will be @xmath299 , where @xmath300 s.t .
any arbitrary state of such a composite system is given as , @xmath301 where @xmath302 and @xmath303 and e and f are dimensions of @xmath296 and @xmath208.also , @xmath304 s.t .
@xmath305 and @xmath306 @xmath128 is called separable if , , @xmath307 otherwise it is entangled .
also , it can be seen that @xmath308 because , @xmath309 and @xmath310
. density matrix of such a composite system is given as @xmath311 s.t .
@xmath312 another important result is that as @xmath313 and as we already saw , @xmath269 and @xmath314 then we have already seen that for a particular non composite system hamiltonian , @xmath271 and @xmath272 , where after transformation @xmath273 we have @xmath274 .
now we check for composite system , we have @xmath9 is equal to @xmath315 which becomes after using previous similarity transform @xmath1 @xmath316 as we know , @xmath276 . then @xmath9 becomes @xmath317 last two brackets become @xmath318 defining @xmath319 and @xmath320 we have @xmath281 where @xmath321 .
now , the von neumann entropy as defined earlier gives @xmath283 substituting @xmath271 it gives @xmath284 hence , we can see that given a composite system of pseudo - hermitian and a hermitian system , their entropy values are same as that of composite system made of their isospectral hermitian partner of pseudo - hermitian systems embedded in it and of respective hermitian systems already present .
this can also be seen for entanglement entropy .
for that we again first need to formalize the act of partial tracing for such systems . the density matrix of a pseudo - hermitian part of composite case 2 system is given by partial tracing , below as seen earlier for hermitian case it can be reduced to , @xmath322 and @xmath323 using definition of von neumann entropy for single pseudo - hermitian systems as derived earlier and defining entanglement entropy for case 2 of composite systems containing these individual systems we see that , entanglement entropy needs to be defined as @xmath288 using the case whichever maybe , also note that @xmath245 is always unitary , instead it is equivalent definition of a unitary matrix . for first definition
it can be seen that @xmath289 and @xmath290 and @xmath324we have @xmath291 where @xmath292 and therefore , entanglement entropy @xmath293 becomes @xmath294 where @xmath219 in h is now a isospectral entangled state of @xmath128 , an entangled state in h. hence we can see that the value of entanglement entropy of an entangled state from composite system of some pseudo - hermitian systems and some hermitian systems will be same as entangled entropy of an entangled state from a hermitian isospectral system of the pseudo - hermitian and the already present hermitian systems . all of these properties can be checked using a bipartite composite system @xmath325
it can now be seen that rate of entanglement of composite systems having pseudo - hermitian systems will have same value as that of an equivalent hermitian isospectral composite system . as per the definition of rate of entanglement production by a particular composite system h for a particular entangled state @xmath17
we write @xmath326 as we already saw , entanglement entropies @xmath327 where @xmath17 is an entangled state which belongs to pseudo - hermitian composite system h and @xmath15 belongs to it s isospectral composite hermitian system h. we have , @xmath328
we learned that theory of pseudo - hermiticity can be consistently applied to dissipative quantum systems , @xcite .
it was presented in density operator language , which has it s own pros and cons . a few clarifications on , @xcitewere provided .
we saw that the von - neumann entropy and entanglement entropy of these non - hermitian systems ( in real eigenvalue regime)are same a their hermitian counterparts .
hence , it does not have effect on the rate of entanglement creation as shown in @xcite , nor can composite systems of these systems be used to distinguish the non - hermitian and their hermitian iso - spectral partners .
here we will give examples of the results obtained earlier .
firstly we give the example of a @xmath329 non - hermitian matrix in the real eigenvalue regime , and show how it is related to an isospectral - hermitian hamiltonian , both algebraically and geometrically .
next we will see by analogy the geometry of a composite pseudo - hermitian system .
the following appendix is based on , @xcite .
we can see in both cases eigenvectors are orthogonal to each other . as we know
the state space of a hermitian @xmath329 system can be projected onto a bloch sphere , always .
we can see that geometrically the act of unitary transformation just rotates the bloch sphere .
it can be seen as follows , by calculating the bloch vector , it shows that for the eigenvector of hamiltonian in eq(41 ) @xmath334 and for eq(42 ) it is @xmath335 which shows that their bloch vectors are only rotated by @xmath105 angle .
it can also be seen that the geodesic distance between both eigenvectors of eq(41 ) and of eq(42 ) have distance @xmath1 and hence they are orthogonal to each other . the geodesic distance between @xmath336 and @xmath337
is given as @xmath338 .
moreover , the eigenvalues and expectation values of the observables of these systems will be real as is already known . however , the hamiltonian is not unitarily equivalent to eq(41 ) or eq(42 ) hamiltonian . instead , if a matrix is untarily equivalent to one matrix then it is unitarily equivalent to all other matrices which are unitarily equivalent to the first matrix .
this , hamiltonian is non - hermitian with eigenvalues @xmath340 the eigenvectors for @xmath341 and @xmath342 in the real eigenvalue regime ( i.e. when @xmath344 where , @xmath345 and eigenvalues become @xmath346 .
it can be easily seen that the eigenvectors are not orthonormal to each other and if we find the eigenvectors of @xmath5 in the real eigenvalue regime then they turn out to be .
it can be seen that @xmath348 and @xmath349 , we can create an operator @xmath350 s.t we get all the eigenvectors of eq(43 ) be orthogonal to each other i.e. @xmath351 and @xmath352 where , now representation of @xmath0 = and square root of @xmath0 which is found by diagonalizing @xmath0 and then taking square roots of the diagonal elements of @xmath0 and then again transforming the root diagonal form to root non - diagonal form .
we , get @xmath354 where @xmath355 and @xmath356 it can be easily be seen that @xmath1 is not unitary ( @xmath357 ) . @xmath358
it can now be checked that h can be transformed to an isospectral hermitian hamiltonian by a similarity transform using @xmath1 s.t .
@xmath273 and @xmath274
. for , this case the hamiltonian turns out to be @xmath359 with eigenvectors , @xmath360 it can be easily seen that it s eigenvalue is same as h , the eigenvectors @xmath16 being , @xmath361 s.t . @xmath362 and @xmath363 .
hence the geodesic distance between @xmath364 and @xmath365 is @xmath1 as their values are same as the orthogonal eigenvectors of h and it is well known that the geodesic distance between orthogonal vectors under metric @xmath75 is @xmath1 , where @xmath1 here is angle in radians .
this can also be checked by using the formula for geodesic distance between two states in the projective space of the state space of a pseudo - hermitian system , given by , @xmath366 check , @xcite .
note now if we try to find the bloch vectors of corresponding eigenvectors of eq(43 ) using usual pauli matrices then @xmath367 gives value @xmath368 which is a complex value .
hence , the expectation value of spin will be complex , so the observables need to be changed .
we will see that indeed we just changed the representation of the observables however they will give same expectation values for corresponding isospectral states . as
, we already saw the observable given by hamiltonian is changed as @xmath273 the pauli matrices will also change as @xmath369 s.t . now it can be easily seen that @xmath370 and hence the projective space of h is same as that of h , which is hermitian .
we can check that only the arrangement of the state space has changed between the pseudo - hermitian and iso - spectral hermtian system .
all the expectation values will match between the systems .
all of the above properties can be visualized as given by fig.1 .
we will have for a separable state of such a composite system that the complex projective space of it can be decomposed into two individual bloch spheres representing the two states making up this seperable state .
also , an entangled state will be represented by an @xmath373 ( unit sphere in 8 dimensions ) as is done in hermitian case .
guo , a. and salamo , g. j. and duchesne , d. and morandotti , r. and volatier - ravat , m. and aimez , v. and siviloglou , g. a. and christodoulides , d. n. , _ observation of pt - symmetry breaking in complex optical potentials_,phys . rev .
103 , ( 2009 ) . | hamiltonian mechanics works for conserved systems and quantum mechanics is given in hamiltonian language .
it is considered that complexifying the quantum hamiltonian a balanced loss and gain model can be created .
the usual mathematics of density operator formalism and entanglement is extrapolated to such systems and the consequences are studied .
namely , a complete formalism using density operators is created for real eigenvalue regime of these non - hermitian systems and correct forms of von - neumann and entanglement entropy are created .
the consequences are studied in this regime and depicted w.r.t recent papers by @xcite . |
let @xmath3 denote the deligne - mumford moduli space of genus @xmath0 stable curves .
our aim is to study the effective cone of @xmath4 . recall that @xmath5 is generated by the hodge class @xmath6 and boundary classes @xmath7.$ ] the study of effective divisors on @xmath3 has a long history dating back to @xcite , @xcite and @xcite . the most important result in this circle of ideas
is the following .
@xmath3 is of general type when @xmath8 .
the idea of the proof is to write the canonical divisor in the form @xmath9 , where @xmath10 and @xmath11 are ample and effective @xmath12-divisors respectively .
then @xmath13 is big and @xmath3 is of general type . in practice , we know that @xmath14 and @xmath15 is ample for @xmath16 , where @xmath17 is the total boundary .
so as long as we obtain an effective divisor @xmath18}b_{i}\delta_{i } } , a , b_{i } > 0 $ ] with @xmath19 for all @xmath20 and @xmath21 , then we can write @xmath22 in the desired form @xmath23 .
define the number @xmath24 to be the slope @xmath25 of the divisor @xmath26 .
then the remaining task is to find effective divisors with small slope .
+ in @xcite , the famous brill - noether divisor was shown to do the trick . recall that when the brill - noether number @xmath27 is equal to @xmath28 , the brill - noether divisor @xmath29 consists of genus @xmath0 curves which possess a @xmath30 .
amazingly , @xmath31 , the so - called brill - noether bound , is independent of @xmath32 , and is less than @xmath33 if @xmath8 . for almost 20 years
after the publication of @xcite , no one succeeded in finding any effective divisor of slope lower than the brill - noether bound .
this led to the formulation , in @xcite , of the following slope conjecture .
the brill - noether bound @xmath34 provides a lower bound for slopes of effective divisors on @xmath3 .
the bound is sharp if and only if @xmath35 is composite .
unfortunately , the slope conjecture is false . in @xcite ,
farkas and popa study genus 10 curves contained in @xmath36 surfaces , or equivalently curves embedded in @xmath37 by @xmath38 s whose images are contained in quadric threefolds .
these curves sweep out a divisor @xmath39 in @xmath40 .
surprisingly , @xmath41 , which is less than the brill - noether bound @xmath42 .
so @xmath43 serves as a first counterexample to the slope conjecture .
recently , in @xcite , @xcite , @xcite and @xcite , more counterexamples are constructed , and in these papers farkas also announces that @xmath44 and @xmath45 are of general type .
+ the above results only tell us one side of the story , where slopes of effective divisors are concerned . on the other side , to the best of the author s knowledge , all effective divisors that have been studied thus far have slope greater than 6 .
so in some sense , the slope conjecture is not far from the truth .
a natural question is to ask : is there a lower bound for slopes of effective divisors on @xmath3 ?
a uniform lower bound @xmath46 for all @xmath0 would enable us to solve the schottky problem , which roughly says that the modular forms of slope @xmath47 cut out @xmath48 in @xmath49 , the moduli space for abelian varieties of dimension @xmath0 , see , e.g. , @xcite .
+ the idea to obtain a lower bound is to construct moving curves in @xmath3 .
an irreducible curve @xmath50 is called a moving curve if the family of its deformations fills in a zariski open set of @xmath3 .
we can define the slope of @xmath51 by @xmath52 . if an effective divisor @xmath26 does not contain @xmath51 , then @xmath53 and @xmath54 hence , the smallest slope of curves in this family bounds the slope of effective divisors , since a divisor can not contain all the deformations of a moving curve ! along these lines , in @xcite , harris and morrison study genus @xmath0 curves as degree @xmath1 covers of @xmath55 for large @xmath1 .
if we vary the branch points in @xmath55 , we just obtain moving curves in @xmath3 .
the slopes of such moving curves do provide good lower bounds when @xmath0 is small .
however , as @xmath56 , these bounds tend to 0 .
therefore , the existence of a uniform positive lower bound for the slope for all @xmath0 remains unknown .
+ the same circle of ideas is mentioned in @xcite . instead of using covers of @xmath55 , coskun , harris and starr
choose canonical curves in @xmath57 to study . in order to get moving curves in @xmath3
, they impose on canonical curves some conditions like contacting fixed linear subspaces with suitable orders .
this method works pretty well for genus up to 6 .
but the general case is hard to analyze since we do not have good knowledge about the geometry of canonical curves and the gromov - witten theory for high genus curves is far from complete .
+ here we push the idea in @xcite further to consider covers of elliptic curves . take a cover @xmath58 that only has one branch point at the marked point @xmath59 of the elliptic curve @xmath11 .
if we vary the complex structure of @xmath11 , @xmath51 also varies and its locus in @xmath3 forms a 1-dimensional subscheme @xmath2 .
the geometry of @xmath2 is our main interest .
we need to do three things as follows .
+ firstly , we want to calculate the slope of @xmath2 . here
we use a very classical approach originally due to hurwitz .
he used symmetric groups to enumerate covers of @xmath60 similarly , the information of a degree @xmath1 branched cover of an elliptic curve can be encoded in a certain equivalence class of solution pairs in @xmath61 .
we fix the ramification type of the cover by a conjugacy class @xmath62 in @xmath61 , i.e. , @xmath63 . let @xmath64 also denote a conjugacy class of @xmath61 whose number of length @xmath20 cycles equals @xmath65 , @xmath66 . for two pairs @xmath67 , we define an equivalence relation by @xmath68 if there exists an element @xmath69 such that @xmath70 and @xmath71
. moreover , let @xmath72 denote the subgroup of @xmath61 generated by @xmath73 and @xmath74 .
define a set @xmath75 @xmath76 and its subset @xmath77 based on the conjugacy type of @xmath74 .
then the cover @xmath78 corresponds to one equivalence class of solution pairs in @xmath79 .
readers should bear in mind that for this cover , @xmath80 correspond to the monodromy images of a standard symplectic basis of @xmath81 . pick a loop @xmath82 around the branched point @xmath83 .
the monodromy image of @xmath84 has to be an element in the ramification class @xmath85 .
the transitivity condition imposed on @xmath72 is because we want the cover to be connected .
see the following picture .
+ let @xmath86 denote the order of the set @xmath87 modulo the equivalence relation , i.e. , @xmath88 when there is no ambiguity , we simply use @xmath89 to denote @xmath90 .
let us further define a total sum @xmath91 and a weighted sum @xmath92 it is clear that @xmath93 and @xmath94 only depend on @xmath95 and @xmath85 .
now we can state the main result about the slope of @xmath2 .
[ slope ] @xmath96 note that @xmath97 only depends on the ratio @xmath98 .
secondly , we have to study the irreducible components of @xmath2 . by using the symmetric group @xmath61 again
, we can see that two covers lie in the same component of @xmath2 if and only if the solution pair corresponding to one can be sent to the solution pair corresponding to the other via a monodromy action .
more precisely , define two actions @xmath99 and @xmath100 acting on the set @xmath89 .
it is not hard to check that both actions are well defined for @xmath101 .
let @xmath102 denote the group of actions generated by @xmath103 and call @xmath102 the monodromy group of @xmath2 .
we give the following criterion .
[ monodromy ] take two covers represented by two solution pairs @xmath104 and @xmath105 in @xmath101
. then the points in @xmath2 corresponding to these two covers are contained in the same component of @xmath2 if and only if there exists an element in the monodromy group @xmath102 that sends @xmath104 to @xmath105 .
finally , we have to verify whether or not @xmath2 is a moving curve in @xmath3 .
readers may already be aware of the fact that @xmath2 is more likely rigid , since a cover corresponding to a point in @xmath2 only has one branched point , whose variation along the target elliptic curve does not change the moduli of this elliptic curve . nevertheless , technically we do not need moving curves . instead , when the degree @xmath1 of the covers varies , as long as the union of all @xmath106 is zariski dense in @xmath3 , an effective divisor can not contain all of them .
so the smallest slope @xmath97 in this union still provides a lower bound for slopes of effective divisors .
luckily , for ramification class @xmath85 of suitable types , we have an affirmative answer to this density problem . [ density ]
when @xmath0 and @xmath62 are fixed but @xmath1 varies , the image of the union of @xmath106 is dense in @xmath3 if and only if the number of ramification points is less than @xmath0 , i.e. , @xmath107 .
we apply the above results to the case @xmath108 . about the slope
, we have the following conclusion . [ g2s ]
when @xmath109 , @xmath85 can be either of type @xmath110 or of type @xmath111 .
in both cases @xmath2 has constant slope @xmath112 independent of @xmath1 . note that 10 is the sharp lower bound for slopes of effective divisors on @xmath113 , cf .
remark [ mc ]
. we will give a direct proof of theorem [ g2s ] in section 4 but only for prime number @xmath1 .
the general case follows from an indirect argument in remark [ mc ] .
+ we can further calculate the genus of @xmath2 .
the following results are only for prime @xmath1 .
[ g21 m ] when @xmath109 , @xmath1 is prime and @xmath85 is of type @xmath110 , @xmath114 where @xmath115 denotes the greatest common divisor of @xmath116 and @xmath117 .
in particular , asymptotically @xmath118 .
[ g22 m ] when @xmath109 , @xmath1 is prime and @xmath85 is of type @xmath111 , @xmath119 @xmath120 in particular , asymptotically @xmath121 .
theorems [ g21 m ] and [ g22 m ] will be proved in section 5 by considering local monodromy actions .
+ when @xmath0 is 3 , the combinatorics becomes harder .
however , we check one case by computers : when @xmath85 is of type @xmath122 and @xmath1 is prime , it looks like that @xmath97 tends to 9 , cf .
conjecture [ g3s ] .
note that 9 is the sharp lower bound for slopes of effective divisors on @xmath123 , cf .
+ unfortunately , although in principle by theorems [ slope ] and [ monodromy ] we know almost everything about @xmath2 , in practice it seems that the combinatorial problems we encounter are quite complicated , especially when @xmath0 is large .
it would be very interesting to obtain more numerical evidences to see if this method does provide a lower bound for slopes of effective divisors .
+ this paper is organized in the following way . in section 2
we study the geometry of @xmath2 , namely we consider slope , monodromy and density .
then we apply the results to the genus 2 case and several examples of the genus 3 case in section 3 . in section 4 ,
a general method is introduced to express @xmath93 and @xmath94 in terms of some generating functions .
then we study the genus and orbifold points of @xmath2 in section 5 .
afterwards , a correspondence between our viewpoint and the viewpoint of square - tiled surfaces is established in section 6 . finally in the appendix , we prove some equalities of certain series appearing in section 5 by a trick of classical number theory . +
* acknowledgements .
* i am grateful to my advisor joe harris who told me about this problem and gave me many valuable suggestions .
i would also like to thank sabin cautis , izzet coskun , noam elkies , thomas lam and curtis t. mcmullen for useful conversations .
in this section , we will focus on the slope , monodromy and density of @xmath2 and prove theorems [ slope ] , [ monodromy ] and [ density ] one by one .
first , let us give a rigorous construction of @xmath2 .
+ let @xmath124 parameterize a general pencil of plane cubics . blow up the 9 base points to obtain a smooth surface @xmath125 that is an elliptic fibration over @xmath126 .
there are 12 rational nodal curves as special fibers over @xmath127 .
fix a section @xmath128 corresponding to the blow up of one of the 9 base points . with this section
, @xmath126 can be considered as a 12-sheeted cover of the moduli space of elliptic curves @xmath129 .
take a general fiber @xmath130 over @xmath131 .
consider all the possible degree @xmath1 covers @xmath132 from a genus @xmath0 connected curves @xmath51 to @xmath11 branched only at @xmath59 .
let @xmath133 , where @xmath134 is a fixed conjugacy class @xmath85 of the symmetric group @xmath61 .
when @xmath135 varies in @xmath136 , the locus of such covers also varies in a 1-dimensional hurwitz scheme @xmath137 .
now , if @xmath135 approaches some point @xmath138 , the cover degenerates to a cover of a rational nodal curve , in the sense of admissible covers .
hence , we can compactify @xmath137 by the space of admissible covers @xmath139 .
when there is no ambiguity , we will simply use @xmath140 and @xmath2 instead . +
the curve @xmath2 is an @xmath93-sheeted cover of @xmath126 , possibly branched at @xmath141 .
@xmath93 is equal to the number of distinct genus @xmath0 degree @xmath1 covers of a general plane cubic only branched at one point of ramification type @xmath85 .
note that this number @xmath93 equals the total sum @xmath142 defined in the introduction section .
there is a natural map from the space of admissible covers @xmath2 to @xmath3 .
so it makes sense to talk about the intersections @xmath143 and @xmath144 by pulling back the corresponding classes to @xmath2 .
now we can prove theorem [ slope ] .
we have to figure out the two numbers @xmath145 and @xmath144 .
the former can be worked out by the following argument .
we want to establish a diagram of maps : @xmath146^\varphi \ar[d ] & s \ar[d ] \\
y \ar[r]^\psi & x } \ ] ] such that over a point @xmath147 mapping to @xmath148 , the fiber over @xmath149 is the corresponding cover of the fiber over @xmath150 , i.e. , t is the universal covering curve over @xmath2 .
see the picture below .
when @xmath150 is in @xmath151 , locally we can always construct @xmath152 .
now in a small neighborhood of @xmath138 , we can identify all the smooth fibers to a fixed elliptic curve @xmath130 .
let @xmath80 be a standard symplectic basis of @xmath153 such that @xmath74 is the vanishing cycle when the smooth elliptic curve degenerates to the rational nodal curve over @xmath138 .
we abuse notation and denote their monodromy images in @xmath61 also by @xmath80 , and assume @xmath154 . now the key observation is that when the elliptic curve degenerates , the combinatorial type of the solution pair corresponding to the smooth cover determines the type of the degenerate cover .
more precisely , the degenerate covering curve will have @xmath65 nodes at which the map @xmath155 is given locally by @xmath156 , where @xmath157 and @xmath158 are the local charts of @xmath152 and @xmath125 respectively .
each such node contributes @xmath159 to the intersection @xmath143 .
therefore , we have @xmath160 the constant 12 comes from the number of singular fibers of @xmath125 .
+ to calculate @xmath144 , by mumford s formula @xmath161 , it suffices to work out the intersection @xmath162 .
the ramification class @xmath85 is of type @xmath163 so we have @xmath116 sections @xmath164 of @xmath152 such that @xmath165 . by riemann - hurwitz ,
the relative dualizing sheaves @xmath166 and @xmath167 satisfy the relation @xmath168 moreover , @xmath169 and @xmath170 so we get @xmath171 and @xmath172 now it is routine to check that @xmath173 so we get the desired formula for @xmath97 . [ bc ] to complete the commutative diagram in the above proof , when the map is given locally by @xmath174 where @xmath157 and @xmath158 are the local charts of @xmath125 and @xmath152 respectively , a base change of degree divisible by @xmath20 is necessary .
so we can make a degree @xmath175 base change once for all to realize such maps globally .
after the base change , @xmath176 and @xmath144 both have to be multiplied by @xmath175 , so the quotient @xmath97 remains the same .
+ a fancier interpretation is that by using a minimal base change , we pass from the space of admissible covers @xmath2 to the stack of admissible covers @xmath177 where a universal covering map lives .
this viewpoint is crucial when we study the local geometry of @xmath2 in section 5 .
[ in ] recall that in our original definition of a cover and its automorphism , we did not allow variation of the target elliptic curve .
so an automorphism of a cover @xmath178 is given by the following commutative diagram , @xmath179^\phi \ar[dr]_\pi & & c \ar[dl]^\pi \\ & e & } \ ] ] where @xmath180 is an automorphism of @xmath51 .
+ however , the elliptic curve @xmath11 always has an automorphism induced by its involution @xmath181 that sends @xmath104 to @xmath182 . if we further identify the two solution pairs @xmath183 , that is , we allow the following commutative diagram to be considered as an automorphism of the cover @xmath78 , @xmath184^{\iota\circ\pi } \ar[d]_\pi & \\ e \ar[r]_\iota & e } \ ] ] we would get a new space @xmath185 .
@xmath2 can be viewed as a double cover of @xmath185 .
more precisely , for a component @xmath186 of @xmath2 , if a general cover in @xmath186 corresponding to a solution pair @xmath104 does not have the automorphism induced by @xmath181 , i.e. , there does not exist an element @xmath187 such that @xmath188 , then @xmath186 is a double cover of the corresponding component @xmath189 of @xmath185 . on the other hand , if there is some @xmath190 such that @xmath188 , then every cover in @xmath186 has the automorphism induced by @xmath181 .
hence , @xmath186 is a double curve and its reduced structure is the same as @xmath189 .
note for the purpose of slope calculation that @xmath191 and @xmath192 are always the same
. the curve @xmath2 may be reducible . actually if two pairs @xmath104 and @xmath105 generate non - conjugate subgroups of @xmath61 , then the corresponding two covers must be contained in different components of @xmath2 . + now let us consider the monodromy of the map @xmath193 .
more precisely , we want to study the @xmath194-monodromy map @xmath195 , where @xmath135 is a fixed base point of @xmath151 , @xmath196 is the fiber over @xmath135 with one marked point and @xmath197 is the orientable outer automorphism group .
the map @xmath198 is surjective and its image @xmath199 .
moreover , the group @xmath102 of @xmath194-monodromy acting on @xmath101 can be generated by two operations : @xmath200 and @xmath201 .
we give an indirect proof using the fact that the @xmath202-monodromy map @xmath203 of a general pencil of plane cubics is surjective .
+ for @xmath202-monodromy , the marked point does not affect homology .
so we consider the map @xmath204 . in order to show that @xmath203 is surjective for a general pencil @xmath126
, it suffices to exhibit a special pencil for which the claim holds .
actually some examples are explicitly worked out in @xcite and the @xmath202-monodromy maps are surjective as we expect .
+ now for @xmath194-monodromy , we can construct the following commutative diagram : @xmath205_\simeq \ar@/^5pc/[ddd]_\simeq \\ \pi_{1}(x^{0},b)\ar[r]^{\rho_{\pi}}\ar[ur]\ar[d]_\simeq & out^{+}(\mathbb z*\mathbb z)\ar[d]_{\phi } \\
\pi_{1}(x^{0},b)\ar[r]^{\rho_{h}}\ar[dr ] & aut^{+}(\mathbb z\oplus\mathbb z ) \\ & \gamma_{1}\ar[u]^\simeq } \ ] ] here @xmath206 and @xmath207 are the mapping class groups for an ordinary torus and a torus with one marked point respectively .
it is well known that @xmath206 and @xmath208 are both isomorphic to @xmath209 .
the map @xmath180 is induced by quotienting out commutators , i.e. , @xmath210)\simeq \mathbb z\oplus\mathbb z$ ] , where @xmath73 and @xmath74 are the generators of the free group @xmath211 .
moreover , we have the isomorphism @xmath212 , cf .
so the conclusion that @xmath198 is surjective follows from the fact that @xmath203 is surjective .
+ finally , @xmath208 is generated by the two basic actions induced from dehn twists along the two loops represented by @xmath73 and @xmath74 .
so correspondingly the monodromy actions send @xmath213 to @xmath214 and @xmath215 respectively .
now theorem [ monodromy ] follows easily from the above lemma . since @xmath216
, the monodromy actions are well defined up to the equivalence relation .
the two actions @xmath217 and @xmath135 correspond to @xmath218 and @xmath219 respectively , which generate @xmath209 .
[ lm ] assume that @xmath74 corresponds to the local vanishing cycle around a nodal fiber over @xmath220 . by the picard - lefschetz formula or kodaira s classification of elliptic fibrations , going along a small loop around @xmath138 once will send a solution pair @xmath213 to @xmath215 . see the following picture .
we call this action as the local monodromy action , in order to distinguish it from the global monodromy .
it is useful when we study the genus of @xmath2 by applying the riemann - hurwitz formula to the map @xmath193 , cf .
section 5 . in this section , whenever we mention density , we mean zariski density .
notation is the same as before .
let @xmath85 be the conjugacy class @xmath134 of @xmath61 .
we assume that @xmath221 are greater than 1 and that the rest of the @xmath222 s are all equal to 1 . by riemann - hurwitz , @xmath223 . now
the pullback of a holomorphic 1-form from the target elliptic curve becomes a holomorphic 1-form @xmath224 on the covering curve @xmath51 and @xmath225 , where @xmath226 and @xmath227 .
so we can fix the ramification type by fixing @xmath221 but adding more length-1 cycles to @xmath85 .
hence , the degree @xmath1 of the map can vary from @xmath228 to infinity .
+ now we consider the union of all possible covers in the above sense .
an equivalent statement of theorem [ density ] is the following .
[ d ] the image of @xmath229 is dense in @xmath48 if and only if the inequality @xmath230 holds , i.e. , @xmath231 .
let @xmath232 be a partition of @xmath233 .
we consider the moduli space @xmath234 parameterizing @xmath235 where @xmath236\in \mathcal m_{g}$ ] , @xmath237 , and @xmath238 .
the dimension of @xmath239 is @xmath240 the space @xmath239 can locally be equipped with a coordinate system .
pick a basis @xmath241 the relative homology of @xmath51 with @xmath242 marked points , such that @xmath243 are the standard symplectic basis of @xmath244 and @xmath245 is a path connecting @xmath246 and @xmath247 , @xmath248 .
the period map @xmath249 is given by @xmath250 which provides a local coordinate system for @xmath251 , cf .
now pick a torus @xmath152 given by @xmath252 , where @xmath253 is a lattice . consider the coordinate @xmath254 .
we cite the following lemma from @xcite .
@xmath255 if and only if the following conditions hold : + ( 1 ) there exists a holomorphic map @xmath256 ; + ( 2 ) @xmath257 ; + ( 3 ) @xmath258 is ramified at @xmath259 with ramification order @xmath260 , @xmath261 ; + ( 4 ) @xmath262 mod @xmath263 , @xmath264 .
by virtue of this lemma , if we want to get a cover @xmath265 branched at one point with ramification type @xmath266 , it is equivalent to @xmath267 mod @xmath263 , @xmath268 .
now we vary @xmath269 and such lattice points @xmath180 are dense in an open domain of @xmath270 .
let @xmath271 be the union of such coverings @xmath235 , so @xmath271 is dense in @xmath234 .
now we have the following diagram : @xmath272 \ar[drr ] & \mathcal h(\mu)\ar@{^{(}->}[r ] & \mathcal h_{g } \ar[d]^{\pi } \\ & & \mathcal m_{g } } \ ] ] where @xmath273 is the hodge bundle over @xmath48 .
as long as @xmath234 dominates @xmath48 , i.e. , for a general @xmath236\in \mathcal m_{g}$ ] there exists @xmath274 such that @xmath275 , then i m @xmath276 must be dense in @xmath48 .
now we have to check that when @xmath230 , i.e. , @xmath231 , @xmath251 does dominate @xmath48 .
+ the case @xmath277 can be reduced to @xmath278 since there is a natural stratification among all the moduli spaces @xmath251 .
when @xmath278 , we apply the de jonquires formula from ( * ? ? ?
* viii 5 ) .
suppose that @xmath279 are distinct integers and @xmath65 appears @xmath280 times in the partition @xmath281 of @xmath233 , @xmath282 and @xmath283 further define @xmath284 and @xmath285 .
then on a genus @xmath0 curve , the virtual number of canonical divisors having @xmath280 points of multiplicity @xmath65 is @xmath286_{t_{1}^{n_{1}}\cdots t_{m}^{n_{m}}}}.\ ] ] as long as this number is nonzero , @xmath234 dominates @xmath48 .
let @xmath287 and @xmath288 .
then @xmath289_{t_{1}^{n_{1}}\cdots t_{m}^{n_{m } } } \\ & = & \left[\left(1+{g \choose 1}a+\cdots+{g\choose g-1}a^{g-1}\right)\left(1-b+b^{2}-\cdots\right)\right]_{t_{1}^{n_{1}}\cdots t_{m}^{n_{m } } } \\ & = & \left[{g\choose g-1}a^{g-1 } - { g\choose g-2}a^{g-2}b + \cdots + ( -b)^{g-1}\right]_{t_{1}^{n_{1}}\cdots t_{m}^{n_{m } } } \\ & = & \left[\frac{a^{g}-(a - b)^{g}}{b}\right]_{t_{1}^{n_{1}}\cdots t_{m}^{n_{m } } } \\ & = & \left[a^{g-1}+a^{g-2}(a - b ) + \cdots + ( a - b)^{g-1}\right]_{t_{1}^{n_{1}}\cdots t_{m}^{n_{m } } } > 0,\end{aligned}\ ] ] since @xmath290 has nonnegative coefficients and @xmath10 has positive coefficients .
the most general case is @xmath291 ; then we want to analyze @xmath292 .
of course in this case the image of @xmath293 is dense in @xmath48 .
the most special case is @xmath294 ; then correspondingly @xmath295 . obviously for dimension reasons ,
the image of @xmath296 has to be contained in a proper subvariety of @xmath48 .
since the slope formula and the monodromy of @xmath2 are relatively simple to analyze when @xmath1 is prime , from now on we will mainly focus on @xmath1 prime for concrete examples .
therefore , the following result is also useful .
the image of @xmath297 is dense in @xmath48 if and only if the inequality @xmath230 holds , i.e. , @xmath231 .
we illustrate the idea of the proof .
use the notation in the proof of theorem [ d ] and write the coordinate @xmath298 .
the degree @xmath1 of the map from @xmath51 to a standard torus @xmath152 equals the area of @xmath51 , which is given by @xmath299 now consider all the integer valued vectors @xmath300 such that @xmath301 is some prime number @xmath1 .
the above claim is equivalent to the density in @xmath302 of their union .
note for a fixed prime @xmath1 that such integer points are always dense in the corresponding hypersurface .
since there are infinitely many such hypersurfaces , the result immediately follows .
in this section , we introduce one ad hoc method to calculate @xmath303 for small @xmath0 . for simplicity
, we only deal with the case when @xmath1 is prime .
it will become clear that the method can also be used for general @xmath1 but with more subtle analysis .
+ first , let us reduce the problem a little bit by getting rid of the equivalence relation . as introduced before , @xmath304 .
@xmath61 acts on @xmath305 by conjugation .
burnside s lemma tells us that @xmath306 where @xmath307 .
[ tau1 ] if @xmath269 commutes with all the elements in a transitive subgroup @xmath308 of @xmath61 , then @xmath269 must be of type @xmath309 in @xmath61 , lm = d .
it suffices to show that for any @xmath310 , if @xmath311 fixes an element in @xmath312 , then @xmath313 . +
suppose @xmath314 .
for any @xmath315 , there exists @xmath316 , such that @xmath317 . but
@xmath269 also commutes with @xmath318 .
by @xmath319 , we know that @xmath320 , so @xmath321 .
when @xmath1 is prime , the above @xmath269 must be either @xmath322 or the long cycle class @xmath323 .
pick a long cycle @xmath324 , i.e. , @xmath20 is sent to @xmath325 . by burnside s lemma
, we have @xmath326 [ tau2 ] @xmath327 . we know that @xmath328 , so @xmath329 , where @xmath330 is an integer independent of @xmath20 .
similarly @xmath331 , and @xmath332 is independent of @xmath333 .
now it is easy to check that @xmath334 , so there is no solution pair in @xmath335 .
hence , finally we obtain the relation : @xmath336 we want to find solution pairs @xmath338 such that @xmath339 and @xmath72 is transitive .
let @xmath340 satisfying the equality @xmath341 .
+ now the key point is , @xmath342 and @xmath74 are in the same conjugacy class , so @xmath343 .
note that @xmath344 can not be contained in three different cycles of @xmath74 , since @xmath345 changes the conjugacy type of @xmath74 .
so only two cases are possible : + ( 1 ) @xmath346 ; + ( 2 ) @xmath347 .
+ using also the condition that @xmath72 is transitive , in case ( 1 ) @xmath74 must be of type @xmath309 , so @xmath74 can only be the long cycle @xmath323 .
pick an element @xmath348 .
there are @xmath349 choices for the cycle @xmath350 ; fix one . now for @xmath73 , we know that @xmath351 , so there are @xmath1 choices for @xmath73 .
overall , we get @xmath352 for case ( 2 ) , we have @xmath353 , so @xmath74 is of type @xmath354 , @xmath355 .
there are @xmath356 choices for @xmath74 .
fix one choice ; then there are @xmath357 choices for the cycle @xmath350 .
fix one of these also ; then because of the transitivity requirement , there are @xmath358 choices for @xmath73 .
multiplying all the numbers together and dividing by @xmath175 , we get @xmath359 therefore , we have @xmath360 and @xmath361 [ g21s
] when @xmath1 is prime and @xmath85 is of type @xmath110 , the slope @xmath362 . by the slope formula , in this case @xmath363 actually when @xmath1 is prime , we have @xmath364 and @xmath365 we will prove these formulae in the appendix by using some tricks from number theory .
we analyze this case similarly .
the condition on @xmath80 is @xmath367 .
now there are 4 possible cases : + ( 1 ) @xmath368 ; + ( 2 ) @xmath369 ; + ( 3 ) @xmath370 ; + ( 4 ) @xmath371 . + for case ( 1 ) , @xmath74 is of type @xmath309 . since @xmath1 is prime , @xmath74 must be the long cycle @xmath323 , so there are @xmath372 choices for @xmath74 .
pick one ; then there are @xmath373 choices for the cycle @xmath374 .
fix one of these also ; then there are @xmath1 choices for @xmath73 .
overall , we get @xmath375 for case ( 2 ) , @xmath74 is of type @xmath376 , where @xmath377 .
+ ( i)@xmath378 .
there are @xmath379 choices for @xmath74 .
fix one of these ; we then have @xmath380 choices for the cycle @xmath374 .
fix one of these too . to keep the transitivity of @xmath72 ,
there are @xmath381 choices for @xmath73 . hence ,
in this case we get @xmath382 ( ii)@xmath383 . since @xmath1 is prime and @xmath384 , we get @xmath385 and @xmath386 .
so @xmath74 is of type @xmath387 and there are @xmath388 choices .
fix @xmath74 ; @xmath389 , so there are @xmath390 choices for the cycle @xmath374 .
fixing one of them , there are @xmath391 choices for @xmath73 .
overall , we get @xmath392 for case ( 3 ) , @xmath393 @xmath394 and either @xmath395 or @xmath396 .
@xmath74 is of type @xmath354 : there are @xmath397 choices .
pick one @xmath74 and assume @xmath398 . since @xmath1 is prime , @xmath399 .
the condition on @xmath59 is that @xmath400 so there are @xmath401 choices for @xmath59 . fix @xmath59 ; then there are @xmath402 choices for the cycle @xmath374 .
fixing one of these also , finally there are @xmath403 choices for @xmath73 . in total , we get @xmath404 + for case ( 4 ) , @xmath405 where @xmath406 @xmath74 has to be of type @xmath354 as in case ( 3 ) . so there are also @xmath356 choices for @xmath74 .
fix one @xmath74 ; then there are @xmath407 choices for the cycle @xmath374 .
pick one ; then there are @xmath358 choices for @xmath73 .
overall , we get @xmath408 + now combining cases ( 3 ) and ( 4 ) , finally we have @xmath409 therefore , we have @xmath410 @xmath411 and @xmath412 @xmath413 [ g22s ] when @xmath1 is prime and @xmath85 is of type @xmath111 , the slope @xmath362 . by the slope formula , we know that @xmath414 actually when @xmath1 is prime , we have @xmath415 and @xmath416 we also postpone the discussion of these formulae to the appendix . combining theorems [ g21s ] and [ g22s ] , we have finished the proof of theorem [ g2s ] for the case @xmath1 prime .
the general case follows from the remark below .
[ mc ] @xmath113 is special in that the following equality holds : @xmath417 hence , for a curve @xmath418 in @xmath113 not entirely contained in the boundary , we always have that its slope @xmath419 . moreover , @xmath420 if and only if @xmath421 .
+ there is an easy but indirect way to show that in general @xmath422 , which was pointed out to the author by curtis t. mcmullen .
the idea is that for a cover @xmath58 , when the vanishing cycle @xmath74 shrinks to a node , any component @xmath84 of @xmath423 is not vanishing in @xmath424 since @xmath74 is not a zero cycle
. then when @xmath84 shrinks , it will form an internal node , i.e. , the degenerate covering curve only lies in @xmath425 , but not in @xmath426 . as in the following picture ,
loop ( 1 ) may belong to @xmath423 , but loop ( 2 ) can not . [ torus ] as @xmath108 , @xmath427 may be reducible , cf .
section 5 , @xcite and @xcite .
however , by the above argument , the slope of each component of @xmath2 is always 10 .
another way to produce a one parameter family of degree @xmath1 covers of elliptic curves is by fixing the @xmath333-invariant of the target elliptic curve and moving one branched point .
for instance , consider degree @xmath1 genus 2 covers of a fixed elliptic curve @xmath11 simply branched at the marked point @xmath428 and another point @xmath429 .
let @xmath429 vary , and then we have a 1-dimensional space @xmath430 of admissible covers which maps to @xmath113 .
when @xmath429 meets @xmath428 , after blowing up we will get some nodal curves as admissible covers of @xmath11 with a 2-marked stable rational tail . using the method in @xcite ,
it is easy to write down the intersection number @xmath431 by the monodromy data and verify that it is not vanishing for @xmath432 . by the same argument in remark [ mc ] , the slope of @xmath430 is strictly smaller than 10 .
therefore , our original one parameter family @xmath2 provides a better lower bound for slopes than the family @xmath430 does , at least for the case @xmath108 .
we still assume that @xmath1 is prime .
since the analysis is almost the same as in the previous examples , we skip the discussion and state the result directly : + @xmath434 @xmath435 @xmath436 now the slope formula says that @xmath437 we calculated by computer for small prime numbers @xmath1 and it seems that @xmath97 in this case decreases to 9 .
this evidence leads to the following conjecture .
[ g3s ] when @xmath85 is of type @xmath122 , we have @xmath438 note that for @xmath123 , it is already known that the hyperelliptic divisor @xmath439 has the smallest slope 9 among all effective divisors .
so if the above conjecture is true , then these curves @xmath440 do provide the sharp lower bound for slopes of effective divisors on @xmath123 .
next , we study in detail the beginning case @xmath441 .
the result is the following .
[ g3d5 ] for @xmath442 and @xmath85 of type @xmath443 , @xmath444 has four irreducible components .
two of them have slope @xmath445 and the other two have slope @xmath446 . the proof is nothing but to enumerate all the possible solution pairs @xmath213 modulo equivalence relation , and further classify the orbits by the monodromy criterion , cf . theorem [ monodromy ] .
+ in total there are 40 non - equivalent solutions in the following list : + ( 1 ) @xmath447 ( 2 ) @xmath448 + ( 3 ) @xmath449 ( 4 ) @xmath450 + ( 5 ) @xmath451 ( 6 ) @xmath452 + ( 7 ) @xmath453 ( 8) @xmath454 + ( 9 ) @xmath455 ( 10 ) @xmath456 + ( 11 ) @xmath457 ( 12 ) @xmath458 + ( 13 ) @xmath459 ( 14 ) @xmath460 + ( 15 ) @xmath461 ( 16 ) @xmath462 + ( 17 ) @xmath463 ( 18 ) @xmath464 + ( 19 ) @xmath465 ( 20 ) @xmath466 + ( 21 ) @xmath467 ( 22 ) @xmath468 + ( 23 ) @xmath469 ( 24 ) @xmath470 + ( 25 ) @xmath471 ( 26 ) @xmath472 + ( 27 ) @xmath473 ( 28 ) @xmath474 + ( 29 ) @xmath475 ( 30 ) @xmath476 + ( 31 ) @xmath477 ( 32 ) @xmath478 + ( 33 ) @xmath479 ( 34 ) @xmath480 + ( 35 ) @xmath481 ( 36 ) @xmath482 + ( 37 ) @xmath483 ( 38 ) @xmath484 + ( 39 ) @xmath485 ( 40 ) @xmath486 + now by monodromy action , it is routine to check that : + ( 2),(10),(13 ) belong to the first component @xmath487 ; + ( 1),(3),(4),(5),(6),(7),(8),(9),(11),(12 ) belong to the second component @xmath488 ; + ( 14),(15),(16),(18),(22),(23),(24),(25),(26),(27),(38),(40 ) belong to the third component @xmath489 ; + ( 17),(19),(20),(21),(28),(29),(30),(31),(32 ) , ( 33),(34),(35),(36),(37),(39 ) belong to the last component @xmath490 .
+ note that the slope formula can be readily used not only for the entire curve @xmath2 but also for its irreducible components .
plugging in the data listed above , we get @xmath491 and @xmath492 for a cover @xmath493 in @xmath494 , @xmath51 necessarily has a holomorphic 1-form with a zero of order 4 .
let @xmath495\in \mathcal m_{3}$ : $ k_{c}$ has a vanishing sequence $ \geq ( 0,1,4)$ at some point $ p\in c$}\}.\ ] ] then we know that i m @xmath496 .
we define another divisor @xmath497 to be the locus of smooth plane quartics that have hyperflexes .
the slope of @xmath498 is @xmath499 , which was worked out by cukierman , cf .
we also know that @xmath500 and @xmath501
. it would be interesting to get some information about the intersection of @xmath439 and @xmath498 at the boundary of @xmath123 . + first , define codimension 2 loci @xmath502 and @xmath503 curves in @xmath10 and @xmath418 are always double covers of rational curves in the sense of admissible covers .
it is not hard to see that @xmath504 + for the divisor @xmath498 , we have the following claim .
@xmath505 first , let us verify that @xmath506 .
take a plane conic @xmath507 and a one parameter family @xmath508 of general plane quartics , such that @xmath509 has a hyperflex line that is also tangent to @xmath507 at the hyperflex point . by stable reduction , as in @xcite , we know that @xmath510 is a general hyperelliptic curve for @xmath511 . then taking the limit @xmath512 amounts to squeezing two weierstrass points together .
so we get a general element in @xmath10 .
+ for @xmath418 , take a banana curve @xmath513 .
consider a sub linear series of @xmath514 that contracts @xmath515 and maps @xmath516 to a tacnodal plane quartic .
we still have one dimension of freedom to impose a hyperflex line .
+ now we want to show that the inclusion in the above claim is proper .
consider the case @xmath517 in claim [ g3d5 ] . when the vanishing cycle @xmath74 is of type @xmath443 , i.e. , cases ( 1 ) , ( 2 ) , ( 3 ) , ( 4 ) , ( 5 ) , ( 6 ) , ( 14 ) and ( 15 ) , the degenerate cover is a 5-sheeted admissible covering map from a 1-nodal geometric genus 2 curve to a rational nodal curve totally ramified at the node and another smooth point .
note that such a cover can be induced from a degree 5 covering map from a smooth genus 2 curve to @xmath55 totally ramified at 3 points @xmath518 and @xmath519 .
these points can not be conjugate to each other simultaneously .
assume that @xmath518 are not conjugate .
identify them and identify their images also . we get a covering curve @xmath236\in$ ] i m @xmath520 .
but by the construction , we know that @xmath236\notin \overline h \cap \delta_{0}=\overline a\cup \overline b$ ] . therefore , we get the desired conclusion
. it would be interesting to do the stable reduction directly for the family @xmath521 .
in this section , we give a method to systematically calculate @xmath522 .
an analogous exposition is givin in the note @xcite . for simplicity , here we only consider the case when @xmath1 is prime and @xmath85 is of type @xmath523 .
the general situation can be solved similarly without further difficulty . + for a cover @xmath58 , an automorphism @xmath155 of this cover is given by the following diagram : @xmath184_{\pi}\ar[rr]^{\varphi } & & c \ar[ld]^{\pi } \\ & e & } \ ] ] if @xmath78 corresponds to one solution pair @xmath524 , then the automorphism @xmath155 corresponds to an element @xmath190 such that @xmath525 .
hence , we have the correspondence @xmath526 , where @xmath527 is the set of stablizers of the @xmath61 conjugate action .
+ in many cases people are interested in the weighted hurwitz numbers , i.e. , counting a cover @xmath528 with weight @xmath529 .
hence , we define a weighted number @xmath530 where @xmath531 denotes the conjugacy class @xmath532 . in particular
when @xmath1 is prime , the cover has no non - trivial automorphism since @xmath533 is transitive , cf .
lemma [ tau1 ] and [ tau2 ] .
so we can pretend to count weighted connected covers instead .
+ now we want to get rid of the transitivity condition imposed on @xmath72 .
so we further define a set @xmath534 and let @xmath535 and look for pairs @xmath536 such that @xmath537 .
for such a pair , there are @xmath538 choices for @xmath73 that satisfy the equality @xmath539 .
hence , latexmath:[\[\widehat{n}_{\wp}^{d , k } = ( * ? ? ? *
a ) , we know that latexmath:[\[|\{(\gamma,\beta)\in \wp\times\wp : \gamma\beta^{-1}=\tau\}| = \frac{|\wp|^{2}}{|s_{d}|}\cdot\left(\sum_{\chi}\frac{1}{\text{deg}(\chi ) } all irreducible characters of @xmath61 . therefore , we get the following expression @xmath541 from @xmath542 .
take a solution pair @xmath543 .
the subgroup @xmath533 of @xmath61 may not be transitive .
consider the orbits and the action of @xmath544 on them , which correspond to the following data : @xmath545 @xmath546 @xmath547 two data @xmath548 and @xmath549 are of the same type if @xmath550 and @xmath551 .
we get the following equality @xmath552 where the condition on the summation is @xmath553 and @xmath554 simplifying the above expression , we obtain @xmath555 if @xmath531 is of type @xmath556 , a datum @xmath548 corresponds to a vector @xmath557 with integer entries .
put all the vectors into a matrix @xmath558 then the summation runs over all possible @xmath559 satisfying @xmath560 now slightly change the notation .
define an index set @xmath561 for @xmath531 of type @xmath562 , write it as @xmath563 .
so it is determined by an element @xmath564 , and @xmath565 .
+ we define two generating functions as follows : @xmath566 and @xmath567 where @xmath568 and @xmath531 is of type @xmath563 for each term . + therefore , finally we obtain the relation between the generating functions for the number of connected and possibly disconnected covers : @xmath569 if @xmath242 is odd or @xmath570 , then obviously @xmath571 . the evaluation of a character @xmath572 on a conjugacy class @xmath531 can be worked out by standard formulae from the representation theory of @xmath61 .
however , when @xmath1 is large , it seems hard to evaluate the quotient @xmath98 even by computer .
so the estimate of @xmath573 remains mysterious to us .
in this section , we study the local monodromy action of the map @xmath193 mentioned in remark [ lm ] , and via it we obtain information like the genus , orbifold points and orbifold euler characteristic of @xmath2 .
+ for each component of @xmath2 , we take its reduced scheme structure . an orbifold point of @xmath2 possibly occurs when a smooth cover degenerates to a singular cover , and its orbifold order depends on the information about the degree of a local minimal base change and the order of its extra automorphisms ( not induced from nearby covers ) , cf .
remark [ bc ] .
the following two examples would illustrate the idea .
* example 1 : @xmath574 * + there are three non - equivalent solution pairs @xmath104 : + ( 1 ) @xmath575 ; + ( 2 ) @xmath576 ; + ( 3 ) @xmath577 . + when @xmath74 corresponds to the vanishing cycle , the local monodromy action @xmath578 can switch sheets ( 1 ) and ( 2 ) but keep ( 3 ) unchanged , and the global monodromy can send all three sheets to each other .
so @xmath2 is a degree 3 connected cover of @xmath126 simply branched at @xmath579 . by riemann - hurwitz ,
@xmath580 so @xmath581 + furthermore , over @xmath138 , sheets ( 1 ) and ( 2 ) meet at the ramification point @xmath519 , and the other pre - image point @xmath330 lies in sheet ( 3 ) .
since in case ( 3 ) , @xmath74 is a length 3 cycle , as in remark [ bc ] a degree 3 base change is necessary to complete the universal covering map locally over sheet ( 3 ) .
so @xmath330 is an orbifold point with structure group @xmath582 .
by contrast , locally around @xmath519 , we need a degree 2 base change since @xmath74 contains length 2 cycles . but sheets ( 1 ) and ( 2 ) meet at @xmath519 , so the cover corresponding to @xmath519 does not have an extra order 2 automorphism compared with nearby smooth covers . hence , @xmath519 is not an orbifold point of @xmath2 . finally , by the orbifold euler characteristic formula , we have @xmath583 by using the results about @xmath584 in section 5.2 and 5.3 , we can work out the orbifold euler characteristic of @xmath106 in the same way when @xmath337 or @xmath111 and @xmath1 is prime .
note that the result for general @xmath1 is obtained in @xcite , although only the case @xmath585 is discussed there .
it is also natural to study the degenerate covers directly and recover the information obtained from the local monodromy action .
the cover corresponding to @xmath330 can be induced from a degree 3 covering map from an elliptic curve to @xmath55 totally branched over 3 points @xmath586 by identifying @xmath587 and their pre - images .
the following picture reveals the idea . a ramification point with order @xmath242
is marked by @xmath588 in the picture .
hence , we look at triples @xmath589 in @xmath590 such that @xmath591 and @xmath592 is of type @xmath593 .
the only possible solution is @xmath594 and therefore this cover has an order 3 automorphism , which coincides with our above discussion . + for @xmath519 , we look at a degree 3 covering map @xmath78 from @xmath595 to @xmath55 simply branched at @xmath587 and totally branched at @xmath59 .
assume that @xmath596 .
. then we can recover the map corresponding to @xmath519 . hence , consider triples @xmath589 in @xmath590 such that @xmath591 , @xmath598 and @xmath599 are simple transpositions but
@xmath600 is a @xmath593 cycle .
modulo the @xmath590 conjugation action , there is a unique solution corresponding to @xmath601 if we switch @xmath598 and @xmath599 , we get back the same cover , so it has the automorphism induced from the involution @xmath181 . but it does not have other automorphisms .
this result also coincides with our previous analysis . +
* example 2 : @xmath602 * + we only focus on the case when the vanishing cycle @xmath74 is of type @xmath443 , i.e. , cases ( 1 ) , ( 2 ) , ( 3 ) , ( 4 ) , ( 5 ) , ( 6 ) , ( 14 ) and ( 15 ) in the proof of claim [ g3d5 ] . the local monodromy acts transitively on ( 1),(3),(4),(5),(6 ) , but keeps ( 2),(14),(15 ) fixed . let @xmath519 be the point over @xmath138 where the sheets ( 1),(3),(4),(5),(6 ) meet , and @xmath603 be the other 3 pre - images of @xmath138 contained in ( 2),(14),(15 ) respectively .
since locally around @xmath138 we need a degree 5 base change , @xmath603 are orbifold points with structure group @xmath604 , and @xmath519 is not an orbifold point .
also note that if we consider @xmath185 instead of @xmath2 , then ( 1)-(6 ) are still distinct sheets in @xmath185 and they all have automorphisms induced from the elliptic involution @xmath181 .
however , ( 14 ) has to be identified with ( 15 ) under the new equivalence relation @xmath183 for @xmath185 .
+ again , we can study the degenerate covers directly as the previous example . in this case , the covers corresponding to @xmath605 can be induced by degree 5 covers from genus 2 curves to @xmath595 totally branched over 3 points @xmath586 .
we then identify @xmath587 and their pre - image points to obtain the desired singular covers .
now we need to analyze the solution triples @xmath589 in @xmath606 such that @xmath607 and they are all @xmath443 cycles .
take @xmath608 , then @xmath599 can only be @xmath609 or @xmath610 , modulo the @xmath606 conjugation action . for the case
@xmath611 , after switching @xmath598 and @xmath599 , we get a cover equivalent to the case @xmath612 so these two covers can be exchanged by the involution of the target rational nodal curve and they correspond to @xmath332 and @xmath613 in sheets ( 14 ) and ( 15 ) . for @xmath614 , switching @xmath598 and @xmath599 , we get the same cover .
moreover , this cover does not have automorphisms except the one induced by the involution @xmath181 .
so it corresponds to the non - orbifold point @xmath519 .
finally , if @xmath615 , this cover has the automorphism induced by @xmath181 and another order 5 automorphism , so it corresponds to @xmath330 in sheet ( 2 ) .
+ now we study in general the orbits of local monodromy for the case @xmath109 and @xmath1 prime .
starting from one solution pair @xmath104 where @xmath74 is the vanishing cycle , the local monodromy action can send @xmath213 to @xmath616 , so these two sheets have to meet at the same degenerate cover over @xmath617 we also assume that @xmath618 where @xmath85 is of type @xmath110 or @xmath111 .
we give a proof of theorem [ g21 m ] .
all the numbers @xmath93 and @xmath522 in the proof below are from section 3.1 .
let @xmath619 be a fixed cycle in @xmath61 .
we look for @xmath242 such that there exists an element @xmath190 , @xmath620 note that such @xmath269 must satisfy @xmath621 since @xmath622 if @xmath74 is of type @xmath323 , from @xmath623 , we get @xmath624 for some integer @xmath116 .
so if @xmath625 , then @xmath626 .
as long as @xmath627 is prime , @xmath269 must be @xmath322 and @xmath628 . since @xmath629 ,
we get @xmath630 orbits each of which has cardinality @xmath1 . from the viewpoint of the covering map
@xmath631 , the @xmath1 sheets in one orbit meet at a degenerate cover which is not an orbifold point of @xmath2 , since locally around such a point we need a degree @xmath1 base change to realize the universal covering map in the proof of theorem [ slope ] .
+ if @xmath74 is of type @xmath354 , @xmath632 we know @xmath633 . without loss of generality , assume that @xmath634 @xmath635 and that @xmath636 from the condition @xmath637 and @xmath1 prime , we can verify that @xmath269 fixes all the elements in the cycles @xmath638 and @xmath639 .
then we have @xmath640 @xmath641 so we can assume that @xmath73 sends the cycle @xmath642 to @xmath643 and the cycle @xmath644 to @xmath645 .
furthermore , assume that @xmath646 but @xmath647 and @xmath648 , i.e. , these actions are twisted by twist parameters @xmath649 and @xmath650 , whose geometric meaning can be more clearly seen in @xcite and the next section .
now , @xmath73 contains a cycle @xmath651 and the corresponding cycle in @xmath652 is @xmath653 since @xmath654 and @xmath655 are all fixed by @xmath269 , we get @xmath656 .
+ similarly we have @xmath657 .
one can check that these two conditions on @xmath242 are also sufficient for the existence of @xmath269 . since @xmath658 , for @xmath74 of such type we just get @xmath659 orbits and each orbit has cardinality @xmath660 + consider the map @xmath193 . by the riemann - hurwitz formula , we have @xmath661 @xmath662 after simplifying , we get the desired expression in the theorem . + furthermore
, we already know that @xmath663 and @xmath664 . in the appendix we will see that @xmath665 has lower order than @xmath666 , so we also obtain the asymptotic result for @xmath584 . in the next section , we will see that the above @xmath2 has two irreducible components @xmath487 and @xmath488 that do not intersect , cf . theorem [ curt ] .
so actually we have @xmath667 note that a similar genus formula for the case @xmath668 and @xmath1 prime is also obtained in @xcite using the technique of square - tiled surfaces .
however , our space of admissible covers @xmath2 differs from the teichmller discs defined in @xcite in that @xmath2 is closed and it is over a pencil of plane cubics rather than @xmath129 .
now we study the other case : @xmath669 , and prove theorem [ g22 m ] .
all the numbers @xmath93 and @xmath670 in the following proof are from section 3.2 .
let @xmath671 be a fixed cycle .
we still look for @xmath242 such that there exists @xmath187 , @xmath672 as before , such @xmath269 must satisfy @xmath673 the discussion is very similar to the one above .
we analyze the solution pairs @xmath104 case by case based on the type of the vanishing cycle @xmath674 + if @xmath74 is of type @xmath323 , assume that @xmath675 . from @xmath623
we get that @xmath269 must be @xmath676 for some integer @xmath116 . but
@xmath677 , since @xmath678 @xmath679 , so in this case we get @xmath680 orbits , each of which contains @xmath1 sheets meeting at one degenerate cover over @xmath138 .
+ if @xmath74 is of type @xmath376 , @xmath681 we can write @xmath74 as @xmath682 @xmath683 @xmath684 and @xmath685 by @xmath686 and @xmath687 co - prime , we know that @xmath269 fixes all the elements in @xmath688 and @xmath689
. then we have @xmath690 @xmath691 @xmath692 so we can assume that @xmath73 sends the cycle @xmath642 to @xmath693 , @xmath694 to @xmath695 , and @xmath696 to @xmath697 @xmath698 but @xmath699 i.e. , these actions are twisted by twist parameters @xmath700 and @xmath701 .
@xmath73 contains a cycle @xmath702 and the corresponding cycle in @xmath652 is @xmath703 since @xmath654 and @xmath704 are all fixed by @xmath269 , we get @xmath656 .
+ similarly , we have @xmath705 .
these conditions are also sufficient .
since @xmath706 , we get @xmath707 orbits in this case , and each orbit contains @xmath708 elements . + if @xmath74 is of type @xmath354 , @xmath709 , we can write @xmath710 @xmath711 if @xmath712 , then @xmath713 @xmath714 we can always assume that @xmath73 sends the cycle @xmath642 to @xmath715 and @xmath644 to @xmath716 .
moreover , @xmath717 and @xmath718 , but @xmath719 and @xmath720 , where @xmath649 and @xmath650 are the twist parameters . + now by @xmath721 , if @xmath722 and @xmath723 , we get @xmath724 , which is impossible .
hence , @xmath269 acts trivially on the elements in the cycles @xmath725 and @xmath726 + the cycle of @xmath73 starting from @xmath727 can be written as @xmath728 and the corresponding cycle of @xmath652 is @xmath729 since @xmath654 and @xmath730 are all fixed by @xmath269 , we get @xmath656 .
+ similarly , we have @xmath657 .
these two conditions on @xmath242 are also sufficient to find a desired @xmath269 . hence in this case
, there are @xmath731 orbits and each orbit contains @xmath732 elements . + there is one more case when @xmath74 is of type @xmath354 , @xmath709 ; namely , @xmath733 and @xmath734 @xmath735 we can also check that @xmath269 acts trivially on @xmath725 and @xmath736 .
+ assume that , similarly to the last case , @xmath73 acts with twist parameters @xmath649 and @xmath650 at the end .
then @xmath73 contains one cycle @xmath737 and the corresponding cycle in @xmath652 is @xmath738 since @xmath654 and @xmath739 are all fixed by @xmath269 , we get @xmath656 .
similarly we have @xmath657 .
so in this case there are @xmath740 orbits and each orbit contains @xmath732 elements .
+ the last case is when @xmath74 is of type @xmath387 .
assume @xmath741 and @xmath742 then we have @xmath743 we can further assume @xmath744 for @xmath745 and @xmath746 since @xmath747 , we only need to check when there exists @xmath269 such that @xmath748 and @xmath749 it is not hard to see that @xmath73 can be of type @xmath750 or @xmath751 if @xmath752 is odd , then @xmath753 can be @xmath754 or @xmath755 respectively .
note that here @xmath753 is not of the same type as @xmath73 , so @xmath269 does not exist .
+ if @xmath752 is even , then @xmath753 can be @xmath756 or @xmath757 here it is easy to check that the desired @xmath269 always exists . + therefore , since @xmath758 , we get @xmath759 orbits and each orbit contains @xmath760 elements . + putting all the results together , we obtain , by the riemann - hurwitz formula , for the map @xmath193 @xmath761 @xmath762 after simplifying , we get exactly the expression in the theorem .
+ the asymptotic behavior of @xmath584 follows from the result about the expression of @xmath93 .
again , the terms involving @xmath763 do not affect the asymptotic order .
in the next section we will see that the above @xmath2 has at least two components , based on the different types of subgroups generated by @xmath544 .
in this section , we establish a correspondence between our method and the work in @xcite from the viewpoint of square - tiled surfaces .
+ the idea is quite simple .
take a standard torus @xmath11 .
if @xmath51 is a cover of @xmath11 , then @xmath51 can be realized as a possibly degenerate lattice polygon with some edges and vertices identified .
it covers @xmath1 unit squares if the degree of the map is @xmath1 .
we will explain the details by some examples for the case @xmath108 .
as before , we only consider the situation when @xmath1 is prime . in this case , if there is a degree @xmath1 covering map @xmath764 only ramified at one point @xmath765 , then @xmath51 can be realized as an octagon of area @xmath1 .
all of its vertices are identified to be the unique ramification point @xmath83 marked with a @xmath766 in the following picture .
take also two loops @xmath73 and @xmath74 of the torus @xmath11 as in the picture .
mark the unit squares covered by the octagon by @xmath767 , and consider the monodromy images in @xmath61 induced from @xmath73 and @xmath74 .
let us look at two examples .
+ * example 1 : * + consider the following octagon
. it should correspond to a degree @xmath768 cover of @xmath11 .
we still abuse notation and use @xmath80 to denote also their monodromy images .
it is easy to see @xmath769 and @xmath770 then we can check that @xmath771 has the desired ramification type . +
* example 2 : * + consider another octagon .
the area is still @xmath768 .
this time we get @xmath772 and @xmath773 .
so @xmath774 has the required ramification type .
+ in @xcite , the square - tiled octagons can be of two types : one - cylinder type and two - cylinder type .
the one - cylinder type corresponds to @xmath775 as in example 1 , and the two - cylinder type corresponds to @xmath776 as in example 2 .
+ moreover , also in @xcite , it is shown that we can mark the 6 weierstrass points of @xmath51 .
1 or 3 out of the 6 points are integer points , which provides two different parities invariant under the monodromy action .
it follows immediately that @xmath777 has at least two components . on the other hand , by our method using @xmath61 , one can check that 1 and 3 integer weierstrass points correspond to @xmath778 and @xmath779 respectively .
+ for instance , in the first example , @xmath780 .
of course the ramification point is one weierstrass point , and we mark the others with a @xmath781 in the following picture . note that only the ramification point is an integer weierstrass point . in the second example , @xmath782 . from the picture below , we can see that there are exactly 3 integer weierstrass points .
in this case we will have 2 ramification points on @xmath51 .
we mark them with a @xmath766 and a @xmath783 respectively . as in @xcite , @xmath51
can be realized as a decagon .
let us look at some examples .
+ * example 3 : * consider a decagon in the following picture and mark the unit squares . this time @xmath784 we get @xmath785 and @xmath786 .
so @xmath787 note that @xmath74 is of type @xmath323 .
in general , this is the one - cylinder type corresponding to @xmath775 in our previous discussion . + * example 4 : * consider another decagon with area 7 .
@xmath788 and @xmath789 so @xmath790 in general , this is the two - cylinder type corresponding to @xmath776 .
+ * example 5 : * for the case @xmath791 , we can consider an example such as the following .
@xmath792 and @xmath793 .
so @xmath794 .
in general , this corresponds to the three - cylinder type .
+ in @xcite , there are two actions @xmath271 and @xmath795 defined by the following pictures .
actually they correspond to our monodromy actions @xmath796 and @xmath797 respectively .
for instance , in the first example , applying the action @xmath271 , the direction of @xmath74 does not change but @xmath73 changes to a direction @xmath798 parallel to the diagonal .
after the action @xmath271 , we get @xmath799 .
hence , the method in @xcite to work out the number of components of @xmath2 by the monodromy actions can be similarly carried out here .
the reader may also be aware of the correspondence between our monodromy actions and the butterfly moves defined in @xcite . actually , the result in @xcite is more general , not only for @xmath1 prime .
we simply cite the result as the following theorem .
[ curt ] the teichmller curve @xmath777 is irreducible for @xmath1 even or @xmath800 , and has exactly two components for @xmath801 odd .
however , to the best of the author s knowledge , for the case @xmath669 the question below is still unknown .
how many irreducible components does the curve @xmath802 have ?
it is relatively easy to get a lower bound for the number of components .
for instance , when @xmath803 is odd , pick a solution pair @xmath804 .
then @xmath805 , and @xmath72 is a subgroup of @xmath779 .
we can take another solution pair @xmath806 .
then @xmath807 , and @xmath778 .
hence , in this case @xmath802 has at least two components .
in this appendix , we will prove the equalities ( 1 ) , ( 2 ) , ( 3 ) and ( 4 ) in section 3 .
first , we introduce the following functions @xmath808 these summations are quasi - modular forms with certain weights .
now define three series @xmath809 @xmath810 and @xmath811 there are some fundamental relations among @xmath812 and @xmath795 the ramanujan differential equations , cf .
@xcite : @xmath813 note that @xmath815_{d},\ ] ] where @xmath816_{d}$ ] means the coefficient of the degree @xmath1 term in the series expansion , and @xmath817 similarly , we have @xmath818 hence , we get @xmath819_{d } = \left [ \big(\frac{1-p}{24}\big)^{2}\right]_{d } = \left [ \frac{1}{24^{2 } } - \frac{1}{24\cdot 12 } + \frac{1}{24^{2}}(q+12q \frac{dp}{dq})\right]_{d}\ ] ] @xmath820 now the equalities ( 1 ) , ( 2 ) for @xmath93 and @xmath94 for @xmath824 of type @xmath110 follow immediately .
similarly one can verify the equalities ( 3 ) and ( 4 ) .
+ now we turn to the asymptotic result for @xmath584 in theorem [ g21 m ] .
it suffices to verify that @xmath825 has asymptotic order less than 3 .
actually in @xcite , the order is proved to be less than @xmath826 for any @xmath827 the asymptotic behavior of @xmath584 in theorem [ g22 m ] can be estimated similarly to show that it is less than 4 . | consider genus @xmath0 curves that admit degree @xmath1 covers to elliptic curves only branched at one point with a fixed ramification type .
the locus of such covers forms a one parameter family @xmath2 that naturally maps into the moduli space of stable genus @xmath0 curves @xmath3 .
we study the geometry of @xmath2 , and produce a combinatorial method by which to investigate its slope , irreducible components , genus and orbifold points . as a by - product of our approach ,
we find some equalities from classical number theory . moreover , a correspondence between our method and the viewpoint of square - tiled surfaces is established .
we also use our results to study the lower bound for slopes of effective divisors on @xmath3 . |
since hosono and coworkers discovered superconductivity in hole - doped lafeaso at @xmath5=26 k,@xcite fe - pnictides have been the cause of much excitement.@xcite the mechanism of the superconductivity has yet to be uncovered , but many have discussed the nesting of two fermi surfaces at the @xmath8 and @xmath11 points.@xcite however , some experimental and theoretical investigations of a few superconducting fe - pnictides have shown the absence of a nested fermiology , though these compounds are expected to share a common mechanism of the superconductivity .
@xcite in the pnictide family , the recently synthesized 21322 systems sr@xmath0@xmath12o@xmath3fe@xmath13 ( @xmath12=transition metal , and @xmath13=p or as ) , with a thick perovskite - like spacer between fe@xmath14 layers , have stimulated a great enthusiasm due to its possible bimetallic character as well as its high two - dimensionality.@xcite sato _ et al .
_ synthesized a member of the 21311 system sr@xmath0(mg@xmath1ti@xmath2)o@xmath3feas in the range of @xmath4.@xcite with 50% mg doping of the ti sites ( _ i.e. _ , @xmath6=0.5 ) , the superconductivity appears at @xmath5= 10 k , but with a small superconducting volume fraction , which thus implies that there is no intrinsic superconductivity . on reducing the mg concentration
, @xmath5 increases sharply to @xmath15 34 k at @xmath16 , and reaches a maximum @xmath15 40 k at @xmath6=0.2 , slightly higher than that in a stoichiometric sr@xmath0vo@xmath3feas . throughout all of the doping range
considered , no significant changes in the lattice constants were observed .
although synthesis of a sample below @xmath6=0.2 has not been successful yet , the ti rich phase may be expected to have a higher @xmath5 .
this implies that the ti ions play a crucial role in the superconductivity of this system . from the viewpoint of formal charge , the @xmath6=0.5 phase ( _ i.e. _ , 50% mg doping ) consists of tetravalent
ti ions of @xmath17 , leading to a wholly insulating block between feas layers . replacing a fraction of the sites containing mg ions with ti ions ,
ti@xmath18 ion becomes metallic @xmath19 ti@xmath20 on the average .
thus , this system is a good example for investigating the effects of metallic and insulating spacers on the superconductivity .
the effects of a metallic spacer have been under debate for sr@xmath0vo@xmath3feas having v ( @xmath21 ) layers.@xcite in this paper , we investigate the @xmath6=0.5 and 0 phases , which represent either pure insulating or metallic spacers , respectively .
so far , no evidence of magnetic ordering has been reported in this compound .
we therefore will focus on the electronic structures and fermi surfaces of the nonmagnetic ( nm ) state , though possible magnetic tendencies will also be discussed briefly .
in our calculations , based on the tetragonal unit cell with @xmath22 space group , the experimentally measured lattice constants @xmath23=3.935 and @xmath24=15.952 were used , resulting in a slightly larger volume by 2% than in sr@xmath0vo@xmath3feas.@xcite in the cell , fe atoms lie at @xmath25 sites ( @xmath26,@xmath27,0 ) , ti / mg , sr , as , and o atoms lie at @xmath28 sites ( @xmath26,@xmath26,@xmath29 ) , and another o atoms sit at the @xmath30 sites ( @xmath26,@xmath27,@xmath29 ) . using the local density approximation ( lda ) , the internal parameters were optimized to 0.3072 for ti , 0.8210 and 0.5898 for two sr sites , 0.0772 for as , 0.4270 for o at @xmath28 sites , and 0.2939 for o at @xmath30 sites .
these values lead to the as - fe - as bond angle of @xmath31=116@xmath32 , which is measured in one side of the fe layer , corresponding to @xmath725 in the literature.@xcite this significant difference from @xmath740 k suggests a crucial role of the sr@xmath0@xmath33o@xmath3 layer .
the mg - rich phase sr@xmath0ti@xmath34mg@xmath34o@xmath3feas was understood with a @xmath35 supercell using our optimized internal parameters . in these optimization processes ,
the residual hellman - feynman forces on the atoms were lower than 1 mev / . in all calculations ,
the local spin density approximation ( lsda ) implemented in the accurate all - electron full - potential local orbital code , fplo , was used.@xcite a regular mesh containing 196 @xmath36 points in the irreducible wedge was used to sample the brillouin zone .
the magnetic tendencies in the @xmath6=0 phase were studied using the fixed spin moment ( fsm ) method,@xcite with a much denser mesh of up to 726 irreducible @xmath36 points .
first , we will address the electronic structure of the @xmath6=0.5 phase having the insulating spacer .
figure [ band21 ] displays the enlarged band structure with the fatband of the ti @xmath37 manifold ( top panel ) and the fatband of the fe @xmath38 manifold ( bottom panel ) .
note that the @xmath39 and @xmath11 points in the unit cell are folded into the @xmath8 and @xmath39 points in the @xmath40 supercell , respectively .
the corresponding dos near the fermi energy @xmath41 is given in fig .
[ dos21 ] .
as expected , in the @xmath6=0.5 phase , the @xmath42 orbitals of the ti ions are completely unoccupied .
the ti @xmath37 manifold expands from 0.5 ev to 2 ev . in this phase ,
the details near @xmath41 are very similar to those observed for other superconducting fe - pnictides.@xcite at the @xmath8 point , three hole pockets derived from the fe @xmath43 , @xmath44 , and @xmath45 bands appear , while two - fold @xmath11-centered electron pockets with the character of these three bands also appear .
considering that there is no intrinsic superconductivity at this phase and the similar fermiology with other superconducting fe pnictides , the nesting effects may not be a necessary ingredient for fe pnictides to be superconductors .
now , we will focus on the @xmath6=0 phase , which contains only the metallic @xmath46 ti@xmath47 ions .
thus , this represents the phase of the range of @xmath48 , consisting of metallic spacers . in this subsection
, we will address the electronic structure of the nm state .
the enlarged band structure near @xmath41 and the fatband of the ti @xmath37 manifold are displayed in the top panel of fig .
the incomplete tio@xmath49as octahedron results in a breaking of the symmetry of the @xmath37 manifold .
the partially filled ti @xmath50 band spreads over the range of 0.7 to 1.6 ev ( here , @xmath41 is set to zero ) , which is a width that is 15% larger than that of the v @xmath50 band in sr@xmath0vo@xmath3feas.@xcite this band can be described by a single - band tight - binding model with nearest neighbor hopping of @xmath51=0.28 ev and next neighbor hopping of @xmath52=0.08 ev , _
i.e. _ , values that are about 15% larger than in sr@xmath0vo@xmath3feas.@xcite the other bands of ti are mostly unoccupied , but there is some @xmath41 crossing in the @xmath44 and @xmath45 bands near the @xmath8 point , as shown in the bottom panel of fig .
although the hybridization of feas layers and the intervening ti layers is negligible in almost the whole regime due to high two - dimensionality , small mixing with fe @xmath45 and @xmath44 leads to this @xmath41 crossing ( see below ) . compared with other superconducting fe - pnictides , @xcite the bands having fe character , as displayed in the middle panel of fig .
[ band ] , show both similarities and remarkable distinctions . similar to other superconducting fe pnictides ,
the @xmath44 and @xmath45 bands lead to electron pockets at the @xmath11 point .
however , instead of three fe - derived hole pockets at the @xmath8 point , only one fe @xmath45-derived hole pocket appears , but this is considerably shrunk in this system compared to other fe - pnictides ( see below ) .
this may imply that the role of fe for the superconductivity is substantially reduced in this phase .
most of the fe bands are separated from the ti - derived bands , reflecting strong two - dimensionality .
exceptions occur around the @xmath11-point and near the @xmath8 point , near @xmath41 . near the @xmath8 point
, the fe and ti @xmath45 bands are hybridized with each other , leading to a 0.1 ev gap at @xmath41 . on the other hand , around the @xmath11 point ,
the ti @xmath45 are more strongly hybridized with the fe @xmath45 and @xmath44 , with the gap of 0.6 ev that occurs above @xmath41 , resulting in surviving @xmath11-centered electron pockets .
this difference to sr@xmath0vo@xmath3feas results from the different @xmath38-orbital filling between v and ti ions in these systems : namely @xmath21 for v@xmath47 and @xmath46 for ti@xmath47 . as shown in the bottom panel of fig .
[ band ] , noticeable features appear close to @xmath41 . at the @xmath8 point ,
two - fold bands at 40 mev and 58 mev , and a single band at 30 mev appear .
one of the two - fold valence bands is dispersionless near the @xmath8 point along the @xmath8@xmath39 line , leading to a van hove singularity at -40 mev . on the other hand , one of the doublet conduction bands almost touches with @xmath41 along the @xmath8@xmath39 line , leading to a van hove singularity almost exactly at @xmath41 .
additionally , one band lying on @xmath41 appears along the @xmath11@xmath8 line ( actually a few mev above @xmath41 ) .
these features are reflected in the total and atom - projected densities of states ( emphasized near @xmath41 ) , given in fig .
a valley appears at -20 mev between the two van hove singularities at -40 mev and @xmath41 , indicating sensitivity to hole ( or electron ) doping .
this may imply that the stoichiometric sample is near to the optimal doping .
@xmath53 is twice as large than for most other fe pnictides due to the greater contribution of ti , but is only two thirds of that found in sr@xmath0vo@xmath3feas.@xcite remarkably , except for the ti contribution , the magnitude of @xmath53 is similar with that of lafeaso .
it is interesting that the @xmath5 of hole - doped lafeaso is comparable with the @xmath54 system described herein , in which ti ions are insulating .
this fact suggests that metallic ti ions can play an important role in the superconductivity of this system . the fermi surfaces ( fss ) consist of five @xmath8- and two @xmath11-centered pockets , as illustrated in fig .
three large @xmath8-centered electron pockets have ti character and the others have fe character , derived from the @xmath45 band .
the character of the @xmath11-centered fss is mostly derived from the @xmath45 and @xmath44 bands and partially from the @xmath43 bands .
the @xmath11-centered electron pockets possess the shape of an elliptical cylinder ( in fig . [ fs](a ) ) and a rhombus ( in fig .
[ fs](b ) ) . in the regime between the two fe - derived @xmath8-centered
fss , electrons reside in a coaxial distorted rectangular box - like shape ( see fig .
[ fs](d ) ) .
a fe - derived @xmath8-centered hole pocket is considered to be a crucial ingredient for the formation of pairings in fe pnictides , but in this system the pocket is shrunk considerably , implying the opposing view that the metallic ti ions have a more important role .
the three @xmath8-centered electron pockets are derived from ti @xmath50 , @xmath44 , and @xmath45 , see figs .
[ fs](a ) to ( c ) .
the first two fss , which are of similar size , are much like mgb@xmath0 .
a large portion of the fss has remarkably low fermi velocity @xmath55 , with the lowest being @xmath56 cm / s .
in particular , throughout the @xmath11-centered elliptical cylinder and the @xmath8-centered concentric fe - derived fss , @xmath55 is close to its lowest value . for the ti - derived @xmath8-centered fss
, @xmath55 is quite low along the @xmath57 line for the two similar sized fss and along the @xmath58 line in the remaining fs given in fig . [
fs](c ) . in this subsection
, we will address briefly the possible magnetic tendencies in the mg - free phase .
the competition between superconductivity and magnetic ordering widely observed in fe pnictides is unclear for this compound , since no magnetic ordering has been observed .
allowing for a ferromagnetic ( fm ) state , fm has an energy that is a little lower by 5 mev / fe than for nm .
the total magnetic moment is 0.46 @xmath59/fe , mostly comprised of contributions from the ti ions ( 0.37 @xmath59 ) .
the fe ions are nearly nonmagnetic , less than 0.1 @xmath59,@xcite suggesting that the magnetic fluctuation due to ti ions is of more importance than that due to the fe ions in this phase .
the fixed spin moment ( fsm ) method was employed to investigate this magnetic behavior , using a cell that allows only fm and nm states . as obtained from lsda calculations
, fsm shows the minimum energy at a total fixed moment of @xmath11=0.45 @xmath59/fe , with an energy 6 mev / fe lower than in the nm state . from @xmath11=0 to around the minimum energy state , the fe moment is quite small and most of the contribution to @xmath11 results from the ti moment .
after the minimum energy is reached , the moment of fe increases monotonically , coinciding with a sharp increment in energy with a slope of @xmath15 220 mev/@xmath11 ( @xmath11 : total moment per formula unit ) , while that of ti stays at around 0.6(@xmath600.05 ) @xmath59 .
the stoner @xmath61 is 0.47 ev , so @xmath62=2.2 with @xmath53=4.6 states per ev per formula unit obtained from the fm calculations , indicating strong magnetic instability .
for the @xmath6=0 phase , the on - site coulomb repulsion @xmath63 to ti ions was applied using both popular double - counting schemes in the lda+u approach,@xcite to investigate whether or not a mott transition in the @xmath46 ti@xmath47 ions occurs . up to @xmath63=7 ev @xmath15 @xmath64 , which is a much larger value than what has been used in perovskite @xmath65tio@xmath3 ( @xmath65=rare earth elements),@xcite the sr@xmath0tio@xmath3 layers are still metallic .
this failure of lda+u has previously been observed in some @xmath66 or @xmath67 systems , in which no integer occupation number can occur due to strong @xmath68-@xmath38 hybridization , since an integer occupation is required to lead to a mott transition by the lda+u method.@xcite in this system , the calculated occupation number is significantly larger than the formal number @xmath46 , resulting from the absence of rigorous definition of oxidation state in solids , as recently discussed.@xcite this may prevent the lda+u approach from driving a mott transition.@xcite in summary , we addressed the electronic structure of sr@xmath0(mg@xmath1ti@xmath2)o@xmath3feas for the @xmath6=0.5 and @xmath6=0 phases , which represent insulating and metallic ti layers , respectively . at @xmath6=0.5 , which shows no intrinsic superconductivity ,
the fermiology is just like that for other fe - pnictides . introducing metallic ti ions ,
the @xmath8-centered fe - derived fss are reduced considerably or disappeared
. instead , three @xmath8-centered ti fss appear , and two of them have similar sizes like in mgb@xmath0 .
our fsm calculations further suggest possible magnetic ordering in the ti ions .
although no magnetic ordering has been observed in this compound , the high temperature resistivity measurement shows a kink for samples below @xmath6=0.5 at @xmath69 k,@xcite thus implying magnetic transition .
consequently , a detailed magnetic measurement is required for this compound .
our results indicate that the metallic ti ions play a crucial role in the superconductivity , suggesting that the mechanism of superconductivity in fe - pnictides should be reconsidered .
we acknowledge w. e. pickett for fruitful communications and h. ogino for clarifying his experimental results on superconductivity of the @xmath6=0.5 phase .
this research was supported by nrf of korea under grant no .
2012 - 0002245 .
yin , z. p. , leb@xmath75gue , s. , han , m. j. , neal , b. p. , savrasov , s. y. , pickett , w. e. : electron - hole symmetry and magnetic coupling in antiferromagnetic lafeaso .
* 101 * , 047001 ( 2008 ) .
arita , r. , ikeda , h.:as compared to other iron - based superconductors is fermi - surface nesting the origin of superconductivity in iron pnictides ? : a fluctuation - exchange - approximation study .
soc . jpn .
* 78 * , 113707 ( 2009 ) .
borisenko , s. v. , zabolotnyy , v. b. , evtushinsky , d. v. , kim , t. k. , morozov , i. v. , yaresko , a. n. , kordyuk , a. a. , behr , g. , vasiliev , a. , follath , r. , bchner , b. : superconductivity without nesting in lifeas .
lett . * 105 * , 067002 ( 2010 ) .
qian , t. , wang , x .-
p . , jin , w .- c .
, zhang , p. , richard , p. , xu , g. , dai , x. , fang , z. , guo , j .-
g . , chen , x .-
l . , ding ,
h. : absence of a holelike fermi surface for the iron - based k@xmath76fe@xmath77se@xmath0 superconductor revealed by angle - resolved photoemission spectroscopy .
rev . lett . * 106 * , 187001 ( 2011 ) .
ogino , h. , matsumura , y. , katsura , y. , ushiyama , k. , horii , s. , kishio , k. , shimoyama , j. : superconductivity at 17 k in ( fe@xmath0p@xmath0)(sr@xmath78sc@xmath0o@xmath79 ) : a new superconducting layered pnictide oxide with a thick perovskite oxide layer .
* 22 * , 075008 ( 2009 ) .
sato , s. , ogion , h. , kawaguchi , n. , katsura , y. , kishio , k. , shimoyama , j. , kotegawa , h. , tou , h. : superconductivity in a new iron pnictide oxide ( fe@xmath0as@xmath0)(sr@xmath78(mg , ti)@xmath0o@xmath79 ) .
* 23 * , 045001 ( 2010 ) .
, h. , sato , s. , kishio , k. , shimoyama , j. , tohei , t. , ikuhara , y. : homologous series of iron pnictide oxide superconductors ( fe@xmath0as@xmath0)[ca@xmath80(sc , ti)@xmath81o@xmath1 ] ( @xmath83 = 3,4,5 ) with extremely thick blocking layers .
. lett . * 97 * , 072506 ( 2010 ) .
ogino , h. , shimizu , y. , ushiyama , k. , kawaguchi , n. , kishio , k. , shimoyama , j. : superconductivity above 40 k observed in a new iron arsenide oxide ( fe@xmath0as@xmath0)(ca@xmath78(mg , ti)@xmath3o@xmath1 ) .
express * 3 * , 063103 ( 2010 ) .
ogino , h. , machida , k. , yamamoto , a. , kishio , k. , shimoyama , j. , tohei , t. , ikuhara , y. : a new homologous series of iron pnictide oxide superconductors ( fe@xmath0as@xmath0)(ca@xmath84(al , ti)@xmath81o@xmath1 ) ( @xmath83 = 2 , 3 , 4 ) .
* 23 * , 115005 ( 2010 ) .
lee , c .- h . , iyo , a. , eisaki , h. , kito , h. , fernandez - diaz , t. m. , ito , t. , kihou , k. , matsuhata , h. , braden , m. , yamada , k. : effect of structure parameters on superconductivity in fluorine - free lnfeaso@xmath2 ( ln = la , nd ) . j. phys .
. jpn . * 77 * , 083704 ( 2008 ) .
mazin , i. i. , johannes , m. d. , boeri , l. , koepernik , k. , singh , d. j. : problems with reconciling density functional theory calculations with experiment in ferropnictides .
b * 78 * , 085104 ( 2008 ) . using the same internal parameters as in svofa@xcite and the experimental lattice parameters in this system@xcite , fm has the large fe moment of 2.16 @xmath59 , resulting in total moment 2.53 @xmath59 , but much higher energy by @xmath15 315 mev / f.u .
than in the state with nearly nonmagnetic fe ions which is obtained with our relaxed internal parameters .
this fact implies strong magneto - phonon coupling on fe ions in this system as observed in lafeaso.@xcite pavarini , e. , biermann , s. , poteryaev , a. , lichtenstein , a. i. , georges , a. , andersen , o. k. : mott transition and suppression of orbital fluctuations in orthorhombic @xmath85 perovskites .
phys . rev .
* 92 * , 176403 ( 2004 ) . | the highly two - dimensional superconducting system sr@xmath0(mg@xmath1ti@xmath2)o@xmath3feas , recently synthesized in the range of @xmath4 , shows an mg concentration - dependent @xmath5 . reducing the mg concentration from @xmath6=0.5 leads to a sudden increase in @xmath5 , with a maximum @xmath740 k at @xmath6=0.2 . using first principles calculations , the unsynthesized stoichiometric @xmath6=0 and the substoichiometric @xmath6=0.5 compounds
have been investigated . for the 50% mg - doped phase ( @xmath6=0.5 )
, sr@xmath0(mg@xmath1ti@xmath2)o@xmath3 layers are completely insulating spacers between feas layers , leading to the fermiology such as that found for other fe pnictides . at @xmath6=0 , representing a phase with metallic sr@xmath0tio@xmath3 layers , the @xmath8-centered fe - derived fermi surfaces ( fss ) considerably shrink or disappear . instead , three @xmath8-centered ti fss appear , and in particular two of them have similar size , like in mgb@xmath0 .
interestingly , fss have very low fermi velocity in large fractions : the lowest being 0.6@xmath910@xmath10 cm / s .
furthermore , our fixed spin moment calculations suggest the possibility of magnetic ordering , with magnetic ti and nearly nonmagnetic fe ions .
these results indicate a crucial role of sr@xmath0(mg@xmath1ti@xmath2)o@xmath3 layers in this superconductivity .
example.eps gsave newpath 20 20 moveto 20 220 lineto 220 220 lineto 220 20 lineto closepath 2 setlinewidth gsave .4 setgray fill grestore stroke grestore |
several new states have been discovered in the last few years @xcite , most of them at b - factories .
the b - factories provide a clean environment to search for and study these new particles either in @xmath8 meson decays or in the production in @xmath9 reactions with initial state radiation . in this article the recent analyses of new states in @xmath8
decays are presented .
both experiments , babar and belle , together have now a total sample of 1.3 billion @xmath10 pairs available .
the oldest member of the family of new charmonium - like states is the @xmath0 , discovered by belle in 2003 @xcite . despite several measurements of its properties at b - factories and at the tevatron the nature of this particle
is still unknown .
it does not fit into the predicted spectrum of conventional chamonium states . since its mass is very close to the @xmath11 threshold the hypothesis of a @xmath11 molecule
is considered @xcite .
another explanation is a bound state of a diquark and an anti - diquark @xcite .
it is also possible , though unlikely , that the @xmath0 is a @xmath12 state and our theoretical models fail to describe it .
therefore the observation of the @xmath13 was received with great interest .
as it is charged it can not be a conventional charmonium state .
the decay @xmath15 with @xmath16 was first observed by belle in 2006 with a significance of 6.4 standard deviations @xcite .
this observation was confirmed by babar last year @xcite . in @xmath17
decays , where the @xmath18 is reconstructed via @xmath19 or @xmath20 , babar sees an excess near threshold with a significance of 4.9@xmath21 ( see fig . [
fig : x3872dds ] left ) .
a template fit is performed that takes into account threshold effects .
while the width of @xmath22 mev is consistent with other measurements , the mass of @xmath23 mev does not agree with of the mass of @xmath24 mev that is obtained by averaging the measurements in the @xmath25 mode .
this may raise the question whether the same particle is observed in both modes or whether there are additional threshold effects .
belle updated their analysis now including @xmath26
@xcite . combining both @xmath27 modes , a signal with a significance of 8.8@xmath21 is observed ( see fig . [
fig : x3872dds ] right ) .
a fit with a relativistic breit - wigner function and mass - dependent resolution yields @xmath28 mev , @xmath29 mev , and @xmath30 . while the mass is closer to the average value in the @xmath31 mode , some uncertainty due to the line shape model remains .
an alternative fit using a flatt function which assumes a mass below threshold @xcite is able to describe the data , too .
mass distribution measured by babar ( left ) and by belle for @xmath26 ( top right ) and @xmath32 ( bottom right ) . ] mass distribution measured by babar ( left ) and by belle for @xmath26 ( top right ) and @xmath32 ( bottom right ) .
evidence for the decay @xmath15 with @xmath34 was first seen in 2005 by belle with a significance of 4.0@xmath21 @xcite and one year later by babar with 3.4@xmath21 @xcite .
now babar presented an updated analysis adding @xmath35 events and considering also @xmath36 decays @xcite .
signals with a significance of 3.6@xmath21 and 3.5@xmath21 are seen in the @xmath37 and @xmath38 modes , respectively ( see fig . [
fig : x3872psigamma ] ) . for the @xmath8 decays to @xmath39 branching ratios of @xmath40 and @xmath41 are measured .
limits for the other three @xmath8 decay modes where no significant signal is observed are quoted
. the determined relative branching ratio of @xmath42 is orders of magnitude higher than expected in @xmath11 molecule models @xcite .
( left ) and @xmath38 ( right ) mass distribution measured by babar .
the four plots for each mode correspond to @xmath8 decays channels with @xmath43 ( a ) , @xmath44 ( b ) , @xmath45 ( c ) , and @xmath46 ( d ) . ] ( left ) and @xmath38 ( right ) mass distribution measured by babar .
the four plots for each mode correspond to @xmath8 decays channels with @xmath43 ( a ) , @xmath44 ( b ) , @xmath45 ( c ) , and @xmath46 ( d ) . ]
in 2008 belle published the observation of a resonance in the @xmath48 invariant mass spectrum in @xmath49 decays with a significance of 6.5@xmath21 @xcite . in this analysis events with a @xmath50 mass consistent with a @xmath51 or @xmath52 were removed and the resulting @xmath53 distribution was fitted . from the fit the mass @xmath54 mev , the width @xmath55 mev , and the branching ratio @xmath56 were extracted .
babar performed a detailed study of @xmath57 and @xmath49 decays @xcite . as the most prominent features in the dalitz plot are the @xmath51 and @xmath52 resonances a good understanding of the @xmath50 mass spectrum is important .
babar is able to describe it well with @xmath58- , @xmath59- , and @xmath60-waves .
this composition and the angular distributions of the components have a strong influence on the @xmath61 mass spectrum . on the other hand they can not create narrow peaks in the @xmath62 distribution .
babar looks at the @xmath63 and @xmath48 mass distributions for all events , events in the @xmath51 and @xmath52 region , and events outside the @xmath64 regions and compares them to the projections of the @xmath50 components ( see fig .
[ fig : zpsipi ] left ) .
no signal is seen in the @xmath65 spectra and no significant signal in the @xmath53 spectra . for the total sample
there is an excess of 2.7@xmath21 in the @xmath48 mass distribution at @xmath66 mev with @xmath67 mev . using the values measured by belle a 95% confidence level limit on the branching ratio of @xmath68
a comparison of the background subtracted @xmath53 distribution with the one published by belle shows good statistical agreement .
belle reanalyzed their data with a fit to the dalitz plot taking into account all known @xmath50 resonances @xcite .
the fit with @xmath69 resonance included ( solid line in fig .
[ fig : zpsipi ] right ) describes the data significantly better than without ( dotted line ) by 6.4@xmath21 .
systematic variation of the fit model yield a significance of 5.4@xmath21 .
the updated measurements of the @xmath70 properties are @xmath71 mev , @xmath72 mev , and @xmath73 . and @xmath48 mass distributions for all events and events inside and outside the @xmath51 and @xmath52 regions measured by babar ( left ) .
invariant @xmath48 mass distributions in different regions of @xmath74 measured by belle ( right ) . ] and @xmath48 mass distributions for all events and events inside and outside the @xmath51 and @xmath52 regions measured by babar ( left ) .
invariant @xmath48 mass distributions in different regions of @xmath74 measured by belle ( right ) . ]
as the first candidate for a charged exotic state was observed as a @xmath48 resonance , this suggests to search for similar charged states in combinations of other charmionia with pions . in a dalitz plot analysis of @xmath76 events including
all known @xmath50 resonances belle observes two new states , called @xmath77 and @xmath78 @xcite ( see fig .
[ fig : zchic1pi ] ) . from the change in the fit likelihood value a significance of the two-@xmath69-state versus the no-@xmath69-state hypothesis of 13.2@xmath21 is determined .
taking into account systematic variations of the fit model the significance is at least 8.1@xmath21 .
compared to the one-@xmath69-state hypothesis the two states are favored by 5.7@xmath21 , or 5.0@xmath21 including systematics .
the masses , widths and branching ratios of the two new states are measured to be @xmath79 mev , @xmath80 mev , @xmath81 mev , @xmath82 mev , @xmath83 , and @xmath84 .
mass distributions for different @xmath74 regions measured by belle with fit projection assuming no @xmath69 state ( left ) and two @xmath69 states ( right ) . ] mass distributions for different @xmath74 regions measured by belle with fit projection assuming no @xmath69 state ( left ) and two @xmath69 states ( right ) . ]
the study of the decay @xmath1 is an interesting topic because it is very close to threshold .
while the @xmath0 mass measured by babar is significantly higher than the average mass measured in the @xmath31 mode , the latest belle measurement agrees with it within 2@xmath21 .
a detailed study of the exact line shape may reveal important information about the nature of the @xmath0 , but requires a much larger data sample .
the first observation of the decay @xmath36 and the measured high branching ratio with respect to the @xmath37 mode disfavor the hypothesis of a @xmath11 molecule .
the first observation of a charged charmonium - like state , the @xmath13 , was confirmed in a dalitz plot analysis by belle .
but so far this particle is not confirmed by any other experiment .
babar does not see a significant signal in the @xmath63 and @xmath48 mass spectra .
although statistical agreement with the belle data is established , the conclusions of both experiments differ .
two further charged states with even higher significance have been observed as @xmath85 resonances by belle .
a clear picture of the @xmath69 states with precise measurements of their properties may only arise at upgraded b - factories .
g. gokhroo _ et al . _ [ belle collaboration ] , phys .
lett . *
97 * , 162002 ( 2006 ) .
b. aubert _ et al . _ [ babar collaboration ] , phys .
011102 ( 2008 ) . i. adachi _ et al . _
[ belle collaboration ] , arxiv : hep - ex/0810.0358 .
c. hanhart , yu .
s. kalashnikova , a. e. kudryavtsev and a. v. nefediev , phys .
d * 76 * , 034007 ( 2007 ) .
k. abe _ et al . _ [ belle collaboration ] , arxiv : hep - ex/0505037 .
b. aubert _ et al . _ [ babar collaboration ] , phys .
d * 74 * , 071101 ( 2006 ) .
b. aubert _ et al . _ [ babar collaboration ] , phys .
lett . * 102 * , 132001 ( 2009 ) .
e. s. swanson , phys .
b * 598 * , 197 ( 2004 ) .
s. k. choi _ et al . _
[ belle collaboration ] , phys .
lett . * 100 * , 142001 ( 2008 ) .
b. aubert _ et al .
_ [ babar collaboration ] , arxiv : hep - ex/0811.0564 .
belle collaboration , kek preprint 2009 - 2 , + ` http://belle.kek.jp/belle/preprint/2009-9/z4430_dalitz_apr29.ps.gz ` | since the discovery of the @xmath0 in 2003 the b - factory experiments babar and belle have been driving forces in the spectroscopy of new charmonium - like states . in this article
the latest results on the decays @xmath1 and @xmath2 , where @xmath3 stands for @xmath4 or @xmath5 , are presented and analyses of @xmath6 and @xmath7 are discussed . |
the global attractor of a dynamical system is the unique compact invariant set that attracts the trajectories starting in any bounded set at a uniform rate
. introduced by billotti & lasalle @xcite , they have been the subject of much research since the mid-1980s , and form the central topic of a number of monographs , including babin & vishik @xcite , hale @xcite , ladyzhenskaya @xcite , robinson @xcite , and temam @xcite .
the standard theory incorporates existence results @xcite , upper semicontinuity @xcite , and bounds on the attractor dimension @xcite .
global attractors exist for many infinite - dimensional models @xcite , with familiar low - dimensional ode models such as the lorenz equations providing a testing ground for the general theory @xcite .
while upper semicontinuity with respect to perturbations is easy to prove , lower semicontinuity ( and hence full continuity ) is more delicate , requiring structural assumptions on the attractor or the assumption of a uniform attraction rate .
however , babin & pilyugin @xcite proved that the global attractor of a parametrised set of semigroups is continuous at a residual set of parameters , by taking advantage of the known upper semicontinuity and then using the fact that upper semicontinuous functions are continuous on a residual set . here
we reprove results on equi - attraction and residual continuity in a more direct way , which also serves to demonstrate more clearly why these results are true .
given equi - attraction the attractor is the uniform limit of a sequence of continuous functions , and hence continuous ( the converse requires a generalised version of dini s theorem ) ; more generally , it is the pointwise limit of a sequence of continuous functions , i.e. a ` baire one ' function , and therefore the set of continuity points forms a residual set .
a semigroup @xmath9 on a complete metric space @xmath10 is a collection of maps @xmath11 such that * @xmath12 ; * @xmath13 for all @xmath14 ; and * @xmath15 is continuous in @xmath16 and @xmath17 .
a compact set @xmath18 is the _ global attractor _ for @xmath19 if * @xmath20 for all @xmath21 ; and * for any bounded set @xmath22 , @xmath23 as @xmath6 , where @xmath24 is the semi - distance @xmath25 .
when such a set exists it is unique , the maximal compact invariant set , and the minimal closed set that satisfies ( a2 ) .
let @xmath0 be a complete metric space and @xmath26 a parametrised family of semigroups on @xmath27 .
suppose that * @xmath26 has a global attractor @xmath2 for every @xmath5 ; * there is a bounded subset @xmath3 of @xmath27 such that @xmath4 for every @xmath5 ; and * for @xmath28 , @xmath29 is continuous in @xmath8 , uniformly for @xmath16 in bounded subsets of @xmath27 .
we can strengthen ( l2 ) and weaken ( l3 ) by replacing ` bounded ' by ` compact ' to yield conditions ( l2@xmath30 ) and ( l3@xmath30 ) .
a wide range of dissipative systems with parameters satisfy these assumptions , for example the 2d navier stokes equations , the scalar kuramoto
sivashinsky equation , reaction - diffusion equations , and the lorenz equations , all of which are covered in @xcite . under these mild assumptions
it is easy to show that @xmath2 is upper semicontinuous , @xmath31 see @xcite , for example .
however , lower semicontinuity , that is @xmath32 requires more : either structural conditions on the attractor @xmath33 ( @xmath33 is the closure of the unstable manifolds of a finite number of hyperbolic equilibria , see hale & raugel @xcite , stuart & humphries @xcite , or robinson @xcite ) or the ` equi - attraction ' hypothesis of li & kloeden @xcite ( see section [ ea ] ) . as a result ,
the continuity of attractors , @xmath34 where @xmath35 is the symmetric hausdorff distance , is only known under restrictive conditions . in this paper
we view @xmath2 as a function from @xmath0 into the space of closed bounded subsets of @xmath27 , given as the limit of the continuous functions @xmath36 ( see lemma [ dncts ] ) .
indeed , note that given any set @xmath37 it follows from the invariance of the attractor ( a1 ) that @xmath38 and so the the attraction property of the attractor in ( a2 ) implies that @xmath39 uniform convergence ( with respect to @xmath8 ) in ( [ limit ] ) is essentially the ` equi - attraction ' introduced in @xcite , and thus clearly related to continuity of the limiting function @xmath2 ( section [ ea ] ) . given only pointwise ( @xmath8-by-@xmath8 ) convergence in ( [ limit ] ) we can still use the result from the theory of baire category that the pointwise limit of continuous functions ( a ` baire one function ' ) is continuous at a residual set to guarantee that @xmath2 is continuous in @xmath8 on a residual subset of @xmath0 ( section [ bc ] ) . for both results
the following simple lemma is fundamental .
we let @xmath40 be the collection of all closed and bounded subsets of @xmath27 , and use the symmetric hausdorff distance @xmath41 defined in ( [ shd ] ) as the metric on this space .
[ dncts ] suppose that @xmath3 is bounded and that ( l3 ) holds .
then for any @xmath28 the map @xmath42 is continuous from @xmath0 into @xmath40 .
the same is true if @xmath3 is compact and ( l3@xmath30 ) holds . given @xmath28 , @xmath43 , and @xmath44 , ( l3 ) ensures that there exists a @xmath45 such that @xmath46 implies that @xmath47 for every @xmath48 .
it follows that @xmath49 and so @xmath50 from which @xmath51 as required .
first we give a simple proof of the results in @xcite on the equivalence between equi - attraction and continuity . in our framework these follow from two classical results : the continuity of the uniform limit of a sequence of continuous functions and dini s theorem in an abstract formulation .
li & kloeden require ( l1 ) , ( l2 ) , a time - uniform version of ( l3 ) ( see corollary [ cor ] ) , and in addition an ` equi - dissipative ' assumption that there exists a bounded absorbing set @xmath52 such that @xmath53 where @xmath54 does not depend on @xmath8 .
they then show that when @xmath0 is compact , continuity of @xmath2 in @xmath8 is equivalent to equi - attraction , @xmath55 we now give our version of dini s theorem . [ dini ] for each @xmath56 let @xmath57 be a continuous map , where @xmath52 is a compact metric space and @xmath58 is any metric space .
if @xmath59 is continuous and is the monotonic pointwise limit of @xmath60 , i.e. for every @xmath61 @xmath62 then @xmath60 converges uniformly to @xmath59 .
given @xmath44 define @xmath63 since @xmath60 and @xmath59 are both continuous , @xmath64 is open and non - decreasing .
since @xmath52 is compact and @xmath65 provides an open cover of @xmath52 , there exists an @xmath66 such that @xmath67 , and so @xmath68 for all @xmath61 for all @xmath69 .
our first result relates continuity to a slightly weakened form of equi - attraction through sequences .
we remark that our proof allows us to dispense with the ` equi - dissipative ' assumption ( [ equid ] ) of @xcite .
[ cty ] assume ( l1 ) and ( l23 ) or ( l2@xmath303@xmath30 ) . if there exist @xmath70 such that @xmath71 then @xmath2 is continuous in @xmath8 for all @xmath5 .
conversely , if @xmath0 is compact then continuity of @xmath2 for all @xmath5 implies that there exist @xmath70 such that ( [ unif ] ) holds .
lemma [ dncts ] guarantees that @xmath72 is continuous for each @xmath73 , and we have already observed in ( [ limit ] ) that @xmath74 therefore @xmath2 is the uniform limit of the continuous functions @xmath36 and so is continuous itself . for the converse ,
let @xmath75 . for each @xmath43
it follows from ( a2 ) and ( l2 ) that there exists a time @xmath76 such that @xmath77 for all @xmath78 .
it follows from ( l3 ) that there exists an @xmath79 such that @xmath80 for every @xmath8 with @xmath81 . since @xmath0 is compact @xmath82 for some @xmath83 and @xmath84 .
if @xmath85 then @xmath86 since any @xmath5 is contained in @xmath87 for some @xmath88 , and @xmath89 for some @xmath90 , from which ( [ second ] ) follows by applying @xmath91 repeatedly ( @xmath92 times ) to both sides of ( [ first ] ) .
it follows from ( [ second ] ) that for every @xmath5 , @xmath93 is a decreasing sequence of sets , and hence the convergence of @xmath93 to @xmath2 , ensured by ( [ limit ] ) , is in fact monotonic in the sense of our theorem [ dini ] .
uniform convergence now follows , and finally the fact that @xmath94 yields @xmath95 uniformly in @xmath8 as @xmath96 . with additional uniform continuity assumptions
we can readily show that continuity implies equi - attraction in the sense of @xcite .
we give one version of this result .
[ cor ] suppose that ( l13 ) hold and that @xmath0 is compact .
assume in addition that @xmath29 is continuous in @xmath16 , uniformly in @xmath8 and for @xmath16 in bounded subsets of @xmath27 and @xmath97 $ ] for any @xmath98
. then continuity of @xmath2 implies ( [ unifkl ] ) .
since @xmath4 and @xmath2 is invariant , given any @xmath44 , by assumption there exists a @xmath45 such that @xmath99 any @xmath100 , for all @xmath5 and @xmath101 .
choose @xmath102 sufficiently large that @xmath103 now for any @xmath104 , @xmath105 , we can write @xmath106 for some @xmath101 , and it follows from ( [ morecty ] ) that @xmath107 with the convergence uniform in @xmath8 as required .
when the rate of attraction to @xmath2 is not uniform in @xmath8 we nevertheless have the convergence in ( [ limit ] ) for each @xmath8 . in general , therefore , one can view the attractor as the ` pointwise ' ( @xmath8-by-@xmath8 ) limit of the sequence of continuous functions @xmath108 . in the case of a sequence of continuous real functions ,
their pointwise limit is a ` baire one function ' , and is continuous on a residual set .
the same ideas in a more abstract setting yield continuity of @xmath2 on a residual subset of @xmath0 .
we use the following abstract result , characterising the continuity of ` baire one ' functions , whose proof ( which we include for completeness ) is an easy variant of that given by oxtoby @xcite . a result in the same general setting as here can be found as theorem 48.5 in munkres @xcite . recall that a set is _ nowhere dense _ if its closure contains no open sets , and a set is _ residual _ if its complement is the countable union of nowhere dense sets .
any residual set is dense .
[ bct ] for each @xmath56 let @xmath109 be a continuous map , where @xmath0 is a complete metric space and @xmath58 is any metric space .
if @xmath59 is the pointwise limit of @xmath60 , i.e. @xmath110 for each @xmath5 ( and the limit exists ) , then the points of continuity of @xmath59 form a residual subset of @xmath0 . before the proof we make the following observation : if @xmath111 and @xmath112 are open subsets of @xmath0 with @xmath113 , then @xmath114 .
otherwise @xmath115 , the complement of @xmath112 in @xmath0 , is a closed set containing @xmath111 , and since @xmath116 is the intersection of all closed sets that contain @xmath111 , @xmath117 , a contradiction . to show that @xmath120 is nowhere dense , i.e. that its closure contains no open set , let @xmath121 note that @xmath64 is closed , @xmath122 , and @xmath123 .
choose any open set @xmath124 , and consider @xmath125 .
since @xmath116 is a complete metric space , it follows from the baire category theorem that there exists an @xmath73 such that @xmath126 contains an open set @xmath127 . from the remark before the proof
, @xmath128 is an open subset of @xmath126 that is in addition a subset of @xmath111 . since @xmath129
, it follows that @xmath130 for all @xmath131 and @xmath132 .
fixing @xmath133 and letting @xmath134 it follows that @xmath135 now , since @xmath136 is continuous in @xmath8 , for any @xmath137 there is a neighbourhood @xmath138 such that @xmath139 thus by the triangle inequality @xmath140 it follows that no element of @xmath141 belongs to @xmath120 , which implies , since @xmath142 that @xmath111 contains an open set that is not contained in @xmath120 .
this shows that @xmath120 is nowhere dense , which concludes the proof .
[ main ] under assumptions ( l13 ) above or ( l1 ) , ( l2@xmath30 ) , and ( l3@xmath30 )
@xmath2 is continuous in @xmath8 for all @xmath143 in a residual subset of @xmath0 .
in particular the set of continuity points of @xmath2 is dense in @xmath0 .
we showed in lemma [ dncts ] that for every @xmath28 the map @xmath144 is continuous from @xmath0 into @xmath145 , and observed in ( [ limit ] ) the pointwise convergence @xmath146 the result follows immediately from theorem [ bct ] , setting @xmath147 and @xmath148 for every @xmath5 .
residual continuity results also hold for the pullback attractors @xcite and uniform attractors @xcite that occur in non - autonomous systems .
we will discuss these results in the context of the two - dimensional navier
stokes equations in a future paper . | let @xmath0 be a complete metric space , and let @xmath1 be a parametrised family of semigroups with global attractors @xmath2 .
we assume that there exists a fixed bounded set @xmath3 such that @xmath4 for every @xmath5 . by viewing the attractors as the limit as @xmath6 of the sets @xmath7 , we give simple proofs of the equivalence of ` equi - attraction ' to continuity ( when this convergence is uniform in @xmath8 ) and
show that the attractors @xmath2 are continuous in @xmath8 at a residual set of parameters in the sense of baire category ( when the convergence is only pointwise ) . |
quantum entanglement is one of the most striking features of quantum physics .
it remarkably reveals nonlocality and may lead to powerful applications in quantum information and quantum computation @xcite .
but entanglement is incredibly fragile under environmental noises , and any local operations can not increase it .
these are serious frustrations for using entanglement .
quantum discord @xcite is another quantum correlation other than entanglement , and also can be used for quantum computation @xcite .
it has been shown that quantum discord is more robust than entanglement @xcite .
furthermore , recent studies @xcite indicate that local environmental noises can create quantum discord .
these are exciting results compared to the case of entanglement .
in @xcite , f. ciccarello and v. giovannetti showed that a zero discord state of two qubits @xmath0 can create quantum discord under an amplitude damping channel , where @xmath1 where @xmath2,\end{aligned}\ ] ] and the noise is the generalized amplitude damping channel .
we know generalized amplitude damping describes temperature effects , so we are especially interested in the temperature effect on the quantum discord .
this paper is organized as follows . in sec .
2 , we briefly recall some basics about quantum discord and generalized amplitude damping as preparations . in sec . 3
, we investigate the quantum discord of two qubits whose intial state as in eq.(3 ) and undergoes a local generalized amplitude damping .
4 is a brief summary .
suppose two quantum systems a and b are described by the hilbert spaces @xmath3 and @xmath4 , respectively .
the composite system ab is then described by the hilbert space @xmath5 @xmath4 . for a state @xmath6 on @xmath5 @xmath4 ,
the quantum discord ( with respect to a ) of @xmath6 is defined as @xcite @xmath7 where @xmath8 , @xmath9 , inf takes all projective measurements @xmath10 on system a , @xmath11 is the von neumann entropy .
it can be proven that @xcite @xmath12 where @xmath13 is an orthonormal basis for @xmath3 , @xmath14 are density operators on @xmath4 , @xmath15 , @xmath16 .
eq.(5 ) is hard to optimize , even for two qubits case till now only few states are found possessing analytical expressions @xcite .
b. dakic , v. vedral and c. brukner proposed a geometric measure of quantum discord as @xcite @xmath17:d_a(\sigma ) = 0\}.\end{aligned}\ ] ] evidently , @xmath18 as an elegant result , @xmath19 allows analytical expressions for all two qubits states @xcite . more specifically , for a two qubits state @xmath20 where , @xmath21 , @xmath22 , @xmath23 are pauli matrices , @xmath24 , @xmath25 , @xmath26 are all real numbers with @xmath27,y_j = tr[\rho i\otimes \sigma _ j],t_{ij}=tr[\rho \sigma _ i\otimes \sigma _ j],\end{aligned}\ ] ] then @xcite @xmath28 where @xmath29 is the largest eigenvalue of the matrix @xmath30 , @xmath31 , @xmath32 is the matrix @xmath33 . generalized amplitude damping describes the effect of dissipation to an environment at finite temperature ( @xcite , 8.3.5 ) .
suppose two qubits systems a and b , given the bases @xmath34 , @xmath34 for @xmath3 and @xmath4 , the operation elements for generalized amplitude damping are @xmath35 where @xmath36 $ ] , @xmath37 $ ] .
@xmath38 and @xmath39 are functions of temperature @xmath32 .
@xmath40 describes the transition probability between @xmath41 and @xmath42 at temperature @xmath32 .
@xmath43 describes temperature effects , @xmath44 is the case of amplitude damping .
we can assume that ( see for example @xcite ) @xmath43 is a monotonically decreasing function when t varies from 0 to @xmath45 , and @xmath46 from eq.(12 ) we can see that p varies from 0 to 0.5 is symmetric to that p varies from 1 to 0.5 in the sense exchanging states @xmath47 and @xmath48 , so we only need to consider the case p varies from 1 to 0.5 . a state @xmath6 after this generalized amplitude damping will be a state @xmath49
now suppose two qubits initially is in the state as in eq.(3 ) , through the generalized amplitude damping channel in eq.(12 ) , the output state @xmath50 can be calculated directly by eqs.(3,12,14 ) , will be @xmath51 where @xmath52 we first ask , for what initial states and what generalized amplitude damping channel , the quantum discord of the output states , @xmath53 , are vanishing ? this is the proposition 1 below .
_ proposition 1 .
_ the quantum discord @xmath53 of the state @xmath6 in eq.(15 ) is zero if one of the conditions below is satisfied ( i ) .
@xmath54 ( ii ) .
@xmath55 ( iii ) .
@xmath56 ( iv ) .
@xmath57 ( v ) . @xmath58 _ proof . _ from eq.(8 ) , we only need to prove @xmath59 using eq.(11 ) to eq.(15 ) , after some calculations , we get @xmath60[(x_3-t_{13})^2+t_{33}^2]}\}.\end{aligned}\ ] ] from eq.(16 ) , it is easy to see that ( vi ) .
if @xmath61 , then @xmath62 ( vii ) .
if @xmath63 , then @xmath64 ( viii ) . if @xmath65 , then @xmath66 ( ix )
if @xmath67 , then @xmath66 ( x ) . if @xmath68 , then @xmath69 .
each of the cases above can result in @xmath70 .
we then complete this proof .
we next investigate the quantum discord @xmath53 for the state @xmath6 in eq.(15 ) for more general cases , that is @xmath71 , @xmath72 , @xmath73 $ ] .
any orthonormal basis of @xmath3 can be expressed as an unitary matrix @xmath74 multiplied by the orthonormal basis @xmath34 .
since @xmath74 can be written as @xmath75 where @xmath76 are all real numbers and satisfy @xmath77 then any orthonormal basis of @xmath3 can be expressed as @xmath78 denote @xmath79 and @xmath80 , then from eqs.(15,20,21 ) , we have @xmath81+\frac{\sigma _ z } 4[2(tb+ac)t_{13}+(t^2+c^2-a^2-b^2)t_{33 } ] , \\
\fl p_0=tr\langle \varphi _ 0|\rho |\varphi _ 0\rangle = \frac 12(1+x_3)-(a^2+b^2)x_3 , \\ \fl p_1\rho _ 1=\frac i4[1+(a^2+b^2-t^2-c^2)x_3]+\frac{\sigma _ z } 4[-2(tb+ac)t_{13}+(a^2+b^2-t^2-c^2)t_{33 } ] , \\ \fl p_1=tr\langle \varphi _ 1|\rho |\varphi _
1\rangle = \frac 12(1-x_3)+(a^2+b^2)x_3.\end{aligned}\ ] ] let @xmath82 then @xmath83 where @xmath84 , \ \ \ \theta _ 1\in [ 0,2\pi ) , \ \ \ \theta _ 2\in [ 0,2\pi ) , \ \ \ \theta _ 3=\theta _ 1+\theta _ 2.\end{aligned}\ ] ] eqs.(22 - 25 ) then read @xmath85+\frac{\sigma _
z}4[2\sqrt{r(1-r)}\sin \theta _ 3t_{13}+(1 - 2r)t_{33 } ] , \\ \fl \ \ \ \ \ \ \ \
p_0=\frac 12[1+(1 - 2r)x_3 ] , \\ \fl \
\ \ \ \ \ \ \ p_1\rho _ 1=\frac i4[1-(1 - 2r)x_3]+\frac{\sigma _
z}4[-2\sqrt{r(1-r)}\sin \theta _ 3t_{13}-(1 - 2r)t_{33 } ] , \\ \fl \ \ \ \ \ \ \ \
p_1=\frac 12[1-(1 - 2r)x_3].\end{aligned}\ ] ] let @xmath86 then eqs.(29 - 32 ) become @xmath87+\frac{\sigma _
z}4[\beta t_{13}+\alpha t_{33 } ] , \\
p_0=\frac 12[1+\alpha x_3 ] , \\ p_1\rho
_ 1=\frac i4[1-\alpha x_3]+\frac{\sigma
_ z}4[-\beta t_{13}-\alpha t_{33 } ] , \\
p_1=\frac 12[1-\alpha x_3],\end{aligned}\ ] ] with @xmath88 from eqs.(34 - 37 ) , it is easy to find that @xmath89 has two eigenvalues @xmath90 and @xmath91 has two eigenvalues @xmath92 thus @xmath93 where @xmath94 is the binary entropy @xmath95 @xmath96 is defined on @xmath97,$ ] @xmath96 is a concave function , and @xmath98 we say @xmath99 is a concave function in @xmath100 , this can be seen from the concavity of @xmath96 and the facts below : ( xi ) .
if @xmath101 and @xmath102 are convex functions , then so is @xmath103 ( @xcite , 3.2.1 ) ; ( xii ) .
convexity is invariant under affine maps : that is , if f(x ) is convex with @xmath104 , then so is @xmath105 with @xmath106 , @xmath107 , @xmath108 an @xmath109 real matrix ( @xcite , 3.2.2 ) ; ( xiii ) . if f(x ) is convex , then so is @xmath110 with @xmath111 ( @xcite , 3.2.6 ) .
it follows that the minimum of @xmath99 in eq.(41 ) over the domain @xmath112 can be achieved on the unit circle @xmath113 further , from eq.(41 ) , it is evident that @xmath114 hence , @xmath115=\min \{f(\theta ) : 0\leq \theta
\leq \pi \},\end{aligned}\ ] ] where @xmath116 finally , we get the expression of @xmath117 as @xmath118 where @xmath119 and @xmath120 can be calculated directly by eq.(15 ) as @xmath121 from these expressions , we have proposition 2 below .
_ proposition 2 .
_ quantum discord @xmath53 in eq.(46 ) for the state @xmath6 in eq.(15 ) has the same value for @xmath122 and @xmath123 , that is , @xmath53 is symmetric about @xmath124 . _
_ from eq.(16 ) , if @xmath122 generates @xmath125 , then @xmath123 generates @xmath126 .
@xmath125 and @xmath126 generate the same value for @xmath127 , see eqs.(47 - 48 ) . also , in @xmath128 of eq.(45 )
, by use of @xmath129 , we can get @xmath130 thus , we can readily attain proposition 2 and end this proof .
and @xmath131 when @xmath132.,title="fig:",width=264 ] and @xmath131 when @xmath132.,title="fig:",width=264 ] fig.1 depicts @xmath53 as a function of @xmath133 at @xmath134 , @xmath135 , @xmath136 , @xmath137 , respectively . from fig.1
we see that , when @xmath138 is close to @xmath124 and @xmath139 close to 1 , @xmath53 is relatively large .
that is to say , more superposition of the states @xmath41 and @xmath42 in eq.(3 ) can boost the creating of quantum discord , while high temperature , conversely , can impede the creating .
fig.2 shows @xmath53 as a function of @xmath140 at @xmath124 .
we investigated the quantum discord of two qubits which initially in completely classical state then experienced a local generalized amplitude damping channel .
we showed that a completely classical state can create quantum discord through a local generalized amplitude damping , while high temperature will impede the creating of quantum discord .
this work was supported by national natural science foundation of china ( grant nos .
10775101 ) .
the author thanks qing hou for helpful discussions . | we show that two qubits initially in completely classical state can create quantum discord through a local generalized amplitude damping channel , but high temperature will impede the creating of quantum discord . |
an understanding peculiarities of alfvn oscillations in solid star models is of particular interest for the asteroseismology of final stage ( fs ) compact objects of stellar evolution , white dwarfs and neutron stars .
the super dense and highly conducting matter of these stars , capable of accumulating magnetic fields of extremely powerful intensity ( e.g. , chanmugam 1992 ) , possesses mechanical elasticity generic to solids .
this suggests that proper account of seismic vibrations of fs stars can be attained working from equations of elastodynamics and magneto - solid - mechanics , contrary to the liquid stars of main sequence ( ms ) whose asteroseismology is adequately modeled by equations of hydrodynamics and magneto - fluid - mechanics .
another important feature distinguishing liquid ms stars from solid fs stars is that the magnetic fields of the ms stars are generated in the dynamo processes whose energy supply comes from central thermonuclear reactive zone ( parker 1979 ) , whereas in the fs stars there are no nuclear energy sources to support persistent convection .
it is commonly believed today that magnetic fields of non - convective fs stars are fossil ( e.g. ferrario and wickramasinghe 2004 ) . in these latter non - convective stars , like white dwarfs ,
the fossil magnetic fields are regarded as frozen in the immobile extremely dense and perfectly conducting matter .
most studies devoted to interpretation of quasi - periodic oscillations ( qpos ) in the electromagnetic emission of fs stars as being produced by lorentz - force - driven alfvn oscillations have been perform within the framework of the fiducial homogeneous model of a solid star with an axisymmetric poloidal internal magnetic field of constant intensity @xmath0,\\ \label{e1.2 } & & w_m({\rm pmf})=\frac{1}{8\pi}\int b^2\,d{\cal v}=\frac{1}{6}b^2r^3 \end{aligned}\ ] ] where @xmath1 is the internal magnetic energy of the star . in particular , in our recent papers ( bastrukov et al 2009a , 2009b )
this model has been invoked to investigate a poorly studied regime of global node - free alfvn oscillations in quaking neutron stars .
based on the well - known argument that magnetic field frozen - in a perfectly conducting matter imparts to stellar material a supplementary portion of elasticity ( e.g. chandrasekhar 1961 ) , and taking into account formal similarity between equations of classical elastodynamics and magneto - solid - mechanics , it has been conjectured that material displacements in a neutron star undergoing node - free oscillations restored by lorentz force of magnetic field stresses should have one and the same form as those for a solid star executing nodeless spheroidal and torsional shear oscillations restored by hooke s elastic force of solid - mechanical stresses whose frequency spectra has early been computed in ( bastrukov et al 2007 , 2008 ) .
following this line of argument it has been shown that this suggestion and the energy variational method of magneto - solid - mechanical theory allow one to get exact spectral formulae for the frequencies of the node - free poloidal and toroidal @xmath2-modes in neutron stars . in the present paper
we discuss application of magneto - solid - mechanical theory of alfvn oscillations developed in ( bastrukov et al 2009b ) to the white dwarf asteroseismology .
the material of this wide class of pulsating solid stars is identified with a metal - like solid - state plasma composed of ions suspended in the degenerate fermi - gas of relativistic electrons .
therefore if one assume that all pulsating white dwarfs , both magnetic and non - magnetic , come into existence with frozen - in magnetic fields , then there seems to be no theoretical reason why the phenomenon of alfvn vibrations should not occur in these stars . to the best of our knowledge , the possibility that magnetic da and db white dwarfs endowed with poloidal magnetic fields ( having dipolar - like external configuration ) are capable of maintaining alfvn oscillations has profoundly been discussed some time ago by lou ( 1995 ) who was the first to present quite convincing arguments that global alfvn oscillations , entrapped in the entire volume of white dwarf , can be self - excited via @xmath3-mechanism of ionization ( of hydrogen in da and helium in db stars ) , that is , by one and the same mechanism which is responsible of excitation in white dwarfs of @xmath4-modes ( winget 1998 , hansen , kawaler and trimble 2004 , fontain and brassard 2008 , winget and kepler 2008 ) . adhering to this line of argument , in this article
we investigate peculiarities of lorentz - force - driven global spheroidal and torsional nodeless vibrations of a solid star with frozen - in purely toroidal and completely intrinsic magnetic field @xmath5 $ ] and discuss expected observational consequences of node - free poloidal and toroidal alfvn oscillations in non - magnetic white dwarfs endowed with toroidal magnetic field completely buried beneath its surface .
first , is the model with the azimuthal component of magnetic field having the form @xmath6 which is pictured , with the aid of computer code maple , in fig.1 .
\sin\theta$ ] entrapped in the star volume.,width=188 ] and second is the model with pictured in fig.2 .
azimuthal component of magnetic field of the form @xmath7^{1/2}}{r}. \end{aligned}\ ] ] it may be worth noting that the store of magnetic energy in the stars with the exactly above forms of toroidal magnetic fields is less than that for a star model with poloidal homogeneous internal and dipolar external magnetic field by @xmath8 in the expressions for @xmath9 and @xmath10 , with constant value of @xmath11 equal to @xmath12 for the former model and @xmath13 for the latter one , leads to energy of magnetic field precisely coinciding with that for a star with uniform internal poloidal magnetic field . ]
@xmath14 the practical usefulness of such models is that their inferences can be utilized in the asteroseismology of wide class of pulsating non - magnetic white dwarfs if we are to understand the impact of toroidal magnetic field , entrapped in the volume of the star , on seismic vibrations restored by lorentz force of magnetic field stresses and to identify characteristic spectral features of these vibrations in the detected qpos of x - ray and optical emission from these stars .
^{1/2}$].,width=188 ] it is worth noting that node - free alfvn oscillations in polytropic self - gravitating gaseous masses ( governed by equations of magneto - fluid - mechanics lying at the base of asteroseimology of liquid ms stars ) with purely toroidal magnetic field have been the subject of long - lasting investigations in the past ( chandrasekhar 1956 , anand & kushwaha r.s .
1962 , roxburgh 1962 , roxburgh and durney 1967 , anand 1969 , sood and trehan 1972 ; miketinac 1973 ; goossens 1976 , goossens and veugelen 1978 , goossens and biront 1980 ) . in most of these focus was on a liquid star model with purely toroidal magnetic field of the form , @xmath15 .
the net outcome of these studies is that the presence of such a field can significantly modify the frequencies of spheroidal non - radial nodeless oscillations such as in kelvin fundamental mode ( bastrukov et al 2009c ) . in the meantime
, the regime of nodeless alfvn oscillations in the solid fs stars with purely toroidal magnetic field remains less studied . in sec.2
, we show that proper account of this case can done on the basis of the energy method of magneto - solid - mechanical theory that has been extensively discussed in our recent work ( bastrukov et al 2009b ) . in sections 3 and 4 , based on this method the detailed calculation of the frequency spectra of node - free poloidal and toroidal @xmath2-modes in a solid stars with intrinsic purely azimuthal magnetic fields of above form
is presented and discussed in the context of current development of the white dwarf asteroseismology .
the newly obtained results are briefly summarized in sec.5 .
the governing equations of solid - magnetics ( formulated as a solid - mechanical counterpart of equations of hydromagnetics ) describing perfectly conducting elastic continuous medium of density @xmath16 pervaded by magnetic field @xmath17 are @xmath18,\\ \label{e1.3 } & & \delta b_i=\nabla_k\,[u_i\,b_k - u_k\,b_i],\\ & & \frac{\partial } { \partial t } \int \frac{\rho { \dot u}^2}{2}\ , d{\cal v}=-\int \delta m_{ik}\,{\dot u}_{ik}\,d{\cal v}. \end{aligned}\ ] ] as in ( bastrukov et al 2009a , 2009b , 2009c ) in what follows we consider vibrations in which material displacements @xmath19 are described by harmonic solenoidal fields obeying vector laplace equation @xmath20 which is regarded as characteristic equation of the regime of node - free shear vibrations , seismic stellar vibrations , governed by equation @xmath21 with shear stresses obeying hooke s law @xmath22 with @xmath23 being the shear modulus of star matter , the laplace equation can be thought of as the long - wavelength limit of helmholtz equation , @xmath24 .
this latter equation emerges from above solid - dynamical equation after substitution @xmath25 and describes standing - wave regime of elastic shear vibrations . understandably that in the limit of long waves @xmath26 ,
i.e. when wave vector @xmath27 , the helmholtz equation is reduced to the laplace equation ( bastrukov et al 2007 ) . ]
. bearing in mind the above mentioned similarity between oscillatory behavior of magneto - active solid - state plasma and elastically deformable solid , the field of instantaneous , i.e. time - independent , material displacements , @xmath28 , which is related to @xmath29 as @xmath30 has been been taken in one and the same form of poloidal @xmath31 and toroidal @xmath32 vector fields as in the case of nodeless spheroidal and torsional shear oscillations restored by hooke s force of elastic shear stresses ( bastrukov et al 2007 , 2008 ) @xmath33,\\ \label{e2.7 } & & { \bf a}_t({\bf r})=a_t\nabla\times [ { \bf r}\chi({\bf r})],\\ \label{e2.8 } & & \chi({\bf r})=r^\ell\,p_\ell(\zeta),\quad \zeta=\cos\theta \end{aligned}\ ] ] where @xmath34 is the legendre polynomial of multipole degree @xmath35 and to accentuate the absence nodes in radial dependence of these fields in the interval @xmath36 $ ] ( from the star center , @xmath37 , to its surface @xmath38 ) , the term node - free vibrations is used .
these fields are basic ingredients of parameters of inertia @xmath39 and stiffness of magneto - mechanical rigidity @xmath40 of the hamiltonian of alfvn oscillations @xmath41 the explicit expressions form of these parameters read @xmath42 where @xmath43,\\ \label{e2.12 } & & b_i({\bf r})=(b_k ( { \bf r})\nabla_k)a_i({\bf r})-(a_k({\bf r})\nabla_k ) b_i({\bf r}),\\ \label{e2.13 } & & a_{ik}({\bf r})=\frac{1}{2}[\nabla_i\,a_k({\bf r})+\nabla_k\,a_i({\bf r } ) ] .
\end{aligned}\ ] ] thus , knowing the density profile @xmath16 and space distribution of equilibrium magnetic field @xmath17 in the vibrating star , the above outlined energy method provides a unified way of computing frequencies @xmath44 of both poloidal and toroidal modes of alfvn node - free oscillations . following this line of argument , in work ( bastrukov et al 2009a ) , the results of which are of particular interest for our discussion here
, it was found that the frequency spectral of poloidal and toroidal @xmath2-modes are given by @xmath45^{1/2},\\ \label{e2.15 } & & \omega(_0a^t_\ell)=\omega_a\left[(\ell^2 - 1)\frac{2\ell+3}{2\ell-1}\right]^{1/2},\\ \label{e2.16 } & & \omega_a=\frac{v_a}{r},\quad v_a=\frac{b}{\sqrt{4\pi\rho}},\\ \label{e2.17 } & & \omega_a = b\sqrt{\frac{r}{3m}},\quad m=\frac{4\pi}{3}\rho r^3 .
\end{aligned}\ ] ] hereafter @xmath46 and @xmath47 stand for the alfvn velocity and frequency , respectively , @xmath48 is the mass and @xmath49 is the radius of a star
. the asymptotic behavior of above spectra , at very high overtones , @xmath50 , is given by @xmath51 and similar asymptotic behavior exhibits frequency spectrum of toroidal alfvn mode in solid star with ferraro s form of non - homogeneous poloidal magnetic field ( bastrukov et 2009b ) .
these spectral formulae provide us information about spacing between overtones of nodeless oscillations which is crucial to identification of @xmath2-modes in spectra of observed qpos in the emission of magnetic white dwarfs . in the reminder of this work we focus on axisymmetric alfvn oscillations in a solid star model with frozen - in toroidal magnetic fields presented which can be relevant to the asteroseismology of non - magnetic white dwarfs .
computation of integral parameters @xmath39 and @xmath40 for both tmf1 and tmf2 solid star models undergoing torsional vibrations about axis of symmetry with toroidal field of differentially rotational displacements @xmath52,\\ \nonumber & & a_r=0,\quad a_\theta=0,\\ \nonumber & & a_{\phi}=a_t\,r^\ell(1-\zeta^2)^{1/2}p'_\ell(\zeta),\quad p'_\ell(\zeta)=\frac{dp_\ell(\zeta)}{d\zeta } \end{aligned}\ ] ] yields @xmath53 the impossibility of sustaining differentially rotational vibrations about axis of symmetry is one of the basic findings of this study that hardly could be foreseen without straightforward computations . from hamiltonian of the eigenfrequency problem in question
it follows that quake - induced perturbation of such displacements should lead to differential rotation with kinetic energy @xmath54 , not vibrations , as is the case of solid star with poloidal magnetic field .
for the former , it worth noting that in the star undergoing axisymmetric spheroidal vibrations under consideration the instantaneous material displacements are described by nodeless poloidal field of instantaneous material displacements which is irrotational @xmath55={\cal a}_p\nabla r^\ell\,p_\ell(\cos\theta),\\ \label{e3.2 } & & { \cal a}_p=\frac{a_p}{\ell+1},\quad \nabla\times { \bf a}^p=0,\\ \nonumber & & a_r={\cal
a}_p\,r^{\ell-1}\ell p_\ell(\zeta),\\ \nonumber & & a_\theta=-{\cal a}_p\,r^{\ell-1}(1-\zeta^2)^{1/2}p'_\ell(\zeta),\quad a_\phi=0 . \end{aligned}\ ] ] the distortions of lines of toroidal magnetic field in the solid star undergoing axisymmetric spheroidal nodeless vibrations are pictured in fig.3 .
the mass parameter of nodeless spheroidal oscillations @xmath39 can conveniently be represented in the form @xmath56
d{\cal v}\\ \label{e3.3 } & = & 4\pi\rho\,{\cal a}_p^2 r^{2\ell+1}\,m_\ell,\quad m_\ell= \frac{\ell}{2\ell+1}\end{aligned}\ ] ] understandably that computed mass parameter has one and the same value for both b(tmf1 ) and b(tmf2 ) models and because the distinction between frequency spectra can only be due to the difference in @xmath40 , we consider computation of this parameter in some details . using computational formulae of above energy method reported in our previous work ( bastrukov et al 2009b ) , one finds that components of instantaneous distortions of magnetic field lines , @xmath57 , are given by @xmath58 the tensor of perturbation - induced magnetic field stresses , @xmath59 $ ] , has only one non - zero component @xmath60\\ \label{e3.5 } & = & \frac{{\cal a}_p\,b^2}{2\pi}\ , \frac{r^{n-2}(r - r)}{r}\ , \ell(1-\zeta^2)\,p_\ell(\zeta ) \end{aligned}\ ] ] so that the integrand of @xmath40 reads @xmath61 .
\end{aligned}\ ] ] integration over spherical polar coordinates @xmath62 and @xmath63 yields @xmath64d{\cal v}=\frac{{\cal a}_p^2\,b^2\,r^{2\ell-1}}{2\ell(2\ell-1 ) } \ell\left[\ell\ , i_1- i_2\right ] , \end{aligned}\ ] ] where @xmath65 and @xmath66 finally , we obtain @xmath67 the frequency spectrum is given by @xmath68 .
\end{aligned}\ ] ] \sin\theta$].,width=302 ] in fig .
4 we plot fractional frequencies and periods as functions of multipole degree of axisymmetric spheroidal alfvn nodeless oscillations of solid star in its own toroidal magnetic field of the tmf1-model .
following the same line of argument , in this case of tmf2-model for instantaneous distortions of magnetic field lines @xmath69 we obtain @xmath70 as in previous case the tensor of @xmath71 has too only one non - zero component @xmath72=-\frac{{\cal a}_p\,b^2}{2\pi}\ , r^{n-2}\ , \zeta\,p'_\ell(\zeta ) \end{aligned}\ ] ] and explicit form of integrand of @xmath40 reads @xmath73 .
\end{aligned}\ ] ] the integral for stiffness taken over @xmath62 and @xmath63 coordinates is given by @xmath74d{\cal v}=\frac{{\cal a}_p^2\,b^2r^{2\ell-1}}{(2\ell-1 ) } \,\left [ i_3- \ell\,i_4\right],\\ \nonumber & & i_3=\int\limits_{-1}^{1 } \zeta^2 \,p'^2_\ell(\zeta ) d\zeta = \frac{\ell\,(\ell+1)\,(2\ell-1)}{(2\ell+1)},\\ \nonumber & & i_4=\int\limits_{-1}^{1}\zeta\,p_\ell(\zeta)\,p'_\ell(\zeta ) d\zeta = \frac{2\,\ell}{(2\ell+1)}\end{aligned}\ ] ] and its resultant expression reads @xmath75 ^{1/2}$].,width=302 ] the frequency spectrum of poloidal alfvn mode in the star with magnetic field of tmf2-model is given by @xmath76 this spectrum is visualized in fig .
it follows from frequency spectra that the lowest overtone is of quadrupole multipole degree , @xmath77 , as should be the case in view of shear character of vibrations at which density remains unaltered .
the excitation of dipole , @xmath78 , displacements results in center - of - mass translation motion of the star , as follows from hamiltonian leads to @xmath79 .
thus , the fact that the toloidal magnetic field vanishes outside ( completely entrapped in the star body ) is crucial for the star capability of maintaining non - radial spheroidal vibrations with nodeless field of displacements in question . ] . for typical of white dwarfs parameters ( the average density of degenerate matter @xmath80 g/@xmath81 , the radius @xmath82 cm and the intensity of intrinsic magnetic field @xmath83 g ) , the basic period of poloidal alfvn mode @xmath84 , where @xmath85 falls in the range @xmath86 sec , that is , in the realm of periods of quasi - periodic oscillations of optical and x - ray emission detected form pulsating non - magnetic white dwarf stars .
the fact that extremely dense metal - like matter of white dwarf stars , composed of ions suspended in the fermi - gas of relativistic electrons , is capable of accommodating magnetic field of very high intensity , suggests that these solid stars could be one of the best objects for studying alfvn oscillations ( lou 1995 ) .
therefore the possibility should not be overlooked that , by not considering such oscillations , we can miss discovering some of important features of the white dwarf asteroseismology . motivated by this line of argument and continuing our investigations of node - free regime of seismic vibrations of degenerate fs
solid stars ( bastrukov et al 2007 - 2009 ) , in this paper focus was laid on search for vibrational peculiarities of node - free alfvn modes in the solid star model with purely toroidal magnetic field entrapped in the star volume .
one of the main purposes was to get a feeling about differences between frequency spectra of global alfvn node - free vibrations in the magnetic white dwarf ( model of solid star endowed with axisymmetric poloidal , homogeneous internal and dipolar , external magnetic field ) and non - magnetic white dwarf ( solid star model with purely toroidal and completely internal magnetic field ) .
what is newly disclosed here is that non - magnetic white dwarf with purely toroidal magnetic field can sustain only non - rotational spheroidal oscillations .
and we found that asymptotic behavior of frequencies of poloidal alfvn modes in a solid star with frozen - in purely azimuthal field is substantially different from that in the con - convective star with homogeneous poloidal magnetic field .
specifically , for b(tmf1)-model , appropriate for a non - magnetic white dwarf , we have found @xmath89^{1/2}\to \frac{1}{\sqrt{4\ell}},\quad \ell>>1.\end{aligned}\ ] ] for a solid star with azimuthal magnetic field of the b(tmf2)-type we found @xmath90^{1/2}\to \sqrt{\ell},\quad \ell>>1.\end{aligned}\ ] ] an absolutely different asymptotic is expected for node - free poloidal @xmath2-mode in a solid star with homogeneous uniform magnetic field , the pmf - model appropriate for magnetic white dwarf @xmath91^{1/2}\to \ell,\quad \ell>>1.\end{aligned}\ ] ] this comparison is visualized in fig.6 .
numerically , the computed periods of @xmath2-modes are in the realm of the qpo periods observed in the optical and x - ray emission of pulsating white dwarfs .
all the above suggests that considered @xmath2-modes of white dwarf vibrations can manifest itself in one line with @xmath4-modes and this feature of the white dwarf asteroseismology worths , therefore , more detailed studying .
we are indebted to referee for valuable suggestions .
this work is a part of investigations on variability of high - energy emission from compact sources supported by nsc of taiwan , grant numbers nsc-96 - 2628-m-007 - 012-my3 and nsc-97 - 2811-m-007 - 003 . | in the context of the white dwarf asteroseismology , we investigate vibrational properties of a non - convective solid star with an axisymmetric purely toroidal intrinsic magnetic field of two different shapes .
focus is laid on regime of node - free global lorentz - force - driven vibrations about symmetry axis at which material displacements have one and the same form as those for nodeless spheroidal and torsional vibrations restored by hooke s force of elastic shear stresses .
particular attention is given to the even - parity poloidal alfvn modes whose frequency spectra are computed in analytic form showing how the purely toroidal magnetic fields completely buried beneath the star surface can manifest itself in seismic vibrations of non - magnetic white dwarfs .
the obtained spectral formulae are discussed in juxtaposition with those for alfvn modes in the solid star model with the poloidal , homogeneous internal and dipolar external , magnetic field whose inferences are relevant to alfvn vibrations in magnetic white dwarfs . |
the long - term goal of this work is to justify the existence and investigate , via computer - aided experiments and suitable mathematical arguments , a prototypical fractal billiard , namely , the koch snowflake billiard . since the koch snowflake curve ( the boundary of the koch snowflake billiard table , see fig .
[ fig : kochconstruction ] ) is nowhere differentiable , it is certainly not clear from the outset that the koch snowflake billiard or , let alone , the associated billiard flow is a legitimate mathematical object of study .
indeed , the first question that springs to mind is `` _ _ how do you reflect a billiard ball off of a nondifferentiable boundary _ _ ? '' on the other hand , because the koch curve is highly symmetric ( indeed , it is _ self - similar _ ) and can be thought of as some kind of ` infinite polygon ' ( see fig .
[ fig : kochconstruction ] ) , it may not be entirely unreasonable to harbor such an expectation . in this paper
, we provide experimental and geometric evidence towards this claim , as well as formulate conjectures and open problems concerning this challenging problem . in a nutshell , the koch snowflake billiard is viewed as a suitable limit of ( rational ) polygonal billiard approximations ; and analogously for the associated billiard ( and geodesic ) flows .
we also identify a variety of ( potential ) periodic and quasiperiodic orbits of the koch snowflake fractal billiard , and discuss some of their properties . in order to present our experimental results and conjectures about the snowflake billiard
, we need to briefly recall several concepts from the theory of dynamical systems ( more specifically , of rational polygonal billiards ) and fractal geometry . a _ polygonal billiard _ @xmath5 is given by a polygon @xmath6 and the bounded region enclosed by @xmath6 . for example , the square billiard is given by the square and the region bounded by the square . when studying a polygonal billiard ,
one is primarily concerned with the orbits of a pointmass moving at unit speed in @xmath5 .
we assume that our billiard ball experiences no friction and perfectly elastic collisions with the smooth portions of the boundary . upon colliding with a smooth portion of the boundary ,
we reflect at the boundary according to snell s law , which says that the angle of incidence equals the angle of reflection ; see , e.g. , @xcite , @xcite , chapter 9 of @xcite , or @xcite .
in general , if the billiard ball collides with a corner of the polygonal billiard table @xmath5 , we must terminate the trajectory because we can not determine reflection at such points .
we call the map that describes the flow on the billiard @xmath6 the _ billiard map_. an important case of a polygonal billiard is a _ rational billiard _ and is defined to be a polygon @xmath6 for which every interior angle is a rational multiple of @xmath7 ; see , e.g. , @xcite , @xcite , @xcite or @xcite .
although a fractal is a shape that lacks many of the appealing qualities of polygons , so long as we are dealing with the right fractal , it is a shape that can be generated by polygons , a fact that is heavily exploited in this paper . the type of fractals we are interested in are self - similar fractal curves in @xmath8 that are nowhere differentiable and have infinite length . specifically , the fractal billiard that we eventually want to define and investigate is the koch snowflake billiard , denoted by @xmath0 ; the boundary of the associated billiard table is the koch snowflake curve , denoted by @xmath1 . the construction of the koch snowflake fractal curve is given in fig .
[ fig : kochconstruction ] .
the _ prefractal _
@xmath2 of @xmath0 is then an @xmath9th level polygonal approximation of @xmath0 .
like many fractals , the koch snowflake curve @xmath1 has a non - integer dimension . in the case of the snowflake ,
that dimension is @xmath10 , which indicates that the curve @xmath1 has infinite length .
( with sides of length 1 ) ; from left to right : @xmath11 , @xmath12 , @xmath13 , @xmath14 .
the process continues ad infinitum , yielding the koch snowflake curve , viewed here as the boundary of the koch snowflake billiard table @xmath0 . for each integer @xmath4
, @xmath2 denotes the @xmath9th prefractal polygonal approximation to @xmath0 .
note that @xmath2 defines a rational billiard because its interior angles are either @xmath15 or @xmath16 . while this property is not required to be a rational billiard , the interior angles of @xmath2 alternate between @xmath15 and @xmath16 .
( depending on the context , @xmath2 also sometimes refers to the polygonal boundary of this rational billiard ; the latter boundary is more correctly denoted by @xmath17 , however . ) ] the koch snowflake is not a rational polygon .
the boundary of the koch snowflake ` fractal billiard ' is nondifferentiable , hence making it very difficult to properly define , and let alone analyze , the billiard map .
however , the behavior of a pointmass may be anything but random , because , as was alluded to just above , @xmath0 has the very special property that there is a sequence of finite polygonal approximations @xmath18 converging to @xmath0 such that for each finite @xmath9 , @xmath2 is a rational billiard ; see fig .
[ fig : kochconstruction ] . in order to overcome the limitations of the koch snowflake
, we may examine the prefractal billiard approximations @xmath2 and attempt to make an argument in support of the existence of periodic orbits of the limiting fractal billiard @xmath0 , based on results concerning the rational polygonal billiards @xmath2 .
therefore , in some sense , the snowflake curve is viewed as an ( infinite ) ` fractal rational polygon ' and the associated billiard table @xmath0 as a ` fractal rational billiard ' . roughly speaking , a compact set @xmath19
is said to be _ self - similar _ if it can be written as a finite union of scaled ( or rather , similar ) copies of itself .
more specifically , this means that @xmath20 is a nonempty compact subset of @xmath8 and there exists a finite collection @xmath21 of contractive similarity transformations of @xmath8 , with @xmath22 , such that @xmath23 .
it then follows that for any nonempty compact subset @xmath24 of @xmath8 , we have that @xmath25 ( in the sense of the hausdorff metric ) , where @xmath26 and @xmath27 denotes the @xmath28th iterate of @xmath29 ; see , e.g. , chapter 9 of @xcite .
for this reason , the self - similar set @xmath20 is called the attractor of the iterated function system @xmath21 .
the koch curve is a perfect example of a self - similar set ; see fig . [
fig : kochcurveconstruction ] . and , in a related context ,
* [ la1,2 ] * , @xcite , @xcite or [ * la - vf * , 12.3.1 & 12.5 ] . ] strictly speaking , the koch snowflake fractal is not self - similar ; specifically , it is the result of pasting together three copies of the koch curve ; see fig .
[ fig:3kochcurves ] . in light of this
, we make a slight abuse of language by referring to @xmath0 ( or rather , its boundary @xmath1 ) as a self - similar fractal . ) . ]
we can exploit the fact that the proposed billiard @xmath0 has a self - similar boundary .
we notice that , in a sense , @xmath2 is constructed from equilateral triangles .
the billiard ball dynamics on the equilateral triangle billiard @xmath30 are well understood .
in particular , the periodic orbits of the equilateral triangle billiard @xmath31 are classified in @xcite .
the resulting theorem , combined with the fact that @xmath2 can be embedded in a tiling @xmath32 of the plane by equilateral triangles with side lengths @xmath33 , will aid us in explaining our experimental results and formulating some of our conjectures .
this paper is structured as follows . in section 2 ,
we discuss the necessary theory of rational billiards so that we may better present our results on the prefractal polygonal billiards @xmath2 ; see fig .
[ fig : kochconstruction ] . in
[ subsec : unfolding ] , we discuss a useful tool for discerning periodic orbits of rational billiards , namely , the unfolding of billiard trajectories . in general , such an unfolding gives rise to a surface that depends on the orbit , but we also state known results for constructing a surface @xmath34 which is independent of the periodic orbit of the rational billiard @xmath6 .
we recall in [ subsec : surface ] that the flow on the rational billiard @xmath6 corresponds to the geodesic flow on the corresponding surface @xmath34 .
in fact , as is explained in [ subsec : singularities ] , vertices constitute singularities of the billiard map and certain copies of the vertices in @xmath34 constitute singularities of the flow on the surface .
we discuss what implications this has for the billiard ball dynamics .
in [ subsec : ksnratbilliard ] , we view the prefractal polygonal billiard @xmath2 as a rational billiard , and discuss the consequences of this fact for the billiard flow at the vertices of @xmath2 .
we also construct the corresponding surface for @xmath12 , @xmath13 and @xmath14 and indicate the singularities of each surface .
we show that , in agreement with the general theory of rational billiards , reflection at acute angles can be defined , but that reflection at obtuse angles can not be determined unambiguously .
in the last part of section 2 , namely [ subsec : baum ] , we discuss the results of @xcite on the equilateral triangle billiard and their implications for the billiard @xmath2 . in doing so ,
we explain how @xmath2 can be embedded in a tiling @xmath32 of the plane by equilateral triangles with side lengths @xmath33 ( as was alluded to above ) and give an equivalence relation on the collection of all periodic orbits of the rational billiard @xmath31 .
we do not seek to generalize the results of @xcite to @xmath2 but instead use these results to provide us with a list of appropriately adjusted initial conditions for testing in @xmath2 . in section 3 ,
we present and discuss our experimental results on the prefractal billiard @xmath2 by raising and sometimes answering a variety of research questions . in general , we examine the behavior of what we call _ induced _ orbits of @xmath2 .
an induced orbit is one for which the initial condition was appropriately adapted from an initial condition in the equilateral triangle billiard @xmath31 . in [ subsec : gammanorbits ] , we examine the behavior of periodic orbits of @xmath2 in the collection @xmath35 of billiard paths determined by the initial conditions @xmath36 , where @xmath37 is a midpoint of a side of @xmath2 and @xmath15 is the initial angle of the periodic orbit at the starting point @xmath37 .
we attempt to organize the orbits in @xmath35 according to their geometrical or dynamical behavior , and provide definitions for what we call _ fagnano orbits _ , _ piecewise fagnano orbits _ and _ primary piecewise fagnano orbits _ of @xmath2 .
while these are very interesting orbits , the elements in @xmath35 do not constitute all of the induced orbits of @xmath2 .
in [ subsec : nongammanorbits ] , we examine the behavior of periodic orbits induced by the initial condition @xmath38 , where @xmath39 or @xmath40 and the orbit remains nonsingular , meaning that it does not hit any vertices of the billiard table .
a particularly interesting example of such a periodic orbit is given by @xmath41 , where @xmath42 is a suitable value ( or vector ) used to perturb the initial basepoint lying at the midpoint @xmath37 of a side of @xmath2 .
because of the nature of the equilateral triangle billiard , the reflection at the vertices of @xmath43 can be determined .
consequently , in [ subsec : singularorbits ] , these singular ( and periodic ) orbits are then used to induce singular orbits of @xmath2 , for some @xmath4 .
finally , in [ subsec : quasiperiodic ] , we close section 3 by discussing a seemingly uninteresting type of orbit , which we call a _ quasiperiodic orbit _ of @xmath2 . such an orbit can be considered as a ` rational approximation ' to a given periodic orbit .
for example , if @xmath38 is an initial condition of a periodic orbit , then a corresponding quasiperiodic orbit would have an initial condition @xmath44 , where @xmath45 is a rational approximation of @xmath46 ( obtained via a continued fraction expansion of @xmath46 ) .
because the boundary of the billiard @xmath2 is changing as @xmath9 increases , the study of quasiperiodic orbits of @xmath2 should enable us to define a suitable notion of ` quasiperiodic orbit ' of @xmath0 .
in section 4 , we conclude our paper by stating several conjectures based on our experimental results and provide a list of open problems and conjectures related to the proposed fractal billiard @xmath0 and a particular collection of its periodic orbits . in particular , we conjecture the existence of what we call the _ primary piecewise fagnano orbit _ of @xmath0 , which we propose would be a suitable limit of primary piecewise fagnano orbits of the prefractal approximations @xmath2 . in the special case of @xmath47 ,
the ` primary piecewise fagnano ' orbit of @xmath0 , of @xmath0 is induced by the fagnano orbit of the original triangle @xmath43 ( i.e. , the shortest periodic orbit of @xmath31 ) , along with its appropriate counterpart in each billiard table @xmath2 ; see [ subsec : gammanorbits ] , along with conjectures [ conj : weakppf ] and [ conj : existenceofselfsimilarperiodicorbits ] of section 4 .
] we also conjecture that its ` footprint ' consisting of all incidence points of the periodic orbit @xmath47 of @xmath0 . ] on the boundary @xmath1 is a self - similar subset of @xmath48 ; more specifically , it is the natural counterpart of the middle - third cantor set strung around the koch snowflake curve @xmath1 .
the other conjectures and open problems stated in section 4 are clearly of a longer term nature .
they concern , for example , the existence of the koch snowflake billiard @xmath0 ( as a proper mathematical object ) and of the associated billiard flow , and of the corresponding geodesic flow ( conjecturally equivalent to the billiard flow on @xmath0 ) ; see [ subsec : surface ] and [ subsec : ksnratbilliard ] , along with parts ( i ) and ( iii ) of conjecture [ conj : fractalbilliardfractalsurface ] .
] or ( in the very long - term ) the relationship between the length spectrum of the elusive koch snowflake fractal billiard and the frequency spectrum of the corresponding koch snowflake drum ( e.g. , [ * la12*,*lanrg*,*lapa * ] , along with 12.3 and 12.5 of @xcite ) .. ] in closing this introduction , we mention that the subject of billiards and particularly , polygonal billiards or even , rational billiards is an active area of research in the field of dynamical systems .
books , survey articles and research papers on various aspects of this topic include , respectively , [ * kaha*,*ta * ] , [ * gastvo*,*gu*,*husc*,*mata*,*sm*,*vo*,*zo * ] and [ * baum*,*guju12*,*kaze*,*ma*,*ve13 * ] . to our knowledge , none of these references deals with the elusive case of ` fractal billiards ' .
it is certainly true , however , that these works ( and the many relevant references therein ) have played a key role in motivating and guiding our investigations on the ` koch snowflake billiard ' ( and its rational polygonal approximations ) reported in the present paper .
in this section , we want to discuss the properties of the billiard flow on @xmath2 and describe how @xmath2 can be viewed naturally as a _ rational billiard_.
a _ rational polygon _ is a polygon with interior angles that are rational multiples of @xmath7 .
if the boundary of a polygonal billiard is a rational polygon , then we call the associated billiard a _ rational billiard_. for us to discuss the flow on the associated closed surface , we need to define a few concepts . if @xmath49 is a regular @xmath50-gon , where @xmath50 is a positive integer , then @xmath51 is the group of symmetries of @xmath49 and has cardinality @xmath52 .
it is called the _ dihedral group _ and is generated by the reflections in the lines through the origin that meet at angles @xmath53 .
let @xmath54 be the group of planar motions generated by the reflections in the sides of a polygon @xmath6 .
furthermore , let @xmath55 denote the subgroup of the orthogonal group @xmath56 consisting of the linear parts of the elements of @xmath54 .
when the polygon @xmath6 has a connected boundary , then we can give an alternate characterization of it being rational .
let @xmath6 be a polygon with connected boundary .
then @xmath6 is a rational billiard if and only if the group @xmath55 is finite . in that case , if the interior angles of the rational polygon @xmath6 are written in the form @xmath57 for @xmath58 , where @xmath59 , @xmath60 are coprime positive integers and @xmath61 is the number of vertices of @xmath6 , then @xmath55 is isomorphic to the dihedral group @xmath51 , where @xmath50 is the least common multiple of @xmath62 .
[ prop : altdefinitionofratpolygon ] if @xmath6 is a rational billiard , then we have at our disposal a method for discerning periodic orbits .
this method was first given in @xcite and reiterated in various forms in the literature ; see , e.g. , the survey articles @xcite , @xcite , @xcite and @xcite .
let @xmath63 be the initial position of the billiard ball and @xmath46 be the initial direction of the billiard ball .
let @xmath64 be the point on the side @xmath65 of the billiard table at which the billiard ball collides , and @xmath66 be the angle at which the billiard ball reflects off of @xmath65 .
if the orbit @xmath67 is periodic , then we want to illustrate this by ` unfolding ' the billiard table and its contents . to unfold the billiard table and its contents , reflect the contents of @xmath5 in the side @xmath65 .
this results in a segment continuing in the direction of @xmath46 .
moreover , we see that we have created a situation where the trajectory passes through @xmath65 and continues on to a segment that is collinear with the initial segment of the trajectory ( see fig . [
fig : unfoldtheorbit ] for an example in the equilateral triangle billiard ) . continuing this process , we see that we can unfold a billiard table and its contents .
when the orbit is a collection of line segments , such an unfolding will be a straight line .
furthermore , the length of the unfolding is exactly the length of the orbit in @xmath5 .
when the orbit unfolds and terminates on an edge of a polygon @xmath69 in the unfolding such that the terminal point is a copy of the initial point and the segment makes an angle with the side of @xmath69 equal to the initial angle , then the original orbit of @xmath6 is periodic .
it is significant that we can unfold an orbit to determine the periodicity of the said orbit . in order to construct a closed surface out of the unfolding ( which is determined by the periodic orbit ) , we identify sides of the unfolding based on where the ball collides . in particular , we identify the terminal side with the initial side since the orientation of @xmath69 and the original polygon @xmath6 are the same .
this results in a closed surface whose geodesic flow corresponds to the billiard flow on @xmath6 .
we can see this clearly in fig .
[ fig : unfoldtheorbit ] . however , this closed surface depends on the initial position and angle .
we want to construct a surface @xmath70 that is independent of these initial conditions .
in addition , the geodesic flow on that surface @xmath34 will turn out to be equivalent to the original billiard flow on @xmath6 .
consider the product @xmath71 .
essentially , @xmath71 contains @xmath52 copies of our rational billiard table @xmath6 .
a graphical representation of @xmath71 can be given by the following .
if we fixed a vertex of a polygon @xmath6 such that reflection in the adjacent sides of @xmath6 generates @xmath52 copies of @xmath6 , then the resulting shape is what is called a _
generalized polygon_. if two sides of the generalized polygon are translates of each other and/or the result of a rotation by @xmath7 , then we consider these two sides to be equivalent .
then , upon modding out by this equivalence relation , we obtain a closed surface @xmath72 that does not depend on any particular orbit .
consider the triangular billiard @xmath6 with boundary given by the triangle with interior angles @xmath73 .
we reflect @xmath6 in the sides emanating from the vertex with angle @xmath74 .
then , as is expected , we get a surface with 16 copies of the billiard @xmath6 ; indeed , @xmath75 and so @xmath76 .
therefore , the associated generalized polygon is an octagon .
two sides are equivalent if they are translates of each other . modding out by this equivalence relation
, we have obtained the associated flat surface @xmath34 as shown in fig .
[ fig : octagonsurface ] .
note , however , that since its genus is greater than @xmath77 , this surface can not be properly embedded in the plane , but that for convenience , we represent @xmath78 as though it could be done ; see fig .
[ fig : octagonsurface ] .
[ exm : triangle ] corresponding to the triangle @xmath79 ; the associated generalized polygon is an octagon .
pairs of appropriate sides should be identified as indicated , in order to obtain @xmath34 , a surface of genus @xmath80 .
note , however , that strictly speaking , this planar representation is not valid .
we illustrate such an embedding here as a heuristic device . ]
the surface @xmath34 is often referred to as a ` translation surface ' or a ` flat surface ' in the literature ; see , e.g. , @xcite , @xcite , @xcite , @xcite and @xcite .
indeed , the geodesic flow on @xmath34 is nothing but the straight - line flow in the associated generalized polygon , with pairs of opposite sides identified as indicated above .
the key fact concerning the geodesic flow on @xmath34 is that it is ( dynamically ) equivalent to the billiard flow on @xmath6 . as mentioned before , the billiard map is only well defined at certain types of vertices of the rational polygonal billiard .
we demonstrate under which conditions the billiard flow can be defined at vertices .
if @xmath6 is a polygon , then the corresponding surface contains copies of the vertices of @xmath6 .
these copies of the vertices are then considered singularities of the flow on the surface and are called _
conic singularities_. in order to understand what a conic singularity is , we must first introduce the notion of a conic angle .
conic angle _ is an angle that measures the radians required to form a closed loop about the origin . in the plane , the conic angle is @xmath82 .
we can form spaces in which the conic angle is not @xmath82 .
suppose we had a space in which the negative upper half plane was glued to the negative lower half plane and the positive lower half plane was glued to the positive upper half plane of a second copy of @xmath8 .
suppose we make similar identifications but instead glue the positive lower half plane to the positive upper half plane of the original copy of @xmath8 .
then we have formed a space in which the conic angle about the origin is @xmath83 .
consequently , there are two types of conic singularities : removable and nonremovable . a conic singularity is called _ removable _ if its conic angle is @xmath82 .
the conic singularity is called _ nonremovable _ if its conic angle is @xmath84 , for some integer @xmath85 . in the context of rational polygonal billiards , we determine the conic singularity of a vertex as follows .
geometrically , if @xmath86 is a vertex of a rational billiard corresponding to a nonremovable singularity , when we sweep out an angle of @xmath82 by continually reflecting the rational billiard in a side , the orientation is not preserved . in fig .
[ fig : weirdgeometry ] , we see that a vertex with an associated conic angle of @xmath83 must sweep out an angle of @xmath83 before returning to a copy with the same orientation .
we calculate the conic angle of a conic singularity as follows .
if , for @xmath87 , @xmath88 is an interior angle of the rational billiard @xmath6 , as in proposition [ prop : altdefinitionofratpolygon ] , then the corresponding conic angle is @xmath89 . in the triangle above
has a conic angle of @xmath83 .
consequently , the only way to recover a correctly oriented triangle by way of reflecting through sides adjacent to this vertex is to sweep out an angle of @xmath83 . ]
consequently , the geodesic flow on @xmath34 does not see removable singularities and continues on unimpeded , but is immediately deterred from progressing across the surface when the geodesic intersects a nonremovable singularity .
the geodesic flow on @xmath34 can not be determined at such a singularity , implying that the billiard flow on @xmath6 at the associated vertex can not be determined either . and the geodesic flow on @xmath34 .
] consider again the triangle @xmath79 , as was done in example [ exm : triangle ] . from the above formula
, the conic angle of the conic singularity corresponding to the vertex with angle @xmath90 is @xmath91 .
consequently , this conic singularity is of the nonremovable type .
this means that the flow on the surface @xmath34 can not be determined through the corresponding copy of the vertex on @xmath34 .
moreover , we can not logically define reflection at this vertex .
the vertex at the center of our surface , however , is of the removable type ; the corresponding conic angle is @xmath92 ; see fig . [
fig : goodflowonoctagon ] . [
cols="^,^,^ " , ]
we propose here several conjectures and open problems regarding the billiard @xmath0 .
as we have done in the previous sections , we begin by first defining a few necessary terms .
our ability to investigate the proposed billiard @xmath0 has been limited by the finiteness of the computer .
hence , our conjectures deal mostly with `` _ _ what happens in the limit _ _ '' .
however , it is not exactly clear what we mean by _ limit_. hence , when we say ` limit ' , we are assuming a suitable notion of limit , most likely ( depending on the context ) an inverse limit of some inverse system or the gromov
hausdorff limit of a sequence of complete metric spaces ; see , respectively , @xcite or @xcite for an introduction to these notions .
once a suitable notion of limit has been defined , we want to consider the ( admittedly , possibly empty ) collection of periodic orbits of the billiard @xmath0 , which we denote by @xmath93 . however , our focus thus far has been on induced orbits of @xmath2 .
we define @xmath94 to be the collection of induced periodic orbits of @xmath0 , and phrase our conjectures in terms of elements in @xmath94 . as in section
[ sec : expresults ] , where we discussed the ` anatomy ' of @xmath2 , we now discuss the anatomy of the proposed billiard @xmath0 and the collection of induced orbits @xmath94 .
we define the _ _ footprint _ _ of an orbit to be the points of the orbit that belong to the boundary of the billiard @xmath0 .
we call the _ ghost set _ of @xmath0 the collection @xmath95 , the union of all ghost sets @xmath96 associated with @xmath2 ( see definition [ def : anatomyofksn](i ) ) . [
def : selfsimorbit ] let @xmath67 be a periodic orbit of the proposed billiard @xmath0 .
then , @xmath67 is said to be a _ self - similar orbit _
if its footprint is a self - similar subset of @xmath99.here and thereafter , when talking about an ifs or a self - similar set , we are making the same slight abuse of language as in [ sec : introduction ] when referring to the koch snowflake curve @xmath1 as being ` self - similar ' , whereas strictly speaking , @xmath1 is obtained by fitting together three copies of a self - similar set , namely , the von koch curve ; see figs . [
fig : kochcurveconstruction ] and [ fig:3kochcurves ] . ]
our first conjecture asserts the existence of a very special periodic orbit of the proposed billiard @xmath0 .
we denote the ( presumed ) ` limit ' of the orbits @xmath101 by @xmath47 , and call this the _ primary piecewise fagnano orbit _ of @xmath0 .
we can see from fig .
[ fig : ifsconstruction ] that there is an iterated function system ( ifs ) , , this ` ifs ' is given by three different ifs s , denoted by @xmath102 , @xmath103 , @xmath104 and each comprised of two contractive similarity transformations of @xmath8 ; see the caption of fig .
[ fig : ifsconstruction ] .
the associated attractor is therefore a ` self - similar set ' of @xmath8 ( also in the spirit of footnote [ foot : abuseoflanguage ] ) . ]
denoted by @xmath20 , which is producing scaled copies of pieces of the orbit , and such that for any @xmath4 , @xmath105 .
observe that for each finite @xmath9 , the chronology of the orbit @xmath101 can be easily recovered . furthermore , if we consider the fixed point attractor of the ifs , then this set @xmath107 should correspond to the footprint of the proposed periodic orbit @xmath47 of @xmath0 .
is a self - similar subset of @xmath48 . ]
defining the orbit @xmath47 to be @xmath107 is not very satisfactory , however , because such a definition does not provide useful information on how to recapture any sense of chronological order .
accordingly , we make the following formal conjecture .
[ conj : weakppf ] under a suitable notion of limit , we conjecture that the chronology of @xmath47 can be naturally realized and hence , that @xmath47 can be considered as a true periodic orbit of the koch snowflake billiard @xmath0 . by means of an iterated function system ( ifs ) : the ifs produces the orbit @xmath101 of @xmath2 by first contracting , rotating and translating the fagnano orbit of @xmath31 , so that what results are three copies of @xmath108 appropriately appended to @xmath108 .
in actuality , there are three ifs s ( @xmath109 ) acting in unison to produce this picture .
specifically , @xmath102 acts on the upper left scaled copy of @xmath108 to produce two scaled , rotated and translated copies of that piece of @xmath110 .
likewise , @xmath103 acts on the upper right copy of @xmath108 to produce two scaled , rotated and translated copies of that piece of the orbit @xmath110 .
finally , @xmath104 behaves similarly on the bottom scaled copy of @xmath108 .
the union of all three images unioned with @xmath110 is then the orbit @xmath111 .
continuing in this manner , we can produce @xmath101 , for every @xmath4 .
] significant analytical evidence in support of @xmath47 being a well - defined orbit of the billiard @xmath0 is the fact that ( i ) @xmath47 is a self - similar orbit ( specifically , its footprint appears to be a self - similar cantor set ) and ( ii ) has finite length .
once we have established the existence of the periodic orbit @xmath47 , we expect that demonstrating the existence of other periodic orbits given as ` limits ' of compatible sequences of induced periodic orbits @xmath112 may not be an unreasonable endeavor .
specifically , we conjecture that orbits emanating from the boundary of the proposed billiard @xmath0 at an angle of @xmath15 can be realized by a suitable generalization of the procedure ( yet to be precisely determined ) that would substantiate our claim that @xmath47 forms a periodic orbit of the koch snowflake billiard .
our next conjecture makes specific conjecture [ conj : weakppf ] and the comments surrounding it .
it also discusses some of the geometric properties of the presumed ` piecewise fagnano orbits ' of @xmath0 , including the ` primary piecewise fagnano orbit ' @xmath47 , the existence of which was asserted in conjecture [ conj : weakppf ] . 1 . _
( the primary piecewise fagnano orbit of @xmath0)_. the periodic orbit @xmath47 of @xmath0 can be realized _ _
( _ _ and its chronology restored _ _ ) _ _ as a suitable limit _
_ ( _ _ possibly , an inverse limit _ _ ) _ _ of the compatible sequence of primary piecewise fagnano orbits @xmath101 of @xmath2 .
furthermore , the periodic orbit @xmath47 is a self - similar orbit _ _ ( _ _ in the sense of definition [ def : selfsimorbit ] _ ) _ ; more specifically , its footprint is the natural self - similar middle - third cantor set , strung around the boundary @xmath1 .
( piecewise fagnano orbits of @xmath0)_. more generally , an arbitrary ` piecewise fagnano periodic orbit ' @xmath67 of @xmath0 can be defined and realized as follows .
there exists a compatible sequence @xmath113 of periodic orbits such that for each @xmath4 , @xmath114 belongs to @xmath115 , and @xmath113 converges _ _ ( _ _ in a suitable sense _ _ ) _ _ to @xmath67 .
that @xmath115 stands for the collection of piecewise fagnano periodic orbits of @xmath2 . ] furthermore , @xmath67 is a self - similar orbit of @xmath0 . on the base of @xmath43 ,
does there exist @xmath63 _ _ ( _ _ other than the midpoint @xmath37 _ _ ) _ _ such that the sequence @xmath113 of compatible non-@xmath35 periodic orbits induced by the initial condition @xmath116 will _ _ ( _ _ i _ ) _ avoid all obtuse angle vertices of @xmath2 , for every @xmath4 and _ _ ( _ _ ii _ ) _ converge to a periodic orbit @xmath67 of @xmath0 ?
moreover , we conjecture that if the initial angle @xmath46 of the trajectory is not @xmath15 , but such that @xmath38 would have induced a nonsingular non-@xmath35 periodic orbit @xmath114 of @xmath2 for all @xmath4 , then the associated compatible sequence @xmath113 ` converges ' to a periodic orbit @xmath67 of @xmath0 ; and correspondingly , the associated sequences of footprints of @xmath113 converges to the footprint of @xmath67 .
in addition , we conjecture that the footprint of @xmath67 is a topological cantor set .. ] a primary candidate for illustrating this conjecture is the periodic orbit induced by the initial condition @xmath117 , for some small perturbation @xmath42 ; see fig .
[ fig : kspi6shifted ] .
a key step towards placing the above conjectures and open problems in a broader context and on firmer mathematical grounds must involve the theory of flat surfaces associated with rational billiards , along with its eventual extension to surfaces of infinite genus ( called here ` fractal flat surfaces ' ) presumably associated with fractal billiards .
of a rational polygonal billiard @xmath6 always has a finite genus .
moreover , a simple calculation based on the known formula for @xmath118 ( see , e.g. , lemma 1.2 , p. 1022 of @xcite ) shows that @xmath119 as @xmath120 .
hence , it is natural to expect that the presumed limiting surface @xmath121 has infinite genus ; see conjecture [ conj : fractalbilliardfractalsurface ] .
] ultimately , we are conjecturing the existence of a suitably defined billiard table @xmath0 .
this amounts to establishing that ( i ) there is a corresponding flat surface @xmath121 and ( ii ) the billiard flow on @xmath0 is equivalent to the billiard flow on @xmath121 . in particular , we conjecture that the ` limit ' of the sequence of prefractal billiard tables @xmath18 would correspond to this billiard @xmath0 , and that similarly , the associated surface @xmath121 would be the ` limit ' of the sequence of prefractal flat surfaces @xmath122 .
if we can prove that there are removable singularities in the proposed surface @xmath121 , then it is not unreasonable to expect the conjecture to be true .
moreover , if the wonderful symmetry of the koch snowflake curve @xmath1 can be exploited so as to provide us with a way of dealing with the ambiguity the billiard ball experiences at non - removable singularities ( see , e.g. , the caption of fig .
[ fig : gammanatobtuseangles ] ) , then we would be further justified in believing the conjecture to be a plausible statement .
we summarize the main aspects of this central conjecture and clearly , long - term open problem in the following more specific form .
naturally , part of the difficulty in dealing with this problem will involve finding the appropriate notions of limits involved in the formulation of the conjecture . 1 . _
( the fractal flat surface @xmath121 , along with the geodesic flow)_. the sequence @xmath122 of prefractal flat surfaces associated with the rational billiard @xmath2 converges _ _ ( _ _ in the gromov
hausdorff sense , see @xcite _ _ ) _ _ to a surface of infinite genus @xmath121 , called the ` fractal flat surface ' associated with @xmath0 .
correspondingly , the ` geodesic flow ' on @xmath121 can be realized as a suitable limit _ _ ( _ _ possibly , an inverse limit _ _ ) _ _ of the geodesic flows on the surfaces @xmath3 .
( the koch snowflake fractal billiard @xmath0 , along with its billiard flow)_. the fractal billiard @xmath0 can be defined as a suitable limit _ _ ( _ _ possibly , an inverse limit _ _ ) _ _ of the prefractal polygonal rational billiards @xmath2 ; essentially , , the billiard tables @xmath2 converge in the sense of the hausdorff metric ( or , equivalently , in the gromov
hausdorff sense ) to the koch snowflake billiard table @xmath0 .
hence , the real issue concerns here the associated billiard flows . ]
this means that the billiard flow on @xmath0 can be realized as a suitable limit _ _ ( _ _ also possibly , an inverse limit _ _ ) _ _ of the corresponding billiard flows on the billiard tables @xmath2 .
( geodesic vs. billiard flow)_. finally , the geodesic flow on the fractal flat surface @xmath121 is equivalent to _ _ ( _ _ and , at first , may be used as a suitable substitute for _ _ ) _ _ the billiard flow on the koch snowflake billiard @xmath0 .
we close this paper by stating the following very long - term problem , which is directly motivated by the questions raised in @xcite concerning the relationship between ` fractal drums ' and ` fractal billiards ' , including the koch snowflake drum and billiard .
see , in particular , @xcite , conjecture 6 , p. 198
, itself motivated by conjectures 2 and 3 , pp .
159 and 163164 , respectively ; see also 12.5.3 of @xcite . for information regarding the koch snowflake drum and other ` fractal drums ' ( viewed as ` drums with fractal boundary ' ) ,
see , e.g. , [ * la12*,*lanrg*,*lapa * ] , [ * la - vf*,12.3 & 12.5 ] , and the relevant references therein . for information regarding trace formulas ( including the gutzwiller and chazarain trace formulas ) in various contexts ,
we refer , e.g. , to * [ gz1,2 ] * , @xcite , @xcite and @xcite .
[ openproblem : fractalbilliardvsfractaldrum ] once the existence of the koch snowflake billiard @xmath0 has been firmly established _
_ ( _ _ as hypothesized in conjecture [ conj : fractalbilliardfractalsurface ] _ ) _ , can one formulate , and eventually establish , a suitable ` fractal trace formula ' in this context ?
presumably , the latter would be a fractal counterpart of the gutzwiller , chazarain and the selberg trace formulas in this context , connecting the length spectrum of the snowflake billiard ( i.e. , essentially equivalently , of the geodesic flow of @xmath121 ; see part ( iii ) of conjecture [ conj : fractalbilliardfractalsurface ] ) . ] and the eigenvalue _ _ ( _ _ or frequency _ _ ) _ _ spectrum of the corresponding snowflake fractal drum .
moreover , can one address the same problem for other fractal billiards _ _ ( _ _ once they have been properly defined _ _ ) _ _ and the associated fractal drums ?
we hope that the study of such open problems and conjectures , aided by a suitable combination of computer experiments and theoretical investigations , will enable us in the future to better understand the elusive nature of the koch snowflake billiard and , eventually , of a variety of other fractal billiards .
_ added note . _ for a version of the paper with crisper pictures , please email the second author .
such a file will , however , be considerably larger and require between thirty and sixty minutes to download over a standard 56kbps connection .
m. l. lapidus , vibrations of fractal drums , the riemann hypothesis , waves in fractal media , and the weyl berry conjecture , in : _ ordinary and partial differential equations _ ( b. d. sleeman and r. j. jarvis , eds . ) , vol .
iv , proc .
twelfth internat .
( dundee , scotland , uk , june 1992 ) , pitman research notes in math .
series , vol .
289 , longman , scientific and technical , london , 1993 , pp .
126209 .
m. l. lapidus , j. w. neuberger , r. j. renka and c. a. griffith , snowflake harmonics and computer graphics : numerical computation of spectra on fractal domains , _ internat .
j. bifurcation & chaos _ * 6 * ( 1996 ) , 11851210 . m. l. lapidus and m. van frankenhuijsen , _ fractal geometry , complex dimensions and zeta functions : geometry and spectra of fractal strings _ , springer monographs in mathematics , springer - verlag , new york , 2006 .
( second revised and enlarged edition to appear in 2010 . )
j. smillie , dynamics of billiard flow in rational polygons , in : _ dynamical systems _ , encyclopedia of math .
sciences , vol .
100 , math .
physics 1 ( ya .
g. sinai , ed . ) , springer - verlag , new york , 2000 , pp . | in this paper , we attempt to define and understand the orbits of the koch snowflake fractal billiard @xmath0 .
this is a priori a very difficult problem because @xmath1 , the snowflake curve boundary of @xmath0 , is nowhere differentiable , making it impossible to apply the usual law of reflection at any point of the boundary of the billiard table .
consequently , we view the prefractal billiards @xmath2 ( naturally approximating @xmath0 from the inside ) as rational polygonal billiards and examine the corresponding flat surfaces of @xmath2 , denoted by @xmath3 . in order to develop a clearer picture of what may possibly be happening on the billiard @xmath0
, we simulate billiard trajectories on @xmath2 ( at first , for a fixed @xmath4 ) .
such computer experiments provide us with a wealth of questions and lead us to formulate conjectures about the existence and the geometric properties of periodic orbits of @xmath0 and detail a possible plan on how to prove such conjectures . |
we are grateful to v.serbo and a.tarasov for useful discussions and interest to the work.we acknowledges the warm hospitality at ect , trento.(e.k ) is grateful to the institute of physics sas ( bratislava ) , where part of this work was done.the work is supported by intas grants 30494,00366 .
let us calculate the total cross section of the lepton pair production off the nuclear coulomb field by virtual photons i.e. the process ^*(k)+ze^+(q_+)+e^-(q_-)+z this allow us to compare the results obtained in our approach with the well known one and show how the integration over phase space can be carried out.for our purposes it is sufficient to calculate the total cross section of this process in the born approximation.the generalization to all orders in @xmath34 using our approach is an easy task .
+ using the same approximations as before , the total cross section of the process ( a.1 ) in the born approximation read @xmath58 dzd^2q_+d^2q;\\ r_s & = \frac{1}{\vec q_+^2+\mu^2}-\frac{1}{\vec q_-^2+\mu^2};\vec r_v=\frac { \vec q_+}{\vec q_+^2+\mu^2}-\frac{\vec q_-}{\vec q_-^2+\mu^2};\nn \vec q & = \vec q_++\vec q_-;\mu^2=m^2+z(1-z)q^2;q_l=\frac{\mu^2+\vec q_+^2 } { 2z(1-z)\omega}.\end{aligned}\ ] ] here @xmath59 are the mass and energy of virtual photons .
the integration over transverse momenta @xmath60 can be done analyticaly.we divide the integration over the two - dimensional transfer momentum q in the two regions : 1)small momentum transfer where the longitudinal momentum @xmath61 has to be taken into account @xmath62 and 2)the region where one can safely ignore it , i.e. , @xmath63 .
+ using this trick the integration over transverse momenta @xmath64 can be carried out.as a result , the total cross section for the lepton pair production by transverse and longitudinal photons in the coulomb field can be represented as _
t&= & ; + _ s&=&q^2 .
the total cross section for the pair production by transversally polarized photons @xmath65 turns out to the bethe - maximon result ( 3 ) at @xmath66=0 .
moreover , the expression ( a.4 ) is in agreement with ( 45 ) from @xcite.as to the pair production by longitudinal photons expressions ( 46,59 ) in @xcite have a misprint : the combination @xmath67 enters them in the first power in contradiction with our result ( a.5 ) .
99 l.d.landau and e.m.lifshitz,phys.z.sowjet.6 ( 1934 ) 244 g. racah , nuovo cim . 14
( 1937 ) 93 .
h. a. bethe and l. maximon , phys .
93 ( 1954 ) 788 .
d.yu . ivanov and k. melnikov ,
d57 ( 1998 ) 4025 .
b.z.kopeliovich,a.v.tarasov,o.o.voskresenskaya,eur.phys.j.a11 ( 2001 ) 345 j.m.bjorken,j.b.kogut,d.e.soper,phys.rev .
d3 ( 1971 ) 1382 a.j.baltz and l.mclerran,phys.rev .
c 58 ( 1998 ) 1679 a.j.baltz,f.gelis,l.mclerran,a.peshier,nucl.phys.a695 ( 2001 ) 395 u.eichmann,j.reinhardt,s.schramm,w.greiner,phys.rev.a59 ( 1999 ) 1223 d.yu .
ivanov , a. schiller , v.g .
serbo , phys.lett.b454 ( 1999 ) 155 .
r.n.lee and a.i.milstein,phys.rev .
a64 ( 2001 ) 032106 e. barto , s. r. gevorkyan , e. a. kuraev and n.n.nikolaev , phys.rev.a66 ( 2002 ) 042720 e.barto,s.r.gevorkyan,e.a.kuraev and n.n.nikolaev , phys .
b538 ( 2002 ) 45 .
i.f.ginzburg,d.yu.ivanov,nucl.phys . b388
( 1998 ) 376 .
i.f.ginzburg,s.l.panfil and v.g.serbo , nucl.phys.b284 ( 1987 ) 685 .
d.yu.ivanov,e .
a. kuraev , v. g. serbo and a. schiller , phys.lett .
b 442 ( 1998 ) 453 e. a. kuraev and l. n. lipatov , yad . fiz . 16
( 1972 ) 1060 .
v.m.budnev,i.f.ginzburg,g.v.meledin,v.g.serbo,phys.rep.c15 ( 1975 ) 181 a.i.akhiezer,v.b.berestetski,quantum electrodynamics , nauka , moscow,1981 | the problem of summation of the perturbation series for the process of lepton pair production in relativistic ion collisions is investigated .
we show that the amplitude of this process can be obtained in the compact analytical form , if one confines to the terms growing as a powers of logarithm energy in the cross section , the approximation of which is completely justified for such colliders as rhic and lhc .
+ using this result we calculate the coulomb corrections to the cross section of the process under consideration .
@xmath0 _ joint institute for nuclear research , 141980 , dubna , russia + @xmath1 _ yerevan physics institute , 375036 , yerevan , armenia _
_ the process of lepton pair production in the collision of charged relativistic particles was investigated many years ago @xcite in the lowest order in the fine structure constant @xmath2 . for the total cross section of the process of lepton pair production in collision of two relativistic particles with charge numbers
@xmath3 z_1+z_2e^+e^-+z_1+z_2 , the famous formula was obtained in 1937 by racah @xcite , which in contemporary symbols reads _
b=[l^3 - 2.2l^2 + 3.84l-1.636],l=(_1_2),_i=,with m electron mass,@xmath4 the center of mass energies and masses of colliding particles .
+ for relativistic heavy ion collisions this cross section is huge and the problem of coulomb corrections ( higher order photon exchanges between the produced lepton pair and the coulomb fields of colliding ions ) becomes relevant .
+ the coulomb corrections ( cc ) in the case of lepton pair photoproduction off the nucleus coulomb field was obtained in @xcite.it was shown that the total cross section for the process of lepton pair photoproduction in the coulomb field of the nucleus with charge number z reads = [ - -f(z)],f(x)=x^2_n=1^. the production of lepton pairs from the nuclear coulomb field by virtual photons was investigated in @xcite . in @xcite the bethe- maximon result
was obtained in the framework of powerful light - cone techniques @xcite , which allows one to trace carefully approximations done at its derivation .
+ as to higher order photon exchanges in the process ( 1 ) , despite the intensive work done on this issue in the last years @xcite the problem of allowing for cc in all orders of perturbation theory is not solved up to now.the goal of the present work is to investigate this issue based on our recent results @xcite obtained for the first terms of perturbation theory for the amplitude of the process ( 1 ) . + as was shown in @xcite , the number of remarkable cancellations between the different terms in the cross section of the process ( 1 ) takes place , which allows one to simplify the obtained expressions and may be a hint of how to solve the problem in the general case . in what follows
we generalize the results of @xcite to all orders of perturbation theory and using the impact factor representation get compact analytical expressions for the cross section of the process ( 1 ) true to all orders of perturbation theory ( up to the constant terms ) .
+ it is convenient to separate the full amplitude of the process ( 1 ) into different terms which lead to a distinct energy dependence in the cross section , namely @xmath5 here @xmath6 is the born amplitude ( see fig.1 ) ; the terms @xmath7 correspond to the pair production by single photon exchange between the ion with charge number @xmath8 and the produced pair and any multiple photon exchanges between the pair of leptons and the ion with @xmath9 ( fig.2 ) ; the term @xmath10 in ( 4 ) describes the cases when the pair interacts with each ion more than once ( fig.3 ) .
+ we do not consider the set of feynman diagrams ( fd ) accounting for the coulomb interactions between the colliding nuclei.such interactions were analyzed in @xcite , where it has been shown that interaction between ions completely canceled when the final ions are not detected .
+ the square of the amplitude ( 4 ) is the combination of different terms which can be classified in the cross section by their energy dependence , i.e. by different powers of big logarithm @xmath11 .
the born approximation leads to the third power of l in the total cross section ( see ( 2 ) ) .
as was discussed in @xcite , the next powers of l come from the square of @xmath12 and @xmath13 and their interference with the born amplitude .
+ as to the interference @xmath14 , it is compensated ( up to the constant ) by the interference between the born term and relevant one @xmath15 , i.e. the combinations of the type @xmath16 .
the same cancellations take place also in the @xmath17 and any interference of the type @xmath18.each such combination leads in the cross section to the terms growing with energy as the first power of l.to find their contribution to the cross section one must take the imaginary part of the complex logarithm @xmath19 which is energy independent .
thus , if one confines to the terms in the cross section rising with energy , the square of the relevant matrix element can be cast in the following form : m ^2=m_b^2+m_s^2 + m^s^2 + 2re(m_b^*m_s+m_b^*m^s ) .
thus to calculate the cross section of the process ( 1 ) ( up to the constant terms ) one needs the knowledge of the born amplitude @xmath6 and terms relevant to single photon exchange with one of the colliding nuclei and any exchanges with the other @xmath7 .
+ let us consider the main results obtained in @xcite for the first terms of such types and generalize them to all orders in @xmath20 .
the amplitude of the process ( 1 ) in the born approximation reads @xcite @xmath21\hat p_2;\\ r_1&=r_1(k_-,k_+)=s(k_-)-s(k_+);\vec{q}_1=\vec{q}_1(k_-,k_+)= \vec{k}_-s(k_-)+\vec{k}_+s(k_+);\nn s(k)&=\frac{1}{\vec{k}^2+\mu^2};\mu^2=m^2+z_+z_-\vec q_1 ^ 2 ; \vec{k}_\pm=\vec{q}_\pm - z_\pm\vec{q}_1;\vec e_1=\frac{\vec q_1 } { \mid \vec q_1\mid } ; z_\pm=\frac{\beta_\pm}{\beta_1}\end{aligned}\ ] ] the quantities @xmath22 for collision of spinless ions @xmath23
so we will omit them later on .
+ we use the sudakov parameterization for the transferred momentum @xmath24 and the created pair momentum @xmath25 in terms of two light - like 4-vectors @xmath26 built from the momenta of colliding ions and invariant energy @xmath27 , and euclidean two - vectors @xmath28 transverse to a beam direction @xmath29 for definiteness we suppose that the created pair moves in the direction of the ion with charge number @xmath30 ; thus @xmath31 .
a similar consideration can be easily carried out for the pairs moving in the opposite direction .
+ expression ( 6 ) depends on lepton transverse momenta @xmath32 through @xmath33 unlike the amplitude for the process of pair photoproduction off the coulomb field.this is a consequence of that in our approach we project all momenta with respect to the axis of colliding ions in their centre of mass system.as to the case of photoproduction , the natural choice is the direction of incident photon .
finally , as was shown in @xcite the racah formula ( 2 ) can be reproduced starting with the above expressions .
+ the spin structure ( 6 ) is the same in all orders of perturbation theory for the matrix elements relevant to single photon exchange with one of the ions ( with any number of exchanges with another ion).this is a direct consequence of lepton helicity conservation at high energy @xcite .
+ using this property we generalize the results obtained in @xcite for the first orders in @xmath34 to all orders of perturbation theory and represent the amplitude relevant to single photon exchange with the ion with charge number @xmath30 and with n attachments to the other ( fig.2 ) : @xmath35\hat{p}_2\\ r_n&=\int\frac{d^2k_1 ... d^2k_n}{\pi^{n-1}\vec{k}_1 ^ 2 ... \vec k _
n^2}\delta^2(\vec k_1+ ... +\vec k_n-\vec q_2)r_n^1(k_-,k_+,k_1, ... k_n),\nn \vec q_n&=\int\frac{d^2k_1 ... d^2k_n}{\pi^{n-1}\vec{k}_1 ^ 2 ... \vec k _ n^2}\delta^2(\vec k_1+ ... +\vec
k_n-\vec q_2)\vec q_n^1(k_-,k_+,k_1, ... k_n).\end{aligned}\ ] ] the quantities @xmath36 can be represented as the combination of s(k ) from ( 7).the expressions for the cases n=2,3 are cited in @xcite.to get the next orders one can use the recurrence relations @xcite : r_n^1(k_n)=r_n-1 ^ 1(k_-,k_+-k_n)-r_n-1 ^ 1(k_k_n , k_+);_n^1(k_n)=_n-1 ^ 1(k_-,k_+-k_n)-_n-1 ^ 1(k_k_n , k_+ ) .
these relations read more effectively in terms of the impact parameter representation which allows one to sum the series like @xmath7 . introducing the fourier transform of the functions @xmath37 g_n^r\{q}(r_1,r_2)=d^2k_-d^2k_+e^i_1 _
+ + i_2_-r_n(k_-,k_+)\{q_n(k_-,k_+)}. one immediately obtains for the relevant structures g_n^r(r_1,r_2)&=&k_0(|_1-_2|)^n ; g_n^q(r_1,r_2)&=&k_1(|r_1-r_2|)^n . here
@xmath38 are the modified bessel functions of the zero and first order . now the summation of perturbation series can be carried out with the result @xmath39;\\ \vec g_q&=\frac{\mu(\vec r_1-\vec r_2)k_1(\mu \mid\vec r_1-\vec r_2\mid ) } { 2(2\pi)^2\mid{\vec r_1-\vec r_2}\mid}[a^{-i\nu}-1+i\nu\ln a];\nn \nu & = \alpha z_2 ; a=\frac{\vec{r}_2 ^ 2}{\vec{r}_1 ^ 2}\end{aligned}\ ] ] applying the inverse fourier transformation to these expressions one obtains @xmath40;\\ \vec q(k_-,k_+)&=\int d^2r_1d^2r_2e^{-i(\vec r_1\vec k_++\vec r_2\vec k_- ) } \frac{\mu(\vec r_1-\vec r_2)}{2\mid \vec r_1-\vec r_2\mid } k_1(\mu|\vec r_1-\vec r_2|)[a^{-i\nu}-1+i\nu\ln a].\end{aligned}\ ] ] the squares of amplitudes entering ( 5 ) are determined by these two structures.really,as was shown in @xcite , ^2= z_+z_- q_1 ^ 2\{(m^2 + 4z_+^2z_-^2q_1 ^ 2 ) r^2 + ( z_+^2+z_-^2)q^2 + 4z_+z_-(z_+-z_-)r q_1q}. as to the amplitude @xmath41 it can be obtained from @xmath42 by simple replacements
@xcite @xmath43
the differential cross section of the process ( 1 ) is connected with the full amplitude ( 4 ) by the relation d=|m|^2 d. the summation is carried over the final particles polarization .
the phase space volume for four particles in the final state can be expressed through the variables introduced above and phase volume of the final leptons d_=^4(q_1+q_2-q_+-q_- ) in the following way @xmath44 the expressions ( 5 - 8,15 - 20 ) allows one to calculate the fully differential cross section of the process under consideration .
+ let us consider the cross section integrated over transverse momenta of final leptons.for this quantity one can obtain the simple analytical formula from which the special cases can be easily deduce .
+ first of all let us note that the integration over created particles transverse momenta in the combinations of type @xmath45 leads to the following structures : @xmath46\nn & = \int d^2r_1d^2r_2 k_0 ^ 2(\mu\mid\vec r_1-\vec r_2\mid ) [ 2-a^{i\nu}- a^{-i\nu}-\nu^2 \ln^2a];\nonumber\\ j_q&=\int d^2r_1d^2r_2 k_1 ^ 2(\mu\mid\vec r_1-\vec r_2\mid ) [ 2-a^{i\nu}-a^{-i\nu}-\nu^2{ln}^2a].\end{aligned}\ ] ] the further integration leads to the relations @xcite j_r =- f();j_q =- f ( ) , where the function @xmath47 is given by ( 3 ) .
+ using the above expressions and representing the cross section as the sum of born term and the term relevant to cc contribution @xmath48 we obtain the following formula : d_c&=&(f(_1)+f(_2 ) ) & & dz . for the sake of simplicity
we consider the case of ions with equal mass @xmath49 and consequently the same lorentz factors @xmath50.the variables @xmath51 and @xmath52 belongs to the intervals 0x < ; ( 1+x ) 1 .
the expression ( 24 ) becomes invalid in the case of pair production by the very soft photons ( @xmath53 in the rest system of one of the ion ) which corresponds to the lower limit of @xmath52 ( we are working in the center of mass system of colliding ions . ) at such low energies one has to take into account the threshold effects and deviation from the eikonal approach @xcite which is out of the present work scope .
+ the integrations in ( 24 ) can be carried out ( for details see appendix ) .
we cite here the expression for cc contribution to the total cross section of the process ( 1 ) _
c= ( f(_1)+f(_2))[^2(_1_2)+(_1 _ 2)+o(const ) ] .
the leading term in this formula @xmath54 coincides with the relevant one found in @xcite . as to the next to leading term ( @xmath55 ) in ( 26 )
it differs from relevant one in @xcite ( see their formula ( 17)).the reason is transparent .
we calculated next to leading terms in the part of the cross section corresponding to the single photon exchange from one of the nuclei ( @xmath56 ) .
as to the formula ( 17 ) in @xcite by derivation it is relevant to the part of the cross section provided by two or more exchanges from both nuclei ( @xmath10 with the preceding notation).as was discussed above and in @xcite such exchanges contribute only to the constant part of the total cross section and are outside of our consideration.there is correspondence between ( 26 ) and relevant formula ( 30 ) in @xcite obtained for the total cross section of @xmath57 pair production in the coulomb field of the nucleus by high energy muons using the generalized weizsacker - williams approximation .
+ in deducing ( 26 ) we integrate over @xmath52 from its lower limit , which is strictly speaking incorrect by the reasons mentioned above .
nevertheless the error in such approach can not be crucial and expression ( 26 ) can be safely used for the simple estimates of the coulomb corrections to the total cross section of the process under consideration.the reason is relative smallness of the near threshold contribution of the pair photoproduction in the coulomb field @xcite .
+ the obtained in the present work expressions can be used for calculation of completely ( four particles in the final state ) or partly ( see ( 24 ) ) differential cross sections the issue which is necessary in any experiment . |
today we are excited by the fast advance of the physics and applied mathematics in the area of research of complex systems in nature and society [ 1 - 8 ] .
an important part of the ground for this success was created more than 60 years ago when l. f. richardson and l. w. lanchester , both famous british scientists , were the first who rose and applied the idea for mathematical modeling of arms races and military combats [ 9 - 12 ] . for many years
the research on wars and other military conflicts was concentrated in the military universities and academies . in the last 20 years and especially after the terrorist attack on 11th of september 2001 this research became actual for many physicists and applied mathematicians too [ 13 - 15 ] . in this paper
we shall follow the terminology used by epstein @xcite who applied ecological models of lotka - volterra kind for description of combats .
let us have two conflicting groups named the `` red group '' @xmath0 and the `` blue group '' @xmath1 .
a general form of model equations of an armed conflict between these groups is @xmath2 where @xmath3 and @xmath4 are the numbers of armed members of the two groups ; @xmath5 and @xmath6 are the `` firing efectivenes '' ( technology level ) of the groups ; and @xmath7 and @xmath8 are linear or nonlinear functions , depending on the character of the conflict .
epstein @xcite proposed the following class of models of the conflict @xmath9 where @xmath10 are real nonnegative coefficients .
if these coefficients are constants the models are called hard models .
if the coefficients depend on the the number of participants or on the parameters of the environment the models are called soft ones .
important characteristics of the model ( [ epst1 ] ) is the casuality exchange ratio or state equation ( the ratio of eliminated members of the `` red '' and `` blue '' groups ) . for ( [ epst1 ] )
the ratio is @xmath11 where @xmath12 and @xmath13 .
the integral form of the above casuality exchange ratio is @xmath14 where @xmath15 and @xmath16 are the numbers of the members of the two groups at the beginning of the conflict ( at @xmath17 ) . for developing of intuition and decision skills it is of interest to know @xmath3 and @xmath4 in closed form .
such analytical solutions are possible only in small number of cases .
section 2 contains the solution of the system of model equations for selected values of the parameters @xmath18 .
several concluding remarks are summarized in section 3 .
the linear model of lanchester describes position conflicts such as the battles for somna and verdune in 1916 .
the coefficients in ( [ epst1 ] ) are @xmath19 , @xmath20 .
the equation of state is quadratic @xmath21 where @xmath22 can be positive , negative , or @xmath23 . for the case
@xmath24 we obtain @xmath25 and the solutions are @xmath26 evidently @xmath27 which means that after endless position conflict the two groups are destroyed and no one of them wins .
however the situation changes if @xmath28 .
let us first assume that @xmath29 . from the state equation ( [ steq1 ] ) @xmath30 and
the solutions of the model system of equations are @xmath31 where @xmath32 .
the obtained solutions satisfy the initial conditions @xmath33 , @xmath34 .
in addition at @xmath35 we obtain @xmath36 and @xmath37 . in other words
the blue group will win the conflict if it lasts long enough .
the red group commanders have to change the strategy if they want to escape the defeat .
if this does not happen after the time @xmath38 from the beginning the red group will be completely destroyed - fig
. 1 . ) .
the conflict ends at finite time by destruction of the red group .
nevertheless the blue group suffers heavy losses.,width=529 ] we note again that this happens when @xmath29 , i.e. , when @xmath39 the condition ( [ sqlaw ] ) ( but with @xmath40 instead of @xmath41 ) for military combats is known as the square law of lanchester : to stalemate and adversay army two times as numerous as yours , your army must be four times as effective .
but in this case our army will be also destroyed .
thus the correct statement of the square law is as follows : to stalemate and adversay army two times as numerous as yours , your army must be four times as effective .
in other words : in position war , in order to stop army that is @xmath42 time larger than yours your army has to be more than @xmath43 technologically better ( to have more than @xmath43 times larger firepower ) .
we now discuss the case @xmath44 .
the solutions of the model equations are @xmath45 now the red group wins if @xmath46 i.e. when @xmath47 , @xmath48 . in this case
the coefficients in the general model are @xmath49 and the equation of state is @xmath50 epstein assumes that stalemate occurs when @xmath51 , i.e. , when @xmath52 .
let us discuss this in more details .
if @xmath53 ( epstein case ) then the solution of the model system of equations is @xmath54 thus at @xmath55 , none of the groups wins , the attack is stopped but the two groups are completely destroyed . this of course is not of favor for the group leaders . in order to consider more realistic scenarios we have to set @xmath56 . in this case
the solutions of the model equations are @xmath57 where @xmath58 and @xmath59 . now
the sign of @xmath60 determines the asymptotic behavior of the number of members of the two groups . if @xmath61 then @xmath62 and @xmath63 .
thus the blue groups wins and the results of the attack is that the red group is defeated -fig.2 ) . , width=529 ] now let @xmath64 .
then @xmath65 in this case as winner from the attack scenario is the red group as @xmath66 , @xmath67 . now let us discuss the following detail .
let us assume that at the beginning of the attack the blue group has more soldiers than the red group : @xmath68 but the firepower of the red group is larger : @xmath69 .
then there must be a moment @xmath70 where the two groups will have equal number of armed members : @xmath71 .
this moment @xmath70 can be determined from the equation of state ( [ steq2 ] ) @xmath72 what is interesting that when @xmath73 a solution exists only if @xmath74 and it is ( see fig.3 ) ) , width=529 ] @xmath75 for this case @xmath76 , @xmath77 .
the equation of state is @xmath78 what is interesting here is that the large group does not win in any case .
let for an example @xmath79 .
then the solutions of the model equations are @xmath80 which means that none wins ( @xmath81 ) and the groups are destroyed .
this of course is not acceptable for both sides .
more realistic are situations where @xmath82 .
let first @xmath83 .
the solution of the model equations is @xmath84 where @xmath85 and @xmath86 .
thus in the asymptotic case @xmath87 and @xmath63 , i.e. the blue groups wins .
quite interesting is the case @xmath88 . in this case
the solution of the model system is @xmath89 where @xmath90 , @xmath91 . in this case at @xmath92
we obtain @xmath93 .
thus at @xmath94 this conflict ends .
blue group is completely destroyed and the red group wins . for these models @xmath95 , @xmath96 and @xmath97 .
epstein @xcite argues that the exponents @xmath18 in the general model ( [ epst1 ] ) should be kept in the interval @xmath98 $ ] .
however there is no evidence to sustain such an assertion .
here we investigate two models for which some of the exponents @xmath18 equal @xmath99 .
let first @xmath100 , @xmath101 .
the model equations are @xmath102 the equation of state is @xmath103 if @xmath24 then the equation of the state together with one of the model equations yield @xmath104 no one of the groups wins at @xmath105 . if @xmath106 we obtain @xmath107 where @xmath108 .
thus @xmath63 and @xmath109 .
if @xmath110 @xmath111 thus @xmath112 and @xmath66 .
let now @xmath113 and @xmath114 . the equilibrium condition ( [ icer ] ) becomes @xmath115^{\epsilon } , \epsilon = \frac{r}{b}\ ] ] the final solution is @xmath116 where @xmath117 is a typical time of decay .
let us form the ratio @xmath118 where @xmath119 or @xmath120 .
assuming @xmath68 but @xmath69 and letting @xmath121 we obtain @xmath122 \hskip.5 cm \kappa = \frac{r+b}{r - b}\ ] ] thus @xmath123 at @xmath124 . nevertheless , both groups , fight to the end ( @xmath125 ) .
it is known from the history of the conflicts that epidemic events had frequently occurred particularly in case of attrition and ambush conflicts .
the simplest way to account for this effect ( removing of conflict participants because of sickness ) is to modify the general model as follows @xmath126 where @xmath127 and @xmath128 are coefficients of morbility ( sick rate ) removal .
generally @xmath129 .
we choose to demonstrate the effect of epidemics on the model ( [ steq3 ] ) in the form @xmath130 hence @xmath131 where @xmath132 @xmath133 @xmath134 in the general case ( [ dop4 ] ) the quantity @xmath135 can be zero , positive or negative .
for an example if @xmath136 the solution of the model system is @xmath137 where @xmath138 these solutions degenerate into ( [ solution31 ] ) at @xmath139 .
a class of mathematical models of armed conflicts ( [ epst1 ] ) was investigated .
the purpose was to identify particular cases with exact simple solutions in analytical form @xmath140 and @xmath141 , where @xmath1 and @xmath0 were the armies numbers .
it was found that these requirements were met by the linear ( lanchester s ) model known from long ago in the form ( [ steq1 ] ) as well as by several nonlinear models described in this paper .
these models demonstrate rich behavior in the time . *
no one of the two groups @xmath1 and @xmath0 wins after endless ( @xmath105 ) attrition conflict- fighting to the finish ( [ res1 ] ) , ( [ sol6y ] ) * one of the groups wins after limited in time or after an endless conflict ( see table 1 ) + -2.5 cm + .summary of model predictions .
[ cols="^,^,^,^,^,^ " , ] both conflicting groups are defined by their initial numbers @xmath15 , @xmath16 and respective firing efectivenesses per shot . the character of the conflict is modeled by the form of the functions @xmath7 and @xmath8 ( [ model1 ] ) which can be linear or nonlinear .
all models can be extended to account for occuring of epidemic events in the fighting groups .
the model ( [ dop2 ] ) is an example .
99 nature , * 337 * , 1989 , 701 - 704 . j. math .
sociology , * 18 * , 1993 , 47 -64 .
studies quarterly , * 37 * , 1993 , 55 - 72 .
usa , * 99 * , 2002 , 7193 - 7194 .
sci.usa , * 102 * , 2005 , 255 - 260 .
theoretical population biology , * 66 * , 2004 , 1 - 12 .
lett . a , * 349 * , 2006 , 350 - 355 .
chaos solitons & fractals , * 33 * , 2007 , 1658 - 1671 .
nature , * 135 * , 1935 , 830 - 831
. nature , * 148 * , 1941 , 598 - 598 .
soc . , * 107 * , 1944 , 242 - 250 .
soc . a , * 109 * , 1946 , 130 - 156 .
int . security , * 12 * , 1998 , 154 - 165 .
american anthropology , * 109 * , 2007 , 318 - 329 .
j. infectious diseases , * 11 * , 2007 , 98 - 108 .
nonlinear dynamics , mathematical biology and social sciences .
addison - wesley , readings , ma , 1997 . | the human society today is far from perfection and conflicts between groups of humans are frequent events .
one example for such conflicts are armed intergroup conflicts .
the collective behavior of the large number of cooperating participants in these conflicts allows us to describe the conflict on the basis of models containing only few variables . in this paper
we discuss several cases of conflicts without use of weapons of non - conventional kind . in the ancient times
the chinese writer sun tsu mentioned that the war is an art .
we can confirm that the conflict is an art but with much mathematics at the background . *
key words * : conflict , attrition , ambush , combat , mathematical models |
we thank paul newman for valuable discussions .
mgr and gw would like to thank the ippp at the university of durham for hospitality .
this work was supported by the royal society , by intas grant 05 - 103 - 7515 , by grant rfbr 04 - 02 - 16073 and by the federal program of the russian ministry of industry , science and technology ss-1124.2003.2 .
h1 collaboration , desy 06 - 048 , arxiv : hep - ex/0606003 .
h1 collaboration , desy 06 - 049 , arxiv : hep - ex/0606004
. j. c. collins , phys .
d * 57 * ( 1998 ) 3051 [ erratum - ibid .
d * 61 * ( 2000 ) 019902 ] .
g. ingelman and p. e. schlein , phys .
b * 152 * ( 1985 ) 256 .
a. donnachie and p. v. landshoff , phys .
b * 296 * ( 1992 ) 227 .
a. d. martin , m. g. ryskin and g. watt , eur .
j. c * 44 * ( 2005 ) 69 .
g. watt , arxiv : hep - ph/0511333 .
a. g. shuvaev _ et al .
_ , phys . rev .
d * 60 * ( 1999 ) 014015 .
s. chekanov _ et al . _
[ zeus collaboration ] , eur .
j. c * 24 * ( 2002 ) 345 .
a. aktas _ et al .
_ [ h1 collaboration ] , eur .
j. c * 46 * ( 2006 ) 585 .
m. gluck , e. reya and a. vogt , z. phys .
c * 53 * ( 1992 ) 651 .
j. f. owens , phys .
d * 30 * ( 1984 ) 943 .
a. d. martin , w. j. stirling and r. s. thorne , phys .
b * 636 * ( 2006 ) 259 .
m. botje , qcdnum version 16.12 , zeus note 97 - 066 ( unpublished ) , available from ` http://www.nikhef.nl/\simh24/qcdcode/ ` .
s. chekanov _ et al . _
[ zeus collaboration ] , eur .
j. c * 38 * ( 2004 ) 43 .
s. chekanov _ et al . _
[ zeus collaboration ] , nucl .
b * 672 * ( 2003 ) 3 .
b. lhr [ zeus collaboration ] , presented at _
diffraction 2006 _ , milos , greece , 510 september 2006 . | we analyse the latest h1 large rapidity gap data to obtain diffractive parton distributions , using a procedure based on perturbative qcd , and compare them with distributions obtained from the simplified regge factorisation type of analysis .
the diffractive parton densities and structure functions are made publically available .
ippp/06/66 + dcpt/06/132 + 7th november 2006 + * diffractive parton distributions from h1 data * a.d .
martin@xmath0 , m.g . ryskin@xmath1 and g. watt@xmath2 + @xmath0 institute for particle physics phenomenology , university of durham , dh1 3le , uk + @xmath3 petersburg nuclear physics institute , gatchina , st .
petersburg , 188300 , russia + @xmath2 department of physics & astronomy , university college london , wc1e 6bt , uk the h1 collaboration have recently released measurements of diffractive deep - inelastic scattering ( ddis ) , @xmath4 , with more precision than hitherto .
they have presented data collected using their forward proton spectrometer ( fps ) @xcite and also data selected using the large rapidity gap ( lrg ) method @xcite .
in addition they have used their lrg data , which has much more statistics , to determine diffractive parton densities assuming that regge factorisation holds .
we call this a `` regge '' fit dependence @xcite . ] . in principle
, this regge factorisation is _ not _ justified theoretically , and the purpose of this note is to present diffractive parton distributions obtained from exactly the _ same _ h1 lrg data @xcite using a more theoretically sound procedure .
it is convenient to call this the perturbative qcd ( `` pqcd '' ) procedure , although it also contains non - perturbative regge - like contributions .
we present the pqcd procedure first .
it is shown schematically in fig .
[ fig : diagrams ] .
there are three components .
first in fig .
[ fig : diagrams](a ) we have the perturbative _ resolved _ pomeron contribution in which the virtualities of the @xmath5-channel partons are strongly ordered as required by dglap evolution : @xmath6 .
the scale @xmath7 at which the pomeron - to - parton splitting occurs can vary between @xmath8 gev@xmath9 and the factorisation scale @xmath10 . for @xmath11 , the representation of the pomeron as a perturbative parton ladder is no longer valid and instead , in the lack of a precise theory of non - perturbative qcd , we appeal to regge theory where the soft pomeron is a regge pole with intercept @xmath12 @xcite ; see fig . [
fig : diagrams](b ) .
in addition to the _ resolved _ pomeron contributions of figs . [
fig : diagrams](a , b ) , we must also account for the _ direct _ interaction of the perturbative pomeron in the hard subprocess , fig .
[ fig : diagrams](c ) , where there is no dglap evolution in the upper part of the diagram .
( a)(b)(c ) + , and ( c ) the perturbative _ direct _ pomeron contribution . ] , and ( c ) the perturbative _ direct _ pomeron contribution.,scaledwidth=80.0% ] , and ( c ) the perturbative _ direct _ pomeron contribution .
] each diffractive dis event is described by four variables , usually taken as @xmath13 , @xmath14 , @xmath10 and @xmath5 . here , @xmath15 is the fraction of the proton s light - cone momentum transferred through the rapidity gap , @xmath16 is the fraction of the pomeron s light - cone momentum carried by the struck quark ( at lo ) , and @xmath17 is the bjorken-@xmath18 variable .
the data are integrated over the squared momentum transfer @xmath19 .
the variables are shown in fig .
[ fig : diagrams](a ) . in the pqcd approach
, the diffractive structure function @xmath20 is given by @xcite @xmath21 where the first _ resolved _ pomeron term corresponds to figs .
[ fig : diagrams](a , b ) while the second _ direct _ pomeron term corresponds to fig .
[ fig : diagrams](c ) .
the @xmath22 are the same coefficient functions as in inclusive dis .
the diffractive , written in in terms of the inclusive parton densities , also contains some admixture from secondary reggeons and not just from the rightmost ( pomeron ) singularity in the complex angular momentum plane .
so everywhere the word `` diffractive '' and the index @xmath23 should be regarded as a symbolic , rather than a literal , notation . ]
parton distribution functions ( dpdfs ) , @xmath24 or @xmath25 , are obtained at a scale @xmath10 by evolving up from @xmath26 with the inhomogeneous evolution equations @xcite , @xmath27 here , @xmath28 $ ] is the fraction of the pomeron s light - cone momentum .
the first term on the right - hand - side of generates the upper part of the ladder in fig .
[ fig : diagrams](a ) , that is , above @xmath7 , via the usual ( homogeneous ) dglap evolution with parton - to - parton splitting functions , @xmath29 .
the second ( inhomogeneous ) term describes the cell where the first @xmath5-channel single parton is generated directly by the pomeron - to - parton splitting functions , @xmath30 , at a scale @xmath7 . in other words , these functions describe the transition from the two @xmath5-channel partons ( that is , the pomeron ) to a single @xmath5-channel parton .
the lo expressions for @xmath30 are given in ref .
@xcite .
the lower parton ladder in fig .
[ fig : diagrams](a ) plays the rle of a perturbative pomeron flux factor , @xmath31 .
these flux factors may be calculated from the familiar global parton densities @xcite .
for example , if the two @xmath5-channel partons are gluons , then @xmath32 ^ 2 , \label{eq : fg}\ ] ] where the diffractive slope parameter @xmath33 comes from the @xmath5 integration , and @xmath34 from the skewedness of the gluon distribution @xcite .
the notation @xmath35 denotes whether these two @xmath5-channel partons are gluons ( @xmath36 ) or sea - quarks ( @xmath37 ) , while the interference term is denoted by @xmath38 .
we neglect the case where the two @xmath5-channel partons are valence quarks , assuming that @xmath39
. we will take @xmath40 gev@xmath41 @xcite , apart from for heavy quark production via a _
direct _
pomeron mechanism , fig .
[ fig : diagrams](c ) , where we take @xmath42 gev@xmath41 @xcite .
now we come to an important observation .
if we neglect the @xmath7 dependence of @xmath43 in , then @xmath44 and the inhomogeneous term would be a higher - twist power - like correction , which is usually neglected and so collinear factorisation would hold in ddis .
however , as @xmath45 , the gluon density grows as @xmath46 , with anomalous dimension @xmath47 , while in the hera domain @xmath48 .
this compensates for the @xmath49 power - like suppression in .
we therefore can not neglect the inhomogeneous term in . using only the perturbative framework ,
that is , just the diagrams in figs .
[ fig : diagrams](a , c ) , down to a very low scale would , of course , be unreliable .
however , it is reasonable to assume that all the contributions originating from these low scales , @xmath11 , are parameterised in terms of an unknown non - perturbative input , which is schematically presented in fig .
[ fig : diagrams](b ) . unlike the unknown input required to analyse inclusive dis which depends on a single momentum fraction ,
here we have an unknown input which depends on two momentum fractions , @xmath13 and @xmath50 . in this non - perturbative domain
it is reasonable to assume a regge behaviour , in which the @xmath13 dependence is given by the sum of regge poles ( including possible interference terms ) , and that each pole generates input diffractive parton distributions with a freely - parameterised @xmath50 behaviour , where the parameters are to be determined by fitting to the ddis data .
note that for very low @xmath13 it would be sufficient to keep just pomeron exchange .
thus in the pqcd approach we consider the dpdfs to be a sum of pomeron and secondary reggeon contributions , @xmath51 where we have neglected the possible interference term , as is the usual practice @xcite .
we parameterise the input quark - singlet and gluon dpdfs for the leading pomeron pole at a starting scale @xmath26 in the form : @xmath52 where the pomeron flux factor is taken from regge phenomenology @xcite , @xmath53 with @xmath54 .
the secondary reggeon contributions may be included in a similar way .
however , in practice , the ddis data at large @xmath13 are insufficient to constrain the input secondary reggeon parameters . instead , following h1 @xcite ,
we fix the @xmath50 and @xmath10 dependence of @xmath55 to be the same as the pionic parton distributions , that is , we neglect the inhomogeneous terms in the evolution , and fit only an overall normalisation factor @xmath56 .
this is reasonable since the secondary reggeons are important only for larger @xmath13 , where the anomalous dimension is not large enough to cancel the power suppression in the corresponding perturbative pomeron flux factors .
moreover , in the perturbative domain , the secondary reggeons are mainly non - singlet quarks which , even at very small @xmath13 , have a small ( or negative ) anomalous dimension .
this completes the summary of the pqcd procedure which is necessary to describe ddis .
now we introduce the regge factorisation approach and relate it to the above pqcd procedure . in the `` regge '' description the last term in and the inhomogeneous term in are neglected .
the factorising regge input , and , is used in the remaining , now homogeneous , dglap evolution equations , .
this assumption is justified for sufficiently large @xmath26 and fixed @xmath13 .
the smaller the value of @xmath13 , the larger the value of @xmath26 has to be to justify the pure regge approach by the linear evolution equations starting from input generated by some set of regge poles . of course
, at a large scale it may be necessary to have a more complicated structure in the complex angular momentum plane than simply the pomeron and secondary reggeon poles .
] . what happens if the regge approach is applied for smaller values of @xmath26 ?
the non - negligible last term in is attempting to be mimicked by enlarging the regge input at @xmath26 .
as a consequence we obtain the wrong @xmath7 dependence of the dpdfs .
we can readily glimpse what the result will be .
experimentally we measure @xmath57 , which is dominantly determined by the diffractive quark density .
thus , to mimic the @xmath7 dependence of the last term in in a pure regge fit , we would expect to obtain a larger gluon at relatively large @xmath50 so as to provide a faster growth of the quark density with increasing @xmath7 .
how different are the diffractive parton distributions obtained from fits to the ddis data using the `` pqcd '' and the simpler `` regge '' approaches ? to see this , we perform fits to the latest h1 lrg data @xcite for the diffractive reduced cross section @xmath58 .
we choose this particular data set because h1 have already performed a `` regge '' parton analysis to precisely these data with @xmath59 gev and @xmath60 gev@xmath9 .
we will use the same kinematic cuts , in order to allow a direct comparison with the h1 2006 analysis @xcite .
we take @xmath61 as a free parameter in and fix @xmath62 gev@xmath41 and @xmath63 gev@xmath41 @xcite , as in the h1 fit .
moreover , the secondary reggeon contributions are also included in a similar way to the h1 2006 analysis @xcite , with @xmath64 , @xmath65 gev@xmath41 and @xmath66 gev@xmath41 @xcite , but using the grv nlo pionic parton distributions
@xcite rather than the owens lo set @xcite . the number of parameters in the pqcd and regge descriptions are the same , since the inhomogeneous term and _ direct _ pomeron contributions are calculated explicitly @xcite , at leading order , and have no free parameters .
we choose to take the input parameterisations at @xmath67 gev@xmath9 , again to keep close to the value used in the h1 analysis @xcite , and because the inclusive parton densities which appear in the pqcd approach are not well known below @xmath68 gev@xmath9 where there are no data included in the global fits .
we use the mrst2004f3 nlo pdfs @xcite in the pqcd fits .
the inhomogeneous evolution is implemented via a modified version of the qcdnum @xcite nlo dglap program . for the pomeron contribution , we include heavy quarks in the fixed flavour number scheme ( ffns ) , where there are no heavy quark dpdfs , with @xmath69 gev and @xmath70 gev , and use a three - flavour @xmath71 with @xmath72 mev @xcite . for the _ resolved _ pomeron case ,
heavy quarks are generated via photon
gluon fusion at nlo @xcite with the renormalisation and factorisation scales taken to be @xmath73 where @xmath74 . for the secondary reggeon case , the contribution to the diffractive structure functions from heavy quarks
is included using the zero - mass variable flavour number scheme , and we take the same values of @xmath71 , @xmath75 and @xmath76 as in the determination of the pion pdfs @xcite .
the _ direct _ pomeron terms @xmath77 , that is , the second terms on the right - hand - side of , are calculated at lo using : @xmath78 where the equation numbers refer to the appendix of ref .
@xcite .
for light quarks , the lower limit of the @xmath7 integration in these equations is replaced by @xmath26 , since the contribution from @xmath11 is included in the input dpdfs taken at a scale @xmath26 . for @xmath79 the light - quark contributions in the limit @xmath80 are subtracted since they are already included in the first term of . this subtraction defines a choice of factorisation scheme .
there is no such subtraction for @xmath81 , which are purely higher - twist , or for the heavy quark contributions since we are working in the ffns .
.input parameters ( corresponding to @xmath82 ) determined in the regge and pqcd fits to h1 lrg data @xcite with @xmath83 gev@xmath9 .
notice that the @xmath84 errors on the parameters are generally smaller in the pqcd fit . [ cols="^,^,^",options="header " , ] gev@xmath9 compared to the h1 2006 fit a @xcite .
all the dpdfs plotted here are normalised to @xmath85 gev by multiplying by a factor 1.23 relative to @xmath86 .
only the pomeron contribution is shown and not the secondary reggeon contributions which are negligible at the value of @xmath87 chosen here .
also shown are the dpdfs obtained by h1 @xcite , using the regge factorisation approach , from a preliminary combined analysis of their inclusive ddis data @xcite with preliminary data on diffractive dijet production @xcite . ]
examples of the diffractive parton densities obtained from these fits to the h1 lrg data are shown in fig .
[ fig : dpdf ] . during the fits , statistical and systematic experimental errors were added in quadrature .
an overall factor of 1.23 @xcite was applied to the theoretical prediction to account for the effect of proton dissociation up to masses @xmath88 gev present in the h1 lrg data @xcite .
the parameters are given in table [ tab : params ] and correspond to @xmath82 .
as is to be expected , there are only very small differences between our `` regge '' fit and the h1 2006 fit a. , @xmath75 , @xmath76 , the heavy quark renormalisation and factorisation scales , and the pion pdfs . however , as seen in fig .
[ fig : dpdf ] , these minor details are not important . ]
the value of the @xmath61 parameter in the pqcd fit is @xmath89 .
it is a bit higher than the value of the intercept 1.08 @xcite coming from soft hadron phenomenology , since the input distributions are taken at @xmath67 gev@xmath9 , and already account for some pqcd contribution which is known to be present at this scale .
for example , in going from @xmath90 to @xmath91 photoproduction , the effective pomeron intercept increases from about 1.09 @xcite to 1.20 @xcite .
note that a detailed comparison of the intercepts would require a study of the correlation between the fitted @xmath61 and the assumed @xmath92 , whereas here we have fixed @xmath93 gev@xmath41 to agree with the h1 analysis @xcite . from fig .
[ fig : dpdf ] we see that the diffractive quark singlet distribution , @xmath94 , is relatively stable in going from the pqcd approach to the simplified regge approach . however , as anticipated , the diffractive gluon distribution , @xmath95 , is greatly suppressed for @xmath96 in the pqcd approach relative to the gluon obtained using the regge approach .
the results obtained here are stable to the inclusion of ddis data from the h1 fps @xcite , the zeus leading proton spectrometer ( lps ) @xcite , and the zeus diffractive charm reduced cross section @xcite , as well as being stable to including h1 lrg data @xcite down to @xmath97 gev@xmath9 . at an input scale of @xmath67 gev@xmath9 , the results are also stable to an extra parameter in the input distributions in the form of an additional factor @xmath98 and to the use of an alternative input parameterisation in terms of chebyshev polynomials ,
although this is no longer the case if a higher input scale is used such as @xmath99 gev@xmath9 .
when the forthcoming zeus lps and lrg data @xcite are released we will perform an analysis of the combined data sets , including also the measurements of diffractive dijet production .
the pqcd diffractive parton densities , @xmath100 and @xmath101 , of the present analysis are publically available in the form of three - dimensional grids in @xmath102 , together with the corresponding predictions for the diffractive structure functions @xmath20 ,
@xmath103 , @xmath104 and @xmath105 .
the _ resolved _
pomeron , _ direct _ pomeron and secondary reggeon contributions to the diffractive structure functions are provided separately in each of these four cases .
measurements of the diffractive charm reduced cross section and the fractional contribution of charm to the total diffractive cross section have recently been shown to be in excellent agreement with these predictions ; see figs .
12 and 13 of ref .
@xcite .
the h1 collaboration have recently presented a preliminary combined fit @xcite , within the regge factorisation framework , of their inclusive ddis data @xcite with preliminary data on diffractive dijet production @xcite .
the dpdfs from this combined fit , shown in fig .
[ fig : dpdf ] , have a much smaller diffractive gluon density at moderate to high @xmath50 compared to the h1 2006 fit a. moreover , the @xmath106 for the 190 inclusive ddis points increases from 158 ( h1 2006 fit a ) to 169 ( h1 combined fit ) on inclusion of the dijet data . therefore , the gluon density determined indirectly from the inclusive ddis data , under the assumption of pure dglap evolution , is different from the gluon density preferred by the dijet data . on the other hand , our pqcd fit naturally has a smaller gluon density at moderate to high @xmath50 than the h1 2006 fit a , due to the presence of the extra inhomogeneous term in the evolution equations .
therefore , the apparent tension between the inclusive ddis and dijet data in the regge factorisation approach is partly alleviated by the inclusion of the perturbative pomeron terms . as a consequence
, we expect that our pqcd fit will give a better description of the diffractive dijet data than the h1 2006 fit a. |
many observations have indicated that the universe was reionised early on . according to the gunn - peterson effect @xcite in the spectra of high-@xmath3 quasars
, the universe was already reionised by @xmath4 @xcite .
in addition , the _ wilkinson microwave anisotropic probe _ ( wmap ) satellite produced further information on the reionisation epoch .
the measured optical depth for the thomson scattering of cmb photons by free electrons is @xmath5 .
assuming an instantaneous reionisation as a toy model , this value indicates that the universe was reionised at @xmath6 @xcite .
moreover , in the last decade , some observed quantities , such as the spectra of gamma - ray burst 050904 @xcite and the evolution of the luminosity function of lyman-@xmath7 emitters ( laes ) @xcite , have also been used to evaluate the neutral hydrogen fraction during the epoch of reionisation .
however the exact reionisation history is still unclear , since we can not directly detect the ionisation state of intergalactic medium ( igm ) even with the latest observational facilities .
numerical simulations are means of increasing our understanding of the physical properties of the undetected igm
. many numerical simulations of the cosmic reionisation have been performed @xcite . moreover , in order to predict the 21-cm signal emitted in intergalactic medium ( igm ) during the reionisation epoch , which should be observed by radio - interferometers such as lofar , mwa and ska , simulations of the cosmic reionisation have been performed .
@xcite .
the nature of the sources of the cosmic reionisation is still uncertain , but stars in galaxies , at least , are expected to contribute to the process , since many galaxies have already been observed at @xmath8 . in a cold dark matter ( cdm ) cosmology , structure formation proceeds in a bottom - up manner . therefore it is expected that low mass galaxies were dominant sources of ionising photons during the early phase of reionisation .
however it is also well known that star formation in low mass galaxies is very sensitive to feedback effects owing to their shallow gravitational potential .
for instance an external ultraviolet background ( uvb ) radiation field which develops as the cosmic reionisation proceeds heats the gas up to @xmath9k @xcite and dissociates molecules @xcite .
moreover @xcite have shown an external uvb can suppress the star formation even in a galaxy that is massive enough to sustain an ionised hot gas .
the suppression in the massive galaxy is mainly due to the fact that photo - heating prevents gravitational inability .
not only the external uvb but also uv feedback from its own stars is likely to regulate star formation in each galaxy . in order to investigate ionising photon escape fraction from high@xmath10 dwarf galaxies , @xcite have performed radiation hydrodynamics ( rhd ) simulations in which the internal uv and sn feedback are included .
they have shown that the internal feedback evacuates a large amount of gas from halos with masses less than @xmath11 , and regulates the star formation rate .
if previous episodes of star formation and feedback effects associated with the star formation , which are not included in their simulations , are taken into account , the impact of the internal feedback is expected to be more drastic .
in addition , such evacuation would lead to the enhancement of the escape fraction of ionising photons , which is an important factor to regulate reionisation history . according to the previous studies
, it is strongly expected that uv feedback plays a very important role in regulating the histories of the cosmic star formation and the reionisation .
but unfortunately it is often hard to implement the internal uv feedback in simulations of the cosmic reionisation , since calculation costs for the rt are very high .
@xcite have explored the reionization history regulated by uvb with high resolution @xmath12-body and post - processing rt calculations , assuming a constant characteristic halo mass of @xmath13 below which stars are forbidden to form in ionised regions . however , this characteristic mass should depend on the halo formation epoch , the strength of the uvb radiation field , and so on @xcite .
moreover the internal uv feedback can not be correctly included due to a lack of spatial resolution in rt calculations , although the spatial and mass resolutions for @xmath12-body simulations are sufficiently high to take into account ionising photons from small structures .
hence the star formation history in their simulations might be inaccurate . @xcite have recently performed simulations of the cosmic reionisation in which the transfer of uv photons is consistently coupled to hydrodynamic calculations .
hence they can consider suppression of star formation in galaxies due to photo - evaporation without assuming a characteristic mass .
however the spatial resolution for their rt calculations is not high enough to correctly take into account the internal uv feedback .
the lack of spatial resolution also implies using an escape fraction parameter , which is frequently assumed to be constant over all masses of galaxies .
since the ionising photon escape fraction likely depends on various properties of the galaxy , it is doubtful that the assumption of a constant value is valid . in addition , there is little agreement on the value of the escape fraction in previous studies @xcite .
hence the employment of the escape fraction results in a large uncertainty of the reionisation history .
@xcite have implemented a moment - based rt solver in gadget , and have simulated the cosmic reionisation . with the code , spatial resolutions for the rt and hydrodynamics calculations are equivalent so that the internal uv feedback can be appropriately considered .
in addition , they didt have to employ a `` galactic '' escape fraction .
however there might be still an unresolved interstellar medium ( ism ) where some photon absorption is expected .
hence they adopted an `` ism '' escape fraction to take into account the photon losses in the unresolved ism .
although some of their models well produce a plausible history of reionisation , the minimum halo mass resolved in their simulations , typically @xmath14 but @xmath15 in the maximum resolution run , seems too massive to discuss the suppression of star formation at the epoch of reionisation @xcite . therefore the impacts of the internal and external uv feedback on the histories of star formation and reionisation have not been well established yet . in this paper
, we aim to clarify the importances of uv feedback on cosmic reionisation history by radiation hydrodynamics simulations .
unlike previous numerical simulations of the cosmic reionisation , we include non - equilibrium chemistry regarding the @xmath0 molecule , which is the most efficient coolant below @xmath16k in a low - metallicity environment .
@xmath0 cooling plays a crucial role not only for less massive halos with @xmath17k but also for massive halos with @xmath18k .
needless to say that halos with @xmath17k can not collapse without @xmath0 cooling in the early universe .
but even in halos with @xmath18k , internal structures would be affected by thermal processes .
we have to trace the dynamical and thermal evolution of gas in the cold @xmath19k regime , since stars are born in cold and dense regions .
in addition , resolving such internal structures is very important to understand how ionising photons escape from a galaxy @xcite .
very recently , @xcite have shown that the global and local photo - dissociating radiation regulates cosmic star formation history in the epoch of reionization .
however the roles of an inhomogeneous reionisation process and the internal photo - ionisation feedback are still unclear , because they simply assumed a homogeneous reionisation model .
we show in this paper not only photo - dissociating radiation but also photo - ionisation radiation plays crucial roles on the reionisation and star formation history .
this paper is organized as follow . in section 2
, we describe our numerical method and the setup of the simulations .
we show the results of the simulations in section 3 . in section 4 ,
we compare the simulation results with observations , and also discuss the expected effect of some processes that we did not consider in this work .
finally we devote section 6 to the conclusions . throughout this paper
we assume a @xmath20 cold dark matter universe with cosmological parameters : @xmath21 , @xmath22 , @xmath23 , @xmath24 , and @xmath25 based on 7-year wmap results @xcite
[ cols="^,^,^,^,^,^,^",options="header " , ] in order to self - consistently simulate the reionisation and star formation history , we have to couple hydrodynamics to the transfer of uv photons from numerous radiation sources . for this purpose
we use a novel radiation hydrodynamics code start ( sph with tree - based accelerated radiative transfer ) @xcite , which enables us to solve such a numerically expensive problem .
we describe the detailed simulation method in the following part .
we adopt the star formation recipe employed in @xcite who explored the impact of uvb on the formation of dwarf galaxies .
we allow each gas particle to form stars , if a gas particle satisfies the following three conditions ; ( i ) @xmath26 , ( ii ) @xmath27k , and ( iii ) @xmath28 .
the stars are born at a rate of @xmath29 where @xmath30 , @xmath31 , and @xmath32 are the star and gas mass density , and the star formation efficiency , respectively .
the dynamical time @xmath33 is evaluated as @xmath34 . here
the condition ( ii ) is the most crucial , since this condition can be satisfied only if the gas is well shielded against uv radiation so that @xmath0 cooling effectively works .
however the change of @xmath35 hardly affects the final results as long as @xmath35 is well smaller than 10,000k , since the cooling time is well shorter than the dynamical time in a self - shielded region @xcite . in previous simulations of galaxy formation
, @xmath36 is frequently employed so as to reproduce the schmidt - kennicutt relation@xcite . in addition , @xcite have shown the final stellar mass fraction in a halo is almost independent on the value of @xmath37 in the case where the star formation is regulated by uv feedback , although the star formation history is slightly delayed for smaller value of @xmath37 .
therefore we carry out all of the simulations with @xmath38 .
we include thermal feedback by sn explosions .
firstly , in advance of the simulations , we compute population synthesis using @xmath39 @xcite . assuming a salpeter imf with masses of @xmath40-@xmath41 and an instantaneous starburst
, we obtain a specific sn rate as a function of a stellar particle age . according to the rate
, we feed the neighboring 50 particles around the star particle with sn energy in thermal form , assuming a released energy of @xmath42 for one sn explosion .
chemical feedback is not included in this study .
the inclusion of metal enrichment enhances cooling rate .
in addition , dust grains likely play an important role ; @xmath0 formation on the surface of dust grains would increases @xmath0 fraction and the absorption of uv photons by the grains leads to a decreased uv photons escape fraction
. we will include chemical feedback in a forthcoming paper .
we use the rhd code start . here
we briefly describe the rt method implemented in the code .
the implementation of ray - tracing in start is basically the same as that adopted in rsph , which has been used for simulations of structure formation in the early universe @xcite .
since the rays are cast from all radiation sources to all sph particles , the rsph scheme allows us to solve rhd problems very accurately @xcite .
moreover , this ray - tracing algorithm enables us to solve rt without losing the spatial resolution of hydrodynamics calculation , since sph particles are directly used for rt grids . with rsph ,
however , the calculation cost is proportional to a number of radiation sources , and it becomes difficult to solve problems in which numerous radiation sources should be considered . in order to avoid this difficulty , a tree - based acceleration algorithm is employed in start . in the tree - based acceleration algorithm , we first make an oct - tree structure @xcite for radiation sources . to make the oct - tree structure , we create the root cell which contains all radiation sources , and recursively divide the cell into eight cells down to the level where each cell contains one radiation source
. then we check if a cell is distant enough from a target sph particle by using following criteria ; @xmath43 where @xmath44 is the distance between the cell and the target sph particle , @xmath45 is the size of the cell , and @xmath46 is the tolerance parameter .
if the above condition is satisfied , the radiation sources in the cell are regarded as one luminous virtual source .
adopting this algorithm , the calculation cost is dramatically reduced .
we have to adopt a sufficiently small value of @xmath46 to obtain accurate results .
@xmath47 is a reasonable choice ( @xcite , see also @xcite ) .
we consider the photo - ionisation of hi , hei , and heii by solving the transfer of ionising photons appropriately .
the cross - sections are respectively given by @xmath48 @xmath49 { \rm cm^2},\ ] ] @xmath50 where @xmath51 , @xmath52 , and @xmath53 are the lyman limit frequencies of hi , hei , and heii , respectively .
the corresponding photo - ionisation rates are expressed as @xmath54 @xmath55 @xmath56 as shown in previous studies @xcite , integrated photoionisation rates can be obtained knowing the optical depths at the three lyman limit frequencies only , since we already know the shape of intrinsic intensity @xmath57 and the dependence of the cross - sections on frequency .
we produce a three dimensional , i.e. , a function of @xmath58 , @xmath59 , and @xmath60 , look - up table by integrating photo - ionisation rates over all frequency ranges as a preprocessing of the simulations . by using the look - up table we can appropriately take into account the dependence of their cross - sections on frequency .
a similar method is applied to calculate photo - heating rates .
hydrogen molecules are easily dissociated by lyman - werner ( lw ) band ( 11.2 - 13.6ev ) photons via the two step solomon process .
we take the photo - dissociation of @xmath0 into account by calculating the column density of @xmath0 and adopting the self - shielding function introduced by @xcite .
then the flux in the lw band can be written as @xmath61 where @xmath62 is the incident flux , @xmath63 is the h@xmath64 column density in units of @xmath65 , and @xmath66 we also include photo - detachment of @xmath67 and photo - dissociation of @xmath68 , assuming these species are optically thin against the radiation owing to their considerably smaller abundances .
these rates can be expressed as @xmath69 where @xmath70ev and @xmath71ev .
the cross - sections are taken from @xcite and @xcite , respectively . here
we should note that the self - shielding function for @xmath0 photo - dissociation was derived with the assumption that a medium is static .
consequently the shielding effect might be stronger than in the case where velocity gradients are properly included .
therefore we additionally simulate two extreme cases , by changing the treatment of @xmath0 photo - dissociation ; ( i ) a run in which we never use the shielding function ( @xmath72 molecules are optically thin in the lw band ) .
( ii ) a run in which photo - dissociation processes ( not only @xmath73 photo - dissociation but also @xmath67 photo - detachment and @xmath74 photo - dissociation processes ) are excluded .
we verify the importances of the photo - dissociation processes by performing these two runs .
we compute the hi , hei and heii ionising photon production rates in @xmath39 @xcite , in the same way that we evaluate the specific sn rate as a function of stellar age .
we plot the photon production rates as a function of stellar age in fig.[rate ] . here , for simplicity , we assume that the shape of spectral energy distribution ( sed ) always corresponds to that of a black body spectrum with an effective temperature of 50,000k .
the assumption is reasonable due to the following reasons ; ( i ) uv radiation is dominantly emitted by young stars .
( ii ) with the imf we choose , the sed of young stellar population can be well approximated by a black body with an effective temperature of 50,000k @xcite .
evaluating reaction rates between uv radiation and matters by the way described above , we implicitly solve a non - equilibrium chemistry regarding nine species , i.e. , @xmath75 , @xmath76 , @xmath77 , @xmath67 , @xmath0 , @xmath68 , @xmath78 , @xmath79 , and @xmath80 .
the initial fractions are taken from @xcite .
finally , there is a synergy effect between sn and uv feedback .
it is often true that thermal feedback by sn explosions can not work efficiently , because radiative cooling processes efficiently work in a high density regions , which are the sites of star formation , and gas in these regions turns out to be cooled quickly . in our simulations ,
a timestep is basically much smaller than the lifetime of massive stars . hence uv feedback works prior to the first sn explosion .
the uv radiation from newly born stars heats the adjacent medium , and the gas density decreases before thermal energy released by sn explosions is poured into the gas .
consequently , sn thermal feedback works more efficiently when uv feedback is included .
each simulation starts from @xmath81 with initial conditions generated with grafic-2 @xcite .
we performed simulations with two different resolutions . in high resolution runs
, we handled @xmath82 particles each for sph and dm particles in a volume of comoving @xmath83 on a side .
in this situation , the particle masses correspond to @xmath84 for sph and @xmath85 for dm , respectively .
hence a small galaxy of @xmath86 can be resolved with 100 particles .
we note that this mass resolution is @xmath87 times better than that in the highest resolution run of @xcite .
the gravitational softening length is set to @xmath88 , which corresponds to 595 comoving pc . in low resolution runs , on the other hand
, we handled @xmath89 particles each for sph and dm particles in the same size box .
hence the minimum halo mass in the low resolution run is @xmath90 which is simply 8 times higher than that in the high resolution run .
since the spatial resolutions of both the rt and hydrodynamics calculations are sufficiently smaller than the typical size of small galaxies , we can appropriately take into account internal uv feedback in each galaxy .
the spatial resolution of the high resolution rt simulation is roughly equivalent to those of previous numerical works that have evaluated ionising photon escape fractions of ionising photons from galactic halos @xcite .
although photon losses in galaxies can automatically be taken into account owing to the high spatial resolution , there still might be absorptions in the spatially unresolved ism .
@xcite have adopted an `` ism '' escape fraction @xmath91 to account for such unresolved absorptions .
we note that this `` ism '' escape fraction has a completely different physical meaning from the `` galactic '' escape fraction which has been usually utilized in previous simulations on reionisation @xcite . in our simulations
we basically work with @xmath92 , but we also carry out simulations with lower values of the ism escape fraction
. we should emphasize that the variance of galactic escape fractions connected to the different properties of each halo is automatically taken into account even if we employ an ism escape fraction .
the validity and dependence of our results on the ism escape fraction will be discussed in section [ escape ] .
parameter sets for all runs are summarized in table1 .
in this section , we show the evolution of global quantities , such as ionised fraction and star formation rate density .
we would like to point out that our simulation box is too small to mitigate the effect of cosmic variance on these quantities .
although the presented global quantities can not be considered to be representative of the whole universe , these quantities are very useful to understand the roles of uv radiation feedback . in addition
, these quantities can be used for the comparison with the results of similar studies . to visually understand a simulated reionization process
, we show the simulated maps of hydrogen number densities , gas temperatures , and ionization degrees at redshifts of 11.1 , 7.58 , and 5.99 in fig .
this is for the rn256l5rt run , but a similar evolution can be seen in other runs .
after stars are born at the early epoch of @xmath93 , well known `` butterfly - shaped '' ionised regions associated with star forming regions appear .
each ionized region grows , and the half of the simulation volume is ionized at @xmath94 .
the reionization is finally completed by @xmath4 . in the top panel of fig.[ionsfdhis ] , we show the simulated ionisation histories .
as we can see from the figure , in both cases of the high and low resolution run , the mass weighted average ionisation fractions of hydrogen are slightly higher than volume weighted ones .
this trend is indicating that high density regions are reionised prior to lower density regions , as in previous studies @xcite
. obviously , ionised fractions at high redshifts are higher in the high resolution run , since the first uv source tends to appear at higher redshift as the mass resolution gets higher . in the bottom panel of fig .
[ ionsfdhis ] , we represent the simulated star formation rate densities ( sfrds ) .
we also plot the sfrds inferred from observed luminosity functions of high-@xmath3 drop - out galaxies in the figure @xcite . it has often been argued that stars are overproduced in numerical simulations
actually our simulations also seem to overproduce stars , if we exclude uv feedback .
it is interesting that the sfrds in both runs are suppressed by a factor of @xmath95 at @xmath96 if we include uv feedback .
this strong suppression makes the simulated sfrds match the sfrds inferred by observations at @xmath97 . at @xmath98 ,
the simulated sfrds remain inconsistent with the observation .
however whether simulations and observations are consistent or not might be inconsequential at this stage , because there are some uncertainty in the both of the numerical and observational studies . concerning our study , as mentioned above , the volume size of the simulation is too small to be regarded as a representative of the universe .
moreover some processes expected to be important for the star formation process , that will be discussed in a later section , havent been included yet . as for observational studies , on the other hand , the shape and amplitude of the luminosity function of faint galaxies are still uncertain especially at high-@xmath3 .
in addition , the fact that our small simulation box does not contain luminous galaxies that are directly observed illustrates the mismatch between our simulations and the observations .
thus here the point to which we should pay attention is that uv feedback significantly affects the cosmic star formation history .
comparing the results of similar previous studies , such as @xcite and @xcite , we find that the suppression seen in our simulation is considerably stronger than those shown in these studies .
we should mention that the former have not appropriately considered the internal uv feedback , while the latter do not have a sufficient mass resolution to discuss the suppression .
also none of these two studies take into account @xmath0 chemistry .
we study the reasons for our strong suppression of star formation in the following sections . in the lower panel of fig .
[ ionsfdhis ] , we can see that the sfrd in rn256l5rt at @xmath99 is lower than that in rn128l5rt , in spite of the fact that sfrds in the corresponding runs without uv feedback , namely rn256l5 and rn128l5 , are almost identical at this epoch .
relating to the discrepancy of the sfrds , the ionization history at the later stage in rn256l5rt is delayed compared to that in rn128l5rt , as shown in the upper panel of fig .
[ ionsfdhis ] .
these discrepancies are thought to be caused by numerical resolution effects . in section [ resef ] ,
we discuss it in detail . in the previous subsection
, we showed that the cosmic star formation rate is strongly suppressed by uv feedback effects . in this section
we analyze some properties of the halos to uncover the reasons for the suppressed star formation . to identify individual halos
, we first use the friends - of - friends method with a linking length of @xmath100 .
then , gradually shrinking a sphere of given radius , we refine the location of the centre of a halo to be the mass centre of it .
when the density within the sphere is equal to the virial density , we define the radius as the virial radius of the halo .
we define the ratio of the amount of baryon particles within a halo s virial radius to the total halo mass as the baryon mass fraction of the halo .
since the suppression of the star formation is the most noticeable at @xmath101 , we here focus on halos at this epoch . in the upper panel of fig.[fb_halo ] , we compare the baryon fractions in the rn256l5rt , rn256l5 , rn128l5rt , and rn128l5 runs . before we show the results in detail , we should mention the spatial resolution of the presented simulations . as described above ,
before the feedback processes start to work , there is a sufficient number of sph particles to resolve the internal feedback in each halo .
however , once the internal feedback starts to work , sph particles are evacuated from halos . as a result ,
the spatial resolutions of the hydrodynamics and rt calculations , especially for low mass halos , gradually decreases as time goes by . as shown by previous studies ,
at least 100 sph particles are required to resolve each halo @xcite .
the baryon mass fraction which satisfies this critical condition can be given by @xmath102 for the high resolution runs .
the corresponding equation for the low resolution runs is easily obtained by multiplying the above equation by eight .
we trace the corresponding lines in the upper panel of fig.[fb_halo ] .
then we can see that the internal structures of halos with @xmath103 ( or @xmath104 in the low resolution run ) can not be discussed with our simulations .
indeed the baryon mass fractions of halos with @xmath105 in rn128l5rt show larger deviations than those in rn256l5rt does .
it is thought to result from a very small number of gas particles in the halos .
therefore we should consider the internal quantities of halos less massive than these mass scales to be unreliable
. roughly speaking , the baryon mass fractions of halos with @xmath106 are insensitive to uv feedback , though the baryon mass fractions are slightly smaller than the value expected from the cosmological parameters @xmath107 . on the other hand , the discrepancy of baryon mass fractions between the runs with and without uv feedback is remarkable for halos with @xmath108 . at @xmath109 , the characteristic mass scale below which photo - evaporation effectively takes place
is roughly consistent with the value shown by @xcite , although they assume an homogeneous external uv feedback and do not consider internal uv feedback and self - shielding .
this consistency likely comes from the following facts : ( i ) the temperature of ionised gas is not changed dramatically even if internal uv feedback is included , ( ii ) for less massive halos , self - shielding is not as efficient as for massive halos @xcite .
therefore , as far as we consider the epoch when the universe is almost ionised , the characteristic mass investigated with a uniform uv background intensity would be approximately correct .
but we should mention here that the internal structure of halos is significantly affected by the internal uv feedback even in massive halos as we will show in section [ impact ] .
we represent the star formation rate as function of halo mass in the lower panel of fig.[fb_halo ] . as expected from the top panel of fig.[fb_halo ] , the star formation rates of halos with @xmath110 is strongly suppressed via photo - evaporation .
an interesting result can be seen in the star formation rates of massive halos .
the star formation in such massive halos is also suppressed , despite the fact that the massive halos hardly lose any mass . in the following two subsections ,
we show reasons for this phenomenon . a possible mechanism which causes the suppression is the decrease of gas density driven by the internal feedback .
soon after the birth of stars in a dense cooled region , the stars emit ionising photons which quickly heat up the surrounding gas up to @xmath9k .
this process leads to the destruction of the dense star - forming region by the high thermal pressure .
also the propagation of the d - type ionisation front makes blow - out by sn explosions effective @xcite . according to the equation ( 1 ) ,
the star formation rate is proportional to @xmath111 .
therefore if the gas density is decreased , this directly indicates that the further star formation in the region is suppressed .
we choose one massive halo with @xmath112 from rn256l5rt .
we also pick the same halo from rn256l5 , and compare them in fig .
[ internal ] .
hereafter we call the former halo - r , and the latter halo - h . in each halo , more than 20,000 sph particles are included .
thus we can discuss their internal density structures . as we can see from fig .
[ internal ] , the internal density structures of these two halos are quite different
. some clumpy structures can be seen in halo - h , that are significantly smoothened by the uv feedback in halo - r . to emphasize the difference in the internal density structures more quantitatively , we show cumulative histograms of the mass fraction where the overdensity @xmath113 exceeds a given value in fig.[clump ] . here
we define the overdensity as @xmath114 , where @xmath115 is the average gas mass density of the massive halo . as shown in fig .
[ clump ] , halo - h has more high density components than halo - r .
this is the direct cause for the suppressed star formation in the massive halo .
the destruction of clumpy structures likely works more effectively in less massive halos . since a massive halo forms through coalescences of smaller halos which are sensitive to the uv feedback in a cdm cosmology , resolving previous episodes of star formation
would leads to stronger suppression of the star formation .
indeed , in our simulations , the sensitivities of the star formation in massive halos to the uv feedback are higher than those shown by @xcite .
it is also expected that the internal uv feedback affects the ionizing photon escape fraction from a galaxy , since the escape fractions strongly depend on the internal structure of a galaxy , such as clumsiness and the covering factor of clumps @xcite .
in addition , stars are usually born in a high density region where a high recombination rate is expected . hence if we do nt take into account the back - reaction by uv radiation emitted by newly born stars , the ionising photons hardly escape from the star forming regions . in @xcite , we have explored the difference between the ionizing photon escape fractions from halo - r and from halo - h , by solving the rt of ionising photons in post - processing . as a result , we have found that the escape fractions of the hi , hei , and heii ionising photons from halo - r are respectively 0.31 , 0.35 , and 0.13 .
these values were indeed higher that those from halo - h by a factor of 2 .
our mass resolution does not allow as to resolve the structure of the low mass clouds where stars are born .
therefore we should consider the possibility that the results presented in this section would be altered if we increase the numerical resolution .
recent high resolution rhd simulations by @xcite have shown that the dynamical effect of photo - ionisation disrupts low - mass star - forming clouds with @xmath116 within a few myr .
this result indicates that further star formation in these regions is suppressed and the feedback enhances the escape fractions of ionizing photons from the star - forming clouds .
thus we expected that the results in this section are qualitatively unaffected by the numerical resolution , even though the quantitative values might be affected . in previous studies of reionisation by numerical simulations ,
the non - equilibrium chemistry of @xmath0 has often been neglected for simplicity .
as mentioned above , if we want to resolve the internal structure of halos , which is closely related with the star formation rate and with the escape probability of ionising photons , we have to consider cooling processes below @xmath16k .
in the low metallicity environment which is prevalent in the early universe , @xmath0 molecules are the most efficient coolants in the low temperature regime .
therefore , if the @xmath0 fraction is considerably decreased , the gas can not cool down to @xmath117k and stars are hardly formed . in this section ,
we show how severely the @xmath0 fraction of each halo is affected by uv feedback , i.e. , the photo - ionisation and photo - dissociation processes .
before we show the results in detail , we briefly introduce some processes closely related to the determination of @xmath0 fraction before we discuss the results .
the destruction of @xmath0 molecules is caused not only by photo - dissociating radiation but also indirectly by photo - ionising radiation . in the photo - ionised hot regions ,
@xmath0 molecules are collisionally dissociated .
moreover , the smoothing of internal density structures shown in the previous section weakens the self - shielding effect . on the other hand
, there is a process which promotes @xmath0 formation . in the absence of dust grains ,
@xmath0 molecules are produced via the @xmath67 process in which electrons work as catalysts .
thus as the @xmath75 fraction increases , @xmath0 formation rate also increases with a given temperature .
actually , the promotion of @xmath0 formation always works when an enhancement of the electron fraction exists , but usually the positive feedback is remarkable only if the photo - ionisation rate is not too high and the medium is only slightly ionized .
otherwise the strong negative photo - ionisation feedback becomes dominant .
as already mentioned , the internal quantities of low mass halos are not reliable due to the very small number of gas particles in the halos
. therefore we will only discuss the internal quantities in massive halos with masses @xmath118 about which we are currently curious .
in addition , we here focus only on high density regions , since stars are expected to form there .
we show the mass weighted average @xmath0 fraction in the high density regions of halos at @xmath109 in fig.[fh2 ] . to emphasize the effects of photo - ionisation
, we also plot the mass averaged ionised fraction and temperature in high density regions in fig .
[ tfhii ] .
we define high density regions as the places where @xmath119 , where the definition of @xmath113 is the same as in the previous section .
we show not only the @xmath0 fraction in the rn128l5 and rn128l5rt runs but also in the rn128l5rt - nolw run , in which the photo - dissociation and photo - detachment processes are excluded , in order to evaluate impact of the photo - dissociating radiations . in the massive halos , the @xmath0 fractions in both the rn128l5rt and rn128l5rt - nolw runs are considerably smaller than in the rn128l5 run .
moreover , the difference in the @xmath0 fractions between in the rn128l5rt and rn128l5rt - nolw runs is not so large .
this result tells us two facts .
first , star formation in massive halos is inhibited by the destruction of @xmath0 molecules .
second , the destruction of @xmath0 is caused by photo - ionisation(-heating ) more than by photo - dissociation .
as obviously shown in the upper panel fig .
[ tfhii ] , the mass weighted ionised fraction tends to increase as the halo mass increases , though the ionisation degrees are rather small .
this trend basically originates in the fact that the specific star formation rate becomes higher as the halo mass increases , as shown by the bottom panel of fig [ fb_halo ] .
a similar trend can also be seen in the average temperatures of halos .
the gas in massive halos is heated up to @xmath120k , in spite of the fact that they are not so highly ionised .
it has been known that such a mild photo - heating is caused by high energy photons .
high energy photons can travel longer paths than lower energy photons , since the cross - sections for photoionisation processes become smaller as the energy of incident photons increases ( equations 3 , 4 , and 5 ) .
in addition , the heating rate by one high energy photon is higher than that by a lower energy photon , since the released electron has a larger kinetic energy . as a result , the gas is mildly heated up even if it is almost neutral .
since star formation in massive halos is continuous , the average @xmath0 fractions in these halos would take an almost - equilibrium value . indeed , the average @xmath0 fractions in the massive halos , about @xmath121 as shown in fig .
[ fh2 ] , are roughly consistent with the equilibrium value at @xmath120k with @xmath122 . here
we should mention that such mildly heated regions are thought to be the places where the star formation is originally possible in rn128l5 run .
it is also true that photo - dissociating radiation can affect higher density regions than photo - ionising radiation can , but the amount of gas residing in the regions solely affected by the photo - dissociating radiation are smaller than that in the mildly heated up regions .
hence we conclude that the suppression of star formation in massive halos is mainly caused by the mild photo - heating and the resultant collisional dissociation in addition to the destruction of clumpy structures shown in the previous section .
if we focus on halos with masses @xmath123 , we notice an interesting fact that the @xmath0 fractions in the rn128l5rt - nolw run are slightly higher than those in the rn128l5 run . this trend might result from the positive feedback due to the enhancement of the electron fraction , but we consider these halos not to be sufficiently resolved to draw this conclusion with any certainty .
as we already mentioned in section [ feedback ] , the self - shielding function we employ might lead to a stronger shielding effect than if we correctly solve the transfer of lw band photons .
hence we should analyse how our results change with the treatment of the photo - dissociating process . in fig.[fs_halo ] , we show the star formation rate of halos . as simply expected from fig .
[ fh2 ] , in massive halos , the star formation rates in rn128l5rt and rn128l5rt - nolw are nearly equivalent . on the other hand , for less massive halos ,
the star formation rates in rn128l5rt - nolw are slightly higher than that in rn128l5rt due to the larger fraction of @xmath0 molecules .
we also plot the star formation rates of the rn128l5rt - nfsh run . since we make the optically thin approximation for lw radiations in the rn128l5rt - nfsh run , the photo - dissociation process is maximal in this run .
we find that the discrepancy between the star formation rates in rn128l5rt and rn128l5rt - nfsh runs is a factor of @xmath124 at most .
we also show the resultant reionisation histories in fig.[fs_kd ] .
we find that the difference of the overlapping redshifts , when the average neutral fraction is 0.5 , in the rn128l5rt and rn128l5rt - nfsh runs is just @xmath125 corresponding to @xmath126 .
therefore we expect that our results would not change dramatically if we solved correctly the transfer of lw photons . in this study
, we never assumed a `` galactic '' escape fraction , since our calculation automatically takes the absorption within a halo into account .
we also assumed no photon losses in the subgrid , unresolved medium , and so worked with an ism escape fraction of @xmath92 .
but unfortunately there is no guarantee that our simulations sufficiently resolve absorption in the ism .
hence we first show how our results change with different choices of @xmath91 .
we show ionisation histories simulated with @xmath127 in the upper panel of fig.[fs_fe ] .
as shown in the figure , ionisation history strongly depends on the choice of @xmath91 .
obviously , reionisation proceeds more quickly as @xmath91 increases .
here we should point out an interesting fact ; in spite of the strong dependence of the reionisation history on @xmath91 , the star formation history is almost independent on @xmath91 as we can see in the bottom panel of fig .
[ fs_fe ] .
this implies that the global star formation at such high redshifts is mainly regulated by the internal feedback processes rather than the external uvb , at least in our simulations .
note however that it might not be always true , since our simulation volume is too small to be representative of the universe .
as shown in @xcite , the global reionisation and star formation histories vary considerably such small volumes .
moreover we can not consider uv radiation emitted from distant sources properly because of the limited volume . in fig.[fhi
] , we plot the simulated neutral hydrogen fractions and compare them with neutral hydrogen fractions observationally estimated from the spectra of 19 qsos at @xmath4 @xcite , from ly@xmath7 damping wing of gamma - ray burst 050904 at @xmath128 @xcite , and from the evolution of the lae luminosity function at @xmath129 @xcite . as shown in the figure
, our simulation results and observed neutral hydrogen fractions are well consistent if we assume @xmath91 greater than @xmath130 , otherwise simulated neutral fractions are larger than the observational values .
the ism escape fraction should approach unity as the mass resolution increases . unfortunately it is still unclear how much resolution we need to remove such an uncertainty .
but we think @xmath131 for our simulations is reasonable , because of the following arguments . @xcite
have shown that an ionizing photon escape fraction from a galaxy with mass of @xmath132 , which roughly corresponds to the minimum halo mass in our low resolution run ( rn128l5 series ) , is @xmath133 . the maximum spatial resolution in their simulations
is 0.1pc so that the clumpiness in a galaxy is possibly sufficiently resolved . @xcite
have also commented that the computed escape fractions should be considered as lower limits , since their simulations have not taken into account previous episodes of star formation which likely decreases the baryon mass fraction as shown in this paper . in addition ,
needless to say that the galactic escape fraction should be smaller than the ism escape fraction since the former includes the latter .
moreover , @xcite have shown the escape fractions from clouds with @xmath116 , which roughly corresponds to the mass resolution of our simulations , are typically greater than 0.6 . in @xcite
, we have computed the escape fractions of hi , hei , and heii ionising photons from the massive halo shown in section [ impact ] .
the computed escape fractions are 0.31 , 0.35 , and 0.13 for hei , hei , and heii ionising photons , respectively .
however we still have many unknowns , e.g. , the sufficient mass resolution to evaluate the escape fraction , a typical value of the escape fraction of a halo with a given mass , the dependence of the escape fraction on halo mass and/or redshift , and so on
. we will do a statistical study on the escape fraction to clarify such unknowns in the future .
the measurement of the optical depth for the thomson scattering of cmb photons by electrons released during the reionisation is frequently applied for evaluating the reionisation epoch .
the wmap 7year have shown that the optical depth is @xmath5 , which corresponds a reionisation at @xmath134 under the assumption of instantaneous reionisation .
in this section , we present the optical depth computed from the simulated reionisation history .
it is likely invalid to compare the optical depth computed from our results with the observational value , since the volume size of the simulations is insufficient . however comparing the optical depths computed from different runs
is useful to understand the simulation results .
we evaluated the optical depths in our simulations by @xmath135 where @xmath136 , @xmath137 , and @xmath138 are the speed of light , the corresponding cross - section , and the electron number density at a given redshift , respectively .
since our simulations are stopped at @xmath109 , we can not evaluate electron fractions at @xmath139 with the simulations .
therefore , to compute the optical depth , we assumed that the h and he in the igm at @xmath139 is fully ionised .
note that we probably overestimate the optical depth at @xmath139 with the assumption , since heii reionization was expected to take place at @xmath140 . in fig .
[ tau ] , we show the optical depths computed in several runs , and compare them with the wmap result .
it is obvious that the optical depths of all runs never reach the observed value .
since the first uv sources would appear at higher redshift as the mass resolution increases , we simply speculate that the total optical depth is higher in a higher resolution run . as we can see in fig .
[ tau ] , the above speculation is indeed correct , but the increment of the optical depth is small .
this originates in the fact that the ionisation history in rn256l5rt run is delayed by uv feedback during the later stage ( see fig .
[ ionsfdhis ] ) . as we can see immediately from fig .
[ tau ] , any the optical depths estimated from our results are not consistent with observed value .
however , no conclusions can be drawn from this , since our simulation volume is not sufficient to directly compare the optical depth .
in addition our simulations did nt include pop iii stars which are expected to have a large contribution to the early stage of the reionsation@xcite . in this paper ,
chemical feedback by sn explosions was not included .
the metal enrichment likely plays an important role .
the increase of metal directly leads to an enhancement of the cooling rate owing to additional cooling processes .
moreover , @xmath0 molecules are efficiently produced on the surface of dust grains .
this also leads to an enhancement of cooling rate below @xmath16k .
not only these effects on cooling rates , but also the absorption of uv photons by grains are closely related to the cosmic star formation and reionisation history . for instance
a direct consequence is that the number of ionising photons escaping from halos decreases .
the absorptions of uv photons by grains would alleviates uv feedback , but push the matter outward by the radiation force .
hence the net effect of metal enrichment on the star formation and reionisation history is still uncertain .
moreover , it is expected that the metal enrichment process regulates the transition from pop iii to pop ii star formation mode .
we will study the complicated role of the metal enrichment in a forthcoming paper ( paper ii ) . in this section ,
we discuss how numerical resolution affects our results .
as we can see from the lower panel of fig .
[ ionsfdhis ] , our simulation result have not converged ; the cosmic star formation rates at the later stage in the high resolution rn256l5rt run is lower than those in the low resolution rn128l5rt run . @xcite
have shown that a considerable fraction of the gas in a small galaxy is evaporated by uvb , but the star formation in high density regions is still possible owing to the self - shielding effect . since our simulations
can not resolve some small clouds where the self - shielding effect originally works , galaxies in the low resolution run are thought to be more sensitive to the uv feedback than those in the high resolution run . due to this type of resolution effect
, we might underestimate the baryon mass fractions and the star formation rates especially in low mass galaxies .
however this resolution effect can not explain why the star formation rates at @xmath99 in the high resolution run is lower than those in the low resolution run .
there is another type of resolution effects in our simulations .
as we increase the mass resolution , the epoch when the first source forms is shifted towards higher redshifts , since our numerical resolution is not enough to resolve pop iii halos which are the first collapsed objects in the universe .
this fact indicates that the feedback processes start to work from an earlier epoch in the higher resolution run . since a massive galaxy forms via coalescences of less massive galaxies which are already damaged by previous feedback episodes , the star formation in the high resolution run is likely delayed compared to that in the low resolution run
this trend caused by the latter type of resolution effect is basically consistent with the trend shown in fig.[ionsfdhis ] .
therefore it is very important to increase the numerical resolution up to resolving the the scale of pop iii halo @xmath141 @xcite , not only for understanding the contribution of pop iii halos to the evolution of the universe , but also for removing this type of resolution effect .
we explored the impact of uv feedback on reionisation history using rhd simulations in which star formation regulated by the internal and external uv feedback in each galaxy is appropriately considered . as a result we found that the cosmic star formation rate is considerably decreased due to the uv feedback . in halos with @xmath1 ,
the star formation is mainly suppressed by photo - evaporation and photo - dissociation .
interestingly , more massive halos also suffers from a strong suppression of star formation , although they hardly lose their baryon components by photo - evaporation . for massive halos ,
the suppression by photo - dissociation is not very effective , but the high density gas clumps are significantly damaged by the internal uv and sn feedback . in addition , mild photo - ionisation in the massive halo turns out to keep the gas hot , and consequently @xmath0 molecules are destroyed via collisional dissociation .
we also found that the degree of the suppression is larger in higher resolution run , because the uv feedback works since an earlier stage , and small galaxies that can be resolved only in the high resolution run are more sensitive to the uv feedback .
therefore the reionisation process turns out to be delayed at @xmath99 in the high resolution run , compared to that in the low resolution run .
this trend is indeed a kind of the resolution effect , but this fact is indicating that resolving pop iii halos is essential to compute the correct star formation and reionisation histories .
we also find that the simulated cosmic star formation rate densities and hi fractions at @xmath142 are well consistent with those inferred from observations , even though our simulation volume is too small to be a representative patch of the universe .
we are grateful to the members of lidau meetings held in paris and strasbourg for fruitful discussion .
we also thank to the anonymous referee for valuable comments .
kh thank m. umemura , t. okamoto , and k. yoshikawa for useful comments on the numerical method .
numerical simulations were performed with t2k - tsukuba at centre for computational sciences in university of tsukuba , with cray xt4 at cfca of naoj , and with blue gene / p babel at the cnrs computing center : idris .
this work was supported in part by grant - in - aid for young scientists b : 24740114 , by mext hpci strategic program , and by the french funding agency anr ( anr-09-blan-0030 ) .
99 ahn k. , iliev i. t. , shapiro p. r. , mellema g. , koda j. , mao y. , 2012 , apj , 756 , l16 aubert d. , teyssier r. , 2010 , apj , 724 , 244 baek s. , di matteo p. , semelin , b. , combes , f. , revaz , y .
, 2009 , a&a , 495 , 389 baek s. , semelin b. , di matteo , p. , revaz , y . , combes , f. , 2010 , a&a , 523 , 4 barnes j. , hut p. , 1986
, nat , 324 , 446 bertschinger , e. 2001 , apjs , 137 , 1 bouwens r j. , et al . , 2011 , apj , 737 , 90 ciardi b. , madau p. , 2003 ,
apj , 596,1 ciardi b. , ferrara a. , white s. d m. , 2003 , mnras , 344,l7 dale j. e. , ercolano b. , bonnel i a. , 2012 , mnras , 424 , 377 draine b. t. , bertoldi r. , 1996 , apj , 468 , 269 fan , x. , carilli , c. l. , keating , b. , 2006 , ara&a , 44 , 415 fan , x. , et al . , 2006 , apj , 132 , 117 fernandez e. r. , shull j. m. , 2011 , apj , 731 , 20 finlator k. , @xmath143zel f. , dave r. , oppenheimer d d. , 2009 , mnras , 400 , 1049 finlator k. , dave r. , @xmath143zel f. , 2011 , apj , 743,169 fioc m. , rocca - volmerange b. , 1997 , a&a , 326 , 950 frenk a. , mellema g. , 1994 , a&a , 289 .
937 galli d. , palla f. , 1998 , a&a , 335 , 403 gnedin n. y. , kravtsov a. v. , chen h. w. , 2008 , apj , 672 , 765 gunn , j. e. , peterson , b. a. , 1965 , apj , 142 , 1633 hasegawa k. , umemura m. , 2010 , mnras , 407 , 2632 hasegawa k. , umemura m. , kitayama t. , 2009 , mnras , 397,1338 hasegawa k. , umemura m. , susa h. , 2009 , mnras , 395,1280 iliev i. t. et al .
, 2006 , mnras , 371 , 1057 iliev i. t. , mellema g. , pen u .-
l . , herz , m. , shapiro p. r. , alvarez , m. a. , 2007 , mnras , 369 1625 iliev i. t. , mellema g. , shapiro p. r. , pen u .-
l . , 2007 , mnras , 376 , 534 iliev i. t. et al . , 2009 , mnras , 400 , 1283 jarosik b. c. l. , et al . 2011 , apjs , 192 , 14 johnson j. l. , vecchia c. d. , khochfar s. , 2012 , mnras accepted ( arxive:1206.5824 ) kitayama t. , susa h. , umemura m. , ikeuchi s. , 2001 , mnras , 326 , 1353 kitayama t. , yoshida n. , 2005 , apj , 630 , 675 kitayama t. , yoshida n. , susa h.,umemura m. , 2004 , apj , 613 , 631 lidz , a. , zahn , o. , mcquinn , m. , zaldarriaga , m. , hernquist , l. 2008 , apj , 680 , 962 mellema g. , iliev i. t. , pen u .-
l . , shapiro p. r. , 2006 , mnras , 372 , 679 nakamoto t. , umemura m. , susa h. , 2001 , mnras , 321 , 593 navarro j. f. , steinmetz m. , 2000 , apj , 538 , 477 okamoto t. , gao l. , theuns t. , 2008 , mnras , 390 , 920 okamoto t. , yoshikawa , k. , umemura , m. , 2011 , mnras ouchi , m. , et al . , 2009 , apj , 706 , 1136 ouchi , m. , et al . , 2010 , apj , 723 , 869 petkova m. , springel v. , 2011 , mnras , 412 , 935 razoumov a. o. , sommer - larsen j. , 2010 , apj , 710 , 1239 ricotti m. , gnedin n. y. , shull , m. , 2001 , apj , 560 , 580 saitoh t. r. , daisaka h. , kokubo e. , makino j. , okamoto t. , tomisaka k. , wada k. , yoshida n. , 2008 , pasj , 60 , 667 santos , m. g. , amblard , a. , pritchard , j. , 2008 , apj , 689 , 1 sokasian a. , yoshida n. , abel t. , hernquist l. , springel v. , 2004 , mnras , 350 , 47 stancil , p. c. 1994 , apj , 430 , 360 susa h. , umemura m. , 2004 , apj , 600 , 1 susa h. , 2006 , pasj , 58 , 455 susa h. , 2007 , apj , 659,908 susa h. , 2008 , apj , 684,226 susa h. , umemura m. , 2006 , apj , 645 , 93l susa h. , umemura m. , hasegawa k. , 2009 , apj , 702 , 480 tajiri y. , umemura m. , 1998 , apj , 502 .
59 tenori - tagle g. , bodenheimer p. , noriega - crespo a. , 1986 , in workshop on model nebulae , d. pequinot ed .
( paris : observatoire de paris ) , p.178 tegmark , m.,silk , j. , rees , m. , blanchard , a. , abel , t. , palla , f.,1997 , apj , 474 , 1 thomas , r. m. et al . , mnras , 393 , 32 thoul a. a. , weinberg d. h. , 1996 , apj , 465 , 608 totani , t. , kawai , n. , kosugi , g. , yamada , t. , iye , m. , ohta , k. , hattori , t. , 2006 , pasj , 58 , 485 trac , h. , cen , r. , 2007 , apj , 671 , 1 umemura m. , susa h. , hasegawa k. , suwa t. , semelin b. , 2012 , progress of theoretical and experimental physics accepted .
vonlanthen , p. , semelin , b. , baek , s. , revaz , y .
, 2011 , a&a , 532 , 97 wise j. h. , cen r. , 2009 , apj , 693 , 984 yajima h. , choi , j - h . , nagamine , k. , 2011 , mnras , 412 , 411 yoshida n. , abel t. , hernquist l. , sugiyama n. , 2003 , apj , 592 , 645 | we explore the impacts of ultraviolet ( uv ) radiation feedback on galaxies during the epoch of reionisation by cosmological simulations in which hydrodynamics and the transfer of the h and he ionising photons are consistently coupled .
moreover we take into account @xmath0 non - equilibrium chemistry , including photo - dissociation .
the most striking feature of the simulations is a high spatial resolution for the radiative transfer ( rt ) calculation which enables us to start considering not only external uv feedback processes but also internal uv feedback processes in each galaxy .
we find that the star formation is significantly suppressed due to the internal uv and supernova ( sn ) feedback . in low mass galaxies with @xmath1 , a large amount of gas is evacuated by photo - evaporation as previous studies have shown , which results in the suppression of star formation .
surprisingly , star formation in massive halos is also strongly suppressed despite the fact that these halos hardly lose any gas by photo - evaporation .
the suppression of star formation in massive halos is mainly caused by following two factors ; ( i ) small scale clumpy structures in the galaxies are smoothened by the internal feedback , ( ii ) although the dense gas in the galaxies is mostly neutral , the @xmath0 formation and cooling processes are disturbed by mild photo - heating .
photo - dissociating radiations actually suppress star formation , but the magnitude of the effect is not so large in massive galaxies . even though our simulation volume is too small to be a representative patch of the universe during reionisation , we find that our simulated star formation rate densities and hi fractions at @xmath2 are consistent with those found in observations .
[ firstpage ] early universe - hydrodynamics - radiative transfer |
in the minimal supersymmetric standard model ( mssm ) @xcite , the masses and mixings of the neutralinos and charginos are given by their mass matrices @xcite @xmath2 respectively , with @xmath3 , @xmath4 and @xmath5 , @xmath6 , with the weak mixing angle @xmath7 .
besides the masses of the @xmath8 and @xmath9 boson , @xmath10 and @xmath11 , respectively , the neutralino and chargino sectors at tree level only depend on the @xmath12 and @xmath13 gaugino masses @xmath0 and @xmath1 , respectively , the higgsino mass parameter @xmath14 , and the ratio @xmath15 of the vacuum expectation values of the two higgs fields .
the neutralino ( chargino ) masses are the square roots of the eigenvalues of @xmath16 ( @xmath17 ) @xcite .
the lep limit on the chargino mass is @xmath18 gev @xcite , from which follows that @xmath19 gev .
if the gut relation @xmath20 is assumed , then @xmath21 gev , such that the lightest neutralino mass is constrained to @xmath22 gev @xcite .
however , if one drops the gut relation , @xmath0 is an independent parameter , allowing to tune the neutralino mass determined from the lowest - order mass matrix @xmath23 freely @xcite .
the neutralino mass is identically zero for @xcite @xmath24\sin(2\beta ) \approx 0.05 \frac{m_z^2}{\mu } = \mathcal{o}(1\,\mathrm{gev } ) .
\label{eq : zeromass } % \end{aligned}\ ] ] ( 200,80 ) ( -5,-10 ) ( 200,-10 ) ( 32,30)@xmath25 gev ( 340,80)@xmath26 ( 365,110)@xmath27 ( 245,108)@xmath28 ( 245,25)@xmath29 ( 230,83)@xmath30 gev ( 320,40)@xmath31 gev for @xmath32 , the neutralino @xmath33 is mainly a bino , i.e. , it couples to hypercharge , and the masses of the other neutralinos and charginos are of the order of @xmath1 and @xmath34 , see fig . [
fig : neutmixandmass ] . in the following ,
we discuss bounds on the neutralino mass from production at lep and from precision observables @xcite , as well as bounds from rare meson decays @xcite .
finally , we summarize bounds from cosmology and astrophysics @xcite .
the opal collaboration @xcite has derived upper bounds on the topological neutralino production cross section @xmath35 at lep with @xmath30 gev , normalized such that @xmath36 .
their observed limit at @xmath37 confidence level in the @xmath38@xmath39 plane is shown in fig .
[ fig : opalbounds](a ) . for @xmath40 gev
, one can roughly read off the upper limit @xmath41 fb , or equivalently , since @xmath42 , @xmath43 fb . this is already a very tight bound , since typical neutralino production cross sections can be of the order of @xmath44 fb . for bino - like neutralinos
, the main contribution to the cross section is due to @xmath45 exchange .
thus one can translate the opal bound on the neutralino production cross section into lower bounds on the selectron mass @xmath46 , for @xmath40 . in fig .
[ fig : opalbounds](b ) we show the contours of @xmath47 in the @xmath14@xmath1 plane , such that along the contours @xmath48 fb .
for example , for a fixed selectron mass of @xmath49 gev , the area below the @xmath50 gev contour in fig .
[ fig : opalbounds](b ) is excluded by lep .
another search channel at lep is radiative neutralino production , @xmath51 . due to the large background from radiative neutrino production @xmath52 , we find that the significance is always @xmath53 for @xmath54 pb@xmath55 and @xmath30 gev @xcite . at the ilc
however , radiative neutralino production will be measurable , due to the significant higher luminosity and the option of polarized beams @xcite .
( 200,100 ) ( 5,-10 ) ( 200,-40 ) ( 40,-5)*(a ) * ( 260,-5)*(b ) *
the invisible @xmath9 width @xmath56 is potentially very sensitive to a light or massless neutralino , due to the contribution from @xmath57 .
however , a light neutralino is mainly bino - like for @xmath58 gev , see fig .
[ fig : neutmixandmass ] . for a pure bino
, the coupling to the @xmath9 boson vanishes at tree level . in fig .
[ fig : invzwidth ] , we show the difference @xmath59 from the measured invisible width @xmath60 mev @xcite , in units of the experimental error @xmath61 mev , to the theoretical prediction @xmath56 . the calculations of @xmath56 include the full @xmath62 sm and mssm contributions , supplemented with leading higher - order terms @xcite .
the deviation from the measured width @xmath63 is larger than @xmath64 only for @xmath65 gev . for decreasing @xmath34
, the increasing higgsino admixture leads to a non - negligible neutralino coupling to the @xmath9 boson .
note that already the sm contribution to @xmath56 is more than @xmath66 larger than the experimental value @xmath63
@xcite .
( 200,150 ) ( -30,0 ) ( 170,0 ) ( 30,-1)*(a ) * ( 240,-1)*(b ) * we have also studied the impact of a massless or light neutralino on the @xmath8 boson mass , the effective leptonic weak mixing angle @xmath67 , the electric dipole moments of the electron , neutron and mercury , and the anomalous magnetic moment of the muon @xmath68 , but have found no significant constraints on the neutralino mass @xcite .
also rare decays like @xmath69 , @xmath70 @xcite , @xmath71 , @xmath72^+ \to \pi^+\tilde\chi^0_1 \tilde\chi^0_1 $ ] , do not constrain @xmath38 @xcite .
the impact of a light neutralino on its thermal relic density has previously been studied @xcite . if the neutralino accounts for the dark matter , its mass has to be @xmath73 gev , in order not to over - close the universe . however
, this bound can be evaded by allowing a small amount of r parity violation @xcite .
one would thus assume that the neutralinos are stable on the time scale of collider experiments , but are not stable on cosmological time scales .
light neutralinos could be thermally produced inside a supernova .
if their mean free path is of the order of the supernova core size or lager , the neutralinos escape freely and lead to an additional cooling of the supernova . to be in agreement with observations of the kamiokande and imb collaborations from sn 1987a , see ref .
@xcite , the cooling must not shorten the neutrino signal .
the energy that is emitted by the neutralinos is much smaller than that emitted by the neutrinos if @xmath74 mev @xcite , with @xmath75 gev . for heavy sleptons , @xmath76 gev ,
however , no bound on the neutralino mass can be set @xcite .
a very light neutralino would be a hot dark matter candidate .
the cowsik - mcclelland bound @xcite gives here @xmath77 ev @xcite , such that a light relativistic neutralino does not disturb the formation of large structures in the universe .
thus , a light or even massless neutralino can be in agreement with constraints from cosmology and astrophysics .
d. choudhury , h. k. dreiner , p. richardson and s. sarkar , phys .
d * 61 * , 095009 ( 2000 ) .
i. gogoladze , j. d. lykken , c. macesanu and s. nandi , phys .
d * 68 * , 073004 ( 2003 ) .
d. hooper and t. plehn , phys .
b * 562 * , 18 ( 2003 ) [ arxiv : hep - ph/0212226 ] .
a. bottino , n. fornengo and s. scopel , phys .
d * 67 * , 063519 ( 2003 ) [ arxiv : hep - ph/0212379 ] .
a. bottino , f. donato , n. fornengo and s. scopel , phys .
d * 68 * , 043506 ( 2003 ) [ arxiv : hep - ph/0304080 ] ; phys .
d * 70 * , 015005 ( 2004 ) [ arxiv : hep - ph/0401186 ] .
g. belanger , f. boudjema , a. cottrant , a. pukhov and s. rosier - lees , jhep * 0403 * , 012 ( 2004 ) [ arxiv : hep - ph/0310037 ] ; g. belanger , f. boudjema , a. pukhov and s. rosier - lees , arxiv : hep - ph/0212227 , j. s. lee and s. scopel , phys .
d * 75 * , 075001 ( 2007 ) [ arxiv : hep - ph/0701221 ] . h. k. dreiner , s. heinemeyer , o. kittel , u. langfeld , a. m. weber , and g. weiglein , in preparation .
g. abbiendi _ et al . _
[ opal collaboration ] , eur .
j. c * 35 * ( 2004 ) 1 [ arxiv : hep - ex/0401026 ] .
h. k. dreiner , o. kittel and u. langenfeld , phys .
d * 74 * ( 2006 ) 115010 [ arxiv : hep - ph/0610020 ] .
h. k. dreiner , o. kittel and u. langenfeld , arxiv : hep - ph/0703009 .
g. a. moortgat - pick _
et al . _ ,
arxiv : hep - ph/0507011 .
the aleph , delphi , l3 , opal , sld collaborations , the lep electroweak working group , + the sld electroweak and heavy flavour groups , hep - ex/0509008 ; hep - ex/0612034 ; lep electroweak working group , see : http://lepewwg.web.cern.ch/lepewwg/welcome.html .
b. mcelrath , phys .
d * 72 * , 103508 ( 2005 ) [ arxiv : hep - ph/0506151 ] .
h. k. dreiner , c. hanhart , u. langenfeld and d. r. phillips , phys . rev .
d * 68 * ( 2003 ) 055004 [ arxiv : hep - ph/0304289 ] ; m. kachelriess , jhep * 0002 * , 010 ( 2000 ) [ arxiv : hep - ph/0001160 ] . | we show that in the minimal supersymmetric standard model , the mass of the lightest neutralino is experimentally unconstrained if the gut relation between the gaugino mass parameters @xmath0 and @xmath1 is dropped .
we discuss what the impact of light or massless neutralinos would be on their production at lep , as well as on electroweak precision data and rare decays .
( 10,10 ) ( 40,260)ippp/07/27 , dcpt/07/54 , bonn - th-2007 - 03 |
cavity quantum electrodynamics ( cqed ) is one of the most important fields of quantum optics , both from a fundamental and an applicative point of view@xcite .
indeed , the interaction of an atom and a single mode of the quantized electromagnetic field inside a high-@xmath0 cavity can be exploited to study the properties of highly non - classical states of the atom - cavity system and to use their non - classical features for quantum information processing@xcite . in this context
it is fundamental to take into account the unavoidable coupling between the atom - cavity system and the environment external to it . in general
, the coupling with an environment gives rise to dissipation and decoherence phenomena@xcite , which can strongly affect the dynamics of the quantum system under study .
in previous papers@xcite we derived a master equation for an atom - cavity system which takes into account from the very beginning a coupling between the atom and the cavity described by the jaynes - cummings ( jc ) model@xcite .
the approach followed in refs . and is different from the one usually reported in the literature to describe cavity losses .
indeed the usual model of cavity losses has been introduced in a phenomenological way , i.e. , one microscopically derives the master equation for the cavity only , described as a quantum harmonic oscillator coupled to a bosonic bath , and then assumes that the presence of the atom inside the cavity does not affect the structure of the dissipator in the master equation@xcite . through the microscopic model derived in refs .
it is possible to give a complete justification of the phenomenological model , which turns out to be valid when the spectrum of the environment is flat and its temperature is zero .
the situation is quite different if the environment has a non - flat spectral density .
this is not an uncommon situation nowadays : indeed there are new - generation cqed experiments which are performed on high - q cavities created inside photonic bandgap ( pbg ) materials@xcite . for this system we proposed a non - markovian model of cavity
losses@xcite which predicts a dynamics completely different from the one predicted by the phenomenological model , giving rise to phenomena such as population trapping due to cavity losses . scope of this paper is to provide some details of the derivation of the non - markovian master equation for the jc model and to show that the main predictions of the theory , namely population trapping due to cavity losses@xcite , does not depend on the shape of the spectral density one chooses , but only on the values of the asymptotic decay rates corresponding to the transitions of interest .
we clarify this point by taking into account two different models of environmental spectral densities and showing that , for the same asymptotic values of the decay rates , the predictions differ only in the details of the very short - time dynamics , while the two densities give the same predictions for longer times .
the paper is structured as follows . in sec . 2 we present the non - markovian model of cavity losses , in sec .
3 we present the calculation of the time - dependent decay rate for a lorentzian spectrum and summarize the dynamics of the system in this case . in sec .
4 the case of a spectrum with a lorentzian gap is presented along with some conclusive remarks .
the system we study consists of a two - level atom interacting with a mode of a cavity coupled to a bosonic environment . calling @xmath1 the bohr frequency of the atom and @xmath2 and @xmath3 its ground and its excited states respectively ,
the interaction between the atom and the cavity mode is described , at resonance and in units of @xmath4 , by the jc hamiltonian@xcite @xmath5 , where @xmath6 ( @xmath7 ) is the creation ( annihilation ) operator of the mode , @xmath8 , @xmath9 , and @xmath10 .
the cavity mode interacts with a bosonic reservoir through an interaction hamiltonian linear in the bosonic operators of the reservoir and of the cavity mode .
more precisely we assume that the dynamics of the total system ( atom - cavity system and environment ) is described by a hamiltonian @xmath11 , where @xmath12 is relative to the atom - cavity system , @xmath13 is relative to the environment and @xmath14 describes the system - environment interaction , with @xmath15 the frequencies of the environment oscillators , @xmath16 ( @xmath17 ) the creation ( annihilation ) operator of quanta in the @xmath18-th environmental mode , and @xmath19 the coupling constants . since the reservoir causing cavity losses
is immersed in the pbg material , we expect its spectrum to be non - flat : then , to be rigorous , the master equation must be derived in the framework of a non - markovian theory . using the second order of the time - convolutionless ( tcl ) expansion@xcite of an exact non - markovian master equation , and neglecting the atomic spontaneous emission and the lamb shifts , it is possible to show that for a weak enough damping the master equation for the system is local in time and that its structure , in the rotating wave approximation ( rwa ) , is the same of the markovian master equation for the same system , with the important difference that now the decay rates are time - dependent .
examples of non - markovian master equations of this type can be found for example in refs . , and .
in our analysis we will focus on the case of one initial excitation only and we will consider a reservoir at zero temperature .
then , the non - markovian master equation for the atom - cavity system density operator @xmath20 , in this case , is the following one@xcite : @xmath21\nonumber\\ \nonumber\\&&+\gamma\left(\omega_0+\omega , t\right ) \left(\frac{1}{2}\left|e_0\right>\left < e_{1,+}\right|\rho(t)\left|e_{1,+}\right>\left < e_0\right| -\frac{1}{4}\left\{\left|e_{1,+}\right>\left < e_{1,+}\right|,\rho(t)\right\}\right)\nonumber\\ \nonumber\\ & & + \gamma\left(\omega_0-\omega , t\right ) \left(\frac{1}{2}\left|e_0\right>\left < e_{1,-}\right|\rho(t)\left|e_{1,-}\right>\left < e_0\right| -\frac{1}{4}\left\{\left|e_{1,-}\right>\left < e_{1,-}\right|,\rho(t)\right\}\right),\end{aligned}\ ] ] where @xmath22 are the eigenstates of @xmath23 with one total excitation , with energy @xmath24 , and @xmath25 is the ground state , with energy @xmath26 . the time - dependent decay rates for @xmath27 and @xmath28 are @xmath29 and @xmath30 respectively . equation ( [ masterequation ] ) is the non - markovian extension of the markovian model introduced , for the same case , in ref . .
by direct substitution , it is possible to show that , if the system starts from the state @xmath31 , i.e. , if the atom is initially excited and the cavity is initially empty , the solution of eq .
( [ masterequation ] ) is the following one : @xmath32 where @xmath33 . from eq .
( [ rhorabinonmarkovian ] ) one can compute all the populations we are going to show in the following .
below we will study the behavior of the non - markovian time - dependent rates @xmath34 , which , through the quantities @xmath35 , lead to non - markovian behavior and to population trapping .
as a first model of environment at zero temperature with non - flat spectrum , we consider the lorentzian distribution @xcite : @xmath36 where @xmath37 is the system - environment coupling strength , and @xmath38 is the width of the distribution , describing also the inverse of the reservoir memory time .
the case of lorentzian spectrum is analytically treatable , while capturing important features of the non - markovian dynamics we are interested in , i.e. , the time - dependence of the decay rates and their different stationary values .
we consider the case in which the spectrum is peaked on the frequency of the state @xmath27 , i.e. , @xmath39 , where @xmath1 is the atomic bohr frequency and @xmath40 is the rabi splitting due to the jc interaction .
the rate @xmath41 for a generic transition with bohr frequency @xmath42 is equal to @xmath43 , where @xmath44 is related to the spectral density @xmath45 through the relation@xcite : @xmath46 this relation gives the right markovian decay rates and lamb shifts for @xmath47 . by performing first the integral with respect to @xmath48 ,
one obtains : @xmath49 which , substituted into eq .
( [ gamma_t_dep ] ) , gives : @xmath50 the second term in eq .
( [ gammatdep_generalj ] ) gives a time - dependent lamb shift which in the following will be neglected .
the first term , i.e. , the real part , is half the time - dependent decay rate @xmath41 .
specializing to the lorentzian spectral density given in eq .
( [ lorentziandensity ] ) , the real part of eq .
( [ gammatdep_generalj ] ) becomes : @xmath51(\omega'-\omega)}\right\},\end{aligned}\ ] ] whose last form is suitable for an evaluation by means of the method of the residues , by closing the integration path with a circle of ray @xmath52 on the upper complex half - plane , where the integral vanishes when @xmath53@xcite .
such an evaluation is straightforward .
it is only worth noting that the pole on the real axis , at @xmath54 , must be circumvented from below , so that its residue is taken for half its value , with positive sign .
the reason of this choice is that , as we will see , it gives a positive stationary value for the decay rate , while the other choice would give a negative stationary decay rate , which would be unphysical . by taking twice the quantity in eq .
( [ regammalorentz ] ) and evaluating the integral , we finally obtain the following expression for the decay rate @xmath41 : @xmath55\mathrm{e}^{-\lambda t}\right\}.\end{aligned}\ ] ] from eq .
( [ gamma_tdep ] ) we clearly see the general behavior of the time - dependent rates : all the rates @xmath41 are zero at @xmath56 , then they oscillate in time with a time - dependent average value , till they reach stationary values for @xmath57 .
these stationary values are proportional to @xmath45 , i.e. , they are equal to the rates one obtains from a markovian theory . for these reasons ,
as anticipated , the quantity @xmath58 can be seen as the memory time of the system - reservoir interaction and non - markovian effects are expected to occur for times shorter than @xmath58 .
in particular , taking the peak of the spectrum in @xmath39 and substituting @xmath59 , one obtains the decay rates for the two dressed states @xmath60 , which , along with the solution of the master equation in eq .
( [ rhorabinonmarkovian ] ) , allow us to compute the dynamics of the atom cavity system presented in ref . ,
to which we refer for the plots of the populations considered .
the population of the ground state of the atom - cavity system @xmath61 increases quadratically for short times , with some oscillations superimposed which are signatures of the oscillations of the decay rates . for larger times ,
i.e. , @xmath57 , it increases in time as the sum of the two exponentials , with rates @xmath62 and @xmath63 respectively .
it is easy to see that the smaller the value of @xmath38 in the spectrum , the smaller the stationary decay rate @xmath64 , while @xmath65 does not change with @xmath38 .
a consequence of this point is the possibility of having situations where , starting with the atom initially excited and the cavity with zero photons , while the state @xmath27 has completely decayed , the population of @xmath28 keeps a constant value for long times , namely the ones in the interval @xmath66 .
a consequence of this fact is that the population of the atom is trapped in the excited state for an amount close to @xmath67 .
one may wonder how strongly this effect could depend on the particular model of spectrum we have chosen .
in fact , the possibility of having a certain amount of population trapping in the system depends on the stationary decay rates only , so that the effect of population trapping involves only the values of the spectrum at the bohr frequencies of the transitions of interest , and not the shape of the spectrum over all the real axis . to clarify this point , in the next section
we show the behavior of the same populations when the spectrum of the environment has a shape different from the one considered in this section .
as a second model for a non - flat spectrum , we take a simple model of structured reservoir , consisting in a lorentzian background with a lorentzian gap@xcite : @xmath68 where the gap is given by the inverted lorentzian peak at frequency @xmath69 . from the calculation of the previous section , it is straightforward to show that each lorentzian peak gives a similar contribution , but with opposite sign , so that the time - dependent decay rate of a transition of bohr frequency @xmath42 is given by : @xmath70\mathrm{e}^{-\lambda_1 t}\right\}\nonumber\\ \nonumber\\ & & -\frac{\alpha_2\lambda_2 ^ 2}{(\omega_2-\omega)^2+\lambda_2 ^ 2}\left\{1+\left[\frac{\omega_2-\omega}{\lambda_2}\sin(\omega_2-\omega)t- \cos(\omega_2-\omega)t\right]\mathrm{e}^{-\lambda_2 t}\right\}.\end{aligned}\ ] ] in the following we will take @xmath71 , and @xmath72 and @xmath73 : in this way we assume that the bohr frequency corresponding to the transition @xmath74 corresponds to the minimum of the spectral density , in analogy with what done in the case of a single lorentzian peak .
the choice of @xmath75 , @xmath76 , @xmath77 and @xmath78 gives a ratio @xmath79 between the stationary values of the two decay rates of interest @xmath80 , a situation which is close to the ideal case of perfect population trapping .
[ cols= " > , < , > , < " , ] as we see from fig .
1-(a ) , the short - time behavior of the rates is rather different from the case of the single lorentzian peak , indeed in the case of the gap the decay rate of @xmath28 initially increases in the same way the rate of @xmath27 does , and , after reaching its maximum value close to @xmath81 , it decays exponentially to its stationary value @xmath82 . anyway
this difference in the behavior of the rates does not lead to observable effects in the dynamics of the atomic populations . indeed , comparing the population of the atomic ground state in fig .
1-(b ) with the corresponding population in ref . , we see that the behavior predicted by both the single - peak lorentzian model and the lorentzian gap model is exactly the same : for a long time a population trapping occurs in the excited atomic state , for an amount of about @xmath67 , no matter which of the two models one chooses . summarizing , in the situation we have analyzed , the essential point to properly choose the model is to check the appropriate ratio between the stationary values of the decay rates . in this sense ,
the most important property is the value of the spectrum at the bohr frequencies of the transitions of interest . on the other hand
, the particular shape of the spectrum one chooses affects only the details of the short - time dynamics , especially the short - time behavior of the decay rates , but the main aspects of the dynamics of the atom - cavity system are not affected by its choice .
we feel that these conclusions are quite general and that they apply to a wide variety of situations involving lossy cavities with non - flat spectra , as in the most advanced quantum electrodynamic systems@xcite .
m.s . thanks the fondazione angelo della riccia for financial support . s.m . ,
acknowledge financial support from the academy of finland ( projects no . 108699 , no .
115682 , and no .
115982 ) , the magnus ehrnrooth foundation and the visl foundation .
a.m. acknowledges partial support by miur project ii04c0e3f3 _ collaborazioni interuniversitarie ed internazionali tipologia c_. | we provide a microscopic derivation for the non - markovian master equation for an atom - cavity system with cavity losses and show that they can induce population trapping in the atomic excited state , when the environment outside the cavity has a non - flat spectrum .
our results apply to hybrid solid state systems and can turn out to be helpful to find the most appropriate description of leakage in the recent developments of cavity quantum electrodynamics . #
1#1 : |
it is a great pleasure to contribute to the volume dedicated to professor e. n. economou with work that attempts to bridge the gap between localization and propagation in complex discrete systems . in the long - standing contributions of prof .
economou , localization is induced via quenched disorder while propagation is a result of either higher dimensions or nonlinearity . in our present work , coherent energy transfer occurs as a result of judicious interlocking of disorder with nonlinearity and an appropriate exploitation of the concept of nonlinear resonance .
resonance is the basic principle for energy propagation in spatially periodic linear systems .
the simplest example in textbooks is obtained for two coupled identical harmonic oscillators . in this case
, it is well known that any amount of energy deposited on one of the two oscillators tunnels back and forth between these two oscillators with a frequency proportional to the coupling . on the other side , in order to keep the energy localized in a given harmonic oscillator , the resonance between this oscillator and the other harmonic oscillators must be broken .
this situation can be reached in a harmonic infinite system for an impurity mode or , more generally , by breaking translational invariance , for instance through strong enough disorder or by quasiperiodic modulations . roughly speaking , finding a harmonic oscillator that is well resonant within the frequency interval @xmath0 with another oscillator
generally requires to go far from the initial one , typically at a distance @xmath1 in a model at dimension @xmath2 .
but since the effective coupling between these two almost resonant oscillators drops exponentially as a function of their distance , the resonance is generally not sufficient for allowing a substantial energy transfer
. this effect blocks energy propagation and causes the anderson localization of the eigenmodes @xcite .
when the system becomes nonlinear , this picture is drastically changed because the frequency of an anharmonic oscillator depends on its amplitude . concerning the time - periodic solutions
, we have shown that nonlinearity plays a double game " @xcite .
first , by adjusting appropriately the amplitude of the nonlinear oscillators , nonlinearity could restore resonances between anharmonic oscillators which were lost in the linearized system .
indeed , small nonlinearity is sufficient to generate exact extended solutions in random systems while the whole linear spectrum is localized .
although these solutions appear spatially intermittent , they can propagate some nonvanishing amount of energy ( much smaller though than standard uniform plane waves ) .
their existence demonstrates that random systems with a small nonlinearity can propagate energy while the same systems without nonlinearity can not .
second , with different amplitudes for the nonlinear oscillators , nonlinearity can also maintain ( or magnify ) the lack of resonance of a localized vibration with the other modes .
thus , the same system can also sustain exact localized nonlinear modes .
moreover , the existence of these modes may not require randomness .
mode localization can also be achieved in a nonlinear spatially periodic system provided the frequency and its harmonics are nonresonant with the extended linear modes ( i.e. , they do not belong to the linear phonon band ) .
these modes are called discrete breathers @xcite .
discrete breathers ( dbs ) , also called intrinsic localized modes , are spatially localized time - periodic solutions of discrete classical nonlinear dynamical systems which could be finite or infinite in arbitrary dimension , spatially periodic or random , etc .
many early works predated their discovery .
for example , polarons are nonlinear objects belonging to the family of dbs introduced long ago by landau @xcite .
the concept of _ local modes _ was independently introduced in chemistry also long ago @xcite for small molecules . in that case ,
discreteness was implicitly necessary .
incidentally , these solutions are recognized to be important for understanding the quantization of molecular vibrations @xcite . renewed interest on these nonlinear concepts but within the framework of integrable models which have exceptional ( but sometimes physically misleading ) properties started during the 1970 s .
the self - trapping equation introduced later in the mid-1980 s already revealed some features characteristic of dbs but only in special cases ( see a recent interesting review by a. scott @xcite ) .
many interesting physical ideas of applications in physics were proposed , but were limited by flaws of the underlying theories .
they could be reconsidered and amended using the more universal concept of db .
it was only in 1988 that sievers and takeno @xcite claimed the existence of dbs as long lifetime solutions in general models , finite or _ infinite _
, on the basis of approximate analytical and numerical calculations ( see review @xcite ) . actually , their claims were definitely confirmed later by a rigorous proof of their existence as exact solutions in klein - gordon systems under rather general conditions of non resonance @xcite .
moreover , dbs were proven to be linearly stable and robust to model perturbations ( i.e. , not restricted to specific models ) .
subsequently , the scope of validity of this proof has been extended to more general models with optical or acoustic phonons and with arbitrary complexity in any dimension @xcite .
this general theory also brought new methods for systematic and accurate numerical investigations @xcite .
the existence of dbs essentially requires _ discreteness and nonlinearity _ ( randomness is not necessary despite the fact that dbs could persist in its presence ) .
thus , these conditions fully take into account two essential characteristics of real matter .
it is discrete , because it is made of atoms , and nonlinear , because the interatomic potentials are nonlinear .
dbs may exist generically as exact solutions when the frequency of a local mode and _ all of its harmonic _ frequencies either belong to phonon gaps or are above the whole phonon spectrum .
otherwise , these local modes would radiate away their energy by the fundamental frequency or some harmonics and thus survive only over a finite time .
however , if the radiation occurs by high order harmonics , the lifetime of this local mode at 0 k may be long but nevertheless finite .
thus , it is clear that db existence as _ exact solution _ crucially depends on the gap structure of the phonon spectrum .
an exact db solution at a given frequency ( or action ) may exist in one model but may disappear in another very similar model just because some high order harmonic becomes resonant with phonons .
another essential property is that dbs come as one parameter family which can be parametrized by their frequency ( or , better , their action which can be used for semiclassical quantization ) .
dbs may have a lot of other amazing properties not fully explored yet but which depend on the considered models .
they scatter elastically linear phonons ( without energy exchange ) like static impurities .
they may filter phonon noise coloring the spectrum @xcite .
inelastic interactions with small amplitude phonons ( generally ) occur at higher order
. then , dbs may slowly decay while in other situations unexpectedly they may grow @xcite by pumping energy from the background .
they are generally pinned to the lattice but may become highly mobile in some special situations @xcite and then they contribute to energy propagation instead of trapping it .
they can also react by fission or fusion @xcite . because they are _ exact solutions _ at 0 k , and the interaction between dbs drops very fast as a function of their distance , at low temperature the only physical way for a quasi - isolated db to decay is by the small interaction with its _ own phonon background_. actually , the real spectrum of this phonon background is essential for determining the true db lifetime .
phenomenological interactions modeled for example by a _
white _ langevin noise and damping ( or , equivalently , a coupling with a large collection of harmonic oscillators with a uniform structureless spectrum ) , systematically destroy the db as an exact solution simply because the phonon spectrum extends to infinity without gaps .
thus , the lifetime obtained by such approaches widely used in physics , is nothing but the result of the assumptions that are made and might be physically misleading .
consequently , when the temperature is sufficiently low ( compared to the db energy ) and the dbs sufficiently far apart , energy trapping by dbs may be substantial and persist over extremely long time in apparent violation of the standard fourier law .
therefore , the most important feature characterizing ( pinned ) dbs ( and distinguishing them from ordinary phonons and local modes ) , is that in appropriate conditions where they can be created ( by external excitations or by other processes breaking initially the thermal equilibrium ) , they may spontaneously appear and _ persist _ as spatially localized dynamical structures with unusually _
long lifetime_. numerical simulations clearly demonstrate that in systems where linearly stable dbs exist , they show up spontaneously under thermal shocks @xcite or when starting from initial states far from thermodynamic equilibrium @xcite .
they trap energy over very long time and generate unusual energy relaxation ( stretched exponentials ) .
this long lifetime of dbs should be relevant for understanding many anomalous energy relaxations especially in disordered matter ( glasses , polymers , biopolymers ) which up to day do not have yet truly satisfactory interpretations .
up to now , experimentalists did not pay much attention to this new paradigm in physics .
one of the reasons might have been that dbs could be easily confused with early concepts inherited from the theory of integrable systems such as ( nondiscrete ) breathers and solitons(perhaps because of the terminology ) .
otherwise , the localization or the anomalous relaxation of energy observed in real experiments , are usually interpreted by randomness , metastable states etc . in spite of the general belief ,
some experimentalists recently observed dbs with different techniques and in different systems , for example in coupled nonlinear optical wave guides @xcite , in coupled arrays of josephson junctions @xcite , in some nonlinear materials @xcite , in magnetic systems @xcite and even possibly in myoglobin @xcite .
after this brief review we now come to the main aim of this paper which is to present a novel property termed targeted db energy transfer that enables possible perfect transfer of energy between two dbs .
this idea was already briefly suggested in @xcite and considered in term of phase dynamics in @xcite .
we give here a precise analytic treatment of this effect through a generalized dimer model .
we should point out that our results do not agree with the scenario imagined in @xcite .
although , this phenomenon can only exist in appropriate models and for appropriate dbs , it could become the key for understanding puzzling phenomena of well - focused energy transfer in physics and especially in biomolecules .
the energy tunneling between two resonant harmonic oscillators is described through the linear superposition of the symmetric and antisymmetric modes .
this is of course not possible when the oscillators are ( strongly ) anharmonic .
we show in this section that we can model the energy transfer in this case by an integrable generalized dimer model .
we shall study more generally the energy transfer between two dbs so that the energy transfer between two oscillators becomes a particular case .
we consider an arbitrary anharmonic system ( d ) ( finite or infinite ) called _ donor molecule _ described by some set of variables and conjugate variables @xmath3 and a hamiltonian @xmath4 .
we assume that this molecule can sustain dbs and select one of them called donor db . as we know
, dbs come as continuous family parametrized for example through their action ( which is @xmath5 integrated on the loop representing the db in the phase space ) .
then , db s energy @xmath6 becomes a function of action and its frequency is @xmath7 .
we also assume that this db family is gapless , i.e. , @xmath8 for @xmath9 when the db is vanishing at zero energy .
we consider a second arbitrary anharmonic system ( a ) ( finite or infinite ) called _ acceptor molecule _ with properties similar to the donor molecule .
we select on the acceptor molecule a family of acceptor db with action @xmath10 and energy @xmath11 . as with the donor system
, we assume that acceptor db is also gapless .
subsequently , we introduce a weak coupling between the donor molecule and the acceptor molecule .
these two systems could physically represent two independent molecules or two linked parts of the same macromolecule relatively far from one another .
the weak coupling interaction could either come from contact ( hydrogen bonds , van der waals forces ) , from ( screened ) coulomb interactions , could be mediated by the solvent , or come directly from the intermediate link of the macromolecule if any , etc .
we would like to find the conditions for complete transfer of the energy of a db from donor molecule with action @xmath12 , after some time , to a db on the acceptor molecule .
since the coupling is weak , this transfer will be slow .
then , at any intermediate time @xmath13 , the dynamics of the system will be well described over a time scale long compared to the period by two dbs with action @xmath14 on the donor and @xmath10 on the acceptor .
this solution is well represented in terms of loop dynamics @xcite as an almost invariant closed loop in the phase space which projects like dbs on the isolated molecules in the complementary donor and the acceptor subspace .
thus , the action of this loop is nothing but the sum of the actions of these two isolated dbs .
because of the weak coupling between donor and acceptor systems , this loop evolves slowly in the phase space .
liouville theorem states that the action of a loop evolving in the phase space is time - constant .
thus , we have @xmath15 another necessary condition is the conservation of the total energy @xmath16 yielding @xmath17 which is a function independent of @xmath14 for @xmath18 . differentiating ( [ eq : consenerg ] ) with respect to @xmath14 readily yields that the frequencies @xmath19 of the donor db and @xmath20 remain identical during the energy transfer , i.e. , the two dbs remain resonant .
actually , this condition ( [ eq : consenerg ] ) does not need to be perfectly fulfilled ( since the coupling energy should also be involved in the energy conservation ) but we shall assume that it is almost fulfilled .
we wish to describe the coupled donor - acceptor system through an effective hamiltonian expressing the energy transfer only as a function of the variables @xmath14 and @xmath10 and their conjugate variables @xmath21 and @xmath22 .
however , the restriction to two pairs of variables will be valid only if condition ( [ eq : consenerg ] ) is not fulfilled precisely or even approximately by any other db of the acceptor molecule for the same total action @xmath12 and the same total energy @xmath23 .
in other words , we assume that the acceptor db is unique . we shall assume also , but only for simplicity , that the donor db is unique .
this condition is imposed for avoiding more complex situations with bifurcating transfer of energy to two or more acceptor db .
it is in principle fulfilled for molecules without any special symmetry . a priori , the small coupling energy , @xmath24 depends on @xmath25 and @xmath26 , on the angle difference
@xmath27 ( conjugate to @xmath28 ) supposed to vary slowly because of the above assumptions ( weak coupling and almost resonance ) , and on the total angle @xmath29 ( conjugate to @xmath30 ) which on the opposite varies fast at the scale of the db periods .
it is then justified to average the coupling over the fast variable @xmath31 ( which varies practically linearly over relatively long time ) and drop the @xmath31 dependence .
next , it is convenient to split the coupling energy as a sum @xmath32 , where @xmath33 is the average of @xmath34 over @xmath35 , while the average of @xmath36 over @xmath35 is zero .
then , the effective hamiltonian regarding the two dbs of the coupled molecules can be written in the form @xmath37 where @xmath38 and @xmath39 which has zero average with respect to @xmath35 . moreover , for @xmath40 , either the acceptor db or the donor db vanishes which removes any @xmath35 dependence .
then , @xmath41 . since this hamiltonian does not depend on @xmath31 , @xmath42 , which implies that the total action @xmath43 is conserved .
because of the averaging over the fast variable @xmath31 , @xmath35 essentially represents the difference between the db phases on the donor and acceptor molecules .
effective hamiltonian ( [ eq : hameff ] ) with @xmath30 a constant parameter is nothing but the hamiltonian for the db phase ( see ref .
@xcite for a formal theory of phase dynamics ) . for donor and acceptor molecules with explicitly given hamiltonians and coupling ,
it is possible to calculate numerically this effective hamiltonian and thus to predict the possibility of complete targeted transfer .
this hamiltonian can be formally written as a generalized discrete nonlinear schrdinger ( dnls ) dimer model when defining the complex variables @xmath44 at the donor site @xmath2 and @xmath45 at the acceptor site @xmath46 .
then , the conservation of the total action @xmath30 is equivalent to the conservation of the total norm @xmath47 .
this approach could be easily modified when the frequency of one of the two dbs is resonant with a harmonic of the other one .
we could then analyze more accurately for dbs the problem of fermi resonance usually considered for quasilinear modes .
this approach can be also extended when there are more than two weakly coupled nonlinear oscillators .
we then obtain a generalized dnls model on a lattice .
it is also extendable for quantum anharmonic oscillators .
we then obtain a quantum generalized dnls model which is nothing but a boson model on a lattice . in this case , the assumption of conservation of the classical action becomes conservation of the number of bosons and nonlinearity becomes many - body interactions . on this basis
we can generalize early calculations based on rotating wave approximations , which were initially restricted to specific models ( e.g. theory of the davydov soliton , see @xcite ) .
expanding the obtained effective hamiltonians at the lowest significant order simply yields standard quartic dnls models .
however , we also learn from our approach that these approximations are only valid in the weak coupling limit , i.e. , close to an anticontinuous limit , which was not clearly stated in the early approaches .
it is thus inconsistent to describe later these dnls models from their continuous limit where they are integrable .
these extended problems will be discussed later in further publications .
we search for solutions of the dynamical system defined by hamiltonian ( [ eq : hameff ] ) , which correspond to a total energy transfer .
these solutions are such that if at the initial time all the energy is on the donor molecule , i.e. , @xmath48 and @xmath49 , it will be completely transferred at a later time @xmath50 on the acceptor molecules , i.e. , @xmath51 and @xmath52 .
then , @xmath53 and @xmath54 .
we say by definition that when this situation occurs , we have targeted energy transfer . since for @xmath55 , as well as for @xmath56 , the angle @xmath35 is undetermined , the topology of the phase space defined by the conjugate variables @xmath57 is the one of a 2d sphere represented in cylindrical coordinates .
any point @xmath58 of this sphere can be represented by the two coordinates which are the angular coordinate @xmath35 ( longitude ) of its projection in the @xmath59 plane and @xmath28 is its @xmath60 coordinate ( latitude ) .
the poles of the sphere are the points where @xmath40 .
there are two time invariants for this hamiltonian which are @xmath30 and the total energy @xmath61 . then , @xmath62 defines contour lines on the sphere @xmath57 which represent the orbits of the dynamical systems in the 2d phase space .
since these contour lines are closed loops , @xmath63 is time - periodic . actually , the solution in the original system consisting of the coupled donor and acceptor molecules , is a two site db the period of which varies periodically with the same period as @xmath63 .
thus , the global solution is quasiperiodic
. the dynamical solutions of the system can be classified in two types .
type 1 corresponds to the orbits on the sphere which are homotopic to zero , i.e. , that can be continuously shrunk to zero without crossing a pole . for these trajectories , @xmath35 oscillates between two bounds .
the limit case where @xmath35 does not oscillate , corresponds to exact dbs .
they correspond to extrema of the energy ( [ eq : orbit ] ) on the sphere at fixed @xmath30 .
type 2 corresponds to the orbits which turn around the axis connecting the two poles .
then , @xmath35 rotates by @xmath64 around the sphere at each period of @xmath63 .
a trajectory corresponding to a targeted energy transfer between the donor db and the acceptor db is such that along the orbit , @xmath63 varies between @xmath30 and @xmath65 .
thus , it is an orbit connecting directly the two poles of the sphere @xmath57 .
it is straightforward to see that a necessary and sufficient condition for having a targeted transfer solution at action @xmath30 , is that the system only accepts type 1 solutions at this total action .
let us explicit more carefully , this condition .
since the two poles must be at the same energy , we have @xmath66 at the uncoupled limit @xmath67 , this condition is fulfilled when ( [ eq : consenerg ] ) is fulfilled .
if @xmath68 is not constant , the contour lines are just circles at constant @xmath28 and are of type 2 and thus , we do not have targeted energy transfer .
when the coupling is turned on , targeted energy transfer requires that there exists an action @xmath12 associated with an energy @xmath23 such that eq.([eq : nlrescond1 ] ) is fulfilled for @xmath69 and @xmath70 .
then , for @xmath70 , this closed orbit on the sphere is defined by implicit equation @xmath71 function @xmath72 vanishes at the poles for @xmath73 and characterizes the detuning between the donor and acceptor dbs during the energy transfer .
we term it _ detuning energy function_. in some sense , @xmath74 extends the concept of peierls - nabarro energy barrier , which concerns only static excitations , to dbs , which are dynamical nontopological excitations .
however , note now that the barrier in energy might be positive or negative . with no restrictions on the choice of the coupling functions @xmath75
, it is rather easy to find sophisticated dependence on @xmath35 ( but a bit unrealistic ) , such that there can not exist any contour line connecting the two poles .
thus , it is convenient to require some properties for the @xmath35 dependence which also make the coupling physically reasonable especially if this coupling is small .
we shall assume that it is sine - like and that is it has only one maximum and one minimum per period for @xmath76 .
this coupling function @xmath75 has two ( and only two ) zeros per period with respect to @xmath35 since its average is zero .
more generally , eq.([eq : orbit2 ] ) determines either two values modulo @xmath64 for @xmath35 or no values . a necessary and sufficient condition for having two solutions @xmath77 and @xmath78 which are continuous functions of @xmath28 on the sphere @xmath57
is simply @xmath79 functions @xmath77 and @xmath78 determine the two segments of the closed contour line which connects the two poles of the sphere @xmath57 .
the corresponding trajectory is a trajectory which realizes a targeted energy transfer .
we note that during the time @xmath50 , the phase difference between the donor and acceptor dbs @xmath80 does not rotate over the trigonometric circle but varies in some interval .
we have just demonstrated ( for a physically reasonable sine - like dependence of the coupling on the phase difference ) that targeted energy transfer solutions do persist when the detuning energy function is not strictly zero provided the detuning energy function is bounded by the max and the min of the coupling function .
when the detuning function is small , this coupling may be also small .
the targeted energy transfer solution at @xmath81 may be lost when varying the model parameters .
it suffices that condition ( [ eq : detuning ] ) is no more fulfilled . at the threshold
, there is a certain value @xmath82 of @xmath28 where @xmath83 .
then , @xmath84 .
consequently , point @xmath85 is an extremum of the energy ( [ eq : orbit ] ) on the sphere so that @xmath86 is time - independent , which in the initial system is time - periodic .
they correspond to exact db solutions of the coupled system made of the donor and acceptor molecules .
thus , the threshold where targeted transfer disappears , can be precisely associated with the appearance of a new db solution of the whole system .
when approaching this threshold , the targeted transfer solution becomes cnoidal , i.e. , the time of transfer diverges because the energy transfer slows down in the vicinity of @xmath82 where the new db solution will appear .
we now investigate explicitly the targeted energy transfer solutions in exactly solvable dimer models which describe quite well the general behavior in the weak coupling limit .
dnls dimer models were studied in many early works @xcite but only in special cases with no special attention to the detuning function .
energy transfer solutions were already found but for relatively large coupling in cases where the detuning function ( [ eq : orbit2 ] ) was not small .
it is thus instructive to reconsider a more general quartic dnls dimer model as an illustration of our general theory .
we consider a dimer with hamiltonian @xmath87 where @xmath88 , @xmath89 , @xmath90 and @xmath91 are parameters .
@xmath92 is the inter - site coupling chosen positive for convenience . @xmath93 and @xmath94 are pairs of conjugate variables . defining first the new pair of conjugate variables ( @xmath95,@xmath96 ) and ( @xmath97,@xmath98 ) by @xmath99 and @xmath100 and next the pair of conjugate variables ( @xmath101 , @xmath102 ) and ( @xmath103 , @xmath104 ) , an equivalent form of hamiltonian ( [ dimhamiltg ] ) is @xmath105 with @xmath106 and the new model parameters @xmath107
complete targeted energy transfer is obtained for @xmath108 the detuning function ( [ eq : orbit2 ] ) and the coupling functions @xmath109 and @xmath110 in eq.([eq : detuning ] ) are @xmath111 so that inequality ( [ eq : detuning ] ) is fulfilled for all @xmath28 when @xmath112 targeted energy transfer is obtained for any nonzero coupling when @xmath113 , i.e. , when the nonlinear coefficients on the two sites are opposite . on the contrary ,
it is never obtained when @xmath114 , i.e. , when the nonlinear coefficients on both sites are equal , except when @xmath115 .
the case @xmath114 , @xmath115 , which was treated in ref.@xcite , is special because @xmath12 is undetermined .
then , there are exact energy transfer solutions for any @xmath30 when the coupling is large enough , i.e. , @xmath116 . exceptionally , energy transfer is not selective in this case .
eq.([eq : orbit ] ) for hamiltonian ( [ eq : energ ] ) yields the model trajectories which appear as contour lines on the sphere @xmath57 defined by @xmath117 they are shown for several values of the total action @xmath30 and for a choice of the model parameters such that targeted energy transfer occurs for a certain initial action @xmath30 and energy @xmath61 ( cf .
fig.[fig1a ] ) .
an example is also shown when targeted energy transfer never occurs at any action @xmath30 and energy @xmath61 ( cf .
fig.[fig1b ] ) . eliminating @xmath35 between eq.([eq : costheta ] ) and the hamilton equation @xmath118 yields @xmath119 where @xmath120 eq.([eq : phsp ] ) describes a single nonlinear oscillator at zero energy with unit mass in a potential @xmath121 . , where @xmath122 is some zero of @xmath121 , which are not solutions of the real dynamical system .
the reason is that the time conservation of @xmath123 is trivial for time - constant solutions but it is not sufficient to insure they fulfill the dynamical equations . ]
the solutions of eq.([eq : phsp ] ) oscillate in an interval determined by two consecutive zeros of @xmath121 and where @xmath121 is negative .
they are determined ( apart from a time shift ) from the knowledge of the two invariants @xmath61 and @xmath30 and the interval of zeros . since in our case
@xmath121 is a fourth degree polynomial , the number of real zeros is even .
note also that all the real zeros @xmath122 of @xmath121 ( @xmath124 ) , necessarily belong to the interval @xmath125 since outside this interval @xmath121 is the sum of two positive terms in eq.([eq : polyn ] ) and thus can not vanish .
the situation where @xmath121 has no zeros never occurs , whatever the initial conditions are , because @xmath126 has to be positive at time @xmath127 .
when @xmath121 has one pair of real zeros , there is a unique solution ( apart from a time shift ) for @xmath63 ( and @xmath128 ) which is time - periodic .
it corresponds to a quasiperiodic solution of the initial dnls equation since @xmath129 when @xmath121 has two pairs of real zeros , there are two time - periodic solutions @xmath63 that , again , correspond to quasiperiodic solutions of the initial dnls equation . time - periodic solutions ( corresponding to dbs ) are obtained when @xmath63 becomes time - constant ( the interval of oscillation shrinks to zero ) , i.e. , when @xmath121 gets a degenerate pair of zeros . for solutions corresponding to a targeted transfer of energy between site @xmath130 and @xmath131
, @xmath63 should oscillate between @xmath132 and @xmath65 which requires that @xmath133 are zeros of @xmath121 .
this condition yields @xmath69 and @xmath70 defined by eqs.([eq : engtg ] ) .
otherwise , @xmath121 must not have any other zeros , yielding for this parameter value @xmath134 which implies @xmath135 equivalent to condition ( [ eq : critcoup ] ) .
then , the solution of eq.([eq : phsp ] ) for @xmath121 given by eq.([eq : polyn2 ] ) is a jacobi elliptic cosine with modulus @xmath136 @xmath137 which describes explicitly the targeted energy transfer between the two sites . then , the frequency of the targeted transfer is @xmath138 where @xmath139 is the first kind complete elliptic integral .
@xmath140 is minimum and proportional to the coupling @xmath92 when @xmath113 , i.e. , when the nonlinear tuning " between the two sites is optimal .
then the oscillation of the difference of action @xmath63 between the two oscillators becomes a pure cosine . in this case , the explicit solution describing the optimal targeted transfer from the donor to the acceptor obtained for @xmath141 and @xmath142 is rather simple @xmath143 fig.[fig2 ] shows the 3d plot of the rate of action transfer @xmath144 between the two sites as a function of the energy @xmath61 and the total action @xmath30 of the solution . of transfer for the action between the two sites of the dimer versus @xmath61 and @xmath30 for hamiltonian ( [ eq : energ ] ) at @xmath113,@xmath145 , @xmath146 , @xmath147 .
coupling @xmath148 is weak .
@xmath149% transfer occurs at the top of the peak while away from the peak , action transfer becomes negligible . ]
the targeted energy transfer solution disappears when its frequency reaches zero , i.e. , @xmath150 becomes equal to @xmath130 .
then , polynomial @xmath121 gets a degenerate zero at @xmath151 which means that a time - periodic solution at @xmath151 appears .
it corresponds to a new db solution which suddenly blocks the targeted transfer . at the limit @xmath152 ,
the elliptic cosine becomes a sech function @xcite and close to this limit , the transfer of energy becomes intermittent .
the time - periodic solutions ( dbs and multibreathers ) are characterized by @xmath153 in eq.([eq : i1 ] ) and @xmath154 yields @xmath155 or @xmath156 and @xmath157 this is a fourth degree equation which may have @xmath131 or @xmath158 real zeros , functions of @xmath30 .
the energy of these solutions @xmath159 determines the boundaries of the domain in the plane of initial conditions @xmath160 that can be realized .
we can also analyze partial energy transfer by considering the set of initial conditions where all the energy is at site @xmath130 .
these conditions are characterized by @xmath53 and @xmath161 .
then , polynomial ( [ eq : polyn ] ) becomes @xmath162 then , @xmath63 oscillates between @xmath30 and the largest zero @xmath163 of this polynomial .
fig.[fig3 ] shows the rate of transfer in action @xmath164 versus @xmath30 for a given set of parameters @xmath165,@xmath166 , @xmath92 and for several values of @xmath167 .
100% transfer is obtained at @xmath168 when condition ( [ eq : critcoup ] ) is fulfilled , i.e. , @xmath169 .
there are discontinuities on these curves where the oscillation regime of @xmath63 becomes cnoidal which are due to the appearance of new time - periodic solutions blocking the energy transfer . for @xmath145 , @xmath147 , @xmath148 and several values of @xmath170 @xmath171 ( thick dashed ) , @xmath172 ( thin
dashed - dotted ) , @xmath127 ( thick full line ) , @xmath173 ( thick dashed ) , @xmath174 ( thin dashed - dotted ) , @xmath175 ( thin full ) . ]
this theory of targeted transfer can be used in practice in complex systems to detect targeted transfer because the functions involved in the effective dimer hamiltonian ( [ eq : hameff ] ) can be explicitly calculated . for that purpose ,
the initial system has to be separated into two coupled subsystems .
the first one will sustain the donor db and the second one the acceptor db .
the coupling between the two systems is not necessarily weak if the considered dbs are far from each other but the effective coupling has to be weak . using the numerical techniques
developed earlier for the calculations of dbs in arbitrarily complex systems @xcite , one chooses a db solution in the isolated donor system and calculate its dynamical configuration @xmath176 , its energy , and its action .
then , varying the frequency , one obtains function @xmath177 .
the same can be done for a db in the acceptor system . for a complete targeted transfer
, we require that both dbs be gapless , i.e. , @xmath177 and @xmath178 can be continued to @xmath151 where @xmath179 .
since the _ effective _ coupling between the two dbs is supposed to be small , @xmath12 can be determined approximately by the equality @xmath180 , which requires that the two curves intersect at @xmath181 . for having targeted transfer at reasonably small coupling
, function @xmath182 should be small .
once these conditions are fulfilled , we have to check that our dimer assumption is valid , i.e. , that there is no other db either on the donor system or the acceptor system with almost the energy functions @xmath177 or @xmath178 in the interval @xmath183 $ ] .
next , the coupling energy between the two systems is considered .
it is originally a function of the global system configuration @xmath3 and @xmath184 it yields the effective coupling in the dimer model by averaging this energy over the fast variable @xmath31 for the system configuration @xmath185 and @xmath186 .
it is thus only a function of @xmath30 , @xmath28 and @xmath35 which has to be small . comparing the precise detuning function and the coupling functions as in eq.([eq : detuning ] )
, we can predict rather accurately the possible existence of targeted transfer solutions between these donor and acceptor dbs . in the cases we tested up to now , where these conditions are correctly fulfilled ,
molecular dynamics confirmed the existence of targeted transfer where almost 100% of the energy oscillates for a very long time between the donor and the acceptor db . as expected also
, this energy transfer is blocked as soon as the initial amount of energy is not optimal or the coupling too weak @xcite .
dbs are out of equilibrium excitations . in real systems , which are not strictly at 0 k
, they interact with the phonon modes and other excitations which should make their lifetime finite .
thermalization should be in principle a consequence of intrinsic nonlinearities of the system .
however , standard approaches usually simulate thermalization and relaxation by a white langevin noise and an extra damping .
this is known to be equivalent to a coupling of the initial system with a fictitious extra system consisting of an infinite set of harmonic oscillators with a uniform distribution of frequencies @xcite .
this is clearly a bias which makes the dbs to systematically radiate energy in this bath and artificially washes out their existence as very long lifetime solutions . actually , the long lifetime of dbs is precisely due to the fact that they have no resonance with the linear phonon spectrum which implies that either their frequencies are above the phonon spectrum or there are gaps in this phonon spectrum .
it is the nonuniform and specific phonon spectrum of the model itself which should be considered for studying db lifetime as we started to do in some recent works @xcite . in other words , real systems
exhibit a selective damping depending on the frequencies of their modes which should be explicitly taken into account to be physically correct .
thus , the thermal bath assumption with a uniform phonon spectrum is not physically founded and , worst , induces the denaturation of the model .
we discussed above the ideal case for targeted energy transfer at 0 k. for obtaining an ideal transfer , besides the adiabaticity hypothesis , the donor and acceptor dbs must be gapless .
this condition requires that there exist no linear modes in the donor ( or the acceptor ) system with frequency in the range of variation of the dbs frequency .
indeed , it has been shown in @xcite that before the frequency of a db collides with a linear mode frequency , it systematically bifurcates with another one . in this case , the db is not strictly continuable to zero amplitude .
there is a frequency discontinuity corresponding to the region where the resonance with the linear mode destroys the db . strictly speaking ,
when there is a linear mode in the range of variation of frequency of a db on the donor ( or acceptor ) system , this db exhibits a nonvanishing energy gap which in principle invalidates our assumptions for describing the coupled system as a generalized dimer model ( [ eq : hameff ] ) . as a consequence , the frequency of the ideal targeted transfer solution precisely oscillates in between the edges of a gap of the linear phonon spectrum .
we would like to briefly discuss here what can happen to targeted energy transfer if there is just a single resonant linear mode in the phonon gap , first in the case of strong coupling , i.e. , this linear mode is spatially located close to the donor ( or the acceptor ) dbs .
as we just said , the db of the donor ( or the acceptor ) will exhibit an energy gap , so that the detuning function can not be defined for the whole interval of variation of @xmath28 .
the energy transfer will become incomplete and irreversible . as the frequency of the targeted energy transfer solution approaches the linear resonance , a substantial fraction of the energy
will be suddenly transferred to this linear mode instead of the acceptor db .
consequently , the frequency of the donor db will change independently of the acceptor db .
the loss of resonance between donor and acceptor db stops immediately the energy transfer .
if one assumes that the energy captured by the linear resonant mode does not return significantly to the donor db , the part of the energy left on the donor system and of the linear mode , could be dissipated into phonons through chaotic processes and finally heat .
then , only a well - defined fraction of the initial energy on the donor has been transferred and the excess of energy has been dissipated into heat .
we obtained only an incomplete but irreversible targeted transfer .
the effect should be much less sharp when the linear mode which is resonant with the donor db is weaker , i.e. , the resonant linear mode is physically located far away from the donor and acceptor dbs , say , in order to fix the ideas , on the donor system . then , there is only a small discontinuity gap in frequency to jump for the donor dbs where function @xmath6 is undefined . we think that despite a small energy loss when crossing the resonance , targeted energy transfer could be continued beyond this resonance and be almost completed .
if there are many such weak resonances they could be described as a frequency dependent damping depending on the distribution and coupling with resonant linear modes .
resonances with bands of extended phonons should have a similar effect .
actually , in realistic systems there are likely some linear modes in the phonon gap causing some energy dissipation and making energy transfer irreversible . on the opposite , at finite temperature , thermal activation of the linear modes
could also help targeted energy transfer by stochastic resonance effects .
these problems will be investigated in the future .
we have demonstrated here that the essential difference between linear and nonlinear resonance ( allowing the targeted energy transfer ) is that nonlinear resonance is _ selective _ in energy ( or action ) , unlike linear resonance , which is not .
targeted energy transfer occurs because the frequencies of the donor oscillator or db and acceptor oscillator or db , although they both vary during the transfer , remain almost equal .
this condition can be achieved at weak coupling , only for a certain initial energy @xmath23 and when a certain detuning function is small enough compared to the coupling energy . in the ideal case ,
an amount of energy @xmath23 deposited on the donor db oscillates slowly and periodically back and forth between donor and acceptor .
it is totally transferred between the donor and the acceptor db at periodic times .
otherwise , although there is some energy oscillation , the energy transfer becomes negligible .
we do not believe that the obtained targeted energy transfer solutions correspond to strictly exact solutions of the initial model because of the hypothesis of adiabaticity which can not be perfectly fulfilled .
however , situations close to the ideal case could exist where the number of energy oscillations between donor and acceptor dbs is very large .
the high sensitivity to the initial energy makes that this energy transfer can be easily stopped by intermediate linear resonance which can be viewed as a kind of transistor effect .
this effect could be used in practical devices ( e.g. , with optical fibers ) .
we may consider situations where , instead of one , there are several dbs on the donor system with almost identical functions @xmath177 and one acceptor db on the acceptor system .
then , if we are in the condition of targeted energy transfer for a single donor db , any donor dbs excited at the energy @xmath23 will transfer their energy to the acceptor db .
however , the transfer has good chances to be irreversible because when the transfer is completed , the acceptor energy could return simultaneously to all donor dbs but then the resonance between the donor dbs and the acceptor db will be broken and the transfer will be blocked .
this situation is interesting for realizing energy funneling analogous to what is observed in chlorophyll @xcite .
we may also consider situations such that for an energy @xmath187 , the acceptor db ( a ) is tuned to another acceptor db ( @xmath188 ) .
then , before the transfer from the initial donor ( d ) to the acceptor ( a ) is completed , this acceptor ( a ) transfer its energy to the next acceptor ( @xmath188 ) with some energy loss ( note that the db @xmath188 acts initially like an intragap linear resonance ) but instead of dissipating energy , it may collect most of the @xmath46 acceptor energy . by this way
, recurrent transfer could be built along a certain well determined sequence of acceptor dbs .
we could then realize irreversible cascades of energy transfer ( with some energy losses however ) .
it is also important to realize now that dbs in molecules may have piezoactive effects , i.e. , they might generate deformation of the molecules as emphasized in @xcite .
this effect could be important for example for understanding the energy conversion of atp into mechanical energy ( molecular motors ) .
the energy deposition from atp could occur first as a db located at a certain receptive site of the moving biomolecule ( e.g. , kinesin adsorbed on a microtubule ) .
then , this db could cascade by targeted energy transfer through a series of dbs located at specific sites of the biomolecule generating a well determined sequence of deformations which lead to a well determined motion of the biomolecule ( step ) along the oriented microtubule after the time @xmath189 necessary for this cascade process . beyond this time , the db energy is finally relaxed into heat .
thus , excitations by atp randomly repeated at time intervals longer than this characteristic time @xmath189 should produce systematically a directive walk of this biomolecule along the microtubule ( this motion should become randomly intermittent in time if the atp concentration is low ) . in the future
, we plan to use and to develop our tools for calculating dbs in nonlinear systems for controlling precisely their targeted energy transfer , funneling , cascades , piezoactivity of dbs and thus for designing artificial models with desired properties mimicking those of biosystems .
let us also note that instead of being atomic vibrations , these dbs could be electronic vibrations ( excitons ) , polarons , or atomic vibrations combined with an electronic charge ( polarobreathers ) @xcite .
thus , there are many possible variations of the targeted transfer principle depending on the considered physical problem ( transfer of energy , charge , spin , etc ) and its scale of energy .
this theory of targeted transfer could be useful , for example , for revisiting the standard theories of fluorescence ( @xcite ) , chemical reactions , catalysis , etc .
we thank r.s .
mackay for frequent and fruitful discussions about dynamical systems .
we also acknowledge m. johansson for his interest and comments concerning this work .
g. kopidakis acknowledges support by greek g.s.r.t .
( e@xmath190et ii ) .
this work has been supported by ec under contract hprn - ct-1999 - 00163 .
anderson , phys .
* 109 * ( 1958 ) 1492 . c. m. soukoulis , e. n. economou , waves random media * 9 * ( 1999 ) 255 .
g. kopidakis , s. aubry , phys .
lett . * 84 * ( 2000 ) 3236 .
g. kopidakis , s. aubry , physica * d 130 * ( 1999 ) 155 .
g. kopidakis , s. aubry , physica * d 139 * ( 2000 ) 247 .
a. j. sievers , s. takeno ( 1988 ) phys .
* 61 * 970 .
campbell , m. peyrard in : d.k .
campbell ( ed . ) , _ chaos - soviet american perspectives on nonlinear science _ , american institute of physics , 1990 .
l. landau , phys .
z. sowjetunion * 3 * ( 1933 ) 664 .
birge , h. sponer , phys .
* 28 * ( 1926 ) 259 .
ellis , phys . rev . * 33 * ( 1929 ) 27
ovchinnikov , n.s .
erikhman , usp.fiz.nauk * 138 * ( 1982 ) 289 ; sov .
* 25 * ( 1982 ) 738 .
farantos , int .
phys . chem . *
15 * ( 1996 ) 345 .
alwyn scott _ nonlinear science , emergence and dynamics of coherent structures _ , oxford university press , oxford , 1999 .
s. flach , c.r .
willis , phys . rep .
* 295 * ( 1998 ) 182 .
mackay , s.aubry , nonlinearity * 7 * ( 1994 ) 1623 .
mackay , j - a .
sepulchre , physica * d 82 * ( 1995 ) 243 .
s. aubry , physica * d 103 * , ( 1997 ) 201 - 250 .
sepulchre , r.s .
mackay , nonlinearity * 10 * ( 1997 ) 679 .
r. livi , m. spicci , r.s .
mackay , nonlinearity * 10 * ( 1997 ) 1421 .
s. aubry , ann .
h. poincar , phys . thor . * 68 * ( 1998 ) 381 .
mari , s. aubry , nonlinearity * 9 * ( 1996 ) 1501 . t. cretegny , s. aubry , s. flach , physica * d 119 * ( 1998 ) 73 .
m. johansson , s. aubry , phys.rev .
* e 61 * ( 2000 ) 5864 .
ding chen , s. aubry , g. tsironis , phys .
letts * 77 * ( 1996 ) 4776 .
s. aubry , t. cretegny , physica * d 119 * ( 1998 ) 34 .
tsironis , s. aubry , phys .
rev . lett .
* 77 * ( 1996 ) 5225 .
a. bikaki , n.k .
voulgarakis , s. aubry , g.p .
tsironis , phys . rev . *
e 59 * ( 1999 ) 1234 .
rasmussen , s. aubry , a.r .
bishop , g.p .
tsironis , eur .
j. * b 15 * ( 2000 ) 169 .
eisenberg , y. silberberg , r. morandotti , a.r .
boyd , j.s .
aitchison , phys .
* 81 * ( 1998 ) 3383 .
r. morandotti , u. peschel , j.s .
aitchison , h.s .
eisenberg , y. silberberg , phys .
* 83 * ( 1999 ) 2726 .
e. trias , j.j .
mazo , t.p .
orlando , phys .
lett . * 84 * ( 2000 ) 741 . p. binder , d. abraimov , a.v .
ustinov , s.flach , y. zolotayuk , phys .
* 84 * ( 2000 ) 745 .
swanson , j.a .
brozik , s.p .
love , g.f .
strouse , a.p .
shreeve , a.r .
bishop , w - z .
wang , m.i .
salkola , phys .
* 82 * ( 1999 ) 3288 .
schwartz , l.q .
english , a.j .
sievers , phys .
* 83 * ( 1999 ) 223 .
a. xie , l. van der meer , w. hoff , r.h .
austin , phys .
* 84 * ( 2000 ) 5435 .
s. aubry , _ discrete breathers in nonlinear lattices : new perspectives in physics and biological physics _ , ( review ) ( 1998 ) unpublished .
t. ahn , r.s .
mackay , j - a .
dynamics of relative phases : generalised multibreathers _ , nonlinear dynamics , to appear .
s. flach , k. kladko , r.s .
mackay , phys .
* 78 * ( 1997 ) 1207 .
floria , j.l .
mari , s. aubry , p.j .
martinez , f. falo , j.j .
mazo , physica * d 113 * ( 1998 ) 387 .
kenkre , d.k .
campbell , phys .
rev.*b 34 * ( 1987 ) 4959 .
kenkre , g.p .
tsironis , phys .
rev.*b 35 * ( 1987 ) 1473 .
gupta , p.a .
sreeram , phys . rev . * b 57 * ( 1998 ) 4358 .
g. kopidakis , s. aubry , g. tsironis , submitted to phys .
r. zwanzig , j.stat.phys .
* 9 * ( 1973 ) 215 .
x. hu , a. damjanovic , t. ritz , k. schulten , p. natl .
( usa ) * 95 * ( 1998 ) 5935 .
j. lakowicz , _ principles of fluorescence spectroscopy _
, plenum press , new york , 1983 . | it is well known that any amount of energy injected in a harmonic oscillator which is resonant and weakly coupled with a second harmonic oscillator , tunnels back and forth between these two oscillators .
when the two oscillators are anharmonic , the amplitude dependence of their frequencies breaks , in general , any eventual initial resonance so that no substantial energy transfer occurs unless , exceptionally , an almost perfect resonance persists .
this paper considers this interesting situation more generally between two discrete breathers belonging to two weakly coupled nonlinear systems , finite or infinite .
a specific amount of energy injected as a discrete breather in a nonlinear system ( donor ) which is weakly coupled to another nonlinear system ( acceptor ) sustaining another discrete breather , might be totally transferred and oscillate back and forth between these donor and acceptor breathers .
the condition is that a certain well defined detuning function is bounded from above and below by two coupling functions .
this targeted energy transfer is selective , i.e. , it only occurs for an initial energy close to a specific value .
the explicit calculation of these functions in complex models with numerical techniques developed earlier for discrete breathers , allows one to detect the existence of possible targeted energy transfer , between which breathers , and at which energy .
it should also help for designing models having desired targeted energy transfer properties at will .
we also show how extra linear resonances could make the energy transfer incomplete and _ irreversible_. future developments of the theory will be able to describe more spectacular effects , such as targeted energy transfer cascades and avalanches , and energy funnels . besides rather short term applications for artificially built devices , this theory might provide an essential clue for understanding puzzling problems of energy kinetics in real materials , chemistry , and bioenergetics .
, , , |
a `` key '' is a semistandard tableau such that the entries in each column also appear in each column to the left .
if a semistandard tableau @xmath1 is not a key itself , then the `` right ( left ) key '' of @xmath1 is a slightly greater ( lesser ) tableau ( via entrywise comparison ) that is associated to @xmath1 .
perhaps the foremost result that uses the notion of the right key of a tableau is a summation formula [ 6 , thm .
1 ] for demazure characters due to lascoux and schtzenberger [ 3 ] . recall a well known description of schur functions : given a partition @xmath2 , the schur function @xmath3 is equal to the sum of the weights of all semistandard tableaux of shape @xmath2 .
more generally , given a highest weight @xmath4 and a permutation @xmath5 , the demazure character ( or `` key polynomial '' ) determined by @xmath4 and @xmath5 of a `` borel subalgebra '' of @xmath0 is equal to the sum of the weights of all semistandard tableaux of shape @xmath4 whose right keys are less than or equal to the key corresponding to @xmath5 .
we now describe a method to quickly write down the right key of a given semistandard tableau @xmath1 .
to do so we need only one special definition beyond the usual terminology .
given a sequence @xmath6 , its _
earliest weakly increasing subsequence _ ( ewis ) is @xmath7 , where @xmath8 and for @xmath9 the index @xmath10 is the smallest index such that @xmath11 .
here is an english description of our method that a non - mathematician could apply : draw an empty young diagram that has the same shape as @xmath1 ; this diagram will receive the entries of the right key of @xmath1 . viewing the bottom entries of the columns of @xmath1 from left to right as forming a sequence , use two index fingers to find the ewis of this sequence . whenever an entry is added to the ewis ,
put a dot above it .
when this ewis ends , write its last member in the diagram for the right key of @xmath1 in the lowest available box in the leftmost available column .
next , repeat the process as if the boxes with dots are no longer part of @xmath1 .
once every box in @xmath1 has a dot , the leftmost column of the right key has been formed . to find the entries of the next column in the right key :
ignore the leftmost column of @xmath1 , erase the dots in the remaining boxes , and repeat the above process .
continue in this manner until the young diagram has been filled with the entries of the right key of @xmath1 .
it can be seen that if there is more than one column of a given length , upon reaching this length one may ignore all but the rightmost of these columns . also , erasing the dots is unnecessary if there is one color of ink available for each distinct column length
. let @xmath1 be the semistandard tableau shown in figure 1 .
first we find the ewiss that begin in the first column of @xmath1 .
to display all six ewiss with just one figure , we use the over - decorations @xmath12{1.00,1.00,1.00}{1}}$ ] , @xmath13{1.00,1.00,1.00}{1}}$ ] , @xmath14{1.00,1.00,1.00}{1}}$ ] , @xmath15{1.00,1.00,1.00}{1}}$ ] , @xmath16{1.00,1.00,1.00}{1}}$ ] and @xmath17{1.00,1.00,1.00}{1}}$ ] to indicate their successive creations and removals . & & & & + & & & & + & & & |8 + & |7 & + 1 & 6 & 6 & 6 & 6 + 4 & 7 & 7 & 7 & 9 + 6 & 8 & 8 & 9 + 7 & 9 & 9 + 8 + 9 so the six ewiss are ( 8,9,9 ) , ( 7,7,8 ) , ( 5,5,6,7 ) , ( 4,5,6 ) , ( 2,3,3,4 ) , and ( 1,1 ) .
hence the entries of the 1st column of the right key of @xmath1 are 9 , 8 , 7 , 6 , 4 , and 1 . to find the ewiss for the later columns
, we repeat the process ( after erasing the over - decorations ) as if the first column is no longer part of @xmath1 .
the four ewiss for the rightmost length 4 column are ( 9,9 ) , ( 6,8 ) , ( 5,7 ) , and ( 3,4,6 ) . continuing the procedure for the length 3 column ,
the three ewiss are ( 8,9 ) , ( 7 ) , and ( 4,6 ) . for the length 2 column ( and in fact the rightmost column of any ssyt ) , the ewiss are just the entries in the column .
writing in the last members from each of these ewiss , we obtain the right key of t , which is also displayed in figure 1 . this paper s main result , theorem 4.5 , states that the output of this method is in fact the right key of @xmath1 .
the proof uses the jeu de taquin description of the right key presented in [ 2 ] .
now we describe the above procedure with mathematical notation . a _ shape _
@xmath18 is a sequence @xmath19 of positive integers such that @xmath20 for some @xmath21 .
each member of a shape is a _ ( column )
length_. the _ young diagram of @xmath18 _ is the diagram with @xmath22 top justified boxes in column @xmath23 for @xmath24 .
semistandard tableau of shape @xmath18 _ is a filling of the young diagram of @xmath18 with positive integers such that the entries are strictly increasing down the columns and weakly increasing across the rows .
henceforth we more simply write _ tableau ( of shape @xmath18 _ ) .
( for lie theoretical purposes , one would bound both the lengths of a shape and the tableau entries by some fixed @xmath25 . ) fix a tableau @xmath1 of shape @xmath18 with @xmath26 columns .
let @xmath27 denote the @xmath23th column of @xmath1 for @xmath24 .
the tableau @xmath1 is a _ key _ if the entries in @xmath27 also appear in @xmath28 for @xmath29 . given a column @xmath30 , let @xmath31 denote its length .
so @xmath32 .
define @xmath33 to be the bottom entry of @xmath27 for @xmath24 .
let @xmath34 for some @xmath35 denote the ewis obtained from @xmath36 .
let @xmath37 denote the collection of boxes in the young diagram of @xmath1 that respectively contain the entries of @xmath38 .
define @xmath39 to be the operator that removes a specified set of boxes ( and their entries ) from a diagram or tableau .
it is easy to see that @xmath40 is a tableau .
write @xmath41 to notate the result of attaching a single entry @xmath42 to the bottom of a column @xmath30 , and @xmath43 to notate the result of prepending a column @xmath30 to the left side of @xmath1 .
the following doubly recursive definition specifies a tableau @xmath44 that has the same shape as @xmath1 ; here @xmath45 denotes the first column of @xmath44 and @xmath46 denotes the bottom entry of @xmath45 .
set @xmath47 ( if @xmath1 is nonempty ; otherwise , for terminating purposes set @xmath48 ) . set @xmath49 .
define @xmath50 .
we call @xmath44 the _ scanning tableau _ of @xmath1 .
the fact that @xmath44 is a key will follow from the main result of this paper .
in this section we present a common definition of the right key of a tableau , as can be found in appendix 5 of [ 2 ]
. we also present a specific method for calculating the right key that uses this definition . given a shape @xmath18 with @xmath26 lengths and a shape @xmath51 with @xmath52 lengths such that @xmath53 for @xmath54 , the _ skew diagram of shape @xmath55 _ is obtained by removing the diagram of @xmath51 from the diagram of @xmath18 .
the sequence of column lengths of @xmath55 is @xmath56
. some of the lengths in a skew diagram may be zero .
skew tableau _ is a filling of a skew diagram with positive integers such that the entries are strictly increasing down the columns and weakly increasing along the rows .
now we assume familiarity with the jeu de taquin ( jdt ) process , as in ( for example ) chapter 1 of [ 2 ] .
rectification _ of a skew tableau [ 2 , p. 15 ] is the tableau that results from successively applying `` jdt slides '' to the `` inside corners '' of the skew tableau ( in any order ) until there are no inside corners left .
it is known that the rectification of a skew tableau is unaffected by any jdt slide or `` reverse jdt ( rjdt ) slide '' .
let @xmath57 be the skew tableau of some shape @xmath55 that is formed by arranging the @xmath26 columns @xmath58 according to @xmath51 .
let @xmath59 and @xmath60 .
using rjdt slides , one may shift all of the entries of each of @xmath61 down @xmath62 rows without otherwise modifying @xmath57 . as @xmath63 runs from 1 to @xmath64 ,
successively perform @xmath62 rjdt slides beginning directly under @xmath65 .
it can be seen that for @xmath66 , each of these rjdt slides will pull each entry of @xmath65 down one row ( without affecting the columns that have already been pulled down ) .
a skew tableau is _ frank _ if its lengths are a rearrangement of the lengths of the shape of its rectification . *
[ 2 , p. 208
] * given a tableau @xmath1 and a skew diagram @xmath55 whose lengths are a rearrangement of the lengths of @xmath1 , there is a unique skew tableau on @xmath55 that rectifies to @xmath1 .
the rightmost column of a frank skew tableau that rectifies to @xmath1 depends only upon the length of that column .
fix a shape @xmath18 with @xmath26 lengths .
let @xmath1 be a tableau of shape @xmath18 . for @xmath24
, define @xmath67 to be the rightmost column of any frank skew tableau whose rightmost column has length @xmath22 and that rectifies to @xmath1 .
right key of _
@xmath1 is @xmath68 .
note that @xmath69 has shape @xmath18 . also , it is known [ 2 , app .
2 ] that @xmath69 is a key . before specifying a method to obtain the individual columns of @xmath69
, we define an operation that will be iterated to produce each of them .
let @xmath57 be a skew tableau of some shape @xmath55 formed from the @xmath26 columns @xmath70 , where @xmath51 has @xmath52 lengths .
for any @xmath71 , set @xmath72 and define @xmath73 to be the number of boxes in @xmath74 that are attached to boxes in @xmath65 . for such @xmath63 , the process of successively shifting @xmath75 down @xmath73 rows and then successively performing @xmath76 rjdt slides on the outside corner of @xmath77 is called the _ jth length swap ( on @xmath57)_. note that the lengths of the resulting skew tableau are @xmath78 .
let @xmath57 be a frank skew tableau of shape @xmath55 that rectifies to a tableau @xmath1 . for @xmath71 ,
the @xmath63th length swap produces a frank skew tableau @xmath79 that also rectifies to @xmath1 .
rectification is preserved by rjdt slides .
since @xmath57 is frank and the lengths of @xmath79 are a rearrangement of the lengths of @xmath57 , we have that @xmath79 is also frank .
let @xmath1 be a tableau of shape @xmath18 with @xmath26 columns and let @xmath80 . successively performing the @xmath23th , @xmath81st , ... , @xmath82st length swaps produces a frank skew tableau with lengths @xmath83 that rectifies to @xmath1 .
its rightmost column is @xmath67 .
apply lemma 3.4 as @xmath63 runs from @xmath23 to @xmath84 .
the method of producing @xmath69 in this section parallels the construction of @xmath44 in section 2 .
let @xmath18 be a shape with @xmath26 lengths . fix a tableau @xmath1 of shape @xmath18 .
@xmath85 .
proposition 3.5 implies that the calculation of @xmath67 does not involve the first @xmath86 columns of @xmath1
. therefore the calculations for @xmath87 are independent of @xmath88 .
we now focus on producing @xmath89 .
define @xmath90 and for @xmath91 define @xmath92 to be the skew tableau obtained by performing the @xmath63th length swap on @xmath93 .
the shape of @xmath93 has lengths @xmath94 .
therefore @xmath95 .
let @xmath96 denote the bottom entry of @xmath89 . recall that @xmath33 is the bottom entry of @xmath27 for @xmath24 , and that @xmath97 .
for @xmath98 , let @xmath99 denote the bottom entry of @xmath100 .
since @xmath101 , we have that @xmath102 .
note that if @xmath103 , then @xmath104 and hence @xmath105 . for @xmath91 ,
the entry @xmath106 produced by the @xmath63th length swap on @xmath93 is one of two possible entries : let @xmath91 .
+ ( i ) if @xmath107 , then @xmath108 .
+ ( ii ) otherwise , @xmath109 . let @xmath91 . the difference in lengths of the two rightmost columns involved in the @xmath63th length swap on @xmath93 is @xmath110 .
first suppose @xmath107 . in this case , by the semistandard conditions @xmath111 is greater than every entry in @xmath112 except perhaps @xmath113 .
thus the first @xmath114 rjdt slides of the @xmath63th length swap will each pull @xmath111 down one row .
then the @xmath76th rjdt slide will compare @xmath113 to @xmath111 and also pull @xmath111 down .
hence @xmath108 . otherwise , @xmath115 . in this case
, @xmath113 is greater than every entry in @xmath112 . as a result
, the @xmath76th rjdt slide will move @xmath113 to the right regardless of what it is compared to .
thus @xmath109 .
note that _
( ii ) _ happens precisely when the final rjdt slide of the @xmath63th length swap moves the bottom entry of @xmath116 across to become the bottom entry of @xmath117 . in this case , the ewis @xmath38 of @xmath118 `` skips '' @xmath119 . the complementary possibility _ ( i ) _ occurs precisely when a new member is appended to @xmath38 .
hence : @xmath120 .
apply these two observations as @xmath63 run from 1 to @xmath84 in lemma 4.2 to see that @xmath121 is the last member of @xmath38 .
recall that @xmath62 denotes the collection of boxes containing @xmath38 : @xmath122 .
let @xmath91 .
we need to show that the entries of @xmath117 above @xmath106 determined by the @xmath63th length swap on @xmath93 are unaffected by the presence of the entries of @xmath38 in @xmath116 .
( these entries are initially located at @xmath62 in @xmath123 . )
it is clear that : @xmath124 for @xmath91 , the presence of @xmath113 will not affect whether or not the entries in @xmath116 above it will be pulled to the right during the @xmath63th length swap . since @xmath125 , it can be safely removed from @xmath1 ( and its successor skew tableaux ) . from lemma 4.2 ,
all but the first member of @xmath38 arise as @xmath111 for some @xmath91 .
suppose @xmath126 and @xmath127 . by the proof of lemma 4.2 we have that @xmath111 was pulled straight down from the bottom of @xmath128 to the bottom of @xmath117 .
so the first move in every rjdt slide executed in the @xmath63th length swap pulls @xmath111 down 1 row .
then the remainder of the rjdt slide continues as it would have if @xmath111 was not originally part of @xmath93 .
now @xmath127 , so by @xmath124 it can be safely removed from @xmath1 ( and its successor skew tableaux ) .
similar reasoning shows that @xmath121 can be safely removed when @xmath129 .
so we see that the presence of the entries in @xmath38 does not affect the determination of @xmath130 .
thus once @xmath96 is found , we can remove all of @xmath62 ( and @xmath38 ) from @xmath1 and then find the rest of @xmath89 iteratively without the entries of @xmath38 being present .
let @xmath1 be a tableau .
the right key @xmath69 is equal to the scanning tableau @xmath44 .
the combination of corollary 4.3 , lemma 4.4 , and lemma 4.1 agrees with the doubly recursive definition of @xmath44 .
here is a method to write down the left key of a given semistandard tableau @xmath1 : draw an empty young diagram that has the same shape as @xmath1 .
let @xmath26 be the number of columns of @xmath1 .
let @xmath131 be the sequence defined as follows : let @xmath132 be the bottom entry of the @xmath26th column . for @xmath29 , let @xmath133 be the largest entry in the @xmath134th column that is less than or equal to @xmath135 .
note that by the semistandard row condition , at least one such entry is guaranteed to exist .
put a dot above each entry of @xmath136 .
place @xmath137 in the lowest available box in the rightmost available column of the young diagram . next , repeat the process as if the boxes with dots and all boxes below them are no longer part of @xmath1 .
once such a sequence has been found for each entry in the rightmost column of @xmath1 , the rightmost column of the left key of @xmath1 has been determined . to find the next column in the left key , ignore the rightmost column of @xmath1 , erase all the dots , and repeat the above process .
continue in this manner until the entire young diagram has been filled .
the proof that the tableau produced by this method is the left key of @xmath1 is similar to the proof of the analogous statement concerning the right key given in this paper .
in fact the proof is simpler , since the sequence produced above necessarily contains exactly one entry from each column to the left of its starting point .
lascoux and schtzenberger introduced [ 3 ] the notions of right and left key in definition 2.9 and developed their relation to demazure characters on pp .
132 - 138 .
the paper [ 6 ] by reiner and shimozono explicitly stated the summation formula for the `` key polynomial '' in theorem 1 .
fulton s book [ 2 ] appeared in 1997 . in [ 4 ] lenart stated ( on p. 280 ) that one can use `` reverse column insertion '' to produce the left key of a tableau , restates [ 6 , thm .
1 ] as theorem 4.1 , and discusses how this result may be proved using material from [ 3 ] . in [ 1 ] , aval showed how to compute the left key of a tableau using manipulations of `` sign matrices '' .
then he obtained the right key from the left key by using the notion of `` complement '' of a young tableau .
mason presented a method in [ 5 ] to produce the right key of a tableau using `` semiskyline augmented fillings '' .
both [ 1 ] and [ 5 ] had larger goals than key computations that led to these procedures . the author would like to thank sarah mason for her remarks and encouragement .
reading [ 5 ] led to the creation of the method presented in this paper , whose original proof referred to her corollary 5.1 .
the author would also like to thank vic reiner for suggesting the possibility of the more direct proof given in this paper .
in addition , both mark shimozono and jim haglund provided valuable perspective to the author s advisor through conversations . finally , the author would like to thank his advisor bob proctor for suggesting many of the aspects of this presentation and for his repeated encouragement . among others ,
these expositional aspects include the english description of the scanning method , notating shapes with column lengths , and structuring the proof of the main result to parallel the scanning method . | the right key of a semistandard young tableau is a tool used to find demazure characters for @xmath0 .
this paper gives methods to obtain the right and left keys by inspection of the semistandard young tableau .
october 27 , 2011 |
the electroweak sector of the standard model ( sm ) contains only one cp - violating phase which arises in the cabibbo - kobayashi - maskawa ( ckm ) quark mixing matrix . adding right - handed neutrinos to account for non zero neutrino masses and their mixing opens up a possibility of new cp - violating phases in the maki - nakagawa - sakata ( mns ) lepton mixing matrix .
in both cases unitarity imposes constraints on mixing matrices , which can be represented graphically as unitarity triangles .
after many years of experimentation with the @xmath1 and @xmath2 mesons , the ckm unitarity triangle @xmath3 is essentially reconstructed and the required amount of cp violation can be accommodated within the sm .
( reconstructing mns triangles will be far more difficult . )
the cp violation in the sm also induces electric dipole moments ( edm ) of elementary particles .
however , in the lepton sector could generate a very large neutron edm , unless @xmath4 is tuned to be very small . ]
they are induced at a multiloop level and , as a result , edm s generated by the ckm phase are extremely small @xcite and beyond current and foreseeable future experiments .
therefore the observation of a lepton edm would be a clear signal of new physics beyond the standard model .
while the cp properties of the @xmath1 and @xmath2 systems appear to be consistent with the sm , another ( indirect ) piece of evidence of cp violation , the baryon asymmetry in the universe , requires a new source of cp violation beyond what is in the sm @xcite .
thus new cp - violating phases must exist in nature .
supersymmetric extensions of the sm based on soft supersymmetry breaking mechanism introduce a plethora of cp phases .
this poses a susy cp problem ( assuming even that the strong cp is solved ) , since if the phases are large @xmath5 , susy contributions to the lepton edm can be too large to satisfy current experimental constraints @xcite .
many models have been proposed @xcite to overcome this problem : fine tune phases to be small , push sparticle spectra above a few tev to suppress effects of large phases on the edm , constrain phases present in the first two generations to be small , arrange for internal cancellations etc . in the absence of any reliable theory that forces in a natural way the phases to be vanishing or small ,
it is mandatory to consider scenarios with some of the phases large and arranged consistent with experimental edm data . in such scenarios
many phenomena will be affected : sparticle masses , their decay rates and production cross sections , susy contributions to sm processes etc .
it is one of the main physics goals of future collider experiments to find susy and verify its cp properties @xcite .
detailed analyses of the neutralino sector can prove particularly useful in this respect . in this note
we will discuss the cp properties of neutralinos strengthening an argument of ref.@xcite that measurements of the neutralino production cross sections can provide a qualitative , unambiguous evidence for non trivial cp phases .
in the minimal supersymmetric extension of the standard model ( mssm ) , the mass matrix of the spin-1/2 partners of the neutral gauge bosons , @xmath6 and @xmath7 , and of the neutral higgs bosons , @xmath8 and @xmath9 , takes the form @xmath10 0 & m_2 & m_z c_\beta c_w & -m_z s_\beta c_w\\[2 mm ] -m_z c_\beta s_w & m_z c_\beta c_w & 0 & -\mu \\[2 mm ] m_z s_\beta s_w & -m_z s_\beta c_w & -\mu & 0 \end{array}\right)\ \label{eq : massmatrix}\end{aligned}\ ] ] here @xmath11 and @xmath12 are the fundamental supersymmetry breaking parameters : the u(1 ) and su(2 ) gaugino masses , and @xmath13 is the higgsino mass parameter . as a result of the electroweak symmetry breaking by the vacuum expectation values of the two neutral higgs fields @xmath14 and @xmath15 ( @xmath16 , @xmath17 where @xmath18 ) , non diagonal terms with @xmath19 and @xmath20 appear and gauginos and higgsinos mix to form the neutralino mass eigenstates @xmath21 ( @xmath22=1,2,3,4 )
. in general the mass parameters @xmath11 , @xmath12 and @xmath13 in the mass matrix ( [ eq : massmatrix ] ) can be complex . by reparameterization of the fields
, @xmath12 can be taken real and positive ; the two remaining non
trivial phases , which are therefore reparameterization invariant , may be attributed to @xmath11 and @xmath13 : @xmath23 since the existence of cp violating phases in supersymmetric theories in general induces electric dipole moments ( edm ) , current experimental bounds can be exploited to derive indirect limits on the parameter space @xcite .
in fact the experimental limits on edm s of the electron , neutron and mercury atom have been used to ( partly ) justify the assumption of real susy parameters , and most phenomenological studies on supersymmetric particle searches have been performed within the cp - conserving mssm .
however , the edm constraints can be avoided assuming masses of the first and second generation sfermions large ( above the tev scale ) , or arranging cancellations between the different susy contributions to the edms . as a result ,
the complex phase of the higgsino mass parameter @xmath13 is much less restricted than previously assumed , while the complex phase of @xmath11 is practically unconstrained . the possibility of non zero cp
phases should therefore be included in phenomenological analyses .
the neutralino mass eigenvalues @xmath24 @xmath25 can be chosen _ positive _ by a suitable definition of the unitary mixing matrix @xmath26 . in general this matrix
involves 6 angles and 10 phases , and can be written as @xcite @xmath27 where @xmath28 are rotations in the [ @xmath29 plane characterized by a mixing angle @xmath30 and a ( dirac ) phase @xmath31 .
for example , @xmath32 with @xmath33 , @xmath34
. one of ( majorana ) phases @xmath35 is nonphysical and , for example , @xmath36 may be chosen to vanish .
none of the remaining 9 phases can be removed by rotating the fields since neutralinos are majorana fermions .
neutralino sector is cp conserving if @xmath13 and @xmath11 are real , which is equivalent to @xmath37 ( mod @xmath38 ) and @xmath39 ( mod @xmath40 ) .
majorana phases @xmath41 indicate only different cp parities of the neutralino states @xcite .
the matrix elements of @xmath26 define the couplings of the mass eigenstates @xmath42 to other particles .
like in the quark sector , it is useful @xcite to represent the unitarity constraints on the elements @xmath43 @xmath44 in terms of unitarity quadrangles . for @xmath45 ,
the above equations define the @xmath46- and @xmath47-type quadrangles in the complex plane .
the @xmath46-type quadrangles are formed by the sides @xmath48 connecting two rows @xmath22 and @xmath49 , and the @xmath47-type by @xmath50 connecting two columns @xmath22 and @xmath49 of the mixing matrix . by a proper ordering of sides
the quadrangles are assumed to be convex with areas given by @xmath51&=&{\textstyle \frac{1}{4 } } ( |j_{ij}^{12}|+|j_{ij}^{23}|+|j_{ij}^{34}|+|j_{ij}^{41}| ) \label{eq : am}\\ { \rm area}[d_{ij}]&=&{\textstyle \frac{1}{4 } } ( |j_{12}^{ij}|+|j_{23}^{ij}|+|j_{34}^{ij}|+|j_{41}^{ij}| ) \label{eq : ad}\end{aligned}\ ] ] where @xmath52 are the jarlskog type cp odd `` plaquettes '' @xcite @xmath53 note that plaquettes , and therefore the areas of unitarity quadrangles , are not sensitive to the majorana phases @xmath35 . unlike in the quark or lepton sector ,
the orientation of all quadrangles is physically meaningful , and determined by the cp - phases of the neutralino mass matrix . in terms of quadrangles , cp is conserved if and only if all quadrangles have null area ( collapse to lines or points ) _ and _ are oriented along either real or imaginary axis .
in principle , the imaginary parts of the complex parameters involved could most directly and unambiguously be determined by measuring suitable @xmath54 violating observables .
neutralinos can copiously be produced at prospective @xmath0 linear colliders @xcite via the @xmath55-channel @xmath56 exchange and @xmath57- and @xmath58-channel selectron exchange .
the polarized differential cross section for the @xmath59 production is given by @xcite @xmath60\ \label{eq : diffx}\end{aligned}\ ] ] where @xmath61=@xmath62 [ @xmath63=@xmath64 is the electron [ positron ] polarization vector ; the electron momentum direction defines the @xmath65axis and the electron transverse polarization
vector the @xmath66axis ; @xmath67[1-(\mu_i-\mu_j)^2]$ ] with @xmath68 .
the coefficients @xmath69 , @xmath70 , @xmath71 and @xmath72 depend only on the polar angle @xmath4 and their explicit form is given in @xcite .
an interesting feature of neutralino production is encoded in the term @xmath72 of eq.([eq : diffx ] ) . unlike @xmath69 , @xmath70 and @xmath71 ,
the @xmath72 is a function of plaquettes @xmath73 \label{sigman}\end{aligned}\ ] ] where tilde means that in calculating @xmath74 the @xmath75 should be replaced by @xmath76 .
the combinations of propagator factors @xmath77 are @xmath78 with @xmath79 , @xmath80 and the @xmath56-boson propagator @xmath81 is taken real by neglecting the @xmath56 width in the limit of high energies .
therefore @xmath72 is nonvanishing only in cp noninvariant theories and already
the detailed measurement of the angular distribution of produced neutralino pairs in collisions of transverse polarized beams could indicate the presence of cp phases .
however , since @xmath72 depends on plaquettes , nonvanishing @xmath72 requires specific form of cp - violation : the area of some unitarity quadrangles has to be non
i.e. _ at least one of dirac phases @xmath82 . moreover , the effect might be quite small due to cancellations between @xmath83 and @xmath84 higgsino components of @xmath42 and @xmath85 in eq.([sigman ] ) , and between t- and u - channel selectron exchanges in eq.([eq : normal1 ] ) .
if the initial beams are not polarized , the cp phases could be inferred from the @xmath86 component of the polarization of the @xmath87 pairs produced in @xmath0 annihilation @xcite .
the polarization vector @xmath88 is defined in the rest frame of the particle @xmath42 , with components parallel to the @xmath42 flight direction in the c.m .
frame , in the production plane , and normal to the production plane , respectively
. the normal component @xmath86 can only be generated by complex production amplitudes in the non
diagonal @xmath59 pair production process with @xmath45 .
for example , the contribution to @xmath86 from the @xmath89 exchange reads @xmath90\end{aligned}\ ] ] the normal polarization can be non
zero even if all the @xmath91-type cp phases vanish , _
i.e. _ it could signal the existence of non
trivial @xmath92-type cp phases .
the non zero values of cp - odd characteristics @xmath72 or @xmath86 would unambiguously indicate cp - violation in the neutralino sector .
however , their experimental measurements will be very difficult .
on the other hand , one can also try to identify the presence of cp - phases by studying their impact on the cp - even quantities , like neutralino masses , branching ratios etc .
since these quantities are non zero in cp conserving case , the detection of cp - odd phases will require a careful quantitative analysis of a number of physical observables .
in particular , for numerically small cp - odd phases , their deviations from cp - even values will also be small .
as an example , in fig.[fig : quad ] the unitarity quadrangles for a particular point in the parameter space ( consistent with all experimental constraints ) are shown .
the phase of @xmath13 is set to zero , and @xmath93 . in this case
the quadrangles are almost degenerate to lines parallel to either real or imaginary axis , and revealing the phase of @xmath11 will be quite difficult . in this respect , as pointed out in ref.@xcite , a clear indication of non zero cp violating phases can be provided by studying the energy behavior of the cross sections for non diagonal neutralino pair production near thresholds . in cp - invariant theories ,
the cp parity of a pair of majorana fermions @xmath87 is given by @xmath94 where @xmath95 is the cp parity of @xmath42 and @xmath96 is the angular momentum @xcite .
therefore neutralinos with the same cp parities ( for example for @xmath97 ) can be excited only in the p - wave via the s - channel @xmath98 and @xmath56 production processes .
the excitation in the s - wave , with the characteristic steep rise @xmath99 of the cross section near threshold , can occur only for non diagonal pairs with opposite cp
parities of the produced neutralinos @xcite .
the power of the selection rule ( [ cpparity ] ) can clearly be seen by inspecting the expressions for the total cross section @xmath100 ( @xmath45 ) near threshold @xmath101 where @xmath102 and the kinematic functions @xmath103 in the cp invariant theory , the imaginary parts of @xmath43 can only be generated by majorana phases @xmath39 and @xmath104 or vice versa , _
i.e. _ the s wave excitation is possible when the cp parities of the produced neutralinos are opposite , as dictated by the eq.([cpparity ] ) .
this immediately implies that if the @xmath105 and @xmath106 pairs are excited in the s wave , the pair @xmath107 must be excited in the p wave characterized by the slow rise @xmath108 of the cross section , fig.[fig : th ] , left panel .
if , however , cp is violated the angular momentum of the produced neutralino pair is no longer restricted by the eq.([cpparity ] ) and all non diagonal pairs can be excited in the s wave .
this is illustrated in fig.[fig : th ] , where the threshold behavior of the neutralino pairs @xmath109 , @xmath110 and @xmath111 for the cp - conserving ( left panel ) case is contrasted to the cp - violating case ( right panel ) . even for relatively small cp
phase @xmath93 , implying small impact on cp even quantities , the change in the energy dependence near threshold can be quite dramatic .
thus , observing the @xmath105 , @xmath106 and @xmath107 pairs to be excited _ all _ in s wave states would therefore signal cp violation .
the supersymmetric extension of the standard model can come with new sources of cp violation . in the absence of natural suppression of the susy cp
phases , their non zero values have to be considered in phenomenological studies . in this paper
we have discussed the cp properties of neutralinos , which are quite peculiar due to their majorana nature .
the cp violation in the neutralino sector can reveal itself in many different ways .
the most ambitious analysis would require the experimental reconstruction of the unitarity quadrangles .
since all phases of the mixing matrix @xmath26 ( at least at the tree level ) are ultimately determined by the phases of the fundamental parameters @xmath13 and @xmath11 , the reconstruction of the quadrangles would provide many consistency checks of the underlying theory .
on the other hand , the first qualitative indication of the cp violation can be provided by the energy dependence of the neutralino production cross sections .
the steep rise of cross sections for the production of at least three different non diagonal neutralino pairs can be interpreted as a first direct signature of the presence of cp violation in the neutralino sector . + * acknowledgments * + work supported in part by the kbn grant 2 p03b 040 24 ( 2003 - 2005 ) .
cohen , d.b .
kaplan and a.e .
nelson , ann .
rev.nucl . part .
* 43 * ( 1993 ) 27 ; b.m .
gavela , p. hernandez , j. orloff , o. pne and c. quimbay , nucl .
phys.b * 430 * ( 1994 ) 382 ; v.a .
rubakov and m.e .
shaposhnikov , usp .
nauk * 166 * ( 1996 ) 493 .
y. kizukuri and n. oshimo , phys .
b * 249 * ( 1990 ) 449 ; t. ibrahim and p. nath , phys .
d * 58 * ( 1998 ) 111301 [ erratum - ibid .
d * 60 * ( 1998 ) 099902 ] [ hep - ph/9807501 ] and hep ph/0107325 ; t. ibrahim and p. nath , phys .
d * 61 * , 093004 ( 2000 ) [ hep - ph/9910553 ] ; m. brhlik , g.j .
good and g.l .
kane , phys .
d * 59 * ( 1999 ) 115004 ; r. arnowitt , b. dutta and y. santoso , phys .
d * 64 * ( 2001 ) 113010 [ hep - ph/0106089 ] ; t. ibrahim and p. nath , arxiv : hep - ph/0210251 .
tesla technical design report , part : iii physics at an @xmath0 linear collider , _ eds .
_ r .- d .
heuer , d. miller , f. richard and p. zerwas , desy 2001 - 011 , arxiv : hep - ph/0106315 ; t. abe _ et al .
_ [ american linear collider working group collaboration ] , `` linear collider physics resource book for snowmass 2001 . 2 : higgs and supersymmetry studies , '' in _ proc .
of the aps / dpf / dpb summer study on the future of particle physics ( snowmass 2001 ) _ ed .
n. graf , arxiv : hep - ex/0106056 ; k. abe _
et al . _ , jlc roadmap report , presented at the acfa lc symposium , tsukuba , japan 2003 , http://lcdev.kek.jp/rmdraft/ s.y .
choi , j. kalinowski , g. moortgat - pick and p.m. zerwas , eur . phys .
j. c * 22 * 563 ( 2001 ) [ hep - ph/0108117 ] , _ ibid .
_ c * 23 * 769 ( 2002 ) [ hep - ph/0202039 ] .
l. chau and w. keung , phys .
* 53 * ( 1984 ) 1802 ; m. gronau and j. schechter , phys .
* 54 * ( 1985 ) 385 [ erratum - ibid .
* 54 * ( 1985 ) 1209 ] ; h. fritzsch and j. plankl , phys . rev .
d * 35 * ( 1987 ) 1732 .
j. ellis , j. m. frre , j. s. hagelin , g. l. kane and s. t. petcov , phys .
b * 132 * ( 1983 ) 436 .
f. del aguila and j. a. aguilar - saavedra , phys .
b * 386 * ( 1996 ) 241 [ hep - ph/9605418 ] ; j. a. aguilar - saavedra and g. c. branco , phys .
d * 62 * ( 2000 ) 096009 [ hep - ph/0007025 ] .
g. moortgat - pick and h. fraas , phys .
d * 59 * ( 1999 ) 015016 [ hep - ph/9708481 ] ; g. moortgat - pick , h. fraas , a. bartl and w. majerotto , eur .
j. c * 9 * ( 1999 ) 521 [ erratum - ibid .
c * 9 * ( 1999 ) 549 ] [ hep - ph/9903220 ] ; g. moortgat - pick , a. bartl , h. fraas and w. majerotto , eur . phys .
j. c * 18 * ( 2000 ) 379 [ hep - ph/0007222 ] ; s. y. choi , h. s. song and w. y. song , phys .
d * 61 * ( 2000 ) 075004 [ hep - ph/9907474 ] .
g. moortgat - pick and h. fraas , eur .
j. c * 25 * ( 2002 ) 189 [ arxiv : hep - ph/0204333 ] . | supersymmetric extensions of the standard model provide new sources of cp violation . here the cp properties of neutralinos are described and possible experimental signatures of cp violation in the neutralino production processes at @xmath0 linear colliders are discussed .
= by -1 |
the zeta function @xmath10 for a finite simple graph @xmath11 is the entire function @xmath0 defined by the positive eigenvalues @xmath1 of the dirac operator @xmath12 , the square root of the hodge laplacian @xmath13 on discrete differential forms .
we study it in the case of circular graphs @xmath14 , where @xmath9 agrees with the zeta function @xmath15 of the classical scalar laplacian like @xmath16\ ] ] of the circular graph @xmath17 . to prove that all roots of the zeta function of @xmath18 converge for @xmath3 to the line @xmath19
, we show that all roots of @xmath20 accumulate for @xmath3 on the line @xmath21 .
the strategy is to deal with three regions : we first verify that for @xmath22 the analytic functions @xmath23 converge to the classical riemann zeta function which is nonzero there .
then , we show that @xmath24 converges uniformly on compact sets for @xmath25 in the strip @xmath26 , where @xmath27 .
finally , on @xmath28 we use that @xmath29 converges to the average @xmath30 as a riemann sum .
while the later is obvious , the first two statements need some analysis .
our main tool is an elementary but apparently new newton - coates - rolle analysis which considers numerical integration using a new derivative @xmath31 called * k - derivative * which has very similar features to the schwarzian derivative .
the derivative @xmath31 has the property that @xmath31 is bounded for @xmath32 if @xmath33 .
+ we derive some values @xmath34 explicitly as rational numbers using the fact that @xmath35 is a fixed point of the linear birkhoff renormalization map @xmath36 this gives immediately the discrete zeta values like the * discrete basel problem * @xmath37 which recovers in the limit the classical riemann zeta values like the * classical basel problem * @xmath38 .
the exact value @xmath39 in the discrete case is of interest because it is the trace @xmath40 of the green function of the jacobi matrix @xmath18 , the laplacian of the circular graph @xmath41 .
the case @xmath42 is interesting especially because @xmath43 , where @xmath44 is the hurwitz zeta function . and this makes also the relation with the @xmath45 function clear , because @xmath46 by the cot - formula of euler .
+ the paper also includes high precision numerical computations of roots of @xmath47 made to investigate at first where the accumulation points are in the limit @xmath3 .
these computations show that the roots of @xmath47 converge very slowly to the critical line @xmath48 and gave us confidence to attempt to prove the main theorem stated here .
+ we have to stress that while the entire functions @xmath49 approximate the riemann zeta function @xmath50 for @xmath51 , * there is no direct relation * on and below the critical line @xmath52 .
the functions @xmath47 continue to be analytic everywhere , while the classical zeta function @xmath53 continues only to exist through analytic continuation .
so , the result proven here * tells absolutely nothing about the riemann hypothesis*. while for the zeta function of the circle , everything behind the abscissa of convergence is foggy and only visible indirectly by analytic continuation , we can fly in the discrete case with full visibility because we deal with entire functions .
+ there are conceptual relations however between the classical and discrete zeta functions : @xmath47 is the dirac zeta function of the discrete circle @xmath2 and @xmath53 is the dirac zeta function of the circle @xmath54 which is the riemann zeta function . at first
, one would expect that the roots of @xmath47 approach @xmath19 ; but this is not the case : they approach @xmath48 . only the laplace zeta function @xmath9 has the roots near @xmath19 . in the proof
, we use that the hurwitz zeta functions @xmath44 are fixed points of birkhoff renormalization operators @xmath55 , a fact which implies the already mentioned fact that @xmath56 fixed point property as a special case because @xmath57 by euler s cotangent formula .
the fixed point property of the hurwitz zeta function is obvious for @xmath22 and clear for all @xmath25 by analytic continuation .
it is known as a summation formula .
+ actually , we can look at hurwitz zeta functions in the context of a central limit theorem because birkhoff summation can be seen as a normalized sum of random variables even so the random variables @xmath58 in the sum @xmath59 depend on @xmath60 . for smooth functions which are continuous up to the boundary
, the normalized sum converges to a linear function , which is just @xmath61 normalized to have standard deviation @xmath62 . when considered in such a probabilistic setup , hurwitz zeta functions play a similar `` central '' role as the gaussian distribution on @xmath63 , the exponential distribution on @xmath64 or binomial distributions on finite sets .
the symmetric zeta function @xmath65 of a finite simple graph @xmath11 is defined by the nonzero eigenvalues @xmath1 of the dirac matrix @xmath12 of @xmath11 , where @xmath66 is the exterior derivative .
since the nonzero eigenvalues come in pairs @xmath67 , all the important information like the roots of @xmath50 are encoded by the entire function @xmath68 which we call the * dirac zeta function * of @xmath11 .
it is the discrete analogue of the dirac zeta function for manifolds , which for the circle @xmath69 , where @xmath70 , agrees with the * classical riemann zeta function * @xmath71 .
the analytic function @xmath10 has infinitely many roots in general , which appear in computer experiments arranged in a strip of the complex plane .
zeta functions of finite graphs are entire functions which are not trivial to study partly due to the fact that for fixed @xmath72 , the function @xmath73 is a bohr almost periodic function in @xmath74 , the frequencies being related to the dirac eigenvalues of the graph . + there are many reasons to look at zeta functions , both in number theory as well as in physics @xcite .
one motivation can be that in statistical mechanics , one looks at partition functions @xmath75 of a system with energies @xmath1 .
now , @xmath76 and the @xmath25th derivative in a distributional sense is @xmath77 .
if @xmath1 are the positive eigenvalues of a matrix @xmath78 , then @xmath79 is a heat kernel .
an other motivation is that @xmath80 is the pseudo determinant of @xmath78 and that the analytic function , like the characteristic polynomial , encodes the eigenvalues in a way which allows to recover geometric properties of the graph .
as we will see , there are interesting complex analytic questions involved when looking at * graph limits * of * discrete circles * @xmath2 which have the circle as a limit .
+ this article starts to study the zeta function for such circular graphs @xmath2 .
the function @xmath81 is of the form @xmath82 with the complex function @xmath83 and where @xmath84 .
if @xmath25 is in the strip @xmath85 , the function @xmath86 is in @xmath87 , but not bounded .
if @xmath88 , then @xmath89 is a riemann sum @xmath90 for some function @xmath91 on the circle @xmath92 which satisfies @xmath93 .
whenever @xmath94 is continuous like for @xmath95 , the riemann sum @xmath96 converges .
however , if @xmath94 is unbounded , this is not necessarily true any more because some of the points get close to the singularity of @xmath97 .
but if - as in our case - the riemann sum is evaluated on a nice grid which only gets @xmath98 close to a singularity of type @xmath99 , one can say more .
in the critical strip the function is in @xmath100 . to the right of the critical line @xmath48 , the function @xmath101 has the property that @xmath23 converges to the classical riemann zeta function @xmath50 .
number theorists have looked at more general sums .
examples are @xcite .
+ studying the zeta functions @xmath47 for discrete circles @xmath2 allows us to compute values of the classical zeta function @xmath50 in a new way . to do so
, we use a result from @xcite telling that @xmath45 is a fixed point of a birkhoff renormalization map ( [ renormalization ] ) .
we compute @xmath102 explicitly for small @xmath103 , solving the basel problem for circular graphs and as a limit for the circle , where it is the classical basel problem .
while these sums have been summed up before @xcite , the approach taken here seems new .
+ to establish the limit @xmath104/n^s$ ] , we use an adaptation of a newton - coates method for @xmath87-functions which are smooth in the interior of an interval but unbounded at the boundary .
this method works , if the * k - derivative * @xmath105 is bounded @xmath106 for @xmath107 .
the k - derivative has properties similar to the * schwarzian derivative*. it is bounded for functions like @xmath108 or @xmath109 which appear in the zeta function of the discrete circle @xmath2 .
a comparison table below shows the striking similarities with the schwarzian derivative . on the critical line @xmath48
, we still see that @xmath110 stays bounded .
+ various zeta functions have been considered for graphs : there is the discrete analogue of the * minakshisundaraman - pleijel zeta function * , which uses the eigenvalues of the scalar laplacian . in a discrete
setting , this zeta function has been mentioned in @xcite for the scalar san diego laplacian . related is @xmath0 , where @xmath1 are the eigenvalues of a graph laplacian like the * scalar laplacian * @xmath111 , where @xmath112 is the diagonal * degree matrix * and where @xmath78 is the * adjacency matrix * of the graph . in the case of graphs
@xmath2 , the dirac zeta function is not different from the laplace zeta function due to poincar duality which gives a symmetry between 0 and 1-forms so that the laplace and dirac zeta function are related by scaling . if the full laplacian on forms is considered for @xmath2 , this can be written as @xmath113 if @xmath50 is the dirac zeta function for the manifold .
the * ihara zeta function * for graphs is the discrete analogue of the * selberg zeta function * @xcite and is also pretty unrelated to the dirac zeta function considered here .
an other unrelated zeta function is the * graph automorphism zeta function * @xcite for graph automorphisms is a version of the * artin - mazur - ruelle zeta function * in dynamical systems theory @xcite .
we have looked at * almost periodic zeta functions * @xmath114 with periodic function @xmath97 and irrational @xmath115 in @xcite .
an example is the polylogarithm @xmath116 .
since we look here at @xmath117 with rational @xmath115 , one could look at @xmath118 , where @xmath119 are periodic approximations of @xmath115 . for special irrational numbers like the golden mean , there are symmetries and the limiting values can in some sense computed explicitly @xcite .
fixed point equations of riemann sums are a common ground for hurwitz zeta functions , polylogarithms and the @xmath45 function .
+ in this article we also report on experiments which have led to the main question left open : to find the limiting set the roots converge to in the critical strip .
we will prove that the limiting set is a subset of the critical line @xmath48 but we have no idea about the nature of this set , whether it is discrete or a continuum .
figure ( [ situation ] ) shows the motion of the roots from @xmath120 to @xmath121 with linearly increasing @xmath60 and then from @xmath122 to @xmath123 , each time doubling @xmath60 . there is a reason for the doubling choice :
@xmath124 can be written as a sum @xmath47 and a twisted @xmath125 and a small term disappearing in the limit .
this allows to interpolate roots for @xmath60 and @xmath126 with a continuous parameter seeing the motion of the roots in dependence of @xmath60 as a flow on time dependent vector field .
this picture explains why the roots move so slowly when @xmath60 is increased .
for a finite simple graph @xmath127 , the dirac zeta function @xmath128 is an analytic function .
it encodes the spectrum @xmath129 of @xmath130 because the traces of @xmath131 are recoverable as @xmath132 . therefore , because the spectrum of @xmath133 is symmetric with respect to @xmath134 , we can recover the spectrum of @xmath133 from @xmath10 .
the hamburger moment problem assures that the eigenvalues are determined from those traces . for circular graphs
@xmath2 , the eigenvalues are @xmath135 together with two zero eigenvalues which belong by hodge theory to the betti numbers @xmath136 of @xmath2 for @xmath137 .
the zeta function @xmath50 is a riemann sum for the integral @xmath138 besides the circular graph , one could look at other classes of graphs and study the limit .
it is not always interesting as the following examples show : for the complete graph @xmath139 , the nonzero dirac eigenvalues are @xmath140 with multiplicity @xmath141 and @xmath142 do not have any roots . for star graphs @xmath143 , the nonzero dirac eigenvalues are @xmath62 with multiplicity @xmath144 and @xmath140 with multiplicity @xmath62 so that the zeta function is @xmath145 which has a root at @xmath146 .
one particular case one could consider are discrete tori approximating a continuum torus . besides graph limits , we would like to know the statistics of the roots for random graphs .
some pictures of zeta functions of various graphs can be seen in figure ( [ diracgraphexamples ] ) .
as already mentioned , we stick here to the zeta functions of the circular graphs .
it already provides a rich variety of problems , many of which are unsettled .
+ unlike in the continuum , where the zeta function has more information hidden behind the line of convergence - the key being analytic continuation - which allows for example to define notions like ray - singer determinant as @xmath80 , this seems not that interesting at first in the graph case because the pseudo determinant @xcite is already explicitly given as the product of the nonzero eigenvalues .
why then study the zeta functions at all ?
one reason is that for some graph limits like graphs approximating riemannian manifolds , an interesting convergence seems to take place and that in the circle case , to the right of the critical strip , a limit of @xmath47 leads to the classical riemann zeta function @xmath53 which is the dirac zeta function of the circle .
as the riemann hypothesis illustrates , already this function is not understood yet . to the left of the critical strip
, we have convergence too and the affine scaling limit @xmath147 exists in the critical strip , a result which can be seen as a central limit result .
still , the limiting behavior of the roots on the critical line is not explained . while it is possible that more profound relations between the classical riemann zeta function and the circular zeta function , the relation for @xmath148 pointed out here is only a trivial bridge .
it is possible that the fine structure of the limiting set of roots of circular zeta functions encodes information of the riemann zeta function itself and allow to see behind the abscissa of convergence of @xmath50 but there is no doubt that such an analysis would be much more difficult .
+ approximating the classical riemann zeta function by entire functions is possible in many ways .
one has studied partial sums @xmath149 in @xcite .
one can look at zeta functions of triangularizations of compact manifolds @xmath150 and compare them with the zeta functions of the later .
this has been analyzed for the form laplacian by @xcite in dimensions @xmath151 .
we will look at zeta functions of circular graphs @xmath152 and their symmetries in form of fixed points of birkhoff sum renormalization maps ( [ renormalization ] ) , and get so a new derivation of the values of @xmath153 for the classical riemann zeta function . while there is a more general connection for @xmath148 for the zeta function of any riemannian manifold @xmath150 approximated by graphs assured by @xcite , we only look at the circle case here , where everything is explicit
. we will prove \a ) for any @xmath25 with @xmath22 we have point wise @xmath154 b ) for @xmath155 we have point wise convergence @xmath156 c ) for @xmath157 , we have point wise convergence of @xmath158/\sigma(\zeta_n)$ ] to a nonzero function .
+ in all cases , the convergence is uniform on compact subsets of the corresponding open regions .
[ maintheorem ] this will be shown below .
an immediate consequence is : for any compact set @xmath5 and @xmath159 , there exists @xmath160 so that for @xmath161 , the roots of @xmath162 are outside @xmath163 .
the main question is now whether @xmath164 has roots converging to a discrete set , to a cantor type set or to a curve @xmath165 on the critical line @xmath52 .
+ if the roots should converge to a discrete set of isolated points , one could settle this with the argument principle and show that a some complex integral @xmath166 along some rectangular paths do no more move .
besides a curve or discrete set @xmath167 the roots could accumulate as a hausdorff limit on compact set or cantor set but the later is not likely . from the numerical data we would not be surprised
if the limiting set would coincide with the critical line @xmath48 .
+ we have numerically computed contour integrals @xmath168 counting the number of roots of @xmath47 inside rectangles enclosed by @xmath169 to confirm the location of roots . that works reliably but it is more expensive than finding the roots by newton iteration . in principle , one could use methods developed in @xcite to rigorously establish the existence of roots for specific @xmath60 but that would not help since we want to understand the limit @xmath3 . if the roots should settle to a discrete set , then @xcite would be useful to prove that this is the case : look at a contour integral along a small circle along the limiting root and show that it does not change in the limit @xmath3 .
we currently have the impression however that the roots will settle to a continuum curve on @xmath48 .
we list now specific values of @xmath170 as a general rule , it seems that one can compute the values @xmath162 for the discrete circle explicitly , if and only the specific values for the classical riemann zeta function @xmath50 of the circle are known .
so far , we have only anecdotal evidence for such a meta rule . for real integer values @xmath171 , we have @xmath172 where @xmath18 is the form laplacian of the circular graph @xmath2
. start with a point @xmath173 and take a simplex @xmath174 ( edge or vertex ) , then choosing a nonzero entry of @xmath175 is either point to the right or up . in order to have a diagonal entry of @xmath176
, we have to hit @xmath177 times @xmath178 and @xmath177 times @xmath179 .
the traces of the laplacian have a path interpretation if the diagonal entries are added as loops of negative length . for the adjacency matrix @xmath78
one has that @xmath180 is the number of closed paths of length @xmath60 in the graph .
we can use that @xmath181 and @xmath130 is the adjacency matrix of a simplex graph @xmath182 .
but it is not @xmath183 . for @xmath184 ,
where @xmath185 but @xmath186 , we have a discrepancy ; the reason is that we have defined the sum @xmath65 which for @xmath187 does not count two cases of @xmath188 , which the trace includes .
formulas like @xmath189 relate combinatorics with analysis .
it is fourier theory in the circular case .
* for @xmath187 we have @xmath190 because the sum has @xmath191 terms @xmath62 . * for @xmath192
we have @xmath193 which as a riemann sum converges for @xmath3 to @xmath194 . * for @xmath195 we have @xmath196 . * for @xmath197
, we see convergence @xmath198 . since @xmath199 this agrees in a limiting case with the convergence of @xmath200 . * for @xmath42
, we deal with the discrete analogue of the * basel problem * and the sum @xmath201 .
we see that @xmath202 which implies @xmath203 . * for @xmath204
, we numerically computed @xmath205 for @xmath206 as @xmath207 .
since no explicit expressions are known for @xmath208 , it is unlikely also that explicit expressions for @xmath47 will be available in the discrete .
* for @xmath209 , we will derive the value @xmath210 .
also this will reestablish the known value value @xmath211 for the riemann zeta function .
the finite analogue of the basel problem is the quest to find an explicit formula for @xmath212 curiously , similar type of sums have been tackled in number theory by hardy and littlewood already @xcite and where the sums are infinite over an irrational rotation number .
we have @xmath213 .
the function @xmath214 is a fixed point of the birkhoff renormalization @xmath215 therefore , @xmath216 or @xmath217 is a fixed point of @xmath218 lets write this out as @xmath219 the right hand side is now analytic at @xmath134 and is @xmath220 .
therefore , the limit @xmath221 gives @xmath222 the following argument just gives the limit @xmath78 .
chose @xmath223 and chose three sums for @xmath224 and @xmath225 : @xmath226 & \to & a \\
\frac{1}{m^2 } [ g(\frac{2}{n } ) + g(\frac{5}{n } ) + \cdots + g ( \frac{n-2}{n } ) ] & \to & g(1/3 ) = 4/3 \\
\frac{1}{m^2 } [ g(\frac{3}{n } ) + g(\frac{6}{n } ) + \cdots + g ( \frac{n-1}{n } ) ] & \to & g(2/3 ) = 4/3 \ ; , \end{aligned}\ ] ] where the limits are @xmath227 . when adding this up we get @xmath228 on the left hand side and @xmath229 on the right hand side in the limit .
from @xmath230 we get @xmath231 .
* example . * for @xmath232
we get @xmath233 .
the scalar laplacian which agrees with the laplacian on @xmath62-forms is displayed in the introduction .
the nonzero eigenvalues are @xmath234 .
the sum of the reciprocals @xmath235 indeed is equal to @xmath236 .
this is the trace of the green function .
note that since in the circle case , the scalar laplacian @xmath237 and the one form laplacian @xmath238 are the same since we have chosen only the positive part of the dirac operator @xmath133 which produces @xmath239 , we have @xmath240 , where @xmath241 is the sum of all nonzero eigenvalues .
we had to divide by @xmath242 because we only summed over the positive eigenvalues of @xmath133 .
+ the formula @xmath243 is true for all @xmath60 : for every @xmath244 , @xmath245 it is a solution of all the arithmetic recursions @xmath246 the previous derivation for @xmath247 can be generalized to @xmath248 .
it shows that @xmath249 , which is solved by @xmath250 .
* remarks . *
+ * 1 ) * @xmath251 satisfies the relation @xmath252 .
@xmath253 satisfies the relation @xmath254 .
+ * 2 ) * the answer to @xmath255 is the variance of uniformly distributed random variables in @xmath256 : @xmath257 . is this an accident ?
@xmath258 . by theorem ( [ maintheorem ] )
, we have @xmath259 . for @xmath42 , we get from the discrete basel problem the answer @xmath260 . lets look at the next larger case and compute @xmath261 : for every @xmath60 , @xmath262 in the limit this gives @xmath263 so that @xmath264 .
\a ) we can get an explicit formula for the sum @xmath265 by renormalization .
we can also get an explicit formula for @xmath266 which allows to get one for @xmath267 . +
while b ) follows from a ) , there is a simpler direct derivation of b ) which works in the case when we are only interested in the limit . with have @xmath268 , then @xmath269 which has the same singularity at @xmath270 than @xmath271 .
now @xmath272 is a fixed point of the birkhoff renormalization operator @xmath273 lets call the limit @xmath78 .
as before , we see that the limit @xmath78 satisfies @xmath274 . since @xmath275 .
we have @xmath276 . if we do the summation with @xmath277 we get @xmath278 that result which is @xmath279 .
in this section we prove the statements in theorem ( [ maintheorem ] ) .
we will use that if a sequence @xmath280 of analytic functions converges to @xmath281 uniformly in an open region @xmath11 and @xmath5 is a subset of @xmath11 and all @xmath280 have a root in @xmath5 .
then @xmath281 has a root in @xmath5 .
this is known under the name hurwitz theorem in complex analysis ( e.g. @xcite and follows from a contour integral argument . + * a ) below the critical strip .
* the first part deals with non - positive values of @xmath282 , where @xmath53 is a riemann sum . for large enough @xmath60 , there are no roots of @xmath162 in @xmath283 . below the critical strip @xmath283 ,
the sum @xmath29 is a riemann sum @xmath284 for a continuous , complex valued function @xmath285 on @xmath286 $ ] .
the function @xmath30 has no roots in @xmath283 . by hurwitz theorem
, also @xmath47 has no roots there for large enough @xmath60 .
there are still surprises .
we could show for example that @xmath104/n^s$ ] even converges even for @xmath287 .
this shows that the riemann sum @xmath29 is very close to the average @xmath30 below the critical strip .
we do not need this result however . + * to the right of the critical strip * + there is a relation between the zeta function of the circular graphs @xmath2 and the zeta function of the circle @xmath288 if @xmath25 is to the right of the critical line @xmath48 . establishing this relation requires a bit of analysis .
+ by coincidence that the name * riemann " * is involved both to the left of the critical strip with riemann sum and to the right with the riemann zeta function .
the story to the right of the critical line is not as obvious as it was in the left half plane @xmath289 .
if @xmath162 is the zeta function of the circular graph @xmath2 and @xmath50 is the zeta function of the circle ( which is the classical riemann zeta function ) , then @xmath290 for @xmath3 .
the convergence is uniform on compact sets @xmath291 and @xmath47 has no roots in @xmath5 for large enough @xmath60 .
we have to show that the sum @xmath292\ ] ] goes to zero . instead , because the riemann sums @xmath284 of @xmath293 and @xmath294 give the same result , we can by symmetry take twice the discrete riemann sum and look at the riemann sum @xmath295 with @xmath296 and @xmath297 we have to show that @xmath298 converges to zero .
this is equivalent to show that @xmath299 converges to zero .
now , @xmath300 where @xmath301 is the anti - derivative of @xmath94 and where @xmath302 is a @xmath5-derivative of @xmath94 . since @xmath53 has no roots in @xmath5 , we know by hurwitz also that @xmath47 has no roots for large enough @xmath60 .
* on the critical line * + on the critical line , we have a sum @xmath303 now @xmath304 is no more in @xmath87 but we have @xmath305 .
this shows that @xmath306 plays now the same role than @xmath307 before .
this only establishes that @xmath308 stays bounded but does not establish convergence .
we measure experimentally that @xmath309 could converge to a bounded function which has a maximum at @xmath134 .
indeed , we can see that the maximum occurs at @xmath310 and that the growth is logarithmic in @xmath60 .
we do not know whether there are roots of @xmath162 accumulating for @xmath3 on @xmath48 . + * inside the critical strip * + the following result uses rolle theory developed in the appendix : for @xmath311 , the limit @xmath312 exists .
there are no roots in the open critical strip . on the interval
@xmath313 $ ] , the function @xmath314 takes values between @xmath315 $ ] and subtracting @xmath316 gives a function @xmath94 on @xmath286 $ ] which averages to @xmath134 .
we have now a smooth bounded function @xmath94 of zero average and have to show that @xmath317 converges .
let @xmath318 be the antiderivative , now use the mean value theorem to have @xmath319 , where @xmath320 are the rolle points satisfying @xmath321 .
deviation of the rolle point from the mean .
we will see that @xmath322 we know that @xmath323/n^s$ ] converges for @xmath3 uniformly for @xmath25 on compact subsets of the strip @xmath85 .
hurwitz assures that there are no roots in the limit . remarks .
+ * 1 ) * we actually see convergence for @xmath324 .
+ * 2 ) * the above proposition does not implie that @xmath162 has no roots outside the critical strip for large enough @xmath60 .
indeed , we believe that there are for all @xmath60 getting closer and closer to the critical line . +
* 3 ) * because of the almost periodic nature of the setup we believe that for every @xmath159 there is a @xmath160 such that for @xmath161 , there are no roots in @xmath325 $ ] .
* a ) * this note has been written with the goal to understand the roots of @xmath47 and to explain their motion as @xmath60 increases .
we still do not know : we think that it the limiting accumulation set could coincide with the entire critical line . if this is not true , where are the gaps ? + *
b ) * the zeta functions considered here do not belong to classes of dirichlet series which are classically considered .
the following question is motivated from the classical symmetry for @xmath326 which satisfies @xmath327 and where @xmath328 is the involution : we think the answer is yes in the limit .
a functional equation which holds for each @xmath60 would be surprising and very interesting .
+ * c ) * the shifted zeta function @xmath329 is different from the twisted zeta function @xmath330 which motivated by the hurwitz zeta function ( @xcite for some integer values of @xmath25 ) .
the function @xmath331 has the property that @xmath332 .
we can now define a vector field @xmath333 which depends on @xmath60 and has the property that the level set @xmath334 moves with @xmath335 .
this applies especially to roots .
if we start with a root of @xmath47 at @xmath336 , we end up with a root of @xmath124 at @xmath337 .
this picture explains why the motion of the roots has a natural unit time - scale between @xmath60 and @xmath126 but it also opens the possibility that we get a limiting field : we expect a limiting vector field @xmath338 to exist in the critical strip and that @xmath338 is not zero almost everywhere on the critical line .
this would show that there are no gaps in the limiting root set .
we report now some values of the root closest to the origin .
these computations were done in the summer of 2013 and illustrate how slowly the roots move when @xmath60 is increased .
the results support the picture that the distance between neighboring roots will go to zero for @xmath3 even so this is hard to believe when naively observing the numerics at first . for integers like @xmath339
, we still see a set which appears indistinguishable from say @xmath340 , giving the impression that the roots have settled . however , the roots still move , even so very slowly : even when observing on a logarithmic time scale , the motion will slow down exponentially .
their speed therefore appears to be asymptotic to @xmath341 for some constants @xmath342 which makes their motion hard to see . for the following data ,
we have computed the first root with @xmath343 digit accuracy and stopped in each case the newton iteration when @xmath344 .
both notions involve the first three derivatives of @xmath97 .
the @xmath5-derivative is invariant under linear transformations @xmath345 and a constant for @xmath346 or @xmath347 or @xmath348 .
note that while the schwarzian derivative is invariant under fractional linear transformations @xmath349 the k - derivative is only invariant under linear transformations only .
it vanishes on quadratic functions .
+ an important example for the present zeta story is the function @xmath350 for which @xmath351 is bounded if @xmath352 . + we can now patch the classical newton - coates in situations , where the second derivative is large but the @xmath5- derivative is small .
this is especially useful for rational trigonometric functions for which the k - derivative is bounded , even so the functions are unbounded .
the following theorem will explain the convergence of @xmath353)/n^s$ ] which is scaled similarly than random variables in the central limit theorem for @xmath354 . here is the main result of this appendix :
assume @xmath97 is @xmath355 on @xmath356 and has no inflection points @xmath357 and satisfies @xmath358 for all @xmath107 .
then , for every @xmath359 and large enough @xmath60 , @xmath360 \ ; .
\end{aligned}\ ] ] we can assume that @xmath361 without loss of generality . let @xmath362 be the rolle points . then @xmath363 .
now use that @xmath364 , where @xmath150 is the upper bound for the k - derivative .
the last equation is a riemann sum .
this especially holds if @xmath97 is smooth in a neighborhood of the compact interval @xmath286 $ ] in which case the limit @xmath365 \label{renormalization1}\ ] ] exists and is linear in @xmath335 .
we will look at this
central limit theorem " story elsewhere .
the result has two consequences , which we will need : we have seen already that the k - derivative of @xmath97 is bounded .
we have @xmath374 \leq c_1 $ ] .
because the function @xmath375 has always the same sign , the rolle estimate shows that @xmath376 . applying the lemma again shows that this is the limit of @xmath377 $ ] .
we see that we have a uniform majorant and so convergence to a function which is analytic inside the strip .
given a smooth @xmath87 function @xmath97 on @xmath378 and @xmath379 , define @xmath380 and @xmath381=\int_0 ^ 1 s_n(y ) \ ; dy$ ] as well as @xmath382 = \int_0 ^ 1 ( s_n-{\rm e}[s_n])^2 \ ; dx$ ] and @xmath383=\sqrt{{\rm var}[s_n]}$ ] .
define @xmath384)}{\sigma(s_n ) } \ ; .\ ] ] * examples . *
+ * 1 ) * the function @xmath293 on @xmath385 has the k - derivative @xmath386 . the riemann sum is @xmath387 .
the integral is @xmath388 .
the theorem tells that @xmath389 is bounded .
of course , we know it converges to @xmath165 the euler - mascheroni constant .
+ * 2 ) * lets take @xmath390 on @xmath391 for which the k - derivative @xmath392 independent of @xmath393 .
assume @xmath336 .
the classical newton - coates would not have give us a good estimate for the riemann sum because the interval increases .
however , the above theorem works . applying the above theorem for @xmath394
gives for every @xmath395 @xmath396 indeed the limit for @xmath397 exists and is again * euler - mascheroni constant*. but the rolle estimate is uniform in @xmath60 .
the bound @xmath398 on the right hand side is not optimal
. one can show it to be @xmath399 . + * 3 ) * for @xmath214 , we have @xmath400
. again assume first @xmath336 .
we have @xmath401 } \cot(\pi k / n)/n + \log ( \sin(\pi x)/\pi)| \leq 1 $ ] for all @xmath60 .
the limit which exists according to the theorem is about @xmath402 .
+ * 4 ) * the theorem implies that the limit @xmath403 exists . actually ,
since this is the birkhoff normalization operator ( a special ruelle transfer operator known in dynamical systems theory ) , the limit is explicitly known to be @xmath404 if we sum from @xmath405 to @xmath144 .
+ * 5 ) * let @xmath406 with @xmath407 , where @xmath150 is @xmath62 .
we also have @xmath408 converging .
the theorem applies and the limit exists because for @xmath409 , the rolle function is bounded .
assume @xmath281 is 5 times continuously differentiable in an interval @xmath410 and that @xmath411 is not zero .
then there is @xmath412 such that the rolle point @xmath413 satisfies @xmath414 furthermore , @xmath415 [ rolleformula ] we start with the implicit equation @xmath416 from which the chain rule gives in a familiar way the implicit differentiation formula @xmath417 . a taylor expansion with lagrangian rest term @xmath418 for some @xmath419 shows that @xmath420 therefore , @xmath421 . and
the rest is clear using @xmath422
. * remark . * + using riemann sums using points which estimate the rolle points can produce a numerical method for integration of functions which are 5 times continuously differentiable in the interior of some interval but possibly unbounded at the boundary .
let @xmath423 denote the end points of a riemann partition and @xmath320 the midpoints and @xmath424 the length of the interval .
then @xmath425 \;\ ] ] has adjusted the midpoint in every interval @xmath426 $ ] by an approximation of @xmath427 . of course , there is a trade off , since we have to compute the @xmath5-derivative @xmath428 at a point .
but the result will be of the order @xmath429 close to the real integral , like simpson .
why is this helpful ?
it is the fact that the @xmath5-derivative is often finite , even so the function and its derivative can be unbounded .
any method , where intervals are divided in a definite way , like simpson , can have larger error in those cases .
an example is the function @xmath430 for @xmath431 which is integrable but where standard integration methods complain about the singularity .
actually , one of our results shows that the difference of riemann sum and integral is of the order @xmath432 .
the function , which appeared in the discrete circle zeta function , has a finite @xmath5-derivative . | we look at entire functions given as the zeta function @xmath0 , where @xmath1 are the positive eigenvalues of the laplacian of the discrete circular graph @xmath2 . we prove that the roots converge for @xmath3 to the line @xmath4 in the sense that for every compact subset @xmath5 in the complement of that line , there is a @xmath6 such that for @xmath7 , no root of the zeta function is in @xmath5 . to prove the result
we actually look at the dirac zeta function @xmath8 where @xmath1 are the positive eigenvalues of the dirac operator of the circular graph . in the case of circular graphs , the laplace zeta function is @xmath9 . |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.